Toward Reliable Sim-to-Real Predictability for MoE-based Robust Quadrupedal Locomotion
This addresses reliability and safety issues in deploying learned locomotion policies for quadruped robots, though it is incremental as it builds on existing MoE and sim-to-sim methods.
The paper tackles the problem of sim-to-real transfer and reward overfitting in quadrupedal locomotion by introducing a Mixture-of-Experts policy and RoboGauge assessment suite, achieving robust locomotion on unseen terrains like snow and obstacles up to 30 cm, with speeds reaching 4 m/s.
Reinforcement learning has shown strong promise for quadrupedal agile locomotion, even with proprioception-only sensing. In practice, however, sim-to-real gap and reward overfitting in complex terrains can produce policies that fail to transfer, while physical validation remains risky and inefficient. To address these challenges, we introduce a unified framework encompassing a Mixture-of-Experts (MoE) locomotion policy for robust multi-terrain representation with RoboGauge, a predictive assessment suite that quantifies sim-to-real transferability. The MoE policy employs a gated set of specialist experts to decompose latent terrain and command modeling, achieving superior deployment robustness and generalization via proprioception alone. RoboGauge further provides multi-dimensional proprioception-based metrics via sim-to-sim tests over terrains, difficulty levels, and domain randomizations, enabling reliable MoE policy selection without extensive physical trials. Experiments on a Unitree Go2 demonstrate robust locomotion on unseen challenging terrains, including snow, sand, stairs, slopes, and 30 cm obstacles. In dedicated high-speed tests, the robot reaches 4 m/s and exhibits an emergent narrow-width gait associated with improved stability at high velocity.