Recent advances in self-supervised learning (SSL) have shown tremendous potential for learning 3D point cloud representations without human annotations. However, SSL for 3D point clouds still faces critical challenges due to irregular geometry, shortcut-prone reconstruction, and unbalanced semantics distribution. In this work, we propose DOS (Distilling Observable Softmaps), a novel SSL framework that self-distills semantic relevance softmaps only at observable (unmasked) points. This strategy prevents information leakage from masked regions and provides richer supervision than discrete token-to-prototype assignments. To address the challenge of unbalanced semantics in an unsupervised setting, we introduce Zipfian prototypes and incorporate them using a modified Sinkhorn-Knopp algorithm, Zipf-Sinkhorn, which enforces a power-law prior over prototype usage and modulates the sharpness of the target softmap during training. DOS outperforms current state-of-the-art methods on semantic segmentation and 3D object detection across multiple benchmarks, including nuScenes, Waymo, SemanticKITTI, ScanNet, and ScanNet200, without relying on extra data or annotations. Our results demonstrate that observable-point softmaps distillation offers a scalable and effective paradigm for learning robust 3D representations.

Large Language Models (LLMs) are increasingly used for decision-making and planning in autonomous driving, showing promising reasoning capabilities and potential to generalize across diverse traffic situations. However, current LLM-based driving agents lack explicit mechanisms to enforce traffic rules and often struggle to reliably detect small, safety-critical objects such as traffic lights and signs. To address this limitation, we introduce TLS-Assist, a modular redundancy layer that augments LLM-based autonomous driving agents with explicit traffic light and sign recognition. TLS-Assist converts detections into structured natural language messages that are injected into the LLM input, enforcing explicit attention to safety-critical cues. The framework is plug-and-play, model-agnostic, and supports both single-view and multi-view camera setups. We evaluate TLS-Assist in a closed-loop setup on the LangAuto benchmark in CARLA. The results demonstrate relative driving performance improvements of up to 14% over LMDrive and 7% over BEVDriver, while consistently reducing traffic light and sign infractions. We publicly release the code and models on https://github.com/iis-esslingen/TLS-Assist.

We present Mixture of Discrete-time Gaussian Processes (MiDiGaP), a novel approach for flexible policy representation and imitation learning in robot manipulation. MiDiGaP enables learning from as few as five demonstrations using only camera observations and generalizes across a wide range of challenging tasks. It excels at long-horizon behaviors such as making coffee, highly constrained motions such as opening doors, dynamic actions such as scooping with a spatula, and multimodal tasks such as hanging a mug. MiDiGaP learns these tasks on a CPU in less than a minute and scales linearly to large datasets. We also develop a rich suite of tools for inferencetime steering using evidence such as collision signals and robot kinematic constraints. This steering enables novel generalization capabilities, including obstacle avoidance and cross-embodiment policy transfer. MiDiGaP achieves state-of-the-art performance on diverse few-shot manipulation benchmarks. On constrained RLBench tasks, it improves policy success by 76 percentage points and reduces trajectory cost by 67%. On multimodal tasks, it improves policy success by 48 percentage points while improving sample efficiency 7-fold. In cross-embodiment transfer, it more than doubles policy success. We make the code publicly available.

Masked autoencoders (MAE) have shown tremendous potential for self-supervised learning (SSL) in vision and beyond. However, point clouds from LiDARs used in automated driving are particularly challenging for MAEs since large areas of the 3D volume are empty. Consequently, existing work suffers from leaking occupancy information into the decoder and has significant computational complexity, thereby limiting the SSL pre-training to only 2D bird's eye view encoders in practice. In this work, we propose the novel neighborhood occupancy MAE (NOMAE) that overcomes the aforementioned challenges by employing masked occupancy reconstruction only in the neighborhood of non- masked voxels. We incorporate voxel masking and occupancy reconstruction at multiple scales with our proposed hierarchical mask generation technique to capture features of objects of different sizes in the point cloud. NOMAEs are extremely flexible and can be directly employed for SSL in existing 3D architectures. We perform extensive evaluations on the nuScenes and Waymo Open datasets for the downstream perception tasks of semantic segmentation and 3D object detection, comparing with both discriminative and generative SSL methods. The results demonstrate that NOMAE sets the new state-of-the-art on multiple benchmarks for multiple point cloud perception tasks.

Forecasting the future trajectories of surrounding agents is crucial for autonomous vehicles to ensure safe, efficient, and comfortable route planning. While model ensembling has improved prediction accuracy in various fields, its application in trajectory prediction is limited due to the multi-modal nature of predictions. In this paper, we propose a novel sampling method applicable to trajectory prediction based on the predictions of multiple models. We first show that conventional sampling based on predicted probabilities can degrade performance due to missing alignment between models. To address this problem, we introduce a new method that generates optimal trajectories from a set of neural networks, framing it as a risk minimization problem with a variable loss function. By using state-of-the-art models as base learners, our approach constructs diverse and effective ensembles for optimal trajectory sampling. Extensive experiments on the nuScenes prediction dataset demonstrate that our method surpasses current state-of-the-art techniques, achieving top ranks on the leaderboard. We also provide a comprehensive empirical study on ensembling strategies, offering insights into their effectiveness. Our findings highlight the potential of advanced ensembling techniques in trajectory prediction, significantly improving predictive performance and paving the way for more reliable predicted trajectories.

Forecasting the semantics and 3D structure of scenes is essential for robots to navigate and plan actions safely. Recent methods have explored semantic and panoptic scene forecasting; however, they do not consider the geometry of the scene. In this work, we propose the panoptic-depth forecasting task for jointly predicting the panoptic segmentation and depth maps of unobserved future frames, from monocular camera images. To facilitate this work, we extend the popular KITTI-360 and Cityscapes benchmarks by computing depth maps from LiDAR point clouds and leveraging sequential labeled data. We also introduce a suitable evaluation metric that quantifies both the panoptic quality and depth estimation accuracy of forecasts in a coherent manner. Furthermore, we present two baselines and propose the novel PDcast architecture that learns rich spatio-temporal representations by incorporating a transformer-based encoder, a forecasting module, and task-specific decoders to predict future panoptic-depth outputs. Extensive evaluations demonstrate the effectiveness of PDcast across two datasets and three forecasting tasks, consistently addressing the primary challenges. We make the code publicly available at https://pdcast.cs.uni-freiburg.de

Tactile sensing is vital for human dexterous manipulation, however, it has not been widely used in robotics. Compact, low-cost sensing platforms can facilitate a change, but unlike their popular optical counterparts, they are difficult to deploy in high-fidelity tasks due to their low signal dimensionality and lack of a simulation model. To overcome these challenges, we introduce PseudoTouch which links high-dimensional structural information to low-dimensional sensor signals. It does so by learning a low-dimensional visual-tactile embedding, wherein we encode a depth patch from which we decode the tactile signal. We collect and train PseudoTouch on a dataset comprising aligned tactile and visual data pairs obtained through random touching of eight basic geometric shapes. We demonstrate the utility of our trained PseudoTouch model in two downstream tasks: object recognition and grasp stability prediction. In the object recognition task, we evaluate the learned embedding's performance on a set of five basic geometric shapes and five household objects. Using PseudoTouch, we achieve an object recognition accuracy 84% after just ten touches, surpassing a proprioception baseline. For the grasp stability task, we use ACRONYM labels to train and evaluate a grasp success predictor using PseudoTouch's predictions derived from virtual depth information. Our approach yields a 32% absolute improvement in accuracy compared to the baseline relying on partial point cloud data. We make the data, code, and trained models publicly available at https://pseudotouch.cs.uni-freiburg.de.

