Real-time tracking of a moving contacting load using the distributive tactile sensing method.

Abstract

Motion measurement is a key function for patient diagnosis, therapy and rehabilitation in clinical disciplines ranging physiology, audiology, orthopaedics, neurophysiology. The equipment can be by video camera recording, machine vision systems and force plates. Motion analysis systems supported by multi-camera machine vision systems are an expensive solution for many applications. Force plates are another measurement tool that can be used in conjunction with a machine vision system or as a separate system. In the case of the latter set-up, the force plate is able to measure the centroid and vector of the ground reaction force, and the data retrieved can be interpreted by clinical staff to determine the nature of balance or stance of a patient. There is valuable information in the transients that can be detected by this approach, however the force plate cannot be utilised to discriminate other spatial factors in the way that the loads are applied by the patient on the surface. This paper reports the distributive approach to tactile sensing applied to infer the 3 dimensional motion of a moving mass in a supporting mechanism placed on the 2 dimensional sensing surface. The distributive approach has the advantage over forceplates on constructional costs and the ability to discriminate many motion metrics of patients. Implementation of the system using only three low cost deflection sensing elements, positioned under the surface with the resulting signals interpreted by neural network implemented on a field programmable gate array (FPGA) output near real time sampling rates greater than 70 KHz. The investigation demonstrates that the performance is sufficiently accurate for the intended clinical application, having backing errors of less than 5% in all three dimensions.