An Approach to Adaptive Control of Robot Manipulators Using the Computed Torque Technique

Abstract

Manipulator’s control system based on computed torque techniques incorporates a model of the manipulator dynamics. The nominal torque, computed using this mathematical model, does not reflect the effects of unknown loadings and uncertainty in modelling the parameters. An approach is presented which compensates for unknown loading and parameter uncertainty. This compensation is based on the “recursive” identification of a new dynamics operator which maps a vector of generalized coordinates into the vector of generalized forces (joint torques). The identification is based on a least-square approximation. Using the identified operator, which provides the compensated nominal torque, the system is controlled in closed-loop to generate regulation of the error in joint coordinates. The regulation is obtained using a common discrete optimization feedback law which is based on a recursive identification of the first order approximation of the dynamics model. The approach is illustrated with simulation results.