• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学 ›› 2022, Vol. 44 ›› Issue (03): 563-570.

• 人工智能与数据挖掘 • 上一篇    

基于动态矩阵的未知环境地图构建与路径规划

张志远,陈海进   

  1. (南通大学信息科学技术学院,江苏 南通 226019)
  • 收稿日期:2020-10-10 修回日期:2020-12-09 接受日期:2022-03-25 出版日期:2022-03-25 发布日期:2022-03-24
  • 基金资助:
    江苏省科技成果转化专项资金(BA2017065)

Unknown environment map construction and path planning based on dynamic matrix

ZHANG Zhi-yuan,CHEN Hai-jin   

  1. (School of Information Science and Technology,Nantong University,Nantong 226019,China)
  • Received:2020-10-10 Revised:2020-12-09 Accepted:2022-03-25 Online:2022-03-25 Published:2022-03-24

摘要: 目前主流的SLAM地图构建方法在环境建模中一般要借助人机交互平台,人工成本高,独立性较差。提出基于动态矩阵的未知环境地图构建算法,可以在完全未知的陌生环境中,基于二维空间栅格地图建模并利用A*算法进行回溯,独立实现地图信息的全覆盖采集。针对传统的局部覆盖路径规划算法存在重复率高、运行效率低的问题,进行了改进设计,一旦检测到封闭区域则优先处理,并采用沿边循迹和牛耕式运动相结合的方法进行子区域路径规划。算法使用Matlab进行仿真设计,通过Webots机器人仿真平台进行了验证,仿真结果表明,改进算法与传统的局部覆盖算法相比,在子区域划分数目、回溯路径总长和路径重复率等指标上有明显提高。


关键词:

Abstract: The SLAM map building method generally relies on human-computer interaction platform in environmental modeling, which has high labor cost and poor independence. This paper proposes an unknown environment map construction algorithm based on dynamic matrix. The algorithm builds a two-dimensional raster map and uses A* algorithm for backtracking in completely unknown environment, so as to realize the full coverage collection of map information independently. Traditional local coverage path planning algorithm has high repetition rate and low operation efficiency. To solve these problems, the presented algorithm deals with the closed area immediately, and combines the edge tracking and boustrophedon coverage methods in the path planning of sub-region. The algorithm was designed using MATLAB, and further verified by the Webots robot simulation platform. Simulation results show, compared with the traditional local coverage algorithm, that the algorithm has significant improvements in the reduced number of sub regions, the total length of backtracking path and the path repetition rate..

Key words: raster map modeling, path planning, dynamic matrix, A* algorithm, edge tracking