A $μ$-Synthesis Based Control for Compliant Maneuvers

This paper deals with a system, which is subjected to very uncertain factors: human and environment. These independent uncertainties are dealt with explicitly on the framework of $μ$-synthesis. We also describe a controller design, which enables a robust force feedback without using a force sensor. The model of human dynamics, environments, and actuators are modeled associated with uncertainties described in the form of weighting functions. A controller is designed based on the $μ$-synthesis so that it maintains robust performance against uncertainties in both environment and human dynamics, which contributes to dexterous manipulation. The controller described here is implemented on the human power extender, which is worn by a human and amplifies the human’s physical strength, while the human’s intelligence remains as the central control system for manipulation. Experimental results conducted on the extender showed that the force estimation worked fine and the control system performed as desired.