Sampling Strategies for Robust Universal Quadrupedal Locomotion Policies
For researchers in legged locomotion, this work provides a practical comparison of sampling strategies to improve policy robustness, though the findings are incremental and domain-specific.
This work investigates sampling strategies for physical robot parameters and joint gains to train a single reinforcement learning policy for quadrupedal locomotion that generalizes across diverse configurations. The results show that significant joint controller gains randomization is necessary for robust sim-to-real transfer, demonstrated via zero-shot deployment on the ANYmal robot.
This work focuses on sampling strategies of configuration variations for generating robust universal locomotion policies for quadrupedal robots. We investigate the effects of sampling physical robot parameters and joint proportional-derivative gains to enable training a single reinforcement learning policy that generalizes to multiple parameter configurations. Three fundamental joint gain sampling strategies are compared: parameter sampling with (1) linear and polynomial function mappings of mass-to-gains, (2) performance-based adaptive filtering, and (3) uniform random sampling. We improve the robustness of the policy by biasing the configurations using nominal priors and reference models. All training was conducted using the RaiSim simulation environment, tested in simulation on a range of diverse quadrupeds, and zero-shot deployed onto hardware using the ANYmal quadruped robot. Compared to multiple baseline implementations, our results demonstrate the need for significant joint controller gains randomization for robust closing of the sim-to-real gap.