# contact-rich manipulation

A robotics manipulation approach for tasks involving sustained or complex physical contact, often using tactile feedback to control interaction forces.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Mind2Act World: A Diagnostic Benchmark for Embodied Mind, Physical Action, and Mind-Action Composition in Robotic Manipulation

DevFeed: [Mind2Act World: A Diagnostic Benchmark for Embodied Mind, Physical Action, and Mind-Action Composition in Robotic Manipulation](<https://devfeed.tech/articles/mind2act-world-a-diagnostic-benchmark-for-embodied-mind-physical-action-and-mind-action-composition-in-robotic-manipulation-145673.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26444>)

Author: Yuzhi Huang, Jie Wu, Fanding Huang, Junchao Lin, Weichen Zhang, Ziyi Xiong, Yang Li, Zhengwei Xie, Zhimin Wang, Zhen Liu, Jiazhen Huang, Yuzheng Zhang, Yuche Liu, Hongyi Yang, Chen Tang, Shixiang Tang

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Robotics](<https://devfeed.tech/topics/robotics.md>), [Robotics Simulation](<https://devfeed.tech/topics/robotics-simulation.md>), [Physical AI](<https://devfeed.tech/topics/physical-ai.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>)

Tags: [act](<https://devfeed.tech/tags/act.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [composition](<https://devfeed.tech/tags/composition.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [diagnostic](<https://devfeed.tech/tags/diagnostic.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [mind](<https://devfeed.tech/tags/mind.md>), [robotic](<https://devfeed.tech/tags/robotic.md>), [robotic-manipulation](<https://devfeed.tech/tags/robotic-manipulation.md>), [world](<https://devfeed.tech/tags/world.md>)

### AI overview

Mind2Act World is a benchmark for diagnosing whether robotic manipulation failures stem from reasoning, physical execution, or combining the two through environmental feedback. It contains 30 simulated tasks across five application domains, with three difficulty levels and Clean or Random scenes (180 configurations). Its Mind view tests decisions from past observations, Act tests precise and timely execution, and Mind2Act tests closed-loop tasks where action outcomes inform later decisions; a diagnosis harness attributes failures to these capabilities or their composition. In evaluations of multimodal agents, vision-language-action models, and world-action models, success in individual views did not transfer to combined tasks: the strongest agent succeeded in 52.7% of Mind episodes and 39.0% of Act episodes, but only 12.0% of Mind2Act episodes. The abstract reports that this agent retained task state but struggled with sustained contact, while end-to-end policies performed practiced motions but retained little task state.

### Source excerpt

arXiv:2611.26444v1 Announce Type: cross Abstract: Embodied intelligence aims to let robots perform complex physical tasks with human-like adaptability and reliability, which requires three complementary capabilities: history-based cognitive reasoning, precise and timely physical execution, and the composition of reasoning and execution through environmental feedback. Yet existing benchmarks either evaluate reasoning and execution within the same tasks or assess them separately without their composition, so they reveal little about whether a failure comes from reasoning, execution, or their composition. We introduce Mind2Act World, a diagnostic benchmark of 30 simulated manipulation tasks from five real-world application domains. Each task has three difficulty levels and two scene settings (Clean and Random), yielding 180 configurations. The tasks form three views: Mind tests decisions based on past observations, Act tests precise and timely execution, and Mind2Act combines both in closed-loop tasks where action outcomes guide later decisions. Comparing the three views shows whether a system lacks a capability or fails to compose capabilities it already has. A diagnosis harness further attributes the failed episodes of three representative systems to Mind, Act, or their composition and summarizes each system in a profile card. Our evaluation of multimodal agents, vision-language-action models (VLAs), and world-action models (WAMs) shows that competence in a single view does not carry over to Mind2Act tasks. Even the strongest agent succeeds in 52.7% of Mind and 39.0% of Act episodes but in only 12.0% of Mind2Act episodes. The strongest agent retains task state but struggles with sustained contact, whereas end-to-end policies execute practiced motions but retain little task state. These findings point to composing reasoning and execution through environmental feedback as the main open problem.

## Task Convergent Flow Matching for Generative Robotics

DevFeed: [Task Convergent Flow Matching for Generative Robotics](<https://devfeed.tech/articles/task-convergent-flow-matching-for-generative-robotics-141794.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22609>)

Author: Lorenzo Paiola, Maximilian St\\"olzle, Giorgio Grioli, Daniela Rus, Cosimo Della Santina

Published: 2026-11-18T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [generative-ai-tools](<https://devfeed.tech/topics/generative-ai-tools.md>)

Tags: [continuous](<https://devfeed.tech/tags/continuous.md>), [convergence](<https://devfeed.tech/tags/convergence.md>), [covariance](<https://devfeed.tech/tags/covariance.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [funnel](<https://devfeed.tech/tags/funnel.md>), [generative](<https://devfeed.tech/tags/generative.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [inference](<https://devfeed.tech/tags/inference.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [motion](<https://devfeed.tech/tags/motion.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [performance](<https://devfeed.tech/tags/performance.md>), [physics](<https://devfeed.tech/tags/physics.md>), [progress](<https://devfeed.tech/tags/progress.md>), [quantile](<https://devfeed.tech/tags/quantile.md>), [robotics](<https://devfeed.tech/tags/robotics.md>), [search](<https://devfeed.tech/tags/search.md>), [source](<https://devfeed.tech/tags/source.md>), [stacking](<https://devfeed.tech/tags/stacking.md>)

### AI overview

The framework refines flow-matching robot policies at inference time by testing candidate action chunks in parallel physics-based rollouts and searching for chunks that make task progress. Performance funnels allow bounded temporary regressions but require net progress; interval-wise Lipschitz bounds are used to verify continuous-time funnel satisfaction from sampled rollouts. For contact-rich dynamics, CMA-ES searches source seeds within radial quantile bands to help retain the learned motion prior. The method supports Euclidean and pose-valued Riemannian Flow Matching policies. Reported success rises from 31.3% to 88.5% on PushT and from 77.0% to 86.0% on three-cube stacking. On PushT, a learned funnel reduces average maximum loss of achieved progress by 19% versus checking chunk endpoints alone, at similar success. Radial-band search yields 11.3% out-of-distribution motion windows, compared with 34.1% for unconstrained search.

### Source excerpt

arXiv:2611.22609v1 Announce Type: cross Abstract: Successful manipulation requires the manipulated objects to progress toward a desired task outcome. Generative policies can produce plausible robot motions without ensuring this progress, allowing successive action chunks to stall or undo earlier changes to the scene. We introduce a task-centric inference-time refinement framework for flow matching policies that evaluates candidate action chunks through parallel physics-based rollouts and searches for motions satisfying sufficient conditions for task convergence. Task-level performance funnels bound temporary regressions while requiring net progress over each chunk. We establish sufficient conditions under which repeated funnel satisfaction yields convergence throughout execution, and use interval-wise Lipschitz bounds to verify continuous-time funnel satisfaction from sampled rollouts. To accommodate contact-rich hybrid dynamics, we refine the flow's source seeds using derivative-free Covariance Matrix Adaptation Evolution Strategy (CMA-ES), constrained to radial quantile bands of the source distribution to help retain the learned motion prior. This formulation supports both Euclidean and pose-valued Riemannian Flow Matching (RMFM) policies. On PushT and three-cube stacking, refinement raises success from 31.3% to 88.5% and from 77.0% to 86.0%. On PushT, whose coverage score responds continuously to the motion of the block, a learned funnel lowers the average maximum loss of achieved progress by 19% relative to constraining the endpoint of each chunk alone, at a similar success rate. With the radial band, 11.3% of the executed motion windows are out of distribution, a rate within the range of held-out demonstrations, compared with 34.1% for an unconstrained search.

## JointGV: Structure-Aware Correction of Visual World Models with Parallel Graph Dynamics for Long-Horizon Modeling

DevFeed: [JointGV: Structure-Aware Correction of Visual World Models with Parallel Graph Dynamics for Long-Horizon Modeling](<https://devfeed.tech/articles/jointgv-structure-aware-correction-of-visual-world-models-with-parallel-graph-dynamics-for-long-horizon-modeling-140048.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22543>)

Author: Tian Lyu, Yichen Jiang, Ziyuan Cheng, Yunda Yan, Xiangyu Pan, Junjie Jiang, Jialin Yu, Boyu Ning, Pietro Li\\\`o, Fan Mo

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [World models](<https://devfeed.tech/topics/world-models.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Graphs](<https://devfeed.tech/topics/graphs.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [actor](<https://devfeed.tech/tags/actor.md>), [contact-rich-manipulation](<https://devfeed.tech/tags/contact-rich-manipulation.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [graph](<https://devfeed.tech/tags/graph.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [structure](<https://devfeed.tech/tags/structure.md>), [world-models](<https://devfeed.tech/tags/world-models.md>)

### AI overview

JointGV is a plug-in that uses a task-conditioned scene graph to correct long-horizon drift in visual world models without sending graph-learning or correction gradients into the original visual dynamics. It rolls out graph and visual predictions in parallel, couples their predicted changes, and uses graph-motion context and predicted node positions to compute a bounded correction. On DINO-WM and DreamerV3, 32-step evaluations report 11-28% lower joint-position error. For DINO-WM spatial tokens, learned graph-visual change coupling outperforms giving every token the same mean graph-motion context, especially when distinct parts look close together. On real-robot inverted-pendulum sequences, DINO-WM's pole-angle error falls from 12.56° to 6.69°; DreamerV3 and LeWorldModel also improve. In simulated pendulum control, an actor using corrected rollouts completes 46 of 50 episodes, versus none with original rollouts. The supplied excerpt does not provide further methodological or evaluation details.

