PREDICTIVE CONTROL OF ROBOTIC MANIPULATORS WITH CONSTRAINTS USING ON-LINE LINEARIZED MODELS

This paper contains a proposal for the implementation of robotic controllers. It seeks to solve the problem of calculating an optimal control sequence at each sampling time that performs within the limitations set by maximum link speed, link acceleration and actuator torque. Accurate predictions are made using a linearized model of the manipulator at each working point, which enable the computational limitations of real-time implementation to be met, without adversely affecting performance. The model is evaluated at each sample-time by linearization of the Lagrange-Euler model and discretization using a bilinear transform.