Collaborative human-object interaction shows dynamic and complex movements that require mutual anticipation and continuous adjustment between participants and the shared object. Understanding and modeling such collaborative multi-human object interaction (MHOI) scenarios requires high-quality data acquisition as a foundational step, however, this is a challenging task due to the inherent complexity of MHOI scenarios where human-human and human-object interactions occur simultaneously. Such complexity leads to noisy MHOI captures characterized by several artifacts: contact misalignment between hands and objects, motion jitter and temporal inconsistencies in the captured sequences, and missing or incomplete finger-level articulation details. To address these challenges, we present MOCHI (MOtion Enhancement of Collaborative Human-object Interactions), a two-stage framework for enhancing noisy MHOI data. Our approach first generates physically plausible hand grasps through optimization from noisy body input, producing grasps that are both physically plausible and semantically consistent with the body pose, where these optimized grasps are extended into complete hand-object interaction sequences. Consequently, the full-body motion for all participants are refined through a diffusion-based noise optimization framework that uses single-person motion priors. During the optimization process, we introduce optimization objectives to encode human-object and human-human interaction information within these single-person priors. Experimental results demonstrate the effectiveness of our pipeline across diverse MHOI data, either acquired by existing capture methods or synthesized by generative models. We further show robustness of our system across varying numbers of participants and types of interactions, and demonstrate various applications including keyframe-based MHOI creation and data augmentation through varying object geometries.

Abstract We present a versatile latent representation that enables physically simulated character to efficiently utilize motion priors. To build a powerful motion embedding that is shared across multiple tasks, the physics controller should employ rich latent space that is easily explored and capable of generating high‐quality motion. We propose integrating continuous and discrete latent representations to build a versatile motion prior that can be adapted to a wide range of challenging control tasks. Specifically, we build a discrete latent model to capture distinctive posterior distribution without collapse, and simultaneously augment the sampled vector with the continuous residuals to generate high‐quality, smooth motion without jittering. We further incorporate Residual Vector Quantization, which not only maximizes the capacity of the discrete motion prior, but also efficiently abstracts the action space during the task learning phase. We demonstrate that our agent can produce diverse yet smooth motions simply by traversing the learned motion prior through unconditional motion generation. Furthermore, our model robustly satisfies sparse goal conditions with highly expressive natural motions, including head‐mounted device tracking and motion in‐betweening at irregular intervals, which could not be achieved with existing latent representations.

We introduce ViSA (Virtual Stunt Actors), an interactive animation system designed to create realistic ballistic stunt actions frequently seen in filmmaking and TV production. By providing spatial constraints suitable for the desired stunt scene, our system generates physically plausible motions satisfying the given constraints. The problem is formulated as a deep reinforcement learning task, incorporating a novel state and action spaces, as well as straightforward yet effective rewards for ballistic stunt actions. Users can receive a fast response within several minutes and continue to choreograph complex stunt scenes in an interactive manner. We demonstrate ballistic stunt scenes resembling those in various films and TV dramas, such as traffic accidents, falling down stairs, and falls from buildings. The effectiveness of the technical components and design choices in our system is demonstrated through extensive comparisons, analyses, and ablation studies.

Given its wide applications, there is increasing focus on generating 3D human motions from textual descriptions. Differing from the majority of previous works, which regard actions as single entities and can only generate short sequences for simple motions, we propose EMS, an elaborative motion synthesis model conditioned on detailed natural language descriptions. It generates natural and smooth motion sequences for long and complicated actions by factorizing them into groups of atomic actions. Meanwhile, it understands atomic-action level attributes (e.g., motion direction, speed, and body parts) and enables users to generate sequences of unseen complex actions from unique sequences of known atomic actions with independent attribute settings and timings applied. We evaluate our method on the KIT Motion-Language and BABEL benchmarks, where it outperforms all previous state-of-the-art with noticeable margins.

