Papers

Motivated by recent challenges in the deployment of robots into customer-facing roles within retail, this work introduces a study of customer activity in physical stores as a step toward autonomous understanding of shopper intent. We introduce an algorithm that computes shoppers' “shelf visits” – capturing their browsing behavior in the store. Shelf visits are extracted from trajectories obtained via machine vision-based 3D tracking and overhead cameras. We perform two independent calibrations of the shelf visit algorithm, using distinct sets of trajectories (consisting of 8138 and 15129 trajectories), collected in different stores and labeled by human reviewers. The calibrated models are then evaluated on trajectories held out of the calibration process both from the same store on which calibration was performed and from the other store. An analysis of the results shows that the algorithm can recognize customers' browsing activity when evaluated in an environment different from the one on which calibration was performed. We then use the model to analyze the customers' “browsing patterns” on a large set of trajectories and their relation to actual purchases in the stores. Finally, we discuss how shelf browsing information could be used for retail planning and in the domain of human-robot interaction scenarios.

How do children respond to repeated robot errors? While prior research has examined adult reactions to successive robot errors, children's responses remain largely unexplored. In this study, we explore children's reactions to robot social errors and performance errors. For the latter, this study reproduces the successive robot failure paradigm of Liu et al. with child participants (N=59, ages 8-10) to examine how young users respond to repeated robot conversational errors. Participants interacted with a robot that failed to understand their prompts three times in succession, with their behavioral responses video-recorded and analyzed. We found both similarities and differences compared to adult responses from the original study. Like adults, children adjusted their prompts, modified their verbal tone, and exhibited increasingly emotional non-verbal responses throughout successive errors. However, children demonstrated more disengagement behaviors, including temporarily ignoring the robot or actively seeking an adult. Errors did not affect participants' perception of the robot, suggesting more flexible conversational expectations in children. These findings inform the design of more effective and developmentally appropriate human-robot interaction systems for young users.

Deformable scenes violate the rigidity assumptions underpinning classical visual–inertial odometry (VIO), often leading to over-fitting to local non-rigid motion or to severe camera pose drift when deformation dominates visual parallax. In this paper, we introduce DefVINS, the first visual-inertial odometry pipeline designed to operate in deformable environments. Our approach models the odometry state by decomposing it into a rigid, IMU-anchored component and a non-rigid scene warp represented by an embedded deformation graph. As a second contribution, we present VIMandala, the first benchmark containing real images and ground-truth camera poses for visual-inertial odometry in deformable scenes. In addition, we augment the synthetic Drunkard's benchmark with simulated inertial measurements to further evaluate our pipeline under controlled conditions. We also provide an observability analysis of the visual-inertial deformable odometry problem, characterizing how inertial measurements constrain camera motion and render otherwise unobservable modes identifiable in the presence of deformation. This analysis motivates the use of IMU anchoring and leads to a conditioning-based activation strategy that avoids ill-posed updates under poor excitation. Experimental results on both the synthetic Drunkard's and our real VIMandala benchmarks show that DefVINS outperforms rigid visual–inertial and non-rigid visual odometry baselines. Our source code and data will be released upon acceptance.

Many multi-agent interaction scenarios can be naturally modeled as noncooperative games, where each agent's decisions depend on others' future actions. However, deploying game-theoretic planners for autonomous decision-making requires a specification of all agents' objectives. To circumvent this practical difficulty, recent work develops maximum likelihood techniques for solving inverse games that can identify unknown agent objectives from interaction data. Unfortunately, these methods only infer point estimates and do not quantify estimator uncertainty; correspondingly, downstream planning decisions can overconfidently commit to unsafe actions. We present an approximate Bayesian inference approach for solving the inverse game problem, which can incorporate observation data from multiple modalities and be used to generate samples from the Bayesian posterior over the hidden agent objectives given limited sensor observations in real time. Concretely, the proposed Bayesian inverse game framework trains a structured variational autoencoder with an embedded differentiable Nash game solver on interaction datasets and does not require labels of agents' true objectives. Extensive experiments show that our framework successfully learns prior and posterior distributions, improves inference quality over maximum likelihood estimation-based inverse game approaches, and enables safer downstream decision-making without sacrificing efficiency. When trajectory information is uninformative or unavailable, multimodal inference further reduces uncertainty by exploiting additional observation modalities.

