Artificial Neural Network Approach for Solving Forward Kinematics of Cable Robots

Abstract

In contrast to serial robots, the forward kinematics of cable parallel robots is more difficult to solve because of their nonlinearity and complexity. For cable robots, the forward kinematics is more difficult to solve because it is also affected by the sagging of the cables and driven system. The solution for forward kinematics based on the dynamic model is quite complex, requiring many processing steps to solve the forward kinematics problem. In cable robot control, the forward kinematics problem is necessary to precisely control the position and velocity of its moving platform. The computational methods give suitable solutions for these cable robots, but these methods also have disadvantages like convergence. This paper describes using a neural network model in proposing a solution for the cable robot with cable sagging because of its weight in its workspace. The experiments conducted with the results show that the solution of the forward kinematics by the neural network model increases the convergence of the solutions with a very small evaluation error. A comparison of the calculation results shows that the used model has achieved prediction accuracy with an error of less than 0.1 mm corresponding to CDPR size 4200×3200×2900 mm.