Can Google Make Stoplights Smarter? (scientificamerican.com) 64
An anonymous reader shares a report: Traffic along some of Seattle's stop-and-go streets is running a little smoother after Google tested out a new machine-learning system to optimize stoplight timing at five intersections. The company launched this test as part of its Green Light pilot program in 2023 in Seattle and a dozen other cities, including some notoriously congested places such as Rio de Janeiro, Brazil, and Kolkata, India. Across these test sites, local traffic engineers use Green Light's suggestions -- based on artificial intelligence and Google Maps data -- to adjust stoplight timing. Google intends for these changes to curb waiting at lights while increasing vehicle flow across busy throughways and intersections -- and, ultimately, to reduce greenhouse gases.
"We have seen positive results," says Mariam Ali, a Seattle Department of Transportation spokesperson. Green Light has provided "specific, actionable recommendations," she adds, and it has identified bottlenecks (and confirmed known ones) within the traffic system.
Managing the movement of vehicles through urban streets requires lots of time, money and consideration of factors such as pedestrian safety and truck routes. Google's foray into the field is one of many ongoing attempts to modernize traffic engineering by incorporating GPS app data, connected cars and artificial intelligence. Preliminary data suggest the system could reduce stops by up to 30 percent and emissions at intersections by up to 10 percent as a result of reduced idling, according to Google's 2024 Environmental Report. The company plans to expand to more cities soon. The newfangled stoplight system doesn't come close to replacing human decision-making in traffic engineering, however, and it may not be the sustainability solution Google claims it is.
"We have seen positive results," says Mariam Ali, a Seattle Department of Transportation spokesperson. Green Light has provided "specific, actionable recommendations," she adds, and it has identified bottlenecks (and confirmed known ones) within the traffic system.
Managing the movement of vehicles through urban streets requires lots of time, money and consideration of factors such as pedestrian safety and truck routes. Google's foray into the field is one of many ongoing attempts to modernize traffic engineering by incorporating GPS app data, connected cars and artificial intelligence. Preliminary data suggest the system could reduce stops by up to 30 percent and emissions at intersections by up to 10 percent as a result of reduced idling, according to Google's 2024 Environmental Report. The company plans to expand to more cities soon. The newfangled stoplight system doesn't come close to replacing human decision-making in traffic engineering, however, and it may not be the sustainability solution Google claims it is.
Mississauga could do this (Score:2)
The city's had its intersections under local autonomous control with central oversite and overrides since I was a kid and I am not young. I remember it being a news story back in the day.
And let me tell you, traffic is not fun there. I tend to avoid the entire city when I can; the highway offramps are just lobster traps for cars.
Re: (Score:2)
the highway offramps are just lobster traps for cars.
Probably due to the new off-ramp metering systems. Based on the original on-ramp metering, city traffic engineers have implemented systems to relieve local street traffic by throttling back the off-ramps.
When people complain about clogged off-ramps, the city just responds with the same line that the highway department used about on-ramp metering backing up city streets: "Not our problem."
Seattle is the worst! (Score:4, Funny)
I grew up in Southern California and have been through LA and Orange County 100s of times. I lived through San Jose traffic during the Dot Com Boom. I've also spent time in Manhattan.
Seattle has the worst fucking traffic I've ever experienced. The streets, traffic lights, and freeways were designed by Satan. There is no worse place in the world, and I'll never go back there or anywhere close.
Re: (Score:2)
Seattle has the worst fucking traffic I've ever experienced.
You've obviously never been to Manila.
Manila isn't that bad (Score:2)
I've driven in Manila. I'd honestly take it over LA. The traffic on the arterial roads in Manila moves slowly, but it never stops moving. Manila drivers also feel more cooperative and less selfish. Manila requires a different attitude to driving, but overall it isn't the worst experience in the world.