We present a novel motion prior, called PhaseMP, modeling a probability distribution on pose transitions conditioned by a frequency domain feature extracted from a periodic autoencoder. The phase feature further enforces the pose transitions to be unidirectional (i.e. no backward movement in time), from which more stable and natural motions can be generated. Specifically, our motion prior can be useful for accurately estimating 3D human motions in the presence of challenging input data, including long periods of spatial and temporal occlusion, as well as noisy sensor measurements. Through a comprehensive evaluation, we demonstrate the efficacy of our novel motion prior, showcasing its superiority over existing state-of-the-art methods by a significant margin across various applications, including video-to-motion and motion estimation from sparse sensor data, and etc.

Synthesizing realistic human movements, dynamically responsive to the environment, is a long-standing objective in character animation, with applications in computer vision, sports, and healthcare, for motion prediction and data augmentation. Recent kinematics-based generative motion models offer impressive scalability in modeling extensive motion data, albeit without an interface to reason about and interact with physics. While simulator-in-the-loop learning approaches enable highly physically realistic behaviors, the challenges in training often affect scalability and adoption. We introduce DROP, a novel framework for modeling Dynamics Responses of humans using generative mOtion prior and Projective dynamics. DROP can be viewed as a highly stable, minimalist physics-based human simulator that interfaces with a kinematics-based generative motion prior. Utilizing projective dynamics, DROP allows flexible and simple integration of the learned motion prior as one of the projective energies, seamlessly incorporating control provided by the motion prior with Newtonian dynamics. Serving as a model-agnostic plug-in, DROP enables us to fully leverage recent advances in generative motion models for physics-based motion synthesis. We conduct extensive evaluations of our model across different motion tasks and various physical perturbations, demonstrating the scalability and diversity of responses.

Transforming neutral, characterless input motions to embody the distinct style of a notable character in real time is highly compelling for character animation. This paper introduces MOCHA, a novel online motion characterization framework that transfers both motion styles and body proportions from a target character to an input source motion. MOCHA begins by encoding the input motion into a motion feature that structures the body part topology and captures motion dependencies for effective characterization. Central to our framework is the Neural Context Matcher, which generates a motion feature for the target character with the most similar context to the input motion feature. The conditioned autoregressive model of the Neural Context Matcher can produce temporally coherent character features in each time frame. To generate the final characterized pose, our Characterizer network incorporates the characteristic aspects of the target motion feature into the input motion feature while preserving its context. This is achieved through a transformer model that introduces the adaptive instance normalization and context mapping-based cross-attention, effectively injecting the character feature into the source feature. We validate the performance of our framework through comparisons with prior work and an ablation study. Our framework can easily accommodate various applications, including characterization with only sparse input and real-time characterization. Additionally, we contribute a high-quality motion dataset comprising six different characters performing a range of motions, which can serve as a valuable resource for future research.

Learning deep neural networks on human motion data has become common in computer graphics research, but the heterogeneity of available datasets poses challenges for training large-scale networks. This paper presents a framework that allows us to solve various animation tasks in a skeleton-agnostic manner. The core of our framework is to learn an embedding space to disentangle skeleton-related information from input motion while preserving semantics, which we call Skeleton-Agnostic Motion Embedding (SAME). To efficiently learn the embedding space, we develop a novel autoencoder with graph convolution networks and provide new formulations of various animation tasks operating in the SAME space. We showcase various examples, including retargeting, reconstruction, and interactive character control, and conduct an ablation study to validate design choices made during development.

Motion retargeting is a promising approach for generating natural and compelling animations for nonhuman characters. However, it is challenging to translate human movements into semantically equivalent motions for target characters with different morphologies due to the ambiguous nature of the problem. This work presents a novel learning-based motion retargeting framework, Adversarial Correspondence Embedding (ACE), to retarget human motions onto target characters with different body dimensions and structures. Our framework is designed to produce natural and feasible character motions by leveraging generative-adversarial networks (GANs) while preserving high-level motion semantics by introducing an additional feature loss. In addition, we pretrain a character motion prior that can be controlled in a latent embedding space and seek to establish a compact correspondence. We demonstrate that the proposed framework can produce retargeted motions for three different characters – a quadrupedal robot with a manipulator, a crab character, and a wheeled manipulator. We further validate the design choices of our framework by conducting baseline comparisons and a user study. We also showcase sim-to-real transfer of the retargeted motions by transferring them to a real Spot robot.

