Position control based on the estimated bending force in a soft robot with tunable stiffness

Abstract

Abstract In this paper, a control method based on the predicted bending force is proposed to increase the position-tracking accuracy of a soft robot when the stiffness is adjusted in open- and closed-loop control systems. For the open-loop stiffness-control system, the predicted bending force and stiffness of the soft robot are modeled, including the major and minor hysteresis effects. For the closed-loop stiffness-control system, a self-sensing method based on resistance is used to represent the stiffness, which cannot be measured directly by the sensor. Then, the bending force of the soft robot can be estimated from the measured resistance. The experimental results show that the proposed method leads to more accurate and robust tracking-control performance than time delay estimation (TDE) when the stiffness changes because the input voltage based on the estimated bending force for the SMA-2 can be properly calculated once the stiffness of the soft robot is changed. This paper proposes a feasible method to simultaneously control the stiffness (based on the self-sensing method) and position, resulting in more accurate tracking performance than the TDE approach without the predicted bending force.