Footage from test runs of Google driverless cars

Google has been testing out its self-driving cars on real roads. This is still a long way from being available for you to purchase, but it's clear that it's working surprisingly well on a technological level.

You can watch some footage, recorded in the driverless cars during their test runs, in the video above. IEEE Spectrum's Erico Guizzo (who, incidentally, says he's a lot less skeptical of Google's goals after seeing this video) explains what makes the system work.

Two things seem particularly interesting about Google's approach. First, it relies on very detailed maps of the roads and terrain, something that Urmson said is essential to determine accurately where the car is. Using GPS-based techniques alone, he said, the location could be off by several meters.

The second thing is that, before sending the self-driving car on a road test, Google engineers drive along the route one or more times to gather data about the environment. When it's the autonomous vehicle's turn to drive itself, it compares the data it is acquiring to the previously recorded data, an approach that is useful to differentiate pedestrians from stationary objects like poles and mailboxes.

The video above shows the results. At one point you can see the car stopping at an intersection. After the light turns green, the car starts a left turn, but there are pedestrians crossing. No problem: It yields to the pedestrians, and even to a guy who decides to cross at the last minute.

Video Link

Via Bryan Walsh


  1. The REAL real world test will be for them to have the cars setup to accept an audio interface, which will be provided by a running Google maps on an appropriate phone, and having the car drive from point A to point B.

    You know with out making a left turn off a bridge or something…

  2. I’ve never understood the point of even looking into this technology.  How much longer will we even have cars before the gas runs out?  What’s the point of having highways full of driverless cars when you could string a railway along the same routes for 10% of the cost? What needs would be fulfilled if we had these?  I can’t think of any.  The only thing I can think of is alpha testing for general purpose robotic navigation as it’s a simplified version of the problems there.

    1. have you heard of renewable energy resources? alternative fuels? i think we’ll have cars of some sort for quite a while. they may be smaller in the future though

    2. Few things here. 
      1) Gas running out isn’t the issue. Gas becoming increasingly expensive is the issue. 
      2) Beyond that, liquid fuels are going to be with us for a long time. 
      3) Driverless cars do have some potential to increase efficiency of fuel use, which is going to be increasingly important. 
      4) A lot of people really like the freedom of choice you get from personal transportation. That’s what driverless cars do that trains don’t. You get to decide where you’re going and when. 
      5) Driverless cars are really expensive and it isn’t clear whether the benefit, environmentally, justifies the cost. So whether this makes sense depends a lot on whether the other benefits (personal convenience, traffic safety) can be proven and whether people like the experience.

    3. Why bother with this technology?  Are you kidding?  If you could have autonomous cars you would eliminate a massive swath of problems that face our civilization.  It would be earth shattering in the potential.

      I only use my car for a good 30 minutes to an  hour each day.  That means that for over 95% of the day it is just sitting there.  I don’t car pool because my schedule is too erratic and I need mildly flexible work hours.  I am pretty typical.

      Now, consider what I might do in the world of autonomous cars.  I wake up each morning and, 5 minutes before heading out, I slap a button on my phone declaring I need a pickup. I am a bit of a cheap bastard and really just want to get to work, so when “my car” pulls up, it is really a small bus.  It is has a dozen folks in it that are a semi-regular crowd of people who all work roughly in the same area as me and who often go to work at roughly the same time.  The crowd changes as people shift their schedules and people need to arrive to work early or late.  The bus takes an optimal path, not really going much out of its way to pick up a couple of other people on the way.  It then drops people off, again, taking an optimal path so that it never really goes out of its way.  The normal highway traffic is gone, as most cars are autonomous and coordinate their movements so that the four lane highway is now overkill.  Add on to that the fact that there are simply fewer cars because many people who commute decide to go cheap and just snag an autonomous bus on the way to work, and you will find the roads are basically empty.  The fare is cheap.  The bus, plans its own rout to maximize speed and passengers.  There are no ‘regular’ routs.  All routs are custom made on the fly based upon people requesting rides.

      When I leave work, I leave late.  Further, I’m grumpy and so not in the mood to share and want to get home quicker.  By default, I normally call a service on the cheaper side, but tonight I decide I am willing to pay a few extra bucks for privacy and order up a personal car.  The personal car comes directly to me and whisks me home.  It is still vastly cheaper than owning my own car, as the “personal” car promptly go finds someone else to ferry around once it is done with me, instead of sitting around rotting on the side of the street.

