ITS vehicle to vehicle demo at ITS World Congress
I'm just back from the "ITS World Congress" an annual meeting of people working on "Intelligent Transportation Systems" which means all sorts of applications of computers and networking to transportation, particularly cars. A whole bunch of stuff gets covered there, including traffic monitoring and management, toll collection, transit operations etc. but what's of interest to robocar enthusiasts is what goes into cars and streets. People started networking cars with systems like OnStar, now known in the generic sense as "telematics" but things have grown since then.
The big effort involves putting digital radios into cars. The radio system, known by names like 802.11p, WAVE and DSRC involves an 802.11 derived protocol in a new dedicated band at 5.9ghz. The goal is a protocol suitable for safety applications, with super-fast connections and reliable data. Once the radios in the car, the car will be able to use it to talk to other cars (known as V2V) or to infrastructure facilities such as traffic lights (known as V2I.) The initial planned figured that the V2I services would give you internet in your car, but the reality is that 4G cellular networks have taken over that part of the value chain.
Coming up with value for V2V is a tricky proposition. Since you can only talk to cars very close to you, it's not a reliable way to talk with any particular car. Relaying through the wide area network is best for that unless you need lots of bandwidth or really low latency. There's not much that needs lots of bandwidth, but safety applications do demand both low latency and a robust system that doesn't depend on infrastructure.
The current approach to safety applications is to have equipped cars transmit status information. Formerly called a "here I am" this is a broadcast of location, direction, speed and signals like brake lights, turn signals etc. If somebody else's car is transmitting that, your car can detect their presence, even if you can't see them. This lets your car detect and warn about things like:
- The car 2 or 3 in front of you, hidden by the truck in front of you, that has hit the brakes or stalled
- People in your blind spot, or who are coming up on you really fast when your're about to change lanes
- Hidden cars coming up when you want to turn left, or want to pass on a rural highway
- Cars about to run red lights or blow stop signs at an intersection you're about to go through
- Privacy is a big issue. The boxes change their ID every minute so you can't track a car over a long distance unless you can follow it over every segment, but is that enough? They say a law is needed so the police don't use the speed broadcast to ticket you, but will it stay that way?
It turns out that intersection collisions are a large fraction of crashes, so there's a big win there, if you can do it. The problem is one of critical mass. Installed in just a few cars, such a system is extremely unlikely to provide aid. For things like blindspot detection, existing systems that use cameras or radars are far better because they see all cars, not just those with radios. Even with 10% penetration, there's only a 1% chance any given collision could be prevented with the system, though it's a 10% chance for the people who seek out the system. (Sadly, those who seek out fancy safety systems are probably less likely to be the ones blowing through red lights, and indeed another feature of the system -- getting data from traffic lights -- already can do a lot to stop an equipped car from going through a red light by mistake.) On its own, even the enthusiasts now agree that decent market penetration is not too likely. The only course to quick adoption is a regulation requiring the systems. And that's in play in the USA. The DoT is conducting a trial in Michigan by equipping 3,000 cars of workers at a hospital with the transmitters. This will create a local zone with decent penetration. If they like the results, in 2013 NHTSA will consider making a regulation requiring the radios. At the speed at which car vendors move, that means it's the 2019 model year which is the first to have the radios standard. Then, over several years, penetration goes up, so that in the mid 2020s you have a decent chance that a car will have the radio.
Robocars (and human drivers) could never depend on such radios as their only way to detect a car. Even with near-100% penetration they will sometimes break or fail. Indeed, as penetration gets high there is a real risk that human drivers would come to depend on the system, figuring they don't need to check their blindspot any more, causing more reckless driving and creating some collisions. Robocars at least won't do that. On the other hand, for the robocars, extra information on the road is generally valuable, particularly information about cars which can't be seen or detected directly by sensors.
Traffic light transmission is also useful to all. Cars can know to not race up to a light that's going to be red, and rather coast and always get a green. (Though this could also be done over the wide area network with synchronized clocks and lights on predictable timers.) It is also planned to let emergency vehicles get signal priority using these radios -- though that interferes with planned timing. Another minor application allows cars to tell signals they are coming, so that the signal does not have to wait for you to go over the magnetic loop to be detected. Late at night that might let you breeze through lights without being an ambulance.
At the ITS congress, there were several demos of this technology. One, done at the Disney speedway by a coalition of most major car vendors, showed the basic functions above in actual operation, with cars from many makers.
In this video I shot, we're in one car following another. We can't see it, but a 3rd car has stopped ahead on the road. The car we're following swerves around it at the last minute, but we get a warning about its presence before that swerve, before we can even see the stalled car.
