Shared Autonomous Vehicle Scheduling Research Based On Connected Vehicle Environment

With the development of deep learning and 5G network technology,connected vehicle system and autonomous vehicle have become the focus of attention.The use of autonomous vehicle to provide urban taxi service can reduce the labor cost and capacity instability of taxi service,which is known as SAV service.The connected vehicle system combined with 5G technology can quickly integrate urban traffic information,and provide decision-making basis for SAV scheduling.Properly scheduling of urban SAVs to provide better service for passengers has become a hot research topic.This paper studies the scheduling problem of SAV based on the Internet of Vehicles environment,and designs a SAV scheduling platform architecture.This paper focuses on the problems of dispatching orders and the matching of zone-based supply and demand in the platform architecture,modeling the dispatch problem as the maximum weighted matching of weighted bipartite graph model,and in the zone-based scheduling,the K-Means clustering algorithm is first used to divide the traffic zone.Based on the timeliness of demand,the zonebased supply and demand matching problem is modeled as the minimum cost maximum flow model to solve zone-based scheduling scheme,and the scheduling rules are designed to convert the zone-based scheduling scheme into the actual available scheduling scheme.Taking the yellow taxi data from March 1st to 7th,2011 in Manhattan,New York City as an example,assumed that the zone-based demand forecast is accurate,this paper compares the order-served rate and average waiting time of different scheduling models.The results show that using minimum cost maximum flow model to balance zone-based demand and supply,the cycle length is 20 minutes,and the weighted bipartite graph model is used to dispatch order for SAV,the operating scheduling system performs optimally.7440 SAVs can meet 98% of the daily travel demand in the Manhattan,the lower quartile of the hourly order-served rate in 7 days is 98%,and the upper quartile of hourly average waiting time is 1.5minutes,which is better than the performance of traditional model.