Shape estimation of a large workspace continuum manipulator with fusion of inertial sensors

Abstract

This paper focuses on shape estimation of a novel multi-section continuum and compliant robotic manipulator with large workspace. Real-time shape estimation is important for closed-loop control of continuum robots. However, due to the continuum deformation of such manipulators, it is not feasible to use conventional angular sensors. On the other hand, in order to increase the motion range in the proposed manipulator structure, each flexible link includes several serially connected sections. Because of the effect of gravitational forces, each section in these long size links could have different curvatures which makes the shape estimation more challenging. In this study, inertial sensors are utilized to estimate the overall shape of the large scale continuum manipulator. MEMS based inertial measurement units (IMUs) including gyroscopes, accelerometers, and magnetometers are installed on each section and the shapes of the sections are determined with Extended Kalman Filter (EKF) fusion algorithm. But the effect of unwanted external disturbances including translational accelerations and magnetic interferences on the observation model of the EKF, severely affects the estimation performance. So, a disturbance rejection algorithm (DRA) is proposed which can detect the presence of disturbances in each measurement axis and excludes that sensor axis from estimation procedure. Performance of the proposed shape estimation method is validated experimentally on the implemented manipulator.