基于ROS四轴飞行器的路径规划
2021-01-25彭求志田丽程彬
彭求志 田丽 程彬



摘 要:针对四轴飞行器很难通过运动控制来完成快速扩展随机树(Rapidly-exploring Random Tree, RRT)所生成运动轨迹的问题,提出动力学和运动学约束的RRT算法.采用固定最终状态和固定最终时间控制器改进算法,通过动力学和运动学约束产生四轴飞行器可执行的路径,对比实际四轴飞行器行走的路径与期望的路径,对RRT重新布线,调节其行走路径,保证算法的渐近最优性.仿真实验及实物测试结果表明,该算法可以运用到四轴飞行器运动规划中,在实际应用中具有一定的价值.
关键词:快速扩展随机树;运动规划;四轴飞行器;ROS
[中图分类号]TP301.6 [文献标志码]A
Path Planning Based on ROS Quadcopter
PENG Qiuzhi,TIAN Li*,CHENG Bin
(Key Laboratory of Advanced Perception and Intelligent Control of High-end Equipment,Ministry ofEducation,Anhui University of Engineering,Wuhu 241000,China)
Abstract:Aiming at the problem that it is difficult for a quadcopter to complete the motion trajectory generated by the Rapidly-exploring Random Tree through motion control,an RRT algorithm with dynamics and kinematics constraints is proposed.The algorithm is improved by the method of fixed final state and fixed final time controller,and the executable path of the quadcopter is generated through the dynamics and kinematics constraint RRT algorithm,and the error between the actual quadcopter's path and the expected path is compared,and the wiring is rerouted,adjust its walking path to ensure the asymptotic optimality of the algorithm. Simulation experiments and physical test results show that the algorithm can be used in the motion planning of quadcopter and has certain value in practical applications.
Key words:rapidly-exploring random tree;path planning;quadcopter;ROS
四軸飞行器自主导航应用广泛,其导航包含感知、定位、路径规划以及运动控制[1]四部分,路径规划(Path Planning)[2-3]是研究的核心之一.路径规划大致分为四类:传统算法、图形学方法、智能算法和其他算法,研究成果丰硕.[4]S.Karaman等证明了基于采样的路径规划算法RRT和PRM具有概率完备性和渐进最优性.[5]Kontoudis George P.等提出了一种基于渐进最优快速探索随机树(RRT),利用积分强化学习,用化直为曲的连续时间线性系统的最优成本和最优策略.[6]Tang Zhilin等提出了一种基于采样的运动规划器,在快速探索随机树(RRT)中,通过考虑初始状态和最终状态对,为每个对确定与系统动力学和约束条件兼容的最佳轨迹,使成本最小化.[7]Moon Chang-bae等提出了一种双向快速探索随机树(DT-RRT),增加节点扩展的成功率,节省计算成本.[8]Chi Wenzheng等提出了一种基于风险的快速探索随机树算法……