Why We Need a Fourth Traffic Light

4thlight Top2
ADVERTISEMENT

If there were a religion based on road markings, their Bible would be the text of the 1968 “Vienna Convention on Road Signs and Signals,” and their Holy Trinity would be the three illuminated components of the standard traffic light: the Red Light, the Yellow Light, and the Green Light — three separate entities, and yet still one whole, the traffic light itself. Well, now researchers at North Carolina State University have proposed a new fourth light for the standard traffic light — a white light, and its usage would be tied to automated vehicles.

Actually, the current announcement is actually an expansion of the original quartet of lights-proposal from February of last year, but this new one factors in pedestrian traffic as well. With 66% of drivers reporting they’re afraid of self-driving cars, it’s not unreasonable to think that one reason is a lack of understanding what an autonomous car might do. The fourth light is designed to help bridge this gap.

Think of it this way: You’re in a regular, non-self-driving car. You pull up to a traffic light at a typical four-way intersection hoping to turn right and there are three self-driving cars in front of you. To make things more complex, there’s also a guy pushing a stroller to your right hoping to cross the street. In a normal situation with no self-driving cars you’d know what’s likely to happen. But what about a situation where many of the cars can speak to one another? In theory, those connected cars could behave differently and more efficiently except… you aren’t a connected car and neither is the pedestrian.

Essentially, the new white light would be used in conjunction with 1. automated, self-driving vehicles with the ability to communicate with other automated vehicles and traffic signal control computers, and 2. human-driven vehicles with similar car-to-car and car-to-infrastructure communication abilities.

Here’s how Ali Hajbabaie, co-author of the paper and an associate professor of civil, construction and environmental engineering at NC State University, describes the concept:

Our earlier work introduced the idea of a fourth traffic signal called a ‘white phase,’ which taps into the computing power of autonomous vehicles (AVs) in order to expedite traffic at intersections – but we had not yet incorporated what this concept would mean for pedestrians. We’ve now expanded our computational modeling to account for foot traffic, and the results are extremely promising for both pedestrians and vehicles.”

To understand what these papers propose, let’s look at what they mean by the “white phase,” which presupposes the existence of automated vehicles that are capable of communication with one another and traffic light signaling systems. For this to be effective, there would need to be a critical mass of these vehicles, along with human-driven vehicles able to connect to the same networks as the automated ones, and neither of these categories of vehicles are here yet.

4thlight 12

Here’s how the abstract to the original paper explains the “white phase,” which is not referring to that summer you swore to only wear alabaster or pearl:

This study presents a vehicle-level distributed coordination strategy to control a mixed traffic stream of connected automated vehicles (CAVs) and connected human-driven vehicles (CHVs) through signalized intersections. We use CAVs as mobile traffic controllers during a newly introduced “white phase”, during which CAVs will negotiate the right-of-way to lead a group of CHVs while CHVs must follow their immediate front vehicle. The white phase will not be activated under low CAV penetration rates, where vehicles must wait for green signals.

Essentially, it works like this: Connected automated vehicles are talking to one another and to traffic lights, and when they all compute an ideal, optimized traffic flow pattern, they can group together and take control of the traffic through an intersection by triggering the white light, which would tell human drivers to just follow the vehicle in front of them through the intersection, which can reduce traffic delays by a pretty wildly varying amount, from 3.2% to 94.06%.

The study also claims a 40.2% to 98.9% reduction in total delay compared to a “fully-actuated signal control obtained from a state-of-practice traffic signal optimization software.”

4thlight 2

So, what do we make of this fourth traffic light? Essentially, this light indicates that some interconnected group of automated cars is taking over traffic flow patterns at that intersection, and human-driven cars should follow.

According to Hajbabaie again,

“Our previous research found that the more AVs there are on the road, the more efficiently the traffic moves,” Hajbabaie says. “To be clear, this improves travel time, fuel efficiency and safety for all of the cars on the road – not just AVs.”

