Slashdot Mirror
New AI Algorithm Beats Even the World's Worst Traffic (vice.com)
"Computer scientists at Nanyang Technological University in Singapore have developed a new intelligent routing algorithm that attempts to minimize the occurrence of spontaneous traffic jams -- those sudden snarls caused by greedy merges and other isolated disruptions -- throughout a roadway network," reports Motherboard. "It's both computationally distributed and fast, requirements for any real-world traffic management system. Their work is described in the April issue of IEEE Transactions on Emerging Topics in Computational Intelligence." From the report: The Nanyang researchers' algorithm starts off by just assuming that, given enough traffic density, shit is going to happen. Someone is going to make a greedy merge -- something is going to cause enough of a traffic perturbation to result in a network breakdown. Breakdown in this context is a technical-ish term indicating that for some period of time the traffic outflow from a segment of roadway is going to be less than the traffic inflow. "We assume that the traffic breakdown model has already been given, and the probability of traffic breakdown occurrence is larger than zero (meaning that traffic breakdowns would occur), and our goal is to direct the traffic flow so that the overall traffic breakdown probability is minimized," Hongliang Guo and colleagues write. Put differently, "our objective is to maximize the probability that none of the network links encounters a traffic breakdown." So, the goal of the algorithm is this maximization, which reduces to a fairly tidy equation. It then becomes a machine learning problem. Things get pretty messy at this point, but just understand that we're taking the current traffic load, adding an unknown additional load that might enter the network at any time, and then coming up with probabilities of network breakdown at each of the network's nodes or intersections. Crunch some linear algebra and we wind up with optimal routes through the network. Crucially, Guo and co. were able to come up with some mathematical optimizations that make this kind of calculation feasible in real-time. They were able to demonstrate their algorithm in simulations and are currently working on a further analysis with BMW, which is providing a vast trove of data from its Munich car-sharing fleet. This may not be as distant a technology as it might seem. As it turns out, only 10 percent of cars in a network need to be driving according to the optimizations for those optimizations to have a positive effect on the entire network.
Traffic 3.0
Life 10.6.1
Does their algorithm account for those people who get all pissy when current GPS implementations route traffic through the publicly funded streets in their neighborhood, and put in speed bumps and other obstructions?
They were able to demonstrate their algorithm in simulations
So they don't beat the world's worst traffic, they beat simulations. Unless someone previously mastered the art of making immensely accurate traffic simulations this is useless.
lucm, indeed.
"Computer scientists at Nanyang Technological University in Singapore CLAIM TO HAVE developed a new intelligent routing algorithm"
It is Singapore.
So if you know a lot about a chaotic system, you can use that knowledge to help avoid catastrophic conditions. That seems quite reasonable, but it doesn't change the essential nature of the problem, which is an infrastructure problem - the coordination and maintenance of a lot of independent pieces. Oh, if only the machines could take care of *that* problem.
there's your algorithm.
Human stupidity will always overcome artificial intelligence.
The mind conceives, the body achieves, the spirit manifests.
Have gnu, will travel.
If you find a way to drive more efficiently, politicians will use it to put off road repair even longer, until the traffic jams are just as bad. For some reason roads are the things that residents get most frustrated about (and indeed, are even willing to pay extra taxes to fix, as seen in elections in California), and yet they are the thing that politicians most would like to delay fixing. I guess that goes for transportation in general.
"First they came for the slanderers and i said nothing."
What would happen if all drivers (or a substantial number of them) use that algorithm? Will still perform that well?
Beats the worst? even the worlds worst?
Im a 70 year old stuck in traffic in Los Angeles. and by Stuck, i mean stuck in the worlds worst. I was born in Inglewood, by which I mean i was conceived and birthed in this car. I grew up a strapping young lad, capable of passing drinks or fetching snacks at a moments notice. As I grew --and as we passed the exit for LAX-- my parents foretold of the one day when I would pilot this car. That day has come.
my one wish before I die --assuming I can merge-- is to see the second sign for the exit to interstate 10. Could this app be the miracle ive prayed for between prayers for the sweet release of death? I sure hope so.
