Robot Learning

Ask about this topic

Overview

Robot learning studies how robots can turn observations and experience into actions, from precise manipulation and locomotion to navigation and human-robot interaction. Common approaches include imitation learning from human or robot demonstrations, reinforcement learning through interaction, model-based control, and vision-language-action models that connect instructions and images to actions. Researchers also use reusable skills, structured physical representations, tactile sensing, and simulation or world models to improve transfer across tasks, objects, and robot bodies. A central challenge is making policies reliable beyond their training data: larger and more varied datasets can help, but physical feasibility, sim-to-real transfer, long-horizon recovery, and evaluation that predicts real-world success remain open concerns.

Latest papers

CardsList

321 papers75 papers in the last four weeks, up 79% on the four weeks before. 0.4% of all new papers.

Mar 3, 2026

Uni-Skill: Building Self-Evolving Skill Repository for Generalizable Robotic Manipulation

While skill-centric approaches leverage foundation models to enhance generalization in compositional tasks, they often rely on fixed skill libraries, limiting adaptability to new tasks without manual intervention. To address this, we propose Uni-Skill, a Unified Skill-centric framework that supports skill-aware planning and facilitates automatic skill evolution. Unlike prior methods that restrict planning to predefined skills, Uni-Skill requests for new skill implementations when existing ones are insufficient, ensuring adaptable planning with self-augmented skill library. To support automatic implementation of diverse skills requested by the planning module, we construct SkillFolder, a VerbNet-inspired repository derived from large-scale unstructured robotic videos. SkillFolder introduces a hierarchical skill taxonomy that captures diverse skill descriptions at multiple levels of abstraction. By populating this taxonomy with large-scale, automatically annotated demonstrations, Uni-Skill shifts the paradigm of skill acquisition from inefficient manual annotation to efficient offline structural retrieval. Retrieved examples provide semantic supervision over behavior patterns and fine-grained references for spatial trajectories, enabling few-shot skill inference without deployment-time demonstrations. Comprehensive experiments in both simulation and real-world settings verify the state-of-the-art performance of Uni-Skill over existing VLM-based skill-centric approaches, highlighting its advanced reasoning capabilities and strong zero-shot generalization across a wide range of novel tasks.

Senwei Xie, Yuntian Zhang, Ruiping Wang +1

Key Laboratory of AI Safety of CAS, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, 100190, China · University of Chinese Academy of Sciences, Beijing, 100049, China

Robot Skill LearningRobot LearningGeneralization in Robotic ManipulationLarge Language Model-Based Robot PlanningVLMs for Robotics
AskPDF
Mar 2, 2026

FATE: Closed-Loop Feasibility-Aware Task Generation with Active Repair for Physically Grounded Robotic Curricula

Recent breakthroughs in generative simulation have harnessed Large Language Models (LLMs) to generate diverse robotic task curricula, yet these open-loop paradigms frequently produce linguistically coherent but physically infeasible goals, stemming from ungrounded task specifications or misaligned objective formulations. To address this critical limitation, we propose FATE (Feasibility-Aware Task gEneration), a closed-loop, self-correcting framework that reimagines task generation as an iterative validation-and-refinement process. Unlike conventional methods that decouple generation and verification into discrete stages, FATE embeds a generalist embodied agent directly into the generation loop to proactively guarantee the physical groundedness of the resulting curriculum. FATE instantiates a sequential auditing pipeline: it first validates static scene attributes (e.g., object affordances, layout compatibility) and subsequently verifies execution feasibility via simulated embodied interaction. Critical to its performance, upon detecting an infeasible task, FATE deploys an active repair module that autonomously adapts scene configurations or policy specifications, converting unworkable proposals into physically valid task instances. Extensive experiments validate that FATE generates semantically diverse, physically grounded task curricula while achieving a substantial reduction in execution failure rates relative to state-of-the-art generative baselines.

