Person Recognition Method in Cross-Country Environment Based on Improved Euclidean Clustering

Abstract

Automated unmanned vehicles can carry equipment for teams, greatly reducing the load of person. Most of the working environment of unmanned vehicles is in the field. The identification of persons in a cross-country environment is the basic requirement for automated driving vehicles, and the problem of identification has received much attention. Aiming at the problem of person identification from lidar point cloud data, particularly the special problem of identification in a cross-country environment, this paper designs an improved identification algorithm based on Euclidean clustering, theoretical analysis, and the geometric and physical characteristics of people. Experiments are carried out on tracked vehicle platforms in a cross-country environment to verify the performance of the algorithm. The experimental results show that the designed lidar personnel identification algorithm can accurately identify personnel using lidar point cloud data, and the recognition rate is about 5% higher than that before the improvement.