A well-designed reward is critical for effective reinforcement learning-based policy improvement. In real-world robotics, obtaining such rewards typically requires either labor-intensive human labeling or brittle, handcrafted objectives. Vision-language models (VLMs) have shown promise as automatic reward models, yet their effectiveness on real robot tasks is poorly understood. In this work, we aim to close this gap by introducing (1) RoboReward, a robotics reward dataset and benchmark built on large-scale real-robot corpora from Open X-Embodiment (OXE) and RoboArena, and (2) vision-language reward models trained on this dataset (RoboReward 4B/8B). Because OXE is success-heavy and lacks failure examples, we propose a negative examples data augmentation pipeline that generates calibrated negative and near-misses via counterfactual relabeling of successful episodes and temporal clipping to create partial-progress outcomes from the same videos. Using this framework, we build a large training and evaluation dataset spanning diverse tasks and embodiments to test whether state-of-the-art VLMs can reliably provide rewards for robot learning. Our evaluation of open and proprietary VLMs finds that no model excels across tasks, highlighting substantial room for improvement. We then train general-purpose 4B- and 8B-parameter models that outperform much larger VLMs in assigning rewards for short-horizon robotic tasks. Finally, we deploy the 8B model in real-robot reinforcement learning and find that it improves policy learning over Gemini Robotics-ER 1.5 while narrowing the gap to RL training with human-provided rewards. We release the full dataset, trained reward models, and evaluation suite on our website to advance the development of general-purpose reward models in robotics: https://crfm.stanford.edu/helm/robo-reward-bench (project website).

An algorithm for 3D terrain-following area coverage path planning is presented. Multiple adjacent paths are generated that are (i) locally apart from each other by a distance equal to the working width of a machinery, while (ii) simultaneously floating at a projection distance equal to a specific working height above the terrain. The complexities of the algorithm in comparison to its 2D equivalent are highlighted. These include uniformly spaced elevation data generation using an Inverse Distance Weighting-approach and a local search. Area coverage path planning results for real-world 3D data within an agricultural context are presented to validate the algorithm.

Reinforcement learning (RL) has strong potential in robotics, but exploration-based training complicates safe deployment on large-scale robots. For such applications, this paper proposes a novel hierarchical goal-reaching framework that integrates stereo visual pose estimation, constrained RL-based motion planning, actuator-level robust adaptive control (RAC), and supervisory safe-return logic. Stereo visual localization is used as the real-time pose-estimation interface with loop closing, map fusion, and relocalization. The RL planner generates smooth, feasible goal-reaching references using a problem-specific reward structure and motion constraints that promote goal progress, reduce oscillations, preserve vision-consistent smoothness, and respect the mechanical limits of a heavy skid-steered robot. At the actuation layer, a scaled conjugate-gradient (SCG)-trained deep neural network (DNN) approximates a quasi-static actuator feedforward map from wheel-speed data to nominal control input. This feedforward map is combined with a logarithmic-barrier-based RAC to compensate for residual modeling errors, slip-induced disturbances, and bounded mismatch between the nominal map and real actuator response. For the actuator-level wheel-tracking subsystem, uniformly ultimately bounded tracking with exponential convergence to a disturbance-dependent residual set is established under bounded uncertainty. A logarithmic safety supervisor monitors execution, detects unsafe operating conditions, including faults and localization inconsistencies, and switches the robot to safe-return mode. Experiments on a 6000 kg robot over asphalt and loose-soil terrain demonstrate approximately 3–4 cm final-position root mean square error (RMSE), accurate tracking of RL-generated commands, improved actuator-level performance over two RAC baselines, and successful autonomous recovery after fault injection.

