基于激光位移传感器的障碍物检测与模式识别

Obstacle detection and pattern recognition based on laser displacement sensor

基于激光位移传感器技术,为植保无人机提出了一种新的避障检测方法,包括数据块提取、障碍物基本参数计算以及障碍物模式识别三个部分。提出了根据距离值的有效性,从数据序列中提取障碍物所对应的数据块,以平均角度、平均距离和宽度为基本参数描述障碍物,以及基于障碍物的宽度、数据块的最大间隙、跳变次数和方差为特征的模式识别分类器的设计。实验表明:方法能够有效检测出未知环境下障碍物的方位和距离并对障碍物类型作出判别。

Based on laser displacement sensor technology,a new obstacle avoidance method is proposed for plant protection unmanned aerial vehicles（UAV）. The method includes the following three parts： data block extraction,obstacle basic parameter calculation and obstacle pattern recognition. Based on the validity of distance value,data block corresponding to obstacle is extracted from data sequence. The average angle,average distance and width are used as the basic parameters to describe the obstacle. The design of pattern recognition classifier based on the width of obstacle,maximum clearance of data block,jump times and variance. Experiments show that the proposed method can effectively detect the position and distance of the obstacle in unknown environment and make a judgment on obstacle type.