Robotics and Embodied Intelligence
AI Faculty Research Area
The Robotics and Embodied Intelligence area studies how to create full-stack intelligence for autonomous robots by tightly integrating perception, learning, planning, and control with a robot's physical or virtual body and its environment. Research in this area focuses on enabling robots to reason and act across multiple spatial and temporal scales, from low-level whole-body control and navigation skills to abstract, long-horizon task planning under uncertainty and multi-robot cooperation. A central theme is understanding how to effectively combine techniques from machine learning, computer vision, automated planning, decision-theoretic reasoning, game theory, and control to produce robust, adaptive behavior in real-world settings. Rather than treating these components in isolation, embodied AI emphasizes end-to-end autonomy grounded in physical interaction, in which intelligence emerges from the continuous coupling among computation, embodiment, and the environment.