Decoupling methods of an FBG-based six-axis force/torque sensor for robotic applications

Six-axis force/torque sensors enable robots to accurately detect force and torque information, allowing them to interact effectively with external environments. This paper presents decoupling methods of a six-axis force/torque sensor using Fibre Bragg Grating (FBG). Decoupling methods of forces and torques were investigated and compared by using the decoupling algorithms of Least Squares (LS), Particle Swarm Optimisation-Back Propagation Neural Network (BP-PSO), and Radial Basis Function Neural Network (RBFNN). Calibration experiments of an FBG-based six-axis force/torque sensor were carried out. The experimental results indicate that the decoupling method using RBFNN achieves the best performance for the sensor, with a maximum nonlinear error of 2.93% and a maximum coupling error of 1.81%.