International Conference on Autonomous Agents and Multi-agent Systems
Abstract
Autonomous Intersection Management (AIM) is a reservation-based intersection control protocol that leverages the capacities of autonomous vehicles to dramatically reduce traffic delay at intersections. AIM was designed for the time when all, or most, of the vehicles on the road are fully autonomous. However, we anticipate that there will be a long transition period during which many cars are still driven by human drivers and/or most vehicles have some but not all capabilities of fully autonomous vehicles. In order to accommodate this transition, this paper introduces a new multiagent protocol called Semi-Autonomous Intersection Management (SemiAIM), which allows vehicles with partially-autonomous features such as adaptive cruise control to make reservations in AIM. We propose a method for vehicles with limited autonomy to make reservations to enter an intersection in an AIM-like style and conduct extensive experiments in simulation to evaluate its effectiveness. Our results show that the delay of semi-autonomous vehicles in SemiAIM can be greatly reduced compared to human-driven vehicles.