DeepRacer: Educational Autonomous Racing Platform for Experimentation with Sim2Real Reinforcement Learning
Authors:
Bharathan Balaji,
Sunil Mallya,
Sahika Genc,
Saurabh Gupta,
Leo Dirac,
Vineet Khare,
Gourav Roy,
Tao Sun,
Yunzhe Tao,
Brian Townsend,
Eddie Calleja,
Sunil Muralidhara,
Dhanasekar Karuppasamy
Abstract:
DeepRacer is a platform for end-to-end experimentation with RL and can be used to systematically investigate the key challenges in develo** intelligent control systems. Using the platform, we demonstrate how a 1/18th scale car can learn to drive autonomously using RL with a monocular camera. It is trained in simulation with no additional tuning in physical world and demonstrates: 1) formulation…
▽ More
DeepRacer is a platform for end-to-end experimentation with RL and can be used to systematically investigate the key challenges in develo** intelligent control systems. Using the platform, we demonstrate how a 1/18th scale car can learn to drive autonomously using RL with a monocular camera. It is trained in simulation with no additional tuning in physical world and demonstrates: 1) formulation and solution of a robust reinforcement learning algorithm, 2) narrowing the reality gap through joint perception and dynamics, 3) distributed on-demand compute architecture for training optimal policies, and 4) a robust evaluation method to identify when to stop training. It is the first successful large-scale deployment of deep reinforcement learning on a robotic control agent that uses only raw camera images as observations and a model-free learning method to perform robust path planning. We open source our code and video demo on GitHub: https://git.io/fjxoJ.
△ Less
Submitted 4 November, 2019;
originally announced November 2019.
