This is a cool video, but the paper itself [1] is pretty light. The substance of the paper is basically (from the abstract):
"Despite simulation-based evidence of the potential benefits of SIs [Slot-based Intersections], a comprehensive, analytical framework to compare their relative performance with traffic lights is still lacking. Here, we develop such a framework."
A handful of other quotes from the body examine the rationale and context for investigating this question.
They compare the results of a platooning-type control strategy with a fair scheduling strategy, and the "usual", fixed-cycle strategy, and find that the platooning one, deemed 'BATCH', scales towards the optimal throughput as the number of vehicles increases. There's formulas and graphs to back this up.
This is good work, but the website is a marvellous example of marketing; subtly promoting the higher implications of the work instead of the work itself.
What about pedestrians? For me, a future in which vehicles have more control of our public spaces is worse than the present, no matter the efficiency of the arrangement. Many American cities are notorious for being designed for the automobile, not the human. Let's not double down. More Ramblas!
Heh. I find it funny that fiction portrays machines wanting to destroy humanity. It may turn out that machines with super human reaction times, never get tired or frustrated, and have infinite patience end up being nicer people than actual people.
It is kind of funny because pedestrian crossings already kind of work like this in China. The car will never ever stop for you, rather, it will adjust its speed and swerve slightly to avoid a collision (if expected, if unexpected, then its a totally different game). Likewise, the pedestrian is expected to move in a consistent direction and adjust their speed to avoid being in the way of a car. The result is that car traffic and pedestrian traffic can keep moving at the same time even when orthogonal.
Of course, it often doesn't work, and it is only possible at all given the slower vehicle speeds of a congested road network. Also, the flow of pedestrians in China is rather more continuous than in America.
In most places in the US it is unlawful for cyclists in the road not to follow traffic law. It is pedestrian hostile in the sense that people don't expect it because it is unlawful.
That it is normal doesn't mean it's not pedestrian hostile. I catch myself cycling over-aggressively. For example, everyone knows that pedestrians have right of way on a zebra-crossing. Good luck with that in The Netherlands. Cyclists will simply not give you right of way, they'll swerve around you, ignore you, and generally make you uncomfortable.
The Netherlands is a cyclists country in much the same way as the U.S. is car country, pedestrians are second class.
I've always thought the reason many cyclists treat red lights and stop signs mere warnings instead of rules is because it's annoying to have to speed up from standstill with your own power.
IMO it's a flaw in human-powered vehicles, although some purists will certainly vehemently disagree such a thing can even exist.
Well.. though I'll admit to cycling aggressively, I don't think I've ever ignored a traffic light unless there was literally no traffic around (and even then only when I was in a hurry), and I don't think many cyclists I know do ignore traffic lights very often.
I definitely get not wanting to stop, and I'll try and optimize. For example in Amsterdam there's traffic lights with counters, and as a cyclist you can usually safely go when there's 5 seconds left, but I don't count that as ignoring the traffic light.
Cutting a red light into slow traffic would be very rude.. I don't think that happens a lot here.
As a cyclist myself, I would never run a red light or stop sign, and I have a history of shouting at fellow cyclists that do. Running traffic signals is an utterly stupid thing for a cyclist to do.
You got it the wrong way around. We only have lights and stop signs because of cars. They are a necessity for motor traffic, not bikes and pedestrians.
The flaw is strictly in the box that extends two more metres forward of the actual driver vision and all around offers terrible visibility.
Interestingly enough, right of way for ships/boats in waterways tends to favor the largest/heaviest over the smallest/lightest. It's oftentimes more dangerous for a larger thing to take evasive action than a smaller thing. I think this should carry over to vehicle/pedestrian right-of-way laws. Pedestrians should assume responsibility for their safety and pay attention (including taking evasive action), just as cyclists should do the same in the presence of cars, and cars should in the presence of semi-trucks, etc. I think that's what you are seeing in The Netherlands. Based on my travels in Europe and Asia, it seems a good portion of the world follows this model, and it's the US that is different.
Your analogy works for 99% of the road, where cars and bicycles are indeed favored and passengers have to risk their lives to use the same pavement. The whole point of traffic crossings is to provide some exception to this rule, wherein pedestrians can use the street without having to exercise athletic skills and hyper-awareness to avoid getting struck. The whole idea that pedestrians are lighter/nimbler is false anyway - a teenager on a bike will react better than a nonagenarian with a walker or a parent with a four-year-old by their hand.
"Widely" ? Maybe in altruism-display places like SoCal, where the expectation for traffic interaction is to stop and have a ten minute meeting about how to proceed so everyone can feel leave feeling superior.
In general it's just called yielding, and that it happens in all four dimensions is implicit. The difference in China is that the tolerances are smaller.
Can you elaborate on where this is coming from? I know you're exaggerating, but in my limited experience I haven't noticed abnormally long yield times at intersections in southern California.
There used to be something called a "Californian stop" that was a partial stop rather than a full stop, but after living in LA for a year, I never saw it happen.
Try walking at a moving car in a parking lot, such that if both of you maintain your velocity you will clearly pass behind the car after it has passed. 9/10 times the driver will "helpfully" stop and "helpfully" wave you on in front of them, delaying both of you.