Visual loop closure detection traditionally relies on place recognition methods to retrieve candidate loops that are validated using computationally expensive RANSAC-based geometric verification. As false positive loop closures significantly degrade downstream pose graph estimates, verifying a large number of candidates in online simultaneous localization and mapping scenarios is constrained by limited time and compute resources. While most deep loop closure detection approaches only operate on pairs of keyframes, we relax this constraint by considering neighborhoods of multiple keyframes when detecting loops. In this work, we introduce LoopGNN, a graph neural network architecture that estimates loop closure consensus by leveraging cliques of visually similar keyframes retrieved through place recognition. By propagating deep feature encodings among nodes of the clique, our method yields high precision estimates while maintaining high recall. Extensive experimental evaluations on the TartanDrive 2.0 and NCLT datasets demonstrate that LoopGNN outperforms traditional baselines. Additionally, an ablation study across various keypoint extractors demonstrates that our method is robust, regardless of the type of deep feature encodings used, and exhibits higher computational efficiency compared to classical geometric verification baselines. We release our code, supplementary material, and keyframe data at https://loopgnn.cs.uni-freiburg.de.

A main bottleneck of learning-based robotic scene understanding methods is the heavy reliance on extensive annotated training data, which often limits their generalization ability. In LiDAR panoptic segmentation, this challenge becomes even more pronounced due to the need to simultaneously address both semantic and instance segmentation from complex, high-dimensional point cloud data. In this work, we address the challenge of LiDAR panoptic segmentation with very few labeled samples by leveraging recent advances in label-efficient vision panoptic segmentation. To this end, we propose a novel method, Limited-Label LiDAR Panoptic Segmentation (L3PS), which requires only a minimal amount of labeled data. Our approach first utilizes a label-efficient 2D network to generate panoptic pseudo-labels from a small set of annotated images, which are subsequently projected onto point clouds. We then introduce a novel 3D refinement module that capitalizes on the geometric properties of point clouds. By incorporating clustering techniques, sequential scan accumulation, and ground point separation, this module significantly enhances the accuracy of the pseudo-labels, improving segmentation quality by up to +10.6 PQ and +7.9 mIoU. We demonstrate that these refined pseudo-labels can be used to effectively train off-the-shelf LiDAR segmentation networks. Through extensive experiments, we show that L3PS not only outperforms existing methods but also substantially reduces the annotation burden. We release the code of our work at https://l3ps.cs.uni-freiburg.de.

Language-Conditioned robotic policies allow users to specify tasks using natural language. While much research has focused on improving the action prediction of language-conditioned policies, reasoning about task descriptions has been largely overlooked. Ambiguous task descriptions often lead to downstream policy failures due to misinterpretation by the robotic agent. To address this challenge, we introduce AmbResVLM, a novel method that grounds language goals in the observed scene and explicitly reasons about task ambiguity. We extensively evaluate its effectiveness in both simulated and realworld domains, demonstrating superior task ambiguity detection and resolution compared to recent state-of-the-art methods. Finally, real robot experiments show that our model improves the performance of downstream robot policies, increasing the average success rate from 69.6% to 97.1%. We make the data, code, and trained models publicly available at https://ambres.cs.uni-freiburg.de.

In autonomous driving, accurate motion prediction is crucial for safe and efficient motion planning. To ensure safety, planners require reliable uncertainty estimates of the predicted behavior of surrounding agents, yet this aspect has received limited attention. In particular, decomposing uncertainty into its aleatoric and epistemic components is essential for distinguishing between inherent environmental randomness and model uncertainty, thereby enabling more robust and informed decision-making. This paper addresses the challenge of uncertainty modeling in trajectory prediction with a holistic approach that emphasizes uncertainty quantification, decomposition, and the impact of model composition. Our method, grounded in information theory, provides a theoretically principled way to measure uncertainty and decompose it into aleatoric and epistemic components. Unlike prior work, our approach is compatible with state-of-the-art motion predictors, allowing for broader applicability. We demonstrate its utility by conducting extensive experiments on the nuScenes dataset, which shows how different architectures and configurations influence uncertainty quantification and model robustness.

LiDAR semantic segmentation models are typically trained from random initialization as universal pre-training is hindered by the lack of large, diverse datasets. Moreover, most point cloud segmentation architectures incorporate custom network layers, limiting the transferability of advances from vision-based architectures. Inspired by recent advances in universal foundation models, we propose BALViT, a novel approach that leverages frozen vision models as amodal feature encoders for learning strong LiDAR encoders. Specifically, BALViT incorporates both range-view and bird’s-eye-view LiDAR encoding mechanisms, which we combine through a novel 2D-3D adapter. While the range-view features are processed through a frozen image backbone, our bird’s-eye-view branch enhances them through multiple cross-attention interactions. Thereby, we continuously improve the vision network with domain-dependent knowledge, resulting in a strong label-efficient LiDAR encoding mechanism. Extensive evaluations of BALViT on the SemanticKITTI and nuScenes benchmarks demonstrate that it outperforms state-of-the-art methods on small data regimes. We make the code and models publicly available at http://balvit.cs.uni-freiburg.de.

Panoptic tracking enables pixel-level scene interpretation of videos by integrating instance tracking in panoptic segmentation. This provides robots with a spatio-temporal understanding of the environment, an essential attribute for their operation in dynamic environments. In this paper, we propose a novel approach for panoptic tracking that simultaneously captures general semantic information and instance-specific appearance and motion features. Unlike existing methods that overlook dynamic scene attributes, our approach leverages both appearance and motion cues through dedicated network heads. These interconnected heads employ multi-scale deformable convolutions that reason about scene motion offsets with semantic context and motion-enhanced appearance features to learn tracking embeddings. Furthermore, we introduce a novel two-step fusion module that integrates the outputs from both heads by first matching instances from the current time step with propagated instances from previous time steps and subsequently refines associations using motion-enhanced appearance embeddings, improving robustness in challenging scenarios. Extensive evaluations of our proposed MAPT model on two benchmark datasets demonstrate that it achieves state-of-the-art performance in panoptic tracking accuracy, surpassing prior methods in maintaining object identities over time. To facilitate future research, we make the code available at http://panoptictracking.cs.uni-freiburg.de.

