基于快速搜索树与凸优化的船舶路径规划与跟踪算法

Ship Path Planning and Tracking Based on Rapidly Exploring Random Tree and Convex Optimization

  • 摘要: 摘 要:【目的】针对欠驱动船舶在障碍物水域的路径规划与路径跟踪问题,提出一种基于随机搜索树与凸优化的船舶路径规划与跟踪算法。【方法】。该方法通过使用快速搜索树算法(Rapidly Exploring Random Tree, RRT),在栅格环境中进行采样并规划出可行路径,得到关键点序列。对可行路径中的关键点序列,基于有限记忆BFGS(Limited Memory BFGS, L-BFGS)凸优化算法与Cubic样条曲线,对曲线的经济性与安全性进行优化,获取时间参数化的更平滑和安全的船舶路径。最后使用模型预测控制算法进行船舶控制输出序列规划,引导船舶安全、经济地避开障碍物,从起始点航行到目标点。【结果】采用该算法对一艘船舶的运动进行模拟,实现了有效的路径规划和跟踪,路径搜索时间低于2e-5秒,路径优化时间低于0.5秒,路径跟踪绝对误差低于0.75米。【结论】研究证明,本文提出的路径规划与跟踪算法能够确保船舶的有效路径搜索与优化,对无人船舶的进一步研究和工程化应用可以提供一定思路。

     

    Abstract: Abstract: ObjectivesTo address the challenges of path planning and tracking for underactuated ships in obstacle-infested waters, this paper proposes a ship path planning and tracking algorithm based on stochastic search trees and convex optimization. MethodsThe method utilizes the Rapidly Exploring Random Tree (RRT) algorithm to sample and plan feasible paths in a grid environment, resulting in a sequence of key points. For the key point sequence in the feasible path, optimization of the economic and safety aspects of the curve is achieved through Limited Memory BFGS (L-BFGS) convex optimization algorithm and Cubic spline curves, yielding smoother and safer ship paths parameterized by time. Finally, Model Predictive Control (MPC) algorithm is employed to plan ship control output sequences, guiding ships to navigate safely and economically around obstacles from the starting point to the destination. ResultsSimulation results demonstrate that with this algorithm, ships can achieve efficient path planning and trajectory tracking, with path search time less than 2e-5 seconds, path optimization time less than 0.5 seconds, and trajectory tracking absolute error less than 0.75 meters. ConclusionsThe simulation concludes that the proposed path planning and tracking algorithm ensures effective path search and optimization for ships, offering insights for further research and industrial applications of autonomous surface vehicles.

     

/

返回文章
返回