A large-scale mobile robot (LSMR) is a high-order multibody system that often operates on loose, unconsolidated terrain, which reduces traction. This paper presents a comprehensive navigation and control framework for an LSMR that ensures stability and safety-defined performance, delivering robust operation on slip-prone terrain by jointly leveraging high-performance techniques. The proposed architecture comprises four main modules: (1) a visual pose-estimation module that fuses onboard sensors and stereo cameras to provide an accurate, low-latency robot pose, (2) a high-level nonlinear model predictive control that updates the wheel motion commands to correct robot drift from the robot reference pose on slip-prone terrain, (3) a low-level deep neural network control policy that approximates the complex behavior of the wheel-driven actuation mechanism in LSMRs, augmented with robust adaptive control to handle out-of-distribution disturbances, ensuring that the wheels accurately track the updated commands issued by high-level control module, and (4) a logarithmic safety module to monitor the entire robot stack and guarantees safe operation. The proposed low-level control framework guarantees uniform exponential stability of the actuation subsystem, while the safety module ensures the whole system-level safety during operation. Comparative experiments on a 6,000 kg LSMR actuated by two complex electro-hydrostatic drives, while synchronizing modules operating at different frequencies.

Multi-robot systems are widely used for coverage tasks that require efficient coordination across large environments. In Multi-Robot Coverage Path Planning (MCPP), the objective is typically to minimize the makespan by generating non-overlapping paths for full-area coverage. However, most existing methods assume uniform importance across regions, limiting their effectiveness in scenarios where some zones require faster attention. We introduce the Priority-Aware MCPP (PA-MCPP) problem, where a subset of the environment is designated as prioritized zones with associated weights. The goal is to minimize, in lexicographic order, the total priority-weighted latency of zone coverage and the overall makespan. To address this, we propose a scalable two-phase framework combining (1) greedy zone assignment with local search, spanning-tree-based path planning, and (2) Steiner-tree-guided residual coverage. Experiments across diverse scenarios demonstrate that our method significantly reduces priority-weighted latency compared to standard MCPP baselines, while maintaining competitive makespan. Sensitivity analyses further show that the method scales well with the number of robots and that zone coverage behavior can be effectively controlled by adjusting priority weights.

This paper presents an end-to-end LLM-based agentic exploration system for an indoor shopping task, evaluated in both Gazebo simulation and a corresponding real-world corridor layout. The robot incrementally builds a lightweight semantic map by detecting signboards at junctions and storing direction-to-POI relations together with estimated junction poses, while AprilTags provide repeatable anchors for approach and alignment. Given a natural-language shopping request, an LLM produces a constrained discrete action at each junction (direction and whether to enter a store), and a ROS finite-state main controller executes the decision by gating modular motion primitives, including local-costmap-based obstacle avoidance, AprilTag approaching, store entry, and grasping. Qualitative results show that the integrated stack can perform end-to-end task execution from user instruction to multi-store navigation and object retrieval, while remaining modular and debuggable through its text-based map and logged decision history.

This paper presents a decentralized control framework for distribution matching in multi-agent systems (MAS), where agents collectively achieve a prescribed terminal spatial distribution. The problem is formulated using optimal transport (Wasserstein distance), which provides a principled measure of distributional discrepancy and serves as the basis for the control design. To avoid solving the global optimal transport problem directly, the distribution-matching objective is reformulated into a tractable per-agent decision process, enabling each agent to identify its desired terminal locations using only locally available information. A sequential weight-update rule is introduced to construct feasible local transport plans, and a memory-based correction mechanism is incorporated to maintain reliable operation under intermittent and range-limited communication. Convergence guarantees are established, showing cycle-wise improvement of a surrogate transport cost under both linear and nonlinear agent dynamics. Simulation results demonstrate that the proposed framework achieves effective and scalable distribution matching while operating fully in a decentralized manner.