Accurate trajectory prediction is crucial for autonomous driving, yet uncertainty in agent behavior and perception noise makes it inherently challenging. While multi-modal trajectory prediction models generate multiple plausible future paths with associated probabilities, effectively quantifying uncertainty remains an open problem. In this work, we propose a novel multi-modal trajectory prediction approach based on evidential deep learning that estimates both positional and mode probability uncertainty in real time. Our approach leverages a Normal Inverse Gamma distribution for positional uncertainty and a Dirichlet distribution for mode uncertainty. Unlike sampling-based methods, it infers both types of uncertainty in a single forward pass, significantly improving efficiency. Additionally, we experimented with uncertainty-driven importance sampling to improve training efficiency by prioritizing underrepresented high-uncertainty samples over redundant ones. We perform extensive evaluations of our method on the Argoverse 1 and Argoverse 2 datasets, demonstrating that it provides reliable uncertainty estimates while maintaining high trajectory prediction accuracy.

Autonomous long-horizon mobile manipulation encompasses a multitude of challenges, including scene dynamics, unexplored areas, and error recovery. Recent works have leveraged foundation models for scene-level robotic reasoning and planning. However, the performance of these methods degrades when dealing with a large number of objects and largescale environments. To address these limitations, we propose MORE, a novel approach for enhancing the capabilities of language models to solve zero-shot mobile manipulation planning for rearrangement tasks. MORE leverages scene graphs to represent environments, incorporates instance differentiation, and introduces an active filtering scheme that extracts task-relevant subgraphs of object and region instances. These steps yield a bounded planning problem, effectively mitigating hallucinations and improving reliability. Additionally, we introduce several enhancements that enable planning across both indoor and outdoor environments. We evaluate MORE on 81 diverse rearrangement tasks from the BEHAVIOR-1K benchmark, where it becomes the first approach to successfully solve a significant share of the benchmark, outperforming recent foundation model-based approaches. Furthermore, we demonstrate the capabilities of our approach in several complex real-world tasks, mimicking everyday activities. We make the code publicly available at https://more-model.cs.uni-freiburg.de.

Autonomous vehicles that navigate in open-world environments may encounter previously unseen object classes. However, most existing LiDAR panoptic segmentation models rely on closed-set assumptions, failing to detect unknown object instances. In this work, we propose ULOPS, an uncertainty-guided open-set panoptic segmentation framework that leverages Dirichlet-based evidential learning to model predictive uncertainty. Our architecture incorporates separate decoders for semantic segmentation with uncertainty estimation, embedding with prototype association, and instance center prediction. During inference, we leverage uncertainty estimates to identify and segment unknown instances. To strengthen the model’s ability to differentiate between known and unknown objects, we introduce three uncertainty-driven loss functions. Uniform Evidence Loss to encourage high uncertainty in unknown regions. Adaptive Uncertainty Separation Loss ensures a consistent difference in uncertainty estimates between known and unknown objects at a global scale. Contrastive Uncertainty Loss refines this separation at the fine-grained level. To evaluate open-set performance, we extend benchmark settings on KITTI-360 and introduce a new open-set evaluation for nuScenes. Extensive experiments demonstrate that ULOPS consistently outperforms existing open-set LiDAR panoptic segmentation methods. We make the code and pre-trained models available at http://ulops.cs.uni-freiburg.de.

Mapping and scene representation are fundamental to reliable planning and navigation in mobile robots. While purely geometric maps using voxel grids allow for general navigation, obtaining up-to-date spatial and semantically rich representations that scale to dynamic large-scale environments remains challenging. In this work, we present CURB-OSG, an open-vocabulary dynamic 3D scene graph engine that generates hierarchical decompositions of urban driving scenes via multi-agent collaboration. By fusing the camera and LiDAR observations from multiple perceiving agents with unknown initial poses, our approach generates more accurate maps compared to a single agent while constructing a unified open-vocabulary semantic hierarchy of the scene. Unlike previous methods that rely on ground truth agent poses or are evaluated purely in simulation, CURB-OSG alleviates these constraints. We evaluate the capabilities of CURB-OSG on real-world multi-agent sensor data obtained from multiple sessions of the Oxford Radar RobotCar dataset. We demonstrate improved mapping and object prediction accuracy through multi-agent collaboration as well as evaluate the environment partitioning capabilities of the proposed approach. To foster further research, we release our code and supplementary material at https://ovcurb.cs.uni-freiburg.de.

Automated parking is a critical feature of Advanced Driver Assistance Systems (ADAS), where accurate trajectory prediction is essential to bridge perception and planning modules. Despite its significance, research in this domain remains relatively limited, with most existing studies concentrating on single-modal trajectory prediction of vehicles. In this work, we propose ParkDiffusion, a novel approach that predicts the trajectories of both vehicles and pedestrians in automated parking scenarios. ParkDiffusion employs diffusion models to capture the inherent uncertainty and multi-modality of future trajectories, incorporating several key innovations. First, we propose a dual map encoder that processes soft semantic cues and hard geometric constraints using a two-step cross-attention mechanism. Second, we introduce an adaptive agent type embedding module, which dynamically conditions the prediction process on the distinct characteristics of vehicles and pedestrians. Third, to ensure kinematic feasibility, our model outputs control signals that are subsequently used within a kinematic framework to generate physically feasible trajectories. We evaluate ParkDiffusion on the Dragon Lake Parking (DLP) dataset and the Intersections Drone (inD) dataset. Our work establishes a new baseline for heterogeneous trajectory prediction in parking scenarios, outperforming existing methods by a considerable margin.

We address the problem of generating long-horizon videos for robotic manipulation tasks. Text-to-video diffusion models have made significant progress in photorealism, language understanding, and motion generation but struggle with long-horizon robotic tasks. Recent works use video diffusion models for high-quality simulation data and predictive rollouts in robot planning. However, these works predict short sequences of the robot achieving one task and employ an autoregressive paradigm to extend to the long horizon, leading to error accumulations in the generated video and in the execution. To overcome these limitations, we propose a novel pipeline that bypasses the need for autoregressive generation. We achieve this through a threefold contribution: 1) we first decompose the high-level goals into smaller atomic tasks and generate keyframes aligned with these instructions. A second diffusion model then interpolates between each of the two generated frames, achieving the long-horizon video. 2) We propose a semantics preserving attention module to maintain consistency between the keyframes. 3) We design a lightweight policy model to regress the robot joint states from generated videos. Our approach achieves state-of-the-art results on two benchmarks in video quality and consistency while outperforming previous policy models on long-horizon tasks.