Bingchuan Wei, Bingqi Huang, Jingheng Ma +2

School of Aerospace Engineering, Tsinghua University, Beijing, China · School of Integrated Circuits, Tsinghua University, Beijing, China · State Key Laboratory of General Artificial Intelligence, Beijing Institute for General Artificial Intelligence (BIGAI), Beijing, China +1

Synthetic Data Generation for RoboticsRobot LearningRobotics Simulation
AskPDF
Mar 2, 2026

Towards Robot Skill Learning and Adaptation with Gaussian Processes

General robot skill adaptation requires expressive representations robust to varying task configurations. While recent learning-based skill adaptation methods refined via Reinforcement Learning (RL), have shown success, existing skill models often lack sufficient representational capacity for anything beyond minor environmental changes. In contrast, Gaussian Process (GP)-based skill modelling provides an expressive representation with useful analytical properties; however, adaptation of GP-based skills remains underexplored. This paper proposes a novel, robust skill adaptation framework that utilises GPs with sparse via-points for compact and expressive modelling. The model considers the trajectory's poses and leverages its first and second analytical derivatives to preserve the skill's kinematic profile. We present three adaptation methods to cater for the variability between initial and observed configurations. Firstly, an optimisation agent that adjusts the path's via-points while preserving the demonstration velocity. Second, a behaviour cloning agent trained to replicate output trajectories from the optimisation agent. Lastly, an RL agent that has learnt to modify via-points whilst maintaining the kinematic profile and enabling online capabilities. Evaluated across three tasks (drawer opening, cube-pushing and bar manipulation) in both simulation and hardware, our proposed methods outperform every benchmark in success rates. Furthermore, the results demonstrate that the GP-based representation enables all three methods to attain high cosine similarity and low velocity magnitude errors, indicating strong preservation of the kinematic profile. Overall, our formulation provides a compact representation capable of adapting to large deviations from a single demonstrated skill.

A K M Nadimul Haque, Fouad Sukkar, Sheila Sutjipto +3

Robot Skill LearningRobot LearningRobotic ManipulationTrajectory Representation Learning
AskPDF
Feb 26, 2026

LeRobot: An Open-Source Library for End-to-End Robot Learning

Robotics is undergoing a significant transformation powered by advances in high-level control techniques based on machine learning, giving rise to the field of robot learning. Recent progress in robot learning has been accelerated by the increasing availability of affordable teleoperation systems, large-scale openly available datasets, and scalable learning-based methods. However, development in the field of robot learning is often slowed by fragmented, closed-source tools designed to only address specific sub-components within the robotics stack. In this paper, we present lerobot, an open-source library that integrates across the entire robot learning stack, from low-level middleware communication for motor controls to large-scale dataset collection, storage and streaming. The library is designed with a strong focus on real-world robotics, supporting accessible hardware platforms while remaining extensible to new embodiments. It also supports efficient implementations for various state-of-the-art robot learning algorithms from multiple prominent paradigms, as well as a generalized asynchronous inference stack. Unlike traditional pipelines which heavily rely on hand-crafted techniques, lerobot emphasizes scalable learning approaches that improve directly with more data and compute. Designed for accessibility, scalability, and openness, lerobot lowers the barrier to entry for researchers and practitioners to robotics while providing a platform for reproducible, state-of-the-art robot learning.

Remi Cadene, Simon Aliberts, Francesco Capuano +14

Hugging Face · University of Oxford

Robot LearningRobotic Control
AskPDF
Feb 22, 2026

TOPReward: Token Probabilities as Hidden Zero-Shot Rewards for Robotics

General-purpose robot learning requires dense, instruction-conditioned feedback that can distinguish meaningful task progress from stalled, failed, or partially completed behavior. Yet obtaining such feedback at scale remains difficult, since existing approaches often rely on manual progress annotations, task-specific demonstrations, or reward models trained on curated robot datasets. We introduce TOPReward, a training-free progress reward method that probes pretrained Video-Language Models (VLMs) through their internal token probabilities rather than asking them to generate numerical progress values. Given a video prefix and a language instruction, TOPReward measures the model's likelihood that the instructed task has been completed, converting latent video-language understanding into a dense reward signal without task-specific reward-model training or manually annotated progress labels. We evaluate TOPReward on ManiRewardBench, our real-world manipulation benchmark spanning 130 unique tasks and four robot platforms, as well as on Open X-Embodiment datasets. Across these settings, TOPReward substantially outperforms prior training-free VLM reward methods on open-source models and is competitive with a trained reward-model baseline on progress-estimation metrics, while requiring no reward-model training. Additional analyses show that the reward is sensitive to the specified instruction and is not explained by time index alone. Finally, TOPReward supports downstream applications including success detection and offline reward-weighted behavior cloning.

