In-Hand Object Rotation via Rapid Motor Adaptation
This work addresses a key problem in robotics for enabling generalized manipulation, though it is an incremental step towards that broader goal.
The paper tackled the challenge of in-hand object rotation by designing a simple adaptive controller trained in simulation on cylindrical objects, which successfully rotated dozens of diverse real-world objects without fine-tuning, achieving stable finger gaits through reinforcement learning.
Generalized in-hand manipulation has long been an unsolved challenge of robotics. As a small step towards this grand goal, we demonstrate how to design and learn a simple adaptive controller to achieve in-hand object rotation using only fingertips. The controller is trained entirely in simulation on only cylindrical objects, which then - without any fine-tuning - can be directly deployed to a real robot hand to rotate dozens of objects with diverse sizes, shapes, and weights over the z-axis. This is achieved via rapid online adaptation of the controller to the object properties using only proprioception history. Furthermore, natural and stable finger gaits automatically emerge from training the control policy via reinforcement learning. Code and more videos are available at https://haozhi.io/hora