Reinforcement learning approach to locomotion adaptation in sloped environments

In this work, Path Integral Policy Improvement with Covariance Matrix Adaptation (PI²-CMA) is implemented and used to address biped locomotion optimization. Central Pattern Generators (CPGs) and Dynamic Movement Primitives (DMPs) are combined to easily produce complex joint trajectories for simulated DARwIn-OP humanoid robot. PI²-CMA seeks optimal DMPs´ weights that maximize frontal velocity when facing different challenges. The simulation environments demand adaptation from the controller in order to successfully walk in different slopes. Elitism was introduced in PI²-CMA in order to improve the convergence property of the algorithm. Results show that these approaches enabled easy adaptation of DARwIn-OP to new situations. The results are very promising and demonstrate the flexibility at generating new trajectories for locomotion.