Found 229 bookmarks
Custom sorting
Rodney Brooks on X: "Back on April 1st I posted my three laws of robotics. Here are my three laws of AI. 1. When an AI system performs a task, human observers immediately estimate its general competence in areas that seem related. Usually that estimate is wildly overinflated. 2. Most successful AI…" / X
Rodney Brooks on X: "Back on April 1st I posted my three laws of robotics. Here are my three laws of AI. 1. When an AI system performs a task, human observers immediately estimate its general competence in areas that seem related. Usually that estimate is wildly overinflated. 2. Most successful AI…" / X
·twitter.com·
Rodney Brooks on X: "Back on April 1st I posted my three laws of robotics. Here are my three laws of AI. 1. When an AI system performs a task, human observers immediately estimate its general competence in areas that seem related. Usually that estimate is wildly overinflated. 2. Most successful AI…" / X
MIT Robotics - Russ Tedrake - Planning with Graphs of Convex Sets (in the age of foundation models)
MIT Robotics - Russ Tedrake - Planning with Graphs of Convex Sets (in the age of foundation models)

Google Gemini summary: According to the video, the lesson learned from the talk is that planning and learning are complementary to each other and both are necessary for robots to achieve intelligence.

The speaker, Russ Tedrake, argues that while foundation models are powerful and will play an important role in the future of robotics, planning is still an important aspect that should not be ignored. He emphasizes that planning can improve the efficiency of learning by guiding the exploration process and focusing on the most promising actions.

The talk also introduces GCS (Graphs of Convex Sets) as a powerful tool for planning in robotics. GCS allows robots to reason about the uncertainties in their environment and make decisions that are robust to these uncertainties.

Here are the key points from the talk:

Planning and learning are complementary: Planning can improve the efficiency of learning by guiding the exploration process. GCS is a powerful tool for planning in robotics: GCS allows robots to reason about uncertainties in their environment. Behavior cloning is a useful tool for robotics: Behavior cloning can be used to learn from human demonstrations. Foundation models are powerful but not a silver bullet: Planning is still important in the age of foundation models.

·youtube.com·
MIT Robotics - Russ Tedrake - Planning with Graphs of Convex Sets (in the age of foundation models)