Shirui Chen, Cole Harrison, Ying-Chun Lee +6

University of Washington · Allen Institute for AI · Amazon +1

Robot Task Progress EstimationRobot LearningMultimodal Reward ModelingVLMs for RoboticsRobot Learning from Video
AskPDF
Feb 22, 2026

The Price Is Not Right: Neuro-Symbolic Methods Outperform VLAs on Structured Long-Horizon Manipulation Tasks with Significantly Lower Energy Consumption

Vision-Language-Action (VLA) models have recently been proposed as a pathway toward generalist robotic policies capable of interpreting natural language and visual inputs to generate manipulation actions. However, their effectiveness and efficiency on structured, long-horizon manipulation tasks remain unclear. In this work, we present a head-to-head empirical comparison between a fine-tuned open-weight VLA model π0 and a neuro-symbolic architecture that combines PDDL-based symbolic planning with learned low-level control. We evaluate both approaches on structured variants of the Towers of Hanoi manipulation task in simulation while measuring both task performance and energy consumption during training and execution. On the 3-block task, the neuro-symbolic model achieves 95% success compared to 34% for the best-performing VLA. The neuro-symbolic model also generalizes to an unseen 4-block variant (78% success), whereas both VLAs fail to complete the task. During training, VLA fine-tuning consumes nearly two orders of magnitude more energy than the neuro-symbolic approach. These results highlight important trade-offs between end-to-end foundation-model approaches and structured reasoning architectures for long-horizon robotic manipulation, emphasizing the role of explicit symbolic structure in improving reliability, data efficiency, and energy efficiency. Code and models are available at https://price-is-not-right.github.io

Timothy Duggan, Pierrick Lorang, Hong Lu +1

Human-Robot Interaction Lab, Tufts University, Medford, MA, USA. · AIT Austrian Institute of Technology GmbH, Center for Vision, Automation & Control, Vienna, Austria.

Long-Horizon Robotic ManipulationRobot LearningVision-Language-Action ModelsNeuro-Symbolic AISymbolic Planning
AskPDF
Feb 20, 2026

Enhancing Goal Inference via Correction Timing

Corrections offer a natural modality for people to provide feedback to a robot, by (i) intervening in the robot's behavior when they believe the robot is failing (or will fail) the task objectives and (ii) modifying the robot's behavior to successfully fulfill the task. Each correction offers information on what the robot should and should not do, where the corrected behavior is more aligned with task objectives than the original behavior. Most prior work on learning from corrections involves interpreting a correction as a new demonstration (consisting of the modified robot behavior), or a preference (for the modified trajectory compared to the robot's original behavior). However, this overlooks one essential element of the correction feedback, which is the human's decision to intervene in the robot's behavior in the first place. This decision can be influenced by multiple factors including the robot's task progress, alignment with human expectations, dynamics, motion legibility, and optimality. In this work, we investigate whether the timing of this decision can offer a useful signal for inferring these task-relevant influences. In particular, we investigate three potential applications for this learning signal: (1) identifying features of a robot's motion that may prompt people to correct it, (2) quickly inferring the final goal of a human's correction based on the timing and initial direction of their correction motion, and (3) learning more precise constraints for task objectives. Our results indicate that correction timing results in improved learning for the first two of these applications. Overall, our work provides new insights on the value of correction timing as a signal for robot learning.

