No label defined (Q4492829)

The authors develop an idea about the significance of view of autonomous geometrical parameters estimation for modular robots, because a very important problem in the design of the robotised flexible manufacturing cell and systems is the differe ce between the nominal geometry and real geometry of the manipulator. In order to identify parameters, authors proposed to linearize geometrical model. An algorithm for parameters estimation of such model, for manipulator based on the recursive least-sq ares method, is presented by authors in this paper. Authors state that persistent excitation condition of the measurement vector automatically guarantees convergence to the true parameters value. The test of geometrical parameters estimation has been c rried out on a manipulator with two degrees of freedom, and authors shows, by computer simulations and computer experiment with proposed algorithm examining several examples, that the estimation converges very fast to the true parameters value, althoug the initial estimation is very bad. In the paper we can see figures with module of prediction error; twist angle estimates, length estimates of different link, offset estimates of different link.