Compositional Learning for Robot Autonomy via Modularity and Abstraction

Explore the cutting-edge research on compositional learning for robot autonomy in this seminar presented by Yuke Zhu, Assistant Professor at UT-Austin and director of the Robot Perception and Learning Lab. Delve into the intersection of robotics, machine learning, and computer vision as Zhu discusses the challenges of building intelligent robots for long-term autonomy. Learn about the importance of functional decomposition through modularity and abstraction in scaling up robot learning methods. Discover recent advancements in developing state and action abstractions from raw signals, and examine neuro-symbolic planners that achieve compositional generalization in long-horizon manipulation tasks. Gain insights into the future of robot intelligence and the potential for creating more robust and adaptable autonomous systems.