Demand-for Graph and Its State Transition Expression Evaluating Traffic Congestion Due to CAVs Control

Automated driving technologies are expected to reduce traffic congestion. Research studies have been conducted on the use of multi-agents as a method for analyzing and resolving traffic congestion involving a mixture of automated and manually driven vehicles using connected automated vehicles (CAVs). Existing methods that use multi-agent flow and density can analyze the occurrence of congestion and its mitigation, but cannot sufficiently characterize the starting point of congestion and the extent of its impact. By contrast, this study analyzes the interaction of communication and actions within agents, in addition to flow and density to identify the starting point of congestion and the extent of its influence. Specifically, we represent traffic flows consisting of manually and automatically driven vehicles as multi-agent systems consisting of cooperative and non-cooperative agents to evaluate the traffic congestion owing to altruistic lane changes by CAVs. Then, identify the starting point of congestion and the extent of its influence, a demand-for graph, which is a graph representation of the interaction of actions and communications among agents, is incorporated into the network search of an already existing asymmetric simple exclusion process. By deriving the state-transition equations of the adjacency matrix of the demand-for graph, this study clarifies the starting point, expansion, and resolution speed of congestion size due to altruistic lane changes. The results of this study enable the derivation of congestion occurrence and resolution conditions using the CAV control method for multiple traffic conditions.