基于图像解析的机车自动驾驶避障系统

打开文本图片集

IMAGEPARSING-BASEDAUTONOMOUSDRIVING OBSTACLEAVOIDANCESYSTEMFORLOCOMOTIVES

Liu He1Wang Yanwei1Hou Jian²Zhang Kun³LiHuabei³Qi Shuo³ （1.Tangshan Huitang IOT Technology Co.，Ltd.TangshanO63009， China; 2.Hegang Digital Technology （Tangshan） Co.，Ltd.TangshanO63OO9， China; 3.Tangshan Iron and Steel Group Co.，Ltd.Logistics BranchTangshanO63009，China）

Abstract：Thisarticle introduces algorithms such as machine vision，industrial Internetof Things，anddeeplearning into the obstacleavoidance system for unmanned locomotives.The Yolov8m algorithmis adopted to detectobstacles around therailwaytracks.Theopticalflowalgorithmisusedtopredictthecollision timebetweenthelocomotiveand the obstacles.Furthermore，the types ofobstacles andcollsion events arefed back as inputs tothe vehiclecontrolunit.The vehiclecontrolunit makescontroldecisionsbasedonthedetectionresultsandsends them tothe actuators toachieve the obstacle avoidance function.The obstacle avoidance system has passed on-site tests，which has improvedthesafetyand reliabilityoflocomotive autonomous driving and isof great significanceforsolvingtheoptimized schedulingoflogistics atthe ironand steel interface.

Key words：machine vision; YOLOv8; optical flow method; obstacle avoidance system

0前言

2024年12月，工信部等三部委印发的《制造业企业数字化转型实施指南》中无人仓储及智能物流典型场景应用中，支持企业基于模型算法开展货物装载、卸载、搬运的路径优化，提高仓储物流效率。（剩余5416字）