External Force Estimation in Reaction Force Sensing Series Elastic Actuator for Human Interaction Control

Abstract

Recently, considering the interaction with human is big issue in robotics area with expanding the area to medical and service applications. The most important thing is a safety in human interaction and series elastic actuator (SEA) has been received great expectation for that need. SEAs have advantages in terms of safety which is having high back drivability for the inherent compliance even though it has high gear ratio while conventional actuator system has low back drivability for high gear ratio. Furthermore SEAs can control the force or torque easily with measuring the deformation of the elastic element (i.e. a spring) even though it has no any force or torque sensor. However, many researchers have applied SEAs to their applications and controlled the robot with assuming that the force which is measured by the spring is same with the external force by the hook’s law. We can recognize that the assumption is wrong by considering exact dynamic model of SEAs. This poor approach can decrease control performance of the applications. Especially, it make problem in terms of safety in applications with human interaction. In this paper, exact dynamic model of Reaction Force sensing Series Elastic Actuator (RFSEA) has derived using the Lagrangian mechanics with considering the load part and dynamic characteristics that the force which is measure by the spring is not same with actual external force is proved. Two methods for estimating the external force are introduced and several experiments are represented for verifying the estimation performance using two methods. ⓒ 2017 DGIST