By on July 22, 2015


The facility was mostly deserted by the time I got there deliberately late to avoid politicians’ speechifying. Between the very realistic — but empty — roadways with functional traffic lights, railway crossings, and even parking meters, on one hand, and the two city blocks of obviously faux buildings, theatrical scrims really, on the other, I felt that at any second, things might switch to black and white and Rod Serling would step out from behind one of the backdrops.

I wasn’t in the Twilight Zone, though. I was on a gentle hillside on the north side of Ann Arbor.

While Serling’s landmark science fiction and fantasy show sometimes gave us a peek at what the future might look like, the 32 acre Mcity autonomous and connected vehicle research facility that opened yesterday on the University of Michigan’s north campus is definitely one of the places where the future of personal transportation will be made.

7/15/15               2015 UM Aerials -July                 MCity, North Campus, Munger Grad                            Residency,Campus construction.

At first glance, Mcity does look more like a movie set than a high tech R&D installation and the $10 million that it cost to build doesn’t seem like much to an industry where it costs a billion dollars to develop a new car, but the facility’s promise is such that almost 50 private sector companies are partners in the project.


The fact that Mcity even exists is because of a unique partnership between the public and private sectors. The state of Michigan, via MDOT, put up $3 million, with UofM, a public university, also kicking in part of the construction costs. It looks like operational costs, at least for the next three years, will be covered by the private sector. Fifteen companies make up what is being called the Leadership Circle, which means they’ve each pledged $1 million to be paid out over three years. The Leadership Circle is a veritable who’s who of the industry including Ford, GM, Honda, Nissan, Toyota, Delphi, Denso, Bosch, Xerox, Verizon and Qualcomm. Another 32 firms have committed $150,000 each.


The members of the Leadership Circle aren’t being completely altruistic. That level of contribution gives those companies access to all the research that will be performed at the facility, regardless of who is doing or sponsoring the work.

That level of cooperation between car companies would not have been possible a generation or two ago, when the U.S. Dept. of Justice had automakers so worried about possible charges of collusion or violating anti-trust statutes that auto executives wouldn’t play golf with their counterparts for fear of an investigation. Of course, that level of cooperation will be necessary for the development of technical standards for vehicle to vehicle communication and other aspects of connected and autonomous driving.


The interest of the auto industry in Mcity is obvious — they can use it to test their tech. The interest of the University of Michigan is also pretty obvious. UoM’s engineering school has long had a very close relationship to the auto industry and the school wants to stay on the cutting edge along with their corporate partners. Mcity is part of the university’s Transportation Research Institute. The state of Michigan’s interest is keeping itself the global center of the auto industry.


With Elon Musk dissing “Detroit” at every opportunity (while staffing Tesla with engineers and designers with the domestic automakers on their resumes), tech behemoth Google testing a fleet of driverless cars, and even rumors of Apple getting into the car biz, there has been boasts from the West Coast and concerns from the rust belt that as computer technology becomes more and more important to how cars work Silicon Valley might displace Detroit as the center of innovation in the auto industry.

Mcity is the state of Michigan saying, “not so fast” to Silicon Valley. So far, it appears to be the only facility of its kind in the world.

The site is realistic down to the handicapped parking spaces and meters. So far no meter maids.

The site is realistic down to the handicapped parking spaces and meters. So far no meter maids.

The test environment (Mcity is adverse to calling it a mere “test track”) is supposed to replicate “the broad range of complexities vehicles encounter in urban and suburban environments.” There are about five miles of roadways including, intersections, traffic signs and signals, sidewalks, benches, simulated buildings (both fixed and moveable), street lights, and obstacles such as construction barriers, “pedestrians”, and railroad crossings. A variety of road surfaces are used including concrete, asphalt and brick paving as well as metal grid and gravel surfaces. There’s a 1,000 foot straight along with a variety of curves and ramps with different radii. There are two, three and four lane roads as well as bike lanes and a roundabout. What looks like a covered bridge is supposed to simulate tunnels and overpasses. One street is covered overhead with netting to reproduce the shade from a canopy of trees. There are parallel and angle parked cars as well as mechanized bicycles, fire hydrants and both fixed and variable lighting and traffic control devices.


The Mcity facility is supported by the University of Michigan, the Michigan Dept. of Transportation, and about 50 corporate partners, including 15 “Leadership Circle” members that have each pledged $1 million.

