Optimal Design of Cable-Driven Parallel Robots by Particle Schemes

Abstract

Cable-driven manipulators are attractive for high payload ratio, low inertia, large workspace, and high-speed duties. The optimal attachment configuration of cable-driven robots is key to attain desirable levels of cost and performance. In this paper, we investigate the optimal configuration of a cable-driven parallel mechanism under topologically distinct tasks by using gradient-free heuristics with distinct modes of exploration and exploitation. Our computational experiments comprising the configuration of IPAnema2, a cable-driven parallel robot with eight cables and 6-DOFs, using five gradient-free particle-based optimization heuristics have shown (1) the multimodal properties of the search space, (2) niching and stagnation avoidance strategies in optimization offer competitive convergence to feasible solutions, and (3) using the cost function based on the sum of square of forces while solving the tension distribution problem leads to feasible yet not always smooth force distributions, implying the need to devise tailored objective functions considering smoothness factors in the quadratic program. Our results has the potential to explore the nature of the search space to build tailored and fast learning schemes for cable-driven mechanisms.