Many problems in robotics involve both continuous and discrete components, and modeling them together for estimation tasks has been a long standing and difficult problem. Hybrid Factor Graphs give us a mathematical framework to model these types of problems, however existing approaches for solving them are based on approximations. In this work, we propose a new framework for hybrid factor graphs along with a novel variable elimination algorithm to produce a hybrid Bayes network, which can be used for exact Maximum A Posteriori estimation and marginalization over both sets of variables. Our approach first develops a novel hybrid Gaussian factor which can connect to both discrete and continuous variables, and a hybrid conditional which can represent multiple continuous hypotheses conditioned on the discrete variables. Using these representations, we derive the process of hybrid variable elimination under the Conditional Linear Gaussian scheme, giving us exact posteriors as a hybrid Bayes network. To bound the number of discrete hypotheses, we use a tree-structured representation of the factors coupled with a simple pruning and probabilistic assignment scheme, which allows for tractable inference. We demonstrate the applicability of our framework on a large scale SLAM dataset and a real world pose graph optimization problem, both with ambiguous measurements which require discrete choices to be made for the most likely measurements. Our demonstrated results showcase the accuracy, generality, and simplicity of our hybrid factor graph framework.

Adapting to unforeseen novelties in open-world environments remains a major challenge for autonomous systems. While hybrid planning and reinforcement learning (RL) approaches show promise, they often suffer from sample inefficiency, slow adaptation, and catastrophic forgetting. We present a neuro-symbolic framework integrating hierarchical abstractions, task and motion planning (TAMP), and reinforcement learning to enable rapid adaptation in robotics. Our architecture combines symbolic goal-oriented learning and world model-based exploration to facilitate rapid adaptation to environmental changes. Validated in robotic manipulation and autonomous driving, our approach achieves faster convergence, improved sample efficiency, and superior robustness over state-of-the-art hybrid methods, demonstrating its potential for real-world deployment.

Space debris is an ever-increasing problem in space travel. There are already many old, no longer functional spacecraft and debris orbiting the earth, which endanger both the safe operation of satellites and space travel. Small nano-satellite swarms can address this problem by autonomously de-orbiting debris safely into the Earth's atmosphere. This work builds on the recent advances of autonomous agents deployed in resource-constrained platforms and shows a first simplified approach how such intelligent and autonomous nano-satellite swarms can be realized. We implement our autonomous agent software on wireless microcontrollers and perform experiments on a specialized test-bed to show the feasibility and overall energy efficiency of our approach.

In domestic environments, robots require a comprehensive understanding of their surroundings to interact effectively and intuitively with untrained humans. In this paper, we propose DVEFormer - an efficient RGB-D Transformer-based approach that predicts dense text-aligned visual embeddings (DVE) via knowledge distillation. Instead of directly performing classical semantic segmentation with fixed predefined classes, our method uses teacher embeddings from Alpha-CLIP to guide our efficient student model DVEFormer in learning fine-grained pixel-wise embeddings. While this approach still enables classical semantic segmentation, e.g., via linear probing, it further enables flexible text-based querying and other applications, such as creating comprehensive 3D maps. Evaluations on common indoor datasets demonstrate that our approach achieves competitive performance while meeting real-time requirements, operating at 26.3 FPS for the full model and 77.0 FPS for a smaller variant on an NVIDIA Jetson AGX Orin. Additionally, we show qualitative results that highlight the effectiveness and possible use cases in real-world applications. Overall, our method serves as a drop-in replacement for traditional segmentation approaches while enabling flexible natural-language querying and seamless integration into 3D mapping pipelines for mobile robotics.

The food packaging industry goes through changes in food items and their weights quite rapidly. These items range from easy-to-pick, single-piece food items to flexible, long and cluttered ones. We propose a replaceable bit-based gripper system to tackle the challenge of weight-based handling of cluttered food items. The gripper features specialized food attachments(bits) that enhance its grasping capabilities, and a belt replacement system allows switching between different food items during packaging operations. It offers a wide range of control options, enabling it to grasp and drop specific weights of granular, cluttered, and entangled foods. We specifically designed bits for two flexible food items that differ in shape: ikura(salmon roe) and spaghetti. They represent the challenging categories of sticky, granular food and long, sticky, cluttered food, respectively. The gripper successfully picked up both spaghetti and ikura and demonstrated weight-specific dropping of these items with an accuracy over 80% and 95% respectively. The gripper system also exhibited quick switching between different bits, leading to the handling of a large range of food items.

