Get rolling with machine learning on AWS DeepRacer
summary
Hands-on ML using a 1/18‑scale WiFi car that trains in a 3D simulator and uploads models to a physical vehicle, claiming it’s the fastest way to start reinforcement learning. Training runs at ~15 camera frames per second, typically converges after ~20 episodes, uses Amazon SageMaker + AWS RoboMaker + S3/Kinesis/CloudWatch, and requires a Python reward function (a Lambda-style snippet) to shape behavior. New DeepRacer Evo adds stereo cameras and LIDAR for object detection. League has AWS DeepRacer open and pro divisions, pro requires top 10% virtual times and wins Evo cars and re Invent trips.
AI-written summary · about
chapters (13)
- 0:00Introduction to AWS DeepRacer
- 0:55Machine learning on AWS
- 1:41Reinforcement learning basics
- 3:54AWS DeepRacer car and Evo
- 8:22Reinforcement learning terms
- 13:30Training process explained
- 16:30Neural network architecture
- 17:58DeepRacer Console overview
- 19:05Writing reward functions
- 20:34Track components in reward functions
- 24:06AWS DeepRacer League 2021
- 25:31Getting started resources
- 26:59Closing and Q&A
related talks
29mGet rolling with ML on AWS DeepRacer
23mGet rolling with machine learning on AWS DeepRacer
24mGet Rolling on Machine Learning
37mAWS DeepRacer: Train, Evaluate, and Tune
37mAccelerate your team’s machine learning journey with DeepRacer
37mAWS DeepRacer:train, Evaluate, and Tune Your Reinforcement Learning Model
29mKickstart Your Machine Learning Career with AWS DeepRacer
30mStart your ML journey with AWS DeepRacer for public sector