Vehicle-in-Virtual-Environment (VVE) Method for Developing and Evaluating VRU Safety of Connected and Autonomous Driving with Year 2 Focus on Bicyclist Safety
Motivated by the shortcomings of using public road development of autonomous driving functions, this project focuses on the Vehicle-in-Virtual-Environment (VVE) method of safe, efficient, and low-cost connected and autonomous driving function development, evaluation, and demonstration. The VVE method places the actual vehicle inside a highly realistic virtual environment with realistic virtual sensor feeds while the vehicle is physically moving in a large and empty testpad. This is as if the vehicle is using a virtual reality headset. It is possible to easily change the virtual development environment and inject rare and difficult events which can be tested very safely. This is the second-year proposal for a two-year project that focuses on the use of the VVE method for development and evaluation of Vulnerable Road User (VRU) safety functions. Pedestrian safety was the focus of the first-year project. The current proposal focuses on bicyclist safety and the use of the VVE method for its development and evaluation. Five FARS (NHTSA’s Fatality Analysis Reporting System) pedestrian crash scenario use cases are being considered in the first-year project. The research team has demonstrated FARS 750 (Crossing Roadway – Vehicle Not Turning) with their research vehicle and real pedestrian as part of the year 1 project and will present their results in the 2024 SAE World Congress and Experience (WCX). The team also started developing and testing a deep reinforcement learning based pedestrian collision avoidance system, the early results of which will also be presented in the 2024 SAE WCX conference. The proposed year 2 project will focus on the bicyclist crash scenario use cases of: Motorist Overtaking Bicyclist (FARS 230), Bicyclists Failed to Yield at Midblock (FARS 310), Bicyclist Failed to Yield at Sign Controlled Intersection (FARS 145), Bicyclist Left Turn / Merge (FARS 220), Motorist Left Turn / Merge (FARS 210). Vehicle-to-VRU communication-based bicyclist detection which also works for non-line-of-sight cases will be combined with perception-based detection within the VVE method in the year 2 project. The Deep Reinforcement Learning (DRL) method the team is developing in the first-year project for pedestrian protection will be further developed for bicyclist protection. The team will be using hierarchical DRL to improve the training times by making DRL generate a collision free trajectory modification of the vehicle while the trajectory tracking control will be treated separately. Vehicle trajectory modification to avoid a possible future collision will be developed and evaluated safely 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. Robust and delay tolerant trajectory control will be developed and evaluated using the VVE method also 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 will be developed and evaluated first in MIL and HIL, followed by development and evaluation using the VVE method. The results will also be applicable and extendable to the safety of scooterists.
- Record URL:
Language
- English
Project
- Status: Active
- Funding: $180158
-
Contract Numbers:
69A3552344811
-
Sponsor Organizations:
Office of the Assistant Secretary for Research and Technology
University Transportation Centers Program
Department of Transportation
Washington, DC United States 20590 -
Managing Organizations:
Carnegie Mellon University
Pittsburgh, PA United StatesSafety21 University Transportation Center
Carnegie Mellon University
Pittsburgh, PA United States 15213 -
Project Managers:
Stearns, Amy
-
Performing Organizations:
Community College of Philadelphia
1700 Spring Garden Street
Philadelphia, PA United States 19130 -
Principal Investigators:
Saxton, Richard
- Start Date: 20240701
- Expected Completion Date: 20250630
- Actual Completion Date: 0
- USDOT Program: University Transportation Centers
Subject/Index Terms
- TRT Terms: Autonomous vehicles; Connected vehicles; Crash avoidance systems; Cyclists; Machine learning; Pedestrian safety; Trajectory control; Virtual reality; Vulnerable road users
- Subject Areas: Highways; Pedestrians and Bicyclists; Safety and Human Factors; Vehicles and Equipment;
Filing Info
- Accession Number: 01933385
- Record Type: Research project
- Source Agency: Safety21 University Transportation Center
- Contract Numbers: 69A3552344811
- Files: UTC, RIP
- Created Date: Oct 12 2024 11:45AM