Autonomous vehicles and wheeled robots are widely used in many applications in both indoor and outdoor settings. In practical situations with limited GNSS signals or degraded lighting conditions, the navigation solution may rely only on inertial sensors and as result drift in time due to errors in the inertial measurement. In this work, we propose WiCHINS, a wheeled and chassis inertial navigation system by combining wheel-mounted-inertial sensors with a chassis-mounted inertial sensor for accurate pure inertial navigation. To that end, we derive a three-stage framework, each with a dedicated extended Kalman filter. This framework utilizes the benefits of each location (wheel/body) during the estimation process. To evaluate our proposed approach, we employed a dataset with five inertial measurement units with a total recording time of 228.6 minutes. We compare our approach with four other inertial baselines and demonstrate an average position error of 11.4m, which is $2.4\%$ of the average traveled distance, using two wheels and one body inertial measurement units. As a consequence, our proposed method enables robust navigation in challenging environments and helps bridge the pure-inertial performance gap.

In vehicle production factories, the vehicle painting process employs multiple robotic arms to simultaneously apply paint to car bodies advancing along a conveyor line. Designing paint paths for these robotic arms, which involves assigning car body areas to arms and determining paint sequences for each arm, remains a time-consuming manual task for engineers, indicating the demand for automation and design time reduction. The unique constraints of the painting process hinder the direct application of conventional robotic path planning techniques, such as those used in welding. Therefore, this paper formulates the design of paint paths as a hierarchical optimization problem, where the upper-layer subproblem resembles a vehicle routing problem (VRP), and the lower-layer subproblem involves detailed path planning. This approach allows the use of different optimization algorithms at each layer, and permits flexible handling of constraints specific to the vehicle painting process through the design of variable representation, constraints, repair operators, and an initialization process at the upper and lower layers. Experiments with three commercially available vehicle models demonstrated that the proposed method can automatically design paths that satisfy all constraints for vehicle painting with quality comparable to those created manually by engineers.

Perching allows unmanned aerial vehicles (UAVs) to reduce energy consumption, remain anchored for surface sampling operations, or stably survey their surroundings. Previous efforts for perching on vertical surfaces have predominantly focused on lightweight mechanical design solutions with relatively scant system-level integration. Furthermore, perching strategies for vertical surfaces commonly require high-speed, aggressive landing operations that are dangerous for a surveyor drone with sensitive electronics onboard. This work presents the preliminary investigation of a perching approach suitable for larger drones that both gently perches on vertical tree trunks and reacts and recovers from perch failures. The system in this work, called SLAP, consists of vision-based perch site detector, an IMU (inertial-measurement-unit)-based perch failure detector, an attitude controller for soft perching, an optical close-range detection system, and a fast active elastic gripper with microspines made from commercially-available slapbands. We validated this approach on a modified 1.2 kg commercial quadrotor with component and system analysis. Initial human-in-the-loop autonomous indoor flight experiments achieved a 75% perch success rate on a real oak tree segment across 20 flights, and 100% perch failure recovery across 2 flights with induced failures.

This paper addresses the critical bottleneck of infrared (IR) data scarcity in Printed Circuit Board (PCB) defect detection by proposing a cross-modal data augmentation framework integrating CycleGAN and YOLOv8. Unlike conventional methods relying on paired supervision, we leverage CycleGAN to perform unpaired image-to-image translation, mapping abundant visible-light PCB images into the infrared domain. This generative process synthesizes high-fidelity pseudo-IR samples that preserve the structural semantics of defects while accurately simulating thermal distribution patterns. Subsequently, we construct a heterogeneous training strategy that fuses generated pseudo-IR data with limited real IR samples to train a lightweight YOLOv8 detector. Experimental results demonstrate that this method effectively enhances feature learning under low-data conditions. The augmented detector significantly outperforms models trained on limited real data alone and approaches the performance benchmarks of fully supervised training, proving the efficacy of pseudo-IR synthesis as a robust augmentation strategy for industrial inspection.

