Autonomous Intersection Management: A Heuristic Approach

Abstract

Self-driving cars do not sound like a part of science fiction movies anymore. With most of the automobile giants committed to launching their autonomous cars as soon as possible, it is expected that self-driving cars will hit the road in a very short time. Increasing the penetration factor of autonomous vehicles will need algorithms to control them in an efficient way in different scenarios. One such scenario is intersection management (IM). We propose an intuitive heuristic to resolve space–time conflicts in vehicle’s trip, enabling vehicles to cross the intersection safely, and with minimum delay. The intersection is modeled as a group of conflict points where intersecting internal links cross each other. For safe and efficient scheduling of vehicles, heuristics are proposed separately for: 1) entering in approach lane; 2) safe traversal in the approach lane; 3) safe and efficient crossing of intersection; and (4) safe departure along depart lane. Vehicle scheduling is done using a three-leveled heuristic in which each level serves a distinct purpose and together guarantee collision-free passing of vehicles. Performance evaluation of the proposed scheme is done in Simulation of Urban MObility simulator and a comparative analysis is done with three other IM schemes. Experiments show that the proposed scheme gives a smaller average trip delay of vehicles compared with other IM schemes.