Search papers, labs, and topics across Lattice.
The paper introduces Competence Manifold Projection (CMP), a novel approach to improve the robustness of whole-body control policies for legged mobile manipulators against out-of-distribution (OOD) inputs. CMP uses a Frame-Wise Safety Scheme to transform safety constraints into manifold inclusion, and an Isomorphic Latent Space to align manifold geometry with safety probability, enabling efficient defense against OOD intents. Experiments show CMP achieves up to a 10x improvement in survival rate in OOD scenarios compared to baselines, with minimal tracking degradation and emergent "best-effort" generalization.
Legged robots can now recover from sensor noise and crazy user commands with 10x greater reliability, thanks to a new method that respects the robot's competence boundaries.
While decoupled control schemes for legged mobile manipulators have shown robustness, learning holistic whole-body control policies for tracking global end-effector poses remains fragile against Out-of-Distribution (OOD) inputs induced by sensor noise or infeasible user commands. To improve robustness against these perturbations without sacrificing task performance and continuity, we propose Competence Manifold Projection (CMP). Specifically, we utilize a Frame-Wise Safety Scheme that transforms the infinite-horizon safety constraint into a computationally efficient single-step manifold inclusion. To instantiate this competence manifold, we employ a Lower-Bounded Safety Estimator that distinguishes unmastered intentions from the training distribution. We then introduce an Isomorphic Latent Space (ILS) that aligns manifold geometry with safety probability, enabling efficient O(1) seamless defense against arbitrary OOD intents. Experiments demonstrate that CMP achieves up to a 10-fold survival rate improvement in typical OOD scenarios where baselines suffer catastrophic failure, incurring under 10% tracking degradation. Notably, the system exhibits emergent ``best-effort'' generalization behaviors to progressively accomplish OOD goals by adhering to the competence boundaries. Result videos are available at: https://shepherd1226.github.io/CMP.
Tsinghua AI