Wednesday, July 29, 2009

Your comments on jerks and traffic jams


Yesterday's post generated a lot of great conversation, so I thought I'd respond to some of the interesting points brought up in the comments section. First, are platoons (chains of several cars that are traveling close together) really that bad?

Anonymous said...
The thing about these "platoons" is that when you're driving that close you're not just watching the car in front of you but the car in front of him. You actually end up monitoring a couple of cars ahead. So if you see something happening to the car 2,3 or 4 cars ahead you start easying off and you have more time to brake when the mud really hits the fan.
Strangely enough, Dr. Appert-Rolland told me that the specific expressway she was looking at rarely had accidents, though that wasn't the focus of her study. With such long platoons, it seems like rear-end collisions would happen pretty frequently, so maybe there's something to the above comment. I'd be interested to know why accidents don't happen more frequently here. When I asked her what drivers should do about this platoon effect , she mentioned not only obeying the three-second rule, but trying to look ahead if you can, as the comment above suggests. So perhaps you can fine-tune how closely you follow based on how far ahead you can see, meaning the writer of the comment below should probably stay away from bigger cars:
That's another issue. I can't see a damn thing from my Ford Focus when any SUV is ahead of me. I end up depending on the SUV driver's reaction time and style, which is, generally speaking, horrid.
Obviously the three-second rule is a rough way to keep your distance larger than your reaction time. (One commenter brought up that reaction times are actually less than one second. Anyone have any evidence for that?) But sometimes even that rule needs to be broken, bringing us back to the first paper the article mentions:
the three-second rule means that most people would never be able to enter the highway. If you attempted to merge, you would break everyone else's three second rule behind you and everyone would have to slow down to let you merge. So you can't say "never" because your highway would be at capacity in no time at all.
This comment brings up the fact that any small disruption in the flow of traffic is going to ripple throughout the rest of the system. The closer your headway, the more the car in front of you is going to affect how you drive. If you're, say, driving through the Mojave at midnight, you're not going to worry about the tail lights you can see half a mile ahead of you, but if you're tailgating a car on the 405 south, you're going to slam on the brakes every time you see your leader's brake lights flicker. So I'd argue that if everyone followed the three-second rule, merging traffic would have less effect on the rest of the freeway. Appert-Rolland also mentioned that it might help to display a "suggested speed" that reflects the average speed, helping drivers going too fast or slow to choreograph their driving. She pointed out that simply having shorter headways gets more cars through. The problem is that when you have a heavy flow, there's a higher risk of a jam. And once you're in that jam, she said, it's really hard to get out of it. If you could somehow homogenize the flow, you'd smooth out the bumps...theoretically. According to Minnhagen's research, homogeneity isn't actually that helpful.
Yahktoe said...
For anybody interested in reading Petter Minnhagen's paper (or at least an abstract of it), you can find it here:

Flow improvement caused by agents who ignore traffic rules

I think it's worth noting how stylized the pedestrian rules are in this study. I haven't sprung for the full paper, but I find myself wondering if maybe the rules themselves could have been changed in such a way as to make rule abiding the optimal "solution."
A comment after my own heart: I often wonder whether the proverbial spherical cow eats grass and moos. So here's a bit more detail on the rules of this model world. So we've got our pedestrian-only street, divided into a grid. Each pedestrian is randomly assigned to move either up or down the street, with each square representing a possible move. You try to move forward if you can, but if there's someone in that square, you can move sideways. If you're a good, rule-abiding citizen, you always try to go right first. If that square's occupied too, then you can go left. But if you're a jerk, you're behavior isn't so rigid. You toss a coin to decide whether to go left or right first. You've got more freedom.
Even though we're talking about pedestrians here, and in a computer simulation at that, it's still very surprising that the best possible situation where you have to get a bunch of people to move in a coordinated way is to have quite a few of them acting outside of the norm. Minnhagen said he was completely astounded by the results. "I was also fascinated when I was a kid by occasions when you were able to break the rules," he told me.
Peter Minnhagen's expertise is, broadly, in statistical mechanics, and he said that this was his group's first big foray into traffic problems, inspired when someone brought up the question while they were tossing around ideas. (Sign me up for that group!) Cecile Appert-Rolland studied fluid dynamics for years before she delved into the physics of tailgating. So what I like about these two papers is that they apply the physics approach the infuriating and seemingly irrational phenomena in every day life. I'm definitely looking forward to seeing better models, but I'm also just fascinated by the fact that you can look at these systems with a physicist's eye and start to get some answers.