Rigorous crop counting is crucial for effective agricultural management and informed intervention strategies. However, in outdoor field environments, partial occlusions combined with inherent ambiguity in distinguishing clustered crops from individual viewpoints poses an immense challenge for image-based segmentation methods. To address these problems, we introduce a novel crop counting framework designed for exact enumeration via 3D instance segmentation. Our approach utilizes 2D images captured from multiple viewpoints and associates independent instance masks for neural radiance field (NeRF) view synthesis. We introduce crop visibility and mask consistency scores, which are incorporated alongside 3D information from a NeRF model. This results in an effective segmentation of crop instances in 3D and highly-accurate crop counts. Furthermore, our method eliminates the dependence on crop-specific parameter tuning. We validate our framework on three agricultural datasets consisting of cotton bolls, apples, and pears, and demonstrate consistent counting performance despite major variations in crop color, shape, and size. A comparative analysis against the state of the art highlights superior performance on crop counting tasks. Lastly, we contribute a cotton plant dataset to advance further research on this topic.

Robotic fleets such as unmanned aerial and ground vehicles have been widely used for routine inspections of static environments, where the areas of interest are known and planned in advance. However, in many applications, such areas of interest are unknown and should be identified online during exploration. Thus, this paper considers the problem of simultaneous exploration, inspection of unknown environments and then real-time communication to a mobile ground control station to report the findings. The heterogeneous robots are equipped with different sensors, e.g., long-range lidars for fast exploration and close-range cameras for detailed inspection. Furthermore, global communication is often unavailable in such environments, where the robots can only communicate with each other via ad-hoc wireless networks when they are in close proximity and free of obstruction. This work proposes a novel planning and coordination framework (SLEI3D) that integrates the online strategies for collaborative 3D exploration, adaptive inspection and timely communication (via the intermit-tent or proactive protocols). To account for uncertainties w.r.t. the number and location of features, a multi-layer and multi-rate planning mechanism is developed for inter-and-intra robot subgroups, to actively meet and coordinate their local plans. The proposed framework is validated extensively via high-fidelity simulations of numerous large-scale missions with up to 48 robots and 384 thousand cubic meters. Hardware experiments of 7 robots are also conducted. Project website is available at https://junfengchen-robotics.github.io/SLEI3D/.

Diffusion-based policies have gained growing popularity in solving a wide range of decision-making tasks due to their superior expressiveness and controllable generation during inference. However, effectively training large diffusion policies using reinforcement learning (RL) remains challenging. Existing methods either suffer from unstable training due to directly maximizing value objectives, or face computational issues due to relying on crude Gaussian likelihood approximation, which requires a large amount of sufficiently small denoising steps. In this work, we propose DIPOLE (Dichotomous diffusion Policy improvement), a novel RL algorithm designed for stable and controllable diffusion policy optimization. We begin by revisiting the KL-regularized objective in RL, which offers a desirable weighted regression objective for diffusion policy extraction, but often struggles to balance greediness and stability. We then formulate a greedified policy regularization scheme, which naturally enables decomposing the optimal policy into a pair of stably learned dichotomous policies: one aims at reward maximization, and the other focuses on reward minimization. Under such a design, optimized actions can be generated by linearly combining the scores of dichotomous policies during inference, thereby enabling flexible control over the level of greediness.Evaluations in offline and offline-to-online RL settings on ExORL and OGBench demonstrate the effectiveness of our approach. We also use DIPOLE to train a large vision-language-action (VLA) model for end-to-end autonomous driving (AD) and evaluate it on the large-scale real-world AD benchmark NAVSIM, highlighting its potential for complex real-world applications.