### Source excerpt

arXiv:2611.22543v1 Announce Type: cross Abstract: Visual world models predict future visual states conditioned on actions, but recursive errors in entity poses and spatial relations limit their long-horizon prediction for robot planning and control. We ask whether a separately predicted relational state can correct this drift without restructuring the visual model. This is challenging because graph nodes and visual states share neither elements nor a feature space. We propose Joint Graph--Visual Dynamics Correction (JointGV), a plug-in based on a cross-granularity dynamics hypothesis: despite their different representations, graph-node and visual-state changes describe the same physical transition. JointGV rolls out a task-conditioned scene graph alongside the visual world model, couples their predicted changes, and combines the resulting graph-motion context with graph-predicted node positions to compute a bounded residual. Graph-learning and correction gradients do not propagate into the original visual dynamics. We attach JointGV to DINO-WM (spatial tokens) and DreamerV3 (a global latent state), evaluating both on DMC Cheetah Run and Walker Walk across three seeds and extending the evaluation to contact-rich manipulation with DMC Finger Turn Hard. Across 32-step rollouts, JointGV improves structural and visual-prediction accuracy, reducing joint-position error by 11--28\%. For spatial tokens, learned graph--visual change coupling gives lower error than broadcasting the same mean graph-motion context to every token, particularly when non-adjacent parts appear close together in the image. On real-robot inverted-pendulum sequences, JointGV reduces DINO-WM's 32-step pole-angle error from $12.56^\circ$ to $6.69^\circ$ and improves DreamerV3 and LeWorldModel as well. In simulated inverted-pendulum control, an actor trained with JointGV-corrected rollouts completes 46 of 50 episodes, compared with none using the original rollouts.

## RUMI: A Robotiq-Matched Universal Manipulation Interface for Robot Learning with Augmented Reality and Haptic Feedback

DevFeed: [RUMI: A Robotiq-Matched Universal Manipulation Interface for Robot Learning with Augmented Reality and Haptic Feedback](<https://devfeed.tech/articles/rumi-a-robotiq-matched-universal-manipulation-interface-for-robot-learning-with-augmented-reality-and-haptic-feedback-132173.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.15802>)

Author: Kevin Avdiu, Seongjin Bien, Yannik Blei, Philine Meister, Johannes Hechtl, Philipp Schmitt, Kai Wurm, Georg von Wichert, Wolfram Burgard

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Glasses & AR Tech](<https://devfeed.tech/topics/glasses-ar-tech.md>), [Virtual reality](<https://devfeed.tech/topics/virtual-reality.md>), [Robotics Simulation](<https://devfeed.tech/topics/robotics-simulation.md>)

Tags: [augmented-reality](<https://devfeed.tech/tags/augmented-reality.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [haptic-feedback](<https://devfeed.tech/tags/haptic-feedback.md>), [interface](<https://devfeed.tech/tags/interface.md>), [learning](<https://devfeed.tech/tags/learning.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [robot](<https://devfeed.tech/tags/robot.md>)

### AI overview

RUMI is a handheld interface matched to the Robotiq 2F-85 that preserves its gripper geometry and uses exchangeable fingertip sensors for natural input before contact and force measurement after contact. Tracking integrated into a VR headset provides real-time augmented reality feedback on whether motions are kinematically feasible for the target robot; a haptic mechanism supports force-aware teleoperation with simulated and real robots. In a user study, episodes with joint-limit violations were fewer with live robot visualization and pose initialization (3/50) than without augmented reality (16/50). The authors also trained imitation policies from demonstrations collected with two fingertip modalities and assessed force-aware data collection in simulation. The supplied text is an abstract excerpt, so it does not provide implementation details or broader evaluation results.

### Source excerpt

arXiv:2611.15802v1 Announce Type: cross Abstract: Many manipulation tasks require carefully controlled contact forces for successful execution, yet large-scale, force-aware datasets remain scarce. Handheld interfaces can help by transmitting interaction forces directly through the operator's arm, but most existing designs are tailored to simple linear grippers and do not match the nonlinear mechanics of widely used robotic grippers. We present RUMI, a handheld universal manipulation interface matched to the Robotiq 2F-85 with exchangeable fingertip sensors that preserves the target gripper geometry while enabling natural input before contact and force measurement after contact. By integrating the tracking system into a virtual reality headset, our method provides real-time augmented reality feedback about the kinematic feasibility of motion on the target robot platform. We further introduce a haptic feedback mechanism that supports force-aware teleoperation for simulated and real robots. A user study shows fewer episodes containing joint-limit violations with live robot visualization and pose initialization (3/50) than without augmented reality (16/50). We additionally train imitation policies from demonstrations, collected using two different fingertip modalities enabled by the gripper-matched hardware, and assess the haptic interface's capability for collecting force-aware teleoperation datasets in simulation.

## DexRemo: A Force-Sensing Hand Platform and Task Suite for Contact-Rich Removal Manipulation

DevFeed: [DexRemo: A Force-Sensing Hand Platform and Task Suite for Contact-Rich Removal Manipulation](<https://devfeed.tech/articles/dexremo-a-force-sensing-hand-platform-and-task-suite-for-contact-rich-removal-manipulation-133057.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.15981>)

Author: Liman Wang, Hanyang Zhong, Sun Zhaole, Wenting Cao, Babar Jamil, Andy Tyrrell, Shan Luo, Jihong Zhu

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [Machine Perception](<https://devfeed.tech/topics/machine-perception.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [axis](<https://devfeed.tech/tags/axis.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [components](<https://devfeed.tech/tags/components.md>), [core](<https://devfeed.tech/tags/core.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [easier](<https://devfeed.tech/tags/easier.md>), [embedded](<https://devfeed.tech/tags/embedded.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [evaluations](<https://devfeed.tech/tags/evaluations.md>), [express](<https://devfeed.tech/tags/express.md>), [extension](<https://devfeed.tech/tags/extension.md>), [frame](<https://devfeed.tech/tags/frame.md>), [fuse](<https://devfeed.tech/tags/fuse.md>), [giving](<https://devfeed.tech/tags/giving.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [platform](<https://devfeed.tech/tags/platform.md>), [rich](<https://devfeed.tech/tags/rich.md>), [sensing](<https://devfeed.tech/tags/sensing.md>)

### AI overview

DexRemo pairs a high-force, three-finger robotic hand with modular fingertips and three-axis force sensing, and evaluates removal manipulation across nine real-robot scenarios. Its learned policies combine multi-view vision, proprioception, fingertip pose, and force measurements to predict closed-loop action chunks from demonstrations. Expressing fingertip forces in a shared wrist-centered frame makes force directions consistent across fingers; ablations on six core tasks report improved stability and generalization, especially when wrist-frame forces are combined with fingertip pose, compared with local-frame force inputs or no-force baselines. Three additional scenarios provide held-out coverage.

### Source excerpt

arXiv:2611.15981v1 Announce Type: cross Abstract: Many manipulation tasks require separating adhered, embedded, contaminated, or mechanically constrained targets. We study this setting as removal manipulation, where progress depends on evolving contact states that are only partially observable from vision and kinematics. We introduce DexRemo, a three-finger robotic hand with high-force actuation, modular fingertips, and three-axis fingertip force sensing, together with a nine-scenario real-robot task suite spanning adhesive detachment, surface cleaning and self-cleaning, and occluded retrieval. To make multi-finger force observations easier to fuse, we express fingertip forces in a shared wrist-centered frame, giving force components consistent geometric meaning across fingers and time. Using an Action Chunking Transformer backbone, policies fuse multi-view vision, proprioception, fingertip pose, and force observations to predict closed-loop action chunks from demonstrations. Evaluation uses six core benchmark tasks for controlled fingertip-input and force-frame ablations, plus three extension scenarios for additional held-out coverage. Across these evaluations, the learned policies achieve consistent removal performance, and ablations show that wrist-frame forces, especially with fingertip pose, improve stability and generalization over local-frame force inputs and no-force baselines. Project website: https://dexremo.github.io

## Learning Dexterous Manipulation Using Contact Wrench Guidance From Human Demonstration

DevFeed: [Learning Dexterous Manipulation Using Contact Wrench Guidance From Human Demonstration](<https://devfeed.tech/articles/learning-dexterous-manipulation-using-contact-wrench-guidance-from-human-demonstration-128126.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.00033>)

Author: Xinghao Zhu, Zixi Liu, Shalin Jain, Chenran Li, Milad Noori, Michael Andres Lin, Huihua Zhao, John Welsh, Mrinal Verghese, Wei Liu, Tingwu Wang, Xingye Da, Zhengyi Luo, Vishal Kulkarni, Naema Bhatti,

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [dexterous-robot-manipulation](<https://devfeed.tech/tags/dexterous-robot-manipulation.md>), [learning](<https://devfeed.tech/tags/learning.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [motion-capture](<https://devfeed.tech/tags/motion-capture.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [simulation](<https://devfeed.tech/tags/simulation.md>)

### AI overview

CHORD guides reinforcement learning for long-horizon dexterous manipulation by comparing human and robot actions through the forces and torques they induce on an object, linking contact-wrench similarity to instantaneous object motion. The authors report a simulation benchmark of 4,739 bimanual tasks assembled from motion-capture data and reconstructed videos. On 1,831 evaluated tasks, CHORD achieves an average 82.12% success rate; for whole-body manipulation using hand-only and third-person demonstrations, it reports 90.77%. The abstract also states that learned policies transfer to real-world tasks in open-loop and closed-loop settings.