Real-time human motion reconstruction from a sparse set of (e.g. six) wearable IMUs provides a non-intrusive and economic approach to motion capture. Without the ability to acquire position information directly from IMUs, recent works took data-driven approaches that utilize large human motion datasets to tackle this under-determined problem. Still, challenges remain such as temporal consistency, drifting of global and joint motions, and diverse coverage of motion types on various terrains. We propose a novel method to simultaneously estimate full-body motion and generate plausible visited terrain from only six IMU sensors in real-time. Our method incorporates 1. a conditional Transformer decoder model giving consistent predictions by explicitly reasoning prediction history, 2. a simple yet general learning target named "stationary body points” (SBPs) which can be stably predicted by the Transformer model and utilized by analytical routines to correct joint and global drifting, and 3. an algorithm to generate regularized terrain height maps from noisy SBP predictions which can in turn correct noisy global motion estimation. We evaluate our framework extensively on synthesized and real IMU data, and with real-time live demos, and show superior performance over strong baseline methods.

Real-time tracking of human body motion is crucial for interactive and immersive experiences in AR/VR. However, very limited sensor data about the body is available from standalone wearable devices such as HMDs (Head Mounted Devices) or AR glasses. In this work, we present a reinforcement learning framework that takes in sparse signals from an HMD and two controllers, and simulates plausible and physically valid full body motions. Using high quality full body motion as dense supervision during training, a simple policy network can learn to output appropriate torques for the character to balance, walk, and jog, while closely following the input signals. Our results demonstrate surprisingly similar leg motions to ground truth without any observations of the lower body, even when the input is only the 6D transformations of the HMD. We also show that a single policy can be robust to diverse locomotion styles, different body sizes, and novel environments.

Motion in-betweening (MIB) is a process of generating intermediate skeletal movement between the given start and target poses while preserving the naturalness of the motion, such as periodic footstep motion while walking. Although state-of-the-art MIB methods are capable of producing plausible motions given sparse key-poses, they often lack the controllability to generate motions satisfying the semantic contexts required in practical applications. We focus on the method that can handle pose or semantic conditioned MIB tasks using a unified model. We also present a motion augmentation method to improve the quality of pose-conditioned motion generation via defining a distribution over smooth trajectories. Our proposed method outperforms the existing state-of-the-art MIB method in pose prediction errors while providing additional controllability. Our code and results are available on our project web page: https://jihoonerd.github.io/Conditional-Motion-In-Betweening.

High-quality motion capture datasets are now publicly available, and researchers have used them to create kinematics-based controllers that can generate plausible and diverse human motions without conditioning on specific goals (i.e., a task-agnostic generative model). In this paper, we present an algorithm to build such controllers for physically simulated characters having many degrees of freedom. Our physics-based controllers are learned by using conditional VAEs, which can perform a variety of behaviors that are similar to motions in the training dataset. The controllers are robust enough to generate more than a few minutes of motion without conditioning on specific goals and to allow many complex downstream tasks to be solved efficiently. To show the effectiveness of our method, we demonstrate controllers learned from several different motion capture databases and use them to solve a number of downstream tasks that are challenging to learn controllers that generate natural-looking motions from scratch. We also perform ablation studies to demonstrate the importance of the elements of the algorithm. Code and data for this paper are available at: https://github.com/facebookresearch/PhysicsVAE

In two-player competitive sports, such as boxing and fencing , athletes often demonstrate efficient and tactical movements during a competition. In this paper, we develop a learning framework that generates control policies for physically simulated athletes who have many degrees-of-freedom. Our framework uses a two step-approach, learning basic skills and learning bout-level strategies, with deep reinforcement learning, which is inspired by the way that people how to learn competitive sports. We develop a policy model based on an encoder-decoder structure that incorporates an autoregressive latent variable, and a mixture-of-experts decoder. To show the effectiveness of our framework, we implemented two competitive sports, boxing and fencing , and demonstrate control policies learned by our framework that can generate both tactical and natural-looking behaviors. We also evaluate the control policies with comparisons to other learning configurations and with ablation studies.

