基于MDP的无人机避撞航迹规划研究  

Research on UAV Collision Avoidance Path Planning Based on MDP

作　　者：阚煌 辛长范[1] 谭哲卿 高鑫 史铭姗 张谦 KAN Huang;XIN Changfan;TAN Zheqing;GAO Xin;SHI Mingshan;ZHANG Qian(College of Mechatronic Engineering,North University of China,Taiyuan 030051,China;Binzhou Institute of Technology,Weiqiao-UCAS Science and Technology Park,Binzhou 256606,China)

机构地区：[1]中北大学机电工程学院,太原030051 [2]滨州魏桥国科高等技术研究院,山东滨州256606

出　　处：《计算机测量与控制》2024年第6期292-298,共7页Computer Measurement &Control

基　　金：教育部产学合作协同育人项目(231106429103427)。

摘　　要：无人机(UAV)进行避撞前提下的目标搜索航迹规划是指在复杂且众多的环境障碍约束中通过合理规划飞行路径,以更快、更高效的形式找到目标;研究了无障碍环境条件下有限位置马尔科夫移动的规律,构建了相应的马尔科夫移动分布模型;在借鉴搜索系统航迹规划的前沿研究成果之上,结合马尔科夫决策过程理论(MDP),引入了负奖励机制对Q-Learning策略算法迭代;类比“风险井”的可视化方式将障碍威胁区域对无人机的负奖励作用直观地呈现出来,构建了复杂障碍约束环境下单无人机目标搜索航迹规划模型,并进行仿真实验证明该算法可行,对航迹规划算法的设计具有一定的参考意义。Target search path planning of UAVon the premise of collision avoidanceis to find the target in the faster and more efficient form by reasonable flight path planning in the complex and numerous environmental obstacles.Discussed the law of finite position Markov movement under barrier-free environment and constructed the corresponding Markov movement distribution model.Based on the cutting-edge research results of search system trajectory planning,combined with the MDP theory,the negative reward mechanism was innovatively introduced to iterate the Q-Learning strategy algorithm,and the single UAV target search model was constructed.By analogy with the“risk well”visualization method,the negative reward effect of the obstacle threat area on the UAV was intuitively presented,and the single UAV target search path planning model under complex obstacle constraint environment was constructed.and the simulation experiment showed that the algorithm is feasible,Which has certain reference significance for the design of the route planning algorithm.

关 键 词：无人机 航迹规划 避撞 静态目标搜索 马尔科夫决策过程(MDP) 风险井 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]