Modeling, Simulation, and Control of a “Sensorless” Cable-Driven Robot

Abstract

This paper focuses on the modeling, simulation, control, and experimental validation of a planar cable-driven robot system that operates on the vertical (x-z) plane. Cable-driven robots have recently gained significant attention due to their successful commercial applications, such as spider cameras in large areas like stadiums. In this study, we model the robotic system as a planar dynamical system driven by visco-elastic tension elements (i.e., cables) attached between the four corners of the rectangular end-effector and corners of the workspace where the electric motors are connected. We assume that limited sensory information is available to the controller, such that we can only measure motor torque, and no direct information is available from the end-effector. This sensorless measurement assumption poses significant control challenges. We propose a novel control approach that adopts a parallel feedforward velocity and feedback force/torque control topology. The control inputs of the system are assumed to be the reference motor velocities, as we utilize industrial servo controllers with built-in velocity control capabilities. We model the motor dynamics as a first-order low-pass filter to account for the phase lag between the reference and actual motor commands. We first simulate the closed-loop system and test the effectiveness of the control policy under different unknown system parameters such as stiffness, damping, and motor lag. We then experimentally verify the topology on an actual experimental setup. We believe that these results are promising for future cable-driven robotic applications, especially for systems with limited sensory equipment.