While there are ramps and signs that look like they do on an interstate highway, there will be no high speed testing at the facility, despite the ‘road train’ of cars and trucks traveling in unison on a highway in Mcity’s animated promotional video. The top speed allowed at the facility is just 40 mph.

Mcity looks very modern and tidy. Everything follows the latest designs for roadways, etc. and I have to wonder how a car developed at Mcity might do in an urban area that’s a bit more vintage — like San Francisco or Rome. I wonder if Mcity has developed a protocol for dealing with New York City levels of litter.

Aerial photo and infographic courtesy of the University of Michigan. All other photos by the author.

Ronnie Schreiber edits Cars In Depth, a realistic perspective on cars & car culture and the original 3D car site. If you found this post worthwhile, you can get a parallax view at Cars In Depth. If the 3D thing freaks you out, don’t worry, all the photo and video players in use at the site have mono options. Thanks for reading – RJS

Get the latest TTAC e-Newsletter!

25 Comments on “Michigan to Stay Ahead of Silicon Valley with “Mcity” for Autonomous & Connected Car R&D...”

  • avatar

    important detail never addressed in articles that i’ve read:

    will robo-cars be programmed to strictly follow your state’s vehicle code? “by the book” nearly everyone breaks some law (speed, following distance, signalling, no passing in the right lane, etc) every hour on the road.

    and for the environmentalists out there…..robo-cars will make sprawl/congestion worse. why would anyone in a big city bother with mass transit or leaving earlier/later for work when you can robo-drive.

    anytime technology makes anything easier (like making kitten videos), humans will do it more often.

    • 0 avatar

      1) I think eventually there will come to be accepted practices where robo-cars slightly shade the laws just like human drivers do today. This won’t be in the areas you reference — robo-cars can do all of that flawlessly — but in the area of handling pedestrian and other vehicle traffic at urban intersections. If you strictly followed crosswalk laws (leave a full lane between yourself and any pedestrian) you could never drive through an intersection in a big city. Eventually robot cars will figure out how to mimic courteous city drivers: yield to pedestrians, but don’t leave quite as much room as a strict reading of the law requires, and inch forward when safely possible.

      2) Parking and capacity at bottlenecks (typically freeway exits) will limit robo-car capacity more or less like they limit car capacity today. Parking will be expensive, and inevitably there will be tolling or congestion pricing that addresses the worst bottlenecks in urban environments. That will ensure there is still an economic incentive for urban governments to operate, and people to use, mass transit systems. Outside of dense urban environments, there is enough room for vehicles and parking, and subsidized robo-car rides could replace transit for the poor and disabled.

      • 0 avatar


        One take-away from the Google talk I posted is that the test-cars they’re driving around are recording all of the time.

        They use the recordings to test their nightly builds of the software in a simulated environment.

        It’s pretty much how the rest of Silicon Valley does things these days: test driven environment, and continuous integration.

        Don’t break the build — or else your co-workers will mock you until you fix it…

      • 0 avatar

        I expect the driverless car to have its place alongside that other success of the modern automotive world, the flying car.

        • 0 avatar

          There are two big reasons the flying car failed: 1) It’s really, really hard to train people to fly, and unless the flying car is 100% autonomous there is no way around that. 2) Flying a single person routinely is very fuel-inefficient.

          Neither of these applies to autonomous vehicles. It’s purely a matter of teaching a computer to drive on public roads, which is not magic. It’s just a slow iterative process requiring lots of data collection and testing to find odd situations and develop rules for handling them.

          • 0 avatar

            You have a spectacular crash with a car without a driver, and the reaction will be a collective “WTF! Why wasn’t there a human there to act as a fail safe???”

            We will not have a situation in which cars will be truly driverless because we won’t tolerate the consequences (and automakers won’t want the liability.) I would expect the driverless car to be akin to cruise control, with a car doing more work than it does now but also with a human who is still ultimately responsible for what happens.

          • 0 avatar

            We’ve had this discussion before, but liability concerns get dealt with. The liability issue has not stopped a single technology from developing, ever. In the end it’s just money, and the potential payoff is easily worth a $10M claim or two.

