Michigan Builds Driverless Town For Testing Autonomous Cars 86
HughPickens.com writes Highway driving, which is less complex than city driving, has proved easy enough for self-driving cars, but busy downtown streets—where cars and pedestrians jockey for space and behave in confusing and surprising ways—are more problematic. Now Will Knight reports that Michigan's Department of Transportation and 13 companies involved with developing automated driving technology are constructing a 30-acre, $6.5 million driverless town near Ann Arbor to test self-driving cars in an urban environment. Complex intersections, confusing lane markings, and busy construction crews will be used to gauge the aptitude of the latest automotive sensors and driving algorithms and mechanical pedestrians will even leap into the road from between parked cars so researchers can see if they trip up onboard safety systems. "I think it's a great idea," says John Leonard, a professor at MIT who led the development of a self-driving vehicle for a challenge run by DARPA in 2007. "It is important for us to try to collect statistically meaningful data about the performance of self-driving cars. Repeated operations—even in a small-scale environment—can yield valuable data sets for testing and evaluating new algorithms." The testing facility is part of broader work by the University of Michigan's Mobility Transformation Facility that will include putting up to 20,000 vehicles on southeastern Michigan roads. By 2021, Ann Arbor could become the first American city with a shared fleet of networked, driverless vehicles. "Ann Arbor will be seen as the leader in 21st century mobility," says Peter Sweatman, director of the U-M Transportation Research Institute. "We want to demonstrate fully driverless vehicles operating within the whole infrastructure of the city within an eight-year timeline and to show that these can be safe, effective and commercially successful."
A little late there, American Car Industry (Score:5, Informative)
Darpa did this a decade ago, google did it 5 years ago, and so did several foreign manufacturers.
I mean, it's a good thing, but it's been more than 30 years since the American Vehicle manufacturing industry was actually on time for a new idea.
Re: (Score:3)
Google is all like, "Look at our shit navigating busy downtown San Francisco streets, dealing with insanity at intersections, crazy drivers, pedestrians, traffic signals, poorly-marked lanes..." They provided a video presentation on the whole thing.
Government is like, "Those fancy new electric cars can self-drive on the highway at 3AM when there's nobody on the road and the lanes are painted in radar-reflective bright white, but they have a lot of trouble navigating when there's a stop sign or another dr
Re: (Score:2)
You talk about a video to represent the Google portion of your post, but the part about the government is left in nebulous ambiguity. From where do you derive your sentiment?
Re: (Score:2)
Perhaps the GP is inferring based on regulatory status of autonomous cars, or from the general pace of acceptance of new technology by governmental bureaucracy.
Re: (Score:2)
But all-in-all regulatory acceptance has been pretty free of obstacles. They did their first dry runs out in Arizona not too long ago, and had a few things that didn't work yet.
Re:A little late there, American Car Industry. (Score:3)
Exactly. Ann Arbor has persistent winter snow and occasional sleet, heavy rain, tornados, and even flooding. Its weather is often a perfect storm for drivers and a far cry from the ideal idyllic settings used so far to test automatic cars.
A2 is the real world. And its mix of academia and auto company proximity make it ideal for this role. Seems like a perfect marriage.
Re: (Score:2)
I work about a mile from this. It's on University of Michigan property within the City of Ann Arbor - which might make it within the City of Ann Arbor, except that no property taxes are paid on it and the University of Michigan police have jurisdiction rather than the City of Ann Arbor police. (The University of Michigan is chartered by the state of Michigan such that it operates much like an independent civic body.) So maybe it's in Ann Arbor or maybe it's not, but that's probably sharing too much about
Re: (Score:2)
Right. With luck this kind of exurban facility will make good use of selective dispensations from the MI DMV to extend their trials off premises and onto roads like you describe.
Dialing up the real world noise is essential to bring these cars up to speed -- missing or obstructed lane markers and signs, poorly marked or uneven road edges, and the introduction of noise like leaves, snow banks, and pools of accumulated rainwater all need to be mastered before automation has any business driving cars, buses, tr
Re: (Score:2)
A good test for these autonomous vehicles will be "drive within a 2 block radius until you find a parking spot."