Generative models have emerged as powerful tools for planning, with compositional approaches offering particular promise for modeling long-horizon task distributions by composing together local, modular generative models. This compositional paradigm spans diverse domains, from multi-step manipulation planning to panoramic image synthesis to long video generation. However, compositional generative models face a critical challenge: when local distributions are multimodal, existing composition methods average incompatible modes, producing plans that are neither locally feasible nor globally coherent. We propose Compositional Diffusion with Guided Search (CDGS), which addresses this mode averaging problem by embedding search directly within the diffusion denoising process. Our method explores diverse combinations of local modes through population-based sampling, prunes infeasible candidates using likelihood-based filtering, and enforces global consistency through iterative resampling between overlapping segments. CDGS matches oracle performance on seven robot manipulation tasks, outperforming baselines that lack compositionality or require long-horizon training data. The approach generalizes across domains, enabling coherent text-guided panoramic images and long videos through effective local-to-global message passing. More details: https://cdgsearch.github.io/

We present GRL-SNAM, a geometric reinforcement learning framework for Simultaneous Navigation and Mapping(SNAM) in unknown environments. A SNAM problem is challenging as it needs to design hierarchical or joint policies of multiple agents that control the movement of a real-life robot towards the goal in mapless environment, i.e. an environment where the map of the environment is not available apriori, and needs to be acquired through sensors. The sensors are invoked from the path learner, i.e. navigator, through active query responses to sensory agents, and along the motion path. GRL-SNAM differs from preemptive navigation algorithms and other reinforcement learning methods by relying exclusively on local sensory observations without constructing a global map. Our approach formulates path navigation and mapping as a dynamic shortest path search and discovery process using controlled Hamiltonian optimization: sensory inputs are translated into local energy landscapes that encode reachability, obstacle barriers, and deformation constraints, while policies for sensing, planning, and reconfiguration evolve stagewise via updating Hamiltonians. A reduced Hamiltonian serves as an adaptive score function, updating kinetic/potential terms, embedding barrier constraints, and continuously refining trajectories as new local information arrives. We evaluate GRL-SNAM on two different 2D navigation tasks. Comparing against local reactive baselines and global policy learning references under identical stagewise sensing constraints, it preserves clearance, generalizes to unseen layouts, and demonstrates that Geometric RL learning via updating Hamiltonians enables high-quality navigation through minimal exploration via local energy refinement rather than extensive global mapping. The code is publicly available on \href{https://github.com/CVC-Lab/GRL-SNAM}{Github}.

Robotic systems operating in dynamic and uncertain environments increasingly require planners that satisfy complex task sequences while adhering to strict temporal constraints. Metric Interval Temporal Logic (MITL) offers a formal and expressive framework for specifying such time-bounded requirements; however, integrating MITL with reinforcement learning (RL) remains challenging due to stochastic dynamics and partial observability. This paper presents a unified automata-based RL framework for synthesizing policies in both Markov Decision Processes (MDPs) and Partially Observable Markov Decision Processes (POMDPs) under MITL specifications. MITL formulas are translated into Timed Limit-Deterministic Generalized Büchi Automata (Timed-LDGBA) and synchronized with the underlying decision process to construct product timed models suitable for Q-learning. A simple yet expressive reward structure enforces temporal correctness while allowing additional performance objectives. The approach is validated in three simulation studies: a $5 \times 5$ grid-world formulated as an MDP, a $10 \times 10$ grid-world formulated as a POMDP, and an office-like service-robot scenario. Results demonstrate that the proposed framework consistently learns policies that satisfy strict time-bounded requirements under stochastic transitions, scales to larger state spaces, and remains effective in partially observable environments, highlighting its potential for reliable robotic planning in time-critical and uncertain settings.

We present SpaceTimePilot, a video diffusion model that disentangles space and time for controllable generative rendering. Given a monocular video, SpaceTimePilot can independently alter the camera viewpoint and the motion sequence within the generative process, re-rendering the scene for continuous and arbitrary exploration across space and time. To achieve this, we introduce an effective animation time-embedding mechanism in the diffusion process, allowing explicit control of the output video's motion sequence with respect to that of the source video. As no datasets provide paired videos of the same dynamic scene with continuous temporal variations, we propose a simple yet effective temporal-warping training scheme that repurposes existing multi-view datasets to mimic temporal differences. This strategy effectively supervises the model to learn temporal control and achieve robust space-time disentanglement. To further enhance the precision of dual control, we introduce two additional components: an improved camera-conditioning mechanism that allows altering the camera from the first frame, and CamxTime, the first synthetic space-and-time full-coverage rendering dataset that provides fully free space-time video trajectories within a scene. Joint training on the temporal-warping scheme and the CamxTime dataset yields more precise temporal control. We evaluate SpaceTimePilot on both real-world and synthetic data, demonstrating clear space-time disentanglement and strong results compared to prior work. Project page: https://zheninghuang.github.io/Space-Time-Pilot/ Code: https://github.com/ZheningHuang/spacetimepilot

