AWS DeepRacer:train, Evaluate, and Tune Your Reinforcement Learning Model
summary
AWS DeepRacer teaches hands‑on reinforcement learning by training 1/18‑scale autonomous car models in the AWS DeepRacer Console using AWS RoboMaker for simulation and Amazon SageMaker for PPO model training, then deploying optimized models with OpenVINO on Intel Atom. Evo adds stereo camera + LIDAR for object avoidance and head‑to‑head formats. The default policy network is a six‑layer CNN; training collects state/action/reward tuples and often runs hours (example shown: eight hours) of iterative tuning of reward functions and hyperparameters to reach convergence.
AI-written summary · about
chapters (6)
- 0:00Introduction to AWS DeepRacer
- 3:44What is reinforcement learning
- 19:49DeepRacer training process
- 24:28Building your first model
- 33:54Evaluating and tuning models
- 36:27DeepRacer League overview
related talks
37mAWS DeepRacer: Train, Evaluate, and Tune
24mGet Rolling on Machine Learning
23mGet rolling with machine learning on AWS DeepRacer
27mGet rolling with machine learning on AWS DeepRacer
29mGet rolling with ML on AWS DeepRacer
42mFrom Zero to Machine Learning in 45 Minutes
37mAccelerate your team’s machine learning journey with DeepRacer
29mKickstart Your Machine Learning Career with AWS DeepRacer