Contribution a l’etude dynamique des robots paralleles hybrides

Abstract

The work presented in this thesis concerns the dynamic modeling of hybrid parallel robots. This work helps to solve one of the real problems of research on hybrid parallel robots that is to nd a compromise between a high dynamic and a high redundancy. Through this, we present a general methodology for analysis and modeling of hybrid modular parallel robot manipulators following procedures and developed techniques for serial robots, tree structured chain or closed loops chain. In geometric modeling, two approaches have been taken to solve the inverse kinematics : the use of the generalized inverse Jacobian matrix of the manipulator, commonly called pseudo-inverse, and moreover, a non linear optimization problem not linear used. The proposed model is original since it achieves a reasonable and satisfactory compromise between the data of the problem, the mathematical model, the constraints imposed on the system, physical limitations and desired results. The dynamics study is, rstly on dynamic modeling of parallel robots. For this we have used the Newton-Euler algorithms for inverse and direct dynamic models.Secondly, from the models developed for the parallel robot, we studied the dynamic modeling of hybrid parallel robots consisting of the concatenation of several parallel modules. The algorithms developed are based on the use of recursive Newton-Euler formalism. The advantage of this model is that it is generic enough to be applied to the modeling, simulation and control of several classes of modular robots architecture. In addition to dynamic modeling, we studied the control aspect of hybrid parallel robots. The aim is to enable accurate prediction with the least possible calculations. Indeed, the inverse dynamic model developed provides a simple and e ective tool for the development of an appropriate control scheme. For this purpose, a control scheme by non-linear decoupling into the joint space was considered. In the developed simulator, all joint variables reported by the optimization algorithm are used as guidelines to meet the target of motion control. Algorithms have been developed and tested in simulation and the results are presented.