A novel method for hand pose estimation based on inertial and magnetic sensors

This paper presents a novel sensing methodology for hand pose estimation, using low cost micro-machined 9-axis inertial and magnetic sensors consisted of gyroscopes, accelerometers and magnetometers. The distributed design that sixteen sensor-boards and one processor-board are respectively deployed on the sections of the hand is proposed, so that the orientations of the palm and each knuckle can be measured. The fast estimation algorithm based on the kalman filter that make full use of sensors´ own advantages is proposed. In static situation, only the measurements of accelerometers and magnetometers are used. The measurements of gyroscopes are just used to give full play to the dynamic characteristic when happens rotation. Meanwhile, the switching mode of the algorithm is deduced. Then the feature of hand is considered, so that the estimated process further reduces the amount of computations. The novel hand pose estimation algorithm is easy to be implemented, the precise and efficient are compared with the common methodology. The results show the effectiveness of the proposed method.