Singularity Avoidance in Human-Robot Collaboration with Performance Constraints

Avoidance of low performance configurations for robotic manipulators such as a singularity during physical human-robot interaction, is a crucial issue for effective cooperation. Performance constraints is a framework for online calculation of repulsive forces in the task space so that the robot does not allow the human to guide it to singularities. However, this repulsive field is configuration dependent and non-conservative, so energy can be injected to the system. In this paper we build upon performance constraints, utilizing a handling mechanism to monitor the energy flow of the system and dissipate the excessive energy. Moreover, the equivalent task space stiffness of performance constraints is determined and the appropriate damping is calculated for the desired dynamic behavior of the robot, so that no oscillations appear that can have a negative effect in the haptic feeling of the operator. This damping is the required minimum so that no over-damped behavior is observed, making the robot cumbersome to manipulate. The proposed method is verified experimentally in a redundant manipulator during physical interaction with a human.