            The details will be worked out by insurers and risk management staff, because they always are. And what the insurers and risk management staff know that the general public won’t (at least at first) is that the total risk will be far, far, far lower with autonomous cars. Remember, they are paying out million-dollar claims every day right now. They want to stop.

          • 0 avatar

            People will want human backup systems because there will be times when the technology fails and we won’t see any justification for not having had humans in the driver seat to intervene. It’s really that simple.

          • 0 avatar

            Pch101, those “exceptional moments” are going to be outweighed by a factor of hundreds (literally) by similar accidents that human drivers cause. The statistics are going to be so obvious and arresting that people will get past the one accident. And if they don’t, the insurance companies will force them to.

            I expect autonomous cars to cause less than 1% as many accidents, probably a small fraction of 1%, as human drivers.

          • 0 avatar

            Exceptional moments don’t get outweighed.

            Imagine that you’re the automotive industry guy or Congressman or whoever who pushed for fully automated cars, and now YOU have to explain to an angry public why Grandma now looks like the stuff in a can of Alpo.

            Do you think that rattling off statistics is going to help you to justify something that could have been avoided had a sober licensed driver been behind the wheel? Do you think that Grandma’s family and their attorneys will be content with “gee whiz, shiite happens”?

          • 0 avatar

            “And if they don’t, the insurance companies will force them to.”

            Why would insurance companies not want a human backup system to exist?

            I’m with PCH on this, ultimate cruise control is as far as this is getting on personal transportion during my lifetime.

          • 0 avatar

            Grandma’s family and their attorneys will have to be happy with “shite happens” when there are 200 other Grandmas and their 200 attorneys saying “how can you still let humans drive when they kill so many people and autonomous cars would have prevented all of these deaths?!”

          • 0 avatar

            ajla, insurance companies will force the transition because humans, backup or not, cost them so much damn money. Money talks.

            If you give a human the capability to act as backup, at least by giving them direct control of the car, they will screw it up much more often than they’ll save the situation. They’ll be unprepared, out of practice, and suffer the usual human recklessness and cluelessness.

            What probably will happen is that when the computer can’t understand how to resolve a situation it will stop in a safe place and ask the human for *advice*, not direct control, about what it should do. The computer will ask something like “Should I try to drive around that log on the left, drive around it on the right, turn around and find a different route, or just wait and call for help?”

        • 0 avatar
          Master Baiter

          Spot on. There’s a lot of money being flushed down the toilet on this technology.

          • 0 avatar

            The Google autonomous car crew wants their system to be statistically safer than a human driver by a huge margin before it goes out in to the world.

            The Tesla people say it’s a labor saving device, and the human driver is still in command of the vehicle.

            You can buy Tesla’s product, if you’re rich enough. You can’t buy Google’s system at any price.

          • 0 avatar

            Google is claiming most of the accidents involving their cars are from being rear-ended.
            As autonomous cars increase,expect heavy,heavy pressure to get human driven cars off the road-or so heavily taxed as to make them impractical for most people.
            Way down the road,but it will happen.

          • 0 avatar

            When Grandma gets run over in a crosswalk by a guy who was playing video games, drinking a Long Island Iced Tea and having a wank while his autonomous car turned her into hamburger, people are going to wonder why he wasn’t there to stop it.

            It makes no difference whether autonomous cars are safer generally (which they should be.) It will be those exceptional moments that lead to the policy that requires a driver. Technology does not work 100% of the time, and we are not going to accept “wasn’t me” as an excuse for allowing a human to get killed by a malfunctioning autopilot.

          • 0 avatar


            You should really listen to the TED talk by Chris Urmson about what Google is trying to achieve with their self-driving car:

            Granted it’s vaporware at this point, but they’ve at least *considered* the point you raise about accidents and taken it to heart.

            I have no idea if they’ll deliver (it’s TED, so grandiose claims are required regardless of their sanity), but they’ve at least recorded a ton of real-world data and are using it to test the nightly builds of their software in simulation.

            Even if you think he’s delusional, the approach they’re using is thought provoking.

            They’ve set a bar for themselves that is deliberately chosen to address exactly the concern you raise.

            Again, I have no idea if they can do it, I’m glad someone is trying it.

          • 0 avatar

            Unless Google has coders who can reprogram people, they can’t do anything to address my point.

            We will not tolerate some dude killing or maiming in his automated car, drinking a beer while he watches it happen right in front of him.