The law as written means you have to yield on a bicycle in exactly the same way you would a car. There’s no room for China-level tolerances, as you describe.
People may also be picturing different scenarios: NYC couriers and delivery people swerving around them inspires scorn, as do cars driving around you at 30mph in a crosswalk even if no contact will be made. Edit: but the former scenario should happen more often than the latter simply because the car is often too big to pull the same stunt without actually causing physical injury.
I love the picture of a 10 minute traffic meeting, though.
This hasn't been my experience. At least twice in the last 7 years, a car has stopped when about to do a right-on-red, to allow me to cross. Sure, that's 0.01% of the time, but not never :)
"The car will ... will adjust its speed and swerve slightly to avoid a collision"
This also does not match my experience. Whether or not I make eye contact with a driver, it's very rare that they make enough of an adjustment to their speed or direction to avoid a collision. I always have to stop, or adjust my speed+/direction.
I thought this whole sensor thing a lot in the past, specifically as an argument to criticize the state of our autonomous vehicle development. We must be very careful handling electronic failures and sensor reliability.
I know someone will bring up "well, traditional car will fail and results in fatal crash too, but technology can help us catch that and predict accidents by looking at 360 degree!". True, but let's not let that promise blinds us.
Also, this senor-led, slot-based system won't work in areas like NYC midtown, so not a one-size fits all solution (that's a given). If we exclude accidents and roadworks, then traffic lights, narrow roads, increase of cars on the road (due to ride-sharing and cars becoming more affordable), and double parking, they all contribute to the majority of traffic jam.
Some futuristic fixes include flying cars, roads in mid-air, roads through buildings, and more underpass might help but they don't scale when you keep adding more cars.
We need to improve our mass transit. If you have ever played Skyline or SimCity, you know that intersection sucks and mass transit is preferred over roads. Too many school buses will slow down traffic. Multi-lane to single-lane can jam the road. The NY subway system coverage is pretty good, but each train runs on one lane and is a single point of failure. Kudos to those working in transportation planning though...
What would people give the odds we achieve a world of autonomous flying drone transport for local travel before a world of autonomous self-driving cars?
It seems right now like we’re really close with self-driving cars but objectively flying ones would be a lot better:
Shorter distances to travel (point to point), easier to implement collision avoidance, faster, no need for this crazy infrastructure we build/maintain of roads, bridges, traffic lights, tunnels, snow plows, street cleaners, etc...
I'm curious if developments like this will mean future cars will not require lights at all. After all, the same communication required for an intelligent traffic intersection like this could just as well be used to exchange breaking, lane changing and position information.
I think a short summary is that the researchers are looking at proposing a few different schedulers for intersections with the aim of implementing them with autonomous cars. The results exist in ideal conditions with little thought given to real-world edge cases, what a transition might look like, how cars would coordinate, or if a square intersection is even a good shape for this kind of thing.
My lingering question after reading the paper is the enforcement mechanism. Process scheduling because there's an omnipotent authority with total control over all processes but whatever system is used needs to work with no central control and if all the cars are programmed to be selfish.
There's been a marked tendency to replace intersections with roundabouts and shared spaces (https://en.wikipedia.org/wiki/Shared_space) in many places, including the outskirts of Lisbon (as well as a few legendary mazes in Britain).
Drawing a (admittedly inaccurate) parallel with shared spaces (which are harder to navigate by sensory deprived humans than standard intersections), I'd say roundabouts (which preserve crossings) would be a more sensible short-term approach overall -- especially considering the remaining human drivers.
Full disclosure: I don't drive anymore. Gave it up decades ago because I prefer to do something useful with my commute times, and resort to public transportation or Uber whenever possible, which (ironically) allows me to arrive earlier to meetings than most of my colleagues (who need to hunt down a parking space).
> Hans Monderman and others have suggested that, by creating a greater sense of uncertainty and making it unclear who has priority, drivers will reduce their speed, in turn reducing the dominance of vehicles, reducing road casualty rates, and improving safety for other road users.
clearly, Hans Monderman has never driven in Mumbai, India.
I think that shared spaces make sense in small towns, university campuses, and other contexts where traffic flow is low due to the inherent context (and, above all, where pedestrians can support each other).
But for modest to high volume traffic, roundabouts are better (although if the traffic flow is constant, you usually have to place the crossings a bit further away from the center of the roundabout, which makes it a little harder on elderly pedestrians to go fully "across" since they have to walk the extra distance).
They actually introduced a shared space at Amsterdam Central Station about 1-2 years ago that I cross daily. The shared space is right next to the incredibly busy ferries and perpendicular to what used to be a bicycle lane. It's now one big square shared by bicycles, pedestrians and scooters.
It does work incredibly well and there are rarely any accidents. The downside however is that it's quite a struggle to cross. It makes me dread to leave the ferry and enter the warzone once again every morning. I understand that it is designed to make me uncomfortable, but it is taken to such an extend that I don't even really want to walk there.
>clearly, Hans Monderman has never driven in Mumbai, India.
I'm reminded of my time in Berkeley. Now, in spite of what Fox News might tell you, Berkeley is one of the most ridiculously upper class places possible. But, it's right next to Oakland, a place that... isn't.
