Cognition-Based Control and Optimization Algorithms for Optimizing Human-Robot Interactions in Power-Assisted Object Manipulation

This article investigates the necessity of inclusion of cognitive information regarding weight perception in the control algorithm of a power assist robotic system (PARS) for object manipulation, and presents optimization algorithms to determine optimum human-robot interactions (HRI) and manipulation performance for the system. Two dynamics models for lifting objects with the system are derived. One model does not include weight perception, but another model does include it. Two different admittance control algorithms based on two dynamics models are derived. A comprehensive evaluation scheme is derived to evaluate the system for HRI and manipulation performance, and optimization algorithms are derived to determine optimum HRI and performance. A test-bed PARS is developed to verify the control and optimization algorithms. During the experiments, the human subjects lift an object with the system for each control algorithm separately. Results show that the system characteristics are unsatisfactory, power assistance is unclear and optimum HRI is not achievable for the control that does not include weight perception. However, power assistance is quantified clearly and optimum HRI is achieved with satisfactory manipulation performance when the control includes weight perception. We then propose to use the results to develop control algorithms of power assist robots to assist humans manipulating heavy objects in industries that may provide optimum HRI and performance.