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Highway driving, which is less complex than city driving, has proved easy enough for self-driving cars, but busy downtown streets—where cars and pedestrians jockey for space and behave in confusing and surprising ways—are more problematic.
This is the justification given for building an urban obstacle course to test self-driving cars: they have trouble driving in the city.
I'm pretty sure Google has been near-exclusively testing in busy cities, not cruising on the highways. The car might go on the highway to get to another city or to navigate from area to area, but it spends a fuckload of time cruising city streets. Highway driving data is meaningless to Google, as it's trivially small and provides little opportunity to improve the car;
Re: (Score:1)
This isn't the car industry (and I wouldn't say just the American car industry, as BMW or Toyota are not doing much either.) Those people have been resting on their laurels since they pushed streetcars off the roads 100 years ago.
It is ironic that tech companies have advanced automobiles further in the past five years, either by autonomous (self-driving) technology or having an extremely usable electric vehicle with a great range... when the rest of the auto industry has done relatively little except desig
Re: (Score:2)
As Xerox PARC showed us, having great ideas doesn't get you anywhere. As Apple showed us, implementing those ideas in a cost-effective way that consumers want is how you make money.
Abandoned America (Score:5, Interesting)
Aren't there any towns hit hard enough to be willing to sell themselves out for this purpose such that building a fake town isn't necessary?
Re: (Score:2)
Probably, but then they'd need to spend twice that much putting in the obstacles and sensors into an existing set of structures. At least, that's what seems to happen in many re-use cases.
Re: (Score:2)
It's almost certainly proximity to all the automotive engineers that made them pick this location. It's only a moderate commute from the suburbs of Detroitus to where this test site is.
Re: (Score:2)
It's almost certainly proximity to all the automotive engineers that made them pick this location. It's only a moderate commute from the suburbs of Detroitus to where this test site is.
I suspect that they wouldn't have to go far [zfein.com].
Re: (Score:2)
If that were true, wouldn't they be picking Detroit instead? I mean, it is in bankruptcy. You should be able to pick it up for pennies on the dollar.
Re: (Score:2)
Pennies on the dollar, sure, but you're underestimating how many dollars went into Detroit in its prime.
Re: (Score:1)
A controlled environment, designed from scratch, probably has more scientific value and is always available, whereas performing experiments in a "live" situation creates costs and also potential risks to other traffic participants. Besides, $6.5 million doesn't sound all that expensive.
Re:Abandoned America (Score:5, Informative)
Aren't there any towns hit hard enough to be willing to sell themselves out for this purpose such that building a fake town isn't necessary?
Detroit or Flint. For $6M they could have (literally) bought a few square miles of uninhabited Detroit for this use. It already has streets, signs, empty houses, etc.
Re: (Score:2)
Detroit or Flint. For $6M they could have (literally) bought a few square miles of uninhabited Detroit for this use. It already has streets, signs, empty houses, etc.
and you might find your fancy autonomous car has no wheels, or battery, etc. after a few weeks. I used to work there.
True story: One of my co-workers had a flat tire on the expressway. While he was taking the flat off, another car pulls up, the driver gets out, and pops the hood of the car exclaiming "You can have the tires, Brother. I'll go for the battery." At least the (potential) battery thief left nicely when informed that the car wasn't abandoned.
There was a reason that the plant at which I worked
Re: (Score:2)
Yes.
Trouble is, would you leave millions of dollars of equpiment virtually unattended in a city that's so poorly managed it had to sell a chunk of itself to survive? Probably wouldn't be a week before everything of value was either stolen or destroyed by vandalism.
Re: (Score:2)
Aren't there any towns hit hard enough to be willing to sell themselves out for this purpose such that building a fake town isn't necessary?
There's a number of abandoned military bases that are used on a routine basis. Both Mythbusters and Top Gear have used them.
The 'problem' is that such areas are normally laid out in logical ways. It specifically mentions that this fake town will have nasty traffic patterns and intersections.
Pick some of the worst designed intersections in the USA, they'll be in the town.
