Path Integral Policy Improvement with Covariance Matrix Adaptation

Freek Stulp, Olivier Sigaud

There has been a recent focus in reinforcement learning on addressing continuous state and action problems by optimizing parameterized policies. PI2 is a recent example of this app…