We introduce Gaussian Garments, a novel approach for reconstructing realistic simulation-ready garment assets from multi-view videos. Our method represents garments with a combination of a 3D mesh and a Gaussian texture that encodes both the color and high-frequency surface details. This representation enables accurate registration of garment geometries to multi-view videos and helps disentangle albedo textures from lighting effects. Furthermore, we demonstrate how a pretrained graph neural network (GNN) can be fine-tuned to replicate the real behavior of each garment. The reconstructed Gaussian Garments can be automatically combined into multi-garment outfits and animated with the fine-tuned GNN.

Human activities are inherently complex, and even simple household tasks involve numerous object interactions. To better understand these activities, it is crucial to model their interactions with the environment captured through dynamic changes. The recent availability of affordable head-mounted cameras and egocentric data offers a more accessible and efficient means to understand dynamic human-object interactions in 3D environments. However, most existing methods for human activity modeling either focus on reconstructing 3D models of hand-object or human-scene interactions or on mapping 3D scenes, neglecting dynamic interactions with objects. The few existing solutions often require inputs from multiple sources, including multi-camera setups, depth-sensing cameras, or kinesthetic sensors. To this end, we introduce EgoGaussian, the first method capable of simultaneously reconstructing 3D scenes and dynamically tracking 3D object motion from RGB egocentric input alone. We leverage the uniquely discrete nature of Gaussian Splatting and segment dynamic interactions from the background. Our approach employs a clip-level online learning pipeline that leverages the dynamic nature of human activities, allowing us to reconstruct the temporal evolution of the scene in chronological order and track rigid object motion. Additionally, our method automatically segments object and background Gaussians, providing explicit 3D representations for both static scenes and dynamic objects. EgoGaussian shows significant improvements in terms of both dynamic object and background reconstruction quality compared to the state-of-the-art. We also qualitatively demonstrate the high quality of the reconstructed models.

Understanding drivers’ decision-making is crucial for road safety. Although predicting the ego-vehicle’s path is valuable for driver-assistance systems, existing methods mainly focus on external factors like other vehicles’ motions, often neglecting the driver’s attention and intent. To address this gap, we infer the ego-trajectory by integrating the driver’s gaze and the surrounding scene. We introduce RouteFormer, a novel multimodal ego-trajectory prediction network combining GPS data, environmental context, and the driver's field-of-view—comprising first-person video and gaze fixations. We also present the Path Complexity Index (PCI), a new metric for trajectory complexity that enables a more nuanced evaluation of challenging scenarios. To tackle data scarcity and enhance diversity, we introduce GEM, a comprehensive dataset of urban driving scenarios enriched with synchronized driver field-of-view and gaze data. Extensive evaluations on GEM and DR(eye)VE demonstrate that RouteFormer significantly outperforms state-of-the-art methods, achieving notable improvements in prediction accuracy across diverse conditions. Ablation studies reveal that incorporating driver field-of-view data yields significantly better average displacement error, especially in challenging scenarios with high PCI scores, underscoring the importance of modeling driver attention. All data and code are available at meakbiyik.github.io/routeformer.

Classifier-free guidance (CFG) is crucial for improving both generation quality and alignment between the input condition and final output in diffusion models. While a high guidance scale is generally required to enhance these aspects, it also causes oversaturation and unrealistic artifacts. In this paper, we revisit the CFG update rule and introduce modifications to address this issue. We first decompose the update term in CFG into parallel and orthogonal components with respect to the conditional model prediction and observe that the parallel component primarily causes oversaturation, while the orthogonal component enhances image quality. Accordingly, we propose down-weighting the parallel component to achieve high-quality generations without oversaturation. Additionally, we draw a connection between CFG and gradient ascent and introduce a new rescaling and momentum method for the CFG update rule based on this insight. Our approach, termed adaptive projected guidance (APG), retains the quality-boosting advantages of CFG while enabling the use of higher guidance scales without oversaturation. APG is easy to implement and introduces practically no additional computational overhead to the sampling process. Through extensive experiments, we demonstrate that APG is compatible with various conditional diffusion models and samplers, leading to improved FID, recall, and saturation scores while maintaining precision comparable to CFG, making our method a superior plug-and-play alternative to standard classifier-free guidance.

Classifier-free guidance (CFG) has become the standard method for enhancing the quality of conditional diffusion models. However, employing CFG requires either training an unconditional model alongside the main diffusion model or modifying the training procedure by periodically inserting a null condition. There is also no clear extension of CFG to unconditional models. In this paper, we revisit the core principles of CFG and introduce a new method, independent condition guidance (ICG), which provides the benefits of CFG without the need for any special training procedures. Our approach streamlines the training process of conditional diffusion models and can also be applied during inference on any pre-trained conditional model. Additionally, by leveraging the time-step information encoded in all diffusion networks, we propose an extension of CFG, called time-step guidance (TSG), which can be applied to *any* diffusion model, including unconditional ones. Our guidance techniques are easy to implement and have the same sampling cost as CFG. Through extensive experiments, we demonstrate that ICG matches the performance of standard CFG across various conditional diffusion models. Moreover, we show that TSG improves generation quality in a manner similar to CFG, without relying on any conditional information.

We propose a physics-based method for synthesizing dexterous hand-object interactions in a full-body setting. While recent advancements have addressed specific facets of human-object interactions, a comprehensive physics-based approach remains a challenge. Existing methods often focus on isolated segments of the interaction process and rely on data-driven techniques that may result in artifacts. In contrast, our proposed method embraces reinforcement learning $(R L)$ and physics simulation to mitigate the limitations of data-driven approaches. Through a hierarchical framework, we first learn skill priors for both body and hand movements in a decoupled setting. The generic skill priors learn to decode a latent skill embedding into the motion of the underlying part. A high-level policy then controls hand-object interactions in these pretrained latent spaces, guided by task objectives of grasping and 3D target trajectory following. It is trained using a novel reward function that combines an adversarial style term with a task reward, encouraging natural motions while fulfilling the task incentives. Our method successfully accomplishes the complete interaction task, from approaching an object to grasping and subsequent manipulation. We compare our approach against kinematics-based baselines and show that it leads to more physically plausible motions. Video and code are available at https://eth-ait.github.io/phys-fullbody-grasp/.

We present a method to estimate human motion in a global scene from moving cameras. This is a highly challenging task due to the coupling of human and camera motions in the video. To address this problem, we propose a joint optimization framework that disentangles human and camera motions using both foreground human motion priors and background scene features. Unlike existing methods that use SLAM as initialization, we propose to tightly integrate SLAM and human motion priors in an optimization that is inspired by bundle adjustment. Specifically, we optimize human and camera motions to match both the observed human pose and scene features. This design combines the strengths of SLAM and motion priors, which leads to significant improvements in human and camera motion estimation. We additionally introduce a motion prior that is suitable for batch optimization, making our approach significantly more efficient than existing approaches. Finally, we propose a novel synthetic dataset that enables evaluating camera motion in addition to human motion from dynamic videos. Experiments on the synthetic and real-world RICH datasets demonstrate that our approach substantially outperforms prior art in recovering both human and camera motions.

We present ArtiGrasp, a novel method to synthesize bimanual hand-object interactions that include grasping and articulation. This task is challenging due to the diversity of the global wrist motions and the precise finger control that are necessary to articulate objects. ArtiGrasp leverages reinforcement learning and physics simulations to train a policy that controls the global and local hand pose. Our framework unifies grasping and articulation within a single policy guided by a single hand pose reference. Moreover, to facilitate the training of the precise finger control required for articulation, we present a learning curriculum with increasing difficulty. It starts with single-hand manipulation of stationary objects and continues with multi-agent training including both hands and non-stationary objects. To evaluate our method, we introduce Dynamic Object Grasping and Articulation, a task that involves bringing an object into a target articulated pose. This task requires grasping, relocation, and articulation. We show our method’s efficacy towards this task. We further demonstrate that our method can generate motions with noisy hand-object pose estimates from an off-the-shelf image-based regressor. Project page: https://eth-ait.github.io/artigrasp/.

The studies of human clothing for digital avatars have predominantly relied on synthetic datasets. While easy to collect, synthetic data often fall short in realism and fail to capture authentic clothing dynamics. Addressing this gap, we introduce 4D-DRESS, the first real-world 4D dataset advancing human clothing research with its high-quality 4D textured scans and garment meshes. 4D-DRESS captures 64 outfits in 520 human motion sequences, amounting to 78k textured scans. Creating a real-world clothing dataset is challenging, particularly in annotating and segmenting the extensive and complex 4D human scans. To address this, we develop a semi-automatic 4D human parsing pipeline. We efficiently combine a human-in-the-loop process with automation to accurately label 4D scans in diverse garments and body movements. Leveraging precise annotations and high-quality garment meshes, we establish several benchmarks for clothing simulation and reconstruction. 4D-DRESS offers realistic and challenging data that complements synthetic sources, paving the way for advancements in research of lifelike human clothing.

Synthesizing natural human motions that enable a 3D human avatar to walk and reach for arbitrary goals in 3D space remains an unsolved problem with many applications. Existing methods (data-driven or using reinforcement learning) are limited in terms of generalization and motion naturalness. A primary obstacle is the scarcity of training data that combines locomotion with goal reaching. To ad-dress this, we introduce WANDR, a data-driven model that takes an avatar's initial pose and a goal's 3D position and generates natural human motions that place the end effec-tor (wrist) on the goal location. To solve this, we intro-duce novel intention features that drive rich goal-oriented movement. Intention guides the agent to the goal, and in-teractively adapts the generation to novel situations without needing to define sub-goals or the entire motion path. Cru-cially, intention allows training on datasets that have goal-oriented motions as well as those that do not. WANDR is a conditional Variational Auto-Encoder (c- VAE), which we train using the AMASS and CIRCLE datasets. We evaluate our method extensively and demonstrate its ability to gener-ate natural and long-term motions that reach 3D goals and generalize to unseen goal locations. Our models and code are available for research purposes at wandr.is.tue.mpg.de.

Since humans interact with diverse objects every day, the holistic 3D capture of these interactions is important to understand and model human behaviour. However, most ex-isting methods for hand-object reconstruction from RGB ei-ther assume pre-scanned object templates or heavily rely on limited 3D hand-object data, restricting their ability to scale and generalize to more unconstrained interaction settings. To address this, we introduce HOLD the first category-agnostic method that reconstructs an articulated hand and an object jointly from a monocular interaction video. We develop a compositional articulated implicit model that can reconstruct disentangled 3D hands and ob-jects from 2D images. We also further incorporate hand-object constraints to improve hand-object poses and con-sequently the reconstruction quality. Our method does not rely on any 3D hand-object annotations while significantly outperforming fully-supervised baselines in both in-the-lab and challenging in-the-wild settings. Moreover, we qualita-tively show its robustness in reconstructing from in-the-wild videos. See here for code, data, models, and updates.

A long-standing goal of 3D human reconstruction is to cre-ate lifelike and fully detailed 3D humans from single-view images. The main challenge lies in inferring unknown body shapes, appearances, and clothing details in areas not visi-ble in the images. To address this, we propose SiTH, a novel pipeline that uniquely integrates an image-conditioned dif-fusion model into a 3D mesh reconstruction workflow. At the core of our method lies the decomposition of the chal-lenging single-view reconstruction problem into generative hallucination and reconstruction subproblems. For the for-mer, we employ a powerful generative diffusion model to hallucinate unseen back-view appearance based on the in-put images. For the latter, we leverage skinned body meshes as guidance to recover full-body texture meshes from the in-put and back-view images. SiTH requires as few as 500 3D human scans for training while maintaining its generality and robustness to diverse images. Extensive evaluations on two 3D human benchmarks, including our newly created one, highlighted our method's superior accuracy and per-ceptual quality in 3D textured human reconstruction.

We present MultiPly, a novel framework to reconstruct multiple people in 3D from monocular in-the-wild videos. Reconstructing multiple individuals moving and interacting naturally from monocular in-the-wild videos poses a challenging task. Addressing it necessitates precise pixel-level disentanglement of individuals without any prior knowledge about the subjects. Moreover, it requires recovering intricate and complete 3D human shapes from short video sequences, intensifying the level of difficulty. To tackle these challenges, we first define a layered neural representation for the entire scene, composited by individual human and background models. We learn the layered neural representation from videos via our layer-wise differentiable volume rendering. This learning process is further enhanced by our hybrid instance segmentation approach which combines the self-supervised 3D segmentation and the promp table 2D segmentation module, yielding reliable instance segmentation supervision even under close human interaction. A confidence-guided optimization formulation is introduced to optimize the human poses and shape/appearance alternately. We incorporate effective objectives to refine human poses via photometric information and impose physically plausible constraints on human dynamics, leading to temporally consistent 3D reconstructions with high fidelity. The evaluation of our method shows the superiority over prior art on publicly available datasets and in-the-wild videos.

We study object interaction anticipation in egocentric videos. This task requires an understanding of the spatio-temporal context formed by past actions on objects, coined action context. We propose TransFusion, a multimodal transformer-based architecture for short-term object interaction anticipation. Our method exploits the representational power of language by summarizing the action con-text textually, after leveraging pre-trained vision-language foundation models to extract the action context from past video frames. The summarized action context and the last observed video frame are processed by the multimodal fusion module to forecast the next object interaction. Experiments on the Ego4D next active object interaction dataset show the effectiveness of our multimodal fusion model and highlight the benefits of using the power of foundation models and language-based context summaries in a task where vision may appear to suffice. Our novel approach outperforms all state-of-the-art methods on both versions of the Ego4D dataset. A project video and code are available at https://eth-ait.github.io/transfusion-proj/.

We present HAAR, a new strand-based generative model for 3D human hairstyles. Specifically, based on textual inputs, HAAR produces 3D hairstyles that could be used as production-level assets in modern computer graphics engines. Current AI-based generative models take advantage of powerful 2D priors to reconstruct 3D content in the form of point clouds, meshes, or volumetric functions. However, by using the 2D priors, they are intrinsically limited to only recovering the visual parts. Highly occluded hair structures can not be reconstructed with those methods, and they only model the “outer shell”, which is not ready to be used in physics-based rendering or simulation pipelines. In contrast, we propose a first text-guided generative method that uses 3D hair strands as an underlying representation. Leveraging 2D visual question-answering (VQA) systems, we automatically annotate synthetic hair models that are generated from a small set of artist-created hairstyles. This allows us to train a latent diffusion model that operates in a common hairstyle UV space. In qualitative and quantitative studies, we demonstrate the capabilities of the proposed model and compare it to existing hairstyle generation approaches. For results, please refer to our project page.

