Using probabilistic models to investigate torque in motors

Abstract

In our current day and age, robots are used in several industries for tasks that are repetitive and are commonly used in contact tasks such as assembly. Humans are able to do these tasks and exert the appropriate amount of force required for the tasks easily. These tasks are however difficult to programme for robots. One important aspect of contact tasks is torque sensing. Torque sensors would inherently sense the intrinsic mechanical torques in addition to the torques that is due to the contact task itself. To isolate the contact task torques, the torques due to the intrinsic mechanics must be regressed and cancelled. This paper analyzes how torque sensed is affected by the intrinsic mechanical sources in the motor with a wheel bearing attached. In this paper, basis functions were utilized to model the torque sensed.