摘要
采用D^*Lite算法规划出的路径并不平滑,且预规路径与障碍物均十分接近.除此之外,在动态环境下时,由D^*Lite算法重规划得到的路径也离障碍物距离很近,十分容易发生碰撞.针对此问题,引入懒惰视线算法与距离变换相结合的方法改进D^*Lite算法.首先,对地图进行距离变换,并引入距离值的启发式代价,使得距离障碍物较远的节点优先被选择.然后,在扩展节点时引入视线算法,增加本地父亲节点和远程父亲节点的概念,使得路径不局限于八邻域扩展,从而进化为任意角度路径规划算法;最后,在遇到未知障碍物时进行局部距离变换,结合启发距离值信息进行重规划,使得重规划得到的路径远离突现的障碍物.仿真实验表明,在不同环境下规划所得到的路径均十分平滑与安全.
The path planned by the D^*Lite algorithm is not smooth,and the preplanned path is very close to the known obstaclse.Besides,the replanned path is very close to unknown dynamic obstacles,so that the collision can happen very easily.To deal with the problem,the thoughts of the Lazy Theta~*algorithm and distance transform are combined with the D^*Lite algorithm.Firstly,the map is processed by the distance transform algorithm to get the heuristic distance value,which makes the nodes that are far away from the obstacles preferred to be selected.Then,a line of sight algorithm is used while expanding nodes.The concepts of local parents and remote parents are added so that the path is more than eight neighbours.Finally,when unknown obstacles are discovered,the local distance transform algorithm is used to speed up the replan process and makes the replanned path safer.The experimental results show that the paths planned in different environments are all smooth and safe.
作者
黄鲁
周非同
HUANG Lu;ZHOU Fei-tong(Department of Electronic Science and Technology,University of Science and Technology of China,Hefei 230027,China)
出处
《控制与决策》
EI
CSCD
北大核心
2020年第4期877-884,共8页
Control and Decision
基金
国家自然科学基金项目(51475462).
作者简介
通讯作者:黄鲁(1961-),男,副教授,从事电路与系统、机器人系统等研究,E-mail:luhuang@ustc.edu.cn;周非同(1994-),男,硕士,从事机器人软件设计、路径规划算法的研究,E-mail:feitong@mail.ustc.edu.cn.
