基于路段全局拓扑的协同感知视野优化分配方法  

An optimized allocation method of cooperative perception view based on road segment global topology

作　　者：刘茜萍[1,2] 吴孝宇 张琳[2] LIU Xiping;WU Xiaoyu;ZHANG Lin(Jiangsu Key Laboratory of Big Data Security and Intelligent Processing,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China)

机构地区：[1]南京邮电大学江苏省大数据安全与智能处理重点实验室,江苏南京210023 [2]南京邮电大学计算机学院,江苏南京210023

出　　处：《南京邮电大学学报（自然科学版）》2024年第6期76-86,共11页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：国网科技项目(5700-202318292A-1-1-ZN)资助项目。

摘　　要：基于协同感知技术,可将路段上若干车载摄像机获取的局部视野上传至路边架设的边缘服务器,进而融合出整个路段全局视野,以助于做出智能辅助驾驶决策。然而,直接将路段上所有车辆的视野数据悉数上传融合将带来极大的通信和计算开销,难以保证时效。为此,提出基于路段全局拓扑的协同感知视野优化分配方法,基于连通组件构建路段全局拓扑,进而在拓扑引导下获取路段视野初始分配方案,并提出分段式并行遗传算法进一步优化该方案,以尽可能减少冗余数据的上传,降低MEC边缘服务器负载,从而提高路段全局视野融合效率,其可行性和有效性通过仿真实验得到了验证。Based on the collaborative perception technology,the local views captured by several onboard cameras on a road segment can be uploaded to the edge servers deployed on the roadside.And these videos can be fused to generate the overall global view of the road segment,benefiting the intelligent driving.However,directly uploading and merging the visual data from all vehicles on the road segment can lead to significant communication and computational overhead,as well as challenges for timeliness.To address this issue,we propose a collaborative perception view optimization and allocation method based on the global topology of the road segment.This method constructs the global topology of the road segment based on connected components.Thus,an initial allocation scheme for the road segment views is obtained according to this topology.This scheme is further optimized by a segmented parallel genetic algorithm.The goal is to minimize redundant data uploads,reduce the load on the mobile edge computing(MEC)edge server,and consequently enhance the efficiency of the global view fusion for the road segment.The feasibility and effectiveness of the proposed approach are validated through simulation experiments.

关 键 词：车联网 协同感知 全局拓扑 近似算法 遗传算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]