Places like Google’s hometown of Mountain View, California, are often associated with self-driving cars. Unfortunately these real-world towns aren’t much help when it comes to putting specific safety features to the test in a highly controlled environment. What’s a researcher to do?

Build a simulated city, of course, which is exactly what researchers at the University of Michigan will be doing this summer. The Mobility Transformation Facility, which will be operated by the university’s Mobility Transformation Center, will consist of 32 acres of simulated city center complete with a four-lane highway.

The test environment will have merge lanes, stoplights, intersections, roundabouts, road signs, a railroad crossing, building facades, construction barrels and eventually a mechanical pedestrian. It will model the kind of connected and automated mobility system that the center aims to enable in Ann Arbor by 2021. Such a system could dramatically reduce crashes, ease traffic and reduce pollution and energy use.

Michigan Engineering researchers will initially use the facility to run tests on an automated Ford Fusion hybrid.

The project will allow researchers to test how automated and networked vehicles respond to rare but dangerous traffic events and road conditions. For example, they will have the ability to trigger events like ill-timed jaywalking and tricky traffic signal timings in an effort to see how vehicles respond.

It’s a step that researchers say is vital in making sure self-driving cars can operate safely in the real world.