Large-scale diffusion generative models are greatly sim-plifying image, video and 3D asset creation from user-provided text prompts and images. However, the challenging problem of text-to-4D dynamic 3D scene generation with diffusion guidance remains largely unexplored. We propose Dream-in-4D, which features a novel two-stage approach for text-to-4D synthesis, leveraging (1) 3D and 2D diffusion guidance to effectively learn a high-quality static 3D asset in the first stage; (2) a deformable neural radiance field that explicitly disentangles the learned static asset from its deformation, preserving quality during motion learning; and (3) a multi-resolution feature grid for the deformation field with a displacement total variation loss to effectively learn motion with video diffusion guidance in the second stage. Through a user preference study, we demon-strate that our approach significantly advances image and motion quality, 3D consistency and text fidelity for text-to-4D generation compared to baseline approaches. Thanks to its motion-disentangled representation, Dream-in-4D can also be easily adapted for controllable generation where appearance is defined by one or multiple images, without the need to modify the motion learning stage. Thus, our method offers, for the first time, a unified approach for text-to-4D, image-to-4D and personalized 4D generation tasks.

While conditional diffusion models are known to have good coverage of the data distribution, they still face limitations in output diversity, particularly when sampled with a high classifier-free guidance scale for optimal image quality or when trained on small datasets. We attribute this problem to the role of the conditioning signal in inference and offer an improved sampling strategy for diffusion models that can increase generation diversity, especially at high guidance scales, with minimal loss of sample quality. Our sampling strategy anneals the conditioning signal by adding scheduled, monotonically decreasing Gaussian noise to the conditioning vector during inference to balance diversity and condition alignment. Our Condition-Annealed Diffusion Sampler (CADS) can be used with any pretrained model and sampling algorithm, and we show that it boosts the diversity of diffusion models in various conditional generation tasks. Further, using an existing pretrained diffusion model, CADS achieves a new state-of-the-art FID of 1.70 and 2.31 for class-conditional ImageNet generation at 256$\times$256 and 512$\times$512 respectively.

Vision-based human-to-robot handover is an important and challenging task in human-robot interaction. Recent work has attempted to train robot policies by interacting with dynamic virtual humans in simulated environments, where the policies can later be transferred to the real world. However, a major bottleneck is the reliance on human motion capture data, which is expensive to acquire and difficult to scale to arbitrary objects and human grasping motions. In this paper, we introduce a framework that can generate plausible human grasping motions suitable for training the robot. To achieve this, we propose a hand-object synthesis method that is designed to generate handover-friendly motions similar to humans. This allows us to generate synthetic training and testing data with 100x more objects than previous work. In our experiments, we show that our method trained purely with synthetic data is competitive with state-of-the-art methods that rely on real human motion data both in simulation and on a real system. In addition, we can perform evaluations on a larger scale compared to prior work. With our newly introduced test set, we show that our model can better scale to a large variety of unseen objects and human motions compared to the baselines.

Advances in latent diffusion models (LDMs) have revolutionized high-resolution image generation, but the design space of the autoencoder that is central to these systems remains underexplored. In this paper, we introduce LiteVAE, a new autoencoder design for LDMs, which leverages the 2D discrete wavelet transform to enhance scalability and computational efficiency over standard variational autoencoders (VAEs) with no sacrifice in output quality. We investigate the training methodologies and the decoder architecture of LiteVAE and propose several enhancements that improve the training dynamics and reconstruction quality. Our base LiteVAE model matches the quality of the established VAEs in current LDMs with a six-fold reduction in encoder parameters, leading to faster training and lower GPU memory requirements, while our larger model outperforms VAEs of comparable complexity across all evaluated metrics (rFID, LPIPS, PSNR, and SSIM).

Volumetric modeling and neural radiance field representations have revolutionized 3D face capture and photorealistic novel view synthesis. However, these methods often require hundreds of multi-view input images and are thus inapplicable to cases with less than a handful of inputs. We present a novel volumetric prior on human faces that allows for high-fidelity expressive face modeling from as few as three input views captured in the wild. Our key insight is that an implicit prior trained on synthetic data alone can generalize to extremely challenging real-world identities and expressions and render novel views with fine idiosyncratic details like wrinkles and eyelashes. We leverage a 3D Morphable Face Model to synthesize a large training set, rendering each identity with different expressions, hair, clothing, and other assets. We then train a conditional Neural Radiance Field prior on this synthetic dataset and, at inference time, fine-tune the model on a very sparse set of real images of a single subject. On average, the fine-tuning requires only three inputs to cross the synthetic-to-real domain gap. The resulting personalized 3D model reconstructs strong idiosyncratic facial expressions and outperforms the state-of-the-art in high-quality novel view synthesis of faces from sparse inputs in terms of perceptual and photo-metric quality.

Abstract Eye gaze and expressions are crucial non‐verbal signals in face‐to‐face communication. Visual effects and telepresence demand significant improvements in personalized tracking, animation, and synthesis of the eye region to achieve true immersion. Morphable face models, in combination with coordinate‐based neural volumetric representations, show promise in solving the difficult problem of reconstructing intricate geometry (eyelashes) and synthesizing photorealistic appearance variations (wrinkles and specularities) of eye performances. We propose a novel hybrid representation ‐ ShellNeRF ‐ that builds a discretized volume around a 3DMM face mesh using concentric surfaces to model the deformable ‘periocular’ region. We define a canonical space using the UV layout of the shells that constrains the space of dense correspondence search. Combined with an explicit eyeball mesh for modeling corneal light‐transport, our model allows for animatable photorealistic 3D synthesis of the whole eye region. Using multi‐view video input, we demonstrate significant improvements over state‐of‐the‐art in expression re‐enactment and transfer for high‐resolution close‐up views of the eye region.

As the number of selectable items increases, point-and-click interfaces rapidly become complex, leading to a decrease in usability. Adaptive user interfaces can reduce this complexity by automatically adjusting an interface to only display the most relevant items. A core challenge for developing adaptive interfaces is to infer user intent and chose adaptations accordingly. Current methods rely on tediously hand-crafted rules or carefully collected user data. Furthermore, heuristics need to be recrafted and data regathered for every new task and interface. To address this issue, we formulate interface adaptation as a multi-agent reinforcement learning problem. Our approach learns adaptation policies without relying on heuristics or real user data, facilitating the development of adaptive interfaces across various tasks with minimal adjustments needed. In our formulation, a user agent mimics a real user and learns to interact with an interface via point-and-click actions. Simultaneously, an interface agent learns interface adaptations, to maximize the user agent's efficiency, by observing the user agent's behavior. For our evaluation, we substituted the simulated user agent with actual users. Our study involved twelve participants and concentrated on automatic toolbar item assignment. The results show that the policies we developed in simulation effectively apply to real users. These users were able to complete tasks with fewer actions and in similar times compared to methods trained with real data. Additionally, we demonstrated our method's efficiency and generalizability across four different interfaces and tasks.

We propose the first framework to learn control policies for vision-based human-to-robot handovers, a critical task for human-robot interaction. While research in Embodied AI has made significant progress in training robot agents in simulated environments, interacting with humans remains challenging due to the difficulties of simulating humans. Fortunately, recent research has developed realistic simulated environments for human-to-robot handovers. Leveraging this result, we introduce a method that is trained with a human-in-the-loop via a two-stage teacher-student framework that uses motion and grasp planning, reinforcement learning, and self-supervision. We show significant performance gains over baselines on a simulation benchmark, sim-to-sim transfer and sim-to-real transfer. Video and code are available at https://handover-sim2real.github.io.

Humans intuitively understand that inanimate objects do not move by themselves, but that state changes are typically caused by human manipulation (e.g., the opening of a book). This is not yet the case for machines. In part this is because there exist no datasets with ground-truth 3D annotations for the study of physically consistent and synchronised motion of hands and articulated objects. To this end, we introduce ARCTIC - a dataset of two hands that dexterously manipulate objects, containing 2 .1M video frames paired with accurate 3D hand and object meshes and detailed, dynamic contact information. It contains bi-manual articulation of objects such as scissors or laptops, where hand poses and object states evolve jointly in time. We propose two novel articulated hand-object interaction tasks: (1) Consistent motion reconstruction: Given a monocular video, the goal is to reconstruct two hands and articulated objects in 3D, so that their motions are spatio-temporally consistent. (2) Interaction field estimation: Dense relative hand-object distances must be estimated from images. We introduce two baselines ArcticNet and InterField, respectively and evaluate them qualitatively and quantitatively on ARCTIC. Our code and data are available at https://arctic.is.tue.mpg.de.

We propose a method that leverages graph neural networks, multi-level message passing, and unsupervised training to enable efficient prediction of realistic clothing dynamics. Whereas existing methods based on linear blend skinning must be trained for specific garments, our method, called HOOD, is agnostic to body shape and applies to tight-fitting garments as well as loose, free-flowing clothing. Furthermore, HOOD handles changes in topology (e.g., garments with buttons or zippers) and material properties at inference time. As one key contribution, we propose a hierarchical message-passing scheme that efficiently propagates stiff stretching modes while preserving local detail. We empirically show that HOOD outperforms strong baselines quantitatively and that its results are perceived as more realistic than state-of-the-art methods.

We present Vid2Avatar, a method to learn human avatars from monocular in-the-wild videos. Reconstructing humans that move naturally from monocular in-the-wild videos is difficult. Solving it requires accurately separating humans from arbitrary backgrounds. Moreover, it requires reconstructing detailed 3D surface from short video sequences, making it even more challenging. Despite these challenges, our method does not require any groundtruth supervision or priors extracted from large datasets of clothed human scans, nor do we rely on any external segmentation modules. Instead, it solves the tasks of scene decomposition and surface reconstruction directly in 3D by modeling both the human and the background in the scene jointly, parameterized via two separate neural fields. Specifically, we define a temporally consistent human representation in canonical space and formulate a global optimization over the background model, the canonical human shape and texture, and per-frame human pose parameters. A coarse-to-fine sampling strategy for volume rendering and novel objectives are introduced for a clean separation of dynamic human and static background, yielding detailed and robust 3D human reconstructions. The evaluation of our method shows improvements over prior art on publicly available datasets.

In this paper, we propose a novel hybrid representation and end-to-end trainable network architecture to model fully editable and customizable neural avatars. At the core of our work lies a representation that combines the modeling power of neural fields with the ease of use and inherent 3D consistency of skinned meshes. To this end, we construct a trainable feature codebook to store local geometry and texture features on the vertices of a deformable body model, thus exploiting its consistent topology under articulation. This representation is then employed in a generative auto-decoder architecture that admits fitting to unseen scans and sampling of realistic avatars with varied appearances and geometries. Furthermore, our representation allows local editing by swapping local features between 3D assets. To verify our method for avatar creation and editing, we contribute a new highquality dataset, dubbed CustomHumans, for training and evaluation. Our experiments quantitatively and qualitatively show that our method generates diverse detailed avatars and achieves better model fitting performance compared to state-of-the-art methods. Our code and dataset are available at https://ait.ethz.ch/customhumans.

In this paper, we take one step further towards real-world applicability of monocular neural avatar reconstruction by contributing InstantAvatar, a system that can reconstruct human avatars from a monocular video within seconds, and these avatars can be animated and rendered at an inter-active rate. To achieve this efficiency we propose a carefully designed and engineered system, that leverages emerging acceleration structures for neural fields, in combination with an efficient empty-space skipping strategy for dynamic scenes. We also contribute an efficient implementation that we will make available for research purposes. Compared to existing methods, InstantAvatar converges 130x faster and can be trained in minutes instead of hours. It achieves comparable or even better reconstruction quality and novel pose synthesis results. When given the same time budget, our method significantly outperforms SoTA methods. In-stantAvatar can yield acceptable visual quality in as little as 10 seconds training time. For code and more demo results, please refer to https://ait.ethz.ch/InstantAvatar.

We present HARP (HAnd Reconstruction and Personalization), a personalized hand avatar creation approach that takes a short monocular RGB video of a human hand as input and reconstructs a faithful hand avatar exhibiting a high-fidelity appearance and geometry. In contrast to the major trend of neural implicit representations, HARP models a hand with a mesh-based parametric hand model, a vertex displacement map, a normal map, and an albedo without any neural components. The explicit nature of our representation enables a truly scalable, robust, and efficient approach to hand avatar creation as validated by our experiments. HARP is optimized via gradient descent from a short sequence captured by a hand-held mobile phone and can be directly used in AR/VR applications with realtime rendering capability. To enable this, we carefully design and implement a shadow-aware differentiable rendering scheme that is robust to high degree articulations and self-shadowing regularly present in hand motions, as well as challenging lighting conditions. It also generalizes to unseen poses and novel viewpoints, producing photo-realistic renderings of hand animations. Furthermore, the learned HARP representation can be used for improving 3D hand pose estimation quality in challenging viewpoints. The key advantages of HARP are validated by the in-depth analyses on appearance reconstruction, novel view and novel pose synthesis, and 3D hand pose refinement. It is an AR/VR-ready personalized hand representation that shows superior fidelity and scalability.

We present Depth-aware Image-based NEural Radiance fields (DINER). Given a sparse set of RGB input views, we predict depth and feature maps to guide the reconstruction of a volumetric scene representation that allows us to render 3D objects under novel views. Specifically, we propose novel techniques to incorporate depth information into feature fusion and efficient scene sampling. In comparison to the previous state of the art, DINER achieves higher synthesis quality and can process input views with greater disparity. This allows us to capture scenes more completely without changing capturing hardware requirements and ultimately enables larger viewpoint changes during novel view synthesis. We evaluate our method by synthesizing novel views, both for human heads and for general objects, and observe significantly improved qualitative results and increased perceptual metrics compared to the previous state of the art. The code is publicly available through the Project Webpage.

We propose GazeNeRF, a 3D-aware method for the task of gaze redirection. Existing gaze redirection methods operate on 2D images and struggle to generate 3D consistent results. Instead, we build on the intuition that the face region and eyeballs are separate 3D structures that move in a coordinated yet independent fashion. Our method leverages recent advancements in conditional image-based neural radiance fields and proposes a two-stream architecture that predicts volumetric features for the face and eye regions separately. Rigidly transforming the eye features via a 3D rotation matrix provides fine-grained control over the desired gaze angle. The final, redirected image is then attained via differentiable volume compositing. Our experiments show that this architecture outperforms naively conditioned NeRF baselines as well as previous state-of-the-art 2D gaze redirection methods in terms of redirection accuracy and identity preservation. Code and models will be released for research purposes.