3D scene understanding has been transformed by open-vocabulary language models that enable interaction via natural language. However, at present the evaluation of these representations is limited to datasets with closed-set semantics that do not capture the richness of language. This work presents OpenLex3D, a dedicated benchmark for evaluating 3D open-vocabulary scene representations. OpenLex3D provides entirely new label annotations for scenes from Replica, ScanNet++, and HM3D, which capture real-world linguistic variability by introducing synonymical object categories and additional nuanced descriptions. Our label sets provide 13 times more labels per scene than the original datasets. By introducing an open-set 3D semantic segmentation task and an object retrieval task, we evaluate various existing 3D open-vocabulary methods on OpenLex3D, showcasing failure cases, and avenues for improvement. Our experiments provide insights on feature precision, segmentation, and downstream capabilities. The benchmark is publicly available at: https://openlex3d.github.io/.

Light detection and rangings (LiDARs) are widely used for mapping and localization in dynamic environments. However, their high cost limits their widespread adoption. On the other hand, monocular localization in LiDAR maps using inexpensive cameras is a cost-effective alternative for large-scale deployment. Nevertheless, most existing approaches struggle to generalize to new sensor setups and environments, requiring retraining or fine-tuning. In this article, we present CMRNext, a novel approach for camera-LiDAR matching that is independent of sensor-specific parameters, generalizable, and can be used in the wild for monocular localization in LiDAR maps and camera-LiDAR extrinsic calibration. CMRNext exploits recent advances in deep neural networks for matching cross-modal data and standard geometric techniques for robust pose estimation. We reformulate the point-pixel matching problem as an optical flow estimation problem and solve the perspective-n-point problem based on the resulting correspondences to find the relative pose between the camera and the LiDAR point cloud. We extensively evaluate CMRNext on six different robotic platforms, including three publicly available datasets and three in-house robots. Our experimental evaluations demonstrate that CMRNext outperforms existing approaches on both tasks and effectively generalizes to previously unseen environments and sensor setups in a zero-shot manner.

Robust Simultaneous Localization and Mapping (SLAM) is a crucial enabler for autonomous navigation in natural, semi-structured environments such as parks and gardens. However, these environments present unique challenges for SLAM due to frequent seasonal changes, varying light conditions, and dense vegetation. These factors often degrade the performance of visual SLAM algorithms originally developed for structured urban environments. To address this gap, we present ROVER, a comprehensive benchmark dataset tailored for evaluating visual SLAM algorithms under diverse environmental conditions and spatial configurations. We captured the dataset with a robotic platform equipped with monocular, stereo, and RGBD cameras, as well as inertial sensors. It covers 39 recordings across five outdoor locations, collected through all seasons and various lighting scenarios, i.e., day, dusk, and night with and without external lighting. With this novel dataset, we evaluate several traditional and deep learning-based SLAM methods and study their performance in diverse challenging conditions. The results demonstrate that while stereo-inertial and RGBD configurations generally perform better under favorable lighting and moderate vegetation, most SLAM systems perform poorly in low-light and high-vegetation scenarios, particularly during summer and autumn. Our analysis highlights the need for improved adaptability in visual SLAM algorithms for outdoor applications, as current systems struggle with dynamic environmental factors affecting scale, feature extraction, and trajectory consistency. This dataset provides a solid foundation for advancing visual SLAM research in real-world, semi-structured environments, fostering the development of more resilient SLAM systems for long-term outdoor localization and mapping. The dataset and the code of the benchmark are available underhttps://iis-esslingen.github.io/rover.

Learning-based manipulation policies from image inputs often show weak task transfer capabilities. In contrast, visual servoing methods allow efficient task transfer in high-precision scenarios while requiring only a few demonstrations. In this work, we present a framework that formulates the visual servoing task as graph traversal. Our method not only extends the robustness of visual servoing, but also enables multitask capability based on a few task-specific demonstrations. We construct demonstration graphs by splitting existing demonstrations and recombining them. In order to traverse the demonstration graph in the inference case, we utilize a similarity function that helps select the best demonstration for a specific task. This enables us to compute the shortest path through the graph. Ultimately, we show that recombining demonstrations leads to higher task-respective success. We present extensive simulation and real-world experimental results that demonstrate the efficacy of our approach.

From dishwashers to cabinets, humans interact with articulated objects every day, and for a robot to assist in common manipulation tasks, it must learn a representation of articulation. Recent deep learning methods can provide powerful vision-based priors on the affordance of articulated objects from previous, possibly simulated, experiences. In contrast, many other works estimate articulation by observing the object in motion, requiring the robot to already be interacting with the object. In this work, we propose to use the best of both worlds by introducing an online estimation method that merges vision-based affordance predictions from a neural network with interactive kinematic sensing in an analytical model. Our work has the benefit of using vision to predict an articulation model before touching the object, while also being able to update the model quickly from kinematic sensing during the interaction. In this paper, we implement a full system using shared autonomy for robotic opening of articulated objects, in particular objects in which the articulation is not apparent from vision alone. We implemented our system on a real robot and performed several autonomous closed-loop experiments in which the robot had to open a door with unknown joint while estimating the articulation online. Our system achieved an 80% success rate for autonomous opening of unknown articulated objects.

Maps have played an indispensable role in enabling safe and automated driving. Although there have been many advances on different fronts ranging from SLAM to semantics, building an actionable hierarchical semantic representation of urban dynamic scenes and processing information from multiple agents are still challenging problems. In this work, we present Collaborative URBan Scene Graphs (CURB-SG) that enable higher-order reasoning and efficient querying for many functions of automated driving. CURB-SG leverages panoptic LiDAR data from multiple agents to build large-scale maps using an effective graph-based collaborative SLAM approach that detects inter-agent loop closures. To semantically decompose the obtained 3D map, we build a lane graph from the paths of ego agents and their panoptic observations of other vehicles. Based on the connectivity of the lane graph, we segregate the environment into intersecting and non-intersecting road areas. Subsequently, we construct a multi-layered scene graph that includes lane information, the position of static landmarks and their assignment to certain map sections, other vehicles observed by the ego agents, and the pose graph from SLAM including 3D panoptic point clouds. We extensively evaluate CURB-SG in urban scenarios using a photorealistic simulator. We release our code at http://curb.cs.uni-freiburg.de.

Current state-of-the-art methods for panoptic segmentation require an immense amount of annotated training data that is both arduous and expensive to obtain posing a significant challenge for their widespread adoption. Concurrently, recent breakthroughs in visual representation learning have sparked a paradigm shift leading to the advent of large foundation models that can be trained with completely unlabeled images. In this work, we propose to leverage such task-agnostic image features to enable few-shot panoptic segmentation by presenting Segmenting Panoptic Information with Nearly 0 labels (SPINO). In detail, our method combines a DINOv2 backbone with lightweight network heads for semantic segmentation and boundary estimation. We show that our approach, albeit being trained with only ten annotated images, predicts high-quality pseudo-labels that can be used with any existing panoptic segmentation method. Notably, we demonstrate that SPINO achieves competitive results compared to fully supervised baselines while using less than 0.3% of the ground truth labels, paving the way for learning complex visual recognition tasks leveraging foundation models. To illustrate its general applicability, we further deploy SPINO on real-world robotic vision systems for both outdoor and indoor environments. To foster future research, we make the code and trained models publicly available at http://spino.cs.uni-freiburg.de.

