Chapter
Self-driving Cars and Reinforcement Learning
1:03:47 - 1:07:14 (03:26)
The use of reinforcement learning in self-driving cars allows them to train completely end to end on user data without the need for lane detection or object detection tasks.
Clips
The lateral policy for self-driving cars can now be trained end-to-end through user data without the need for lane detection or object detection tasks.
1:03:47 - 1:06:02 (02:15)
self-driving cars
Summary
The lateral policy for self-driving cars can now be trained end-to-end through user data without the need for lane detection or object detection tasks. This method involves managing resources on the device such as logging, recording, thermal management and disk space.
ChapterSelf-driving Cars and Reinforcement Learning
Episode#132 – George Hotz: Hacking the Simulation & Learning to Drive with Neural Nets
PodcastLex Fridman Podcast
The CEO of Comma.ai, George Hotz, discusses a reinforcement learning framework for supervised autonomous driving and how the company looks at the improvement rate of disengagements to ensure it remains supervised.
1:06:03 - 1:07:14 (01:10)
Autonomous Driving
Summary
The CEO of Comma.ai, George Hotz, discusses a reinforcement learning framework for supervised autonomous driving and how the company looks at the improvement rate of disengagements to ensure it remains supervised.