### Source excerpt

arXiv:2607.00033v3 Announce Type: replace-cross Abstract: Dexterous robot manipulation can benefit from the abundance of human demonstrations, but transferring such demonstrations to robot policies remains challenging. We present Contact Wrench Guidance from Human Demonstration in Robotic Dexterous Manipulation (CHORD), a framework for long-horizon manipulation of rigid and articulated objects with reinforcement learning. The key idea is object-centric contact wrench space guidance: we represent human and robot motions by the forces and torques they can induce on the object, enabling similarity to be measured by the induced instantaneous motions. This guidance makes reinforcement learning more scalable for contact-rich dexterous manipulation. We further introduce a large-scale simulation benchmark with 4,739 bimanual dexterous manipulation tasks, constructed from motion-capture datasets and reconstructed in-house videos. Evaluated on 1,831 benchmark tasks, CHORD achieves an average success rate of 82.12%, demonstrating strong scalability. CHORD also generalizes to whole-body manipulation from hand-only and third-person demonstrations, achieving a 90.77% success rate, and the learned policies transfer to the real world in both open-loop and closed-loop settings.

## InfCoiL: Coordinated Planner-Controller Learning for Closed-Loop Physics-Based Human-Object Interaction

DevFeed: [InfCoiL: Coordinated Planner-Controller Learning for Closed-Loop Physics-Based Human-Object Interaction](<https://devfeed.tech/articles/infcoil-coordinated-planner-controller-learning-for-closed-loop-physics-based-human-object-interaction-125399.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.11517>)

Author: Yude Zou, Yifei Yao, Yiwei Hao, Hanqing Wang, Xing Gao

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>)

Tags: [animation](<https://devfeed.tech/tags/animation.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [controller](<https://devfeed.tech/tags/controller.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [flow](<https://devfeed.tech/tags/flow.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [framework](<https://devfeed.tech/tags/framework.md>), [generative](<https://devfeed.tech/tags/generative.md>), [humanobject-interaction](<https://devfeed.tech/tags/humanobject-interaction.md>), [interface](<https://devfeed.tech/tags/interface.md>), [learning](<https://devfeed.tech/tags/learning.md>), [online](<https://devfeed.tech/tags/online.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [physics](<https://devfeed.tech/tags/physics.md>), [planner](<https://devfeed.tech/tags/planner.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [semantic](<https://devfeed.tech/tags/semantic.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [sequences](<https://devfeed.tech/tags/sequences.md>), [systems](<https://devfeed.tech/tags/systems.md>), [text](<https://devfeed.tech/tags/text.md>)

### AI overview

InfCoiL trains a human-object interaction planner and physics-based controller together in a closed loop, allowing execution feedback to reshape future plans. A morphology-aware flow-matching planner and a unified multi-morphology controller share an interface and are alternately fine-tuned with reinforcement learning on the same physical rollouts. Its Semantic Flow Policy Optimization method stochasticizes selected denoising steps and uses group-relative, sequence-level rewards from complete executions. On held-out OMOMO sequences, the abstract reports higher goal-reaching success and physical tracking accuracy than open-loop hierarchical baselines.

### Source excerpt

arXiv:2611.11517v1 Announce Type: cross Abstract: Text- and goal-conditioned human-object interaction (HOI) provides an intuitive interface for specifying high-level intents in physics-based character animation. Mapping such intents to physically plausible, contact-rich behaviors naturally calls for a generative planner and a physics-based controller, yet existing planner-controller systems either execute plans open-loop or close the loop while keeping the planner fixed, so execution feedback never reshapes what is planned. We present InfCoiL, a closed-loop framework for physics-based HOI from coordinated planner-controller learning. InfCoiL couples a morphology-aware flow-matching planner with a unified multi-morphology controller through a shared morphology interface in an autoregressive plan-and-execute loop, and alternately fine-tunes both components with reinforcement learning from the same physical rollouts. To optimize the flow-matching planner online, we introduce Semantic Flow Policy Optimization (SFPO), which stochasticizes selected denoising transitions with channel-wise semantic covariance over heterogeneous HOI features and assigns group-relative, sequence-level rewards from complete closed-loop executions. On held-out OMOMO sequences, InfCoiL outperforms open-loop hierarchical baselines in goal-reaching task success and physical tracking accuracy. Project page: https://infcoil.github.io/

## Grasp-TAGS: Task-Aware Grasp Selection via Critic Distillation for Prehensile Loco-Manipulation

DevFeed: [Grasp-TAGS: Task-Aware Grasp Selection via Critic Distillation for Prehensile Loco-Manipulation](<https://devfeed.tech/articles/grasp-tags-task-aware-grasp-selection-via-critic-distillation-for-prehensile-loco-manipulation-123466.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.09864>)

Author: Xun Tu, Minghao Zou, Karthik Desingh

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Post-training optimization](<https://devfeed.tech/topics/post-training-optimization.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [distillation](<https://devfeed.tech/tags/distillation.md>), [framework](<https://devfeed.tech/tags/framework.md>), [learning](<https://devfeed.tech/tags/learning.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [mobile-manipulators](<https://devfeed.tech/tags/mobile-manipulators.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [network](<https://devfeed.tech/tags/network.md>), [orientation](<https://devfeed.tech/tags/orientation.md>), [ppo](<https://devfeed.tech/tags/ppo.md>), [random](<https://devfeed.tech/tags/random.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [results](<https://devfeed.tech/tags/results.md>), [robotics](<https://devfeed.tech/tags/robotics.md>), [selection](<https://devfeed.tech/tags/selection.md>), [warehouse-automation](<https://devfeed.tech/tags/warehouse-automation.md>)

### AI overview

Grasp-TAGS selects an initial grasp for goal-conditioned large-object rearrangement before contact, aiming to improve downstream loco-manipulation. It trains a shared PPO policy across multiple initial grasps, then distills its critic into a lightweight network that ranks candidate grasps using grasp configuration and goal information. In experiments, the approach achieved 89.84% average success, versus 79.87% for the best fixed single-grasp policies, and reduced final orientation errors across all evaluated objects. The abstract also reports improvement over random grasp selection but gives no numerical comparison for that baseline.

### Source excerpt

arXiv:2611.09864v1 Announce Type: cross Abstract: Loco-manipulation, which requires mobile manipulators to coordinate locomotion and manipulation over long horizons, is a key capability for large-object rearrangement in applications such as service robotics and warehouse automation. Despite recent progress, reliably executing these tasks remains challenging because performance depends on complex robot-object and object-environment interaction dynamics. Existing approaches address this challenge either through model-based methods, which often require accurate dynamics models, or through learning-based methods, which reduce this requirement but typically assume a fixed or predefined initial grasp. However, different geometrically stable grasps can lead to substantially different downstream manipulation performance. Therefore, we propose Grasp-TAGS, a task-aware grasp selection framework for goal-conditioned large-object rearrangement. We train a unified PPO-based reinforcement learning policy across multiple initial grasp configurations and distill the learned critic into a lightweight grasp utility network that ranks candidate grasps using only the candidate grasp configuration and goal information. At inference time, we select the grasp with the highest predicted downstream utility for the downstream policy to complete the task, before even making a contact. Experiments demonstrate that Grasp-TAGS improves rearrangement success over both fixed single-grasp policies and random grasp selection, achieving an average success rate of 89.84% compared with 79.87% for the best-performing single-grasp policies, an improvement of 9.97 percentage points. Grasp-TAGS also reduces final orientation errors across all evaluated objects. For more qualitative results, refer to the supplementary webpage: https://rpm-lab-umn.github.io/Grasp-TAGS-web-page/

## NoContactNoWorries: Estimating Contact through Vision and Proprioception for In-Hand Dexterous Manipulation

DevFeed: [NoContactNoWorries: Estimating Contact through Vision and Proprioception for In-Hand Dexterous Manipulation](<https://devfeed.tech/articles/nocontactnoworries-estimating-contact-through-vision-and-proprioception-for-in-hand-dexterous-manipulation-124446.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.24450>)

Author: Soham Patil, Avirup Das, Sourabh Bhosale, Spandan Roy

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [novel](<https://devfeed.tech/tags/novel.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [robot](<https://devfeed.tech/tags/robot.md>), [sensors](<https://devfeed.tech/tags/sensors.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [transformer](<https://devfeed.tech/tags/transformer.md>), [vision](<https://devfeed.tech/tags/vision.md>)

### AI overview

NoContactNoWorries uses a transformer to combine RGB-D vision and robot proprioception to estimate binary hand-object contact as a pseudo-tactile signal. The authors train one contact predictor across multiple objects and report that its output supports reinforcement learning for in-hand object reorientation, including generalization to novel objects. They validate the approach in simulation and on a real-world robot. The abstract frames this as a scalable alternative for binary contact estimation, citing cost, fragility, and integration challenges with dedicated tactile hardware; it does not establish broader tactile replacement.

