Examining the Safety and Efficiency Implications of Several Autonomous Car-following Models

Research Lead: Wenlong Jin

UC Campus(es): UC Irvine

Problem Statement: Autonomous vehicles hold promises for revolutionizing transportation, but they also face many challenges in both safety and efficiency. Traffic system managers need to know what kind of autonomous vehicles should be allowed to operate on California’s roads and what driving behavior to expect from them. Based on daily driving experiences, traffic rules, vehicle characteristics, and empirical observations, sound driving behavior should be collision-free, cars should be safely spaced and travel at reasonable speeds. Autonomous cars’ driving behaviors should satisfy these principles to be safe, efficient, and human-like. These principles have been used to develop car-following models and have been incorporated into standard Adaptive Cruise Control (ACC) systems, but these models tend to be too complicated to analyze, and there is no guarantee that they satisfy safe driving criteria.

Project Description: This research will systematically examine the safety and efficiency of different existing car-following models, ACC models, Artificial Intelligence-autonomous vehicle driving models, as well as the performance of Positioning, Navigation and Timing (PNT) and other sensing technologies. Using real world data, the researchers will calibrate the braking distances and the speed profiles of human-driven vehicles at signalized intersections. The project will then develop a new driving model to describe human drivers’ behaviors and prescribe autonomous vehicles’ driving policies. Further, the team will examine the impacts of PNT accuracy on the safety and efficiency of the model. This research may help to increase users’ confidence in the safety and performance of the transportation system.

Status: In Progress

Budget: $87,000