Vehicle-in-Virtual-Environment Method for Autonomous Driving System Development and Evaluation
This technology is a Vehicle-in-Virtual Environment (VVE) method for testing and evaluating Autonomous Vehicles (AV). The VVE approach is a safe, reliable, repetitive, and scalable method of testing AVs, decreasing the risks and costs of testing AVs on public roads.
The demand for autonomous vehicles continues to grow as companies and cities realize the effect on the costs, safety, and efficiency that AVs have on everyday lives. Initiatives such as MCity in Detroit, Michigan, and Smart City in Columbus, Ohio, show a strong commitment to the development of autonomous vehicles and increasing demand. With that increasing demand is a necessity for safety and evaluation of the vehicle and pre-deployment demonstration of operation. AV testing and development relies heavily on simulation of the software, hardware and then combined with the vehicle. After a brief period of testing on a controlled proving ground area, the bulk of testing is done on public roads. Proving grounds are limited to what can be physically recreated in a closed lot, so the randomness and variety of public roads cannot be sufficiently recreated. It takes a large amount of time and resources to recreate physical scenarios in a closed lot. These issues lead to AVs not being thoroughly tested before being put on public roads leading to increased costs and risk of accident. Therefore, there is a need to test AVs more thoroughly in realistic environments before they are put on public roads.
Researchers at The Ohio State, led by Dr. Bilin Aksun Guvenc and Dr. Levent Guvenc , have developed a Vehicle-in-Virtual Environment (VVE) method to improve testing of AVs before the AV is evaluated on public roads. VVE technology creates a virtual environment that can simulate public roads and random events in a closed parking lot. The car is "seeing" the virtual environment but is driving around a closed lot. The virtual environment has different test scenarios including buildings, vegetation, infrastructure, other road users, weather effects, road irregularities, irregularly parked obstacles, deep billboard effects, suddenly appearing obstacles, false positives, false negatives, weather effects on sensor outputs and other sensor falsification effects. VVE provides for significantly more scenarios than what can be physically recreated at a fraction of the time and cost.
The Automated Driving Lab (ADL) at The Ohio State University where this technology was developed has expertise in automotive control, ADAS and connected and autonomous driving. They specialize in path planning and path tracking control, collision avoidance maneuvering, pedestrian and other vulnerable road user safety, data-driven end to mid autonomous driving, safe handling of connected intersections by autonomous vehicles, cooperative autonomous driving, fuel/energy efficiency using connectivity and cooperation and planning and evaluation of autonomous vehicle deployments. The ADL specializes in evaluation, testing and demonstration of connected and autonomous driving functions and deployments and has conducted the realistic simulation-based pre-deployment evaluation of the recent Linden Residential Area autonomous shuttle deployment of Smart Columbus.