Simultaneous Estimation of Upper Limb Pose and Joint Torque Based on Upper Arm Deformation

The upper limb motion plays a very important role when manipulating various daily objects. Therefore, the measurement of upper limb motion helps to understand human dexterity. The richer the measurement is, such as that including poses and joint torque, the deeper the understanding is. In this paper, we propose a novel method to simultaneously estimate the pose and joint torque considering external load using only the upper arm deformation measured by a distance sensor array. The upper arm deformation is a biosignal which provides information about the complex activities of muscles and tendons. We designed a distance sensor array consisting of eight distance sensor units to measure the upper arm deformation. Based on the deformation, the pose and joint torque are estimated using Deep Neural Networks. In the experiments, we evaluate the performance of the proposed method with five participants. The results show that the proposed method can estimate the pose and joint torque with an RMSE of less than 8.7 degrees and 4.1 Nm, respectively.