      Which brings us to how these cars would have changed cities.  There are no longer any cars on the side of the street.  Cities are no longer giant parking lots.  When night falls, most autonomous cars disperse out into the burbs to be packed into large parking lots that double park all cars in.  Some stay in the city, but they too double park themselves into garages instead of polluting the streets.  When it gets near work time, they will slip out and move to likely spots to snag routs.  Fewer roads are used for traffic, and far more have become ped ways.

      As a society, we consume vastly less fuel.  Almost everyone who isn’t in need of privacy “car pools”.  Hell, we might not even consume fuel at this point.  Waiting 8 hours for an electric car might be a pain if you own your own car, but if you use an autonomous car, when one car needs to charge, another rotates in to take its place.

      So lets go over what a world of autonomous cars looks like:
      1) Vastly cheaper transportation costs as you drop the 1 car for every person thing.
      2) Vastly less fuel is used as you can set up a dynamic car pools with the press of a button.
      3) Vastly less traffic as fewer people physically drive their cars AND there are fewer cars on the road to begin with.
      4) Cities suddenly open up and become even more friendly to non-motorized transportation when we stop using them as giant parking lots.

  3. Hmmm.  

    The demo looks good, but until Google Maps can correctly figure out the numbered street system of Utah — where if you know the address, you know how to get there — accurately,  I wouldn’t trust it with my life.

  4. I’m ready for one now! I especially don’t want to get cut off by a VW Robot while I’m driving a manually operated vehicle.

  5. I wonder if driverless cars could be used as some kind of taxi service.

    Anyways, as long as there is a need to drive ahead and record stationary objects this will not work in the long run. This because streets change as objects come and go along the road sides.

    1. I think you can probably use subsequent passes by the automated cars to gather more data, meaning as long as the street gets some use by the car network such problems could be avoided.

      1. Good point. It becomes similar to some traffic monitoring systems that have been tested out that used gps and mobile phone networks.

  6. All it needs is for one person, or heaven forbid a dog or kitty being run over and google is the devil.

    I like it though, being my age having grown up in the 80’s where you saw movies of cars driving themselves and now we actually have cars that drive themselves is damn cool. Not many people will go through this experience…

  7. Barking at the wrong tree Google: auto-pilots are already quite good so shouldn’t you concentrate you millions on the flying car ?

  8. Are these really ultimately intended for sale to consumers? Or is Google’s plan to eliminate its team of Streetview picture cars and drivers and replace them with a fleet of these autonomous drones, constantly prowling the world’s streets and paths, continuously updating, and adding to, Streetview?

  9. This is super cool and some day I hope all of the bands in vans can use this to get some shut eye…. 
    but do they need a special permit for the car to test this on actual city streets? If they kill or injure some one while testing the car who is prosecuted? 

  10. I would like to have an auto pilot function for long drives. In the city it would be more problematic but I can see it being easier to establish auto pilot lanes on a highway. Set it, put your seat back and enjoy the ride.

  11. I’ve been a safety operator for driverless vehicles — I push the emergency stop button before the thing hits something (person, bush, cliff, fence, etc). When they work, it is the most amazing thing. I can only imagine that this technology will get better at an alarming rate.
    You can blind these things with an infra-red spotlight.

  12. With all of the distracted drivers on the road, these things could potentially have a much better safety record than humans do.

  13. “When it’s the autonomous vehicle’s turn to drive itself, it compares the
    data it is acquiring to the previously recorded data, an approach that
    is useful to differentiate pedestrians from stationary objects like
    poles and mailboxes.”

    So based on that, I have to wonder how easily fooled these cars would be by the fake tunnels painted on rockfaces by Wile E. Coyote.

  14. I just wonder how long it will take until google acquires velodyne, given the fact that their demand for HDL-64 and HDL-32 scanners will increase substantial once they want to deploy several cars and also have backups available.

  15. Vehicle accidents are the leading cause of accidental death in the U.S., amounting to 44.3% of all deaths according to

    What you’re looking at is a future with a drastic reduction in vehicle fatalities, most of which are caused by people breaking the laws (speeding, running lights, inattentive driving). Automated vehicles will not break the laws, and will cooperate nicely with each other. The more of these we have on the streets, the safer we’ll be. If collisions do occur, they will most likely be minor, and the reasons for those collisions can be researched and fixed. Mechanical failure is always a possibility (as it is with current cars), but statistically this will be far safer than what we have now.

    We should be pouring tons of money into this research. I can imagine a day when nobody drives at all. Plus it will save gas, open up green space, and cops can find better things to do besides sitting around issuing traffic tickets.

  16. Sometimes I feel like I’m the only BoingBoing reader that actually enjoys cars and driving. I enjoy driving immensely, and cannot imagine not driving. I mean if you guys like being chauffeured around, that’s cool with me, its just not something I want to do.

Comments are closed.