基于多传感器融合的无人机自主避障技术研究

打开文本图片集

关键词：多传感器融合；自主避障；路径规划；动态障碍；无人机导航

中图分类号： 文献标志码：A 文章编号：2095-2945（2025）34-0037-04

Abstract：InresponsetotheautonomousobstacleavoidanceneedsofUAVsincomplexenvironments，thisresearchproposes acolaborativesensinganddecision-makingarchitecturebasedonmulti-sensorfusion.Throughheterogeneoussensorspatiotemporalcalibrationandfeature-levelfusionmethods，aLiDAR-vision-IMUjointsensingmodelisconstructed，andahybridpath planningstrategyofdynamicthreatballandrolingtimedomainoptimizationisdesigned.Simulationtestsshowthatthesystem achievesanobstacleavoidancesuccessrateof89.7%indynamicobstacleraidscenarios，andthetrajectorylengthratiois optimizedto1.23.Intheactualflightverification，thefusionpositioningerorintheurbanbuildingcomplexenvironmentwas controlled within 1.2m ，and the identificationaccuracy of dead branches and obstacles in forest scenes reached 92 % ．The research results provide technical support for safe navigation of UAVs：in GNSS denial environments.

Keywords：multi-sensor fusion；autonomousobstacleavoidance；path planning;dynamicobstacle;UAV navigation

随着无人机在物流巡检、应急救援等领域的广泛应用，复杂动态环境下的自主避障能力成为制约其发展的关键技术瓶颈。（剩余5087字）