We present X-Avatar, a novel avatar model that captures the full expressiveness of digital humans to bring about life-like experiences in telepresence, AR/VR and beyond. Our method models bodies, hands, facial expressions and appearance in a holistic fashion and can be learned from either full 3D scans or RGB-D data. To achieve this, we propose a part-aware learned forward skinning module that can be driven by the parameter space of SMPL-X, allowing for expressive animation of X-Avatars. To efficiently learn the neural shape and deformation fields, we propose novel part-aware sampling and initialization strategies. This leads to higher fidelity results, especially for smaller body parts while maintaining efficient training despite increased number of articulated bones. To capture the appearance of the avatar with high-frequency details, we extend the geometry and deformation fields with a texture network that is conditioned on pose, facial expression, geometry and the normals of the deformed surface. We show experimentally that our method outperforms strong baselines both quantitatively and qualitatively on the animation task. To facilitate future research on expressive avatars we contribute a new dataset, called X-Humans, containing 233 sequences of high-quality textured scans from 20 participants, totalling 35,500 data frames. Project page: https://ait.ethz.ch/X-Avatar.

We propose Hi4D, a method and dataset for the automatic analysis of physically close human-human interaction under prolonged contact. Robustly disentangling several in-contact subjects is a challenging task due to occlusions and complex shapes. Hence, existing multi-view systems typically fuse 3D surfaces of close subjects into a single, connected mesh. To address this issue we leverage i) individually fitted neural implicit avatars; ii) an alternating optimization scheme that refines pose and surface through periods of close proximity; and iii) thus segment the fused raw scans into individual instances. From these instances we compile Hi4D dataset of 4D textured scans of 20 subject pairs, 100 sequences, and a total of more than 11 K frames. Hi4D contains rich interaction-centric annotations in 2D and 3D alongside accurately registered parametric body models. We define varied human pose and shape estimation tasks on this dataset and provide results from state-of-the-art methods on these benchmarks. Hi4D dataset can be found at https://ait.ethz.ch/Hi4D.

The ability to create realistic animatable and relightable head avatars from casual video sequences would open up wide ranging applications in communication and entertainment. Current methods either build on explicit 3D morphable meshes (3DMM) or exploit neural implicit representations. The former are limited by fixed topology, while the latter are non-trivial to deform and inefficient to render. Furthermore, existing approaches entangle lighting and albedo, limiting the ability to re-render the avatar in new environments. In contrast, we propose PointAvatar, a deformable point-based representation that disentangles the source color into intrinsic albedo and normal-dependent shading. We demonstrate that PointAvatar bridges the gap between existing mesh- and implicit representations, combining high-quality geometry and appearance with topological flexibility, ease of deformation and rendering efficiency. We show that our method is able to generate animatable 3D avatars using monocular videos from multiple sources including hand-held smartphones, laptop webcams and internet videos, achieving state-of-the-art quality in challenging cases where previous methods fail, e.g., thin hair strands, while being significantly more efficient in training than competing methods.

NeRFs have enabled highly realistic synthesis of human faces including complex appearance and reflectance effects of hair and skin. These methods typically require a large number of multi-view input images, making the process hardware intensive and cumbersome, limiting applicability to unconstrained settings. We propose a novel volumetric human face prior that enables the synthesis of ultra high-resolution novel views of subjects that are not part of the prior’s training distribution. This prior model consists of an identity-conditioned NeRF, trained on a dataset of low-resolution multi-view images of diverse humans with known camera calibration. A simple sparse landmark-based 3D alignment of the training dataset allows our model to learn a smooth latent space of geometry and appearance despite a limited number of training identities. A high-quality volumetric representation of a novel subject can be obtained by model fitting to 2 or 3 camera views of arbitrary resolution. Importantly, our method requires as few as two views of casually captured images as input at inference time.

While progress in 2D generative models of human appearance has been rapid, many applications require 3D avatars that can be animated and rendered. Unfortunately, most existing methods for learning generative models of 3D humans with diverse shape and appearance require 3D training data, which is limited and expensive to acquire. The key to progress is hence to learn generative models of 3D avatars from abundant unstructured 2D image collections. However, learning realistic and complete 3D appearance and geometry in this under-constrained setting remains challenging, especially in the presence of loose clothing such as dresses. In this paper, we propose a new adversarial generative model of realistic 3D people learned from 2D images. Our method captures shape and deformation of the body and loose clothing by adopting a holistic 3D generator and integrating an efficient, flexible, articulation module. To improve realism, we train our model using multiple discriminators while also integrating geometric cues in the form of predicted 2D normal maps. We experimentally find that our method outperforms previous 3D- and articulation-aware methods in terms of geometry and appearance. We validate the effectiveness of our model and the importance of each component via systematic ablation studies.

We present EMDB, the Electromagnetic Database of Global 3D Human Pose and Shape in the Wild. EMDB is a novel dataset that contains high-quality 3D SMPL pose and shape parameters with global body and camera trajectories for in-the-wild videos. We use body-worn, wireless electromagnetic (EM) sensors and a hand-held iPhone to record a total of 58 minutes of motion data, distributed over 81 indoor and outdoor sequences and 10 participants. Together with accurate body poses and shapes, we also provide global camera poses and body root trajectories. To construct EMDB, we propose a multi-stage optimization procedure, which first fits SMPL to the 6-DoF EM measurements and then refines the poses via image observations. To achieve high-quality results, we leverage a neural implicit avatar model to reconstruct detailed human surface geometry and appearance, which allows for improved alignment and smoothness via a dense pixel-level objective. Our evaluations, conducted with a multi-view volumetric capture system, indicate that EMDB has an expected accuracy of 2.3 cm positional and 10.6 degrees angular error, surpassing the accuracy of previous in-the-wild datasets. We evaluate existing state-of-the-art monocular RGB methods for camera-relative and global pose estimation on EMDB. EMDB is publicly available under https://ait.ethz.ch/emdb.

We propose a method to estimate 3D human poses from substantially blurred images. The key idea is to tackle the inverse problem of image deblurring by modeling the forward problem with a 3D human model, a texture map, and a sequence of poses to describe human motion. The blurring process is then modeled by a temporal image aggregation step. Using a differentiable renderer, we can solve the inverse problem by backpropagating the pixel-wise re-projection error to recover the best human motion representation that explains a single or multiple input images. Since the image reconstruction loss alone is insufficient, we present additional regularization terms. To the best of our knowledge, we present the first method to tackle this problem. Our method consistently outperforms other methods on significantly blurry inputs since they lack one or multiple key functionalities that our method unifies, i.e. image deblurring with sub-frame accuracy and explicit 3D modeling of non-rigid human motion.

Real-world robotic manipulation tasks remain an elusive challenge, since they involve both fine-grained environment interaction, as well as the ability to plan for long-horizon goals. Although deep reinforcement learning (RL) methods have shown encouraging results when planning end-to-end in high-dimensional environments, they remain fundamentally limited by poor sample efficiency due to inefficient exploration, and by the complexity of credit assignment over long horizons. In this work, we present Efficient Learning of High-Level Plans from Play (ELF-P), a framework for robotic learning that bridges motion planning and deep RL to achieve long-horizon complex manipulation tasks. We leverage task-agnostic play data to learn a discrete behavioral prior over object-centric primitives, modeling their feasibility given the current context. We then design a high-level goal-conditioned policy which (1) uses primitives as building blocks to scaffold complex long-horizon tasks and (2) leverages the behavioral prior to accelerate learning. We demonstrate that ELF-P has significantly better sample efficiency than relevant baselines over multiple realistic manipulation tasks and learns policies that can be easily transferred to physical hardware.

Neural fields have revolutionized the area of 3D reconstruction and novel view synthesis of rigid scenes. A key challenge in making such methods applicable to articulated objects, such as the human body, is to model the deformation of 3D locations between the rest pose (a canonical space) and the deformed space. We propose a new articulation module for neural fields, Fast-SNARF, which finds accurate correspondences between canonical space and posed space via iterative root finding. Fast-SNARF is a drop-in replacement in functionality to our previous work, SNARF, while significantly improving its computational efficiency. We contribute several algorithmic and implementation improvements over SNARF, yielding a speed-up of 150×. These improvements include voxel-based correspondence search, pre-computing the linear blend skinning function, and an efficient software implementation with CUDA kernels. Fast-SNARF enables efficient and simultaneous optimization of shape and skinning weights given deformed observations without correspondences (e.g. 3D meshes). Because learning of deformation maps is a crucial component in many 3D human avatar methods and since Fast-SNARF provides a computationally efficient solution, we believe that this work represents a significant step towards the practical creation of 3D virtual humans.

Hugs are complex affective interactions that often include gestures like squeezes. We present six new guidelines for designing interactive hugging robots, which we validate through two studies with our custom robot. To achieve autonomy, we investigated robot responses to four human intra-hug gestures: holding, rubbing, patting, and squeezing. A Total of 32 users each exchanged and rated 16 hugs with an experimenter-controlled HuggieBot 2.0. The robot’s inflated torso’s microphone and pressure sensor collected data of the subjects’ demonstrations that were used to develop a perceptual algorithm that classifies user actions with 88% accuracy. Users enjoyed robot squeezes, regardless of their performed action, they valued variety in the robot response, and they appreciated robot-initiated intra-hug gestures. From average user ratings, we created a probabilistic behavior algorithm that chooses robot responses in real time. We implemented improvements to the robot platform to create HuggieBot 3.0 and then validated its gesture perception system and behavior algorithm with 16 users. The robot’s responses and proactive gestures were greatly enjoyed. Users found the robot more natural, enjoyable, and intelligent in the last phase of the experiment than in the first. After the study, they felt more understood by the robot and thought robots were nicer to hug.

Our goal is to efficiently learn personalized animatable 3D head avatars from videos that are geometrically accurate, realistic, relightable, and compatible with current rendering systems. While 3D meshes enable efficient processing and are highly portable, they lack realism in terms of shape and appearance. Neural representations, on the other hand, are realistic but lack compatibility and are slow to train and render. Our key insight is that it is possible to efficiently learn high-fidelity 3D mesh representations via differentiable rendering by exploiting highly-optimized methods from traditional computer graphics and approximating some of the components with neural networks. To that end, we introduce FLARE, a technique that enables the creation of animatable and relightable mesh avatars from a single monocular video. First, we learn a canonical geometry using a mesh representation, enabling efficient differentiable rasterization and straightforward animation via learned blendshapes and linear blend skinning weights. Second, we follow physically-based rendering and factor observed colors into intrinsic albedo, roughness, and a neural representation of the illumination, allowing the learned avatars to be relit in novel scenes. Since our input videos are captured on a single device with a narrow field of view, modeling the surrounding environment light is non-trivial. Based on the split-sum approximation for modeling specular reflections, we address this by approximating the prefiltered environment map with a multi-layer perceptron (MLP) modulated by the surface roughness, eliminating the need to explicitly model the light. We demonstrate that our mesh-based avatar formulation, combined with learned deformation, material, and lighting MLPs, produces avatars with high-quality geometry and appearance, while also being efficient to train and render compared to existing approaches.

We present a method for inferring diverse 3D models of human-object interactions from images. Reasoning about how humans interact with objects in complex scenes from a single 2D image is a challenging task given ambiguities arising from the loss of information through projection. In addition, modeling 3D interactions requires the generalization ability towards diverse object categories and interaction types. We propose an action-conditioned modeling of interactions that allows us to infer diverse 3D arrangements of humans and objects without supervision on contact regions or 3D scene geometry. Our method extracts high-level commonsense knowledge from large language models (such as GPT-3), and applies them to perform 3D reasoning of human-object interactions. Our key insight is priors extracted from large language models can help in reasoning about human-object contacts from textural prompts only. We quantitatively evaluate the inferred 3D models on a large human-object interaction dataset and show how our method leads to better 3D reconstructions. We further qualitatively evaluate the effectiveness of our method on real images and demonstrate its generalizability towards interaction types and object categories.

We introduce TempCLR, a new time-coherent contrastive learning approach for the structured regression task of 3D hand reconstruction. Unlike previous time-contrastive methods for hand pose estimation, our framework considers temporal consistency in its augmentation scheme, and accounts for the differences of hand poses along the temporal direction. Our data-driven method leverages unlabelled videos and a standard CNN, without relying on synthetic data, pseudo-labels, or specialized architectures. Our approach improves the performance of fully-supervised hand reconstruction methods by 15.9% and 7.6% in PA-V2V on the HO-3D and FreiHAND datasets respectively, thus establishing new state-of-the-art performance. Finally, we demonstrate that our approach produces smoother hand re-constructions through time, and is more robust to heavy occlusions compared to the previous state-of-the-art which we show quantitatively and qualitatively. Our code and models will be available at https://eth-ait.github.io/tempclr.

To make 3D human avatars widely available, we must be able to generate a variety of 3D virtual humans with varied identities and shapes in arbitrary poses. This task is chal-lenging due to the diversity of clothed body shapes, their complex articulations, and the resulting rich, yet stochas-tic geometric detail in clothing. Hence, current methods that represent 3D people do not provide a full generative model of people in clothing. In this paper, we propose a novel method that learns to generate detailed 3D shapes of people in a variety of garments with corresponding skin-ning weights. Specifically, we devise a multi-subject forward skinning module that is learned from only a few posed, unrigged scans per subject. To capture the stochastic nature of high-frequency details in garments, we leverage an adversarial loss formulation that encourages the model to capture the underlying statistics. We provide empirical evi-dence that this leads to realistic generation of local details such as wrinkles. We show that our model is able to gen-erate natural human avatars wearing diverse and detailed clothing. Furthermore, we show that our method can be used on the task of fitting human models to raw scans, out-performing the previous state-of-the-art.

We introduce the dynamic grasp synthesis task: given an object with a known 6D pose and a grasp reference, our goal is to generate motions that move the object to a target 6D pose. This is challenging, because it requires reasoning about the complex articulation of the human hand and the intricate physical interaction with the object. We propose a novel method that frames this problem in the reinforcement learning framework and leverages a physics simulation, both to learn and to evaluate such dynamic interactions. A hierarchical approach decomposes the task into low-level grasping and high-level motion synthesis. It can be used to generate novel hand sequences that approach, grasp, and move an object to a desired location, while retaining human-likeness. We show that our approach leads to stable grasps and generates a wide range of motions. Furthermore, even imperfect labels can be corrected by our method to generate dynamic interaction sequences. Video and code are available at: https://eth-ait.github.io/d-grasp/.

