Optimal design of parallel manipulators based on their dynamic performance

This paper presents an approach for the design of parallel manipulators, from the perspective of optimal dynamic performance over a given safe working zone. It is shown that no single index is sufficient to fully quantify the dynamic performance of such a system. Two different indices are therefore developed from the study of the generic equation of motion of such systems. The resulting bi-objective optimisation problem is solved using a genetic algorithm-based numerical optimiser, namely, NSGA-II. The proposed algorithm is applied to an existing planar 3-RRR parallel manipulator, to obtain an improved design. It is found that the improved design, though obtained by optimising the intrinsic performance indices, perform better in terms of extrinsic dynamic indices, such as the required torque for tracking a set of given trajectories.