Essentially driverless cars. There’s not much point in them behaving this way at most lights because some cars will be turning which breaks the pattern.
The discussion in this thread so far seems to assume that the cars will move off like a train, the distance between cars remaining the same as they increase speed. But this is not a necessary consequence of starting to move at the same time.
It’s possible for all the cars in the line to start moving at the same time, each accelerating in such a way as to maintain at all times a safe distance from the car in front as they increase to full cruising speed.
I feel like I saw or heard something about Germans all taking their foot off the brake as soon as the light turns green. I can’t find any confirmation on Google.
Specifically at traffic lights there usually are those cars about to continue straight ahead, and those about to turn right. The latter will often accelerate then brake a bit (or suddenly brake sharply if they have not anticipated a pedestrian). A recipe for being rear ended if the following car has not opened the distance.
That‘s not the case. As presumably in other countries, we take our foot off the brake when the brake lights of the car before us turn dark. And direct a laser stare to any drivers‘ backs of head visible before us to get going dammit.
That’s assuming similar acceleration characteristics. A sports car vs. an econobox or even two of the same model of pickup, one with just a driver vs. one loaded & pulling a trailer are accelerate/decelerate/handle very differently.
The ultimate end goal will be an enhanced version of ADS-B currently used by aircraft. It will broadcast speed & probably speed trend - (maintaining, increasing, decreasing), direction, position, vehicle length (I can safely pass behind a crossing motorcycle way sooner than a tractor trailer or a pickup towing a camper/boat). It will probably also broadcast info about obstructions/braking - there’s a big difference between slowing down due to an upcoming turn/exit & panic braking due to a child/animal in the road.
The question is how to handle it in all of the places one doesn’t have GPS today. I drive thru the Lincoln tunnel sometimes. My speed slows down approaching the entrance due to all of the merging lanes after the now-unused tollbooths, it then typically picks up early in the tunnel & then slows again on the far side when we catch up to the traffic ahead of us. GPSs know your speed (& position) going in but loose signal once far enough in the tunnel; it assume I’m maintaining a constant speed so it’s almost always wrong on predicting how far thru I am/when I’ll exit as my car moves along the map. We could probably place sensors in the tunnel to simulate GPS to keep your car up to date but you have the same issues in city ‘canyons’ & on twisty-turny mountain roads where you loose a clear view of the sky for a bit.
ADS-B isn’t required of all aircraft today, only those with an onboard electrical system so balloons, gliders, powered parachutes, skydivers, etc. don’t have it The cars will still need radar/lidar/cameras to handle things on the road that don’t have it, cyclists, pedestrians, kids, pets, & wildlife that may unexpectedly get into your way, as well as the occasional mattress, chair, box, pail, tool or other item that was not properly secured & is now in the middle of the travel lane.
I think it’ll happen after we have nuclear fusion power plants.
It doesn’t make any difference. Each car can still maintain a safe distance from the car in front.
That is a cool video but ignores the need for increasing following distance as you start from a stop, then accelerate up to speed.
But the thing I found most interesting was the traffic “snake” and “phantom intersections”. This also happens with people gawking when there is an accident on the other side of a highway. But the video says people have to change behavior, which ain’t gonna happen.
That’s why things as dangerous as directing a multi-ton steel vehicle down the highway shouldn’t be left to our dumb little monkey wetware.
If they maintain that same distance when stopped for a stop light, assuming the continuous-flow intersection of AI-augmented driverless car fantasies is a non-starter. The reason, as mentioned upthread, that everyone can’t go as soon as the light turns green is because they stopped too close to the car in front of them and thus need time to reestablish a safe following speed when traffic starts moving again. However, if all cars stop at the same time when the light turns red, rather than snugging up to one another, then they’ll be able to all go when the light turns green too.
That’s true but you still wouldn’t get cars through the intersection any faster that way. By the time cars cross into the intersection they are going the same speed at the same separation in either scenario.
I am not sure how realistic it is to model even computer-controlled cars as instantaneously accelerating from a stop.
The classical approach to traffic modelling posits or derives a hyperbolic conservation law where disturbances propagate with finite speed (indeed, why should a red light instantly influence vehicles 10 miles away)?
Neither should directing the same volume of multi-ton steel vehicles several hundred feet ABOVE the highway.
I’ve observed too much of John Q. Public’s earthbound behavior over the years to ever get on board with the whole “flying cars” dream.
As it should be, since Indonesia has one of the world’s worst traffic fatality records, with pedestrians and motorcyclists making up a disproportionate share of the deaths.
That’s the real-life answer, of course. But if reality perfectly mimicked the model we can imagine, surely it would work just fine.
Here’s your thought experiment: instead of individual cars that inevitably do not move in perfect synch, imagine a single, solid metal bar extending from the stoplight to whatever distance accounts for all the cars lined up. Mark the bar with blocks of paint that represent one car-length each, spaced out as far apart at the light as real cars would be. Put the bar on wheels.
Now the light turns green and the bar moves forward. The blocks of paint don’t crash into each other. If we could get our independent cars to move EXACTLY like the paint blocks do, nothing would crash.
I know this is a stupid musing, but that’s the kind of place my brain goes when I’m stuck in traffic.
I think based on what we’re seeing with the fledgeling “air taxi” industry that any realistic flying car is going to be self piloted. I am sure the vast majority (or even all) of the companies currebtly going for it will fail, but is anyone even attempting a model with a driver?
It’s true that a computer-driven car still needs to maintain a safe distance from the car in front of it. But what constitutes a safe distance for a computer-driven car is much shorter than a safe distance for a human-driven car. And much less yet, if all of the cars are computer-driven and all communicating with each other.
But the length of a safe distance changes; a sports car with big brakes need a lot less distance behind a tractor-trailer than the tractor-trailer need behind a sports car.
For routine driving SDV are probably better than humans, it’s some of the oddball one-offs where humans are better about thinking & adapting than the computers are. What we’ve seen in SF with the Waymos & some of the other SDVs is when they don’t know how to handle a situation they just…stop & one of the techs has to come out & assist with getting it going again. Can exactly do that 1000’ up. (the tech’s don’t have a long enough ladder. 
)
Traffic lights work slightly differently in Germany. While the red is still on, the yellow turns on for the last second or so. This gives the drivers at the front of each lane a heads up to get ready to roll. It no doubt signals other drivers to be ready to take their foot off the brake.
This whole thread is a good argument about why we need more roundabouts and fewer traffic lights.