In Berkeley? If you even look like you were about to step out into the road in the middle of the street, people would stop for you. Over in El Cerrito or Oakland? you could be in the crosswalk, and nobody even slows down.
> Drawing a (admittedly inaccurate) parallel with shared spaces (which are harder to navigate by sensory deprived humans than standard intersections), I'd say roundabouts (which preserve crossings) would be a more sensible short-term approach overall -- especially considering the remaining human drivers.
I actually find roundabouts easier to navigate than large crossroads, I think this is partly because they actually are objectively easier and safer to navigate (one stream of traffic) and partly (but mostly) because they are dominant in the UK.
But for the same reason (familiarity) I expect many in the US would find roundabouts hard to navigate (and maybe unsafe as a result?), it's more about what you are used to and what you had to learn as part of your driving test.
It wasn't part of my driving test in Virginia, at least when I took it. I'm guessing if your state has a low number of roundabouts [0] then it's pretty unlikely you'll be tested on it. I'd be interested to hear if Marylanders (#1 in roundabouts:intersections) have it on their test.
Personally, I freak out when I see a roundabout coming up, and slow down dramatically. Or, I completely avoid them.
This is generally how it works in the UK, except for newbies who tend to always stop completely rather than try to time the merge where possible... I've heard that this is actually taught in advanced driving in the UK (to maintain flow when possible, time merges etc). Frankly there are some roundabouts in cities at rush hour which are frightening for new drivers and you just can't join unless you have good timing and can get up to speed quickly (or get a lucky smooth merge).
Roundabouts are not part of the driving test in DC. Indeed, roundabouts are so scary to Americans that most of the ones in DC have traffic lights. We also have roundabouts with roads cutting across them: https://ddot.dc.gov/sites/default/files/dc/sites/ddot/ward%2....
hah, that is completely insane! the design completely looses all the benefits of it being a roundabout. Even those crazy triple mini roundabouts make more sense than that one.
I grew up in Canberra, known as the roundabout capital of Australia [1] as well as being the actual capital of Australia. It has a population of ~350,000, 406 roundabouts and 276 traffic lights.
In my experience they keep traffic moving really well, as long as the traffic load into each entrance is roughly equal - otherwise it's hard to enter the roundabout from a lesser used entrance.
They take up more space than traffic lights in most situations, however Canberra is a planned suburbia so that wasn't an issue there.
This looks like an awesome project, and I hope this receives significant attention from city planners. I imagine as cars become more connected, we'll have better interfaces for integrating mobile traffic sensors with stationary traffic sensors for optimal management of vehicle flow along congested streets. But I do worry about the pace at which city planners adopt such technology at a time when local government budgets are tight.
This is a good video example of how manufacturing works in lean factories. By reducing or avoiding inventories, fetching from temporary warehouses and machining stops (traffic lights here) you can often increase the production rate significantly.
Thanks! Kiva/Amazon is a great real-time optimization challenge. Digikey logistics is a good example of how it is well done with non-machines:
https://youtu.be/pYqzccJdgVY
You're watching it wrong: It's a dystopian future in which all humans died out after years of being forced to take part in a city scale version of frogit with perma-death against autonomous vehicles.
On roads where you can drive 30+MPH (50 KMPH+), I would suggest removing any possiblity for pedestrians to cross. Either make the road 20MPH (30KMPH) max where all space is shared with pedestrians, bicycles and cars - or make much better, safer & faster 50KMPH+ roads.
I imagine you don't even have to put too many of them in, before any pedestrians give up and drive---having tired of needing to climb and descend four flights of stairs or, worse, walk an extra block along ramps, just to cross the street.
If I'm in a car, I'd like to not brake-accelerate-brake-accelerate but rather move in a constant tempo. If I'm a pedestrian, I'd like a car-less safe environment to walk.
In many large cities in Asia, the solution has been to elevate the pedestrian layer entirely using a system of skybridges, at least at the most congested roads. Of course, this would require a large investment in infrastructure.
This reminds me of another article on HN a while ago, about really wide multilane road outside of a school with a pedestrian crossing. The argument was against the design of this road, generally you shouldn't have pedestrians crossing giant roads like this because it's so big and comfortable for vehicle to go fast. I imagine this animation could be reality on a large crossroad without road level crossings, but on smaller ones the vehicles would have to navigate signals still... although they would have the benefit of being able to interface with the pedestrian light system to A) wait to let immediate stream of traffic pass and B) slow down the next lot approaching to increase efficiency by preventing complete stop while the pedestrians cross.
I live in the UK where roundabouts are almost everywhere. While they are definitely more efficient than crossroads they still have some flow control issues in some scenarios: each joining road effectively controls the flow of the next joining road further along the roundabout through priority, this is both a strength (automatic flow control) and a weakness (unbalanced flow of joining roads): Imagine a simple roundabout in place of an crossroad with 4 joining roads, if only two joining roads next to each other are active and of those two, the one to the right has an endless contiguous flow, then all traffic on the road to the left is blocked - it's basically waiting for road to the right to be blocked by someone coming onto the roundabout from the next road along.
As a result we have traffic lights on roundabouts as well sometimes, which prevents this unfairness but also takes them back closer to the efficiency of a crossroad.
