摘要
针对辅助驾驶系统的研究热点,提出了一种基于多核ARM嵌入式平台的辅助驾驶系统。系统基于NXP公司的i.MX 8QM嵌入式平台进行开发,通过两路摄像头分别采集前方路面以及驾驶员上半身的视觉信息,实现车道偏离预警与基于卷积神经网络(CNN)和支持向量机(SVM)的驾驶员疲劳监测功能,并且能实时显示车道线以及驾驶员状态信息。结合多核ARM架构特性,利用并行计算等手段对系统进行了实时优化。使用实车采集的真实场景数据进行测试,验证了辅助驾驶系统功能的准确性与实时性。
Aiming at the research hotspot of assisted driving system,an assisted driving system based on multi-core ARM embedded platform was proposed.The system was developed based on NXP’s i.MX 8QM embedded platform.The visual information of the road ahead and the driver’s upper body are collected through two cameras to realize lane departure warning and driver fatigue monitoring function based on convolutional neural network(CNN)and supporting vector machine(SVM),and display lane lines and driver status information in real time.Combining the characteristics of the multi-core ARM architecture,the system is optimized for real-time performance by means of parallel computing.The test was conducted using real scene data collected from real vehicles to verify the accuracy and real-time performance of the assisted driving system.
作者
郁晓冬
戴毅
王俊杰
张峰
Yu Xiaodong;Dai Yi;Wang Junjie;Zhang Feng(National Experimental Teaching Demonstration Center of Electrical Engineering and Electronics,Shanghai Jiao Tong University,Shanghai 200240,China)
出处
《电气自动化》
2022年第1期102-104,107,共4页
Electrical Automation
关键词
辅助驾驶系统
车道识别
疲劳监测
多核并行
assisted driving system
lane recognition
fatigue monitoring
multi-core parallel