Good people go to bed earlier.
Soylent Green...is cars!
You are welcome on my lawn.
Cannot be repealed. So the improvement in traffic will encourage more people to drive instead of WFH or carpooling or finding alternative transportation, until you have the same gridlock with more vehicles on the road than pre-AI.
Nice. Add to Waze. Next.
.... It fixes Bangalore Traffic.
Nice to see a patent violation posted publically.
can you spell "lawsuit"? I thought you could...
All kinds of efficiencies can be obtained if you drive through living rooms.
If you are ever unsure that a street will go through a neighborhood and take you where you are trying to go, the presence of speed bumps tells you a lot. People wouldn't put in speed bumps if the road wasn't a shortcut, after all.
I wonder how well this would work for computer networks. Handling flash congestions is a very big issue in networks. Those congestions propagate and bring down whole swathes of network areas, just like traffic jams.
I've always made it a rule to keep node usage below 30% capacity to handle such congestions gracefully, but with a more optimal system we could increase that number and thus make the whole system more cost efficient.
i think fix bangllore http://www.stylecollects.com/b...
banglore traffic http://pakshatworl.com/google-...
I already work to normalize the traffic I drive in and if AI did this too in place of just a small percentage of bad drivers we would all benefit. It generally does not help me get to my destination quicker, but does help but all the people in traffic behind me. When I notice bad drivers engaging in start stop traffic in front of me I start leaving a expanding and contracting gap between the car in front and work to keep my car always moving at a constant speed that does not necessitate hitting the brakes. Sure sometimes someone will jump into the gap, but in the end all the traffic behind me starts flowing smoothly and spending less energy braking and accelerating due to some idiot in traffic that tailgates everyone and over reacts with heavy starts and stops.
I provide a simple solution. Allow less cars on the roads and traffic jams will cease to take place except for accidents and sink holes.
The underlying issue of traffic is housing. If you build a bunch of new housing, and don't have public transportation, ya need better roads. I'd *thought* I'd heard that San Diego had some deal where developers were required to improve roads to support the estimated capacity as a condition of building new housing.
... prove this won't work?
What I noticed is that traffic on a highway gets messed up where ever traffic merges when the rates goes pass a certain amount. The solution is to keep cars off the highway once the flow of cars on the highway reaches critical flow. The way you do that is to increase the length of the red lights on the streets that feed the highway. I know it sounds like it sucks but you weren't going to get there faster if the highway is a parking lot.
It's nice to see an algorithmic solution to the well-known traffic wave issue.
http://trafficwaves.org/
Of course if everyone also tried harder at driving better and unselfishly, we would not need the algorithmic solution.
Roundabouts solve this inflow to outflow problem. You can't get in because there's already too many cars trying to get out. Every one to the left of you gets to go before you, and since it's a circle, no one gets to go in. Unlike a 4way stop, where everyone has to take turns going thru.
And to think some idiot wants to do a roundabout runway at a major airport.
Does this mean that if we get to 10% usage of self-driving cars we'll start to see a big difference in traffic congestion?
-
I can't find an exact link but the experiment was in relation with the "Dynamic decision making" topic.
There is a famous experiment/game called "The beer distribution game":
https://en.wikipedia.org/wiki/...
Here's what an expert in this field concluded:
Subjects generate large amplitude oscillations with stable phase and gain relationships among the variables. [...] Analysis shows the subjects fall victim to several 'misperceptions of feedback' identified in prior experimental studies of dynamic decisionmaking. Specifically, they fail to account for control actions which have been initiated but not yet had their effect. More subtle, subjects are insensitive to the presence of feedback from their decisions to the environment
and attribute the dynamics to exogenous variables, leading their normative efforts away from the source of difficulty.
https://dspace.mit.edu/bitstre...
lucm, indeed.