I think the best combinations would be autonomous + roundabouts, I think fast crossroads are just dangerous (autonomous or not). Roundabouts force vehicles to slow down due to the forced turn and negotiate one stream of traffic going in one direction rather than two streams going in up to 4 directions at speed. The potential injury from the later is so much higher it's a non-argument.
I also live in the UK, so I'm familiar with roundabouts for the same reasons you are, but I'm glad you've given the explanation for those that are not.
In your example, "an crossroad with 4 joining roads, if only two joining roads next to each other are active and of those two, the one to the right has an endless contiguous flow, then all traffic on the road to the left is blocked", this is a theoretical weakness, but in practice it rarely occurs. For starters, the chances of zero traffic from the other two roads is small, and "continuous" traffic is rarely busy to the level where cars from the road to the left are blocked completely.
Furthermore, if such bottlenecks occur, they can be mitigated against by diverting traffic, through the addition of changes to surrounding roads.
Roundabouts may not be something to be implemented without consideration to the surrounding road infrastructure, but they seem like they would be decent step forward for many busy intersections with traffic coming from multiple directions.
1. Roundabouts are safer and better with human drivers but bad from a usability standpoint. It's annoying to drive, slower, and makes people feel nauseous when driving through many. When we have road usage of autonomous vehicles surpassing 50%, I suggest to remove them.
2. Two-Leveled intersections without left turns seem cost effective and highly advantageous in cities. To lower the cost even further, it may be possible to design them for vehicles only (much lower ceilings needed, so much lower cost). Left turns can be achieved (especially in the US this is viable) bu going right three times (going around the block). Two-leveled intersections without left turns can reach 100% capacity in all lanes from all directions.
3. Lane-specific traffic light already may increase efficiency on intersections. Imagine a 2x2 intersection from both directions: the right turns can be green (where cars turn right on the rightmost lane) while through-traffic can use the left lane.
I feel like so many similar proposals tend to ignore how saturated our roads are, how expensive the road infrastructure maintenance is, and how fast people expect to drive (because of long commutes from endless suburbs).
We need to find innovative ways to implement forms of collective transportation (rail or what not) in already dense areas. Rail/subways have already proven how effective they are. Autonomous vehicles are necessary to improve road safety, maybe they can help slightly reduce the amount space used by roads, but they won't solve most of the problems we have today on our unsustainable roads.
Unsustainable compared to what? New York is clearly demonstrating that it’s incapable of sustaining a subway, and it has just about the best conditions possible for such a project. You think a place that can’t even manage roads can swing a subway?
Is it sustainable to all move
closer together? Not if the residents of existing cities have anything to say about how those cities are governed. They don’t want the extra population.
unsustainable is an absolute, not a relative. A road network doesn't have to be unsustainable relative to anything in order for it to be full to capacity at and for there to be no room for any more roads, which is the situation New York finds itself in. It is physically impossible for NYC to have sustained substantial growth to their private vehicle infrastructure, hence "unsustainable". If they want to grow their transportation network, they need to do it via other modes.
And i reject your assertion that New York has the best possible conditions for transportation infrastructure. A lack of space, a huge demand, and a corrupt mixture of local government and unions don't make for an idea development environment.
Maintaining existing roads is certainly not beyond the economic capacity of the United States. It just costs more than you're willing to pay. That's valid, but the alternatives don't appear to be cheaper. New York also finds maintaining its subways in working order, something that's certainly physically possible, to cost more than it's willing to pay.
High density and high transportation demand are prerequisites for having a public transit system at all. And what dense US city's government isn't corrupt or incompetent?
So a system that requires all vehicles on the road to have flawless sensor and communications systems can achieve some improvement over current-day technology (from the paper, approximately a "doubling of capacity" for intersections with this implemented.)
This is a profoundly uninteresting result, given the current-day technology in question is "a dozen lights attached to a timer." Add devices that are several orders of magnitude more expensive and yeah you'll get something better- that they only achieve a factor of two improvement (in theory!) reads as a stunning failure of autonomous vehicle technology to pull their weight in this regard. (Granted, more clever schemes might be produced that might get better results from this scenario- it's hard to prove that they can't be effective.)
Better idea- instead of having the world pay for sensor/communication systems on every car, have the same sensor packages installed on the traffic lights so that they can operate in a smarter fashion. Orders of magnitude less expensive, and quite possibly as effective in reducing traffic backup.
What I don't understand is why every traffic light doesn't have cameras and some image recognition/machine learning (whether local or remote) to improve the timing - it would be such a cheap and easy fix compared to putting a sensor in each car.
Even OpenCV running on a Raspberry Pi can pick out cars in a moving image - a bad offline solution (with a sensible fallback) would be better than a timer.
On a large remote setup I think you are right though.
OpenCV on a Raspberry does not sound like something that is reliable and easy to deploy.
It‘s a long way to go from something that you hack together in the living room to something you can deploy on hundreds of traffic lights, that works in every weather, that can easily be configured for different use cases, and that can be serviced by the technicians.
You are right. I'm trying to say I think the problem is development bound and not insurmountable (secure, reliable, good fallback to timer) rather than "moving, storing, and processing data" bound as per the parent comment.
I agree with you that developing a prototype wouldn't be too hard. But actually rolling it out within a few cities (and getting cities to pay for it) are where I think the complexities lie. I think it's a social or bureaucratic problem more than a technical one.
