On the optimum design of 3-RPR parallel mechanisms

This paper deals with the optimization of 3-RPR planar parallel mechanisms based on different performance indices including the kinematic sensitivity, the workspace and the singularity locus. Since the kinematic sensitivity has been proposed only recently, more emphasis is placed on how it should be adapted for the optimization of 3-RPR parallel mechanisms. The optimization is implemented in sequence using first a single objective technique, differential evolution, and then resorting to a multi-objective optimization concept, the so-called Non-dominated sorting genetic algorithm-II. A Pareto-based multi-objective approach helps to overcome the problem of unit-inconsistent objectives in the optimization algorithm. Moreover, an ∞-norm is used in computation of the kinematic sensitivity which has perhaps the clearest physical interpretation.