Dexterity is in contact
Abstract: The human hand is a marvel of dexterity. Modern robotics has long equated dexterity with anthropomorphism—building hands that look like ours and replicate our grasping postures. Yet, despite decades of research in articulation, robots still struggle with simple tasks like adjusting a key before insertion or shifting a pen into a writing grip. Why? Because we have mimicked the form of the hand while ignoring the mechanism that governs the physics at the contact. In this talk, I argue that true dexterity lies not in articulation, but in friction modulation. By dynamically controlling friction, we can selectively slide and roll the object within the hand. First, I briefly present our novel gripper design with embedded vibration modules. By modulating friction, it offers a compact, scalable approach that can dramatically enhance the gripper's manipulation workspace. I then introduce a novel planning framework that simultaneously plans for the object and the finger surface. We demonstrate complex, physically realizable trajectories that were previously "invisible" to traditional object-centric planners. While sliding is often treated as incidental, this work demonstrates a foundational step toward its deliberate use.
Advisor: Professor Berk Calli
Committee: Professor Jing Xiao, Professor Constantinos Chamzas
Zoom link: https://wpi.zoom.us/j/99934570121