A Raspberry Pi is a surprisingly useful and cost effective device when it comes to IOT. Most industrially hardened devices of similar capacity cost a bit north of an order of magnitude more.
There are really only a couple of things that would need to be addressed, a good weather-tight case that allowed for heat dissipation, decent storage (most microSD has a short lifespan), and a good 5v power supply.
In many cases, designing a system that is cheap, expendable, and expected to fail is way less costly than the alternative.
Sone of what your saying is kind of making the point of the parent though.
Yes, you can make the software deployment go easier, and resin looks like an interesting project from that front. But that really is only a small part of the problen, and those hardware problems you described aren't trivial. The testing you described to convincingly show that your device can handle the environmental requirements isn't trivial. (And by environmental, I dont just mean weather, but also things like available power quality, vibration, etc.)
These industrial devices aren't expensive just because the suppliers aren't aware of RasPi, or are looking to gouge you. Its because they put in a lot of work to qualify these devices for their intended conditions. The RasPi, hacked up solution might actually be the WAY more expensive option over it's lifetime, if I have to roll maintenance crews every couple of months to replace SD cards or fix a flaky connector. And god forbid our code hangs in a way that makes all of the lights at an intersection green. "Oh you have a way to prevent that? Prove it." Designed to fail works fine for microservices where I can spin up a new instance rapidly, but it isn't the best match for industrial or infrastructure in most cases.
Apologies if I'm coming across overly snarky here. I genuinely love the RaspPi, and its got a great niche for prototyping, hacking up one-off projects, or even as the base for many types of connected projects. But I've also worked on industrial and infrastructure types of systems, and more than once I've seen someone say that they've got a cheaper solution to X, when really they've got something that may look cheaper if your only looking at upfront costs, but they are completely ignoring the lifetime costs of the solution.
No apologies necessary. I wasn't vehemently disagreeing with the parent so much as saying that the cost-benefit analysis is worthwhile.
I do some IIoT stuff, so I have a little experience here. Many of the lower cost PLCs and industrial edge devices have no better environmental characteristics than a Raspberry Pi. They just happen to be more expensive by comparison because they're competing in a market with much more expensive devices. In many cases, as you suggested, buying those devices is a false economy.
My point is simply that they're not that bad, and like you said, one should truly consider the lifecycle costs.
A traffic light would only need to solve a much simpler problem than an autonomous car- after image-recognition, given the number of cars waiting/approaching at each direction, decide which lane(s) to give a green light to. Finding a reasonably optimal policy could be hard, yes, but once one is found that could be applied to all (or most) traffic lights, and it couldn't plausibly be more computationally intensive to run than the image-recognition step.
Meanwhile, if conversation involves a plausible future where all/most vehicles (inevitably including police cars and other government-owned vehicles) are autonomous and have the same kind of sensors, its hard to imagine the anti-surveillance angle being a decisive factor. I'm not for pervasive surveillance, and perhaps the NSA will be nice and claim really convincingly that they can't access records from traffic lights... but that bird has flown, anyway.
Lots of traffic lights in my area already have sensors (induction loops to detect waiting cars or approaching bicycles, IR motion detectors to detect pedestrians). In some cases this is really useful (eg. a traffic light for a bike path does not need to switch very often in winter, so it only switches when a bike approaches)
I think that the problem is that often there is too much traffic from all directions and even with all the sensors in the world the traffic light can‘t change that.
I believe the induction loops only sense if the first car is present. This is helpful, but can't detect the number of cars which I think is the next most important metric.
There's also an amusing "failure mode" of induction loops: impatient people pulling up further into an intersection -- for example, aiming for an opportunistic left turn -- and they move off of the loop. The lights cycle; the left turn green arrow which would give them an exclusive phase never materializes.
At busy intersections, there will typically be a loop at the 1st spot and a second loop further back - say, at the 5th spot. It's not perfect, but it's pretty good.
Many traffic lights do have cameras/sensors to detect waiting traffic and measure wait times.
It's not always quite so simple as using ML to adjust light timing, however. In big cities, at least, traffic lights are sometimes deliberately intended to slow traffic down in order to reduce volumes and speeds on the street, or to reduce congestion elsewhere on connecting roads.
As others are indicating, this is tech exists, and in some places it's widely deployed, where in some others it's nowhere to be found. I've seen them in several US cities and suburbs, in many states, usually along major arterials and covering the busiest intersections.
Missouri Department of Transportation's St Louis District has one of the best FAQs [1] on this tech that shows how these cameras work: exactly as you'd suspect from some OpenCV proof-of-concept.
The cameras are a more robust replacement of the in-pavement induction loops, which are much more widely deployed, but give you a subset of the information at a higher cost.
Sydney actually developed an integrated traffic light system titled "Sydney Coordinated Adaptive Traffic System" (SCATS) which has been exported around the world. It adjusts timing of traffic light systems using a variety of sensors - if a bus is running late it will affect the phasing to give it more green lights to try and get it back on time.
I believe the only car sensor for SCATS is an inductive loop so it doesn't know the amount of backed up traffic, and it was developed in the 1970s and rolled out across Australia in the 80s - it's better than a timer but I don't think it should be what the world aspires to.
