Cooperative Human-Robot Grasping with Extended Contact Patches

Grasping large and heavy objects with a robot assistant is one of the most interesting scenarios in physical Human-Robot Interaction. Many solutions have been proposed in the last 40 years, focusing not only on human safety, but also on comfort and ergonomics. When carrying objects with large planar surfaces, classical contact models developed in robotic grasping cannot be used. This is why we conceived a contact model explicitly considering contact area properties and thus suitable to deal with grasps requiring large contact patches from the robot side. Together with the model, this work proposes a decentralized control strategy to implement cooperative object handling between humans and robots. Experimental results showed that the proposed method performs better than a simpler one, which does not account for contact patch properties. The control strategy is decentralized and needs minimal exteroceptive sensing (force/torque sensor at the robot wrist), so the proposed approach can easily be generalized to human-robot teams composed of more than two agents.