Subtle grip force estimation from EMG and muscle stiffness--relationship between muscle character frequency and grip force.

A number of upper limb amputees experience difficulty in picking up a food bowl during a meal, because grip force estimation using EMG currently does not provide sufficient accuracy for this task. In this paper, we propose a grip force estimation system that allows amputees to pick up a bowl with a prosthetic hand by using the properties of muscle stiffness in addition to EMG. We have chosen a tray holding task to evaluate the proposed system. A weight is dropped on the tray and the subjects are expected to control the tray's attitude during the task. Actual grip force, EMG, and muscle stiffness are measured, and the actual measured grip force is compared with the estimated grip force for evaluation. As a result, the proposed algorithm is found to be able to estimate grip force with an error of just 18[N], which is 30% smaller than in the method that uses only EMG. From the result that the response time estimated by proposed system is even less than a human's mechanical reaction time, the effectiveness of the proposed method has been validated.