Search papers, labs, and topics across Lattice.
This study integrates reinforcement learning with a humanoid robot equipped with passive inline skating wheels, enabling it to achieve dynamic locomotion strategies that mimic human skating. By employing a unique reward structure and various geometric wheel models, the researchers successfully trained a control policy that reduces the Cost of Transport (CoT) by up to 50% compared to traditional walking gaits. The resulting policy not only transfers effectively to physical hardware but also demonstrates advanced capabilities in dynamic balance and agile movement in real-world scenarios.
Achieving a 50% reduction in Cost of Transport, this humanoid robot skates with unprecedented agility and balance, redefining locomotion strategies in robotics.
As humanoid robots become increasingly dynamic, coupling them with reinforcement learning offers a promising approach to solving the complex, underactuated mechanics of passive inline skating. Equipping a humanoid robot with passive inline skating wheels presents an opportunity to combine the versatile agility of humanoids with the high-speed, energy-efficient locomotion strategies utilized by human skaters. In this paper, we train and deploy a reinforcement learning control policy that enables novel locomotion strategies for a humanoid robot modified to equip consumer inline skates instead of conventional feet. Unlike previous work limited to quadrupedal robots or actively driven wheels, our system allows for precise 6-DoF control of the skates to execute dynamic, edge-driven propulsion strategies. Our skating strategies emerge entirely from our reward structure, without reliance on human motion data, imitation learning, or kinematic priors. We overcome the inherent instability of passive wheels and simulation contact artifacts by utilizing different geometric wheel models (spherical and ellipsoidal) during training and validation, along with a custom success-based command curriculum and a specialized rolling reward. Consequently, our policy demonstrates up to a 50% reduction in Cost of Transport (CoT) compared to standard walking gaits. The resulting policy successfully transfers zero-shot to the physical Booster T1 hardware. Real-world deployments demonstrate dynamic balance, the ability to reject active physical perturbations, and agile locomotion strategies capable of turning at speed. A video of our results can be found at https://www.youtube.com/watch?v=-_APcOS7uFo.
