A Hierarchical Reinforcement Learning Approach for Adaptive Quadruped Locomotion of a Rat Robot.

Small robots encounter considerable difficulties in learning effective motions on complex terrains owing to their underactuated nature and nonlinear dynamics. In this paper, we present a novel approach for robot motion generation that implements reinforcement learning, based on simplified exploration of the robot’s action and time slice conduction. Our approach controls the robot’s actions using normalized signals and hierarchical mappings on mathematical space, which facilitates the learning process. We execute action in the timeslice to make efficient interaction with the environment. The effectiveness of our methodology is evaluated across a diverse range of simulated terrain scenarios, supplemented by physics validation. Our results show that our approach performs effective on complex terrains that are designed for small-sized robots.