Sequential identification of the dynamic parameters of a robot

Presents a method for the parameters identification of a dynamic robot model. This method consists of the simultaneous estimation of a minimum number of parameters by using a particular path and motion along this path, where a minimum number of axes is moving simultaneously. A reduced model is built from the complete dynamic model, for each of these paths. Starting from the properties of the dynamic model, the authors investigate the reduced dynamic models. Then, applying the recursive least square algorithm to these reduced models, the authors obtain a good convergence for the parameters in spite of measurement noises.<>