You are here

Google not alone with robocar advances


This weekend's announcement that Google had logged 140,000 miles of driving in traffic with their prototype robocars got lots of press, but it's not the only news of teams making progress. A team at TU Braunschweig in Germany has their own model which has been driving on ordinary city streets with human oversight. You can watch a video of the car in action though there is a lot of B-roll in that video, so seek ahead to 1:50 and particularly 3:20 for the inside view of the supervisor's hands hovering just over the self-turning steering wheel. There is some information on Stadtpilot here, but we can see many similarities, including the use of the Velodyne 64 line LIDAR on the roof and a typical array of sensors, and more use of detailed maps.

The team at Vislab in Milan has completed most of their Milan to Shanghai autonomous car journey which I have been following. You can read their blog or watch video (sometimes live) of their trip. A lot of the blog has ended up being not about the autonomous challenges, but just the challenges of taking a fleet of very strange looking vehicles in a convoy across Eastern Europe and Asia. For example, they have trucks which can carry their robocars inside, and once decided it was simpler to cross a border into Hungary this way. However, they left driving the vehicles, and the exit officials got very concerned that there was no record of the robocars coming into the country. I presume it wasn't hard to convince them they were not smuggling Hungarian robocars out.

The Vislab challenge is impressive but aimed at very different goals. They have a focus on making more near term "driver assist" technology. This makes sense and is a definite stop on the roadmap to Robocars. However, it means their trek is not as autonomous as you might imagine. The lead vehicle in each pair is usually human driven, in particular because there are not accurate maps of the route they are taking. The second vehicle is usually autonomous, and either follows visually or by receiving sets of GPS waypoints from the lead.

They plan to reach Shanghai for the close of the world's faire there.

Alberto Broggi on the Vislab team tells me that one of the big challenges in some areas -- in particular Moscow -- has been really erratic driving by locals. (That's a lot, coming from Italians!) I have certainly seen how in some places in Eastern Europe and Asia the lines on the road (including the center one) are "just a suggestion." Their software is not up to dealing with this and they've gone to wholly manual there.

The Vislab team has smaller 4-line LIDARs and has not used the larger Velodyne everybody else uses, though they agree it's the best sensor available. They want to see what they can develop for driver assist that is at a lower price point, particularly what they can do with machine vision. Vision sensors are indeed cheap, and in fact they are becoming stock equipment on many cars for other purposes. "...given someone else is investing on this lidar, research-wise I feel some value in investing on another technology," says Professor Broggi.

My prediction is that with Moore's law and economies of scale, all sensors will get cheap over time, and that all robocars will want to have as many sensors as they can afford. That means that all efforts to gain more out of each type of sensor are worthwhile.

More Google Notes

A small part of the press on Google's robocars was overshadowed by the question of why a company like Google is doing this. I think the answer there lies in the extensive use of mapping and local data in improving the performance of their cars. Google has given its cars a really, really good sense of where they are because they only drive streets that were scanned before in 3-D by similar vehicles. With a complete 3-D map of each street, knowing the location of every curb and tree and building by shape, the cars are able to very accurately figure out just where they are without anything more than a loose position from GPS and the odometer. The GPS may only report position within a few meters, and may in fact not even be working in some locations, but the car knows where it is because that building is here, and that other building is over there, and the lane markers on this particular street are at X, Y and Z. In effect, the google car is able to navigate the street because it has been there before and remembered every little detail of it -- taking out the moving things. While I don't know how good the system is, such a technique should be fairly robust -- even if a building is knocked down, there are lots of other buildings. In rural areas with a certain sameness, I imagine the dashed lane markings, phone poles, ditches, road edge shapes and other clues will never look the same over a short stretch. The GPS, odometer and intertial systems will always tell you roughly where you are, but the street memory will tell you exactly.

And this is something Google is one of the world experts on, particularly the StreetView team. There are two broad problem spaces in autonomous driving. One is the tactics -- figuring out what to do in real time based on changing conditions, avoiding obstacles and so on. The other is the navigation (what might be called the strategy) of knowing where you are and picking lanes and turns. They are not disjoint, of course, but Google has become heavily involved in navigation tools and their breakthrough involves not just improving them, but making them so good that the sense of where you are helps the tactical driving system figure out what it's going to do.

I've been advocating the development of what I call a trillion mile test suite. Google has also demonstrated the value of precise mapping data, which is easy to get -- just drive down a street a few times. With a proper test suite, your future robocar will have already driven -- virtually -- down every street in your region before it ever physically takes you down them.


I can understand how they might have had trouble with traffic in various places. I remember how awfully scary traffic in Moscow was. But the place that truly terrified me was Istanbul. (I didn't drive in either place. I was driven around by local friends.) My first car ride in Istanbul, I remember a place where the freeway was marked out in 4 lanes. Traffic was driving 7 abreast. At freeway speeds. With random lane changes. And jaywalkers! I closed my eyes and hid for the rest of the drive.

