Research Team: Hao Liu (lead) and Xiao-Yun Lu
UC Campus(es): UC Berkeley
Problem Statement: Testing the performance of Connected Automated Vehicles (CAV) in various road facilities and traffic scenarios is a critical step towards successful deployment of this advanced technology. Since large scale field CAV tests are difficult to conduct due to resource limitations and safety constraints, the hardware-in-the-loop (HIL) experiment is an alternative approach to expedite the CAV performance evaluation and system implementation. In a HIL test, researchers often put real-world CAV vehicles or their subsystems (e.g., vehicle dynamics controllers) in a controllable test environment (e.g., a test track or a chassis dynamometer). They use computer simulation tools to generate repeatable traffic flow conditions and inject the virtual flow into the test environment. The performance of the CAVs is then measured as the real system interacts with the artificial environment. Developing a HIL test requires comprehensive knowledge in vehicle dynamics control, traffic flow modeling, machine learning, and communication. However, few research teams have expertise in all those fields. ITS Berkeley has pioneered the development of CAV systems for over three decades and has made profound contributions on CAV control system design, traffic flow modeling, and development of deep learning algorithms for automated vehicles. Since those research deliverables are open to the public, they offer essential building blocks for any California institute to develop a HIL testbed.
Project Description: This project will develop a HIL test tool that integrates the existing CAV systems, communication algorithms, and traffic flow models, and streamline their combined applications. The HIL test tool will allow users to directly call programs that generate the virtual traffic environment. Users of the tool will not be required to possess detailed knowledge on traffic flow modeling. They will only need to provide basic traffic flow inputs (e.g., traffic demand, vehicle origins and destinations, and fleet composition). The tool will automatically execute the traffic simulation program to establish the simulated traffic scenario. In addition, the proposed tool will provide channels to enable real-time data interchange among individual systems in a HIL test and offer database functions to store raw data sets collected from the test systems and generate performance metrics based on the raw data. An example CAV test at the Richmond Field Station will demonstrate the detailed data collection and performance evaluation procedure.