Optical flow estimation is very challenging in situations with transparent or occluded objects. In this work, we address these challenges at the task level by introducing Amodal Optical Flow, which integrates optical flow with amodal perception. Instead of only representing the visible regions, we define amodal optical flow as a multi-layered pixel-level motion field that encompasses both visible and occluded regions of the scene. To facilitate research on this new task, we extend the AmodalSynthDrive dataset to include pixel-level labels for amodal optical flow estimation. We present several strong baselines, along with the Amodal Flow Quality metric to quantify the performance in an interpretable manner. Furthermore, we propose the novel AmodalFlowNet as an initial step toward addressing this task. AmodalFlowNet consists of a transformer-based cost-volume encoder paired with a recurrent transformer decoder which facilitates recurrent hierarchical feature propagation and amodal semantic grounding. We demonstrate the tractability of amodal optical flow in extensive experiments and show its utility for downstream tasks such as panoptic tracking. We make the dataset, code, and trained models publicly available at http://amodal-flow.cs.uni-freiburg.de.

Localization is paramount for autonomous robots. While camera and LiDAR-based approaches have been extensively investigated, they are affected by adverse illumination and weather conditions. Therefore, radar sensors have recently gained attention due to their intrinsic robustness to such conditions. In this paper, we propose RaLF, a novel deep neural network-based approach for localizing radar scans in a LiDAR map of the environment, by jointly learning to address both place recognition and metric localization. RaLF is composed of radar and LiDAR feature encoders, a place recognition head that generates global descriptors, and a metric localization head that predicts the 3-DoF transformation between the radar scan and the map. We tackle the place recognition task by learning a shared embedding space between the two modalities via cross-modal metric learning. Additionally, we perform metric localization by predicting pixel-level flow vectors that align the query radar scan with the LiDAR map. We extensively evaluate our approach on multiple real-world driving datasets and show that RaLF achieves state-of-the-art performance for both place recognition and metric localization. Moreover, we demonstrate that our approach can effectively generalize to different cities and sensor setups than the ones used during training. We make the code and trained models publicly available at http://ralf.cs.uni-freiburg.de.

Teaching robots new skills quickly and conveniently is crucial for the broader adoption of robotic systems. In this work, we address the problem of one-shot imitation from a single human demonstration, given by an RGB-D video recording. We propose a two-stage process. In the first stage we extract the demonstration trajectory offline. This entails segmenting manipulated objects and determining their relative motion in relation to secondary objects such as containers. In the online trajectory generation stage, we first re-detect all objects, then warp the demonstration trajectory to the current scene and execute it on the robot. To complete these steps, our method leverages several ancillary models, including those for segmentation, relative object pose estimation, and grasp prediction. We systematically evaluate different combinations of correspondence and re-detection methods to validate our design decision across a diverse range of tasks. Specifically, we collect and quantitatively test on demonstrations of ten different tasks including pick-and-place tasks as well as articulated object manipulation. Finally, we perform extensive evaluations on a real robot system to demonstrate the effectiveness and utility of our approach in real-world scenarios. We make the code publicly available at http://ditto.cs.uni-freiburg.de.

Multi-task perception networks hold great potential as they can improve performance and computational efficiency compared to their single-task counterparts, facilitating online deployment. However, current multi-task architectures in point cloud perception combine multiple task-specific point cloud representations, each requiring a separate feature encoder, making the network significantly large and slow. In this work, we propose PAttFormer, an efficient multi-task learning architecture for joint semantic segmentation and object detection in point clouds, only relying on a point-based representation. The network builds on transformer-based feature encoders using neighborhood attention and grid-pooling, complemented with a query-based detection decoder using a novel 3D deformable-attention detection head topology. Unlike other LiDAR-based multi-task architectures, our proposed PAttFormer does not require separate feature encoders for multiple task-specific point cloud representations, resulting in a network that is 3× smaller and 1.4× faster while achieving competitive performance on the nuScenes and KITTI benchmarks for autonomous driving perception. We perform extensive evaluations that show substantial improvement from multi-task learning, achieving +1.7% in mIoU for LiDAR semantic segmentation and +1.7% in mAP for 3D object detection on the nuScenes benchmark compared to the single-task models.

Sample efficient learning of manipulation skills poses a major challenge in robotics. While recent approaches demonstrate impressive advances in the type of task that can be addressed and the sensing modalities that can be incorporated, they still require large amounts of training data. Especially with regard to learning actions on robots in the real world, this poses a major problem due to the high costs associated with both demonstrations and real-world robot interactions. To address this challenge, we introduce BOpt-GMM, a hybrid approach that combines imitation learning with own experience collection. We first learn a skill model as a dynamical system encoded in a Gaussian Mixture Model from a few demonstrations. We then improve this model with Bayesian optimization building on a small number of autonomous skill executions in a sparse reward setting. We demonstrate the sample efficiency of our approach on multiple complex manipulation skills in both simulations and real-world experiments. Furthermore, we make the code and pre-trained models publicly available at http://bopt-gmm.cs.uni-freiburg.de.

Semantic scene segmentation from a bird’s-eye-view (BEV) perspective plays a crucial role in facilitating planning and decision-making for mobile robots. Although recent vision-only methods have demonstrated notable advancements in performance, they often struggle under adverse illumination conditions such as rain or nighttime. While active sensors offer a solution to this challenge, the prohibitively high cost of LiDARs remains a limiting factor. Fusing camera data with automotive radars poses a more inexpensive alternative but has received less attention in prior research. In this work, we aim to advance this promising avenue by introducing BEVCar, a novel approach for joint BEV object and map segmentation. The core novelty of our approach lies in first learning a point-based encoding of raw radar data, which is then leveraged to efficiently initialize the lifting of image features into the BEV space. We perform extensive experiments on the nuScenes dataset and demonstrate that BEVCar outperforms the current state of the art. Moreover, we show that incorporating radar information significantly enhances robustness in challenging environmental conditions and improves segmentation performance for distant objects. To foster future research, we provide the weather split of the nuScenes dataset used in our experiments, along with our code and trained models at http://bevcar.cs.uni-freiburg.de.