We present a novel method to learn Personalized Implicit Neural Avatars (PINA) from a short RGB-D sequence. This allows non-expert users to create a detailed and personal-ized virtual copy of themselves, which can be animated with realistic clothing deformations. PINA does not require complete scans, nor does it require a prior learned from large datasets of clothed humans. Learning a complete avatar in this setting is challenging, since only few depth observations are available, which are noisy and incomplete (i.e. only partial visibility of the body per frame). We propose a method to learn the shape and non-rigid deformations via a pose-conditioned implicit surface and a deformation field, defined in canonical space. This allows us to fuse all partial observations into a single consistent canonical representation. Fusion is formulated as a global optimization problem over the pose, shape and skinning parameters. The method can learn neural avatars from real noisy RGB-D sequences for a diverse set of people and clothing styles and these avatars can be animated given unseen motion sequences.

Traditional 3D morphable face models (3DMMs) provide fine-grained control over expression but cannot easily capture geometric and appearance details. Neural volumetric representations approach photorealism but are hard to animate and do not generalize well to unseen expressions. To tackle this problem, we propose IMavatar (Implicit Morphable avatar), a novel method for learning implicit head avatars from monocular videos. Inspired by the fine-grained control mechanisms afforded by conventional 3DMMs, we represent the expression- and pose-related deformations via learned blendshapes and skinning fields. These attributes are pose-independent and can be used to morph the canonical geometry and texture fields given novel expression and pose parameters. We employ ray marching and iterative root-finding to locate the canonical surface intersection for each pixel. A key contribution is our novel analytical gradient formulation that enables end-to-end training of IMavatars from videos. We show quantitatively and qualitatively that our method improves geometry and covers a more complete expression space compared to state-of-the-art methods. Code and data can be found at https://ait.ethz.ch/projects/2022/IMavatar/.

Garments with the ability to provide kinesthetic force-feedback on-demand can augment human capabilities in a non-obtrusive way, enabling numerous applications in VR haptics, motion assistance, and robotic control. However, designing such garments is a complex, and often manual task, particularly when the goal is to resist multiple motions with a single design. In this work, we propose a computational pipeline for designing connecting structures between active components—one of the central challenges in this context. We focus on electrostatic (ES) clutches that are compliant in their passive state while strongly resisting elongation when activated. Our method automatically computes optimized connecting structures that efficiently resist a range of pre-defined body motions on demand. We propose a novel dual-objective optimization approach to simultaneously maximize the resistance to motion when clutches are active, while minimizing resistance when inactive. We demonstrate our method on a set of problems involving different body sites and a range of motions. We further fabricate and evaluate a subset of our automatically created designs against manually created baselines using mechanical testing and in a VR pointing study.

Abstract Kinesthetic garments provide physical feedback on body posture and motion through tailored distributions of reinforced material. Their ability to selectively stiffen a garment's response to specific motions makes them appealing for rehabilitation, sports, robotics, and many other application fields. However, finding designs that distribute a given amount of reinforcement material to maximally stiffen the response to specified motions is a challenging problem. In this work, we propose an optimization‐driven approach for automated design of reinforcement patterns for kinesthetic garments. Our main contribution is to cast this design task as an on‐body topology optimization problem. Our method allows designers to explore a continuous range of designs corresponding to various amounts of reinforcement coverage. Our model captures both tight contact and lift‐off separation between cloth and body. We demonstrate our method on a variety of reinforcement design problems for different body sites and motions. Optimal designs lead to a two‐ to threefold improvement in performance in terms of energy density. A set of manufactured designs were consistently rated as providing more resistance than baselines in a comparative user study.

A unique challenge in creating high-quality animatable and relightable 3D avatars of real people is modeling human eyes, particularly in conjunction with the surrounding periocular face region. The challenge of synthesizing eyes is multifold as it requires 1) appropriate representations for the various components of the eye and the periocular region for coherent viewpoint synthesis, capable of representing diffuse, refractive and highly reflective surfaces, 2) disentangling skin and eye appearance from environmental illumination such that it may be rendered under novel lighting conditions, and 3) capturing eyeball motion and the deformation of the surrounding skin to enable re-gazing. These challenges have traditionally necessitated the use of expensive and cumbersome capture setups to obtain high-quality results, and even then, modeling of the full eye region holistically has remained elusive. We present a novel geometry and appearance representation that enables high-fidelity capture and photorealistic animation, view synthesis and relighting of the eye region using only a sparse set of lights and cameras. Our hybrid representation combines an explicit parametric surface model for the eyeball surface with implicit deformable volumetric representations for the periocular region and the interior of the eye. This novel hybrid model has been designed specifically to address the various parts of that exceptionally challenging facial area - the explicit eyeball surface allows modeling refraction and high frequency specular reflection at the cornea, whereas the implicit representation is well suited to model lower frequency skin reflection via spherical harmonics and can represent non-surface structures such as hair (i.e. eyebrows) or highly diffuse volumetric bodies (i.e. sclera), both of which are a challenge for explicit surface models. Tightly integrating the two representations in a joint framework allows controlled photoreal image synthesis and joint optimization of both the geometry parameters of the eyeball and the implicit neural network in continuous 3D space. We show that for high-resolution close-ups of the human eye, our model can synthesize high-fidelity animated gaze from novel views under unseen illumination conditions, allowing to generate visually rich eye imagery.

We propose a novel Transformer-based architecture for the task of generative modelling of 3D human motion. Previous work commonly relies on RNN-based models considering shorter forecast horizons reaching a stationary and often implausible state quickly. Recent studies show that implicit temporal representations in the frequency domain are also effective in making predictions for a predetermined horizon. Our focus lies on learning spatio-temporal representations autoregressively and hence generation of plausible future developments over both short and long term. The proposed model learns high dimensional embeddings for skeletal joints and how to compose a temporally coherent pose via a decoupled temporal and spatial self-attention mechanism. Our dual attention concept allows the model to access current and past information directly and to capture both the structural and the temporal dependencies explicitly. We show empirically that this effectively learns the underlying motion dynamics and reduces error accumulation over time observed in auto-regressive models. Our model is able to make accurate short-term predictions and generate plausible motion sequences over long horizons. We make our code publicly available at https://github.com/eth-ait/motion-transformer.

In natural conversation and interaction, our hands often overlap or are in contact with each other. Due to the homogeneous appearance of hands, this makes estimating the 3D pose of interacting hands from images difficult. In this paper we demonstrate that self-similarity, and the resulting ambiguities in assigning pixel observations to the respective hands and their parts, is a major cause of the final 3D pose error. Motivated by this insight, we propose DIGIT, a novel method for estimating the 3D poses of two interacting hands from a single monocular image. The method consists of two interwoven branches that process the input imagery into a per-pixel semantic part segmentation mask and a visual feature volume. In contrast to prior work, we do not decouple the segmentation from the pose estimation stage, but rather leverage the per-pixel probabilities directly in the downstream pose estimation task. To do so, the part probabilities are merged with the visual features and processed via fully-convolutional layers. We experimentally show that the proposed approach achieves new state-of-the-art performance on the InterHand2.6M [33] dataset. We provide detailed ablation studies to demonstrate the efficacy of our method and to provide insights into how the modelling of pixel ownership affects 3D hand pose estimation.

Capturing the dynamically deforming 3D shape of clothed human is essential for numerous applications, including VR/AR, autonomous driving, and human-computer interaction. Existing methods either require a highly specialized capturing setup, such as expensive multi-view imaging systems, or they lack robustness to challenging body poses. In this work, we propose a method capable of capturing the dynamic 3D human shape from a monocular video featuring challenging body poses, without any additional input. We first build a 3D template human model of the subject based on a learned regression model. We then track this template model’s deformation under challenging body articulations based on 2D image observations. Our method outperforms state-of-the-art methods on an in-the-wild human video dataset 3DPW. Moreover, we demonstrate its efficacy in robustness and generalizability on videos from iPER datasets.

We present Hand ArticuLated Occupancy (HALO), a novel representation of articulated hands that bridges the advantages of 3D keypoints and neural implicit surfaces and can be used in end-to-end trainable architectures. Unlike existing statistical parametric hand models (e.g. MANO), HALO directly leverages the 3D joint skeleton as input and produces a neural occupancy volume representing the posed hand surface. The key benefits of HALO are (1) it is driven by 3D keypoints, which have benefits in terms of accuracy and are easier to learn for neural networks than the latent hand-model parameters; (2) it provides a differentiable volumetric occupancy representation of the posed hand, (3) it can be trained end-to-end, allowing the formulation of losses on the hand surface that benefit the learning of 3D keypoints. We demonstrate the applicability of HALO to the task of conditional generation of hands that grasp 3D objects. The differentiable nature of HALO is shown to improve the quality of the synthesized hands both in terms of physical plausibility and user preference.

To create aesthetically pleasing aerial footage, the correct framing of camera targets is crucial. However, current quadrotor camera tools do not consider the 3D extent of actual camera targets in their optimization schemes and simply interpolate between keyframes when generating a trajectory. This can yield videos with aesthetically unpleasing target framing. In this paper, we propose a target framing algorithm that optimizes the quadrotor camera pose such that targets are positioned at desirable screen locations according to videographic compositional rules and entirely visible throughout a shot. Camera targets are identified using a semi-automatic pipeline which leverages a deep-learning-based visual saliency model. A large-scale perceptual study (N ≈ 500) shows that our method enables users to produce shots with a target framing that is closer to what they intended to create and more or as aesthetically pleasing than with the previous state of the art.

Receiving a hug is one of the best ways to feel socially supported, and the lack of social touch can have severe negative effects on an individual's well-being. Based on previous research both within and outside of HRI, we propose six tenets ("commandments") of natural and enjoyable robotic hugging: a hugging robot should be soft, be warm, be human sized, visually perceive its user, adjust its embrace to the user's size and position, and reliably release when the user wants to end the hug. Prior work validated the first two tenets, and the final four are new. We followed all six tenets to create a new robotic platform, HuggieBot 2.0, that has a soft, warm, inflated body (HuggieChest) and uses visual and haptic sensing to deliver closed-loop hugging. We first verified the outward appeal of this platform in comparison to the previous PR2-based HuggieBot 1.0 via an online video-watching study involving 117 users. We then conducted an in-person experiment in which 32 users each exchanged eight hugs with HuggieBot 2.0, experiencing all combinations of visual hug initiation, haptic sizing, and haptic releasing. The results show that adding haptic reactivity definitively improves user perception a hugging robot, largely verifying our four new tenets and illuminating several interesting opportunities for further improvement.

Deep generative models can synthesize photorealistic images of human faces with novel identities. However, a key challenge to the wide applicability of such techniques is to provide independent control over semantically meaningful parameters: appearance, head pose, face shape, and facial expressions. In this paper, we propose VariTex - to the best of our knowledge the first method that learns a variational latent feature space of neural face textures, which allows sampling of novel identities. We combine this generative model with a parametric face model and gain explicit control over head pose and facial expressions. To generate complete images of human heads, we propose an additive decoder that adds plausible details such as hair. A novel training scheme enforces a pose-independent latent space and in consequence, allows learning a one-to-many mapping between latent codes and pose-conditioned exterior regions. The resulting method can generate geometrically consistent images of novel identities under fine-grained control over head pose, face shape, and facial expressions. This facilitates a broad range of downstream tasks, like sampling novel identities, changing the head pose, expression transfer, and more.

Neural implicit surface representations have emerged as a promising paradigm to capture 3D shapes in a continuous and resolution-independent manner. However, adapting them to articulated shapes is non-trivial. Existing approaches learn a backward warp field that maps deformed to canonical points. However, this is problematic since the backward warp field is pose dependent and thus requires large amounts of data to learn. To address this, we introduce SNARF, which combines the advantages of linear blend skinning (LBS) for polygonal meshes with those of neural implicit surfaces by learning a forward deformation field without direct supervision. This deformation field is defined in canonical, pose-independent, space, enabling generalization to unseen poses. Learning the deformation field from posed meshes alone is challenging since the correspondences of deformed points are defined implicitly and may not be unique under changes of topology. We propose a forward skinning model that finds all canonical correspondences of any deformed point using iterative root finding. We derive analytical gradients via implicit differentiation, enabling end-to-end training from 3D meshes with bone transformations. Compared to state-of-the-art neural implicit representations, our approach generalizes better to unseen poses while preserving accuracy. We demonstrate our method in challenging scenarios on (clothed) 3D humans in diverse and unseen poses.

In this paper we contribute a simple yet effective approach for estimating 3D poses of multiple people from multi-view images. Our proposed coarse-to-fine pipeline first aggregates noisy 2D observations from multiple camera views into 3D space and then associates them into individual instances based on a confidence-aware majority voting technique. The final pose estimates are attained from a novel optimization scheme which links high-confidence multi-view 2D observations and 3D joint candidates. More-over, a statistical parametric body model such as SMPL is leveraged as a regularizing prior for these 3D joint candidates. Specifically, both 3D poses and SMPL parameters are optimized jointly in an alternating fashion. Here the parametric models help in correcting implausible 3D pose estimates and filling in missing joint detections while updated 3D poses in turn guide obtaining better SMPL estimations. By linking 2D and 3D observations, our method is both accurate and generalizes to different data sources because it better decouples the final 3D pose from the interperson constellation and is more robust to noisy 2D detections. We systematically evaluate our method on public datasets and achieve state-of-the-art performance. The code and video will be available on the project page: https://ait.ethz.ch/projects/2021/multi-human-pose/.

Fully immersive experiences in AR/VR depend on re-constructing the full body pose of the user without restricting their motion. In this paper we study the use of body-worn electromagnetic (EM) field-based sensing for the task of 3D human pose reconstruction. To this end, we present a method to estimate SMPL parameters from 6-12 EM sensors. We leverage a customized wearable system consisting of wireless EM sensors measuring time-synchronized 6D poses at 120 Hz. To provide accurate poses even with little user instrumentation, we adopt a recently proposed hybrid framework, learned gradient descent (LGD), to iteratively estimate SMPL pose and shape from our input measurements. This allows us to harness powerful pose priors to cope with the idiosyncrasies of the input data and achieve accurate pose estimates. The proposed method uses AMASS to synthesize virtual EM-sensor data and we show that it generalizes well to a newly captured real dataset consisting of a total of 36 minutes of motion from 5 subjects. We achieve reconstruction errors as low as 31.8 mm and 13.3 degrees, outperforming both pure learning- and pure optimization-based methods. Code and data is available under https://ait.ethz.ch/projects/2021/em-pose.