Busy construction crews? (Score:2)
and busy construction crews
Just where will that be experienced in the real world? All I see is a mile or more of blocked lanes for two guys sitting on the tailgate of a pickup...
Or are they testing how the computers can deal with something that just doesn't compute?
Re: (Score:2)
Of course you do. When you're on the road, it's usually close to breakfast/lunch/diner.
When I'm on the road, it usually is nowhere near any of these things. Yet I usually see about four idle "construction" workers for every one guy actually doing something. Mostly they are jaw-jacking. Possibly this isn't actually the workers' fault, perhaps the system is set up so inefficiently that they actually have nothing to do, but I don't care whose fault it is.
Re: (Score:2)
Depends on the task and area. Having watched some construction efforts over time, I don't see many people idle 'long term' in my local area. The latest had a guy standing around much of the time looking idle - but every 4 minutes or so he was busy for 30 seconds directing the latest truck where to dump it's fill dirt. You also had a couple flaggers working to ensure that the truck was able to get back on the road to head back for more dirt.
After that, well, you're going to have a safety monitor who's job
Re: (Score:2)
Weather and Potholes (Score:5, Funny)
If there is one thing Michigan is good for it's testing vehicles in horrible winter conditions with a crumbling road infrastructure. You get the weather for free and if they can just hire MDOT crews to build the roads you should have no problem testing out some of the most difficult conditions a driverless car will face.
Re: (Score:1)
Will the snow plows be automated also?
Why build a town (Score:2)
...they could just have used Detroit
Re: (Score:3)
The real test of driverless cars (Score:2)
The article is pretty short on details, but implies that the only cars in the test bed will be driverless. It strikes me that a better test would be a mix of driven and driverless cars, since that scenario is both more complex and more realistic. Algorithms developed and perfected in a 'simulated' real world stand a very good chance of falling apart in the 'real' real world - after all, public roads aren't going to be *totally* driverless for a long, long time.
Re:The real test of driverless cars (Score:5, Insightful)
Yeah, throw in the random human factor ... cut off the car, turn right from the left lane, ignore speed limits, tailgate, drift from one lane to the other while texting, make unsafe rolling stops through an intersection where even though the car has a green light it's going to have to jam on its brakes, children randomly running into the street, and cyclists who alternate between acting like they're entitled to drive on the road and driving anywhere else that suits them, pedestrians who come out from between cars and don't look.
Hell, put it behind a dump truck spraying gravel. Find a city bus which is going to jam into your lane whether you're in it or not. Ambulances at intersections. A construction detour which technically has you not following any identifiable lanes. Have people run red lights and blow through stop signs.
You know, the kind of stuff we all see every day. Try like hell to find out what its corner cases are. I'm sure they're there.
Teaching it the rules of the road only goes so far. Because many drivers and pedestrians seem oblivious to those.
Re: (Score:2)
You forgot school busses with the "No passing when red light flashes" signs on them.
Kids LOOOVE to just run around the front of the bus and into oncoming traffic.
Re: (Score:2)
Yeah, throw in the random human factor
The cars in the fake city may be driverless, but the simulation is still being controlled by humans.
I can tell you from first hand experience, when the sim director is in a bad mood, the simulation gets very, very "interesting".
Re: (Score:2)
Good. The job of the sim director is to ALWAYS be in a bad mood for something like this.
It's like QA, your job is to break it, in every way you can think of. If the people who built it think that's harsh or unfair, too damned bad.
I've never seen one of these cars in action, but my assumption is there are probably cases in which it will fail quite spectacularly.
And in this case, the "oh, the running over the old lady bug, yeah, that's fixed in car 1.3, you should upgrade" is not gonna cut it.
Re: (Score:2)
I think that part of the "selling point" of such a community is that the fleet of driverless cars would obviate the need for public transit, except for underground subways. The fare you pay represents only a portion of the actual costs to municipal, state/provincial, and federal subsidies. The biggest cost for a public transit system is salaries, benefits, and pensions, not capital costs.
