Pose Estimation and Tracking Control of a Pneumatic Soft Robotic Hand

Abstract The use of soft robotics offers opportunities which cannot be achieved with conventional rigid robots, including adaptive interactions with humans (Kim et al. (2013)). This article presents the state estimation and tracking control for a soft robotic hand with 12 degrees of freedom (DOF). In the work, we achieve orientation estimation of phalanges and palm using a Multiplicative Extended Kalman-Filter (MEKF) yielding an average mean absolute error of less than 3.5°. Additionally, we use the estimated orientations for a tracking control for the finger poses. Experiments show that the estimated control variable can follow a sine trajectory as well as small precise step trajectories with an estimated control error of less than 3° which we consider sufficient to precisely target objects or copy gestures in real-time.