Hacker News new | past | comments | ask | show | jobs | submit login
In-Car Algorithm Could Rapidly Dissolve Traffic Jams (technologyreview.com)
85 points by munin on Sept 15, 2011 | hide | past | web | favorite | 67 comments

My hunch is that ultimately, ceding control of the car to an automated system is going to produce better results, at least when driving on major "pipeline" roads. The biggest source of traffic jams isn't necessarily the pure density of cars, but the inefficiency of the individual driver. If each person takes that extra second to accelerate, that gums up the whole works (in NYC we are trained to be more efficient - taxis will honk at us if we take more than .05 seconds ;)

Imagine this: You enter a major arterial roadway -- I-95 or Route 80 and you program your destination -- an exit #, a rest area, etc. Your car joins the pipeline, syncs up with all the other cars on the road, accelerates to a (fast) cruising speed (100+ mph), and you sit back and enjoy the ride. No lane-changing, no "human" failures (like slowing down just because the sun changes direction) - and when it comes time to arrive at your waypoint, the system safely gets you out of the flow and into a "manual" lane. Then you take over from there.

Could be awesome...

Interestingly, this seems to converge towards traditional ways of mass transit, like trains. Except for two crucial differences with regards to the US: Insane connectivity, seeing how good the highway system is in the US, and of course the fact that once you get off the freeway, you still have a car to navigate through the vast suburbia (even if you have to do it manually). I wonder if this will be the replacement for good mass transit in the US, at least for now.

I think the final piece of this puzzle is going to have electrified highways so you can travel long distances without the need for large battery packs. It would still be more efficient to uses busses/trains but we already have a lot of roads and cars which make retrofitting seem like the best option.

Yes - and we could replace the rubber/tarmac interface with a steel/steel one for even better efficiency.

Highly automated steel-wheeled vehicle on steel tracks with continuous electricity supply along the way... I wonder how we'll call that.

So long as I don't have to walk a mile to get to the steel/steel in the 110+(F) sun, I don't care what you call it. ;)

When we got a talk from a renewable energy / car battery researcher at my university at the end of the talk I asked him "don't you think that given the costs and difficulties with scarce materials that it would be cheaper and more energy efficient to create power lines for cars on every road than to put a battery in every car?" -- and he agreed, except that this strategy doesn't have a good business model unless governments take major steps.

Seems like it would be a pretty good business model for utility companies.

One of the other benefits is that a car controlled by a computer should be able to draft off the car in front it. This should significantly improve gas mileage. We might even design the shape of cars specifically for this since aerodynamics don't matter as much at slow speeds.

In the era of rising energy costs, there's still the fact that you have a one-to-three ton steel cage just to transport a 100-300 pound person. There's no way that's a sustainable solution.

It's still pretty low-density though; using something like four square meters of space per person is ridiculous.

There was a system kind of like this in the movie Minority Report. It takes place in a future LA.


It's worth noting that this is not a new idea. In the late 60s there were at least a dozen research projects working on this.


Maybe technological advances have made it more feasible. I guess we'll see.

I agree. The only problem is that you still need to leave room in case something catastrophic happens in front of you (deer, tire flat, etc.) Of course, if the system detects these events as slowing down of one vehicle, it could quickly slow others down, so you'd need less distance than with a human driver. But it only works so far. There are laws of physics dictating the distance needed if one vehicle has a sudden deceleration, and without sufficient distance you'd end up with one spectacular pileup.

I'd hate to use such a system in the rain. The road is worn down (i.e. is lower) where the tires usually run. That means the pavement is drier if you drive on a "rain line" that is slightly offset from the usual line. This is often enough to keep the car from hydroplaning.

Years ago I've read an article that recommended doing this manually as a way to assist the flow of traffic.

I thought its a nice idea and tried it. Unfortunately, I tried this in Tel-Aviv, where drivers from other lanes immediately moved their cars into the space that opened between my car and the one ahead of me. It didn't take me long to figure out that I'm getting nowhere.

Two years later, when I moved to the bay area, I tried it on the 101 during rush hour traffic. To my shock, it worked. It was exceedingly rare to have any car move into that space. Or move lanes in general. For some reason (laziness? safety?), California drivers don't switch lanes as much as Israeli drivers do.

Moral: There's time and place for every algorithm.