Despite significant progress, we show that state of the art 3D human pose and shape estimation methods remain sensitive to partial occlusion and can produce dramatically wrong predictions although much of the body is observable. To address this, we introduce a soft attention mechanism, called the Part Attention REgressor (PARE), that learns to predict body-part-guided attention masks. We observe that state-of-the-art methods rely on global feature representations, making them sensitive to even small occlusions. In contrast, PARE’s part-guided attention mechanism overcomes these issues by exploiting information about the visibility of individual body parts while leveraging information from neighboring body-parts to predict occluded parts. We show qualitatively that PARE learns sensible attention masks, and quantitative evaluation confirms that PARE achieves more accurate and robust reconstruction results than existing approaches on both occlusion-specific and standard benchmarks. The code and data are available for research purposes at https://pare.is.tue.mpg.de/

Due to the lack of camera parameter information for in-the-wild images, existing 3D human pose and shape (HPS) estimation methods make several simplifying assumptions: weak-perspective projection, large constant focal length, and zero camera rotation. These assumptions often do not hold and we show, quantitatively and qualitatively, that they cause errors in the reconstructed 3D shape and pose. To address this, we introduce SPEC, the first in-the-wild 3D HPS method that estimates the perspective camera from a single image and employs this to reconstruct 3D human bodies more accurately. First, we train a neural network to estimate the field of view, camera pitch, and roll given an input image. We employ novel losses that improve the calibration accuracy over previous work. We then train a novel network that concatenates the camera calibration to the image features and uses these together to regress 3D body shape and pose. SPEC is more accurate than the prior art on the standard benchmark (3DPW) as well as two new datasets with more challenging camera views and varying focal lengths. Specifically, we create a new photorealistic synthetic dataset (SPEC-SYN) with ground truth 3D bodies and a novel in-the-wild dataset (SPEC-MTP) with calibration and high-quality reference bodies. Code and datasets are available for research purposes at https://spec.is.tue.mpg.de/.

Encouraged by the success of contrastive learning on image classification tasks, we propose a new self-supervised method for the structured regression task of 3D hand pose estimation. Contrastive learning makes use of unlabeled data for the purpose of representation learning via a loss formulation that encourages the learned feature representations to be invariant under any image transformation. For 3D hand pose estimation, it too is desirable to have invariance to appearance transformation such as color jitter. However, the task requires equivariance under affine transformations, such as rotation and translation. To address this issue, we propose an equivariant contrastive objective and demonstrate its effectiveness in the context of 3D hand pose estimation. We experimentally investigate the impact of invariant and equivariant contrastive objectives and show that learning equivariant features leads to better representations for the task of 3D hand pose estimation. Furthermore, we show that standard ResNets with sufficient depth, trained on additional unlabeled data, attain improvements of up to 14.5% in PA-EPE on FreiHAND and thus achieves state-of-the-art performance without any task specific, specialized architectures. Code and models are available at https://ait.ethz.ch/projects/2021/PeCLR/

In this paper we propose a convolutional autoencoder to address the problem of motion infilling for 3D human motion data. Given a start and end sequence, motion infilling aims to complete the missing gap in between, such that the filled in poses plausibly forecast the start sequence and naturally transition into the end sequence. To this end, we propose a single, end-to-end trainable convolutional autoencoder. We show that a single model can be used to create natural transitions between different types of activities. Furthermore, our method is not only able to fill in entire missing frames, but it can also be used to complete gaps where partial poses are available (e.g. from end effectors), or to clean up other forms of noise (e.g. Gaussian). Also, the model can fill in an arbitrary number of gaps that potentially vary in length. In addition, no further post-processing on the model's outputs is necessary such as smoothing or closing discontinuities at the end of the gap. At the heart of our approach lies the idea to cast motion infilling as an inpainting problem and to train a convolutional de-noising autoencoder on image-like representations of motion sequences. At training time, blocks of columns are removed from such images and we ask the model to fill in the gaps. We demonstrate the versatility of the approach via a number of complex motion sequences and report on thorough evaluations performed to better understand the capabilities and limitations of the proposed approach.

The task of estimating the 6D pose of an object from RGB images can be broken down into two main steps: an initial pose estimation step, followed by a refinement procedure to correctly register the object and its observation. In this paper, we propose a new method for 6D pose estimation refinement from RGB images. To achieve high accuracy of the final estimate, the observation and a rendered model need to be aligned. Our main insight is that after the initial pose estimate, it is important to pay attention to distinct spatial features of the object in order to improve the estimation accuracy during alignment. Furthermore, parts of the object that are occluded in the image should be given less weight during the alignment process. Most state-of-the-art refinement approaches do not allow for this fine-grained reasoning and can not fully leverage the structure of the problem. In contrast, we propose a novel neural network architecture built around a spatial attention mechanism that identifies and leverages information about spatial details during pose refinement. We experimentally show that this approach learns to attend to salient spatial features and learns to ignore occluded parts of the object, leading to better pose estimation across datasets. We conduct experiments on standard benchmark datasets for 6D pose estimation (LineMOD and Occlusion LineMOD) and outperform previous state-of-the-art methods.

This paper proposes an approach to detect information relevance during decision-making from eye movements in order to enable user interface adaptation. This is a challenging task because gaze behavior varies greatly across individual users and tasks and ground-truth data is difficult to obtain. Thus, prior work has mostly focused on simpler target-search tasks or on establishing general interest, where gaze behavior is less complex. From the literature, we identify six metrics that capture different aspects of the gaze behavior during decision-making and combine them in a voting scheme. We empirically show, that this accounts for the large variations in gaze behavior and out-performs standalone metrics. Importantly, it offers an intuitive way to control the amount of detected information, which is crucial for different UI adaptation schemes to succeed. We show the applicability of our approach by developing a room-search application that changes the visual saliency of content detected as relevant. In an empirical study, we show that it detects up to 97% of relevant elements with respect to user self-reporting, which allows us to meaningfully adapt the interface, as confirmed by participants. Our approach is fast, does not need any explicit user input and can be applied independent of task and user.

We present a generative model for stroke-based drawing tasks which is able to model complex free-form structures. While previous approaches rely on sequence-based models for drawings of basic objects or handwritten text, we propose a model that treats drawings as a collection of strokes that can be composed into complex structures such as diagrams (e.g., flow-charts). At the core of the approach lies a novel auto-encoder that projects variable-length strokes into a latent space of fixed dimension. This representation space allows a relational model, operating in latent space, to better capture the relationship between strokes and to predict subsequent strokes. We demonstrate qualitatively and quantitatively that our proposed approach is able to model the appearance of individual strokes, as well as the compositional structure of larger diagram drawings. Our approach is suitable for interactive use cases such as auto-completing diagrams. We make code and models publicly available at https://eth-ait.github.io/cose.

Many computer vision tasks rely on labeled data. Rapid progress in generative modeling has led to the ability to synthesize photorealistic images. However, controlling specific aspects of the generation process such that the data can be used for supervision of downstream tasks remains challenging. In this paper we propose a novel generative model for images of faces, that is capable of producing high-quality images under fine-grained control over eye gaze and head orientation angles. This requires the disentangling of many appearance related factors including gaze and head orientation but also lighting, hue etc. We propose a novel architecture which learns to discover, disentangle and encode these extraneous variations in a self-learned manner. We further show that explicitly disentangling task-irrelevant factors results in more accurate modelling of gaze and head orientation. A novel evaluation scheme shows that our method improves upon the state-of-the-art in redirection accuracy and disentanglement between gaze direction and head orientation changes. Furthermore, we show that in the presence of limited amounts of real-world training data, our method allows for improvements in the downstream task of semi-supervised cross-dataset gaze estimation. Please check our project page at: https://ait.ethz.ch/projects/2020/STED-gaze/

We present Omni, a self-contained 3D haptic feedback system that is capable of sensing and actuating an untethered, passive tool containing only a small embedded permanent magnet. Omni enriches AR, VR and desktop applications by providing an active haptic experience using a simple apparatus centered around an electromagnetic base. The spatial haptic capabilities of Omni are enabled by a novel gradient-based method to reconstruct the 3D position of the permanent magnet in midair using the measurements from eight off-the-shelf hall sensors that are integrated into the base. Omni's 3 DoF spherical electromagnet simultaneously exerts dynamic and precise radial and tangential forces in a volumetric space around the device. Since our system is fully integrated, contains no moving parts and requires no external tracking, it is easy and affordable to fabricate. We describe Omni's hardware implementation, our 3D reconstruction algorithm, and evaluate the tracking and actuation performance in depth. Finally, we demonstrate its capabilities via a set of interactive usage scenarios.

We introduce an optimal control method for electromagnetic haptic guidance systems. Our real-time approach assists users in pen-based tasks such as drawing, sketching or designing. The key to our control method is that it guides users, yet does not take away agency. Existing approaches force the stylus to a continuously advancing setpoint on a target trajectory, leading to undesirable behavior such as loss of haptic guidance or unintended snapping. Our control approach, in contrast, gently pulls users towards the target trajectory, allowing them to always easily override the system to adapt their input spontaneously and draw at their own speed. To achieve this flexible guidance, our optimization iteratively predicts the motion of an input device such as a pen, and adjusts the position and strength of an underlying dynamic electromagnetic actuator accordingly. To enable real-time computation, we additionally introduce a novel and fast approximate model of an electromagnet. We demonstrate the applicability of our approach by implementing it on a prototypical hardware platform based on an electromagnet moving on a bi-axial linear stage, as well as a set of applications. Experimental results show that our approach is more accurate and preferred by users compared to open-loop and time-dependent closed-loop approaches.

Abstract 3D gaze tracking from a single RGB camera is very challenging due to the lack of information in determining the accurate gaze target from a monocular RGB sequence. The eyes tend to occupy only a small portion of the video, and even small errors in estimated eye orientations can lead to very large errors in the triangulated gaze target. We overcome these difficulties with a novel lightweight eyeball calibration scheme that determines the user‐specific visual axis, eyeball size and position in the head. Unlike the previous calibration techniques, we do not need the ground truth positions of the gaze points. In the online stage, gaze is tracked by a new gaze fitting algorithm, and refined by a 3D gaze regression method to correct for bias errors. Our regression is pre‐trained on several individuals and works well for novel users. After the lightweight one‐time user calibration, our method operates in real time. Experiments show that our technique achieves state‐of‐the‐art accuracy in gaze angle estimation, and we demonstrate applications of 3D gaze target tracking and gaze retargeting to an animated 3D character.

We propose an approach to capture subjective first-person view (FPV) videos by drones for automated cinematography. FPV shots are intentionally not smooth to increase the level of immersion for the audience, and are usually captured by a walking camera operator holding traditional camera equipment. Our goal is to automatically control a drone in such a way that it imitates the motion dynamics of a walking camera operator, and, in turn, capture FPV videos. For this, given a user-defined camera path, orientation, and velocity, we first present a method to automatically generate the operator’s motion pattern and the associated motion of the camera, considering the damping mechanism of the camera equipment. Second, we propose a general computational approach that generates the drone commands to imitate the desired motion pattern. We express this task as a constrained optimization problem, where we aim to fulfill high-level user-defined goals, while imitating the dynamics of the walking camera operator and taking the drone’s physical constraints into account. Our approach is fully automatic, runs in real time, and is interactive, which provides artistic freedom in designing shots. It does not require a motion capture system, and works both indoors and outdoors. The validity of our approach has been confirmed via quantitative and qualitative evaluations.

Deep neural networks provide powerful tools for pattern recognition, while classical graph algorithms are widely used to solve combinatorial problems. In computer vision, many tasks combine elements of both pattern recognition and graph reasoning. In this paper, we study how to connect deep networks with graph decomposition into an end-to-end trainable framework. More specifically, the minimum cost multicut problem is first converted to an unconstrained binary cubic formulation where cycle consistency constraints are incorporated into the objective function. The new optimization problem can be viewed as a Conditional Random Field (CRF) in which the random variables are associated with the binary edge labels. Cycle constraints are introduced into the CRF as high-order potentials. A standard Convolutional Neural Network (CNN) provides the front-end features for the fully differentiable CRF. The parameters of both parts are optimized in an end-to-end manner. The efficacy of the proposed learning algorithm is demonstrated via experiments on clustering MNIST images and on the challenging task of real-world multi-people pose estimation.

We propose a method to produce a continuous stream of novel views under fine-grained (e.g., 1 degree step-size) camera control at interactive rates. A novel learning pipeline determines the output pixels directly from the source color. Injecting geometric transformations, including perspective projection, 3D rotation and translation into the network forces implicit reasoning about the underlying geometry. The latent 3D geometry representation is compact and meaningful under 3D transformation, being able to produce geometrically accurate views for both single objects and natural scenes. Our experiments show that both proposed components, the transforming encoder-decoder and depth-guided appearance mapping, lead to significantly improved generalization beyond the training views and in consequence to more accurate view synthesis under continuous 6-DoF camera control. Finally, we show that our method outperforms state-of-the-art baseline methods on public datasets.

Human motion prediction is a challenging and important task in many computer vision application domains. Existing work only implicitly models the spatial structure of the human skeleton. In this paper, we propose a novel approach that decomposes the prediction into individual joints by means of a structured prediction layer that explicitly models the joint dependencies. This is implemented via a hierarchy of small-sized neural networks connected analogously to the kinematic chains in the human body as well as a joint-wise decomposition in the loss function. The proposed layer is agnostic to the underlying network and can be used with existing architectures for motion modelling. Prior work typically leverages the H3.6M dataset. We show that some state-of-the-art techniques do not perform well when trained and tested on AMASS, a recently released dataset 14 times the size of H3.6M. Our experiments indicate that the proposed layer increases the performance of motion forecasting irrespective of the base network, joint-angle representation, and prediction horizon. We furthermore show that the layer also improves motion predictions qualitatively. We make code and models publicly available at https://ait.ethz.ch/projects/2019/spl.

Gaze redirection is the task of changing the gaze to a desired direction for a given monocular eye patch image. Many applications such as videoconferencing, films, games, and generation of training data for gaze estimation require redirecting the gaze, without distorting the appearance of the area surrounding the eye and while producing photo-realistic images. Existing methods lack the ability to generate perceptually plausible images. In this work, we present a novel method to alleviate this problem by leveraging generative adversarial training to synthesize an eye image conditioned on a target gaze direction. Our method ensures perceptual similarity and consistency of synthesized images to the real images. Furthermore, a gaze estimation loss is used to control the gaze direction accurately. To attain high-quality images, we incorporate perceptual and cycle consistency losses into our architecture. In extensive evaluations we show that the proposed method outperforms state-of-the-art approaches in terms of both image quality and redirection precision. Finally, we show that generated images can bring significant improvement for the gaze estimation task if used to augment real training data.