### Source excerpt

arXiv:2606.24450v2 Announce Type: replace-cross Abstract: Perceiving physical contact is fundamental to dexterous manipulation. While robots often rely on dedicated hardware tactile sensors, humans exhibit a remarkable ability to infer contact by integrating visual information with an innate sense of their body's pose and movement. Inspired by this embodied perceptual skill, we investigate whether a robot can learn to infer contact from vision, an approach that also offers a scalable alternative to tactile hardware specifically for binary contact estimation, which faces practical challenges in cost, fragility, and integration. We present NoContactNoWorries, a transformer-based multimodal framework that fuses RGB-D vision with the robot's proprioception to infer binary contact states as a pseudo-tactile signal for hand-object interactions. We validate by training a single contact prediction model on multiple objects and show that the inferred contact signal supports downstream reinforcement learning agents for in-hand object reorientation, generalizing to novel objects. Experiments in both simulation and on a real-world robot validate our approach, highlighting the feasibility of inferring contact from vision and proprioception. Project Page: https://soham2560.github.io/no-contact-no-worries/

## MATE: Asynchronous Multi-Rate Diffusion Policy Composition

DevFeed: [MATE: Asynchronous Multi-Rate Diffusion Policy Composition](<https://devfeed.tech/articles/mate-asynchronous-multi-rate-diffusion-policy-composition-124456.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.17257>)

Author: Zihao He, Hongjie Fang, Shirun Tang, Cewu Lu, Haoshu Fang

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>)

Tags: [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [composition](<https://devfeed.tech/tags/composition.md>), [contact-rich-manipulation](<https://devfeed.tech/tags/contact-rich-manipulation.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [denoising](<https://devfeed.tech/tags/denoising.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [latency](<https://devfeed.tech/tags/latency.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [performance](<https://devfeed.tech/tags/performance.md>), [policies](<https://devfeed.tech/tags/policies.md>), [policy](<https://devfeed.tech/tags/policy.md>), [vision](<https://devfeed.tech/tags/vision.md>)

### AI overview

LAG-Fusion composes multimodal diffusion policies asynchronously, letting each modality-specific policy run at its native inference rate and provide denoising guidance when available. To align delayed guidance under relative action representations, it rebases diffusion variables into a shared reference frame before fusion. The authors apply this approach to contact-rich manipulation by combining low-frequency vision with high-frequency force input. In experiments with heterogeneous modality latencies, they report better policy responsiveness and task performance than synchronous fusion and specially designed force-aware baselines.

### Source excerpt

arXiv:2607.17257v2 Announce Type: replace-cross Abstract: Diffusion policies have shown strong potential for robotic imitation learning, and recent extensions incorporate additional modalities to improve manipulation performance. However, these modalities often differ not only in information content but also in sensing rates and inference latencies. Existing multimodal diffusion policies typically rely on synchronous fusion or manually designed multi-frequency architectures, which either slow down high-frequency feedback or limit extensibility to new modality combinations. We propose LAG-Fusion, a latency-aware guidance fusion framework for asynchronous multimodal diffusion policy composition. LAG-Fusion allows modality-specific policies to operate at their native inference rates and contribute denoising guidance whenever available. To make asynchronous composition consistent, we derive a reference-frame rebasing rule for diffusion variables under relative action representations, enabling delayed guidance to be aligned before fusion. We instantiate LAG-Fusion in contact-rich manipulation by composing a low-frequency vision policy with a high-frequency force policy. Experiments under heterogeneous modality latencies show that LAG-Fusion improves policy responsiveness and task performance over synchronous fusion and specially designed force-aware baselines.

## VTAM: Efficient Visuo-Tactile World-Action Models for Contact-Rich Manipulation without Tactile Pretraining

DevFeed: [VTAM: Efficient Visuo-Tactile World-Action Models for Contact-Rich Manipulation without Tactile Pretraining](<https://devfeed.tech/articles/vtam-efficient-visuo-tactile-world-action-models-for-contact-rich-manipulation-without-tactile-pretraining-120007.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.23481>)

Author: Haoran Yuan, Weigang Yi, Zhenyu Zhang, Wendi Chen, Yuchen Mo, Jiashi Yin, Xinzhuo Li, Yue Luo, Jialin Li, Xiangyu Zeng, Chuan Wen, Cewu Lu, Katherine Driggs-Campbell, Ismini Lourentzou

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [World models](<https://devfeed.tech/topics/world-models.md>)

Tags: [contact-rich-manipulation](<https://devfeed.tech/tags/contact-rich-manipulation.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [flow](<https://devfeed.tech/tags/flow.md>), [inference](<https://devfeed.tech/tags/inference.md>), [latency](<https://devfeed.tech/tags/latency.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [pretraining](<https://devfeed.tech/tags/pretraining.md>), [robot](<https://devfeed.tech/tags/robot.md>), [sensors](<https://devfeed.tech/tags/sensors.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [transformer](<https://devfeed.tech/tags/transformer.md>), [vae](<https://devfeed.tech/tags/vae.md>), [vision-language-action](<https://devfeed.tech/tags/vision-language-action.md>), [world-model](<https://devfeed.tech/tags/world-model.md>), [zero-shot](<https://devfeed.tech/tags/zero-shot.md>)

### AI overview

VTAM is a visuo-tactile world-action model for contact-rich manipulation that uses task demonstrations without tactile or action pretraining. It first adapts a pretrained video backbone to predict future visual and tactile latents, then fine-tunes it with an action expert that predicts actions, proprioceptive states, and a flow-force proxy from tactile optical flow. The abstract reports better results than all baselines on four real-robot tasks across two platforms and higher average success than baselines without tactile pretraining on eight UniVTAC tasks. Its reported UniVTAC success is comparable to models trained on over 30,000 hours of largely visuo-tactile data: 83.4% versus 84.5% for N0-TWAM. At inference, it is about 6x faster than N0-TWAM and 16x faster than LingBot-VA without compilation; compiled latency is comparable to Fast-WAM. Ablations found that bypassing tactile world modeling failed on every tested task, while the flow-force proxy helped prevent the policy from ignoring touch. The tactile encoder transferred zero-shot between GelSight Mini and Xense Photon sensors.

### Source excerpt

arXiv:2603.23481v2 Announce Type: replace-cross Abstract: Tactile vision-language-action models (VLAs) and tactile world-action models (WAMs) are a promising new paradigm for contact-rich manipulation, but existing models either underutilize touch, rely on large-scale tactile pretraining, or incur high inference latency due to iterative video generation. We present VTAM, an efficient visuo-tactile world-action model for contact-rich manipulation that requires no tactile or action pretraining. RGB and tactile observations share a frozen video VAE and the pretrained video transformer weights. VTAM first adapts the video backbone to jointly predict future visual and tactile latents, then fine-tunes it end-to-end with a randomly initialized action expert that predicts actions, proprioceptive states, and a light flow-force proxy computed from tactile optical flow. Trained only on task demonstrations, VTAM outperforms all baselines on four real-robot tasks across two platforms with different tactile sensors, outperforms all baselines without tactile pretraining in average success on eight UniVTAC simulation tasks, and is comparable to tactile VLAs and WAMs pretrained on over 30K hours of largely visuo-tactile robot data (83.4% vs. 84.5% for N0-TWAM). Without full video denoising at inference, VTAM is approximately 6x and 16x faster than N0-TWAM and LingBot-VA without compilation, with compiled latency comparable to Fast-WAM. Ablations show that feeding frozen-VAE tactile features directly to the action expert, bypassing tactile world modeling, fails on every tested task, that Stage I visuo-tactile world-model fine-tuning improves UniVTAC success on most tasks, and that flow-force proxy keeps the policy from fully ignoring touch. The tactile encoder transfers zero-shot across GelSight Mini and Xense Photon sensors.