In some other videos such as this one of a DENSO test you can see an attempt at doing a map with a green band to show where to drive to get a green. Also of interest in this video is the two DSRC equipped cars driving around our bus, shown on the screen. One thing that may concern you is the way they dance around -- GPS is not very accurate, and in fact not even accurate enough to be entirely sure what lane another vehicle is. They've had to work to avoid that causing false alarms, and they still happen. One way they could fix that would be to have the radios share GPS solution data, so you can be sure both GPSs used the same algorithms and satellites for their solution. In that case the relative error of GPS is much smaller, and you can tell if another car is in your lane or the next one. They are not doing that at present.
Here are some of the other problems they face:
- Security is a big issue. They have a plan to digitally sign transmissions, and a PKI for these signatures. But the internet hasn't got its own PKI for TLS working perfectly after 20 years and these guys won't do much better.
- Even if you can certify the messages, you have to seal and certify the whole system. What if bored teenagers send out signed messages from a fooled GPS that says a car is stalled in the middle of the highway, and everybody gets a collision warning, or worse, their cars auto-brake?
- Even without malice, I am not sure how well the system handles a car parked on an overpass, transmitting the truth about where it is. (Altitude accuracy in GPS is much lower.) And of course there is no GPS in tunnels and many urban canyons.
- Many of the current radios just don't link-up fast enough or over long enough distances to really be good for safety applications.
All in all it's a tough problem. It would be nice to see this happen but as yet it's hard to see how that comes to be.
Where were the robocars?
One thing that surprised me about the ITS congress was the general absence of robocars and discussion of them. Oh, the keynote from GM predicted them and GM had their autonomous EN-V doing demos in the parking lot. And a few papers out of a thousand talked about autonomous vehicles and related technologies. Most were unaware of the progress in robocar technologies and sensors or placed the technology many decades out. That's an issue because many of the safety systems aimed at assisting drivers are much less important if fewer people are driving, particularly the ones who are early adopters of fancy safety systems.
Today's safety systems are another matter. Known as ADAS (Advanced Driver Assist Systems) many of these technologies are precursors for robocars. What began with radar based adaptive cruise control (ACC) now has blossomed, and things like blindspot warnings, lane departure warnings, collision detection and pedestrian detection can be found in high-end cars from most vendors, though overall penetration is still quite low. It is extensions of these technologies -- combining lane-keeping with ACC that is behind the announcements from several car vendors of their plans to have limited self-driving in highway or stop-and-go situations just a few years from now.
AISIN, a Japanese supplier of advanced in-car technology, described a system I found quite interesting. They have worked hard to make systems for getting more accurate locations for cars. GPS is just not accurate enough or reliable enough for many applications. AISIN's system starts with GPS but uses a map of the road and turning sensors to place the car in specific locations when the car turns. It then counts the stripes on the road combined with the odometer to figure out just how far the car has gone since the last identified turn to get a good handle on where the car is to within a meter or two along the road. Then it does something I discussed a while ago -- shares information about problems with the road. If one car hits a pothole, it uploads the location of the pothole to the cloud. Other cars download it, and when they are approaching the pothole, the car softens the suspension. It also detects positions in on-ramps and off-ramps and tells the automatic transmission to shift differently based on that knowledge.
More benefit from low adoption
These tools might do better if they see what they can do with low adoption. For example, many cars, not just robocars, are gaining sensors to detect other cars and pedestrians for various purposes. If those cars have DSRC radios and transmit their own view of the world, suddenly you don't need to have nearly as much penetration to get safety benefits -- if you can trust the other reports. One car ahead of you might sense several cars around it and tell you about all of them, and suddenly it's like a lot of cars are broadcasting.
This is also true for traffic signals. Each traffic signal should broadcast not just its own timings but those of the signals around it. For close signals, that means you don't need to wire them all up. For more distant signals, each one still needs a transponder in theory, but you could learn about upcoming signals much further away than you could with a direct connection.
Another big theme in the ITS world is traffic management. And they are starting to see the role of automated driving here, even if it's only ACC. To my surprise, 60% of congestion on Japanese highways is caused by "sags." Sags are stretches of highway through a gentle depression, so that the cars go down for a bit and then up for a bit. In gentle sags, the drivers aren't really paying that much attention but they slow down as the up-grade section starts, and this slowdown causes a shockwave back through the traffic and a tie-up. The use of ACC avoids this, because the ACC maintains speed on the upgrade. (On steeper sags, the drivers notice and accelerate when they hit the climb.) This research points to something I have intuitively believed for some time: That robocars will cause major improvements in congestion because they avoid a variety of human driving behaviours. They will not just maintain speed, they won't pause to look at accidents or distractions on the road, and they will perform more orderly merges. And of course they will cause fewer accidents. These together are the major causes of congestion. The other big cause is simply having more demand than there is available roadspace, and I have detailed other solutions for how robocars can solve that.
Congestion currently costs $100B per year and burns 1.9 billion gallons of gasoline. At the conference, it was stated that in a few years this will climb to $130B and 2.5 billion gallons. They hope to solve that with ITS, but robocars will do an even better job.