Inter-personal anatomical differences limit the accuracy of person-independent gaze estimation networks. Yet there is a need to lower gaze errors further to enable applications requiring higher quality. Further gains can be achieved by personalizing gaze networks, ideally with few calibration samples. However, over-parameterized neural networks are not amenable to learning from few examples as they can quickly over-fit. We embrace these challenges and propose a novel framework for Few-shot Adaptive GaZE Estimation (Faze) for learning person-specific gaze networks with very few (≤ 9) calibration samples. Faze learns a rotation-aware latent representation of gaze via a disentangling encoder-decoder architecture along with a highly adaptable gaze estimator trained using meta-learning. It is capable of adapting to any new person to yield significant performance gains with as few as 3 samples, yielding state-of-the-art performance of 3.18-deg on GazeCapture, a 19% improvement over prior art. We open-source our code at https://github.com/NVlabs/few_shot_gaze.

In this paper, we propose a method for training control policies for human-robot interactions such as handshakes or hand claps via Deep Reinforcement Learning. The policy controls a humanoid Shadow Dexterous Hand, attached to a robot arm. We propose a parameterizable multi-objective reward function that allows learning of a variety of interactions without changing the reward structure. The parameters of the reward function are estimated directly from motion capture data of human-human interactions in order to produce policies that are perceived as being natural and human-like by observers. We evaluate our method on three significantly different hand interactions: handshake, hand clap and finger touch. We provide detailed analysis of the proposed reward function and the resulting policies and conduct a large-scale user study, indicating that our policy produces natural looking motions.

Among the currently available grasp-type selection techniques for hand prostheses, there is a distinct lack of intuitive, robust, low-latency solutions. In this paper we investigate the use of a portable, forearm-mounted, video-based technique for the prediction of hand-grasp preshaping for arbitrary objects. The purpose of this system is to automatically select the grasp-type for the user of the prosthesis, potentially increasing ease-of-use and functionality. This system can be used to supplement and improve existing control strategies, such as surface electromyography (sEMG) pattern recognition, for prosthetic and orthotic devices. We designed and created a suitable dataset consisting of RGB-D video data for 2212 grasp examples split evenly across 7 classes; 6 grasps commonly used in activities of daily living, and an additional no-grasp category. We processed and analyzed the dataset using several state-of-the-art deep learning architectures. Our selected model shows promising results for realistic, intuitive, real-world use, reaching per-frame accuracies on video sequences of up to 95.90% on the validation set. Such a system could be integrated into the palm of a hand prosthesis, allowing an automatic prediction of the grasp-type without requiring any special movements or aiming by the user.

We present AudioTouch, a minimally invasive approach for sensing micro-gestures using active bio-acoustic sensing. It only requires attaching two piezo-electric elements, acting as a surface mounted speaker and microphone, on the back of the hand. It does not require any instrumentation on the palm or fingers; therefore, it does not encumber interactions with physical objects. The signal is rich enough to detect small differences in micro-gestures with standard machine-learning classifiers. This approach also allows for the discrimination of different levels of touch-force, further expanding the interaction vocabulary. We conducted four experiments to evaluate the performances of AudioTouch: a user study for measuring the gesture recognition accuracy, a follow-up study investigating the ability to discriminate different levels of touch-force, an experiment assessing the cross-session robustness, and, a systematic evaluation assessing the effect of sensor placement on the back of the hand.

An ideal Mixed Reality (MR) system would only present virtual information (e.g., a label) when it is useful to the person. However, deciding when a label is useful is challenging: it depends on a variety of factors, including the current task, previous knowledge, context, etc. In this paper, we propose a Reinforcement Learning (RL) method to learn when to show or hide an object's label given eye movement data. We demonstrate the capabilities of this approach by showing that an intelligent agent can learn cooperative policies that better support users in a visual search task than manually designed heuristics. Furthermore, we show the applicability of our approach to more realistic environments and use cases (e.g., grocery shopping). By posing MR object labeling as a model-free RL problem, we can learn policies implicitly by observing users' behavior without requiring a visual search model or data annotation.

We present an optimization-based approach for Mixed Reality (MR) systems to automatically control when and where applications are shown, and how much information they display. Currently, content creators design applications, and users then manually adjust which applications are visible and how much information they show. This choice has to be adjusted every time users switch context, i.e., whenever they switch their task or environment. Since context switches happen many times a day, we believe that MR interfaces require automation to alleviate this problem. We propose a real-time approach to automate this process based on users' current cognitive load and knowledge about their task and environment. Our system adapts which applications are displayed, how much information they show, and where they are placed. We formulate this problem as a mix of rule-based decision making and combinatorial optimization which can be solved efficiently in real-time. We present a set of proof-of-concept applications showing that our approach is applicable in a wide range of scenarios. Finally, we show in a dual-task evaluation that our approach decreased secondary tasks interactions by 36%.

We introduce TacTiles, light (1.8g), low-power (130 mW), and small form-factor (1 cm3) electromagnetic actuators that can form a flexible haptic array to provide localized tactile feedback. Our novel hardware design uses a custom 8-layer PCB, dampening materials, and asymmetric latching, enabling two distinct modes of actuation: contact and pulse mode. We leverage these modes in Virtual Reality (VR) to render continuous contact with objects and the exploration of object surfaces and volumes with spatial haptic patterns. Results from a series of experiments show that users are able to localize feedback, discriminate between modes with high accuracy, and differentiate objects from haptic surfaces and volumes even without looking at them.

We propose a hardware and software pipeline to fabricate flexible wearable sensors and use them to capture deformations without line-of-sight. Our first contribution is a low-cost fabrication pipeline to embed multiple aligned conductive layers with complex geometries into silicone compounds. Overlapping conductive areas from separate layers form local capacitors that measure dense area changes. Contrary to existing fabrication methods, the proposed technique only requires hardware that is readily available in modern fablabs. While area measurements alone are not enough to reconstruct the full 3D deformation of a surface, they become sufficient when paired with a data-driven prior. A novel semi-automatic tracking algorithm, based on an elastic surface geometry deformation, allows us to capture ground-truth data with an optical mocap system, even under heavy occlusions or partially unobservable markers. The resulting dataset is used to train a regressor based on deep neural networks, directly mapping the area readings to global positions of surface vertices. We demonstrate the flexibility and accuracy of the proposed hardware and software in a series of controlled experiments and design a prototype of wearable wrist, elbow, and biceps sensors, which do not require line-of-sight and can be worn below regular clothing.

We propose a stretch-sensing soft glove to interactively capture hand poses with high accuracy and without requiring an external optical setup. We demonstrate how our device can be fabricated and calibrated at low cost, using simple tools available in most fabrication labs. To reconstruct the pose from the capacitive sensors embedded in the glove, we propose a deep network architecture that exploits the spatial layout of the sensor itself. The network is trained only once, using an inexpensive off-the-shelf hand pose reconstruction system to gather the training data. The per-user calibration is then performed on-the-fly using only the glove. The glove's capabilities are demonstrated in a series of ablative experiments, exploring different models and calibration methods. Comparing against commercial data gloves, we achieve a 35% improvement in reconstruction accuracy.

We introduce a new method that efficiently computes a set of viewpoints and trajectories for high-quality 3D reconstructions in outdoor environments. Our goal is to automatically explore an unknown area and obtain a complete 3D scan of a region of interest (e.g., a large building). Images from a commodity RGB camera, mounted on an autonomously navigated quadcopter, are fed into a multi-view stereo reconstruction pipeline that produces high-quality results but is computationally expensive. In this setting, the scanning result is constrained by the restricted flight time of quadcopters. To this end, we introduce a novel optimization strategy that respects these constraints by maximizing the information gain from sparsely sampled viewpoints while limiting the total travel distance of the quadcopter. At the core of our method lies a hierarchical volumetric representation that allows the algorithm to distinguish between unknown, free, and occupied space. Furthermore, our information gain-based formulation leverages this representation to handle occlusions in an efficient manner. In addition to the surface geometry, we utilize free-space information to avoid obstacles and determine collision-free flight paths. Our tool can be used to specify the region of interest and to plan trajectories. We demonstrate our method by obtaining a number of compelling 3D reconstructions, and we provide a thorough quantitative evaluation showing improvement over previous state-of-the-art and regular patterns.

Digital ink promises to combine the flexibility and aesthetics of handwriting and the ability to process, search and edit digital text. Character recognition converts handwritten text into a digital representation, albeit at the cost of losing personalized appearance due to the technical difficulties of separating the interwoven components of content and style. In this paper, we propose a novel generative neural network architecture that is capable of disentangling style from content and thus making digital ink editable. Our model can synthesize arbitrary text, while giving users control over the visual appearance (style). For example, allowing for style transfer without changing the content, editing of digital ink at the word level and other application scenarios such as spell-checking and correction of handwritten text. We furthermore contribute a new dataset of handwritten text with fine-grained annotations at the character level and report results from an initial user evaluation.

Developing cross-device multi-user interfaces (UIs) is a challenging problem. There are numerous ways in which content and interactivity can be distributed. However, good solutions must consider multiple users, their roles, their preferences and access rights, as well as device capabilities. Manual and rule-based solutions are tedious to create and do not scale to larger problems nor do they adapt to dynamic changes, such as users leaving or joining an activity. In this paper, we cast the problem of UI distribution as an assignment problem and propose to solve it using combinatorial optimization. We present a mixed integer programming formulation which allows real-time applications in dynamically changing collaborative settings. It optimizes the allocation of UI elements based on device capabilities, user roles, preferences, and access rights. We present a proof-of-concept designer-in-the-loop tool, allowing for quick solution exploration. Finally, we compare our approach to traditional paper prototyping in a lab study.

The human hand moves in complex and high-dimensional ways, making estimation of 3D hand pose configurations from images alone a challenging task. In this work we propose a method to learn a statistical hand model represented by a cross-modal trained latent space via a generative deep neural network. We derive an objective function from the variational lower bound of the VAE framework and jointly optimize the resulting cross-modal KL-divergence and the posterior reconstruction objective, naturally admitting a training regime that leads to a coherent latent space across multiple modalities such as RGB images, 2D keypoint detections or 3D hand configurations. Additionally, it grants a straightforward way of using semi-supervision. This latent space can be directly used to estimate 3D hand poses from RGB images, outperforming the state-of-the art in different settings. Furthermore, we show that our proposed method can be used without changes on depth images and performs comparably to specialized methods. Finally, the model is fully generative and can synthesize consistent pairs of hand configurations across modalities. We evaluate our method on both RGB and depth datasets and analyze the latent space qualitatively.

Conventional feature-based and model-based gaze estimation methods have proven to perform well in settings with controlled illumination and specialized cameras. In unconstrained real-world settings, however, such methods are surpassed by recent appearance-based methods due to difficulties in modeling factors such as illumination changes and other visual artifacts. We present a novel learning-based method for eye region landmark localization that enables conventional methods to be competitive to latest appearance-based methods. Despite having been trained exclusively on synthetic data, our method exceeds the state of the art for iris localization and eye shape registration on real-world imagery. We then use the detected landmarks as input to iterative model-fitting and lightweight learning-based gaze estimation methods. Our approach outperforms existing model-fitting and appearance-based methods in the context of person-independent and personalized gaze estimation.

In this video we present soft self-sensing capacitive arrays and demonstrate their use in capturing dense surface deformations without requiring line of sight. The capacitive arrays are made of two electrode patterns embedded into a single silicone compound. The overlaps of the electrode strip patterns form local capacitors. As the sensor is stretched the local capacitance measurements change. We introduce a fabrication technique that allows to produce such sensors while only requiring hardware readily available in modern fablabs. The resulting sensors are able to densely capture area changes as they deform. Since they do not directly measure bend, a prior is required to fully reconstruct the underlying motion. We propose a deep neural network regressing the sensor geometry from the local area measurements. A motion capture system is used for training data acquisition. At runtime vertex positions are predicted and used as positional constraints to deform a mesh using a state-of-the-art elastic surface energy. The flexibility and accuracy of the introduced sensors is demonstrated in a series of controlled experiments and by fabricating a prototype sensor and applying it to capture deforming skeletal and non-skeletal objects.

We introduce DextrES, a flexible and wearable haptic glove which integrates both kinesthetic and cutaneous feedback in a thin and light form factor (weight is less than 8g). Our approach is based on an electrostatic clutch generating up to 20 N of holding force on each finger by modulating the electrostatic attraction between flexible elastic metal strips to generate an electrically-controlled friction force. We harness the resulting braking force to rapidly render on-demand kinesthetic feedback. The electrostatic brake is mounted onto the the index finger and thumb via modular 3D printed articulated guides which allow the metal strips to glide smoothly. Cutaneous feedback is provided via piezo actuators at the fingertips. We demonstrate that our approach can provide rich haptic feedback under dexterous articulation of the user's hands and provides effective haptic feedback across a variety of different grasps. A controlled experiment indicates that DextrES improves the grasping precision for different types of virtual objects. Finally, we report on results of a psycho-physical study which identifies discrimination thresholds for different levels of holding force.

In this paper we first contribute a large scale online study ( N ≈ 400) to better understand aesthetic perception of aerial video. The results indicate that it is paramount to optimize smoothness of trajectories across all keyframes. However, for experts timing control remains an essential tool. Satisfying this dual goal is technically challenging because it requires giving up desirable properties in the optimization formulation. Second, informed by this study we propose a method that optimizes positional and temporal reference fit jointly. This allows to generate globally smooth trajectories, while retaining user control over reference timings. The formulation is posed as a variable, infinite horizon, contour-following algorithm. Finally, a comparative lab study indicates that our optimization scheme outperforms the state-of-the-art in terms of perceived usability and preference of resulting videos. For novices our method produces smoother and better looking results and also experts benefit from generated timings.

