摘要
智能小车通过红外光电传感器采集信号,得到小车前方障碍物的灰度图,根据该灰度图进行避障判断,这种灰度图由于环境的复杂性,导致很多智能小车避障困难,甚至无法判断,给智能车的安全运行带来很大不确定性。本文设计了一种有效提高智能小车的红外避障精度的方法,首先将采集图片进行分类,对智能小车进行不断地训练,使其从不同的类别图像中找到最佳路径,从而达到提高避障的精确度,通过实验将本文方法与WALSH、CNN经典图像处理算法进行对比,得出本文提出方法的高效性。
The smart car collects the signal through the infrared photoelectric sensor,and obtains the grayscale image of the obstacle in front of the car.According to the grayscale image,the obstacle avoidance judgment is made.Due to the complexity of the environment,many smart cars are difficult to avoid obstacles,and even Judging,it brings great uncertainty to the safe operation of smart cars.This paper designs a method to effectively improve the infrared obstacle avoidance accuracy of smart cars.Firstly,the collected pictures are classified and continuously trained by the smart car to find the best path from different types of images,thereby improving the accuracy of obstacle avoidance.The experiments are compared with the classic image processing algorithms of WALSH and CNN,and the efficiency of the proposed method is verified.
作者
李振锋
黎敬涛
许博文
Li Zhenfeng;Li Jingtao;Xu Bowen(Kunming University of Science&Technology,School of Information Engineering and Automation,Kunming 650051,China)
出处
《电子测量技术》
2020年第8期108-111,共4页
Electronic Measurement Technology
关键词
智能小车
避障
HOG基分类模型
Smart car
obstacle avoidance
HOG based classification model