That being said, I had some interesting conversations with a person who worked for Transport for NSW who said some of the signalling backend was recently improved to streamline future integration with self driving cars, I can't find a reference for this though.
This is usually called adaptive traffic control, and has been implemented (minus machine learning, I'm guessing) in a bunch of places, including a part of New York City [1] and, more famously, Australia [2]. The latter system, pioneered by the city of Sydney, has been sold to other countries such as Singapore. Not sure how technologically interesting these systems are. The Sydney system was developed in the 1970s, but has been improved with more modern tech (such as cellular networking) since then.
> that they only achieve a factor of two improvement (in theory!) reads as a stunning failure of autonomous vehicle technology to pull their weight in this regard.
Sounds like a profoundly interesting result to me -- strong evidence that we'd need to look elsewhere to increase transit carrying capacity. Maybe AI cars can't reduce traffic that much afterall.
It's a profoundly uninteresting result because, you know, we can already build autonomous trains. If you restrict the environment to what works for autonomous cars, we could have had them centuries ago. They will probably run on rails and look a lot like trains, but there you go..
>Better idea- instead of having the world pay for sensor/communication systems on every car, have the same sensor packages installed on the traffic lights so that they can operate in a smarter fashion.
Human drivers preclude such a system. The article actually is about replacing the intersection but you still need the self driving cars to actually perform the optimizations.
I think people are focused on maximizing the wrong metric. Traffic projects like this tend to want to maximize car throughput, when what we really need is to minimize travel distance/time. Maybe then society would start looking at the effects that cars have made to sprawl, and the self-reinforcement cycle that emerges.
I honestly think the real toll is the mental concentration required and the stress caused by driving a car. Self-driving cars automatically fix that problem by removing the need for the passenger to concentrate on not killing people. When I take the bus I don't worry so much about my commute time, because I spend the ride relaxing and reading.
I think it is a both thing rather than an either-or thing.
As long as roads continue to exist in any form, it seems crazy not to aim for efficiency. What is to be gained by passing up opportunities to make them efficient?
So this is not a wrong metric, it is just that both metrics need to be taken into account.
This all works okay if you can trust every car on the road, but what about XYZ hacker broadcasting false information to slow everything down? Or more nefarious, they could cause an accident.
>> With today's traffic volumes, queues would vanish and travel delays would be cut to almost zero.
I’m pretty sure it’s critical in traffic engineering to consider how the design would hold up in the event of larger volume given greater efficiency. There’s a single lane exit on the Mass Pike about 5 miles outside of Boston that’s always backed up onto the highway. I’d always wondered why they hadn’t added a lane, put a divider, or otherwise designed the exit differently. A few years ago a former colleague met the engineer who did the traffic analysis for that exit. Apparently adding capacity to achieve greater throughput would’ve resulted in worse traffic at the exit. According to the models they developed, the more constrained solution was actually optimal, although to the frustrated driver (me!) it still seems poorly designed.
Granted, this is an example of increasing capacity rather than improving efficiency. Regardless, I’d be interested to see how the MIT solution holds up to increased traffic volume.
These slots look way too small in the animation. Every slot needs a bit of a safety buffer. Getting t-boned is more dangerous than getting rear-ended.
So it seems like they should be trying to amortize that cost by grouping cars together, both horizontally (side by side) and vertically (front to back).
So in other words, have a cluster of, say, 5 or 10 cars that cross the intersection together.
That can be done by adjusting the speed of cars well in advance of the intersection to group them, so it's not necessary to come to a stop as with stoplights.
123 comments
[ 3.3 ms ] story [ 198 ms ] thread"Despite simulation-based evidence of the potential benefits of SIs [Slot-based Intersections], a comprehensive, analytical framework to compare their relative performance with traffic lights is still lacking. Here, we develop such a framework."
A handful of other quotes from the body examine the rationale and context for investigating this question.
They compare the results of a platooning-type control strategy with a fair scheduling strategy, and the "usual", fixed-cycle strategy, and find that the platooning one, deemed 'BATCH', scales towards the optimal throughput as the number of vehicles increases. There's formulas and graphs to back this up.
This is good work, but the website is a marvellous example of marketing; subtly promoting the higher implications of the work instead of the work itself.
[1] http://senseable.mit.edu/papers/pdf/20160316_Tachet_etal_Rev...
Of course, it often doesn't work, and it is only possible at all given the slower vehicle speeds of a congested road network. Also, the flow of pedestrians in China is rather more continuous than in America.
The Netherlands is a cyclists country in much the same way as the U.S. is car country, pedestrians are second class.
IMO it's a flaw in human-powered vehicles, although some purists will certainly vehemently disagree such a thing can even exist.
I definitely get not wanting to stop, and I'll try and optimize. For example in Amsterdam there's traffic lights with counters, and as a cyclist you can usually safely go when there's 5 seconds left, but I don't count that as ignoring the traffic light.
Cutting a red light into slow traffic would be very rude.. I don't think that happens a lot here.
The flaw is strictly in the box that extends two more metres forward of the actual driver vision and all around offers terrible visibility.
In general it's just called yielding, and that it happens in all four dimensions is implicit. The difference in China is that the tolerances are smaller.