We demonstrate a novel deep neural network capable of reconstructing human full body pose in real-time from 6 Inertial Measurement Units (IMUs) worn on the user's body. In doing so, we address several difficult challenges. First, the problem is severely under-constrained as multiple pose parameters produce the same IMU orientations. Second, capturing IMU data in conjunction with ground-truth poses is expensive and difficult to do in many target application scenarios (e.g., outdoors). Third, modeling temporal dependencies through non-linear optimization has proven effective in prior work but makes real-time prediction infeasible. To address this important limitation, we learn the temporal pose priors using deep learning. To learn from sufficient data, we synthesize IMU data from motion capture datasets. A bi-directional RNN architecture leverages past and future information that is available at training time. At test time, we deploy the network in a sliding window fashion, retaining real time capabilities. To evaluate our method, we recorded DIP-IMU, a dataset consisting of 10 subjects wearing 17 IMUs for validation in 64 sequences with 330 000 time instants; this constitutes the largest IMU dataset publicly available. We quantitatively evaluate our approach on multiple datasets and show results from a real-time implementation. DIP-IMU and the code are available for research purposes. 1

We propose a real-time method for the infrastructure-free estimation of articulated human motion. The approach leverages a swarm of camera-equipped flying robots and jointly optimizes the swarm's and skeletal states, which include the 3D joint positions and a set of bones. Our method allows to track the motion of human subjects, for example an athlete, over long time horizons and long distances, in challenging settings and at large scale, where fixed infrastructure approaches are not applicable. The proposed algorithm uses active infra-red markers, runs in real-time and accurately estimates robot and human pose parameters online without the need for accurately calibrated or stationary mounted cameras. Our method i) estimates a global coordinate frame for the MAV swarm, ii) jointly optimizes the human pose and relative camera positions, and iii) estimates the length of the human bones. The entire swarm is then controlled via a model predictive controller to maximize visibility of the subject from multiple viewpoints even under fast motion such as jumping or jogging. We demonstrate our method in a number of difficult scenarios including capture of long locomotion sequences at the scale of a triplex gym, in non-planar terrain, while climbing and in outdoor scenarios.

We propose a new architecture for the learning of predictive spatio-temporal motion models from data alone. Our approach, dubbed the Dropout Autoencoder LSTM (DAELSTM), is capable of synthesizing natural looking motion sequences over long-time horizons1 without catastrophic drift or motion degradation. The model consists of two components, a 3-layer recurrent neural network to model temporal aspects and a novel autoencoder that is trained to implicitly recover the spatial structure of the human skeleton via randomly removing information about joints during training. This Dropout Autoencoder (DAE) is then used to filter each predicted pose by a 3-layer LSTM network, reducing accumulation of correlated error and hence drift over time. Furthermore to alleviate insufficiency of commonly used quality metric, we propose a new evaluation protocol using action classifiers to assess the quality of synthetic motion sequences. The proposed protocol can be used to assess quality of generated sequences of arbitrary length. Finally, we evaluate our proposed method on two of the largest motion-capture datasets available and show that our model outperforms the state-of-the-art techniques on a variety of actions, including cyclic and acyclic motion, and that it can produce natural looking sequences over longer time horizons than previous methods.

Mass adoption and innovation in the field of the Internet of Things has transformed the environments we live in, from stale siloes of technologies into rich interactive playgrounds. Nevertheless, the vast majority of surface area in these spaces are being overlooked and under-utilized in today's research. Surface imaging provides the means to extend and include typically out-of-reach, disconnected objects into these playgrounds. However, existing surface imaging technologies are impractical to embed in everyday environments, restricting researchers from exploring the design and interaction opportunities they can afforded these spaces. In this paper, we propose IRIS, a modular surface imaging prototype capable of providing scalable, low-cost, high-resolution surface imaging. We describe a real-world case study where IRIS is used to identify and track fresh fruit produce being prepared -- a task that is typical infeasible with existing technologies. Through IRIS, we hope to enable the community to exploit these under-explored surface areas and enhance the rich, interactive, connected environments we inhabit.

Deep ConvNets have been shown to be effective for the task of human pose estimation from single images. However, several challenging issues arise in the video-based case such as self-occlusion, motion blur, and uncommon poses with few or no examples in the training data. Temporal information can provide additional cues about the location of body joints and help to alleviate these issues. In this paper, we propose a deep structured model to estimate a sequence of human poses in unconstrained videos. This model can be efficiently trained in an end-to-end manner and is capable of representing the appearance of body joints and their spatio-temporal relationships simultaneously. Domain knowledge about the human body is explicitly incorporated into the network providing effective priors to regularize the skeletal structure and to enforce temporal consistency. The proposed end-to-end architecture is evaluated on two widely used benchmarks for video-based pose estimation (Penn Action and JHMDB datasets). Our approach outperforms several state-of-the-art methods.

We present MagTics, a novel flexible and wearable haptic interface based on magnetically actuated bidirectional tactile pixels (taxels). MagTics' thin form factor and flexibility allows for rich haptic feedback in mobile settings. We propose a novel actuation mechanism based on bistable electromagnetic latching that combines high frame rate and holding force with low energy consumption and a soft and flexible form factor. We overcome limitations of traditional soft actuators by placing several hard actuation cells, driven by flexible printed electronics, in a soft 3D printed case. A novel EM-shielding prevents magnet-magnet interactions and allows for high actuator densities. A prototypical implementation comprising of 4 actuated pins on a 1.7 cm pitch, with 2 mm travel, and generating 160 mN to 200 mN of latching force is used to implement a number of compelling application scenarios including adding haptic and tactile display capabilities to wearable devices, to existing input devices and to provide localized haptic feedback in virtual reality. Finally, we report results of a psychophysical study, conducted to inform future developments and to identify possible application domains.

We propose a method for automated aerial videography in dynamic and cluttered environments. An online receding horizon optimization formulation facilitates the planning process for novices and experts alike. The algorithm takes high-level plans as input, which we dub virtual rails, alongside interactively defined aesthetic framing objectives and jointly solves for 3D quadcopter motion plans and associated velocities. The method generates control inputs subject to constraints of a non-linear quadrotor model and dynamic constraints imposed by actors moving in an a priori unknown way. The output plans are physically feasible, for the horizon length, and we apply the resulting control inputs directly at each time-step, without requiring a separate trajectory tracking algorithm. The online nature of the method enables incorporation of feedback into the planning and control loop, makes the algorithm robust to disturbances. Furthermore, we extend the method to include coordination between multiple drones to enable dynamic multi-view shots, typical for action sequences and live TV coverage. The algorithm runs in real-time on standard hardware and computes motion plans for several drones in the order of milliseconds. Finally, we evaluate the approach qualitatively with a number of challenging shots, involving multiple drones and actors and qualitatively characterize the computational performance experimentally.

Researchers often introduce visual tools to programming environments in order to facilitate program comprehension, reduce navigation times, and help developers answer difficult questions. Syntax highlighting is the main visual lens through which developers perceive their code, and yet its effects and the effects of richer code presentations on code comprehension have not been evaluated systematically. We present a rigorous user study comparing mainstream syntax highlighting to two visually-enhanced presentations of code. Our results show that: (1) richer code visualizations reduce the time necessary to answer questions about code features, and (2) contrary to the subjective perception of developers, richer code visualizations do not lead to visual overload. Based on our results we outline practical recommendations for tool designers.

We present a novel optimization-based algorithm for the design and fabrication of customized, deformable input devices, capable of continuously sensing their deformation. We propose to embed piezoresistive sensing elements into flexible 3D printed objects. These sensing elements are then utilized to recover rich and natural user interactions at runtime. Designing such objects is a challenging and hard problem if attempted manually for all but the simplest geometries and deformations. Our method simultaneously optimizes the internal routing of the sensing elements and computes a mapping from low-level sensor readings to user-specified outputs in order to minimize reconstruction error. We demonstrate the power and flexibility of the approach by designing and fabricating a set of flexible input devices. Our results indicate that the optimization-based design greatly outperforms manual routings in terms of reconstruction accuracy and thus interaction fidelity.

In this paper we propose a computational design tool that allows end-users to create advanced quadrotor trajectories with a variety of application scenarios in mind. Our algorithm allows novice users to create quadrotor based use-cases without requiring deep knowledge in either quadrotor control or the underlying constraints of the target domain. To achieve this goal we propose an optimization-based method that generates feasible trajectories which can be flown in the real world. Furthermore, the method incorporates high-level human objectives into the planning of flight trajectories. An easy to use 3D design tool allows for quick specification and editing of trajectories as well as for intuitive exploration of the resulting solution space. We demonstrate the utility of our approach in several real-world application scenarios, including aerial-videography, robotic light-painting and drone racing.

We present a tracking system based on ultra-wideband (UWB) radio tranceivers mounted on a robot and a target. In comparison to typical UWB localization systems with fixed UWB tranceivers in the environment we only require instrumentation of the target with a single UWB tranceiver. Our system works in GPS-denied environments and does not suffer from long-term drift and limited fields of view. This paper reports the localization algorithm and implementation details. Additionally, we demonstrate a quantitative evaluation of the accuracy (10cm average position error for a square with side-length of 4m) and application scenarios with a quadrotor flying in close proximity to a person and handling occlusion of the target. Finally, we release our implementation as open-source software.

We present a Visual Inertial Odometry system that enables the autonomous flight of Micro Aerial Vehicles in GPS denied and unstructured environments. The system relies on commercially available and affordable hardware both for sensing and computation. The algorithm runs in real time on an ARM based embedded micro-computer on-board an MAV. In experiments, we demonstrate the performance of the system both indoors and outdoors, in hand held an in-flight scenarios. The achieved accuracy of the experiments is competitive with other research which uses custom designed hardware and desktop-grade processors.

This paper proposes a novel machine learning architecture, specifically designed for radio-frequency based gesture recognition. We focus on high-frequency (60]GHz), short-range radar based sensing, in particular Google's Soli sensor. The signal has unique properties such as resolving motion at a very fine level and allowing for segmentation in range and velocity spaces rather than image space. This enables recognition of new types of inputs but poses significant difficulties for the design of input recognition algorithms. The proposed algorithm is capable of detecting a rich set of dynamic gestures and can resolve small motions of fingers in fine detail. Our technique is based on an end-to-end trained combination of deep convolutional and recurrent neural networks. The algorithm achieves high recognition rates (avg 87%) on a challenging set of 11 dynamic gestures and generalizes well across 10 users. The proposed model runs on commodity hardware at 140 Hz (CPU only).

We propose a novel approach to digital character animation, combining the benefits of tangible input devices and sophisticated rig animation algorithms. A symbiotic software and hardware approach facilitates the animation process for novice and expert users alike. We overcome limitations inherent to all previous tangible devices by allowing users to directly control complex rigs using only a small set (5-10) of physical controls. This avoids oversimplification of the pose space and excessively bulky device configurations. Our algorithm derives a small device configuration from complex character rigs, often containing hundreds of degrees of freedom, and a set of sparse sample poses. Importantly, only the most influential degrees of freedom are controlled directly, yet detailed motion is preserved based on a pose interpolation technique. We designed a modular collection of joints and splitters, which can be assembled to represent a wide variety of skeletons. Each joint piece combines a universal joint and two twisting elements, allowing to accurately sense its configuration. The mechanical design provides a smooth inverse kinematics-like user experience and is not prone to gimbal locking. We integrate our method with the professional 3D software Autodesk Maya ® and discuss a variety of results created with characters available online. Comparative user experiments show significant improvements over the closest state-of-the-art in terms of accuracy and time in a keyframe posing task.

Data structure algorithms are of fundamental importance in teaching and software development, yet are difficult to understand. We propose a new approach for understanding, debugging and developing heap manipulating data structures. The key technical idea of our work is to combine deep parametric abstraction techniques emerging from the area of static analysis with interactive abstraction manipulation. Our approach bridges program analysis with HCI and enables new capabilities not possible before: i) online automatic visualization of the data structure in a way which captures its essential operation, thus enabling powerful local reasoning, and ii) fine grained pen and touch gestures allowing for interactive control of the abstraction -- at any point the developer can pause the program, graphically interact with the data, and continue program execution. These features address some of the most pressing challenges in developing data structures. We implemented our approach in a Java-based system called FluiEdt and evaluated it with $27$ developers. The results indicate that FluiEdt is more effective in helping developers find data structure errors than existing state of the art IDEs (e.g. Eclipse) or pure visualization based approaches.

We present a machine learning technique to recognize gestures and estimate metric depth of hands for 3D interaction, relying only on monocular RGB video input. We aim to enable spatial interaction with small, body-worn devices where rich 3D input is desired but the usage of conventional depth sensors is prohibitive due to their power consumption and size. We propose a hybrid classification-regression approach to learn and predict a mapping of RGB colors to absolute, metric depth in real time. We also classify distinct hand gestures, allowing for a variety of 3D interactions. We demonstrate our technique with three mobile interaction scenarios and evaluate the method quantitatively and qualitatively.

We propose a novel monocular visual inertial odometry algorithm that combines the advantages of EKF-based approaches with those of direct photometric error minimization methods. The method is based on sparse, very small patches and incorporates the minimization of photometric error directly into the EKF measurement model so that inertial data and vision-based surface measurements are used simultaneously during camera pose estimation. We fuse vision-based and inertial measurements almost at the raw-sensor level, allowing the estimated system state to constrain and guide image-space measurements. Our formulation allows for an efficient implementation that runs in real-time on a standard CPU and has several appealing and unique characteristics such as being robust to fast camera motion, in particular rotation, and not depending on the presence of corner-like features in the scene. We experimentally demonstrate robust and accurate performance compared to ground truth and show that our method works on scenes containing only non-intersecting lines.

We present an interactive design system that allows casual users to quickly create 3D-printable robotic creatures. Our approach automates the tedious parts of the design process while providing ample room for customization of morphology, proportions, gait and motion style. The technical core of our framework is an efficient optimization-based solution that generates stable motions for legged robots of arbitrary designs. An intuitive set of editing tools allows the user to interactively explore the space of feasible designs and to study the relationship between morphological features and the resulting motions. Fabrication blueprints are generated automatically such that the robot designs can be manufactured using 3D-printing and off-the-shelf servo motors. We demonstrate the effectiveness of our solution by designing six robotic creatures with a variety of morphological features: two, four or five legs, point or area feet, actuated spines and different proportions. We validate the feasibility of the designs generated with our system through physics simulations and physically-fabricated prototypes.

We present a new type of augmented mechanical keyboard, capable of sensing rich and expressive motion gestures performed both on and directly above the device. Our hardware comprises of low-resolution matrix of infrared (IR) proximity sensors interspersed between the keys of a regular mechanical keyboard. This results in coarse but high frame-rate motion data. We extend a machine learning algorithm, traditionally used for static classification only, to robustly support dynamic, temporal gestures. We propose the use of motion signatures a technique that utilizes pairs of motion history images and a random forest based classifier to robustly recognize a large set of motion gestures on and directly above the keyboard. Our technique achieves a mean per-frame classification accuracy of 75.6% in leave-one-subject-out and 89.9% in half-test/half-training cross-validation. We detail our hardware and gesture recognition algorithm, provide performance and accuracy numbers, and demonstrate a large set of gestures designed to be performed with our device. We conclude with qualitative feedback from users, discussion of limitations and areas for future work.

