Listening to the Siren Call: Virginia Tech Works with NVIDIA to Test AV Interactions with Emergency Vehicles

Move over, self-driving cars.

The Virginia Tech Transportation Institute has received a federal grant from the U.S. Department of Transportation to study how autonomous vehicles interact with emergency vehicles and public safety providers.

VTTI, the second largest transportation research institute in the country, will use vehicles equipped with the NVIDIA DRIVE Hyperion platform to conduct these evaluations on public roads.

Emergencies or unexpected events can change the flow of traffic in a matter of minutes. Human drivers are trained to listen for sirens and watch for police officers directing traffic; however, this behavior may not be as instinctual to autonomous vehicles.

VTTI is working with NVIDIA as well as a consortium of automotive manufacturers organized through Crash Avoidance Metrics Partners (CAMP LLC) to study challenging and dynamic scenarios involving automated driving systems, such as encounters with public safety providers. Participating CAMP LLC members include General Motors, Ford, Nissan and Daimler. The team will also address ways to facilitate communications between these systems and with their supporting physical infrastructure.

The project will identify solutions and build highly automated Level 4 reference vehicles retrofitted with autonomous driving technology, as well as connected infrastructure to support them. In the final phase, VTTI and its partners will hold demonstrations on Washington, D.C., area highways to showcase the technology safely navigating challenging scenarios.

Safety First Safely maneuvering around emergency vehicles, including ambulances, fire trucks and police vehicles, is a key component to everyday driving.

The consequences of not doing so are serious. Over the past decade, ambulances experienced an average of about 170 crash-related delays per year, costing precious time in responding to and transporting emergency patients.

Additionally, not moving over for emergency vehicles is illegal. Every state has a move over law, requiring vehicles passing stopped police cars, ambulances or utility vehicles to vacate the nearest lane and slow down while passing.

Autonomous vehicles must comply with these traffic norms to deploy safely and at scale. AV fleets will need to be able to identify emergency vehicles, recognize whether lights or sirens are running and obey officers directing traffic.

Leveling Up with DRIVE Hyperion VTTI will use Level 4 autonomous test vehicles to study how this technology will behave in emergency scenarios, helping determine what measures must be taken in development and infrastructure to facilitate seamless and safe interactions.

NVIDIA DRIVE Hyperion is an autonomous vehicle data collection and perception platform. It consists of a complete sensor suite and NVIDIA DRIVE AGX Pegasus in-car AI computing platform, along with the full software stack for autonomous driving, driver monitoring and visualization.

The high-performance, energy-efficient DRIVE AGX Pegasus AI computer achieves an unprecedented 320 trillion operations per second. The platform is designed and built for Level 4 and Level 5 autonomous systems, like those being tested in the VTTI pilot.

The DRIVE Hyperion developer kit can be integrated into a test vehicle, letting developers use DRIVE AV software and perform data collection for their autonomous vehicle fleet.

Using this technology, researchers can quickly develop a test fleet without having to build from the ground up. The ability to collect data with DRIVE Hyperion also ensures an efficient pipeline of conducting tests and studying the results.

With the collaboration among NVIDIA, VTTI and its automotive partners, this pilot program is slated to significantly advance research on the safe integration of autonomous driving technology into U.S. roadways.

LINK:	https://blogs.nvidia.com/blog/2020/10/27/virginia-tech-test-av-emergen...
	See more stories from nvidia

North America Stories