🤖 AI Summary
Existing neural motion planners (NMPs) exhibit poor generalization across robotic morphologies, particularly failing zero-shot adaptation to unseen manipulators.
Method: We propose XMoP, a whole-body control policy that enables the first zero-shot cross-morphology generalization of NMPs to unknown robot bodies. XMoP is trained end-to-end on over 3 million procedurally generated synthetic manipulator datasets, jointly modeling implicit kinematics from visual and proprioceptive inputs—without requiring fine-tuning for deployment. Its implicit representation encodes diverse kinematic constraints, enabling seamless sim-to-real transfer.
Results: XMoP achieves an average planning success rate of 70% across seven commercial manipulators. Crucially, it successfully executes three novel tasks—including dynamic obstacle avoidance—on two previously unseen physical robots, robustly demonstrating zero-shot cross-body generalization.
📝 Abstract
Classical manipulator motion planners work across different robot embodiments. However they plan on a pre-specified static environment representation, and are not scalable to unseen dynamic environments. Neural Motion Planners (NMPs) are an appealing alternative to conventional planners as they incorporate different environmental constraints to learn motion policies directly from raw sensor observations. Contemporary state-of-the-art NMPs can successfully plan across different environments. However none of the existing NMPs generalize across robot embodiments. In this paper we propose Cross-Embodiment Motion Policy (XMoP), a neural policy for learning to plan over a distribution of manipulators. XMoP implicitly learns to satisfy kinematic constraints for a distribution of robots and $ extit{zero-shot}$ transfers the planning behavior to unseen robotic manipulators within this distribution. We achieve this generalization by formulating a whole-body control policy that is trained on planning demonstrations from over three million procedurally sampled robotic manipulators in different simulated environments. Despite being completely trained on synthetic embodiments and environments, our policy exhibits strong sim-to-real generalization across manipulators with different kinematic variations and degrees of freedom with a single set of frozen policy parameters. We evaluate XMoP on $7$ commercial manipulators and show successful cross-embodiment motion planning, achieving an average $70%$ success rate on baseline benchmarks. Furthermore, we demonstrate our policy sim-to-real on two unseen manipulators solving novel planning problems across three real-world domains even with dynamic obstacles.