智能科学与技术学报 ›› 2022, Vol. 4 ›› Issue (2): 264-276.doi: 10.11959/j.issn.2096-6652.202229

• 学术论文 • 上一篇    下一篇

基于改进RRT*-Smart的复杂动态环境下的无人艇路径规划

董璐1, 熊爱玲2   

  1. 1 东南大学网络空间安全学院,江苏 南京 211189
    2 东南大学自动化学院,江苏 南京 210096
  • 出版日期:2022-06-15 发布日期:2022-06-01
  • 作者简介:董璐(1990− ),女,博士,东南大学网络空间安全学院副研究员,主要研究方向为多智能体强化学习
    熊爱玲(1996− ),女,东南大学自动化学院硕士生,主要研究方向为无人艇路径规划
  • 基金资助:
    国家自然科学基金资助项目(62173251);广东省智能决策与协同控制重点实验室开放课题资助项目;中央高校基本科研业务费专项资金项目

Path planning for unmanned surface vehicle in complex dynamic environment based on improved RRT*-Smart

Lu DONG1, Ailing XIONG2   

  1. 1 School of Cyber Science and Engineering, Southeast University, Nanjing 211189, China
    2 School of Automation, Southeast University, Nanjing 210096, China
  • Online:2022-06-15 Published:2022-06-01
  • Supported by:
    The National Natural Science Foundation of China(62173251);Guangdong Provincial Key Laboratory of Intelligent Decision and Cooperative Control;Fundamental Research Funds for the Central Universities

摘要:

针对在多移动障碍船的复杂动态环境下无人水面艇的路径规划问题,设计了一种基于改进 RRT*-Smart的无人艇路径规划方法(RTSNew)。该方法首先优化了节点采样方式,在以无人艇为原点的极坐标系中进行采样,采用椭圆形的采样点范围约束以避免无效采样,增加了历史路径缓存池以充分利用历史路径。该优化大幅降低了计算量,提高了路径规划的速度。其次,改进了节点拓展方式,为了避免将动态障碍作为静态障碍处理,给每个拓展节点增加时间信息以实现动态碰撞检测,对动态障碍运动信息的充分利用大幅提高了路径的可行性。同时,考虑到无人艇的操纵性,增加了节点拓展的角度约束以保证路径平滑。最后,对移动障碍船增加虚拟障碍,使得规划路径符合《国际海上避碰规则》(COLREGS)。基于 VREP 平台进行了无人艇航行仿真实验和对比实验,结果表明,在多障碍船的复杂动态环境下,该方法能够使无人艇高效、安全地抵达终点,相较于人工势场法和传统RRT*-Smart方法,该方法在规划效率、路径优化、安全性方面均表现优异,并且保证运动路径遵守COLREGS。提出的方法较好地解决了传统方法中将动态障碍当作静态障碍处理、未考虑COLREGS、计算量大且效率低、不适用于多移动障碍船的复杂动态环境等问题。

关键词: 无人水面艇, RRT*-Smart, 路径规划, 动态碰撞检测, 《国际海上避碰规则》

Abstract:

Aiming at the path planning problem of unmanned surface vehicle (USV) in complex dynamic environment with moving multi-obstacle ships, a path planning method based on improved RRT*-Smart (RTSNew) was designed.Firstly, the sampling mode of nodes was optimized, nodes were sampled in the polar coordinate system with USV as the origin, an elliptical sampling range constraint was adopted to avoid invalid sampling, and a historical path buffer pool was used to make full use of the historical path.The optimization greatly reduced the amount of calculation and improved the speed of path planning.Secondly, the expansion mode of nodes was improved.In order to avoid treating dynamic obstacles as static obstacles, time information was added to each node to realize dynamic collision detection and the full use of dynamic obstacles motion information greatly improves the feasibility of the planned path.At the same time, considering the maneuverability of USV, the angle constraint was added in expansion of nodes to ensure smooth path.Finally, virtual obstacles were designed to mobile obstacle ships to make the planned path comply with International Regulations for Preventing Collisions at Sea (COLREGS).Based on VREP platform, the USV navigation simulation experiments and comparative experiments were carried out.The results show that RTSNew can make USV reach the destination efficiently and safely, and it performs better in planning efficiency, path optimization and path security than traditional methods in complex dynamic environment with multi-obstacle ships.RTSNew ensures that the motion path complies with COLREGS, and solves the problems of traditional methods: treating the dynamic obstacles as static obstacles, ignoring COLREGS, large amount of calculation and low efficiency, not suitable for the complex dynamic environment with moving multi-obstacle ships.

Key words: unmanned surface vehicle, RRT*-Smart, path planning, dynamic collision detection, COLREGS

中图分类号: 

No Suggested Reading articles found!