Machine learning (ML) has significantly enhanced the abilities of robots, enabling them to perform a wide range of tasks in human environments and adapt to our uncertain real world. Recent works in various ML domains have highlighted the importance of accounting for fairness to ensure that these algorithms do not reproduce human biases and consequently lead to discriminatory outcomes. With robot learning systems increasingly performing more and more tasks in our everyday lives, it is crucial to understand the influence of such biases to prevent unintended behavior toward certain groups of people. In this work, we present the first survey on fairness in robot learning from an interdisciplinary perspective spanning technical, ethical, and legal challenges. We propose a taxonomy for sources of bias and the resulting types of discrimination due to them. Using examples from different robot learning domains, we examine scenarios of unfair outcomes and strategies to mitigate them. We present early advances in the field by covering different fairness definitions, ethical and legal considerations, and methods for fair robot learning. With this work, we aim to pave the road for groundbreaking developments in fair robot learning.

Lane graph estimation is an essential and highly challenging task in automated driving and HD map learning. Existing methods using either onboard or aerial imagery struggle with complex lane topologies, out-oj-distribution scenar-ios, or significant occlusions in the image space. Moreover, merging overlapping lane graphs to obtain consistent large-scale graphs remains difficult. To overcome these challenges, we propose a novel bottom-up approach to lane graph esti-mation from aerial imagery that aggregates multiple over-lapping graphs into a single consistent graph. Due to its modular design, our method allows us to address two complementary tasks: predicting ego-respective successor lane graphs from arbitrary vehicle positions using a graph neural network and aggregating these predictions into a consistent global lane graph. Extensive experiments on a large-scale lane graph dataset demonstrate that our approach yields highly accurate lane graphs, even in regions with severe occlusions. The presented approach to graph aggregation proves to eliminate inconsistent predictions while increasing the overall graph quality. We make our large-scale urban lane graph dataset and code publicly available at http://urban1anegraph.cs.uni-freiburg.de

Bird's-Eye-View (BEV) semantic maps have become an essential component of automated driving pipelines due to the rich representation they provide for decision-making tasks. However, existing approaches for generating these maps still follow a fully supervised training paradigm and hence rely on large amounts of annotated BEV data. In this work, we address this limitation by proposing the first self-supervised approach for generating a BEV semantic map using a single monocular image from the frontal view (FV). During training, we overcome the need for BEV ground truth annotations by leveraging the more easily available FV semantic annotations of video sequences. Thus, we propose the SkyEye architecture that learns based on two modes of self-supervision, namely, implicit supervision and explicit supervision. Implicit supervision trains the model by enforcing spatial consistency of the scene over time based on FV semantic sequences, while explicit supervision exploits BEV pseudolabels generated from FV semantic annotations and self-supervised depth estimates. Extensive evaluations on the KITTI-360 dataset demonstrate that our self-supervised approach performs on par with the state-of-the-art fully supervised methods and achieves competitive results using only 1 % of direct supervision in BEV compared to fully supervised approaches. Finally, we publicly release both our code and the BEV datasets generated from the KITTI-360 and Waymo datasets.

We present CARTO, a novel approach for reconstructing multiple articulated objects from a single stereo RGB observation. We use implicit object-centric representations and learn a single geometry and articulation decoder for multiple object categories. Despite training on multiple categories, our decoder achieves a comparable reconstruction accuracy to methods that train bespoke decoders separately for each category. Combined with our stereo image encoder we infer the 3D shape, 6D pose, size, joint type, and the joint state of multiple unknown objects in a single forward pass. Our method achieves a 20.4% absolute improvement in mAP 3D IOU50 for novel instances when compared to a two-stage pipeline. Inference time is fast and can run on a NVIDIA TITAN XP GPU at 1 HZ for eight or less objects present. While only trained on simulated data, CARTO transfers to real-world object instances. Code and evaluation data is available at: carto. cs. uni - freiburg. de

Uncertainty estimation is crucial in safety-critical settings such as automated driving as it provides valuable information for several downstream tasks including high-level decision making and path planning. In this work, we propose EvCenterNet, a novel uncertainty-aware 2D object detection framework using evidential learning to directly estimate both classification and regression uncertainties. To employ evidential learning for object detection, we devise a combination of evidential and focal loss functions for the sparse heatmap inputs. We introduce class-balanced weighting for regression and heatmap prediction to tackle the class imbalance encountered by evidential learning. Moreover, we propose a learning scheme to actively utilize the predicted heatmap uncertainties to improve the detection performance by focusing on the most uncertain points. We train our model on the KITTI dataset and evaluate it on challenging out-of-distribution datasets including BDD100K and nuImages. Our experiments demonstrate that our approach improves the precision and minimizes the execution time loss in relation to the base model.

Abstract The success of deep learning in recent years has lead to a rising demand for neural network architecture engineering. As a consequence, neural architecture search (NAS), which aims at automatically designing neural network architectures in a data-driven manner rather than manually, has evolved as a popular field of research. With the advent of weight sharing strategies across architectures, NAS has become applicable to a much wider range of problems. In particular, there are now many publications for dense prediction tasks in computer vision that require pixel-level predictions, such as semantic segmentation or object detection. These tasks come with novel challenges, such as higher memory footprints due to high-resolution data, learning multi-scale representations, longer training times, and more complex and larger neural architectures. In this manuscript, we provide an overview of NAS for dense prediction tasks by elaborating on these novel challenges and surveying ways to address them to ease future research and application of existing methods to novel problems.

Despite its importance in both industrial and service robotics, mobile manipulation remains a significant challenge as it requires seamless integration of end-effector trajectory generation with navigation skills as well as reasoning over long-horizons. Existing methods struggle to control the large configuration space and to navigate dynamic and unknown environments. In the previous work, we proposed to decompose mobile manipulation tasks into a simplified motion generator for the end-effector in task space and a trained reinforcement learning agent for the mobile base to account for the kinematic feasibility of the motion. In this work, we introduce Neural Navigation for Mobile Manipulation (N$^{2}$M$^{2}$), which extends this decomposition to complex obstacle environments, extends the agent's control to the torso joint and the norm of the end-effector motion velocities, uses a more general reward function and, thereby, enables robots to tackle a much broader range of tasks in real-world settings. The resulting approach can perform unseen, long-horizon tasks in unexplored environments while instantly reacting to dynamic obstacles and environmental changes. At the same time, it provides a simple way to define new mobile manipulation tasks. We demonstrate the capabilities of our proposed approach in extensive simulation and real-world experiments on multiple kinematically diverse mobile manipulators.

