This paper analyzes the possible contribution of actuation redundancy in obtaining very high acceleration with Parallel Kinematic Machines (PKM). This study is based on redundant and non-redundant Delta/Par4-like robots (frequently used for pick-and-place applications). The dynamic model, valid for both redundant and non-redundant robots, is used to analyze the traveling plate acceleration capabilities: (i) at zero speed and in any directions, (ii) at zero speed in the “best” direction. The results show that actuation redundancy allows to homogenize the dynamic capabilities throughout the workspace and to increase the traveling plate acceleration capability. Finally, the design of a redundant Delta/Par4-like robot optimized for typical pick-and-place trajectories is presented.

This paper presents the design and the optimization of a parallel machine-tool composed of (i) an 3-dof actuated parallel mechanism (a linear Delta) and (ii) a 6-dof measuring parallel mechanism (a Gough platform).The measuring mechanism is used to control the actuated mechanism. The interest to use a measuring device independent of the actuation device is shown and the modeling of both devices used for the optimization is explained. Then, the optimization of both mechanisms is presented. The optimization of the actuated mechanism is performed to have an homogeneous behavior in velocity while the goal of the measuring device optimization is to obtain the best resolution for the measuring system evaluated at the tool level.

This paper presents an improved method for the geometrical calibration of parallel robots for which the structure is based upon some parallelogram mechanisms. Its originality is to identify the complete geometry of the mechanism's parallelograms, and to compensate the positioning error of the TCP (tool centre point), due to the infinitesimal rotation of the traveling plate, induced by the parallelogram geometrical errors. The main difficulties are: (i) to derive the calibration model relative to all geometrical parameters, and (ii) to reuse the identified errors in a control model in order to compensate the positioning errors. In fact, the position relationship taking into account the full geometry of the parallelograms is difficult, not to say impossible, to derive in a close form; therefore a linear approximation of the model is proposed. The formulas necessary to run the method on a Delta robot are given. Then a simple mechanism is used to illustrate the benefits of this method compared to classical ones