Anjiabei Wang, Shuangge Wang, Tesca Fitzgerald

Yale University, New Haven, CT, United States

Robot LearningHuman-Robot CollaborationRL from Human Feedback
AskPDF
Feb 20, 2026

EgoPush: Egocentric Multi-Object Rearrangement for Mobile Robots via Constrained Teacher Observability

Humans rearrange objects in cluttered environments using egocentric perception, actively moving to keep task-relevant spatial cues in view. Mobile robots have not matched this: rearrangement is usually built on a global pose estimate or a map, which is exactly what a robot carrying one camera lacks, while pushing keeps changing the scene it would have to be built from. We present EgoPush, which pushes objects into anchor-relative formations from onboard RGB-D alone, with no global localization, external tracking, or map at deployment, and transfers zero-shot to a TurtleBot in controlled and visually cluttered scenes. What makes this learnable turns out to be a property of the teacher rather than of the student: three privileged teachers trained with identical rewards, architecture, and hyperparameters all exceed 98%98\% success, yet their distilled egocentric students reach 0%0\%, 54.8%54.8\%, and 87.3%87.3\%, the only variable being the teacher's observation function. EgoPush therefore trains the teacher under egocentric observability constraints, restricting it to visibility-limited cues and revealing target references only when the anchor is centrally visible, so that its supervision is recoverable by a depth-based student distilled online. Making the teacher trainable in the first place needs two further pieces: a role-grouped object-centric interface shared by teacher and student, and stage-wise temporally decayed rewards for long-horizon credit assignment. Videos, the playable task, and code are available at https://ai4ce.github.io/EgoPush/.

Boyuan An, Zhexiong Wang, Yipeng Wang +4

New York University

Robotic ManipulationRobot LearningMobile ManipulationRobotic Object PushingPolicy DistillationZero-Shot Robotic Manipulation
AskPDF
Feb 18, 2026

SparTa: Sparse Graphical Task Models from a Handful of Demonstrations

Learning long-horizon manipulation tasks efficiently is a central challenge in robot learning from demonstration. Unlike recent endeavors that focus on directly learning the task in the action domain, we focus on inferring what the robot should achieve in the task, rather than how to do so. To this end, we represent evolving scene states using a series of graphical object relationships. We propose a demonstration segmentation and pooling approach that extracts a series of manipulation graphs and estimates distributions over object states across task phases. In contrast to prior graph-based methods that capture only partial interactions or short temporal windows, our approach captures complete object interactions spanning from the onset of control to the end of the manipulation. To improve robustness when learning from multiple demonstrations, we additionally perform object matching using pre-trained visual features. In extensive experiments, we evaluate our method's demonstration segmentation accuracy and the utility of learning from multiple demonstrations for finding a desired minimal task model. Finally, we deploy the fitted models both in simulation and on a real robot, demonstrating that the resulting task representations support reliable execution across environments.

Adrian Röfer, Nick Heppert, Abhinav Valada

University of Freiburg, Germany · Zuse School ELIZA

Robot LearningLearning from DemonstrationLong-Horizon Robotic ManipulationData-Efficient Robot LearningGraph Representation Learning
AskPDF
Feb 18, 2026

EgoScale: Scaling Dexterous Manipulation with Diverse Egocentric Human Data

Human behavior is among the most scalable sources of data for learning physical intelligence, yet how to effectively leverage it for dexterous manipulation remains unclear. While prior work demonstrates human to robot transfer in constrained settings, it is unclear whether large scale human data can support fine grained, high degree of freedom dexterous manipulation. We present EgoScale, a human to dexterous manipulation transfer framework built on large scale egocentric human data. We train a Vision Language Action (VLA) model on over 20,854 hours of action labeled egocentric human video, more than 20 times larger than prior efforts, and uncover a log linear scaling law between human data scale and validation loss. This validation loss strongly correlates with downstream real robot performance, establishing large scale human data as a predictable supervision source. Beyond scale, we introduce a simple two stage transfer recipe: large scale human pretraining followed by lightweight aligned human robot mid training. This enables strong long horizon dexterous manipulation and one shot task adaptation with minimal robot supervision. Our final policy improves average success rate by 54% over a no pretraining baseline using a 22 DoF dexterous robotic hand, and transfers effectively to robots with lower DoF hands, indicating that large scale human motion provides a reusable, embodiment agnostic motor prior.

