Force-Torque Sensing in Robotics

Being able to perform dynamic motions repeatably and reliably is an active research topic. The present thesis aims to contribute to this by improving the accuracy of force-torque sensing in robots. It focuses primarily on six axis force-torque sensors, although other sources of force-torque sensing are explored. Force sensing technologies, calibration procedures of these sensors and the use of force-torque sensing in robotics are described with the aim to familiarize the reader with the problem to solve. The problem is tackled in two ways: improving the accuracy of six axis force-torque sensors and exploring the use of tactile sensor arrays as force-torque sensors. The contributions of this thesis are : the development of the Model Based In situ calibration method for improving measurements of sensors already mounted on robots and the improvement in performance of the robot as a consequence; the design of a calibration device to improve the reliability and speed of calibration; and the improvement of force sensing information of a capacitive tactile array and its use on a robot as force-torque information source. The developed algorithms were tested on the humanoid robotic platform iCub.