基于滚动时域的无人机动态航迹规划

PDF下载HTML

Title:

Dynamic UAV trajectory planning based on receding horizon

作者:

王文彬^1,2, 秦小林^1,2,3, 张力戈^1,2, 张国华^1,2

1. 中国科学院 成都计算机应用研究所, 四川 成都 610041;
2. 中国科学院大学 计算机与控制学院, 北京 100080;
3. 广州大学 智能软件研究院, 广东 广州 510006

Author(s):

WANG Wenbin^1,2, QIN Xiaolin^1,2,3, ZHANG Lige^1,2, ZHANG Guohua^1,2

1. Chengdu Institute of Computer Applications, Chinese Academy of Sciences, Chengdu 610041, China;
2. School of Computer and Control Engineering, University of Chinese Academy of Sciences, Beijing 100080, China;
3. Academy of Intelligent Software, Guangzhou University, Guangzhou 510006, China

关键词:

航迹规划; 滚动时域控制; VORONOI图; 变权重; 粒子群优化; 人工势场

Keywords:

trajectory planning; receding horizon control; VORONOI graph; variable weight; particle swarm optimization; artificial potential field

分类号:

TP18;V279

DOI:

10.11992/tis.201708031

摘要:

针对带有动力学约束的多旋翼无人机航迹规划问题，提出了一种基于滚动时域控制和快速粒子群优化(RHC-FPSO)方法。该方法引入了基于VORONOI图的代价图方法说明从航迹端点到达目标点的距离估计。根据滚动时域和人工势场法的思想，将路径规划问题转化为优化问题，以最小距离和其他性能指标为代价函数。设计评价函数准则，按照评价准则使用变权重粒子群优化算法求解。针对无人机靠近危险区飞行的问题，将斥力场引入到代价函数中，提升其安全性。仿真实验结果显示，使用文中方法可以有效地在满足约束条件下穿过障碍物区域，以及在复杂环境下可以动态计算。

Abstract:

Using receding horizon control and fast particle swarm optimization (RHC-FPSO), in this paper, we propose an algorithm for unmanned aerial vehicle (UAV) trajectory planning with dynamic constraints. We introduce the cost map method based on the VORONOI graph to estimate the distance from the end point of the trajectory to the target point. Using the concept of receding horizon control and the artificial potential field method, the path planning problem is transformed into an optimization problem, with the minimum distance and other performance indicators as cost functions. We design the evaluation function criteria based on the evaluation criteria and obtain the solution using a particle swarm optimization algorithm with variable weight. To address the problem in which a UAV approaches a danger zone, we introduce a repulsion field into the cost function to ensure safety. The simulation results show that the proposed method can effectively avoid obstacles within the constraint conditions and perform dynamic calculations in a complicated environment.