基于FSAC多傳感器賽道錐桶建圖算法研究

打開文本圖片集

主題詞：方程式賽車激光雷達工業(yè)相機組合慣導系統(tǒng)RGB錐桶地圖中圖分類號：U469.696 文獻標志碼：A DOI：10.19620/j.cnki.1000-3703.20240403

Research on Cone Barrel Mapping Algorithm Based on FSAC MultiSensor Track

LiYilong1，LiGang1，DengWeiwen2，XuLong1 （1.Liaoning Universityof Technology，Jinzhou121ooo;2.Beijing Universityof AeronauticsandAstronautics， Beijing 102206）

【Abstract】Forthe problem of mapping failure in the high-speed tracking and figure-eight scenariosof the Formula StudentAutonomous China （FSAC）duetothelimitedrecognitionand lowaccuracyofsingle-sensorcone detection，this paper proposesacone mapping algorithmbasedon theloosecoupling of LiDAR，industrial cameras，andacombined inertial navigation system.Byprojecting LiDARdataontothecamera cordinatesystem，the similaritybetweenthetargetdetection boundingboxes from thecamera’sdep learning framework （YOLOv5）andthe LiDARcone bounding boxes is matched.The fused point cloud，containing RGBcolor information，isthen transformed from the LiDAR cordinatesystem to themap coordinatesystem.Tereal-timevehicleposecalculatedbythecombined inertialnavigationsystemisusedtoupdatethefused cone pointcloud map.Real-vehicle comparative test results show thatthe algorithm achieves anaverage recallrateof 98.6% andanaverageprecisionof99.1%，enabling thedistinction betweentheinnerandoutertracksof theconemap，thereby enhancing the vehicle's perception，anticipation capabilities and path planning efficiency.

KeyWords：Formula RacingCar，LiDAR，Industrial Cameras，Combined InertialNavigation Systems，RGB Cone Barrel Maps

【引用格式】李逸龍，李剛，鄧偉文，等.基于FSAC多傳感器賽道錐桶建圖算法研究[J].汽車技術(shù)，2025（5）：29-38. LIYL，LI G，DENG W W，et al. Research on Cone Barel Mapping Algorithm Based on FSAC Multi-Sensor Track[J]. Automobile Technology，2025（5）： 29-38.

1前言

目前，自動駕駛技術(shù)已成為全球汽車產(chǎn)業(yè)轉(zhuǎn)型升級的核心驅(qū)動力和戰(zhàn)略制高點，中國大學生無人駕駛方程式大賽（Formula Student AutonomousChina，F(xiàn)SAC）依托高校，推動自動駕駛技術(shù)的創(chuàng)新實踐。（剩余12758字）