Traffic Control of the Future

petra13 writes "A high point of the Autonomous Agents and Multi Agent Systems conference this past week was Kurt Dresner and Peter Stone's paper 'Multiagent Traffic Management: A Reservation-Based Intersection Control Mechanism.' They designed an automated system where cars reserve a time to pass through an intersection as they approach it and are then sped up or slowed down to ensure their arrival at exactly the right time. This allows traffic to enter the intersection from all directions simultaneously, eliminating the need for traffic lights and considerably reducing delays caused by stopping traffic. On their website, you can find Java applet simulations to illustrate the system. Especially impressive looking is the six lanes of heavy traffic in all directions simulation. I would love to see this in real life (from a safe distance of course)."

What about.....

[BWJones]

I have to wonder if these simulations or plans account for bicycles or pedestrians?

Hmm..

[bigattichouse]

I wonder if you can apply the same logic to items of in a processor. or in a kernel thread scheduler... hmm.

Um

[Apreche]

That simulation was pretty impressive when I looked at it. Until I realized something. None of the cars are turning left or right. Theories and math and simulations work great and are often impressive. But real world factors will almost always mess them up.

So one day when there is a way to get from everywhere on earth to every other place on earth without turning left or right give me a call. Until then, let's stop and let people turn left.

Great!!

[SeaDour]

Now all we have to do is convince the general population that their cars are safe in the autonomous control of computers rather than their own two hands. Sure, *I* know that having automobiles controlled by a sophisticated traffic network would be safer and more efficient -- I read Slashdot, after all -- but I doubt very many people in this country would be so thrilled about the idea of giving up their grip on the steering wheel.

Re:What about.....

[transient]

Doesn't look like it. Nor do they account for, as someone else pointed out, turning. Even more importantly, at no point during the simulation does a dog run out into the street, a hubcap fall off, or a tire blow out. At the end of their report, the authors mention that humans probably aren't capable of driving within the tolerances required by their system, but they never consider distress/emergency situations.

But, in spite of its limitations, this is an impressive technique and I'm sure that someone will be able to build on it.

Re:What about.....

[Keck]

I have to wonder if these simulations or plans account for bicycles or pedestrians?

They probably account for them by saying this is only for highways, where bicyclists and pedestrians aren't legally allowed (at least in the US) anyway. Besides, you have to start *somewhere* :). In their paper, they list assumptions even greater than !bicycles and !pedestrians:

• no TURNS
• everybody goes roughly the same speed (not a bad assumption on highway)

Overall, a very worthy bit of research IMHO.

Standard vehicles in controlled areas

[kindofblue]

This may not be practical for general traffic, but I could see it being very useful for places where one can control a fleet of individual cooperative vehicles. This could be on a factory floor with robotic delivery vehicles (e.g. in an Amazon-type warehouse), baggage haulers on airport runways, at airports with the airplanes themselves to get to runways, construction sites with heavy machinery, companies with fleets of similar vehicles like at UPS, FedEx, Walmart, military sites with tanks and humvees (using encrypted channels of course), etc.

There are lots of places where you have a need for traffic control with big or many vehicles, in tight spaces. Such resource allocation is a huge part of many problems. That's where they should market this first, I think.

Re:Breakdown?

[nuclear305]

The same thing that would happen without the system...the other people either 1) Stop, 2) Stop, and help move your vehicle if necessary or 3) Drive right into you because they weren't paying attention.

Too many things that could go wrong

[momerath2003]

What happens if a pedestrian walks into the intersection? If a car's brakes fail or it doesn't accelerate as fast as it should?

This would require that every car on the road has both extremely precise acceleration and precise location reference (possible with GPS, but even that only has resolution of a few meters).

In short, this tech certainly won't be around anytime soon.

Roundabouts?

[Anonymous Coward]

I'd like to see this compared to the roundabout intersection.

Solving the wrong problem.

[Moderation abuser]

The problem with all these traffic management systems is that they are attempting to solve the wrong problem. What they should be doing is asking why there are so many people on the road at the same time all going in the same direction.

Hybrid

[t_allardyce]

What about mixing traffic lights with a reservation system? as you get near you signal the computer your intentions early (left/right/straight) and it starts giving you a speed to match, the speed would be tuned to try and prevent you needing to stop or slow down too much which makes everything quicker for everyone, if you did break or you didnt have the system installed (or it malfunctioned) you would just drive like normal and obviously stop if there was a car infront of you or a red light. Technically this already exists - its called 'figuring out how fast you should go' but people either dont bother or get it wrong and end up stopping - the advantage would be that the computer _knows_ exactly when the lights are going to change because its the one doing the changing, there would be no safety issues and the whole thing would be optional? It would be like automated air-traffic-control for cars with the backup feature that cars can stop if needed.

