The Inverse Kinematics of Cable-Driven Parallel Robot with More Than 6 Sagging Cables Part 2: Using Neural Networks

A cable-driven parallel robot (CDPR) has n cables whose lengths are used to control the pose of the platform. We assume here a 6 d.o.f CDPR with \(n>6\) cables that have elasticity and mass i.e. are sagging cables. In that case the inverse kinematics (IK, i.e. finding the cable lengths to reach a given platform pose) has usually an infinite number of solutions. It is then common to look for an IK solution that satisfies some optimality condition usually related to the cable tensions. Determining the global optimum with sagging cables is very difficult as we have to deal with highly non linear constraints. In a companion paper we have proposed an algorithm for finding possibly the optimal solution with an iterative method that has a relatively large computation time. In this paper we investigate the use of neural networks for speeding up the solving time.