Dynamic identification of stairs in exoskeleton robot's service environment
In this article,in order to accurately identify the stair features in the exoskeleton robot's service environment,a method for dynamic identification of stairs is proposed,which is based on the combination of machine vision and the ultrasonic distance-measurement technology.Firstly,with the stairs'structural features taken into consideration,the stair-identification model is set up according to the Faster R-CNN target detection network,in combination with the Hough transformation algorithm.Then,the ultrasonic sensor is used to obtain the distance from the exoskeleton robot system to the stairs;the up-and down-stairs are identified based on the measured value and the pre-defined value.Finally,the up-and-down stairs are subject to a series of experiment,with different light and in special cases.The results show that the video which is input to the identification model en-sures better dynamic identification.Compared to the original Faster R-CNN model,this new model has some advantages.The stair-identification rate with normal light increases from 82.5％to 96.17％,and higher identification robustness is obtained in special cases.In addition,the accurate identification rate of up-and-down stairs is 96.67％.This study provides basis and data support for intelligent control and decision making of exoskeleton robots,which is important to improve coordination of the human-robot system,efficiency of exoskeleton assistance,as well as safety and comfort.
dynamic identification of stairsmachine visionultrasonic technologyexoskeleton robotenvironmental per-ception
NETL
NSTL
万方数据
