Real-time Terrain Mode Recognition Module for the Control of Lower-limb Prosthesis

This paper is aimed to develop and evaluate a realtime terrain mode recognition system to assist the control of the lower-limb prosthesis. A wearable terrain recognition device, composed of an inertia measurement unit and a 2D-laser, was built to classify the types of terrain in front of the lower-limb prosthesis while walking. The terrain mode recognition module could not only detect the types of terrain in front of the wearers but also provide the detailed environment information such as the height of the stair, the angle of the slope. The built terrain mode recognition algorithm was evaluated on an able-bodied subject in the study. The results showed that the algorithm provided the terrain information accurately and quickly: terrain mode recognition accuracy was over 95% and the types of terrain were classified every 150ms which was quick enough for the control of the lower-limb prosthesis in the future. Compared with the previous work on the terrain mode recognition, the approach mentioned on this paper could be applied in more complex environment and furthermore should be considered for operating wearable lower-limb prosthesis and orthotics.