On the Control of Human-Robot Bi-Manual Manipulation

As robots begin to permeate the everyday human workspace to collaborate in innumerable and varied tasks, the robotic structure must adhere and replicate human-like gestures for effective interaction. Whether rehabilitation or augmentation, upper arm human-robot interaction is some of the most prevalent and investigated forms of collaboration. However, currently robotic control schemes fail to capture the true intricacies of anthropomorphic motion and intent during simple bi-manual manipulation tasks. This paper focuses on the introduction of bio-inspired control schemes for robot manipulators that coordinate with humans during dual arm object manipulation. Using experimental data captured from human subjects performing a variety of every-day bi-manual life tasks, we propose a bio-inspired controller for a robot arm, that is able to learn human inter- and intra-arm coordination during those tasks. Using dimensionality reduction techniques to make comprehensible the linear correlations of both arms in joint space we fit and utilize potential fields that attract the robot to human-like configurations. This method is then tested using real experimental data across multiple bi-manual tasks with a comparison made between the bio-inspired and traditional inverse kinematic controllers. Using a robotic kinematic chain identical to the human arm, models are evaluated for anthropomorphic configurations.