Feed-forward friction and inertia compensation for improving backdrivability of motors

Abstract

Backdrivability is an important property in applications like haptics, where force or torque is exerted by the user onto the motor. Gears cause higher friction, which results in a reduction of the backdrivability. This paper investigates how the backdriving torque can be reduced without the additional use of expensive force-torque sensors. The friction compensation uses a predetermined mapping, that adapts the motor's supporting torque depending on the measured velocity. The inertia compensation depends on the acceleration multiplied by the motor's moment of inertia. The method was objectively evaluated by using a robot. Kinetic friction compensation with inertia compensation significantly reduced the backdriving torque by 66.67 % over all median, and 23.58 % over all average values from measurements with different velocity and acceleration profiles. However, the variance and torque peaks were increased. The inertia compensation showed slight benefits in comparison to kinetic compensation alone, but not throughout all measurements.