I have no doubt that the mathematical models used for this study do in fact show that traffic can flow better and easier with fleets of AVs evaluating things and talking to one another. But, the issue here is that in the actual, real, grimy world we live in, so far the companies working on automated vehicles do not appear to be focused on developing common vehicle-to-vehicle (V2V) communication standards necessary for this to work. This lack of progress on universal V2V has been noted before by far more reputable sources than myself.

For this fourth light to actually make sense, a lot has to happen. We need more automated vehicles, a lot more, and we need a car-to-car communication standard and much wider-spread car-to-traffic-light-systems, sort of similar to what Audi uses for their Traffic Light Information system.

I don’t think this white light will be added anytime soon, because for it to make any sense, we’d need large numbers of AVs and a comprehensive and widely accepted car-to-car communication system, and we’re just not there yet. It’s a good idea for finding a way to incorporate human-driven cars into the traffic flow optimizations of AVs in a relatively low-cost, well-understood way (people know to pay attention at traffic lights, and the only rule, follow the car ahead of you, is easy enough to convey) so in that sense I think the white light is a smart plan.

Oh, and if you don’t like a white light, the authors of the study are understanding:

“And, just to be clear, the color of the ‘white light’ doesn’t matter. What’s important is that there be a signal that is clearly identifiable by drivers.”

Yeah, it should be something other than white. How about ivory? Or eggshell?

Relatedbar

Internal Report Shows Cruise Didn’t Think Its Robotaxi Dragging A Pedestrian Was A Big Enough Deal To Fix The Cars

Cruise Stopping Its Driverless Taxi Service Reveals What Self-Driving Cars Need To Focus On

A Video Showing A Police Officer Yelling At An Autonomous Car Has Me Worried About Robocar Emergency Overrides

104 thoughts on “Why We Need a Fourth Traffic Light

  1. Professor of civil engineering: “but we had not yet incorporated what this concept would mean for pedestrians.”

    Sounds about normal for that field.

  2. Jason, I take great offense at your skepticism of the existence of a religion centered around traffic lights. We’re very closely related to your taillight-focused cult! In fact, we follow the same basic deities and adhere to the same core concept that all car-related communication should be quickly and clearly understood, without requiring language markers or other arcane symbologies.

  3. Beyond stupid idea and the topshot image is a joke. The science of human behavior would tell you that this will cause more problems than it solves.

  4. 100% this makes sense if we want to be automated driving, no accidents as cars all talk and hopefully avoid accidents with that communication, all while recording everything.

    While on the outside, that’s great! No more accidents.

    No more freedom either, just another step away from being able to cruise to the store without everyone in the government and private data firms knowing you farted on 3rd street and ate a whole bag of share size m&m’s on the way home.

    I hate this with all my heart and soul.

    1. then your phone pops up with “Have you been a victim of dispensing unwanted flatulence? Indigestion need no longer ‘air’ its foul breath with our new product, especially formulated for [INSERT NAME HERE]’s budget and convenience!”

  5. So two things. I don’t think we’ve all mastered the 3 colour system we currently have.

    Second, we don’t need a 4th light, we just need to program the lights to suit the flow required, if everything is talking to each other, it’s fine. Pretty sure to an extent this tech already exists with loops and cameras, this is just removing the car to appear to physically flick the switch…

    1. Yeah. If we’re connecting the lights to the cars anyways, we can just make the lights cycle through their existing colours in the appropriate way to maximize flow. How is a fourth light an improvement?

    2. What it would allow is for groups of turning traffic to travel through at the same time without the addition of turning arrows for all lanes.

      If the concept is optimal flow, then this allows the CAVs to determine that multiple lanes from different directions could turn through or out of the intersection all at once.

    3. Completely agree. Just use the AV’s inputs to change the actual existing traffic light timings. Nobody needs to adopt a new system, we just make better use of the system that every driver has been taught.

  6. If we’re gonna start adding features needing connected vehicles, can we add a voting feature? If, during your driving day, no one down votes you, you get a gold star! Get enough gold stars, you get a cut in your insurance rates!

