Evolving Robot Affordances
Raja explains how robots initially discover simple interactions, like pushing or holding objects, which evolve into more complex affordances through reinforcement learning. By associating actions with motivations, such as turning on a light, robots can develop a more abstract understanding of their environment. This process highlights the potential for robots to build upon basic interactions to achieve more sophisticated tasks, drawing parallels to concepts of common sense in AI.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Towards Abstract Robotic Understanding with Raja Chatila - #118
Related Questions
Is the interaction of robots in the physical world a signal that could be used in reinforcement learning, as discussed in the episode Pieter Abbeel: Deep Reinforcement Learning | Lex Fridman Podcast #10 and the clip Robot Psychology?
Is the interaction of robots in the physical world a signal that could be used in reinforcement learning as discussed in the episode Pieter Abbeel: Deep Reinforcement Learning | Lex Fridman Podcast #10 and the clip Teaching Robots Skills?
As we have robots interact in the physical world, is that a signal that could be used in reinforcement learning in the context of the episode Pieter Abbeel: Deep Reinforcement Learning | Lex Fridman Podcast #10 and the clip Hierarchical Reasoning Challenges?