Chapter
Comma AI's approach to self-driving cars
Comma AI uses end-to-end learning on actual driving data from users to train its system for successful control of the vehicle while maintaining the lane and longitudinal control. Their approach differs greatly from the rest of the self-driving car industry, including Tesla.
Clips
Comma AI uses a massive amount of data on successful vehicle control to allow mastery of each task in self-driving, such as lane detection and stop sign recognition.
1:07:14 - 1:11:10 (03:55)
Summary
Comma AI uses a massive amount of data on successful vehicle control to allow mastery of each task in self-driving, such as lane detection and stop sign recognition. Their approach differs from Tesla's and the wider self-driving car world.
ChapterComma AI's approach to self-driving cars
Episode#132 – George Hotz: Hacking the Simulation & Learning to Drive with Neural Nets
PodcastLex Fridman Podcast
The host wonders about the future of autonomous driving and compares it to a game of chess, discussing the importance of specialization and how driving compares to the complexity of chess.
1:11:10 - 1:12:02 (00:52)
Summary
The host wonders about the future of autonomous driving and compares it to a game of chess, discussing the importance of specialization and how driving compares to the complexity of chess.
ChapterComma AI's approach to self-driving cars
Episode#132 – George Hotz: Hacking the Simulation & Learning to Drive with Neural Nets
PodcastLex Fridman Podcast
The concept of end-to-end learning in AI is a compelling notion that eliminates the need for traditional feature engineering approaches and aims to teach machines to learn tasks from start to finish.
1:12:02 - 1:13:39 (01:36)
Summary
The concept of end-to-end learning in AI is a compelling notion that eliminates the need for traditional feature engineering approaches and aims to teach machines to learn tasks from start to finish. This higher level of abstraction is expected to replace current feature engineering techniques in the future.
ChapterComma AI's approach to self-driving cars
Episode#132 – George Hotz: Hacking the Simulation & Learning to Drive with Neural Nets
PodcastLex Fridman Podcast
This podcast discusses the idea of simulating human-like objects without going through the entirety of evolution and mentions Mu Zero as a solution to this.
1:13:40 - 1:14:52 (01:12)
Summary
This podcast discusses the idea of simulating human-like objects without going through the entirety of evolution and mentions Mu Zero as a solution to this.