Chapter
The Limits of Reinforcement Learning with Robots
41:54 - 45:34 (03:40)
The concept of letting reinforcement learning figure everything out is counter-intuitive to how humans, especially children, operate in their environment. This raises questions on how much people can introspect what’s going on, and whether reinforcement learning could mimic the neural network inside a child's mind.
Clips
The speaker describes their experience at DeepMind where they learned about an experiment with robots using reinforcement learning to manipulate objects.
41:54 - 43:10 (01:15)
DeepMind
Summary
The speaker describes their experience at DeepMind where they learned about an experiment with robots using reinforcement learning to manipulate objects. However, the robots were kept secret from visitors.
ChapterThe Limits of Reinforcement Learning with Robots
Episode#217 – Rodney Brooks: Robotics
PodcastLex Fridman Podcast
The possibility of reinforcement learning in the mind's imagination, and the ability for introspection to explore mechanisms in different lock mechanisms.
43:10 - 45:34 (02:24)
Reinforcement Learning
Summary
The possibility of reinforcement learning in the mind's imagination, and the ability for introspection to explore mechanisms in different lock mechanisms.