4 comments:

  1. I would suggest identifying those maneuvers in traffic that require cooperation to succeed, and developing either a mutually accepted rule or a communication technology to allow coordination between the cars involved.

    A fast car (fast acceleration, good cornering. great braking) makes it easy to interact with traffic as lesser machines can be out-maneuvered without needing cooperation, as their slower responses may be taken out of consideration (in the proper situation)

    When this is not done well, it is a danger and more than an annoyance.

    The rule I think most important is that of sufficient acceleration while passing so as to not cut off the passed car.

    Usually it is not lack of acceleration that causes drivers to violate this, it seems it is due to infantile bravado, but I suppose ignorance could be preset as well.

    So another rule is to slow down when being passed, as the other driver is unpredictable and may cut you off.

    But even if one ignores the impulse to race the pacing car, and thus risk accelerating into his trunk, their is the risk that a driver set to maneuver will be cut off by an unseen racer, and disaster will result.

    Accelerate far enough of the car you are passing!

    ReplyDelete
  2. Firstly, a model that has two kinds of pedestrian is not much of the real world. Precise details of the two behaviors chosen will presumably make big differences to the steady states and responses of the system to change.

    As far as actual use of the "slow" lane is concerned, I am dissuaded from using it in heavy traffic for two reasons. 1. incoming traffic at intersections rarely yields to traffic on the highway that can't move over. 2. Traffic in the "slow" lane can't move over because traffic in the 2nd lane is moving too close together. Consequence: no-one uses the slow lane except those driving aggressively enough not to care whether the gap they move into in the 2nd lane is too small.

    In the UK, the recommended rule for distance is one car length for every 10 miles per hour. This is much more easily judged than the three second rule. At 70 mph, this is 7 car lengths. 5 is OK, 3 is problem, 1 car length is. 5 car lengths enables people to move into a gap, gaps can be adjusted without too much trouble, and all lanes on the highway can be used.

    ReplyDelete
  3. First commenter: given that (and I agree) it would seem reasonable to require different licensing for different kinds of vehicles. This might also help to reduce the number of monstrous gas guzzlers on the road down to those who actually have some need for them. Tiered licensing that places a 5000lb SUV in the slow lane with laws requiring reasonable distance to the next car, limited passing scenarios, a lower speed limit, and driving tests specific to handling a large vehicle seem like good starts. Likewise another tier that places a sub 2500lb sports car with requirements regarding reaction times, handling training, and a higher speed limit to allow better freedom of maneuvering through the traffic to help keep things moving.

    Not that anecdote is necessarily data, but...
    I have two cars, one is my "daily driver." It's slow, boring, mediocre brakes, poor handling, etc- but gets great mileage and is very reliable.

    The second car is a toybox on wheels. It's high power, reasonably light, great handling and grip, and brakes that can leave you with a seatbelt bruise on a dry road. Not something I take out every day, but there when I "need" it.

    Point is, I've noticed my driving style changes drastically depending on which vehicle I'm in. The first car is relatively unstable at higher speeds. I don't feel confident in its brakes. I can't get up and down in speed reasonably, and use a lot of gas when I try.

    This results, naturally, in less passing. More following from a distance and generally less hoonery.

    In the Toy That Is- I'm a bigger hoon. It's hard to judge speed without checking the gauge. Often It's much easier to see my way through congested situations where the risk is higher. The safest thing might well be to just let me watch the road and traffic without mind to how fast I'm actually going (within limits, of course. One still needs to be at a predictable rate for cars entering and leaving given situations).

    Note: I have professional racing experience. My adjustments between cars is automatic for me, but I'm not sure it is for everyone else. Given the number of people who try to drive a Ford Explorer like it's a Porsche cup car, I suspect not. Tiered licensing sounds nice in this regard.

    ReplyDelete
  4. Basically in metro cities we have found the common problems of traffic jams; traffic jams are basically appear just because of number of vehicles. Therefore in metro cities people are facing several kinds of problems; while moving for office from home. Every time they used to face the same traffic jam problems. So in order to deal with traffic jam problems; we have found number of traffic rules and regulation and also some digital techniques to deal with traffic jam problems.

    ReplyDelete