Teaching-Robots-to-RUN-in-Simulation-by-AI

Introduction

Boston Dynamics’ BigDog and WildCat all are real life hardware…real world robots who walk similar to a horse or any quadruped…but with increasing hardware costs…its difficult to repair them over and over again if they are damaged…so, one of the better ways is to train a robot to do tasks in a simulation.

This project does that. It used one of the cutting edge algorithms in AI(which surpassed even Reinforcement Learning in this case): the Augmented Random Search(ARS) algo.