A Geometry Recognition-Based Strategy for Locomotion Transitions Early Prediction of Prosthetic Devices

Abstract

Prediction of locomotion modes is an integral part of control system that determines the kinematics and dynamics of powered prosthetic and orthotic devices for lower limb amputees. The traditional prediction strategies exhibit a transitional delay while detecting the transition of locomotion modes, which is undesirable. In this paper, we propose a novel prediction strategy that recognizes the geometry of the terrains (i.e., plane, wall, stairs, sloping surface, and obstacle) lying ahead of the user in order to identify the locomotion mode transitions in advance. Using this methodology, the transitional delay is eliminated. To achieve this goal, a wearable prototype, which is equipped with an array of range sensors, a force sensitive resistor (FSR), and a microcontroller, is developed. A classification algorithm along with a postprocessing technique is utilized to achieve a prediction accuracy almost equal to 99%. The proposed strategy has been successfully tested in real time on normal human subjects. Moreover, the performance of the proposed methodology is found to be robust to the transition of locomotion modes, change in dimensions of the terrains and different subjects. In addition, it also gives some rough idea about the dimensions of the terrains.