## DexCanvas: Bridging Human Demonstrations and Robot Learning for Dexterous Manipulation

DevFeed: [DexCanvas: Bridging Human Demonstrations and Robot Learning for Dexterous Manipulation](<https://devfeed.tech/articles/dexcanvas-bridging-human-demonstrations-and-robot-learning-for-dexterous-manipulation-116645.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2510.15786>)

Author: Xinyue Xu (Daisy), Jieqiang Sun (Daisy), Jing (Daisy), Dai, Siyuan Chen, Lanjie Ma, Ke Sun, Bin Zhao, Jianbo Yuan, Sheng Yi, Haohua Zhu, Irvine Lu

Published: 2026-11-04T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [hand-tracking](<https://devfeed.tech/tags/hand-tracking.md>), [hugging-face](<https://devfeed.tech/tags/hugging-face.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [learning](<https://devfeed.tech/tags/learning.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [motion-capture](<https://devfeed.tech/tags/motion-capture.md>), [optical-motion-capture](<https://devfeed.tech/tags/optical-motion-capture.md>), [physics](<https://devfeed.tech/tags/physics.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [policies](<https://devfeed.tech/tags/policies.md>), [policy](<https://devfeed.tech/tags/policy.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [rl](<https://devfeed.tech/tags/rl.md>), [robot](<https://devfeed.tech/tags/robot.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [simulator](<https://devfeed.tech/tags/simulator.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>)

### AI overview

DexCanvas augments 70 hours of optical motion capture across 21 manipulation types into a reported 7,000 hours of hand-object interaction with per-frame contact points and forces. A reinforcement learning policy drives an actuated MANO hand in simulation to reproduce captured object trajectories, while the simulator measures contacts. It reproduces 87.5% of demonstrations, succeeds on 80.9% of rollouts from perturbed initial poses, and succeeds 72.5% on six YCB objects absent from capture. In 18,522 trials, a Diffusion Policy using DexCanvas contact features succeeds 82.2% of the time versus 67.5% without them. Contact-driven retargeting is also reported for two robot hands. The excerpt says a first release includes 111,857 force-annotated trajectories on Hugging Face.

### Source excerpt

arXiv:2510.15786v3 Announce Type: replace-cross Abstract: A dexterous hand succeeds or fails through contact, yet every scalable record of human manipulation (video, vision-based hand tracking, motion capture) observes how the hand moves and misses how it pushes. We present DexCanvas, a hybrid real-synthetic dataset of 7,000 hours of dexterous hand-object interaction with per-frame contact points and contact forces, grown from 70 hours of optical motion capture that spans 21 manipulation types, from power and precision grasps to in-hand manipulation. The forces come from a real-to-sim pipeline: one reinforcement learning (RL) policy, conditioned on object geometry and manipulation type, drives an actuated MANO hand in physics simulation to reproduce each captured object trajectory, and the simulator measures the contacts that produce the reproduced motion. The policy reproduces 87.5% of the demonstrations and still succeeds on 80.9% of rollouts from perturbed initial object poses; these perturbed rollouts expand the 70 captured hours a hundredfold. On six YCB objects absent from the capture set it succeeds 72.5% of the time. The contact channels make better policies: a Diffusion Policy that observes DexCanvas contact features succeeds 82.2% of the time, against 67.5% without them, over 18,522 trials. Contact-driven retargeting carries the same demonstrations to two dexterous robot hands, where the RL stage again produces stable rollouts. A first release of 111,857 force-annotated trajectories is publicly available on Hugging Face.

## Closing the Dynamics Gap through Self-Directed Robotic Interaction

DevFeed: [Closing the Dynamics Gap through Self-Directed Robotic Interaction](<https://devfeed.tech/articles/closing-the-dynamics-gap-through-self-directed-robotic-interaction-115355.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.02691>)

Author: Xilun Zhang, Liam Achenbach, Luis Pabon, Matthew Foutter, Marco Pavone

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Robotics Simulation](<https://devfeed.tech/topics/robotics-simulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [Physical AI](<https://devfeed.tech/topics/physical-ai.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [billiards](<https://devfeed.tech/tags/billiards.md>), [camera](<https://devfeed.tech/tags/camera.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [gap](<https://devfeed.tech/tags/gap.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [physics](<https://devfeed.tech/tags/physics.md>), [pool](<https://devfeed.tech/tags/pool.md>), [robotic](<https://devfeed.tech/tags/robotic.md>), [robotics](<https://devfeed.tech/tags/robotics.md>), [simulation](<https://devfeed.tech/tags/simulation.md>)

### AI overview

CUE addresses the "dynamics gap" in Real2Sim2Real systems by learning contact physics through self-directed interaction, since camera observations alone do not reveal properties such as friction and restitution. Given a high-level instruction and no resets, it builds its own perception, estimates physical parameters, and chooses interactions where its simulation is least accurate. In billiards simulations across 25 hidden-physics conditions, CUE reduced held-out ball-position error by 72% relative to the strongest baseline; the authors also report deployment on a real pool table.

### Source excerpt

arXiv:2611.02691v1 Announce Type: cross Abstract: Agentic Real2Sim2Real systems can reconstruct a scene in simulation from camera observations, but contact-rich dynamics such as friction and restitution cannot be observed without interaction. We call this the dynamics gap and present CUE, an agentic system that closes it by interacting with the environment itself. Given only a high-level instruction and no resets, the agent builds its own perception, identifies physical parameters, and targets interactions where its simulation is least accurate. On billiards, CUE reduces held-out ball-position error by 72% relative to the strongest baseline across 25 hidden-physics conditions in simulation, and we further deploy it on a real pool table. More details can be found on the project website: https://cue-dynamics.github.io/

## FlowTouch: View-Invariant Visuo-Tactile Prediction

DevFeed: [FlowTouch: View-Invariant Visuo-Tactile Prediction](<https://devfeed.tech/articles/flowtouch-view-invariant-visuo-tactile-prediction-111385.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.08255>)

Author: Seongjin Bien, Carlo Kneissl, Tobias J\\"ulg, Frank Fundel, Thomas Ressler-Antal, Florian Walter, Bj\\"orn Ommer, Gitta Kutyniok, Wolfram Burgard

Published: 2026-11-02T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [Computer vision](<https://devfeed.tech/topics/computer-vision.md>), [3D](<https://devfeed.tech/topics/3d.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Image](<https://devfeed.tech/topics/image.md>)

Tags: [3d](<https://devfeed.tech/tags/3d.md>), [camera](<https://devfeed.tech/tags/camera.md>), [contact-rich-manipulation](<https://devfeed.tech/tags/contact-rich-manipulation.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [image-generation](<https://devfeed.tech/tags/image-generation.md>), [invariant](<https://devfeed.tech/tags/invariant.md>), [mesh](<https://devfeed.tech/tags/mesh.md>), [model](<https://devfeed.tech/tags/model.md>), [prediction](<https://devfeed.tech/tags/prediction.md>), [sensor](<https://devfeed.tech/tags/sensor.md>), [sim-to-real](<https://devfeed.tech/tags/sim-to-real.md>), [tasks](<https://devfeed.tech/tags/tasks.md>), [vision](<https://devfeed.tech/tags/vision.md>)

### AI overview

FlowTouch predicts tactile patterns from visual information to make tactile feedback useful before contact. It uses an object's local 3D mesh to reduce dependence on scene-specific camera views, combining scene reconstruction with Flow Matching-based image generation. The authors report bridging the sim-to-real gap, generalizing to new sensor instances, and using the predicted tactile images for downstream grasp stability prediction.

### Source excerpt

arXiv:2603.08255v2 Announce Type: replace-cross Abstract: Tactile sensation is essential for contact-rich manipulation tasks. It provides direct feedback on object geometry, surface properties, and interaction forces, enhancing perception and enabling fine-grained control. An inherent limitation of tactile sensors is that readings are available only when an object is touched. This precludes their use during planning and the initial execution phase of a task. Predicting tactile information from visual information can bridge this gap. A common approach is to learn a direct mapping from camera images to the output of vision-based tactile sensors. However, the resulting model will depend strongly on the specific setup and on how well the camera can capture the area where an object is touched. In this work, we introduce FlowTouch, a novel model for view-invariant visuo-tactile prediction. Our key idea is to use an object's local 3D mesh to encode rich information for predicting tactile patterns while abstracting away from scene-dependent details. FlowTouch integrates scene reconstruction and Flow Matching-based models for image generation. Our results show that FlowTouch is able to bridge the sim-to-real gap and generalize to new sensor instances. We further show that the resulting tactile images can be used for downstream grasp stability prediction. Our code, datasets and videos are available at https://flowtouch.github.io/

## Tactus: Tactile & Pose Sensing Glove for Natural Dexterity Data Collection

DevFeed: [Tactus: Tactile & Pose Sensing Glove for Natural Dexterity Data Collection](<https://devfeed.tech/articles/tactus-tactile-pose-sensing-glove-for-natural-dexterity-data-collection-108152.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.39076>)

Author: Li Guan, Rachael Granberry, Sahar Aseeri, Sahar Rostami, Taylor Niehues, Sai Dhanooshi Marri, Mia Huang, Dingtian Zhang, Ziyi Kou, Jason Pursel, Eric Whitmire, Vatsal Mehta, Ashish Deshpande, Tristan

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [World models](<https://devfeed.tech/topics/world-models.md>)

Tags: [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [capacitive](<https://devfeed.tech/tags/capacitive.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [data](<https://devfeed.tech/tags/data.md>), [data-collection](<https://devfeed.tech/tags/data-collection.md>), [dexterity](<https://devfeed.tech/tags/dexterity.md>), [hand-tracking](<https://devfeed.tech/tags/hand-tracking.md>), [imu](<https://devfeed.tech/tags/imu.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [sensing](<https://devfeed.tech/tags/sensing.md>), [textile](<https://devfeed.tech/tags/textile.md>), [wearable](<https://devfeed.tech/tags/wearable.md>), [world-models](<https://devfeed.tech/tags/world-models.md>)

### AI overview

Tactus combines a thin, textile-integrated glove with capacitive touch sensors and IMU-fused hand-pose tracking to record manipulation while aiming to preserve natural movement and track through hand-object occlusion. Its calibration pipeline supports consistency across gloves and correction for individual sessions. In standardized tests against two commercial tactile gloves, Tactus detected forces down to 0.59 N at 95% confidence--1.6-2.7 times lower than the baselines--and achieved 0.87 cross-participant consistency. During manipulation, fingertip keypoint relative error was 11.0 mm under the heaviest occlusion. The authors report that the combined signals support object-property estimation, slip detection, and hand-object interaction recordings for contact-aware robotics research.