Bay area and LA drivers are very different. LA driving is more similar to Tel Aviv driving, if you want a taste of home :)

Bay area drivers have this weird sense of "lane ownership". They don't change lanes... and they don't let anyone come in! It's not the sort of "Won't let you cut in front of me!" that you sometimes see in Israel, but a general thing. Even if you need to change lanes to exit and it's clear you're going to exit right away, I've had countless drivers speed up and block me from merging to the right.

This weird quirk does make it useful for applying traffic wave theory, so at least there's one good outcome.

I've used a variant of this successfully when there's a merge coming up. Basically, if I'm in the lane that's disappearing, I slow down as if to merge, matching speed with the car next to me, but then don't merge until the last moment. When I do this I invariably get some hotshot honking behind me who wants to zip up to the merge point and slow down traffic there, but such people are essentially the cause of the problem, so I don't worry too much.

What generally happens is that from the moment I start doing this, overall traffic speeds up just a bit and the lane ahead of me clears fairly quickly. It's win-win for everyone but the hotshot behind me.

Actually, this is suboptimal.

The optimal flow is for everyone to use both lanes up until the last moment, and then to merge by alternation (one car from one lane goes, then one car from the other lane goes).

Unfortunately, this only works when a substantial majority of the drivers understand that this is how it should work. If you don't hit that threshold, then the people who use the disappearing lane are going to get glares.

Here's a link that describes the theory behind optimal merge patterns:


Actually he may be forcing the lanes-full optimal mode to arise.

If one or a few drivers in the empty lane start pacing the cars in the full lane, cars will build up behind them. When they arrive at the end of their lane, the drivers in the full lane won't be so ready to block merges (since nobody was cheating by racing down to the end.) Perhaps this could trigger an outbreak of zipper-merging.

So, what you're saying is: your solution is optimal in theory but suboptimal in practice, and the grandparent's theory is suboptimal in theory but optimal in practice. Right?


He has updated it since '98, and even bought a domain! But yeah, it does work.

I'll bet that you're thinking that this will have to wait a decade or two, until every car on the road has a new generation of computers installed. Look up a Mercedes product called Distronics Plus and a company called Autoliv. These guys are already mounting stereo cameras and long-wavelength IR (heat) cameras on production luxury cars to detect turn signals, pedestrians, and license plates of other cars. When 50% of cars have radars linked to their cruise control, it's a matter of iteration to let them share data, then to characterize the remaining vehicles on the road based on the aggregate observational data.

Your '95 Civic may not be announcing its velocity, but when the car behind you and the car in front of you have both read your license plate and agree on your speed, you may as well be.

Prizewinning 2004 research by LC Davis of Umich showed that stop/go traffic jams were eliminated by adding in a certain percentage of cars with Adaptive Cruise Control (ACC.) It only took a little less than 1 in 5 cars.



What does ACC do? It imposes proper safe driving headway distance, as well as providing near-instant reaction time for braking. No tailgating, no accumulation of 1-sec human reaction times.

But couldn't human drivers do the same by simply padding out their 2-sec spacing to 3-sec? And only a few need to do it.

Huh, I've been using an "algorithm" for 30 years. I noticed on the freeway that when traffic slows in front, drivers overreact and brake more than necessary. This causes the driver behind to also brake more than necessary, eventually slowing traffic to a complete stop.

The point where traffic stops then flows backwards along the freeway like a wave. You can see this if you have a long view of the freeway ahead.

So what I'd do is slow down in advance of the wave reaching me, and try to time it so that the traffic starts moving again before I reach the stoppage. Many times, I've been able to stop the wave from progressing by doing this.

Rather than 100 automated cars, just use one daily commuter.

If you commute along a stretch of highway every day, then you know exactly where the "stop waves" tend to appear. That's the place where you stop driving like a Wolf, and instead drive like a Shepherd: stop speeding and tailgating, and instead open up a large buffer.

I've found that this works well in some places. The other drivers seem to know what you're doing, and sometimes you'll even see others using the same technique. But in other spots everyone seems clueless, and they'll race to fill up the tiniest hole.

Also, as with the Koshima monkeys washing potatos or Blue Tits & milk bottles, in the last ten years the trick seems to be spreading. Some of our "clueless" regions have experienced a state change, and the daily jams seem to appear less often.

I wonder if the articles below have a simple cause:




Absolutely. I watch out for those waves of red lights and adapt ahead of time. They're really dangerous around here if you're not watching out for them, actually. You can help a lot of people behind you avoid problems too if you slow a bit ahead of time, though idiots who blow past are always dangerous. Then again, it's a lot better not to have those morons behind you, so you still win in the end.

yep, that's what I do. Just try to drive the average speed.

