Cooperative object transport in unstructured environments remains challenging for assistive humanoids due to strong, time-varying interaction forces, which can render tracking-centric whole-body control unreliable, especially in close-contact support tasks. This paper proposes a bio-inspired, interaction-oriented whole-body control (IO-WBC) that acts as an artificial cerebellum—a motor agent that translates upstream (skill-level) commands into stable, physically consistent whole-body behavior under contact.
The approach structurally separates upper-body interaction execution from lower-body support control, enabling the robot to maintain balance while shaping force exchange in a tightly coupled robot-object system. A trajectory-optimized reference generator (RG) provides kinematic prior, while a reinforcement learning (RL) policy governs body responses under heavy-load interactions and disturbances. The policy is trained in simulation with randomized payload mass/inertia and external perturbations and deployed via asymmetric teacher-student distillation, allowing the student to rely solely on proprioceptive histories at runtime.
Extensive experiments show that IO-WBC maintains stable whole-body behavior and physical interaction even when precise velocity tracking becomes infeasible, enabling compliant object transport across a wide range of scenarios.
Blogger's Review: This research significantly enhances the object transport capabilities of humanoid robots through a bio-inspired control strategy, demonstrating the immense potential of combining reinforcement learning with control theory, which is worthy of attention and further exploration.