Trustworthy Power Assistance in Object Manipulation with a Power Assist Robotic System

A 1-DOF power assist robotic system is developed for grasping and lifting objects tied with it by human operator. The dynamics of object manipulation with the system is derived, and a simple admittance-type feedback position control for the system is developed. Human operator's cognition in the form of weight perception is reflected when determining the control parameters. A computational model of power assistance provided by the system in object manipulation is proposed. Another computational model of operator's trust in the power assistance is also proposed. Then, the tradeoff between power assistance and trust is determined through experimentation. Results show that higher power assistance does not always generate higher trust of the operator in the robotic system. Instead, the tradeoff follows a special pattern. The reasons behind such tradeoff are analyzed. The results can be used to develop human-friendly intelligent power assist robotic systems for heavy object manipulation in different industries especially manufacturing and assembly.