Humans have the remarkable ability to perceive objects as a whole, even when parts of them are occluded. This ability of amodal perception forms the basis of our perceptual and cognitive understanding of our world. To enable robots to reason with this capability, we formulate and propose a novel task that we name amodal panoptic segmentation. The goal of this task is to simultaneously predict the pixel-wise semantic segmentation labels of the visible regions of stuff classes and the instance segmentation labels of both the visible and occluded regions of thing classes. To facilitate research on this new task, we extend two established benchmark datasets with pixel-level amodal panoptic segmentation labels that we make publicly available as KITTI-360-APS and BDD100K-APS. We present several strong baselines, along with the amodal panoptic quality (APQ) and amodal parsing coverage (APC) metrics to quantify the performance in an interpretable manner. Furthermore, we propose the novel amodal panoptic segmentation network (APSNet), as a first step towards addressing this task by explicitly modeling the complex relationships between the occluders and occludes. Extensive experimental evaluations demonstrate that APSNet achieves state-of-the-art performance on both benchmarks and more importantly exemplifies the utility of amodal recognition. The datasets are available at http://amodal-panoptic.cs.uni-freiburg.de.

$A$core challenge for an autonomous agent acting in the real world is to adapt its repertoire of skills to cope with its noisy perception and dynamics. To scale learning of skills to long-horizon tasks, robots should be able to learn and later refine their skills in a structured manner through trajectories rather than making instantaneous decisions individually at each time step. To this end, we propose the Soft Actor- Critic Gaussian Mixture Model (SAC-GMM), a novel hybrid approach that learns robot skills through a dynamical system and adapts the learned skills in their own trajectory distribution space through interactions with the environment. Our approach combines classical robotics techniques of learning from demonstration with the deep reinforcement learning framework and exploits their complementary nature. We show that our method utilizes sensors solely available during the execution of preliminarily learned skills to extract relevant features that lead to faster skill refinement. Extensive evaluations in both simulation and real-world environments demonstrate the effectiveness of our method in refining robot skills by leveraging physical interactions, high-dimensional sensory data, and sparse task completion rewards. Videos, code, and pre-trained models are available at http://sac-gmm.cs.uni-freiburg.de.

Existing Deep Learning (DL) frameworks typically do not provide ready-to-use solutions for robotics, where very specific learning, reasoning, and embodiment problems exist. Their relatively steep learning curve and the different methodologies employed by DL compared to traditional approaches, along with the high complexity of DL models, which often leads to the need of employing specialized hardware accelerators, further increase the effort and cost needed to employ DL models in robotics. Also, most of the existing DL methods follow a static inference paradigm, as inherited by the traditional computer vision pipelines, ignoring active perception, which can be employed to actively interact with the environment in order to increase perception accuracy. In this paper, we present the Open Deep Learning Toolkit for Robotics (OpenDR). OpenDR aims at developing an open, non-proprietary, efficient, and modular toolkit that can be easily used by robotics companies and research institutions to efficiently develop and deploy AI and cognition technologies to robotics applications, providing a solid step towards addressing the aforementioned challenges. We also detail the design choices, along with an abstract interface that was created to overcome these challenges. This interface can describe various robotic tasks, spanning beyond traditional DL cognition and inference, as known by existing frameworks, incorporating openness, homogeneity and robotics-oriented perception e.g., through active perception, as its core design principles.

Loop closure detection is an essential component of simultaneous localization and mapping (SLAM) systems, which reduces the drift accumulated over time. Over the years, several deep learning approaches have been proposed to address this task; however, their performance has been subpar compared to handcrafted techniques, especially while dealing with reverse loops. In this article, we introduce the novel loop closure detection network (LCDNet) that effectively detects loop closures in light detection and ranging (LiDAR) point clouds by simultaneously identifying previously visited places and estimating the six degrees of freedom relative transformation between the current scan and the map. LCDNet is composed of a shared encoder, a place recognition head that extracts global descriptors, and a relative pose head that estimates the transformation between two point clouds. We introduce a novel relative pose head based on the unbalanced optimal transport theory that we implement in a differentiable manner to allow for end-to-end training. Extensive evaluations of LCDNet on multiple real-world autonomous driving datasets show that our approach outperforms state-of-the-art loop closure detection and point cloud registration techniques by a large margin, especially while dealing with reverse loops. Moreover, we integrate our proposed loop closure detection approach into a LiDAR SLAM library to provide a complete mapping system and demonstrate the generalization ability using different sensor setup in an unseen city.

Panoptic segmentation of point clouds is a crucial task that enables autonomous vehicles to comprehend their vicinity using their highly accurate and reliable LiDAR sensors. Existing top–down approaches tackle this problem by either combining independent task-specific networks or translating methods from the image domain ignoring the intricacies of LiDAR data and thus often resulting in suboptimal performance. In this article, we present the novel top–down efficient LiDAR panoptic segmentation (EfficientLPS) architecture that addresses multiple challenges in segmenting LiDAR point clouds, including distance-dependent sparsity, severe occlusions, large scale-variations, and reprojection errors. EfficientLPS comprises of a novel shared backbone that encodes with strengthened geometric transformation modeling capacity and aggregates semantically rich range-aware multiscale features. It incorporates new scale-invariant semantic and instance segmentation heads along with the panoptic fusion module which is supervised by our proposed panoptic periphery loss function. Additionally, we formulate a regularized pseudolabeling framework to further improve the performance of EfficientLPS by training on unlabeled data. We benchmark our proposed model on two large-scale LiDAR datasets: nuScenes, for which we also provide ground truth annotations, and SemanticKITTI. Notably, EfficientLPS sets the new state-of-the-art on both these datasets.

Attributes of sound inherent to objects can provide valuable cues to learn rich representations for object detection and tracking. Furthermore, the co-occurrence of audiovisual events in videos can be exploited to localize objects over the image field by solely monitoring the sound in the environment. Thus far, this has only been feasible in scenarios where the camera is static and for single object detection. Moreover, the robustness of these methods has been limited as they primarily rely on RGB images which are highly susceptible to illumination and weather changes. In this work, we present the novel self-supervised MM-DistillNet framework consisting of multiple teachers that leverage diverse modalities including RGB, depth and thermal images, to simultaneously exploit complementary cues and distill knowledge into a single audio student network. We propose the new MTA loss function that facilitates the distillation of information from multimodal teachers in a self-supervised manner. Additionally, we propose a novel self-supervised pretext task for the audio student that enables us to not rely on labor-intensive manual annotations. We introduce a large-scale multimodal dataset with over 113,000 time-synchronized frames of RGB, depth, thermal, and audio modalities. Extensive experiments demonstrate that our approach outperforms state-of-the-art methods while being able to detect multiple objects using only sound during inference and even while moving.