    If you get upvoted, maybe with enough, you get a cash award!

    But if you’re getting consistently down voted, maybe (with credit to Brian Regan), a helicopter with a big magnet hanging underneath flies over your car, picks it up, and just chucks it into a field so you can be a dickhead OVER THERE.

  7. I concur with all the comments that if these things are so smart just have them extend the green lights, don’t overcomplicate it.

    People already struggle a distressing amount with the system we have, so we reeeeeally think adding another light is a good option!?

  8. So, first of all … has everyone forgotten about green arrows? There’s already a fourth light. Second, if this system is smart enough to coordinate with vehicles, just display the correct light so that there isn’t a confusing decision tree – still just go or don’t go.

        1. Don’t get me started on flashing amber arrows… I’m still yet to fully understand the rational where adding a signal is a better option than just a normal green light and absence of a green turn arrow.

        2. It is a bit disconcerting to get a flashing amber turn arrow before the green particularly when the straight-thru lights are still red. My first thought was the light was malfunctioning before realizing with repetition that oh, that’s just how it is now.

  9. I don’t think the CAV controlled phase needs an extra color.

    If the traffic light is connected to the CAV network, it can just extend the normal green phase as needed.

    Typical academics making things more complicated so they can publish some rubbish.

  10. This is objectively a terrible idea. You’d be introducing random control in a setting were humans all ready act erratically. And who takes priority? If you’re outside of an event with a lot of foot traffic, is the band of self driving Kias going to actively extend a light? Also, look at any city street, there is so many variations on speed between modes. In a combined road/bike lane/sidewalk you can speed variation of 2-50 mph. How far out due you make adjustments when you can’t really predict how fast traffic will be going. Will you adjust light length for large pedestrian event? How many people does it take to adjust? What about say you have a station exit, a major bike path exit and a main drag all within the same intersection. Minneapolis does this a lot. That’s thousands of variations of movement happening every minute, there’s just no current way for AI to predict this accurately though 1’s and 0’s. This is just precision scheduling for cars on the road. It’s only already ruined trains, why not let it ruin your morning commute too.

    1. They are building lots of roundabouts near me and this is good. The problem is that traffic engineers believe they can improve them so they will add a right turn bypass here, and an outer lane there. It is hard enough for people to learn and get comfortable with them, let alone if every design is different.

      1. I hate the multilane round abouts. I like the normal ones a lot, but the multi ones break peoples brains and its really tough for folks who don’t think quickly.

      2. It is better, but the number of people that pull up short in absolute terror or stop mid-roundabout is deeply distressing, as is the number of people who feel that feeding input into the steering wheel is just too much work and drive directly through and over the roundabout and median.

    2. The most mind-blowing (not in a good way) traffic control I’ve ever seen was a roundabout with traffic lights controlling all the entrances.

      This was in Washington, DC, by the way, if that helps explain it at all.

  11. Instead, have the traffic light with an AI camera do all the work. We’ll never get cars all talking together (and there will always be some older ones without the feature). The Camera can detect the cars and people and adjust the flow and change the signals accordingly. no more standard light cycles when no other cars are around. if there is a heavy flow of traffic, keep the light green so all can get through. the one car on the side street can wait an extra 10 seconds.

    also this would work for pedestrian crossing. a person needs more time to cross, hold the light. if it’s a bicyclist and rode through fast, end the walk signal. or if the cross button is pressed but the pedestrian leaves before the light changes, then don’t enable the walk light.

    1. I think the idea is to offload the traffic detection and information processing from the traffic signals to the connected cars. Otherwise they seem do be doing what you said. I don’t disagree with that idea, I’m just not convinced the application in the article is necessarily an improvement.

      1. Yeah, but that’s the wrong direction to shift the burden. There are far fewer traffic signals than cars so if you can centralize the logic in the signals it’s far less new hardware to be rolled out. It also can be rolled out piecemeal – you don’t need critical mass of AVs in order to install a smarter traffic light, and then every existing car on the road benefits too.

