Project: #484 Vehicle-in-Virtual-Environment (VVE) Method for Developing and Evaluating VRU Safety of Connected and Autonomous Driving with Year 2 Focus on Bicyclist Safety Progress Report - Reporting Period Ending: March 31, 2025 Principal Investigator: Levent Guvenc Status: Active Start Date: July 1, 2024 End Date: June 30, 2025 Research Type: None Grant Type: Research Applied Grant Program: US DOT BIL, Safety21, 2023 - 2028 (4811) Grant Cycle: Safety21 : 24-25 Progress Report (Last Updated: March 27, 2025, 7:41 p.m.) % Project Completed to Date: None % Grant Award Expended: None % Match Expended & Document: None USDOT Requirements Accomplishments The main goal and objective of this project was to use the Vehicle-in-Virtual-Environment (VVE) method to develop and evaluate approaches for the safety of vulnerable road users (VRU) in their interactions with connected and autonomous vehicles. While pedestrian safety was treated in our first-year project, this second-year project focuses on bicyclist safety. Vehicle-to-VRU communication-based bicyclist detection which also works for non-line-of-sight cases was combined with perception-based detection in our year 2 project. The Deep Reinforcement Learning (DRL) method developed in the first-year project for pedestrian protection was further developed for bicyclist protection. We used the hierarchical DRL to improve the training times by making DRL generate a collision free trajectory modification of the vehicle while the trajectory tracking control was treated separately. Work on vehicle trajectory modification to avoid a possible future collision and its safe evaluation using the VVE approach with the vehicle and bicyclist at separate locations physically but on a collision risk path in the virtual environment which will enable very realistic evaluation of the designed VRU safety function is in progress. MIL simulations of the considered high crash risk scenarios were conducted. We have been working on our HIL simulator, upgrading it and adding/updating the Unreal Engine co-simulation capability as a preparation for our VVE deployment. We used the parameter space and disturbance observer methods to develop robust and delay tolerant trajectory control for executing the calculated collision free modified vehicle trajectory which may involve slowing down, braking, or braking and steering. Virtual environments and collision risk scenarios were developed and evaluated in MIL while HIL evaluations are in progress. The HIL evaluations will be finished and the VVE evaluations will be conducted in the second half of the current project. While one doctoral student worked as a graduate research assistant funded by the project, two other doctoral students in the Department of Mechanical and Aerospace Engineering of the Ohio State University also took part partially in the project research activities, thus, receiving research training and professional development. Project results were disseminated in 1 journal paper (published), 1 conference paper (accepted), two invited talks (Smart Safety Connection Seminar Series, OTEC) and one interview (by students of Public Policy course at OSU). We also participated in the Deployment Partner Consortium Symposium with a poster. Impacts As part of transportation workforce development, one doctoral graduate student took part directly and two other doctoral graduate students took part partially in project work. The first graduate student was supported as a GRA. These three students were trained on research on bicyclist/pedestrian safety, safety of connected and autonomous driving, robust and time delay tolerant path tracking control, HIL simulations, and the Vehicle-in-Virtual-Environment method. One undergraduate student (Mechanical and Aerospace Engineering) and one M.S. degree student (Electrical and Computer Engineering) were trained on Unreal Engine scene generation and perception sensor data collection for the VVE method and MIL/HIL simulations. It is expected that they will continue their training which will also be the basis of their undergraduate B.S. thesis and M.S. Plan B thesis, respectively. We are also recruiting other undergraduate and M.S. degree students for training. One SAE WCX 2025 paper was accepted for publication and presentation with very high review marks and will be presented by the doctoral students on April 8 in Detroit. The SAE meeting is well attended by the automotive and transportation industries. One journal paper was published in Electronics (impact factor 2.6). PI L. Guvenc taught the course ME 8322 Vehicle System Dynamics and Control during the Autumn 2024 semester and is currently teaching the ECE 5553 Autonomy in Vehicles course this Spring 2025 semester. Both courses treat the safety of road transport and help with the transportation workforce development goal of the UTC. PI L. Guvenc also taught the ME 8352 Robust Control of Mechatronic Systems course in the Autumn 2024 semester and this course is related to robust and delay tolerant controls and hence also helps with transportation workforce development. Pi L. Guvenc is also teaching ME 3360 System Integration and Control during Spring 2025 which is helping in recruiting undergraduate students. Other Courses: ME 8322 Vehicle System Dynamics and Control ECE 5553 Autonomy in Vehicles ME 8352 Robust Control of Mechatronic Systems ME 8360 System Integration and Control Further development of the Vehicle-in-Virtual-Environment method and evaluation environments and tests. Outcomes New Partners None at the moment but there will be new partners for our Year 3 project. Issues The DRL training and Unreal Engine simulations require high performance computing. We have been re-configuring the PCs we use with the more capable GPU boards we have as an intermediate solution. Our HIL setup was using older versions of Matlab, dSpace software and Unreal Engine as we wanted to be compatible with the real time CarSim vehicle model we were using. As updating CarSim turned out to be too expensive for the project budget, we switched to our own higher fidelity Simulink vehicle modeling and updated all other software accordingly. We are now working on synchronized real time operation of all our models in the HIL setup which we use as a preliminary step before VVE testing.