People may also be picturing different scenarios: NYC couriers and delivery people swerving around them inspires scorn, as do cars driving around you at 30mph in a crosswalk even if no contact will be made. Edit: but the former scenario should happen more often than the latter simply because the car is often too big to pull the same stunt without actually causing physical injury.
I love the picture of a 10 minute traffic meeting, though.
This hasn't been my experience. At least twice in the last 7 years, a car has stopped when about to do a right-on-red, to allow me to cross. Sure, that's 0.01% of the time, but not never :)
"The car will ... will adjust its speed and swerve slightly to avoid a collision"
This also does not match my experience. Whether or not I make eye contact with a driver, it's very rare that they make enough of an adjustment to their speed or direction to avoid a collision. I always have to stop, or adjust my speed+/direction.
As for (b), it only counts if traffic is going slow enough, and it should be part of a human wave. Otherwise, it is a bit like a game of frogger.
Yes, like a game of frogger.
Maybe the intersection looks like this once you add pedestrians and cyclists: https://vimeo.com/106226560
Edit: also worth checking the earlier discussion from a couple years ago: https://news.ycombinator.com/item?id=11336009
I know someone will bring up "well, traditional car will fail and results in fatal crash too, but technology can help us catch that and predict accidents by looking at 360 degree!". True, but let's not let that promise blinds us.
Also, this senor-led, slot-based system won't work in areas like NYC midtown, so not a one-size fits all solution (that's a given). If we exclude accidents and roadworks, then traffic lights, narrow roads, increase of cars on the road (due to ride-sharing and cars becoming more affordable), and double parking, they all contribute to the majority of traffic jam.
Some futuristic fixes include flying cars, roads in mid-air, roads through buildings, and more underpass might help but they don't scale when you keep adding more cars.
We need to improve our mass transit. If you have ever played Skyline or SimCity, you know that intersection sucks and mass transit is preferred over roads. Too many school buses will slow down traffic. Multi-lane to single-lane can jam the road. The NY subway system coverage is pretty good, but each train runs on one lane and is a single point of failure. Kudos to those working in transportation planning though...
It seems right now like we’re really close with self-driving cars but objectively flying ones would be a lot better:
Shorter distances to travel (point to point), easier to implement collision avoidance, faster, no need for this crazy infrastructure we build/maintain of roads, bridges, traffic lights, tunnels, snow plows, street cleaners, etc...
My lingering question after reading the paper is the enforcement mechanism. Process scheduling because there's an omnipotent authority with total control over all processes but whatever system is used needs to work with no central control and if all the cars are programmed to be selfish.
There was even a BBC report on the latter (which I can't find right now, as usual, only a follow-up: http://www.bbc.com/news/uk-33303031).
Drawing a (admittedly inaccurate) parallel with shared spaces (which are harder to navigate by sensory deprived humans than standard intersections), I'd say roundabouts (which preserve crossings) would be a more sensible short-term approach overall -- especially considering the remaining human drivers.
Full disclosure: I don't drive anymore. Gave it up decades ago because I prefer to do something useful with my commute times, and resort to public transportation or Uber whenever possible, which (ironically) allows me to arrive earlier to meetings than most of my colleagues (who need to hunt down a parking space).
> Hans Monderman and others have suggested that, by creating a greater sense of uncertainty and making it unclear who has priority, drivers will reduce their speed, in turn reducing the dominance of vehicles, reducing road casualty rates, and improving safety for other road users.
clearly, Hans Monderman has never driven in Mumbai, India.
But for modest to high volume traffic, roundabouts are better (although if the traffic flow is constant, you usually have to place the crossings a bit further away from the center of the roundabout, which makes it a little harder on elderly pedestrians to go fully "across" since they have to walk the extra distance).
It does work incredibly well and there are rarely any accidents. The downside however is that it's quite a struggle to cross. It makes me dread to leave the ferry and enter the warzone once again every morning. I understand that it is designed to make me uncomfortable, but it is taken to such an extend that I don't even really want to walk there.
You mean, no cars?
https://en.wikipedia.org/wiki/High_Five_Interchange
That crosses 16 lanes with 14 lanes. (3,5,5,3 with 3,4,4,3)
Or Naples, Italy.
I'm reminded of my time in Berkeley. Now, in spite of what Fox News might tell you, Berkeley is one of the most ridiculously upper class places possible. But, it's right next to Oakland, a place that... isn't.
In Berkeley? If you even look like you were about to step out into the road in the middle of the street, people would stop for you. Over in El Cerrito or Oakland? you could be in the crosswalk, and nobody even slows down.
https://www.theguardian.com/cities/2016/may/17/superblocks-r...
I actually find roundabouts easier to navigate than large crossroads, I think this is partly because they actually are objectively easier and safer to navigate (one stream of traffic) and partly (but mostly) because they are dominant in the UK.
But for the same reason (familiarity) I expect many in the US would find roundabouts hard to navigate (and maybe unsafe as a result?), it's more about what you are used to and what you had to learn as part of your driving test.
Are roundabouts part of driving tests in the US?
Personally, I freak out when I see a roundabout coming up, and slow down dramatically. Or, I completely avoid them.
[0] https://www.citylab.com/transportation/2016/03/america-traff...
