Sentinel: An Onboard Lane Change Advisory System for Intelligent Vehicles to Reduce Traffic Delay during Freeway Incidents

Goodarz Mehr,Azim Eskandarian

DOI: https://doi.org/10.1109/TITS.2021.3087578

2021-09-06

Abstract:This paper introduces Sentinel, an onboard system for intelligent vehicles that guides their lane changing behavior during a freeway incident with the goal of reducing traffic congestion, capacity drop, and delay. When an incident blocking the lanes ahead is detected, Sentinel calculates the probability of leaving the blocked lane(s) before reaching the incident point at each time step. It advises the vehicle to leave the blocked lane(s) when that probability drops below a certain threshold, as the vehicle nears the congestion boundary. By doing this, Sentinel reduces the number of late-stage lane changes of vehicles in the blocked lane(s) trying to move to other lanes, and distributes those maneuvers upstream of the incident point. A simulation case study is conducted in which one lane of a four-lane section of the I-66 interstate highway in the U.S. is temporarily blocked due to an incident, to understand how Sentinel impacts traffic flow and how different parameters - traffic flow, system penetration rate, and incident duration - affect Sentinel's performance. The results show that Sentinel has a positive impact on traffic flow, reducing average delay by up to 37%, particularly when it has a considerable penetration rate. Working alongside Traffic Incident Management Systems (TIMS), Sentinel can be a valuable asset for reducing traffic delay and potentially saving billions of dollars annually in costs associated with congestion caused by freeway incidents.

Robotics,Signal Processing