### Source excerpt

arXiv:2610.39076v1 Announce Type: cross Abstract: Collecting naturalistic human manipulation demonstrations with accurate contact force and hand pose is essential for robot dexterous manipulation policy and foundational world model learning - yet no existing wearable platform provides both calibrated fingertip force sensing and occlusion-resilient hand tracking in a form factor that preserves natural manipulation behavior. We present Tactus, a textile-integrated sensing glove that co-optimizes three design pillars: (1) a thin, conformal 3D-knit form factor (1.5 mm fingertip stackup) designed to minimize interference with natural manipulation, (2) strategically placed capacitive tactile sensors at manipulation-relevant regions informed by empirical contact studies, and (3) IMU-fused pose tracking that maintains accuracy through the hand-object occlusions where manipulation data is most valuable. Tactus includes a documented calibration pipeline for cross-glove consistency and per-session calibration for subject-dependent correction. To enable objective comparison, we introduce standardized benchmarking protocols for tactile force regression and hand pose estimation, and apply them to Tactus and two commercially available tactile gloves - to our knowledge the first quantitative cross-platform evaluation of this kind. Tactus resolves a minimum detectable force of 0.59 N at 95% confidence, 1.6 - 2.7x lower than the commercial baselines, and with 0.87 cross-participant consistency, while our pose tracking sustains 11.0 mm fingertip keypoint relative error during manipulation, when occlusion is heaviest. We show that these combined signals support object property estimation, slip detection, and representative recording of hand-object interaction - yielding demonstration data with the fidelity required for contact-aware manipulation policies and world models.

## Learning from Relative Action Credit for Escaping Local Minima

DevFeed: [Learning from Relative Action Credit for Escaping Local Minima](<https://devfeed.tech/articles/learning-from-relative-action-credit-for-escaping-local-minima-108632.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.38034>)

Author: Giovanni Minelli, Giulio Turrisi, Lorenzo Amatucci, Victor Barasuol, Claudio Semini

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [all](<https://devfeed.tech/tags/all.md>), [alternatives](<https://devfeed.tech/tags/alternatives.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [better](<https://devfeed.tech/tags/better.md>), [budget](<https://devfeed.tech/tags/budget.md>), [can](<https://devfeed.tech/tags/can.md>), [combined](<https://devfeed.tech/tags/combined.md>), [compare](<https://devfeed.tech/tags/compare.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [data](<https://devfeed.tech/tags/data.md>), [discrimination](<https://devfeed.tech/tags/discrimination.md>), [events](<https://devfeed.tech/tags/events.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [extend](<https://devfeed.tech/tags/extend.md>), [failure](<https://devfeed.tech/tags/failure.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [improvement](<https://devfeed.tech/tags/improvement.md>), [learning](<https://devfeed.tech/tags/learning.md>), [local](<https://devfeed.tech/tags/local.md>)

### AI overview

The authors attribute PPO's difficulty on long-horizon, contact-rich manipulation tasks to state-value critics' inability to distinguish actions when a policy reaches a partial solution and outcomes converge. They extend PPO with an action-conditioned value critic that compares the executed action with alternatives sampled at the same state. Combined with Generalized Advantage Estimation and bounded exploration noise, the method reportedly escapes this plateau and completes precise peg insertion within a training budget in which PPO with identical exploration fails. The supplied text does not provide further experimental details.

### Source excerpt

arXiv:2610.38034v1 Announce Type: new Abstract: Reinforcement learning holds the promise of robots that learn manipulation skills from interaction alone, without pretrained priors or demonstration data. In practice, on-policy methods such as PPO routinely stall on long-horizon, contact-rich tasks: dense reward shaping carries the policy through stages, yet stitching them together and discovering sparsely rewarded events rarely succeeds. We trace this failure to the missing action discrimination in state-value credit assignment: a state-value critic judges progress only by episode outcomes, so when the policy converges to a partial solution all outcomes coincide, and improvement stops because no action can prove better than another. We therefore extend PPO with an action conditioned value critic which, can compare the executed action with alternatives sampled at the same state and aid the learning signal. Combined with Generalized Advantage Estimation and a bound on exploration noise, we show how the extended algorithm escapes the plateau and completes precise peg insertion within a training budget where PPO with identical exploration fails.

## Same Label, Different Interface: Beyond Action-Space Labels in Robot Learning

DevFeed: [Same Label, Different Interface: Beyond Action-Space Labels in Robot Learning](<https://devfeed.tech/articles/same-label-different-interface-beyond-action-space-labels-in-robot-learning-108645.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.38104>)

Author: Arshia Sangwan, Zhongyang Zhang

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [cs-sy](<https://devfeed.tech/tags/cs-sy.md>), [eess-sy](<https://devfeed.tech/tags/eess-sy.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [interfaces](<https://devfeed.tech/tags/interfaces.md>), [learning](<https://devfeed.tech/tags/learning.md>), [position](<https://devfeed.tech/tags/position.md>), [robot](<https://devfeed.tech/tags/robot.md>), [tasks](<https://devfeed.tech/tags/tasks.md>), [velocity](<https://devfeed.tech/tags/velocity.md>)

### AI overview

Across three contact-rich robot tasks, the same action label can produce different learning outcomes because policy outputs pass through scaling, filtering, feedback, timing, and effort limits. With a shared learner and training budget, the study compared five joint-space interfaces using ten independently trained policies per condition. Changing only physical action scale under the same position-feedback law shifted median evaluation success to 0.645-1.000 across tasks; removing exponential moving average filtering had task-dependent effects. Matching velocity scale to position's local raw-torque command gain raised median success on GearMesh and NutThread, with no median change on PegInsert, but residual gaps remained on PegInsert and GearMesh; NutThread's interval included zero. Because feedback and held-action dynamics remained unmatched, this calibration did not establish equivalent trained outcomes. The authors recommend reporting the executed action pathway and the physical quantity matched.

### Source excerpt

arXiv:2610.38104v1 Announce Type: cross Abstract: Robot-learning studies often compare position and velocity actions as though each label identified a single experimental treatment, yet the label alone does not determine what the robot executes. Before affecting robot motion, bounded policy output passes through filtering, physical scaling, reference construction, feedback, timing, and effort limits. We study how these omitted properties affect attribution under one shared learner and training budget. Across three contact-rich Factory tasks (PegInsert, GearMesh, and NutThread), we compare five joint-space interfaces, with ten independently trained policies per condition. We first remain within the same position-feedback law and vary the executed interface: substituting only the physical action scale as the directly modified component changes sample-median evaluation success by 0.645-1.000 across tasks, while removing an exponential moving average produces task-dependent changes. We then move to a cross-law comparison and choose a velocity scale that exactly matches the position interface's local raw-torque command gain. This calibration raises the sample median on GearMesh and NutThread, with no median change on PegInsert. Yet clear residual sample-median gaps remain on PegInsert and GearMesh, while the NutThread interval includes zero. Thus, even this explicit equality leaves feedback and held-action dynamics unmatched and does not establish position-velocity trained-outcome equivalence. Action-interface comparisons should therefore report the executed pathway and state the physical quantity they match.

## ActiveAlign: Aligning Egocentric Manipulation via Active Perception

DevFeed: [ActiveAlign: Aligning Egocentric Manipulation via Active Perception](<https://devfeed.tech/articles/activealign-aligning-egocentric-manipulation-via-active-perception-103199.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.36859>)

Author: Ri-Zhao Qiu, Bowen Wen, Yifeng Jiang, Chen Tessler, Eric Heiden, Xue Bin Peng, Xiaolong Wang, Stan Birchfield, Erwin Coumans

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Physical AI](<https://devfeed.tech/topics/physical-ai.md>)

Tags: [alignment](<https://devfeed.tech/tags/alignment.md>), [buttons](<https://devfeed.tech/tags/buttons.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [framework](<https://devfeed.tech/tags/framework.md>), [gap](<https://devfeed.tech/tags/gap.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [planning](<https://devfeed.tech/tags/planning.md>), [precision](<https://devfeed.tech/tags/precision.md>), [rendering](<https://devfeed.tech/tags/rendering.md>), [robot](<https://devfeed.tech/tags/robot.md>), [robot-manipulation](<https://devfeed.tech/tags/robot-manipulation.md>), [sim-to-real](<https://devfeed.tech/tags/sim-to-real.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [sliders](<https://devfeed.tech/tags/sliders.md>), [unitree](<https://devfeed.tech/tags/unitree.md>)

### AI overview

ActiveAlign uses interactions to reduce Real2Sim2Real alignment gaps associated with imperfect passive observations. Its ActiveAlign-Waving module calibrates the hand and eye using active hand motions and differentiable rendering; ActiveAlign-Touching refines object alignment with tactile feedback, with the authors reporting millimeter-level precision. They report using Waving to curate simulation environments for open-source uncalibrated manipulation datasets, and demonstrating on a real Unitree G1 that the framework supports millimeter-scale tasks such as pressing buttons and moving sliders. The supplied text is an abstract-level account and does not provide further evaluation details.

