近地大惯性旋翼无人机高速自主避障方法研究

中图分类号:TP242.6

摘要:

针对大惯性旋翼无人机在近地高速飞行中面临复杂障碍与通信受限的双重挑战，提出一种改进导航地图构建与避障轨迹规划的自主避障方法。该方法首先采用改进的ESDF+八叉树地图融合的建图策略，以兼顾全局搜索与局部优化的环境表征需求；在此基础上，通过改进的BIT*算法进行全局路径搜索，然后利用B样条插值对轨迹进行局部优化，并将动力学、平滑性及跟踪精度约束嵌入优化过程，确保生成轨迹满足大惯性旋翼无人机平台的可跟踪性；基于Gazebo环境开展了仿真实验，验证了该算法的地图构建与导航规划避障能力，结果表明与传统路径规划方法相比，本算法的平均规划速度提升了约31%，且10次测试的避障成功率可达100%；在室外真实环境开展了无人机避障飞行试验，验证了该算法在实际场景下的可行性与应用效果。

Abstract:

To address the dual challenges of navigating complex obstacles and managing communication constraints during high-speed, near-ground flight of large-inertia rotor UAVs, this paper proposes an autonomous obstacle avoidance method based on improved navigation map construction and trajectory planning. First, an enhanced mapping strategy integrating the Euclidean Signed Distance Field (ESDF) and OctoMap is developed to satisfy the environmental representation requirements for both global search and local optimization. Subsequently, an improved Batch Informed Trees (BIT*) algorithm is employed for global path search, while B-spline interpolation is utilized for local trajectory refinement. Dynamic constraints, smoothness requirements, and tracking accuracy are embedded within the optimization process to ensure the generated trajectory is trackable for the large-inertia rotor UAV platform. Simulation experiments conducted in the Gazebo environment validate the mapping and navigation capabilities of the proposed algorithm. The results demonstrate that, compared to traditional path planning methods, the average planning speed is improved by approximately 31%, and the obstacle avoidance success rate reaches 100% across ten trials. Furthermore, outdoor flight experiments in a real-world environment confirm the feasibility and practical effectiveness of the algorithm.