Torque-based gravity compensation using regression for tracing trajectory

Abstract

The surge in the number of robotic applications in the industry 4.0 era has led to myriads of applications, research experiments involving human-robot interaction. Hence it is essential to move towards actuators and robots that are compliant for the safety of the humans involved in the workspace. Gravity compensation in a robotic arm, aids to decrease the load on the actuator and increases the robustness of the actuator in tracing a reference trajectory. This report focuses on gravity compensation using torque-based control complemented with position control for a compliant actuator without any prior knowledge of the description of the payload. Multivariate linear regression was used to regress the commanded torque using torque control with an accuracy of 98.9%.