Manipulator performance constraints in human-robot cooperation

Physical human-robot cooperation is a rapidly emerging field that aims to support humans in industrial and everyday environments. In order to facilitate effective cooperation, the kinematic and dynamic capabilities of the robot must be taken into account, such as the ability of the end-effector to move and apply forces to all directions. An online approach is proposed in this paper to address the problem of providing feedback to the operator about the robot's performance under Cartesian Compliance control. The introduced performance constraints prevent the operator from guiding the robot to low performance configurations by effectively restricting the cooperative movement towards such configurations. The constraints are in the form of spring forces/torques expressed in the Cartesian tool frame that are applied by the robot to the operator. A numerical approximation algorithm is used to determine the gradient of configuration dependent performance indices and calculate the constraints online. Various performance indices are compared regarding their ability to indicate the distance from a singularity and render force constraints during cooperation. Experimental results conducted with a 7-DOF serial LWR manipulator and a number of subjects demonstrate significant improvement of the proposed method in low effort cooperation compared to others from the literature.