Ruijie Zheng, Dantong Niu, Yuqi Xie +12

NVIDIA · University of California, Berkeley · University of Maryland

Robotic ManipulationRobot LearningTransfer LearningLong-Horizon Robotic ManipulationCross-Embodiment Robot LearningData-Efficient Robot Learning
AskPDF
Feb 17, 2026

Feasibility-aware Imitation Learning from Observation with Multimodal Feedback

Imitation learning frameworks that learn robot control policies from demonstrators' motions via hand-mounted demonstration interfaces have attracted increasing attention. However, due to differences in physical characteristics between demonstrators and robots, this approach faces two limitations: i) the demonstration data do not include robot actions, and ii) the demonstrated motions may be infeasible for robots. These limitations make policy learning difficult. To address them, we propose Feasibility-Aware Behavior Cloning from Observation (FABCO). FABCO integrates behavior cloning from observation, which complements robot actions using robot dynamics models, with feasibility estimation. In feasibility estimation, the demonstrated motions are evaluated using a robot-dynamics model, learned from the robot's execution data, to assess reproducibility under the robot's dynamics. The estimated feasibility is used for multimodal feedback and feasibility-aware policy learning to improve the demonstrator's motions and learn robust policies. Multimodal feedback provides feasibility through the demonstrator's visual and haptic senses to promote feasible demonstrated motions. Feasibility-aware policy learning reduces the influence of demonstrated motions that are infeasible for robots, enabling the learning of policies that robots can execute stably. We conducted experiments with 15 participants on two tasks and confirmed that FABCO improves imitation learning performance by more than 3.2 times compared to the case without feasibility feedback.

Kei Takahashi, Hikaru Sasaki, Takamitsu Matsubara

Division of Information Science, Graduate School of Science and Technology, Nara Institute of Science and Technology (NAIST), Ikoma-shi, Nara 630-0192, Japan

Imitation LearningRobot Imitation LearningRobot Learning
AskPDF
Feb 16, 2026

BPP: Long-Context Robot Imitation Learning by Focusing on Key History Frames

Many robot tasks require attending to the history of past observations. For example, finding an item in a room requires remembering which places have already been searched. However, the best-performing robot policies typically condition only on the current observation, limiting their applicability to such tasks. Naively conditioning on past observations often fails due to spurious correlations: policies latch onto incidental features of training histories that do not generalize to out-of-distribution trajectories upon deployment. We analyze why policies latch onto these spurious correlations and find that this problem stems from limited coverage over the space of possible histories during training, which grows exponentially with horizon. Existing regularization techniques provide inconsistent benefits across tasks, as they do not fundamentally address this coverage problem. Motivated by these findings, we propose Big Picture Policies (BPP), an approach that conditions on a minimal set of meaningful keyframes detected by a vision-language model. By projecting diverse rollouts onto a compact set of task-relevant events, BPP substantially reduces distribution shift between training and deployment, without sacrificing expressivity. We evaluate BPP on four challenging real-world manipulation tasks and three simulation tasks, all requiring history conditioning. BPP achieves 70% higher success rates than the best comparison on real-world evaluations. Videos are available at https://bigpicturepolicies.github.io/

Max Sobol Mark, Jacky Liang, Maria Attarian +4

Carnegie Mellon University · Google DeepMind · Princeton University

Robot Imitation LearningRobot LearningRobot Policy Generalization
AskPDF
Feb 8, 2026

USER: A Unified and Extensible System for Real-World Online Policy Learning in Embodied AI

Online policy learning directly in the physical world is a promising yet challenging direction for embodied intelligence. Unlike simulation, real-world systems cannot be arbitrarily accelerated, cheaply reset, or massively replicated, suggesting that real-world policy learning is not merely an algorithmic problem, but inherently a systems problem. We present USER, a Unified and extensible SystEm for real-world online policy leaRning. On the systems side, USER introduces a hardware abstraction layer for unified robot management and an adaptive communication plane that enables efficient cloud-edge training. On the learning side, USER adopts a fully asynchronous training framework, designs a persistent and cache-aware replay buffer, and provides extensible abstractions for rewards, algorithms, and policies. Experiments in both simulation and the real world demonstrate that USER supports multi-robot coordination, heterogeneous manipulators, cloud-edge training with large models, and long-running asynchronous training. Together, these capabilities establish USER as a unified and extensible systems foundation for real-world online policy learning.

