Design of a Lower Limb Assisted Walking Device Based on Machine Vision

Abstract

Aiming at the problem that elderly people and patients with leg diseases are prone to falls in stair scenarios, this paper proposes a stair recognition method based on traditional computer vision, deployed on the Canaan Technology CanMV K230 embedded platform. The method first uses the Canny operator to extract edges from grayscale images, then employs the Standard Hough Transform to detect line segments from the edge map, clusters the inclination angles of each line segment, and determines whether there exist multiple groups of approximately parallel stair edge structures, thereby identifying stair scenarios and transmitting the results via UART to the STM32 lower computer to control the assistance level of the auxiliary motor. Experimental results show that the method achieves a recognition rate of over 90% in typical indoor and outdoor stair scenarios, with a running frame rate of no less than 20 FPS. It exhibits high real-time performance and low system overhead, requires no large-scale annotation data or deep learning training, and is suitable for deployment on resource-constrained embedded platforms, effectively improving the stability and safety of elderly walking assist devices in complex scenarios.