            It is natural and human to demand that people do something, and not just act like bystanders. And that means having a driver.

            If the unintended acceleration cases tell us anything, it’s that we have this underlying fear that the machines are going to have a mind of their own, rise up and kill us. We will not tolerate HAL being behind the wheel without someone to stop him.

          • 0 avatar

            Incidentally, I just watched your video. Sorry, but it doesn’t address my point at all.

            Google is just assuming that everyone will be pleased with a lower fatality rate, and that there won’t be any desire to blame the machines for the crashes that do happen. They also seem to think that people will believe that it’s acceptable for a driverless car user to kill others without making any attempt to intervene. That’s horribly naive.

    • 0 avatar

      I don’t know what the fully autonomous car people are doing. It seems like they’d be by-the-book — but I don’t see the Google cars going slower than traffic around Palo Alto, so maybe they’ve been programmed go with the flow.

      Tesla’s autopilot is more like cruise control. You can adjust your following distance. It will see and read speed limit signs, but there’s probably a “speeding factor” that you can set, too. Oh, and it will steer to keep the car in the lane and brake when the lane is obstructed.

      The TED talk that I posted below ( is basically a product manager from Google telling the world why he thinks Tesla’s incremental approach isn’t the right way — and that a moonshot like Google is doing is better. I particularly like the story about the person in a wheelchair chasing a duck.

      The technology is young yet, and the question about whether these things can bend traffic rules is something that can be tuned easily with a software update. I suspect that the bias will be toward driving by the book, but with some respect for reality.

      • 0 avatar

        I have a car with Adaptive Cruise Control. I spent the extra $$$ to have something that makes it easier on the freeway to let the CAR figure out how efficiently to drive, and not my admittedly heavy right foot! (It also helps me maintain my sanity when I set it to 7 over in a 25 or less, 5 over in a 30 or more, 8 over on the freeway, when I come up on the inevitable driver whose only regard for traffic laws is some random underposted number on a sign!)

        My biggest fear is that I end up in one of these autonomous cars, helplessly belted into the passenger seat, as a child runs out right in front of me on a two-lane, 40mph street, and the options are to go head-on into the oncoming semi, or headlong into the stout oak groves to the right! If I judged it right, perhaps I’d try hitting the brakes while steering; would an automated system do the same?

        I don’t want to be the beta tester/guinea pig for this situation!

  • avatar

    “Michigan to Stay Ahead of Silicon Valley with “Mcity” for Autonomous & Connected Car R&D”

    LOL, no. That’s not how this works.

    It’s a useful facility for companies anywhere who’s willing to put their test vehicle on a car and drive it there.

    Tesla and Google seem to be focusing on making the car react to what it sees, rather than on the connected car.

    Here’s what Google wants to tell the public:

    Here’s what Tesla wants to tell the public:

    The connected car is a “nice to have” for these guys.

    I haven’t seen anything comparable from the old-school auto folks.

    • 0 avatar

      I think of it this way:

      Silicon Valley is a cluster of expertise on information technology.

      Michigan is a cluster of expertise on manufacturing, with other expertise there to support manufacturing.

      You need both to make an autonomous car. I don’t actually know, but I suspect that Tesla has been dealing with this by hiring engineers from the old-school auto industry, and managers and programmers from Silicon Valley. At least that’s what I’d do if I were in Musk’s shoes.

      I don’t know how the big 2.5 are dealing with this. They’re mostly making more modern/efficient versions of the same old stuff, and selling a lot of them.

  • avatar

    Guy in that hacked Cherokee had an autonomous car for a while.

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • dal20402: Utterly gorgeous, and such a nice antidote to most of the cars built in the period. Such a shame that...
  • Russycle: My high school had one of these in the 70s. We had a TV studio and this was our mobile production van, in...
  • Funky D: So will the “premium” Mercedes herald the return of the bank-vault build quality, the...
  • ToolGuy: Now that GM can guarantee Zero Crashes, I welcome the return to the U.S. market of cabover light-duty...
  • Funky D: I had been jonesing for a Gladiator since they first came out, but wasn’t sure if I could live with it...

New Car Research

Get a Free Dealer Quote

Who We Are

  • Adam Tonge
  • Bozi Tatarevic
  • Corey Lewis
  • Jo Borras
  • Mark Baruth
  • Ronnie Schreiber