Explore the latest advancements in dexterous robotic manipulation in this 59-minute colloquium talk by Matei Ciocarlie from Columbia University. Delve into the convergence of sensor design, mechanisms, and computational motor learning that has finally enabled true robotic dexterity. Learn about the "Hardware as Policy" approach, which optimizes underactuated hand transmission mechanisms alongside grasping policies. Discover the innovative optics-based tactile finger design, providing accurate touch information across multi-curved surfaces. Examine the development of tactile-based policies for in-hand manipulation and object recognition through motor learning. Consider the future implications of these breakthroughs, including ways to enhance robustness, versatility, and generalization of dexterous manipulation, as well as potential new applications in robotics.
We (Finally) Have Dexterous Robotic Manipulation - Now What? - 2024 Winter Robotics Colloquium
Paul G. Allen School via YouTube
Overview
Syllabus
2024 Winter Robotics Colloquium: Matei Ciocarlie (Columbia University)
Taught by
Paul G. Allen School