Abstract Understanding the scene in which an autonomous robot operates is critical for its competent functioning. Such scene comprehension necessitates recognizing instances of traffic participants along with general scene semantics which can be effectively addressed by the panoptic segmentation task. In this paper, we introduce the Efficient Panoptic Segmentation (EfficientPS) architecture that consists of a shared backbone which efficiently encodes and fuses semantically rich multi-scale features. We incorporate a new semantic head that aggregates fine and contextual features coherently and a new variant of Mask R-CNN as the instance head. We also propose a novel panoptic fusion module that congruously integrates the output logits from both the heads of our EfficientPS architecture to yield the final panoptic segmentation output. Additionally, we introduce the KITTI panoptic segmentation dataset that contains panoptic annotations for the popularly challenging KITTI benchmark. Extensive evaluations on Cityscapes, KITTI, Mapillary Vistas and Indian Driving Dataset demonstrate that our proposed architecture consistently sets the new state-of-the-art on all these four benchmarks while being the most efficient and fast panoptic segmentation architecture to date.

Mobile robots operating in unknown urban environments encounter a wide range of complex terrains to which they must adapt their planned trajectory for safe and efficient navigation. Most existing approaches utilize supervised learning to classify terrains from either an exteroceptive or a proprioceptive sensor modality. However, this requires a tremendous amount of manual labeling effort for each newly encountered terrain as well as for variations of terrains caused by changing environmental conditions. In this article, we propose a novel terrain classification framework leveraging an unsupervised proprioceptive classifier that learns from vehicle-terrain interaction sounds to self-supervise an exteroceptive classifier for pixelwise semantic segmentation of images. To this end, we first learn a discriminative embedding space for vehicle-terrain interaction sounds from triplets of audio clips formed using visual features of the corresponding terrain patches and cluster the resulting embeddings. We subsequently use these clusters to label the visual terrain patches by projecting the traversed tracks of the robot into the camera images. Finally, we use the sparsely labeled images to train our semantic segmentation network in a weakly supervised manner. We present extensive quantitative and qualitative results that demonstrate that our proprioceptive terrain classifier exceeds the state-of-the-art among unsupervised methods and our self-supervised exteroceptive semantic segmentation model achieves a comparable performance to supervised learning with manually labeled data.

Indoor localization is one of the crucial enablers for deployment of service robots. Although several successful techniques for indoor localization have been proposed, the majority of them relies on maps generated from data gathered with the same sensor modality used for localization. Typically, tedious labor by experts is needed to acquire this data, thus limiting the readiness of the system as well as its ease of installation for inexperienced operators. In this paper, we propose a memory and computationally efficient monocular camera-based localization system that allows a robot to estimate its pose given an architectural floor plan. Our method employs a convolutional neural network to predict room layout edges from a single camera image and estimates the robot pose using a particle filter that matches the extracted edges to the given floor plan. We evaluate our localization system using multiple real-world experiments and demonstrate that it has the robustness and accuracy required for reliable indoor navigation.

Localization is an indispensable component of a robot's autonomy stack that enables it to determine where it is in the environment, essentially making it a precursor for any action execution or planning. Although convolutional neural networks have shown promising results for visual localization, they are still grossly outperformed by state-of-the-art local feature-based techniques. In this work, we propose VLocNet, a new convolutional neural network architecture for 6-DoF global pose regression and odometry estimation from consecutive monocular images. Our multitask model incorporates hard parameter sharing, thus being compact and enabling real-time inference, in addition to being end-to-end trainable. We propose a novel loss function that utilizes auxiliary learning to leverage relative pose information during training, thereby constraining the search space to obtain consistent pose estimates. We evaluate our proposed VLocNet on indoor as well as outdoor datasets and show that even our single task model exceeds the performance of state-of-the-art deep architectures for global localization, while achieving competitive performance for visual odometry estimation. Furthermore, we present extensive experimental evaluations utilizing our proposed Geometric Consistency Loss that show the effectiveness of multitask learning and demonstrate that our model is the first deep learning technique to be on par with, and in some cases outperforms state-of-the-art SIFT-based approaches.

Robust scene understanding of outdoor environments using passive optical sensors is a onerous and essential task for autonomous navigation. The problem is heavily characterized by changing environmental conditions throughout the day and across seasons. Robots should be equipped with models that are impervious to these factors in order to be operable and more importantly to ensure safety in the real-world. In this paper, we propose a novel semantic segmentation architecture and the convoluted mixture of deep experts (CMoDE) fusion technique that enables a multi-stream deep neural network to learn features from complementary modalities and spectra, each of which are specialized in a subset of the input space. Our model adaptively weighs class-specific features of expert networks based on the scene condition and further learns fused representations to yield robust segmentation. We present results from experimentation on three publicly available datasets that contain diverse conditions including rain, summer, winter, dusk, fall, night and sunset, and show that our approach exceeds the state-of-the-art. In addition, we evaluate the performance of autonomously traversing several kilometres of a forested environment using only the segmentation for perception.

Interpreting the semantics and motion of objects are prerequisites for autonomous robots that enable them to reason and operate in dynamic real-world environments. Existing approaches that tackle the problem of semantic motion segmentation consist of long multistage pipelines and typically require several seconds to process each frame. In this paper, we present a novel convolutional neural network architecture that learns to predict both the object label and motion status of each pixel in an image. Given a pair of consecutive images, the network learns to fuse features from self-generated optical flow maps and semantic segmentation kernels to yield pixel-wise semantic motion labels. We also introduce the Cityscapes-Motion dataset which contains over 2,900 manually annotated semantic motion labels, which is the largest dataset of its kind so far. We demonstrate that our network outperforms existing approaches achieving state-of-the-art performance on the KITTI dataset, as well as in the more challenging Cityscapes-Motion dataset while being substantially faster than existing techniques.

Hand-Drawn sketches are natural means by which abstract descriptions of environments can be provided. They represent weak prior information about the scene, thereby enabling a robot to perform autonomous navigation and exploration when a full metrical description of the environment is not available beforehand. In this paper, we present an extensive evaluation of our navigation system that uses a sketch interface to allow the operator of a robot to draw a rough map of an indoor environment as well as a desired trajectory for the robot to follow. We employ a theoretical framework for sketch interpretation, in which associations between the sketch and the real world are modeled as local deformations of a suitable metric manifold. We investigate the effectiveness of our system and present empirical results from a set of experiments in real-world scenarios, focusing both on the navigation capabilities and the usability of the interface.

This paper addresses the problem of human body part segmentation in conventional RGB images, which has several applications in robotics, such as learning from demonstration and human-robot handovers. The proposed solution is based on Convolutional Neural Networks (CNNs). We present a network architecture that assigns each pixel to one of a predefined set of human body part classes, such as head, torso, arms, legs. After initializing weights with a very deep convolutional network for image classification, the network can be trained end-to-end and yields precise class predictions at the original input resolution. Our architecture particularly improves on over-fitting issues in the up-convolutional part of the network. Relying only on RGB rather than RGB-D images also allows us to apply the approach outdoors. The network achieves state-of-the-art performance on the PASCAL Parts dataset. Moreover, we introduce two new part segmentation datasets, the Freiburg sitting people dataset and the Freiburg people in disaster dataset. We also present results obtained with a ground robot and an unmanned aerial vehicle.