摘要
以Dijkstra算法求解移动机器人路径规划(mobile robot path planning,MRPP)问题已得到广泛的应用,但在复杂工况下无法保证求解的正确性和全局最优性.而基于蚁群算法的移动机器人路径规划模型,在一定条件下能可靠地获得全局最优解,但存在求解时间过长的问题.因此,提出一种结合Dijkstra算法和蚁群算法模型两者优势求解MRPP问题的融合优化方法,以实现在短时间内获得全局最优解的目标.首先,应用Dijkstra快速算法在机器人工作环境中粗略寻迹得到最短路径次优解,然后,在次优解路径附近进行工作环境的精确划分;最后,利用蚁群算法在次优解附近精确寻迹,使最终的寻迹结果无限逼近最短路径.仿真结果表明,该融合优化方法既克服了经典蚁群算法求解时间过长的缺点,又能无限逼近全局最优解,寻迹时间较蚁群算法可缩短90%以上.
Dijkstra algorithm has been widely used for the mobile robot path planning(MRPP). However, under the complex conditions it can not make sure that the results are correct or optimal. Though with the ant colony algorithm, the mobile robot path planning model can give the global optimal solution under certain conditions, but it needs a long time for the solution. Therefore, a fusion algorithm combining Dijkstra algorithm with ant colony algorithm is proposed. This hybrid algorithm contains the advantage of Dijkstra and ant colony algorithm to obtain the optimal MRPP in shorter time. Firstly, a suboptimal solution of the shortest path can be found with the Dijkstra fast algorithm when the robot is working. Secondly, near the suboptimal solution the working environment is accurately divided. Finally, the shortest path can be gained by the ant colony algorithm. From the simulation results , more than 90% of the tracing time of the fusion algorithm can be reduced when compared with ant colony algorithm. It is apparent that this fusion optimization algorithm not only reduces the time consumption of the ant colony algorithm, but also obtains the reliable global optimal solution.
作者
谭会生
廖雯
贺迅宇
Tan Huisheng;Liao Wen;He Xunyu(School of Traffic Engineering, Hunan University of Technology, Zhuzhou 412007, China;School of Electrical and Information Engineering, Hunan University of Technology,Zhuzhou 412007, China;The Graduate School, Hunan University of Technology, Zhuzhou 412007, China)
出处
《动力学与控制学报》
2019年第2期104-111,共8页
Journal of Dynamics and Control
基金
国家自然科学基金资助项目(6167224)
湖南省自然科学基金资助项目(2016JJ6036)~~
作者简介
通讯作者:谭会生,E-mail:huisheng21nd@163.com.