Humanoid robots hold great promise for operating in human-centric environments, yet achieving robust whole-body coordination across the head, hands, and legs remains a major challenge. We present a system that combines a modular teleoperation interface with a scalable learning framework to address this problem. Our teleoperation design decomposes humanoid control into intuitive submodules, which include hand-eye coordination, grasp primitives, arm end-effector tracking, and locomotion. This modularity allows us to collect high-quality demonstrations efficiently. Building on this, we introduce Choice Policy, an imitation learning approach that generates multiple candidate actions and learns to score them. This architecture enables both fast inference and effective modeling of multimodal behaviors. We validate our approach on two real-world tasks: dishwasher loading and whole-body loco-manipulation for whiteboard wiping. Experiments show that Choice Policy significantly outperforms diffusion policies and standard behavior cloning. Furthermore, our results indicate that hand-eye coordination is critical for success in long-horizon tasks. Our work demonstrates a practical path toward scalable data collection and learning for coordinated humanoid manipulation in unstructured environments.

Vision Language Models (VLMs) are increasingly adopted as central reasoning modules for embodied agents. Existing benchmarks evaluate their capabilities under ideal, well-lit conditions, yet robust 24/7 operation demands performance under a wide range of visual degradations, including low-light conditions at night or in dark environments, a core necessity that has been largely overlooked. To address this underexplored challenge, we present DarkQA, an open-source benchmark for evaluating perceptual primitives under multi-level low-light conditions in embodied scenarios. DarkQA evaluates single-view egocentric observations across controlled degradation levels, isolating low-light perceptual failures before they are entangled with complex embodied tasks. The benchmark contains 9.4K deterministically generated and verifiable question-image pairs spanning five visual-primitive families. A key design feature of DarkQA is its physical fidelity: visual degradations are modeled in linear RAW space, simulating physics-based illumination drop and sensor noise followed by an ISP-inspired rendering pipeline; we further validate the synthesis against real paired low-light camera data. We evaluate representative VLMs and Low-Light Image Enhancement (LLIE) preprocessing methods. Results show consistent VLM degradation under low illumination and sensor noise, while LLIE provides severity-dependent but unstable recovery. We demonstrate the utility of DarkQA by evaluating a wide range of state-of-the-art VLMs and Low-Light Image Enhancement (LLIE) models, and systematically reveal VLMs' limitations when operating under these challenging visual conditions. Our code and benchmark dataset will be released upon acceptance. Project website: https://darkqa-benchmark.github.io

Deformable linear objects (DLOs) manipulation presents significant challenges due to DLOs' inherent high-dimensional state space and complex deformation dynamics. The wide-populated obstacles in realistic workspaces further complicate DLO manipulation, necessitating efficient deformation planning and robust deformation tracking. In this work, we propose a novel framework for DLO manipulation in constrained environments. This framework combines hierarchical deformation planning with neural tracking, ensuring reliable performance in both global deformation synthesis and local deformation tracking. Specifically, the deformation planner begins by generating a spatial path set that inherently satisfies the homotopic constraints associated with DLO keypoint paths. Next, a path-set-guided optimization method is applied to synthesize an optimal temporal deformation sequence for the DLO. In manipulation execution, a neural model predictive control approach, leveraging a data-driven deformation model, is designed to accurately track the planned DLO deformation sequence. The effectiveness of the proposed framework is validated in extensive constrained DLO manipulation tasks.