摘要
近年来,移动边缘计算和深度学习在自动驾驶的应用场景中引起业界的强烈关注.移动边缘计算通过将计算任务卸载至边缘服务器以降低网络负荷,减少自动驾驶信息传输时延;深度学习可有效提高障碍物检测的准确性,从而提升自动驾驶的稳定性和安全性.首先介绍MEC的基本概念和参考架构以及深度学习中常用的模型算法,之后从目标检测、路径规划、碰撞避免三个方面对MEC和深度学习在自动驾驶中的应用进行归纳总结,最后对目前研究中存在的问题与挑战进行讨论和展望.
In recent years,the application of mobile edge computing and deep learning in automatic driving has aroused enormous attention in the industry.Mobile edge computing is able to reduce the network load and the transmission delay of autopilot information by offloading the computing task to the server nearby.Deep learning can improve the accuracy of obstacle detection and improve the stability and safety of automatic driving.It first introduces the basic concepts and reference architecture of MEC,as well as the general used model algorithms in deep learning.After that,the application of MEC and deep learning in Autonomous driving is summarized from target detection,path planning,and collision avoidance.Finally,the problems and challenges in the current research are discussed and forecasted.
作者
黄磊
郑艺峰
张文杰
HUANG Lei;ZHENG Yifeng;ZHANG Wenjie(School of Computer Science,Minnan Normal University,Zhangzhou,Fujian 363000,China;Key Laboratory of Data Science and Intelligence Application,Zhangzhou,Fujian 363000,China)
出处
《闽南师范大学学报（自然科学版）》
2021年第4期39-47,共9页
Journal of Minnan Normal University：Natural Science
基金
福建省自然科学基金(2021J011004,2021J011002)
福建省教育厅中青年项目(JAT190392)。
关键词
移动边缘计算
深度学习
自动驾驶
车联网
mobile edge computing
deep learning
autonomous driving
Internet of vehicles
