$25 Million Michigan Project Hopes to Add Cars to Internet of Things

17 9 Loading

The US Department of Transportation (DOT) is conducting a 12-month, $25 million study to see if cars sending data to each other over Wi-Fi can make driving safer. Cars talking to each other and maybe braking or swerving to avoid collisions? Very cool. Spending $25 million on a relatively limited test? Less cool. But we’ll get to that momentarily.

The test, funded by the US Department of Transportation links 3,000 volunteer cars in Ann Arbor, Michigan over Wi-Fi. The cars talk to each other and infrastructure built into portions of the road. It is the biggest “road test” of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication ever conducted. Put another way, it’s the only such large scale test.

V2V and V2I warns drivers if another similarly equipped car infringes on their lane, or runs a red light, or stops suddenly ahead of them.

DOT believes such warnings could prevent “four out of five unimpaired vehicle crashes.” And further, they hope it may eventually provide real time traffic monitoring, allowing commuters to choose the path of least resistance.

According to DOT Secretary Ray Lahood, “This cutting-edge technology offers real promise for improving both the safety and efficiency of our roads.”

It’s a good idea that deserves study. (Almost as good as these Dutch glow-in-the-dark smart highway designs.) But there are a few details worth a raised eyebrow.

Take for example, the $25 million in funding. For 3,000 devices, we’re talking over $8,000 per system. You might expect that price tag on a revolutionary new bit of tech.

Maybe a system that takes control of the car when the data warns of an impending collision. Or even something that drives the car all the time. (Which, by the way, is awesome and already happening. You can read about it here and here.)

Whatever is being installed in these cars, it’s not that. Undoubtedly, part of the cost is because the equipment is hooked into the car’s systems to monitor them directly. But otherwise, V2V will talk to other cars over Wi-Fi and warn drivers of detected hazards.

And that’s about it.

Thing is, smartphones already do this. They have GPS and an accelerometer; they are user-programmable; and many people have them. Think of how broad (and cheap) a test you could run with smartphones, a clever app, and a special wired connection to the car’s computer.

There’s reason to be skeptical of the study’s effectiveness given its scale too. The Department of Transportation (DOT) is billing it as the “largest-ever road test of connected vehicle crash avoidance technology.”

But large is relative. 3,000 cars will be running the system in Ann Arbor. How does that compare to all the cars on the road?

Ann Arbor has a population of 114,925. On average, 43.9% of Americans own a car. That implies Ann Arbor has something like 50,452 cars. Probably not exact—and not counting out-of-towners—but close enough.

Under those conditions, the study will equip six out of every hundred cars on the road. Given the average rate of accidents among licensed drivers is 8%, what’s the probability two cars equipped with these devices meet in an “almost” accident?

Even if it’s not zero, a handful of data points in a year aren’t enough to make conclusive statements worthy of $25 million.

And one more thing about traffic monitoring. To make that at all effective, you need way more cars in the study. But why bother when Google Maps already has it covered?

Again, this is a good idea—but for a startup or other private firm. For an organization answerable to investors, passionate about making the best possible technology, and realistically bringing it to market.

And there are organizations out there doing this research already.

Google’s self-driving cars may not converse with each other—but do they need to if they are equipped with radar? Probably not. And ostensibly, that could be a more elegant solution because you don’t have to compel every car on the road to adopt the technology and you don’t have to outfit all the roads with sensors.

Cars talking to each other on a dedicated Wi-Fi network is potentially a great idea. But allocating $1 million each to 25 plausible ideas seems more productive than allocating $25 million to just one plausible idea.

Discussion — 9 Responses

  • Chris Law December 1, 2012 on 11:32 am

    I agree with the author on this subject.

  • Brian Krassenstein December 1, 2012 on 11:54 am

    This coupled with Google’s system could really be a force to reckon with. In 5-7 years I think there will be far fewer accidents, and in 10-12 years , the number of accidental vehicle deaths should be so low it will barely show up on statistics.

  • Craig J. Townsend December 1, 2012 on 1:48 pm

    the STATE, high waste at a low level of productivity.

  • eldras December 3, 2012 on 6:18 am

    A sufficient map pf a sufficient number of Things will enable programmes to be available online to download into 3D printing devices.

    • turtles_allthewaydown eldras December 8, 2012 on 7:59 pm

      What does that have anything to do with this article?

  • faulksb December 3, 2012 on 8:22 am

    Government waste at it’s finest!

  • Enigmas Zauberer December 3, 2012 on 10:12 am

    You know I’m all for testing out new technologies that have the potential to make the road ways safer for all of us, but I agree with the author. With the current technology available on cell phones, why are they spending so much money and only 3,000 cars? Would the study not be better off in an area with a smaller population? I think that they could do several studies in several small towns with a population of say not more than 10,000 total, so that the results could be a bit more realistic. Ann Arbor is certainly to big for such a small number of controls.

    • turtles_allthewaydown Enigmas Zauberer December 8, 2012 on 8:31 pm

      I think you and the author don’t understand the engineering behind all of this. It is not “available on cell phones”. First, GPS for a mobile device is typically accurate to about 6-10 meters (differential WAAS when available brings that down to 10 feet/3 meters). That’s fine for determining what highway you’re on when traveling across the country, but that type of error is the difference between being safely behind the white line or in the middle of an intersection, of being in the same lane as another vehicle or across a median from that car. So this requires new tech, I’m not sure exactly how they have it set up, but it probably involves the V2I portion.

      Also, response times need to be very fast. A second or two delay doesn’t matter in a text message or even streaming audio, but it is too late when it comes to avoiding an accident. And there are complexities in determining who to talk to and how to coordinate when the units talk, when the vehicles and infrastructure around you is constantly changing and there is no one specific device in charge (a quickly and constantly changing ad hoc network). IPass/EZPass transponders for toll collection works pretty good today, but it doesn’t care what direction a car is, or what direction it’s facing, just that it it approaches and leaves. That doesn’t work for accident avoidance.

      There are a thousand details that have to be done right to give warnings that make sense, especially in the unexpected things that people sometimes do – travel the wrong way in one-ways, turn around in the middle of a block, veer across lanes or drive on roads not on the map. What if you lose GPS signal, does the whole system go down? This is an ambitious program, and will eventually probably be replaced by a much cheaper commercial system, but right now the expensive engineering and prototyping work is happening.

      This is most definitely not “available on cell phones”.

  • turtles_allthewaydown December 8, 2012 on 8:42 pm

    To Jason Dorrier: How does “Google Maps” have this covered in any way?

    This is not an easy task, and I think people are really underestimating the challenges involved. I don’t see how Google Maps has any part of this real-time accident avoidance covered. Anything worth doing is going to take some up-front money, and $25 million really isn’t all that much for the engineering, prototyping and wide-scale deployment and analysis. Yes, a private company could probably do the tech part much cheaper, but they would have all kinds of legal hoops and liability issues that they would have to incur to test this on public streets.

    How much money has Google spent on their auto-driving cars? Why don’t we have better real-time traffic reporting on our GPS units yet? That’s a pretty simple feature, but is only available on major roads, and not with high resolution.