        1. I get what you’re saying except for one thing, updating and upgrading hardware and software. The hardware and associated software of the intersections are almost never upgraded more often than the average age of the vehicles using it so I’d want that to be the relatively “dumb” part of the system.
          Kind of the same idea as Carplay or AndroidAuto, the phone (the more frequently upgraded part) does all the processing and tells the relatively “dumb” display what the show and do.

          1. EDIT (because I missed the window to do it to my reply): I think we’re on a similar train of thought on this. As you mentioned in another reply I saw while looking for this one, I don’t really trust municipalities to either upgrade in a timely manner or keep the systems up to date either. That’s why I’d suggest having the decision making done at the side that will be likely to stay more relatively current through normal car turnover than fixed infrasructure. In areas or instances where there isn’t a critical mass of connected cars, there’s no reason the signals couldn’t revert to operating like they already do.

  12. I remember years ago some folks wanting blue signals as red is bad for the colour blind, I am only partially so I wouldn’t be affected but this could be bad.

  13. There is no way this can’t fail, it’s perfect! Ever see a person come down the on-ramp and then slam on the brakes instead of merging? Yeah, this will work out real well.

  14. “would tell human drivers to just follow the vehicle in front of them through the intersection” Yeah, this can’t go wrong. I’ll stick with being a free range driver.

  15. AND the CHV will need some way of assuring that said human has their head up & is ‘on the bounce’ so to speak. Look around at your next light: at least 1 in 5 around here has their head down looking at a phone—aside from normal stuff like administering dope-slaps to their silly friends in the backseat.

  16. Lavender. A nice, calming color. Maybe with lavender scent released by your vehicle at the same time so you’re not just relying on your eyes.

    1. I like the purple idea – distinct enough, and dovetails nicely with the pink signage by me for automated-toll express lane marking. Pink and purple as “this is tech” colors.

    1. Someone needs to teach the drivers in Louisville to not slam on their brakes as hard as possible the second they see a yellow light, even if they’re right about to cross the intersection and going 50mph. I’ve seen someone hit their brakes so late they ended up halfway through the intersection before stopping.

      But then we recently made a list of the top 10 cities with the worst drivers, so I guess I shouldn’t have expected anything else.

  17. Oh, please. Two counties near me recently implemented “flashing yellow left arrows” with signs that say “left turn yield on flashing yellow”. And 90% of these idiots can’t figure it out.
    Besides, we already have a fourth, white-colored traffic signal. It flashes when emergency vehicle pre-emption is active.

    1. I have seen those in the US in the downtown of the city I live in. My feeling is that if you have to read an adjacent sign first in order to interpret a signal while driving it is a bad design.

    2. Flashing yellow is totally logical coming from someone who doesn’t have that locally.

      And a fourth, here, is used as bus/train priority light.

      I think, honestly, better/safer/efficient public transit might be a more noble cause than automated vehicles on public roadways and likely end up with better results for everyone.

    3. I was driving somewhere recently, can’t remember where exactly, and I saw this flashing white light you speak of. I had no idea what it was for. I treated it like a bonus yellow and floored it.

    4. What’s always intrigued to me is the jurisdictional difference in what you do in turn lanes – where I live, absent an arrow’s direction but still on green, you pull into the intersection and wait until it’s clear to turn, doing so before/as it turns red.

      But I guess in other parts of the U.S., you can’t pull into the intersection until it’s clear for your turn?

    5. There’s one not far from me where two lanes are through lanes and the two outer lanes turn off and do not continue. All four lanes are one way. For an absolutely unfathomable reason, when you’re getting green through the intersection, the other two lanes have flashing yellow arrows.

      The first time I saw it slowed down considerably, not quite a panic stop, as I thought it was an emergency signal or something. Nope. Just two flashing lights for normal, legal turns. Why? It’s not even like an unprotected turn, since all traffic is one way on that street. Just, what, why?

Leave a Reply