Quadrupedal robots can successfully perform locomotion tasks in complex environments and terrains. A significant extension of these abilities is to make robots able to manipulate objects as animals do. A recent study on arXiv.org proposes a framework that enables a four-legged robot to perform circus tasks such as rotating a ball.
It can turn the ball in roll, pitch, and yaw directions with a different velocity. The manipulation points are only on feet, and the robot’s base does not touch the ball. The policy is trained with model-free reinforcement learning, and zero-shot transfer to a real robot is achieved.
Image credit: courtesy of the researchers: Fan Shi at al., arXiv:2011.08811
The robot was able to manipulate a ball continuously for two minutes and recover in the case of external disturbances such as pocking the ball. This work is an important step into dexterous full-limbs manipulation on a quadrupedal robot.
Quadrupedal robots are skillful at locomotion tasks while lacking manipulation skills, not to mention dexterous manipulation abilities. Inspired by the animal behavior and the duality between multi-legged locomotion and multi-fingered manipulation, we showcase a circus ball challenge on a quadrupedal robot, ANYmal. We employ a model-free reinforcement learning approach to train a deep policy that enables the robot to balance and manipulate a light-weight ball robustly using its limbs without any contact measurement sensor. The policy is trained in the simulation, in which we randomize many physical properties with additive noise and inject random disturbance force during manipulation, and achieves zero-shot deployment on the real robot without any adjustment. In the hardware experiments, dynamic performance is achieved with a maximum rotation speed of 15 deg/s, and robust recovery is showcased under external poking. To our best knowledge, it is the first work that demonstrates the dexterous dynamic manipulation on a real quadrupedal robot.