Re:An alternative mechanism

[s7uar7]

The problem with roundabouts (and I'm from the UK by the way), is that there has to be a roughly equal traffic flow from each entry point, otherwise the system falls down. If the majority of traffic is following a particular route, say going straight across, and there is very little traffic they have to give way to (as happens during rush hour), then it's almost impossible to join if they have right of way. The only solution is to start putting traffic lights up on them, and that defeats the whole object.

Wrong!

[TiggertheMad]

Actually you do, and heres why. If he should slam on his breaks, you will almost instantly hit him. The force you hit him with will be minimal, as he will not have had any time to slow down. Basic physics says if you rear end someone who is doing 68mph, when you are doing 70, will produce a 2mph impact.

Now, you say, wouldn't it be better to have enough room to stop completely, and NOT hit them at all? An excellent idea, but you have to have quite a bit of space to go from 70 to 0 + plus the delta distance you travel in the two tenths of a second that is required for you to react.

Now that is a far mor ideal sutiation, but if you have driven on a freeway in any mahor city, you know that the volume of traffic during a busy period will preclude a 50 foot spacing between each car. With a 15 foot spacing, you only insure that when the person infront of you slams on the brakes, that you will hit them pretty hard.

Lesser of two evils, I'll take the 1 foot spacing.

Re:Solving the wrong problem.

[Anonymous Coward]

Because they are only coincidentally sharing one common segment of the road. This is not the same as having the same origin and destination, which you must have for mass transportation.

When you log onto Slashdot, does all your other Internet traffic stop, and are all your neighbors forced to view Slashdot along with you? No. But, oddly enough, all of your packets are going the same way at the same time on the same wire.

Re:Wrong!

[greenrd]

The force you hit him with will be minimal, as he will not have had any time to slow down. Basic physics says if you rear end someone who is doing 68mph, when you are doing 70, will produce a 2mph impact.

I agree with the second sentence but I'm not sure about the first one. How do you figure that the front car will only have slown down by 2mph by the time the back car hits it?

as many have pointed out:

[itzdandy]

as many have pointed out:
no turning
no dogs
no breakdowns
no bicycles

and as i'm pointing out:
no lane changes
no variable sized cars/busses
no emergency vehicles!

=

turning can be solved, the outer most lanes are for turning, and would theirfor not place a lease on the forward motion but would place a lease on the crossing lane so any oncoming traffic the crosses in the turning lane would be told accordingly.

lane changes would have to be allowed only far between intersections, and disallowed in the intersections.

no generic vehicle size could be accounted for, but every vehicle must state it's size when placing lease, so busses could get more intersection time. ALSO, busses should have a higher priority and that could be stated with conditions to acceptance while placing lease.

accidents can be handled via a motion detection system at the intersection seeing non-leased action and routing traffic to other lanes around the incident. if their are 6 lanes, and an accident or breakdown occurs blocking 2 lanes, then the other 4 lanes must be routed for traffic instantly.

Emergency vehicles(EV) must take top priority and must also place a lease as they arrive. other traffic would route around the EV.

pedestrians should not be allowed and high walls and fences should protect such roadways. also, the incedent detection system should be able to see non-lease activity and if it is moving. Then adjust traffic speeds accordingly and signal for human intervention.

=

though these intersections would be autonomous, they would require human monitoring of signaled events, and human can make deccisions and lower traffic speed to adapt.

Re:What about.....

[SammysIsland]

It doesn't account for different length vehicles either, or different slow down/speed up capabilities of different vehicles.

Of course there are weather conditions to factor in as well.

It just looks plain old dangerous to me!

Cars versus airplanes

[mec]

In the USA at least, commercial airline travel is much much safer than riding in a conventional automobile.

And yet people don't care. They think air travel is dangerous but thinking nothing of their cars that kill 30,000 per year and injure millions per year. In terms of human life, there's a WTC catastrophe *every month* on the highways.

So it's not about safety. It doesn't matter whether an automatic system is safer than a human-controlled system or not. People want contro and don't actually care about safety.

Re:What about.....

[Anonymous Coward]

I think "most" may be a bit of an exaggeration. All the peds that I saw get nearly run down were above ground.