Predictive-based method for modelling and control of robot arm

To simulate the motion of a manipulator, one must make use of a model of real robot dynamics, where the robot dynamics is usually given by the Newton-Euler equations. A reliable model of a robot is essential not only for simulation purposes but also to let the designer to develop better control algorithms. In literature, the models usually used are given by state-equations; the basic disadvantage of using these models is that they create cumulative errors in simulation. In this study, a new method is proposed which determines the error between the discrete, linear, time-variant model and the real robot dynamics over torques. This error dynamic which possess a predictive character produced very efficient results in obtaining error-free model of a robot.