Neat idea, I like the idea of a somewhat intelligent "traffic swarm", but the article doesn't address what effects a malicious player could induce on the scheme. What's to stop people from randomly broadcasting "Traffic Jam Ahead" signals to cause slowdowns, or maybe even "Traffic Jam Behind" to cause the vehicles to suddenly accelerate? Considering that, I think it would be possible to come up with an aftermarket module to regulate the behavior of the traffic swarm such that the user would reach his destination faster at the expense of slowing the rest of the swarm down. Any traffic swarming scheme needs to take this type of behavior into account, or else I don't think it would work out very well in practice.

What's to stop someone from spoofing your bank site while you're on the free wifi at Starbucks? I can imagine an aftermarket module designed to do just that, so free wifi is clearly not viable.

In practice, they'll implement encryption and authentication, then make it a federal crime to mess with it.

This is already possible with traffic lights. Emergency Vehicles are equipped to switch traffic lights in their path to green. It's called Traffic Signal Preemption. Of course, the criminal penalties for abusing the system are pretty stiff...

As a child, I used to dream of such a device as I sat impatiently at traffic lights in the car (naturally asking my parents "are we there yet" every 5 seconds). I was pretty excited to learn that it is a real thing!


I thought the article was just saying that the cars will broadcast their speed. In this way cars will know if they are slowing down ahead of them, and it will be built in, so it won't be easily hackable like you described.

Things like this could possibly save Millons of hours a year. Quite powerful.

If it can be hacked, it will be hacked. Just because it's built in means nothing. In fact, I bet it makes it a bigger target.

Ya I know. I meant hacked as in regular drivers randomly broadcasting signals saying traffic jam ahead. Obviously it can be hacked, but it will have to be a lot more sophisticated then that.

in my opinion it is far more important to guard against the actions of the stupid and helpless than the malicious. there are far more stupid people on the road than malicious people, so ...

Sounds like this technique can be imperfectly implemented without any new technology.

Just encourage people to use defensive behavior when they see a jam up ahead, and optimistic behavior as they leave the jam.

Sure, that's limited to people's line of sight (in the case of approaching the jam), but it's better than nothing.

Interestingly, I've been assuming both optimal behaviors intuitively. When I see a jam up ahead, I slow down well ahead of time, hoping that by slowing now, I'll avoid a full stop up ahead; similarly, as soon as I get out of the jam, I speed up quickly. Seems like the natural thing to do after a traffic delay.

We have that system already (at least, here in the Netherlands), where automated overhead signs urge you to decrease your speed in case of traffic jams.

In my experience, not a single person follows those signs. Everybody just keeps on going at full speed until they're forced to stop.

It would be a lot better (for everyone!) if cars were driven by machines that followed those rules properly.

Yup. Game theory exists for a reason, people doing public policy should think about its predictions.

People doing public policy not thinking about its predictions, is one of its predictions.

SCIENCE HOBBYIST: Traffic Waves, physics for bored commuters http://trafficwaves.org/

I always try to do the "average speed" thing in stop and go traffic (if for no other reason that I have a manual transmission and it's really annoying to have to stop and start...) and it has another effect that this page did not mention. The spaces that open up in front of you can often be quite large, and a substantial fraction of the drivers behind you (who presumably have not learned about travelling waves) will honk, flash their light, or even cut into the other lane, pass, cut back in front of you, speed up to 60mph and then slam to a stop, just because you are obviously "not keeping up with traffic".

Why hasn't someone designed a BUMPER STICKER publicizing Traffic Waves? Bumper stickers seem like the perfect medium for spreading the word to other drivers.

How would one explain the why and how of Traffic Waves in such a small space? And don't forget to include a shout-out to trafficwaves.org.

"I drive nice 'n' slow to avoid the stop 'n' go."

"I drive slow 'n' steady when traffic's getting heavy."


That is an awesome website if you've not seen it before.

I-5 into Seattle has networked variable Speed Limit signs that reduce the speed for people approaching traffic jams. Same idea as this article I think, but no need for transceivers in every car.

It won't work while you have aggressive drivers messing with the system.

There's a legal solution - create cameras that detect tailgaters, and fine them. That will encourage people to leave more space, which will give everyone more room to maneuver. Country-town car spacing = country town speeds, not the ~10mph you get in city centers.

