Infrastructure Guided Self-driving Vehicles Flow Optimization in Street Network with Batch Algorithm and Access Control

Abstract

Metro cities all around the world are facing increasing congestion problems. In congested city traffic, most of the delay and fuel consumption occur at the intersections. Infrastructure guided self-driving (IGSD) technology raised the potential to control the path and speed of IGSD vehicles in such a way that traffic flow efficiency is greatly improved. IGSD vehicles approaching the same intersection are grouped into batches based on non-conflict phases, each of which traverse the intersection without stopping. The greater challenge is to factor into the knock-on effect among intersections, and to create a coordinated solution for IGSD vehicle flow in the entire network. We propose a Batch Algorithm which is used in rolling horizon fashion in a 100% IGSD environment, that determines the velocities of vehicles in such a way that no vehicle has to stop while traversing the network, and that minimizes the total travel time. An Access-control (AC) method is also proposed to minimize the knock-on effect and consequent gridlock, which can be applied with most traffic management mechanisms with minor modification. The approaches are experimented in a dynamic and stochastic setting with the open-source simulator SUMO for Sioux Falls network and compared with other solutions proposed in the literature. The numerical results demonstrate that Batch algorithm with AC method significantly outperforms other traffic control methods and increases the throughput of the network by 45%.