Some aerial tasks are achieved more efficiently and at a lower cost by a group of independently controlled micro aerial vehicles (MAVs) when compared to a single, more sophisticated robot. Controlling formation flight can be cast as a two-level problem: stabilization of relative distances of agents (formation shape control) and control of the center of gravity of the formation. To date, accurate shape control of a formation of MAVs usually relies on external tracking devices (e.g. fixed cameras) or signals (e.g. GPS) and uses centralized control, which severely limits its deployment. In this paper, we present an environment-independent approach for relative MAV formation flight, using a distributed control algorithm which relies only on embedded sensing and agentto- agent communication. In particular, an on-board monocular camera is used to acquire relative distance measurements in combination with a consensus-based distributed Kalman filter. We evaluate our methods in- and outdoors with a formation of three MAVs while controlling the formation's center of gravity manually.

We present a novel machine learning based algorithm extending the interaction space around mobile devices. The technique uses only the RGB camera now commonplace on off-the-shelf mobile devices. Our algorithm robustly recognizes a wide range of in-air gestures, supporting user variation, and varying lighting conditions. We demonstrate that our algorithm runs in real-time on unmodified mobile devices, including resource-constrained smartphones and smartwatches. Our goal is not to replace the touchscreen as primary input device, but rather to augment and enrich the existing interaction vocabulary using gestures. While touch input works well for many scenarios, we demonstrate numerous interaction tasks such as mode switches, application and task management, menu selection and certain types of navigation, where such input can be either complemented or better served by in-air gestures. This removes screen real-estate issues on small touchscreens, and allows input to be expanded to the 3D space around the device. We present results for recognition accuracy (93% test and 98% train), impact of memory footprint and other model parameters. Finally, we report results from preliminary user evaluations, discuss advantages and limitations and conclude with directions for future work.

Articulation of 3D characters requires control over many degrees of freedom: a difficult task with standard 2D interfaces. We present a tangible input device composed of interchangeable, hot-pluggable parts. Embedded sensors measure the device's pose at rates suitable for real-time editing and animation. Splitter parts allow branching to accommodate any skeletal tree. During assembly, the device recognizes topological changes as individual parts or pre-assembled subtrees are plugged and unplugged. A novel semi-automatic registration approach helps the user quickly map the device's degrees of freedom to a virtual skeleton inside the character. User studies report favorable comparisons to mouse and keyboard interfaces for the tasks of target acquisition and pose replication. Our device provides input for character rigging and automatic weight computation, direct skeletal deformation, interaction with physical simulations, and handle-based variational geometric modeling.

Systems that detect the unaugmented human body allow players to interact without using a physical controller. But how is interaction altered by the absence of a physical input device? What is the impact on game performance, on a player's expectation of their ability to control the game, and on their game experience? In this study, we investigate these issues in the context of a table tennis video game. The results show that the impact of holding a physical controller, or indeed of the fidelity of that controller, does not appear in simple measures of performance. Rather, the difference between controllers is a function of the responsiveness of the game being controlled, as well as other factors to do with expectations, real world game experience and social context.

We present a novel yet simple technique that mitigates the interference caused when multiple structured light depth cameras point at the same part of a scene. The technique is particularly useful for Kinect, where the structured light source is not modulated. Our technique requires only mechanical augmentation of the Kinect, without any need to modify the internal electronics, firmware or associated host software. It is therefore simple to replicate. We show qualitative and quantitative results highlighting the improvements made to interfering Kinect depth signals. The camera frame rate is not compromised, which is a problem in approaches that modulate the structured light source. Our technique is non-destructive and does not impact depth values or geometry. We discuss uses for our technique, in particular within instrumented rooms that require simultaneous use of multiple overlapping fixed Kinect cameras to support whole room interactions.

HoloDesk is an interactive system combining an optical see through display and Kinect camera to create the illusion that users are directly interacting with 3D graphics. A virtual image of a 3D scene is rendered through a half silvered mirror and spatially aligned with the real-world for the viewer. Users easily reach into an interaction volume displaying the virtual image. This allows the user to literally get their hands into the virtual display and to directly interact with an spatially aligned 3D virtual world, without the need for any specialized head-worn hardware or input device. We introduce a new technique for interpreting raw Kinect data to approximate and track rigid (e.g., books, cups) and non-rigid (e.g., hands, paper) physical objects and support a variety of physics-inspired interactions between virtual and real. In particular the algorithm models natural human grasping of virtual objects with more fidelity than previously demonstrated. A qualitative study highlights rich emergent 3D interactions, using hands and real-world objects. The implementation of HoloDesk is described in full, and example application scenarios explored. Finally, HoloDesk is quantitatively evaluated in a 3D target acquisition task, comparing the system with indirect and glasses-based variants.

This paper describes a field study of an interactive surface deployed in three family homes. The tabletop technology provides a central place where digital content, such as photos, can be easily archived, managed and viewed. The tabletop affords multi-touch input, allowing digital content to be sorted, triaged and interacted with using one or two-handed interactions. A physics-based simulation adds dynamics to digital content, providing users with rich ways of interacting that borrows from the real-world. The field study is one of the first of a surface computer within a domestic environment. Our goal is to uncover people's inter-actions, appropriations, perceptions and experiences with such technologies, exploring the potential barriers to use. Given these devices provide such a revolutionary shift in interaction, will people be able to engage with them in everyday life in the ways we intend? In answering this question, we hope to deepen our understanding of the design of such systems for home and consumer domains.

Digits is a wrist-worn sensor that recovers the full 3D pose of the user's hand. This enables a variety of freehand interactions on the move. The system targets mobile settings, and is specifically designed to be low-power and easily reproducible using only off-the-shelf hardware. The electronics are self-contained on the user's wrist, but optically image the entirety of the user's hand. This data is processed using a new pipeline that robustly samples key parts of the hand, such as the tips and lower regions of each finger. These sparse samples are fed into new kinematic models that leverage the biomechanical constraints of the hand to recover the 3D pose of the user's hand. The proposed system works without the need for full instrumentation of the hand (for example using data gloves), additional sensors in the environment, or depth cameras which are currently prohibitive for mobile scenarios due to power and form-factor considerations. We demonstrate the utility of Digits for a variety of application scenarios, including 3D spatial interaction with mobile devices, eyes-free interaction on-the-move, and gaming. We conclude with a quantitative and qualitative evaluation of our system, and discussion of strengths, limitations and future work.

Steerable displays use a motorized platform to orient a projector to display graphics at any point in the room. Often a camera is included to recognize markers and other objects, as well as user gestures in the display volume. Such systems can be used to superimpose graphics onto the real world, and so are useful in a number of augmented reality and ubiquitous computing scenarios. We contribute the Beamatron, which advances steerable displays by drawing on recent progress in depth camera-based interactions. The Beamatron consists of a computer-controlled pan and tilt platform on which is mounted a projector and Microsoft Kinect sensor. While much previous work with steerable displays deals primarily with projecting corrected graphics onto a discrete set of static planes, we describe computational techniques that enable reasoning in 3D using live depth data. We show two example applications that are enabled by the unique capabilities of the Beamatron: an augmented reality game in which a player can drive a virtual toy car around a room, and a ubiquitous computing demo that uses speech and gesture to move projected graphics throughout the room.

We present a system for accurate real-time mapping of complex and arbitrary indoor scenes in variable lighting conditions, using only a moving low-cost depth camera and commodity graphics hardware. We fuse all of the depth data streamed from a Kinect sensor into a single global implicit surface model of the observed scene in real-time. The current sensor pose is simultaneously obtained by tracking the live depth frame relative to the global model using a coarse-to-fine iterative closest point (ICP) algorithm, which uses all of the observed depth data available. We demonstrate the advantages of tracking against the growing full surface model compared with frame-to-frame tracking, obtaining tracking and mapping results in constant time within room sized scenes with limited drift and high accuracy. We also show both qualitative and quantitative results relating to various aspects of our tracking and mapping system. Modelling of natural scenes, in real-time with only commodity sensor and GPU hardware, promises an exciting step forward in augmented reality (AR), in particular, it allows dense surfaces to be reconstructed in real-time, with a level of detail and robustness beyond any solution yet presented using passive computer vision.

We present Vermeer, a novel interactive 360° viewable 3D display. Like prior systems in this area, Vermeer provides viewpoint-corrected, stereoscopic 3D graphics to simultaneous users, 360° around the display, without the need for eyewear or other user instrumentation. Our goal is to over-come an issue inherent in these prior systems which - typically due to moving parts - restrict interactions to outside the display volume. Our system leverages a known optical illusion to demonstrate, for the first time, how users can reach into and directly touch 3D objects inside the display volume. Vermeer is intended to be a new enabling technology for interaction, and we therefore describe our hardware implementation in full, focusing on the challenges of combining this optical configuration with an existing approach for creating a 360° viewable 3D display. Initially we demonstrate direct involume interaction by sensing user input with a Kinect camera placed above the display. However, by exploiting the properties of the optical configuration, we also demonstrate novel prototypes for fully integrated input sensing alongside simultaneous display. We conclude by discussing limitations, implications for interaction, and ideas for future work.

KinectFusion enables a user holding and moving a standard Kinect camera to rapidly create detailed 3D reconstructions of an indoor scene. Only the depth data from Kinect is used to track the 3D pose of the sensor and reconstruct, geometrically precise, 3D models of the physical scene in real-time. The capabilities of KinectFusion, as well as the novel GPU-based pipeline are described in full. Uses of the core system for low-cost handheld scanning, and geometry-aware augmented reality and physics-based interactions are shown. Novel extensions to the core GPU pipeline demonstrate object segmentation and user interaction directly in front of the sensor, without degrading camera tracking or reconstruction. These extensions are used to enable real-time multi-touch interactions anywhere, allowing any planar or non-planar reconstructed physical surface to be appropriated for touch.

The Family Archive device is an interactive multi-touch tabletop technology with integrated capture facility for the archiving of sentimental artefacts and memorabilia. It was developed as a technology probe to help us open up current family archiving practices and to explore family archiving in situ. We detail the deployment and study of three of these devices in family homes and discuss how deploying a new, potentially disruptive, technology can foreground the social relations and organizing systems in domestic life. This in turn facilitates critical reflection on technology design.

In this paper we discuss some of our recent research work designing tabletop interfaces for co-located photo sharing. We draw particular attention to a specific feature of an interface design, which we have observed over an extensive number of uses, as facilitating an under-reported but none-the-less intriguing aspect of the photo-sharing experience - namely the process of 'getting sidetracked'. Through a series of vignettes of interaction during photo-sharing sessions we demonstrate how users of our tabletop photoware system used peripheral presentation of topically incoherent photos to artfully initiate new photo-talk sequences in on-going discourse. From this we draw implications for the design of tabletop photo applications, and for the experiential analysis of such devices.

Although interactive surfaces have many unique and compelling qualities, the interactions they support are by their very nature bound to the display surface. In this paper we present a technique for users to seamlessly switch between interacting on the tabletop surface to above it. Our aim is to leverage the space above the surface in combination with the regular tabletop display to allow more intuitive manipulation of digital content in three-dimensions. Our goal is to design a technique that closely resembles the ways we manipulate physical objects in the real-world; conceptually, allowing virtual objects to be 'picked up' off the tabletop surface in order to manipulate their three dimensional position or orientation. We chart the evolution of this technique, implemented on two rear projection-vision tabletops. Both use special projection screen materials to allow sensing at significant depths beyond the display. Existing and new computer vision techniques are used to sense hand gestures and postures above the tabletop, which can be used alongside more familiar multi-touch interactions. Interacting above the surface in this way opens up many interesting challenges. In particular it breaks the direct interaction metaphor that most tabletops afford. We present a novel shadow-based technique to help alleviate this issue. We discuss the strengths and limitations of our technique based on our own observations and initial user feedback, and provide various insights from comparing, and contrasting, our tabletop implementations

In this paper we investigate tangible interaction on interactive tabletops. These afford the support and integration of physical artefacts for the manipulation of digital media. To inform the design of interfaces for interactive surfaces we think it is necessary to deeply understand the benefits of employing such physical handles, i.e., the benefits of employing a third spatial dimension at the point of interaction.To this end we conducted an experimental study by designing and comparing two versions of an interactive tool on a tabletop display, one with a physical 3D handle, and one purely graphical (but direct touch enabled). Whilst hypothesizing that the 3D version would provide a number of benefits, our observations revealed that users developed diverse interaction approaches and attitudes about hybrid and direct touch interaction.

This paper explores the intersection of emerging surface technologies, capable of sensing multiple contacts and of-ten shape information, and advanced games physics engines. We define a technique for modeling the data sensed from such surfaces as input within a physics simulation. This affords the user the ability to interact with digital objects in ways analogous to manipulation of real objects. Our technique is capable of modeling both multiple contact points and more sophisticated shape information, such as the entire hand or other physical objects, and of mapping this user input to contact forces due to friction and collisions within the physics simulation. This enables a variety of fine-grained and casual interactions, supporting finger-based, whole-hand, and tangible input. We demonstrate how our technique can be used to add real-world dynamics to interactive surfaces such as a vision-based tabletop, creating a fluid and natural experience. Our approach hides from application developers many of the complexities inherent in using physics engines, allowing the creation of applications without preprogrammed interaction behavior or gesture recognition.

Collaborative creativity is traditionally supported by formal techniques, such as brainstorming. These techniques improve the idea-generation process by creating group synergies, but also suffer from a number of negative effects. Current electronic tools to support collaborative creativity overcome some of these problems, but introduce new ones, by either losing the benefits of face-to-face communication or the immediacy of simultaneous contribution. Using an interactive environment as a test bed, we are investigating how collaborative creativity can be supported electronically while maintaining face-to-face communication. What are the design-factors influencing such a system? We have designed a brainstorming application that uses an interactive table and a large wall display, and compared the results of using it to traditional paper-based brainstorming in a user study with 30 participants. From the considerations that went into the design and the observations during the study we derive a number of design guidelines for collaborative systems in interactive environments.

In this paper we present a wall-sized input system with high accuracy input in the center and lower precision tracking on the remaining parts of the wall. This work complements the concept of focus plus context displays where output quality in the center is better than in the periphery. Our contribution is to realize a similar concept for the input direction. A high precision tracking unit in the center provides a workspace with high interactivity, while four additional cameras placed in the corners of the wall provide lower precision tracking and hence much coarser interactivity on the entire wall. In our prototype this multi-precision input solution is combined with a focus plus context display, thus providing two levels of interactivity in both the input and output direction. With this setup, we have implemented two example applications to demonstrate the benefits of variable precision input

User interfaces for ubiquitous augmented reality incorporate a wide variety of concepts such as multi-modal, multi-user, multi-device aspects and new input/output devices. In this paper we present a twofold approach that consists of an execution engine for ubiquitous augmented reality user interfaces and a runtime development environment that enables rapid prototyping and live system adaption for such advanced user interfaces.