One thing that amused me while travelling was that the same joke kept coming up everywhere. I heard it in Russia, Ireland, Estonia, The Netherlands, Turkey. It was always claimed as a local joke based on local driving conditions. Sometimes they had trouble translating it from their language into English for me: My friend was driving me and we came to a traffic light. Just as we arrived, it turned red. He sped through. "The light was red. Why didn't you stop?" "Don't worry. My brother does this all the time!" We came to another traffic light. The same thing happened. "Don't worry! My brother does this all the time!" We came to a third light. This one turned green. He slammed on the brakes and screeched to a stop. "What are you doing? The light is green. Go!" He turned to me and said, "Are you crazy? My brother might be coming!"

Could You please comment more on google car?
Several friends of mine who work in this area claimed that there is nothing new in all those thousands miles hyped by google PR team
Is google robocar a big achievement vs robocars of other research institutions?
What is new and remarkable achieved by google team and not achieved by other teams
I am interested in new technologies rather then financial comparisons
I do not have any expertise in this area

Google demonstrated serious real-world performance, in traffic, with other cars, pedestrians, cyclists -- the works. A lot of miles of it, and a lot of continuous miles without intervention. This is a large step above the Urban grand challenge which had only very limited traffic and a fairly controlled environment.

It is not complete, still far to go, but a major step.

I think another important factor for robocars will be the High Accuracy-Nationwide Differential Global Positioning System that is being tested in Maryland and Pennsylvania ( . The target for that project is an accurate position down to 10cm. I personally figure that a system that is accurate to 30 cm would be as good or better than the average driver. I system like that would let a car accurately follow a given lane using nothing but a accurate survey. A robocar that could manage freeways wouldn't be much more than this and current radar systems, already available on some cars, to maintain current a proper spacing from the cars in front of them.

The problem is how to get the driver's attention quickly enough when some judgment is required to deal with a situation. Frankly that is quite a big problem now with manually steered cars. Having the robocars networked so they can relay information about problems back down the road would help a lot. At least we could avoid multi-car pile ups.

Actually, networking of the cars comes last. The first robocar on the road has nobody to talk with. Car-to-car networking is only good for things that are OK if they are intermittent.

In theory, robocars should have a very low risk of multi-car pile-ups. They will have a very quick reaction time. Humans take at least 0.75 seconds to react and we follow too closely for the conditions. The robot should constantly be aware of traction due to wheel sensors. Radar will mean they see cars far away even in fog, including cars stopped in a pile-up.

But yes, it would still be handy to know if there is ice ahead, for example.

Google's new techniques eliminate the need for HA-NDGPS, though it will still always be handy. But there will always be places where there is no GPS signal and you have to be able to go there.

The point about a critical mass of networked cars is a good one, but what about a networked traffic light that not only broadcasts it current condition, but tells you how long before it turns read or green so the car can adjust it's speed? Speed limit signs that not only broadcast the current speed limit, but road conditions ahead. A school zone sign that knows when school is letting out.

I disagree about HA-NDGPS. It appears to me that SLAM depends on GPS to narrow it's search, although it can probably extrapolate though small dead zones. If I understand correctly, the GPS signal is actually the weakest part of the system. I personally would be pretty happy with a robocar that can drive autonomously 90% of the time, as long as I get adequate warning before it requires manual control. Such a system will be a precursor for future driverless vehicles.

On a related topic, Google had another interesting press release:

Unless someone is spinning like crazy, this is being treated like they are putting one of their senior executives into what they think is an important future growth area. There has already been a lot written about localized search, but what about an advertising engine that knows where you are going and your route? If you are driving home, it can suggest a take out place on the way and email them your order. If you want to cook it will suggest a menu and a grocery store on the way and download a shopping list to your phone sorted by aisle.

I think local stuff is quite important to Google.

Traffic lights and other sources will be broadcasting soon. There is even a protocol (802.11p) defined but not widely yet used. I expect a lot of data broadcasts from cities (and streets) in various ways, though I suspect a lot of it will just be the better build out of faster and more reliable data networks, which then can be used to fetch all sorts of data.

Right now you have to assume GPS can be off for long stretches, and also inaccurate. More accurate GPS systems will be great but it's a while before cars could depend on them to do things like lane positioning, so the SLAM/LIDAR appraoch is needed and Google seems to show it's working well. More data when you can get it is not likely to hurt. I don't know if anybody has done it, but I would bet if you were to distill things like the intervals of dashed lines on the road and the presence of driveway cut-curbs and a few other things, you could probably fingerprint the whole world to the point that plopped down on an urban street you could figure out where you were without GPS. (There are lots of other things besides GPS to locate yourself too, such as the presence of TV channels, wifi MACs and everything in between.)

Google's project seems to show we can move "where precisely are you?" to the mostly solved category. Still lots to work on, though.

Does one refer to the name of the town in the language which is spoken there,
or in one's own language? Generally, for big cities the former, for smaller
towns the latter.

So, not a judgement, just an observation: English for Braunschweig is Brunswick,
as in New Brunswick.

Add new comment

Subscribe to Comments for "Google not alone with robocar advances"