Re: (Score:2)
I've been to pretty much every major city in the US, some many times, and LA traffic was heavy in places but always cleared up eventually. Chicago is just eternal gridlock and horrible highways under construction that never seem to be fixed. And you pay expensive tolls to sit in that traffic. Boston... is all streets that were laid out centuries ago before cars were a consideration, and the drivers are horrible.
Re: (Score:2)
I live in Bellevue, across the lake from Seattle, and used to work in Renton. If I left at 7:15 I-405 would be decent. If I left at 7:25 it would be a parking lot. One day I saw why.
I got on the Coal Creek onramp behind a very small middle aged Asian woman, who proceeded to drive in the middle lane three exits down the freeway at 35 mph and exit. She seemed utterly oblivious to the snarling chaos surrounding her, horns, obscene gestures, tailgaters, nothing made an impact. A few minutes later the radio
Camera Input (Score:3)
Local traffic engineers use Green Light's suggestions -- based on artificial intelligence and Google Maps data -- to adjust stoplight timing.
Optimized timing sounds great, but how about a simple camera system so a light will not turn red (on a timer) unless cross traffic is actually coming?
Re: (Score:2)
In a city like Seattle, there's almost always cross traffic. Lots of modern intersections use sensors for cross traffic and otherwise stay green for the main thoroughfare. No camera needed.
Re: (Score:2)
In a city like Seattle, there's almost always cross traffic. Lots of modern intersections use sensors for cross traffic and otherwise stay green for the main thoroughfare. No camera needed.
There are sensors (induction coils) embedded at intersections to detect if traffic is queued, but these do not detect approaching traffic. You'll still be forced to stop before a green light is triggered.
Re:Camera Input (Score:4, Interesting)
Re: (Score:2)
Re: (Score:2)
but these do not detect approaching traffic
The location of the sensors is the only factor here to detect approaching traffic. I live in a European city where every night every light in every direction is red. They switch to green on approach at every intersection unless another green is interlocking it. If you aren't speeding you will get a green with time to stop for a red in case you don't.
This is achieved simply by putting a second sensor loop further along the road away from the traffic light.
Re: (Score:2)
ideally, the light would transmit data *to* the approaching vehicles, possibly with instruction on speed/decelleration, for optimal approach, based upon when the light will change.
Re: (Score:2)
Seattle is already deriving a profit stream having installed camera/rfid systems that allow privileged (rich) users access to exclusive lanes of traffic bypassing the hoi polli.
This will will be yet another seattle system that enriches the rich and demeans the common.
Re: (Score:2)
Lights in my area generally have main routes green and timers only kick in after someone stops over the sensor loop on a side street (or a pedestrian hits the walk button).
There are photo sensors (not cameras) to detect the coded strobes of fire and ambulance vehicles to give them green lights on demand. It seems to work fairly well, and is probably simpler and more energy efficient than cameras.
If we were going to add cameras... I'd want them to turn the lights green ahead of time for cars approaching fro
Re: (Score:2)
There are photo sensors (not cameras) to detect the coded strobes of fire and ambulance vehicles
Where can you buy those coded strobe emitters? Asking for a friend who says he needs them.
Re: (Score:2)
They're right next to the free criminal charges on the shelf in the "you shouldn't buy this" aisle.
Re: (Score:2)
Not a fan of the camera idea. Too easily co-opted into government surveillance on all of us. Between government and major corporations I already dont feel like I have as much privacy as I should.
Re: (Score:2)
Not a fan of the camera idea. Too easily co-opted into government surveillance on all of us. Between government and major corporations I already dont feel like I have as much privacy as I should.
Fair point. Although with the already existing proliferation of cameras around most cities, along with things like license plate scanners on cop cars, I don't think there's much privacy left to protect.
Re: (Score:2)
Completely agree. I think if current cameras can be duel purposed into helping with timing some stop lights then your idea is a fantastic one since the camera is there anyways (might as well make the best of the camera even if I'd rather it not be there). No one likes sitting in front of an intersection with zero traffic moving through it after all and your idea is definitely realistic.