Hongzhi Zang, Shu'ang Yu, Hao Lin +14

Tsinghua University · Shanghai AI Laboratory · Infinigence AI +3

Robot Policy LearningReinforcement LearningDistributed RLRobot Learning
AskPDF
Dec 30, 2025

World In Your Hands: A Large-Scale and Open-Source Ecosystem for Learning Human-Centric Manipulation in the Wild

We introduce World In Your Hands (WIYH), a large-scale open-source ecosystem comprising over 1,000 hours of human manipulation data collected in-the-wild with millimeter-scale motion accuracy. Specifically, WIYH includes (1) the Oracle Suite, a wearable data collection kit with an auto-labeling pipeline for accurate motion capture; (2) the WIYH Dataset, featuring over 1,000 hours of multimodal manipulation data across hundreds of skills in diverse real-world scenarios; and (3) extensive annotations and benchmarks supporting tasks from perception to action. Furthermore, experiments based on the WIYH ecosystem show that integrating WIYH's human-centric data improves robotic manipulation success rates from 8% to 60% in cluttered scenes. World In Your Hands provides a foundation for advancing human-centric data collection and cross-embodiment policy learning. All data and hardware design will be open-source.

Yupeng Zheng, Jichao Peng, Weize Li +22

TARS Robotics

Human-Object InteractionRobot Skill LearningRobotic Data CollectionCross-Embodiment Robot LearningRobotic ManipulationRobot Manipulation Benchmarks+2
AskPDF
Dec 20, 2025

Learning Semantic Atomic Skills for Multi-Task Robotic Manipulation

Scaling imitation learning to diverse multi-task robot manipulation remains challenging due to suboptimal demonstrations, behavioral multi-modality, and destructive interference across tasks. While skill-based methods offer a promising direction by decomposing behaviors into reusable abstractions, existing approaches often learn skills that are either biased toward linguistic structure or lack semantic alignment across tasks, limiting generalization. In this work, we propose AtomSkill, a novel framework that learns a semantically aligned Atomic Skill Space from demonstrations and enables robust long-horizon execution through keypose imagination. Our method introduces: (1) semantic contrastive skill alignment, which partitions demonstrations into variable-length atomic skills and employs a contrastive objective to jointly enforce semantic consistency and temporal coherence, yielding a compact and reusable skill library; and (2) action decoding with keypose imagining, where the policy predicts both a skill's terminal keypose and immediate actions, thereby supporting progress-aware skill transitions. During inference, an atomic skill diffusion sampler generates plausible skill sequences, while predicted keyposes autonomously trigger smooth skill chaining. Extensive experiments in simulation and real-world settings show that AtomSkill consistently outperforms state-of-the-art imitation learning and skill-based baselines. Project page: https://atom-skill.github.io.

Yihang Zhu, Weiqing Wang, Shijie Wu +2

ShanghaiTech University, Shanghai, China · InstAdapt

Long-Horizon Robotic ManipulationRobot Skill LearningGeneralization in Robotic ManipulationMulti-Task Robotic ManipulationRobotic ManipulationRobot Imitation Learning+1
AskPDF
Nov 21, 2025

METIS: Multi-Source Egocentric Training for Integrated Dexterous Vision-Language-Action Model

Building a generalist robot that can perceive, reason, and act across diverse tasks remains an open challenge, especially for dexterous manipulation. A major bottleneck lies in the scarcity of large-scale, action-annotated data for dexterous skills, as teleoperation is difficult and costly. Human data, with its vast scale and diverse manipulation behaviors, provides rich priors for learning robotic actions. While prior works have explored leveraging human demonstrations, they are often constrained by limited scenarios and a large visual gap between human and robots. To eliminate these limitations, we propose METIS, a vision-language-action (VLA) model for dexterous manipulation pretrained on multi-source egocentric datasets. We first construct EgoAtlas, which integrates large-scale human and robotic data from multiple sources, all unified under a consistent action space. We further extract motion-aware dynamics, a compact and discretized motion representation, which provides efficient and expressive supervision for VLA training. Built upon them, METIS integrates reasoning and acting into a unified framework, enabling effective deployment to downstream dexterous manipulation tasks. Our method demonstrates exceptional dexterous manipulation capabilities, achieving highest average success rate in six real-world tasks. Experimental results also highlight the superior generalization and robustness to out-of-distribution scenarios. These findings emphasize METIS as a promising step toward a generalist model for dexterous manipulation.

