关键词: 关键词：密集障碍, 分层轨迹规划, 急转大机动目标, 能见性约束, 自适应机制

Abstract: Abstract: Unmanned Aerial Vehicles (UAVs) can achieve robust tracking of moving targets in obstacle environments through real-time environmental perception and planning. However, for tracking of target executing a sharp turn in dense-obstacle environment, there are still some problems such as response lag, target occlusion, and high risk of obstacle collision. To address this issue, a hierarchical trajectory planning method for visibility-enhanced tracking of target executing a sharp turn is proposed. The front-end path planning constructs a heuristic function that integrates the target's historical and predict-ed trajectories to incorporate the target's real-time motion state and improve the response speed of tracking UAV. The back-end trajectory optimization incorporates visibility constraints of the target and obstacles to achieve stable tracking. At the back-end, a sequence of ellipsoid is used for modeling field-of-view to improve detection accuracy of occlusion. And the tracking distance is adaptively adjusted according to the seen obstacle density to enhance target visibility. Besides, the angle between the UAV's line of sight and UAV's motion direction is constrained to ensure continuous visibility of obstacles for reducing collision risk. Simulation experiments show that the proposed method improves the tracking success rate compared with the methods of vis-planner and fast-tracker, and it can achieve a success rate of over 80% for target executing a sharp turn tracking in dense-obstacle environment.

Key words: Keywords: dense-obstacle environment, hierarchical trajectory planning, target executing a sharp turn, visibility con-straints, adaptive mechanism