Human characters with a broad range of natural looking and physically realistic behaviors will enable the construction of compelling interactive experiences. In this paper, we develop a technique for learning controllers for a large set of heterogeneous behaviors. By dividing a reference library of motion into clusters of like motions, we are able to construct experts , learned controllers that can reproduce a simulated version of the motions in that cluster. These experts are then combined via a second learning phase, into a general controller with the capability to reproduce any motion in the reference library. We demonstrate the power of this approach by learning the motions produced by a motion graph constructed from eight hours of motion capture data and containing a diverse set of behaviors such as dancing (ballroom and breakdancing), Karate moves, gesturing, walking, and running.

We present a novel and general framework for the design and control of underwater soft-bodied animals. The whole body of an animal consisting of soft tissues is modeled by tetrahedral and triangular FEM meshes. The contraction of muscles embedded in the soft tissues actuates the body and limbs to move. We present a novel muscle excitation model that mimics the anatomy of muscular hydrostats and their muscle excitation patterns. Our deep reinforcement learning algorithm equipped with the muscle excitation model successfully learned the control policy of soft-bodied animals, which can be physically simulated in real-time, controlled interactively, and resilient to external perturbations. We demonstrate the effectiveness of our approach with various simulated animals including octopuses, lampreys, starfishes, stingrays and cuttlefishes. They learn diverse behaviors such as swimming, grasping, and escaping from a bottle. We also implemented a simple user interface system that allows the user to easily create their creatures.

Recently, deep reinforcement learning (DRL) has attracted great attention in designing controllers for physics-based characters. Despite the recent success of DRL, the learned controller is viable for a single character. Changes in body size and proportions require learning controllers from scratch. In this paper, we present a new method of learning parametric controllers for body shape variation. A single parametric controller enables us to simulate and control various characters having different heights, weights, and body proportions. The users are allowed to create new characters through body shape parameters, and they can control the characters immediately. Our characters can also change their body shapes on the fly during simulation. The key to the success of our approach includes the adaptive sampling of body shapes that tackles the challenges in learning parametric controllers, which relies on the marginal value function that measures control capabilities of body shapes. We demonstrate parametric controllers for various physically simulated characters such as bipeds, quadrupeds, and underwater animals.

Simulating believable virtual crowds has been an important research topic in many research fields such as industry films, computer games, urban engineering, and behavioral science. One of the key capabilities agents should have is navigation, which is reaching goals without colliding with other agents or obstacles. The key challenge here is that the environment changes dynamically, where the current decision of an agent can largely affect the state of other agents as well as the agent in the future. Recently, reinforcement learning with deep neural networks has shown remarkable results in sequential decision-making problems. With the power of convolution neural networks, elaborate control with visual sensory inputs has also become possible. In this paper, we present an agent-based deep reinforcement learning approach for navigation, where only a simple reward function enables agents to navigate in various complex scenarios. Our method is also able to do that with a single unified policy for every scenario, where the scenario-specific parameter tuning is unnecessary. We will show the effectiveness of our method through a variety of scenarios and settings.

