Walk These Ways: Tuning Robot Control for Generalization with Multiplicity of Behavior
This addresses the need for efficient robot control adaptation in robotics, offering a practical solution to bypass slow retraining cycles.
The paper tackles the problem of learned locomotion policies failing to adapt quickly to out-of-distribution environments by proposing a single policy that encodes multiple locomotion strategies, enabling real-time selection for new tasks without retraining. The result is a fast, robust controller that supports diverse gaits and tasks like crouching, hopping, and stair traversal.
Learned locomotion policies can rapidly adapt to diverse environments similar to those experienced during training but lack a mechanism for fast tuning when they fail in an out-of-distribution test environment. This necessitates a slow and iterative cycle of reward and environment redesign to achieve good performance on a new task. As an alternative, we propose learning a single policy that encodes a structured family of locomotion strategies that solve training tasks in different ways, resulting in Multiplicity of Behavior (MoB). Different strategies generalize differently and can be chosen in real-time for new tasks or environments, bypassing the need for time-consuming retraining. We release a fast, robust open-source MoB locomotion controller, Walk These Ways, that can execute diverse gaits with variable footswing, posture, and speed, unlocking diverse downstream tasks: crouching, hopping, high-speed running, stair traversal, bracing against shoves, rhythmic dance, and more. Video and code release: https://gmargo11.github.io/walk-these-ways/