So you could have a designated area where cars are not allowed, same as you have areas designated as pedestrians only,
Re: (Score:2)
Retirement communities would be big on this, since part of the problem is getting gramps to admit that he can't drive any more.
Intentional communities can simply have a tramway that runs along a loop route, possibly bidirectionally. They can be designed around the transportation, rather than the other way around, with positive ramifications all around.
Re: (Score:2)
Re: (Score:2)
Here's the thing - People with limited mobility need something a bit more personal than a tramway on a fixed schedule that doesn't give them time to get on and off.
So use a PRT-style tramway on a flexible schedule which waits for people to get on or off.
Re: (Score:2)
The article is pretty short on details, but implies that the only cars in the test bed will be driverless. It strikes me that a better test would be a mix of driven and driverless cars, since that scenario is both more complex and more realistic.
Realistic, but very hard to learn from since it'll all be one-time events. I suspect what they want is repeatable "randomness" so more like scripted cars where you can tweak the algorithms, reset the scenario and try again. Not to mention that if you're intentionally trying to break the computer's algorithms, having actual people in the mix seems like a bad idea. Having people drive cars around like they do with drones could be a good compromise though.
Winter (Score:5, Insightful)
Re: (Score:2)
Sorry, Detroit is not Winter "in all it's harshness", unless you're specifically referring to all of the salt they put on the roads to dissolve your car over time. I lived in the Detroit area for a few years and would laugh at all of the complaining about the cold weather. Minneapolis/St. Paul got far colder in my mind. And this is just from south of the Canadian border. I'm sure they get much colder Up North.
Re: (Score:2)
Driverless Town (Score:1)
"Michigan Builds Driverless Town For Testing Autonomous Cars"
It's called Detroit.
BOOM! Thank you, I'll be here all week!!!
Re: (Score:2)
Why the expense? (Score:2)
Why are they building this and paying millions for it?
Couldn't they just drive around most parts of Detroit? That'd be the perfect test, driving in a ghost town dodging random roving street gangs.
Great but lower priority (Score:2)
This whole field is interesting to me and so I'm glad to see research being done. That said, for me the place where I most want driverless cars is on the open highway. The problem with driving on the open highway is that it's just complex enough to require some of your attention, but not interesting enough to keep you mentally engaged, so it's really easy to get bored or sleepy or just plain distracted. In other words, it's a perfect candidate for automation.
In an urban environment I have no problem staying
AWESOME!!! (Score:2)
Re: (Score:2)
Besides, if there is a 'public' asset like this, then I am sure that MDOT and the other 13 companies can probably rent it out fo
They should just use detroit... (Score:2)
That way the cars can learn to dodge being shot at and handle having the wheels stolen when they stop at a stop light.
Just drive in New Jersey (Score:1)
Confusing enough there making a right to make a left turn lol
Re: (Score:2)
Highway driving easy enough for driverless cars? (Score:2)
Then they should release this right away. A lot of people would jump at the chance to be free to do other things during highway driving. I'd love to watch a movie during the 3 hour drive to the ski resort.
And think of an automated RV that drives itself on highways! These things spend like 90% of their time on the road on highways anyways.
Guys, I have a great joke (Score:1)
Why not use Detroit?
Funny huh?
I'm thinking (Score:2)
Will there be roundabouts? (Score:2)
I was in Carmel Indiana, a northern Indianapolis suburb, last week. Since the 1990's they have been replacing all of the main intersections with roundabouts. They have over 60 of them now.
While roundabouts have been proven to be safer for average drivers, how easy are they for autonomous vehicles to navigate vs your standard intersection? Is a roundabout an asset to the adoption of autonomous vehicles, a hinderance or a wash?
Re: (Score:2)
Re: (Score:2)
Detroit is on it's own. Seriously, why would Michigan (which is nearly broke) drive itself to bankruptcy to attempt to bail out a city that is beyond saving.
Michigan could money on this by charging the auto companies. That is if they weren't a government, with the built in disregard for costs that entails (because they are not 'costs' to the politicians, they are opportunities to grease palms with other peoples money).
um, it's Michigan (Score:2)