Re: (Score:2)
Too late, that horse has left the barn. One day my wife and my niece were going to the mall while I was at work. Just out of curiosity I started looking at the traffic cameras between home and the mall, and saw them on 5 of the 8 on their route. This was over a decade ago, camera quality has improved dramatically while cost has plummeted, and connectivity and processing are now ubiquitous. It would be trivial to program something to monitor the movements of a specific vehicle today. Recording the video
Re: (Score:2)
It is being done already. There are UHF antennas at most exits around here. Formally for traffic management , e.g. counting cars. It is done by reading the Bluetooth and RFIDs from cars as they pass by.
Re: (Score:2)
Your anecdote is all well and good but there are still plenty of places in this country where traffic cameras are not omni present and my core argument is that it would be nice if they stayed that way. Rolling out new cameras for "traffic efficiency" would be good way to make them present everywhere though
If you scroll through my conversation with the parent you'll even find me conceding that converting existing surveillance to traffic service would at least be doing something I approve of with those camera
Google has canceled your stop lights (Score:5, Insightful)
Imagine the hell that will break loose when google decided to aburptly exit the traffic control business.
Re: (Score:3)
In this case, Google is acting in an advisory capacity. They are giving guidance on how to program the existing systems. If they leave, it just goes back to being bad. Worse for a while because of any additional traffic that the efficiency has encouraged, but someone else could do similar things for the same municipalities by collecting data with roadside sensors.
Someone should, but not Google. (Score:4, Insightful)
Sounds like a good idea but I wouldn't trust Google with doing it. We need an independent, fair government body to oversee the process and the model.
Google probably would (Score:3)
replace the red-light sign by an advertisement.
Re: (Score:2)
As long as the City gets a cut of ad revenue they would probably go for it.
Re: (Score:2)
Red light ads (Score:5, Funny)
Re: Red light ads (Score:2)
Do cities really want this? (Score:2)
Cities across the U.S. have been changing traffic patterns & street layouts in order to discourage automobile traffic. Changing four-lane streets to two, for example. Where I live there is a section of a 4 lane interstate which bottlenecks down to two and has been traffic-troublesome for decades. Now there's a plan in the works to finally expand that section of freeway and, of course, people are lining up to protest the expansion.
I'm not sure cities with this anti-auto traffic management philosophy trul
Re: (Score:3)
It won't help. Widening roads has been tried before. Usual you get a brief period of respite then induced demand takes over and it fills back up.
Hopefully not real time adjustments (Score:2)
From my experience, Google routing completely falls apart when things get too busy. I'd be a bit concerned what would happen to the lights with real time adjustments.
Re: (Score:3)
Hopefully not. The best systems I've seen use fixed time delays, tuned to optimize traffic flow at some predetermined speed. The whole "real-time adjustment" thing is what you get when Karen shows up to a city council meeting and complains about having to wait to exit her neighborhood for the next full light cycle.
Applying machine learning to discover the optimum timing cycle isn't a bad idea. But then just set the lights and leave them alone.
Deck chairs on the Titanic (Score:3)
Throwing somebody in a multi-ton vehicle by themselves or best with four other people and then expecting them to operate it efficiently on two weeks of training when they were 15 is not a viable transportation system. It is however a fantastic business model if you sell gasoline or own an automotive company.
Re: (Score:2)
None of this stuff works because of induced demand
Maybe to a slight extent, but there's hardly ever any traffic in the town where I live and I rarely have an inkling to just go drive around simply because traffic is low. Where is this fantastical city where people have the time, money, and interest in just 'driving around for fun' because traffic is slightly lower than usual?
Re: (Score:2)
That's either because nobody wants to live there or because your town won't allow new stuff to get built.
Traffic congestion is a sign of success just like a waiting list at a restaurant.
Re: (Score:3)
None of this stuff works because of induced demand.
False. Creating intelligent traffic lights does far more than induce demand. It also increases safety and reduces unnecessary delays in times of low demand (induced demand being a largely peak hour concept).