I think a technical solution for a technical problem is better than a legal solution to a technical problem, especially since the hardware and infrastructure for such a widespread camera system (not to mention the associated state bureaucracy that would be required to operate it and process the fines) would be unavoidably expensive. Even if it were somehow cheap and simple, monitoring every piece of road would be impossible. And then you would need to legally define "tailgating" and correct for edge cases such as when the person in front brakes suddenly, but the camera still sees it as tailgating.

Interesting idea, although the part where the car starts to "accelerate away quickly using automated cruise control", to leave the jam, sounds a bit scary..

Ford Explorers (and possibly other) have adaptive (automated) cruise control. It's not really scary at all because it measures the distance between your car and the car in front of you, and determines the proper speed to drive. It's no different than you performing that calculation in your head and making the correct adjustment to the gas peddle.

Still a bit scary... Did you see the Volvo demo video of the exact same technology you describe go full speed into the back of a truck?

Yes, and "if Captain Sully had been flying an Airbus he would have crashed because they don't allow manual control". The objective is to reduce the total number of crashes, even if that means that you feel like you have less control of when you do crash.

If you want to play with a simple traffic flow cellular automaton, check out this quick C++ gist: https://gist.github.com/1199391

More discussion at reddit: http://www.reddit.com/r/math/comments/k6ked/can_you_guess_wh...

Here's my algorithm: everyone drives "defensive" all the time.

No new computers needed, and fewer "optimists" rear ending people.

I like that idea, and I'd add it to mine:

People drive at a constant speed.

The root cause of traffic jams, in my opinion, is that people brake much quicker than they accelerate.

This leads to traffic bunching up, and then stretching out as people regain speed.

In the UK, the variable speed limits on major routes work if people obey them, as it cuts down this emergent behaviour.

I think the really cool potential for traffic jam abatement is when basically everyone uses on-line GPS routing (like Google maps), which can dynamically route vehicles based on the traffic density to approach a global minimum.

There is actually research about this, like http://papers.ssrn.com/sol3/papers.cfm?abstract_id=577201

You can do this yourself. Just leave space. Others will see you do this and leave space, too. While making this a science project sounds like fun, that seems unnecessary.

If the traffic pattern is so saturated that people rush into the space you leave, leave space anyway. Leave more space.

People seem to want to optimize for "average distance from the destination". That doesn't get them there any quicker.

> Others will see you do this and leave space, too.

You certainly don't drive where I live.

Reminds me of this:


Looks like he's redesigned his homepage since I got my degree, so I can't immediately find the animations he used to have to illustrate the method.

It also requires a little more automated on-board control than cars currently have but not an unrealistic amount.

That's going to be the biggest problem. Automatic driving of cars isn't a technical problem where you need to tweak something, it's a legal problem.

What happens if driving becomes painless. Won't you have more drivers on the road?

Quite possibly. Someone noticed that the actual speed of traffic in London is always 9 miles per hour. If traffic improves, more people drive until it gets down to 9 mph. If traffic gets worse, people decide driving is too much trouble and traffic improves to 9 mph :)

Likely but by that time we probably wont care because we would be able to sign up for a car online, tell it how many occupants, have it pick us up automatically, join on to other vehicles to form car trains on the roads for optimal efficiency, join mesh networks with neighboring vehicles to organize traffic, never have to stop for signs/lights because cars would seemlessly inter-weave through intersections, all while everyone texts safely.

This solution will never work. Almost all cars should be equipped with the system, which is very difficult task. And just look at this sentence: "At the same time, vehicles leaving the jams are made to accelerate away quickly using automated cruise control." The car accelerating by itself... this is madness.

A lot better way to dissolve would be to show the drivers the traffic jam on the map so they take another way that is free. This is the main idea of this system:


I have used it and it really works, because you can see the traffic jam in advance and plan another route (or the navi can do it for you).

Automated cars = less traffic jams. That was easy.

Installing speed rely in every car might be unpractical and expensive. How about install road sensors and displays flow speed on the side of the road?

Currently people use brake light to signal and detect slowdown ahead, but sometimes people slow down without hitting the brakes. Displaying flow speed every 500 or 1000 feet on the side of the road would give fore warning on traffic slowdown ahead.

"That's an interesting and simple approach that could be implemented relatively easily in the next generation of cars."

Yep, just as soon as we have cars that can not only drive themselves, but we trust enough to cede control. Next generation? I doubt it.

Google has cars that drive themselves in stop and go traffic with pedestrians present already. Next Generation? Closer than you think.

Why do I get the impression that trying to stop traffic jams is about as futile as trying to predict the stock market?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact