RoboSpan: Learning Modular Coordination Policies for High-DoF Whole-Body Mobile Manipulation with Teleoperation
RoboSpan: Learning Modular Coordination Policies for High-DoF Whole-Body Mobile Manipulation with Teleoperation
This research introduces RoboSpan, a high-DoF whole-body teleoperation platform, and the Hierarchical Modular Framework (HMF) for learning modular coordination policies. The HMF decouples the 29-DoF action space into hand manipulation, arm pose adjustment, and mobile-base configuration branches, using the Coordinated Modular Embodied Transformer (COMET) for dynamic information exchange. A full-palm 2D tactile representation enables cross-module tactile empowerment. Experiments on a physical 29-DoF robot show a 20% higher task success rate compared to mainstream baselines in long-horizon whole-body tasks.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
Source:arXiv Robotics — research abstracts · arxiv.org