Yankai Fu, Ning Chen, Junkai Zhao +5

State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University · Beijing Academy of Artificial Intelligence

Robotic ManipulationVLMs for RoboticsRobot LearningGeneralization in Robotic ManipulationCross-Embodiment Robot LearningRobustness of VLA Models
AskPDF
Oct 4, 2025

Model-Based Adaptive Precision Control for Tabletop Planar Pushing Under Uncertain Dynamics

Data-driven planar pushing methods have recently gained attention as they reduce manual engineering effort and improve generalization compared to analytical approaches. However, most prior work targets narrow capabilities (e.g., side switching, precision, or single-task training), limiting broader applicability. We present a model-based framework for non-prehensile tabletop pushing that uses a single learned model to address multiple tasks without retraining. Our approach employs a recurrent GRU-based architecture with additional non-linear layers to capture object-environment dynamics while ensuring stability. A tailored state-action representation enables the model to generalize across uncertain dynamics, variable push lengths, and diverse tasks. For control, we integrate the learned dynamics with a sampling-based Model Predictive Path Integral (MPPI) controller, which generates adaptive, task-oriented actions. This framework supports side switching, variable-length pushes, and objectives such as precise positioning, trajectory following, and obstacle avoidance. Training is performed in simulation with domain randomization to support sim-to-real transfer. We first evaluate the architecture through ablation studies, showing improved prediction accuracy and stable rollouts. We then validate the full system in simulation and real-world experiments using a Franka Panda robot with markerless tracking. Results demonstrate high success rates in precise positioning under strict thresholds and strong performance in trajectory tracking and obstacle avoidance. Moreover, multiple tasks are solved simply by changing the controller's objective function, without retraining. While our current focus is on a single object type, we extend the framework by training on wider push lengths and designing a balanced controller that reduces the number of steps for longer-horizon goals.

Aydin Ahmadi, Baris Akgun

Computer Engineering Department, Ko¸c University, Rumelifeneri Yolu, Istanbul, 34450, Istanbul, Turkey.

Contact-Rich ManipulationNon-Prehensile ManipulationRobot Dynamics LearningRobotic ManipulationRobot Learning
AskPDF
Sep 23, 2025

Agentic Scene Policies

Designing or learning robot policies that generalize zero-shot across a range of language instructions and objects is a core problem in robotics. Vision-Language-Action models (VLAs) learn such policies end-to-end by repurposing existing Vision-Language Models (VLMs), but generalization to new instructions and objects remains challenging. An alternative is to implement a modular policy by leveraging an explicit VLM-based 3D scene representation and motion planning. While modular policies show strong zero-shot potential, they typically retrieve objects based on semantics without explicit spatial reasoning, severely restricting their overall grounding capabilities. They also interact with objects using basic grasping and navigation skills. In this work, we address these limitations by unifying grounding capabilities and robot skills in a single agentic action space through a scene-agent tool interface. By leveraging part-level affordances, our skills generalize across diverse objects and enable zero-shot interactions such as unplugging chargers and opening drawers. We name the resulting framework Agentic Scene Policies (ASP). Through extensive real-world experiments, we show how ASP consistently outperforms leading VLAs in the zero-shot setting. We also demonstrate the extensibility of our framework by introducing a mobile version of ASP to tackle room-level queries. See our project page (https://montrealrobotics.ca/agentic-scene-policies.github.io/) for more results.

