Background
Despite the success of Autonomous Vehicles (AVs) in reducing crash risks has been proved by many field tests, their safety performance, under adverse driving conditions in winter seasons, still lacks comprehensive evaluations. Hence, to avoid costly mistakes before the widespread implementations, there is an urgent need to build a reliable cyberinfrastructure, a stochastic simulation platform more specifically, to pre-evaluate the capability of Automated Driving Systems (ADS) algorithms in dealing with icy/snowy driving conditions. To overcome those challenges, the major goals of the project are listed as follows:
Research Goals
Goal 1: Design a novel microscopic model to simulate human-driven vehicle (HV) behaviors on icy/snowy roads with the Physics Regularized Gaussian Process (PRGP) technique.
Goal 2: Develop an efficient, reliable, and accurate model to predict the crash risks of AVs, in a mixed-HV-AV environment, under adverse driving conditions.
Goal 3: Integrate the models and algorithms into an open-source software package, with comprehensive documentation and plenty of application cases.
Goal 4: Validate the simulation model with field data and develop transition-to-practice plans.