Optimal Reconfiguration of a Limited Parallel Robot for Forward Singularities Avoidance

The positioning of the anchoring points of a Parallel Kinematic Manipulator has an important impact on its later performance. This paper presents an optimization problem to deal with the reconfiguration of a Parallel Kinematic manipulator with four degrees of freedom and the corresponding algorithms to address such problem, with the subsequent test on an actual robot. The cost function minimizes the forces applied by the actuators along the trajectory and considers singular positions and the feasibility of the active generalized coordinates. Results are compared among different algorithms, including evolutionary, heuristics, multi-strategy and gradient-based optimizers.