Also who said it had to be limited to car? When the goal is to move people all you need to do is take into account the people instead of cars. The same kind of intelligent traffic management system that is making cars flow more smoothly can also be used to ensure pedestrians have a green light straight a
NO. (Score:4, Insightful)
And then ... (Score:3)
Obviously no (Score:3)
It would violate the law, Betteridge's law on headlines stating that such questions are always to be answered with 'No'.
Re: (Score:2)
What exact part will violate which law and where? Be specific. You're talking about a program that already exists and has been deployed successfully in multiple countries, so no Betteridge law doesn't apply, instead there should be another law about the necessity to create a clickbait headline phrased as a question just to cause some commentor to incorrectly invoke Betteridge - for the clicks.
Yes, please. (Score:2)
For a long time this has seemed to me to be a great application for AI. So much time wasted at a red light with no cross traffic. Racing up to a light only to have it turn red just soon enough so I have to stop, when there's no other traffic around. Just missing the poll of the left turn sensor (which happens well before the light actually changes) and having to wait an entire cycle. This wastes time, raises stress levels, and causes pollution.
Not gonna happen around here (Score:1)
Many cities will just refuse because it will speed up traffic. They intentionally make traffic suck as a form of social engineering - "if you used mass transit, this wouldn't happen to you." Of course, mass transit is far, far worse (and much slower, a ten minute to work would take me an hour and a half on the bus - and don't forget your towel to wipe the homeless guy's urine off the seat before you sit down), but they don't mention that in the press releases.
Re: (Score:2)
Many cities will just refuse because it will speed up traffic.
Stoplight automation is far more than just moving a car. In fact Google's system has been test deployed in Hamburg where it was used to speed up public transport sharing the road (busses and trams).
Speaking of busses, show me on the doll where a bus touched you.
Re: (Score:2)
I could drive to downtown Seattle in 13-25 minutes depending on traffic, and then spend 10-15 minutes and $20 to find parking. Or I can go to the Park & Ride up the hill, catch a bus within 8 minutes, be downtown in 18 minutes, and spend $3 doing it. Guess which one I prefer? (Especially because I hate driving.)
Get rid of the lights with roundabouts? (Score:3)
Didn't we just see an article about how most traffic lights can be elliminated using traffic circles, which require significantly less ongoing costs or any computer control?
https://www.washingtonpost.com... [washingtonpost.com]
https://usa.streetsblog.org/20... [streetsblog.org]
Re: (Score:2)
That might work fine for towns that don't have a lot of traffic. But when you live in a city like Houston or LA where city streets often have 4-5 lanes of traffic each direction, a traffic circle is not going to cut it. https://www.google.com/maps/@2... [google.com]
Improve (Score:2)
To reduce traffic congestion, make roads narrower, add bike lanes, decent pavements/sidewalks, and decent public transport.
Easy. Cheaper. Fewer cars means less congestion.
Re: (Score:2)
That sounds great, if you don't live in a city that regularly hits 100 degrees in the summer. Nobody wants to be walking around on sidewalks that can literally burn your feet. And in a city like LA or Houston, public transport will easily double or triple your commute time, compared to driving, even though driving commutes are regularly horrendous.
No solution that is so grand, is easy or cheap.
Get rid of as many stoplights as possible (Score:2)
only supposed to blow the bloody doors off (Score:2)
Let's call it ARTIE (Score:2)
I've got a good feeling about this.
All fun 'n' games til it glitches/shuts down (Score:2)
Literally anyone could make them smarter (Score:2)
The lights where I used to live (Dallas, Texas) switched to blinking red after every storm knocked out power. Never mind that they had enough power to display blinking red; they went into full on panic mode, bringing busy intersections to a crawl.
This is inexcusable. A program for an intersection should not vaporize with each loss of power. This is not a clock radio!
elephant gun to kill a tree (Score:2)
Just replace stoplights with over- and under-passes.
No (Score:2)