An optimized allocation method of cooperative perception view based on road segment global topology
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.
国家科技期刊平台
NSTL
万方数据