### Source excerpt

arXiv:2610.36859v1 Announce Type: cross Abstract: Real-to-Sim-to-Real (Real2Sim2Real) enables structured world representations for planning, interpretability, and safe execution. Most existing approaches rely on passive perception, which creates real2sim gaps due to imperfect observation. Inspired by how humans perceive actively through actions, we propose ActiveAlign, a framework that leverages active perception to align the Real2Sim2Real gap from interactions. ActiveAlign introduces two complementary modules: (i) ActiveAlign-Waving, which performs hand-eye calibration by exploiting active hand motions and differentiable rendering, and (ii) ActiveAlign-Touching, which refines object alignment through tactile feedback to achieve millimeter-level precision. Together, these capabilities enable strong real2sim alignment. We demonstrate that ActiveAlign-Waving can be used to curate simulation environments for open-sourced uncalibrated robot manipulation datasets, and demonstrate on a real Unitree G1 robot that ActiveAlign enables fine manipulation of millimeter-scale objects such as pressing buttons and moving sliders.

## RePO-VLA: Recovery-Driven Policy Optimization for Vision-Language-Action Models

DevFeed: [RePO-VLA: Recovery-Driven Policy Optimization for Vision-Language-Action Models](<https://devfeed.tech/articles/repo-vla-recovery-driven-policy-optimization-for-vision-language-action-models-102958.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.09410>)

Author: Weijia Liufu, Xiaoyu Guo, Ruiyi Chen, Jingzhi Liu, Kaidong Zhang, Xiwen Liang, Jianqi Lin, Dawei Sun, Yuze Wang, Rongtao Xu, Bingqian Lin, Bowen Yang, Tongtong Cao, Bowen Peng, Dongyu Zhang, Guangrun

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [smolvla](<https://devfeed.tech/topics/smolvla.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [adversarial](<https://devfeed.tech/tags/adversarial.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [contact-rich-manipulation](<https://devfeed.tech/tags/contact-rich-manipulation.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [data](<https://devfeed.tech/tags/data.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [failure](<https://devfeed.tech/tags/failure.md>), [failure-recovery](<https://devfeed.tech/tags/failure-recovery.md>), [framework](<https://devfeed.tech/tags/framework.md>), [imitation](<https://devfeed.tech/tags/imitation.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [planner](<https://devfeed.tech/tags/planner.md>), [policy](<https://devfeed.tech/tags/policy.md>), [progress](<https://devfeed.tech/tags/progress.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [robustness](<https://devfeed.tech/tags/robustness.md>), [vision-language-action](<https://devfeed.tech/tags/vision-language-action.md>)

### AI overview

RePO-VLA trains vision-language-action models to recover from execution errors by learning from success, recovery, and failure trajectories. Recovery-Aware Initialization turns corrective suffixes into episodes starting from adverse states; a Progress-Aware Semantic Value Function labels useful prefixes and drift or breakdowns; and Value-Conditioned Refinement favors high-progress actions. Corrective rollouts can be planner-generated or human-collected. At deployment, a fixed high-value token biases actions toward the learned success manifold, without online failure detectors or heuristic retries. On the new Failure-Recovery Benchmark and simulated and real-world bimanual tasks, the authors report average adversarial success rising from 20% to 75%, and up to 80% in scaled real-world trials.

### Source excerpt

arXiv:2605.09410v2 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models remain brittle in long-horizon, contact-rich manipulation because success-only imitation provides little supervision for execution drift, while failed rollouts are often discarded. We introduce RePO-VLA, a recovery-driven policy optimization framework that assigns distinct roles to success, recovery, and failure trajectories. RePO-VLA first applies Recovery-Aware Initialization (RAI), slicing each corrective suffix into a new episode so corrective actions depend on the current adverse state rather than the preceding failure. It then learns a Progress-Aware Semantic Value Function (PAS-VF), aligning spatiotemporal trajectory features with instructions and successful references. The resulting labels salvage useful failure prefixes via reliability decay, while low-value labels mark drift and terminal breakdowns, teaching differences among nominal, failed, and corrective actions. The data engine turns adverse states into planner-generated or human-collected corrective rollouts, teaching recovery to the success manifold. Value-Conditioned Refinement (VCR) trains the policy to prefer high-progress actions. At deployment, a fixed high-value token biases actions toward the learned success manifold without online failure detectors or heuristic retries. We introduce the Failure-Recovery Benchmark (FRBench), with standardized error injection and recovery-focused evaluation. Across simulated and real-world bimanual tasks, RePO-VLA improves robustness, raising adversarial success from 20% to 75% on average and up to 80% in scaled real-world trials. Code and video results are available at https://1018weijia.github.io/RePO-VLA-Web/.

## SEVI: Safe Evidential Real-World Reinforcement Learning

DevFeed: [SEVI: Safe Evidential Real-World Reinforcement Learning](<https://devfeed.tech/articles/sevi-safe-evidential-real-world-reinforcement-learning-99515.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.31829>)

Author: Shaekh Mohammad Shithil, Lantao Liu

Published: 2026-10-26T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>)

Tags: [constraint](<https://devfeed.tech/tags/constraint.md>), [contact-rich-manipulation](<https://devfeed.tech/tags/contact-rich-manipulation.md>), [counts](<https://devfeed.tech/tags/counts.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [low-latency](<https://devfeed.tech/tags/low-latency.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [online-learning](<https://devfeed.tech/tags/online-learning.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [safety](<https://devfeed.tech/tags/safety.md>), [uncertainty](<https://devfeed.tech/tags/uncertainty.md>), [usb](<https://devfeed.tech/tags/usb.md>)

### AI overview

SEVI filters actions during real-world reinforcement learning using an Evidential Safety Critic that estimates future constraint-violation risk and epistemic uncertainty in one forward pass. Before execution, it replaces actions predicted unsafe or insufficiently supported with an offline-trained recovery policy intended to guide the robot toward a lower-risk region. Tests covered two simulated manipulation tasks and a real USB insertion task, including distribution-shifted conditions. The abstract reports the strongest trajectory-level safety prediction, highest final task success, and fewest unsafe events among the compared task-learning and safety-critic baselines. Relative to recovery-based safe RL, SEVI reduced unsafe events by 13.4-57.1% while retaining low-latency execution-time filtering.

### Source excerpt

arXiv:2610.31829v1 Announce Type: cross Abstract: Real-world reinforcement learning has enabled contact-rich manipulation tasks to be trained directly on physical robots, but unsafe interactions during online learning remain a key barrier to practical deployment. We propose Safe Evidential Real-World Reinforcement Learning (SEVI), a framework for reliability-aware safety intervention during real-world policy learning. SEVI introduces an Evidential Safety Critic that estimates both future constraint-violation risk and epistemic uncertainty in a single forward pass. During online learning, SEVI evaluates each task-policy action before execution; if the action is predicted to be unsafe or the prediction is insufficiently supported by evidence, SEVI overrides it with an offline-trained recovery policy that guides the robot toward a lower-risk region. We evaluate SEVI on two simulated manipulation tasks and a real-world USB insertion task under in-distribution and distribution-shifted conditions. Across task-learning and safety-critic baselines, SEVI achieves the strongest trajectory-level safety prediction, the highest final task success, and the lowest unsafe-event counts. SEVI also reduces unsafe events by 13.4 to 57.1 percent relative to recovery-based safe RL while preserving low-latency execution-time filtering. The project page is available at: https://sevi-safe.github.io/

## AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems

DevFeed: [AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems](<https://devfeed.tech/articles/agibot-world-colosseo-a-large-scale-manipulation-platform-for-scalable-and-intelligent-embodied-systems-94660.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2503.06669>)

Author: AgiBot World Contributors, Qingwen Bu, Jisong Cai, Li Chen, Xiuqi Cui, Yan Ding, Siyuan Feng, Shenyuan Gao, Xindong He, Xuan Hu, Xu Huang, Shu Jiang, Yuxin Jiang, Cheng Jing, Hongyang Li, Jialu Li, Ch

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [Physical AI](<https://devfeed.tech/topics/physical-ai.md>), [AI-Ready Data](<https://devfeed.tech/topics/ai-ready-data.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [big-data](<https://devfeed.tech/topics/big-data.md>)

Tags: [agibot](<https://devfeed.tech/tags/agibot.md>), [complex](<https://devfeed.tech/tags/complex.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [data](<https://devfeed.tech/tags/data.md>), [data-volume](<https://devfeed.tech/tags/data-volume.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [domain](<https://devfeed.tech/tags/domain.md>), [generalist](<https://devfeed.tech/tags/generalist.md>), [genie](<https://devfeed.tech/tags/genie.md>), [grippers](<https://devfeed.tech/tags/grippers.md>), [human-in-the-loop](<https://devfeed.tech/tags/human-in-the-loop.md>), [improvement](<https://devfeed.tech/tags/improvement.md>), [intelligence](<https://devfeed.tech/tags/intelligence.md>), [large](<https://devfeed.tech/tags/large.md>), [magnitude](<https://devfeed.tech/tags/magnitude.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [models](<https://devfeed.tech/tags/models.md>), [platform](<https://devfeed.tech/tags/platform.md>), [scale](<https://devfeed.tech/tags/scale.md>)

### AI overview

AgiBot World combines more than 1 million robot trajectories across 217 tasks and five deployment scenarios with a standardized collection pipeline and human verification. The platform supports grippers, dexterous hands, and visuo-tactile sensors. Its generalist policy, Genie Operator-1 (GO-1), uses latent action representations; the authors report predictable scaling as data volume increases. Policies pretrained on the dataset improve average performance by 30% over policies trained on Open X-Embodiment in both in-domain and out-of-distribution settings. GO-1 exceeds 60% success on complex real-world dexterous and long-horizon tasks and outperforms the prior RDT approach by 32%. These are abstract-reported results; the supplied text does not include evaluation details.

