In-hand object reorientation, i.e., picking an object, reorienting it in hand, and then placing it or using it as a tool, is a common maneuver for humans. However, this task in robotics remains an active area of research due to several unsolved issues.

Object re-orientation is a difficult task for machines.

Object re-orientation is a difficult task for machines. Image credit: Jeremy Tarling via Wikimedia, CC-BY-SA-2.0

A recent paper on proposes a system for the object reorientation problem with a multi-finger hand. It allows manipulation with hands facing upward or downward and is able to use external surfaces to aid manipulation. Moreover, the system can reorient objects of novel shapes to arbitrary orientations. It operates from sensory data that can be easily obtained in the real world.

In order to achieve that, the system uses model-free reinforcement learning with three key components: teacher-student learning, gravity curriculum, and stable initialization of objects. The researchers provide evidence that the learned policies can be transferred to the real world in the future.

In-hand object reorientation has been a challenging problem in robotics due to high dimensional actuation space and the frequent change in contact state between the fingers and the objects. We present a simple model-free framework that can learn to reorient objects with both the hand facing upwards and downwards. We demonstrate the capability of reorienting over 2000 geometrically different objects in both cases. The learned policies show strong zero-shot transfer performance on new objects. We provide evidence that these policies are amenable to real-world operation by distilling them to use observations easily available in the real world. The videos of the learned policies are available at: this https URL.

Research paper: Chen, T., Xu, J., and Agrawal, P., “A System for General In-Hand Object Re-Orientation”, 2021. Link: