基于树莓派的不良驾驶行为AI监测系统设计  

Design of risky driving behavior AI monitoring system based on Raspberry Pi

作　　者：潘菊 兰晓红[1] 刘敏 李文杰 PAN Ju;LAN Xiaohong;LIU Min;LI Wenjie(College of Computer and Information Science,Chongqing Normal University,Chongqing 401331,China)

机构地区：[1]重庆师范大学计算机与信息科学学院,重庆401331

出　　处：《现代电子技术》2023年第7期176-179,共4页Modern Electronics Technique

基　　金：重庆市教育科学“十三五”规划课题(2019⁃GX⁃329);重庆市研究生科研创新项目:“一核三层四段三念融会贯通”青少儿编程教育模式研究与实践(CYS22572);重庆市大学生创新创业训练计划项目(S202110637072)。

摘　　要：为了规范驾驶员的驾驶行为,预防及减少机动车驾驶过程中交通事故的发生,设计并实现了一种基于树莓派的不良驾驶行为AI监测系统。该系统以树莓派为核心控制器,运用百度AI开放平台训练不良驾驶行为模型,通过搭载的摄像头实时获取驾驶行为图像,再与模型进行比对分析,监测到不良驾驶行为则予以相应的警示提醒,且提醒强度随着行为程度的加深而变化。同时,运用语音识别技术实现了语音交互、智能点歌等功能,有利于驾驶员提神解乏。实验结果表明:该系统运行稳定可靠,对各类不良驾驶行为识别率达98%以上,能够实现智能监测并进行警示提醒,控制灵活方便,应用效果好。文中研究工作对于驾驶员不良驾驶行为的监测预警、改善行车安全具有理论指导和实践应用价值。In order to regulate drivers′driving behaviors,prevent and reduce the occurrence of traffic accidents during motor vehicle driving,a risky driving behavior AI(artificial intelligence)monitoring system based on Raspberry Pi is designed and implemented.In the system,the Raspberry Pi is taken as the core controller and the Baidu AI open platform is used to train a model of risky driving behaviors.The mounted camera is used to obtain images of driving behaviors in real time and then are compared with the model for analysis.When risky driving behaviors are detected,corresponding alerts are given,and the intensity of the alert changes as the behavior deepens.The voice recognition technology is used to realize the voice interaction,intelligent song ordering and other functions,which is conducive to refreshing drivers.The experimental results show that the system is stable and reliable,and its recognition rate of all kinds of risky driving behavior is above 98%.It can realize intelligent monitoring and warning,flexible and easy control,and good application effect.The research work has the oretical guidance and practical application value for monitoring and early warning of drivers′risky driving behavior and the improvement of driving safety.

关 键 词：不良驾驶行为 监测系统 树莓派 人工智能 深度学习 图像识别 实时监测 行车安全 

分 类 号：TN911.73-34[电子电信—通信与信息系统] U495[电子电信—信息与通信工程] TP391.41[交通运输工程—交通运输规划与管理]