### Source excerpt

arXiv:2503.06669v5 Announce Type: replace-cross Abstract: We explore how scalable robot data can address real-world challenges for generalized robotic manipulation. Introducing AgiBot World, a large-scale platform comprising over 1 million trajectories across 217 tasks in five deployment scenarios, we achieve an order-of-magnitude increase in data scale compared to existing datasets. Accelerated by a standardized collection pipeline with human-in-the-loop verification, AgiBot World guarantees high-quality and diverse data distribution. It is extensible from grippers to dexterous hands and visuo-tactile sensors for fine-grained skill acquisition. Building on top of data, we introduce Genie Operator-1 (GO-1), a novel generalist policy that leverages latent action representations to maximize data utilization, demonstrating predictable performance scaling with increased data volume. Policies pre-trained on our dataset achieve an average performance improvement of 30% over those trained on Open X-Embodiment, both in in-domain and out-of-distribution scenarios. GO-1 exhibits exceptional capability in real-world dexterous and long-horizon tasks, achieving over 60% success rate on complex tasks and outperforming prior RDT approach by 32%. By open-sourcing the dataset, tools, and models, we aim to democratize access to large-scale, high-quality robot data, advancing the pursuit of scalable and general-purpose intelligence.

## Eka is worth $1.45B thanks to a physical AI model that has experts raving

DevFeed: [Eka is worth $1.45B thanks to a physical AI model that has experts raving](<https://devfeed.tech/articles/eka-is-worth-1-45b-thanks-to-a-physical-ai-model-that-has-experts-raving-90933.md>)

Original publisher: [Read original article](<https://techcrunch.com/2026/10/20/eka-is-worth-1-45-billion-thanks-to-a-physical-ai-model-that-has-experts-raving/>)

Author: Tim Fernholz

Published: 2026-10-20T13:00:00Z

Content type: news

Language: en

Sources: [TechCrunch | AI News & Artificial Intelligence](<https://devfeed.tech/sources/techcrunch-ai-news-artificial-intelligence.md>)

Topics: [Physical AI](<https://devfeed.tech/topics/physical-ai.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [Robotics](<https://devfeed.tech/topics/robotics.md>), [Robotics Simulation](<https://devfeed.tech/topics/robotics-simulation.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-model](<https://devfeed.tech/tags/ai-model.md>), [ai-robotics](<https://devfeed.tech/tags/ai-robotics.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [eka-robotics](<https://devfeed.tech/tags/eka-robotics.md>), [exclusive](<https://devfeed.tech/tags/exclusive.md>), [physical-ai](<https://devfeed.tech/tags/physical-ai.md>), [robotic-arm](<https://devfeed.tech/tags/robotic-arm.md>), [robotics](<https://devfeed.tech/tags/robotics.md>)

### AI overview

Eka Robotics pairs a force-sensitive gripper with a vision-force-action model trained through reinforcement learning in physics simulation. The approach uses contact and force sensing to guide the final grasp, and the founders say simulation can generate thousands of hours of experience in minutes, avoiding reliance on human demonstration data. In the reported demonstrations, the robot adapted to many objects without retraining, handled about 1,500 objects weighing up to 1 kg per hour, and picked reliably more than 95% of the time; these are company claims and observed demos, not proof of industrial performance. Eka is piloting pick-and-place and wire manipulation with industrial customers. Open questions include reliability over year-long production use, generalization to tasks beyond those shown, and improving gripper life from a rated 5 million duty cycles toward the 20 million standard cited for industrial arms.

### Source excerpt

Eka Robotics may be the first robotic arm that is actually good at picking a wide range of objects up, and it is so smooth and human-like, it is wowing the industry.

## Industrial Next wants factory robots to switch tasks in minutes, not weeks

DevFeed: [Industrial Next wants factory robots to switch tasks in minutes, not weeks](<https://devfeed.tech/articles/industrial-next-wants-factory-robots-to-switch-tasks-in-minutes-not-weeks-86466.md>)

Original publisher: [Read original article](<https://techcrunch.com/2026/10/13/industrial-next-wants-factory-robots-to-switch-tasks-in-minutes-not-weeks/>)

Author: Aisha Malik

Published: 2026-10-13T23:00:00Z

Content type: news

Language: en

Sources: [TechCrunch | AI News & Artificial Intelligence](<https://devfeed.tech/sources/techcrunch-ai-news-artificial-intelligence.md>)

Topics: [Robotics](<https://devfeed.tech/topics/robotics.md>), [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [dexterity](<https://devfeed.tech/tags/dexterity.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [industrial](<https://devfeed.tech/tags/industrial.md>), [industrial-next](<https://devfeed.tech/tags/industrial-next.md>), [robotics](<https://devfeed.tech/tags/robotics.md>), [robots](<https://devfeed.tech/tags/robots.md>), [startup-battlefield-200](<https://devfeed.tech/tags/startup-battlefield-200.md>), [techcrunch-disrupt](<https://devfeed.tech/tags/techcrunch-disrupt.md>)

### AI overview

Industrial Next says its AI-driven robotic stations can be reconfigured for new manufacturing tasks in minutes. The stations use high-fidelity simulation and demonstration to learn delicate, high-precision work, and the company announced a partnership with BMW Group Technology Office to evaluate its robotic manipulation technology.

### Source excerpt

Industrial Next helps manufacturers automate complex, delicate tasks that have traditionally required human labor.

## Gemini 4 Argon for Robotics

DevFeed: [Gemini 4 Argon for Robotics](<https://devfeed.tech/articles/gemini-4-argon-for-robotics-86357.md>)

Original publisher: [Read original article](<https://deepmind.google/blog/gemini-4-argon-for-robotics/>)

Author: Kanishka Rao

Published: 2026-10-13T16:01:58Z

Content type: release

Language: en

Sources: [Google DeepMind News](<https://devfeed.tech/sources/google-deepmind-news.md>)

Topics: [Robotics](<https://devfeed.tech/topics/robotics.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [Robotics Simulation](<https://devfeed.tech/topics/robotics-simulation.md>), [contact-rich manipulation](<https://devfeed.tech/topics/contact-rich-manipulation.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [Humanoid Robots](<https://devfeed.tech/topics/humanoid-robots.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [3D](<https://devfeed.tech/topics/3d.md>), [robot sense of touch](<https://devfeed.tech/topics/robot-sense-of-touch.md>), [World models](<https://devfeed.tech/topics/world-models.md>), [cURL](<https://devfeed.tech/topics/curl.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [gemini-4-argon](<https://devfeed.tech/tags/gemini-4-argon.md>), [robotics](<https://devfeed.tech/tags/robotics.md>)

### AI overview

The source presents Gemini 4 Argon as a general-purpose physical reasoning model for robotics, citing an 87% score on the ERQA benchmark and describing workflows across control, simulation, annotation, and data generation. Examples include controlling a Spot robot and a simple arm, iteratively steering the Gemini Omni world model, tuning MuJoCo assets and contact behavior against recorded episodes, retargeting human motions, and producing task progress annotations. It also describes benchmark and simulation-generation prompts, including a claimed increase in real-world success from 56% to 75% and a reduction in completion time from 35 to 24 seconds. These are claims in the supplied material; it provides no independent validation. The closing text says developers can access Argon as `gemini-4-argon` through the Interactions API and `google-genai` SDKs.

### Source excerpt

Discover Gemini 4 Argon for robotics: leading the ERQA benchmark with physical reasoning, zero-shot control, Real2Sim, and agentic workflows.

[Next page](<https://devfeed.tech/topics/contact-rich-manipulation.md?cursor=WyIyMDI2LTEwLTEzVDE2OjAxOjU4KzAwOjAwIiwgIjQ2NmRkNWFjLTEzZjQtNGE2OC04YjUzLTk5YjMzOWY4MDZmOSJd>)