aiqbench analysis
This research introduces WACL, a hierarchical framework that enables robots to perform whole-arm contact manipulation by learning from bare-hand human demonstrations without teleoperation. By decoupling task semantics from specific robot kinematics, the method allows for robust adaptation to new end-effector constraints and safety requirements while maintaining high success rates in complex physical interactions. The approach represents a significant step toward more adaptable and efficient robotic systems that mimic natural human whole-body interaction strategies.
Beyond the End-Effector: Learning Whole-Arm Contact Manipulation from Human Demonstrations
arXiv:2610.17802v1 Announce Type: cross Abstract: Robot manipulation has traditionally relied on end-effector-centric interactions. In contrast, humans routinely exploit whole-arm contacts to improve efficiency, robustness, and adaptability, especially when the hands are occupied or when precise grasping is unnecessary. Enabling similar capabilities in robots remains challenging: the robot must select where on its arm to make contact, generate effective non-prehensile interactions, and satisfy task-specific pose or safety constraints on the end-effector. We propose Whole-Arm Contact Learning (WACL), a hierarchical framework that describes each whole-arm contact task with three robot-agnostic representations and hands them to a quasi-static surface-contact optimizer: semantic keypoints that summarize the scene, a reference contact trajectory that specifies where and how the arm presses on the object, and safety-aware path constraints that a vision-language model generates from the task description to protect the end-effector's task. The robot's kinematic chain enters only in this optimizer, which grounds these representations onto the arm by jointly searching over which link makes contact, where on that link, and the joint configuration that realizes it. This decomposition lets WACL learn the contact trajectory from 10 bare-hand human demonstrations per task collected without teleoperation, while new end-effector constraints change only the optimizer's input rather than any trained component. Real-world experiments on a 7-DoF arm across four whole-arm tasks show that WACL sustains contact under end-effector constraints where a strong link-pose baseline fails (78% vs. 33% success) and generalizes to new constraints without retraining. Website: https://wacl-robot.github.io
Abstract:Robot manipulation has traditionally relied on end-effector-centric interactions. In contrast, humans routinely exploit whole-arm contacts to improve efficiency, robustness, and adaptability, especially when the hands are occupied or when precise grasping is unnecessary. Enabling similar capabilities in robots remains challenging: the robot must select where on its arm to make contact, generate effective non-prehensile interactions, and satisfy task-specific pose or safety constraints on the end-effector. We propose Whole-Arm Contact Learning (WACL), a hierarchical framework that describes each whole-arm contact task with three robot-agnostic representations and hands them to a quasi-static surface-contact optimizer: semantic keypoints that summarize the scene, a reference contact trajectory that specifies where and how the arm presses on the object, and safety-aware path constraints that a vision-language model generates from the task description to protect the end-effector's task. The robot's kinematic chain enters only in this optimizer, which grounds these representations onto the arm by jointly searching over which link makes contact, where on that link, and the joint configuration that realizes it. This decomposition lets WACL learn the contact trajectory from 10 bare-hand human demonstrations per task collected without teleoperation, while new end-effector constraints change only the optimizer's input rather than any trained component. Real-world experiments on a 7-DoF arm across four whole-arm tasks show that WACL sustains contact under end-effector constraints where a strong link-pose baseline fails (78% vs. 33% success) and generalizes to new constraints without retraining. Website: this https URL
Accepted to the 10th Conference on Robot Learning (CoRL 2026). 23 pages, 8 figures, 5 tables. Project page: this https URL
Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2610.17802 [cs.RO]
(or arXiv:2610.17802v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.17802 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 360 words
Related coverage
Force-Aware Reinforcement Learning with Hybrid Sensorless Force Estimation for Wheeled-Legged Loco-Manipulation
arXiv · Sep 15, 2026
PanTact: Teaching Whole-Body Dexterity with a Wearable Contact Interface
arXiv · Oct 12, 2026
WT-UMI: Tactile-based Whole-Body Manipulation via Force-Supervised Contact-Aware Planning
arXiv · Jun 12, 2026
PolyUMI: Accessible Visual-Tactile-Audio Data Collection for Object Inference and Manipulation
arXiv · Sep 25, 2026