Sacha Morin, Kumaraditya Gupta, Mahtab Sandhu +4

Université de Montréal · Mila - Quebec AI Institute · Sapienza University of Rome

Robot Policy GeneralizationRobot LearningZero-Shot Robotic ManipulationTool-Using Vision-Language Agents
AskPDF
Sep 15, 2025

Learning Contact Dynamics through Touching: Action-conditional Graph Neural Networks for Robotic Peg Insertion

We present a learnable physics-based model that predicts motion of the robot end effector and reaction force-torque in contact-rich manipulation. The model represents the end effector and the environment as interacting meshes in a graph structure, and conditions its prediction explicitly on the applied control input. It predicts object-level pose update directly, while the reaction torque emerges from a per-vertex force field. Training is self-supervised using only joint encoder and force-torque data while the robot is randomly touching the environment without task context. In simulation, our model transfers to peg insertion with unseen concave geometry, where an MPC agent using it reaches up to 98% success rate, and after fine-tuning on self-collected data matches an agent planning with the ground truth dynamics at the tightest 1 mm clearance. In the real world, it outperforms the system-identified MuJoCo model by 45% in position and by 74% and 63% in force and torque error.

Zongyao Yi, Joachim Hertzberg, Martin Atzmueller

German Research Center for Artificial Intelligence, DFKI Niedersachsen, Osnabrück, Germany · Osnabrück University, Institute of Computer Science, Osnabrück, Germany

Graph Neural NetworksContact-Rich Robotic ManipulationRobot Dynamics LearningRobotic InsertionRobot LearningContact Force Estimation
AskPDF
May 28, 2025

Prior Reinforce: Goal-Conditioned Dynamic Manipulation with Limited Trials

Embodied robots have achieved strong performance in many real-world manipulation tasks, yet agile dynamic manipulation remains challenging due to high sensitivity to motion parameters and sparse outcome-level feedback. Tasks such as shooting a basketball into a hoop require precise control of fast open-loop motions, where small trajectory variations can lead to large outcome deviations, making data-efficient adaptation difficult for existing methods that rely on large-scale interaction, reward engineering, or accurate dynamic modeling. We propose Prior Reinforce (P.R.), a simple and practical framework for goal-conditioned dynamic manipulation. The method first learns a structured motion manifold from a small set of demonstrations using a conditional diffusion model, and then adapts motions toward new goals through feedback-driven optimization in a low-dimensional condition space. By separating motion generation from outcome-driven adaptation, the framework enables efficient refinement using only a small number of real-world trials under noisy perception. Experiments on multiple real-world dynamic manipulation tasks demonstrate that P.R. reliably achieves new goals within as few as ten total trials while remaining robust to perception noise and hardware uncertainty, suggesting a practical approach for low-trial real-world robot adaptation. Project website: https://adap-robotics.github.io/.

Yihang Hu, Pingyue Sheng, Yuyang Liu +2

IIIS, Tsinghua University. · Shanghai Qi Zhi Institute. · Spirit AI.

Robot Policy AdaptationRobot Skill LearningGoal-Conditioned Robotic ManipulationRobot Motion GenerationRobotic ManipulationDynamic Robotic Manipulation+1
AskPDF
Oct 24, 2024

Data Scaling Laws in Imitation Learning for Robotic Manipulation

Data scaling has revolutionized fields like natural language processing and computer vision, providing models with remarkable generalization capabilities. In this paper, we investigate whether similar data scaling laws exist in robotics, particularly in robotic manipulation, and whether appropriate data scaling can yield single-task robot policies that can be deployed zero-shot for any object within the same category in any environment. To this end, we conduct a comprehensive empirical study on data scaling in imitation learning. By collecting data across numerous environments and objects, we study how a policy's generalization performance changes with the number of training environments, objects, and demonstrations. Throughout our research, we collect over 40,000 demonstrations and execute more than 15,000 real-world robot rollouts under a rigorous evaluation protocol. Our findings reveal several intriguing results: the generalization performance of the policy follows a roughly power-law relationship with the number of environments and objects. The diversity of environments and objects is far more important than the absolute number of demonstrations; once the number of demonstrations per environment or object reaches a certain threshold, additional demonstrations have minimal effect. Based on these insights, we propose an efficient data collection strategy. With four data collectors working for one afternoon, we collect sufficient data to enable the policies for two tasks to achieve approximately 90% success rates in novel environments with unseen objects.

Fanqi Lin, Yingdong Hu, Pingyue Sheng +3

Tsinghua University · Shanghai Qi Zhi Institute · Shanghai Artificial Intelligence Laboratory

Generalization in Robotic ManipulationRobotic ManipulationRobot Imitation LearningRobot Policy GeneralizationRobot Learning
AskPDF