Parallel robots know a recent development in industry. They have several advantages (stiffness, dynamic characteristics...), however their kinematics is complicated in particular because the coupling of their motions. This is why the robots with decoupled motions represent interesting solutions and will be studied within the framework of this thesis, in particular from the point of view of their accuracy. We propose a methodology in two stages. Initially, we will adopt the assumption of the non-deformable bodies and we will seek to integrate the assembly and manufacturing defects in the forward kinamtic model of the robot. The study analyzes the effects of these defects on the decoupling characteristics of the robot, as on the accuracy of the pose of the moving platform. In the second time, with the assumption of the deformable bodies it is matter of calculating the stiffness of the robot with the finite element method, then to propose a faster analytical approach based on the principle of the sub-structuring. The integration of the assembly and manufacturing defects in the model of the deformable bodies may find it beneficial to visualize the effect of prestressing on the accuracy and the stiffness of the robot. This work concerned robots of the Isoglide family developed in LaMI with the collaboration of the LASMEA. The modeling of several versions with different degrees of overconstraints show the effect of the "hyperstaticity". Analysis of the behavior, as well static as dynamic, of a robot with parallel structure is a complex task. It is necessary to be able to follow several bodies simultaneously and to measure with a high degree of accuracy the complete space pose of the moving platform. Work presented implements a system of metrology by vision multi-sensors, allowing the analysis of the behavior of independent bodies, then of the complete structure of the robot. A study of correlation between the behavior expected by simulation and measurements, releases from the prospects on the coupling mechanics and vision.