Flying creatures in animated films often perform highly dynamic aerobatic maneuvers, which require their extreme of exercise capacity and skillful control. Designing physics-based controllers (a.k.a., control policies) for aerobatic maneuvers is very challenging because dynamic states remain in unstable equilibrium most of the time during aerobatics. Recently, Deep Reinforcement Learning (DRL) has shown its potential in constructing physics-based controllers. In this paper, we present a new concept, Self-Regulated Learning (SRL) , which is combined with DRL to address the aerobatics control problem. The key idea of SRL is to allow the agent to take control over its own learning using an additional self-regulation policy. The policy allows the agent to regulate its goals according to the capability of the current control policy. The control and self-regulation policies are learned jointly along the progress of learning. Self-regulated learning can be viewed as building its own curriculum and seeking compromise on the goals. The effectiveness of our method is demonstrated with physically-simulated creatures performing aerobatic skills of sharp turning, rapid winding, rolling, soaring, and diving.

Imaginary winged creatures in computer animation applications are expected to perform a variety of motor skills in a physically realistic and controllable manner. Designing physics-based controllers for a flying creature is still very challenging particularly when the dynamic model of the creatures is high-dimensional, having many degrees of freedom. In this paper, we present a control method for flying creatures, which are aerodynamically simulated, interactively controllable, and equipped with a variety of motor skills such as soaring, gliding, hovering, and diving. Each motor skill is represented as Deep Neural Networks (DNN) and learned using Deep Q-Learning (DQL). Our control method is example-guided in the sense that it provides the user with direct control over the learning process by allowing the user to specify keyframes of motor skills. Our novel learning algorithm was inspired by evolutionary strategies of Covariance Matrix Adaptation Evolution Strategy (CMA-ES) to improve the convergence rate and the final quality of the control policy. The effectiveness of our Evolutionary DQL method is demonstrated with imaginary winged creatures flying in a physically simulated environment and their motor skills learned automatically from user-provided keyframes.

Shadow theatre is a genre of performance art in which the actors are only visible as shadows projected on the screen. The goal of this study is to generate animated characters, the shadows of which match a sequence of target silhouettes. This poses several challenges. The motion of multiple characters are carefully coordinated to form a target silhouette on the screen, and each character's pose should be stable, balanced, and plausible. The resulting character animation should be smooth and coherent spatially and temporally. We formulate the problem as nonlinear constrained optimization with objectives, which were designed to generate plausible human motions. Our optimization algorithm was primarily inspired by the heuristic strategies of professional shadow theatre actors. Their know-how was studied and then incorporated into our optimization formulation. We demonstrate the effectiveness of our approach with a variety of target silhouettes and 3D fabrication of the results.

In many application areas, such as animation for pre-visualizing movie sequences and choreography for dance or other types of performance, only a high-level description of the desired scene is provided as input, either written or verbal. Such sparsity, however, lends itself well to the creative process, as the choreographer, animator or director can be given more choice and control of the final scene. Animating scenes with multi-character interactions can be a particularly complex process, as there are many different constraints to enforce and actions to synchronize. Our novel 'generate-and-rank' approach rapidly and semi-automatically generates data-driven multi-character interaction scenes from high-level graphical descriptions composed of simple clauses and phrases. From a database of captured motions, we generate a multitude of plausible candidate scenes. We then efficiently and intelligently rank these scenes in order to recommend a small but high-quality and diverse selection to the user. This set can then be refined by re-ranking or by generating alternatives to specific interactions. While our approach is applicable to any scenes that depict multi-character interactions, we demonstrate its efficacy for choreographing fighting scenes and evaluate it in terms of performance and the diversity and coverage of the results.

We present a physically based controller that simulates the flapping behavior of a bird in flight. We recorded the motion of a dove using marker-based optical motion capture and high-speed video cameras. The bird flight data thus acquired allow us to parameterize natural wingbeat cycles and provide the simulated bird with reference trajectories to track in physics simulation. Our controller simulates articulated rigid bodies of a bird's skeleton and deformable feathers to reproduce the aerodynamics of bird flight. Motion capture from live birds is not as easy as human motion capture because of the lack of cooperation from subjects. Therefore, the flight data we could acquire were limited. We developed a new method to learn wingbeat controllers even from sparse, biased observations of real bird flight. Our simulated bird imitates life-like flapping of a flying bird while actively maintaining its balance. The bird flight is interactively controllable and resilient to external disturbances.