Mcity says open-source digital twin enables cheaper AV testing

The first open-source digital twin of the Mcity Test Facility—the University of Michigan’s center for connected and autonomous vehicles and technologies—is now available to the public. The university said the digital twin gives researchers around the world access to a new free tool.

A digital twin is a virtual representation of a physical environment that also exchanges data with that environment, enabling simulation and testing.

Mcity claimed that its new digital twin, developed with support from the National Science Foundation (NSF), is the first open-source digital twin for mobility systems testing. The organization said it provides a faster, safer, and less expensive way to test autonomous and connected vehicle software.

Researchers anywhere can use the facility’s features—with a variety of road materials, markings, signals, and intersections—to test their autonomous driving algorithms without having to make the trip to Ann Arbor, Mich.

“This takes our almost 10-year-old track and puts the digital replica directly over it,” said Greg Stevens, Mcity’s director of research. “That’s a living, breathing manifestation of that physical track where people can do mixed reality testing and development.”

The digital twin works with TeraSim, an open-source traffic simulator developed by Mcity researchers. It introduces other road users, such as pedestrians, cyclists, and other drivers, and generates safety-critical events like potential collisions.

Using traffic behavioral models calibrated with real-world data, TeraSim can simulate both normal and high-risk driving scenarios.

Register today to save 40% on conference passes!

About the Mcity Test Facility

Opened in 2015, Mcity is a purpose-built proving ground for connected and automated vehicles. Its physical features include:

• A simulated downtown area with urban streets
• A 1,000-foot straightaway, plus access ramps, a curve, and a traffic circle
• Multiple road surfaces with a variety of road markings and crossing types
• Traffic signals and traffic signs
• Bridge deck, underpass, guardrails, barriers, and crash attenuators
• House and garage exterior with accessibility ramp for first-mile/last-mile testing, deliveries, and ride-hailing

Mcity said it is continuously adding new features. Officials have worked to make the facility and its technologies available to a larger group of researchers.

In 2022, a $5.1 million grant from the NSF funded the development of Mcity’s digital infrastructure, enabling remote use of the Mcity Test Facility—dubbed Mcity 2.0. The 2022 grant also funded the creation of the digital twin.

The potential impact of remote testing 

Mobility officials have touted autonomous vehicles (AVs), along with connected vehicles and infrastructure, as a path to safer, more efficient, and equitable transportation.

Still, a great deal of research remains before the technologies will produce everyday benefits for a large portion of the population, acknowledged Mcity. Testing in the real world takes time and money and raises a host of safety concerns, but virtual testing can enable control software to demonstrate a high level of safety ahead of that stage.

“You can drive millions of miles in your AV in a digital twin built off of a real-world environment before your AV actually touches the real world,” said Darian Hogue, an Mcity software engineer who helped develop the digital twin. “With this, we can control all kinds of factors. That includes controlling and manipulating simulated pedestrian traffic—a factor that is random in the real world. This focuses and accelerates simulated testing.”

Mcity officially launched its remote use capabilities in October. Researchers operating from their home bases can test their autonomous algorithms in virtual and mixed-reality environments by connecting to Mcity’s cloud-based digital infrastructure.

Using 5G wireless communications, researchers can control physical vehicles and traffic signals on Mcity’s test streets and receive real-time data in return—all while protecting proprietary information.