You can see the segment here[0]
They measured throughput for both 4-way stops and roundabouts, with both people who hadn’t used them before and people who had.
Unsurprisingly (for me at least) roundabouts were significantly better, even for people unfamiliar with them.
0: https://m.wimp.com/mythbusters-four-way-stop-vs-roundabout/
People will learn to use turn signals quickly if they are fined.
Works well in Europe. Though I can understand why this wouldn't work in the US.
In my experience they keep traffic moving really well, as long as the traffic load into each entrance is roughly equal - otherwise it's hard to enter the roundabout from a lesser used entrance.
They take up more space than traffic lights in most situations, however Canberra is a planned suburbia so that wasn't an issue there.
[1]: http://www.abc.net.au/news/specials/curious-canberra/2016-12...
You're watching it wrong: It's a dystopian future in which all humans died out after years of being forced to take part in a city scale version of frogit with perma-death against autonomous vehicles.
My city put bridges over a lot of the major roads for people to cross. A few roads have a tunnels underneath as well.
As a result we have traffic lights on roundabouts as well sometimes, which prevents this unfairness but also takes them back closer to the efficiency of a crossroad.
I think the best combinations would be autonomous + roundabouts, I think fast crossroads are just dangerous (autonomous or not). Roundabouts force vehicles to slow down due to the forced turn and negotiate one stream of traffic going in one direction rather than two streams going in up to 4 directions at speed. The potential injury from the later is so much higher it's a non-argument.
In your example, "an crossroad with 4 joining roads, if only two joining roads next to each other are active and of those two, the one to the right has an endless contiguous flow, then all traffic on the road to the left is blocked", this is a theoretical weakness, but in practice it rarely occurs. For starters, the chances of zero traffic from the other two roads is small, and "continuous" traffic is rarely busy to the level where cars from the road to the left are blocked completely.
Furthermore, if such bottlenecks occur, they can be mitigated against by diverting traffic, through the addition of changes to surrounding roads.
Roundabouts may not be something to be implemented without consideration to the surrounding road infrastructure, but they seem like they would be decent step forward for many busy intersections with traffic coming from multiple directions.
“Without tunnels we’ll be in traffic hell forever” -Elon Musk
1. Roundabouts are safer and better with human drivers but bad from a usability standpoint. It's annoying to drive, slower, and makes people feel nauseous when driving through many. When we have road usage of autonomous vehicles surpassing 50%, I suggest to remove them.
2. Two-Leveled intersections without left turns seem cost effective and highly advantageous in cities. To lower the cost even further, it may be possible to design them for vehicles only (much lower ceilings needed, so much lower cost). Left turns can be achieved (especially in the US this is viable) bu going right three times (going around the block). Two-leveled intersections without left turns can reach 100% capacity in all lanes from all directions.
3. Lane-specific traffic light already may increase efficiency on intersections. Imagine a 2x2 intersection from both directions: the right turns can be green (where cars turn right on the rightmost lane) while through-traffic can use the left lane.
We need to find innovative ways to implement forms of collective transportation (rail or what not) in already dense areas. Rail/subways have already proven how effective they are. Autonomous vehicles are necessary to improve road safety, maybe they can help slightly reduce the amount space used by roads, but they won't solve most of the problems we have today on our unsustainable roads.
Is it sustainable to all move closer together? Not if the residents of existing cities have anything to say about how those cities are governed. They don’t want the extra population.
And i reject your assertion that New York has the best possible conditions for transportation infrastructure. A lack of space, a huge demand, and a corrupt mixture of local government and unions don't make for an idea development environment.
High density and high transportation demand are prerequisites for having a public transit system at all. And what dense US city's government isn't corrupt or incompetent?
This is a profoundly uninteresting result, given the current-day technology in question is "a dozen lights attached to a timer." Add devices that are several orders of magnitude more expensive and yeah you'll get something better- that they only achieve a factor of two improvement (in theory!) reads as a stunning failure of autonomous vehicle technology to pull their weight in this regard. (Granted, more clever schemes might be produced that might get better results from this scenario- it's hard to prove that they can't be effective.)
Better idea- instead of having the world pay for sensor/communication systems on every car, have the same sensor packages installed on the traffic lights so that they can operate in a smarter fashion. Orders of magnitude less expensive, and quite possibly as effective in reducing traffic backup.
Let alone concerns with people about being recorded in public, even if red light cameras are permitted in some states today.
On a large remote setup I think you are right though.
It‘s a long way to go from something that you hack together in the living room to something you can deploy on hundreds of traffic lights, that works in every weather, that can easily be configured for different use cases, and that can be serviced by the technicians.
https://resin.io
A Raspberry Pi is a surprisingly useful and cost effective device when it comes to IOT. Most industrially hardened devices of similar capacity cost a bit north of an order of magnitude more.
There are really only a couple of things that would need to be addressed, a good weather-tight case that allowed for heat dissipation, decent storage (most microSD has a short lifespan), and a good 5v power supply.
In many cases, designing a system that is cheap, expendable, and expected to fail is way less costly than the alternative.
Yes, you can make the software deployment go easier, and resin looks like an interesting project from that front. But that really is only a small part of the problen, and those hardware problems you described aren't trivial. The testing you described to convincingly show that your device can handle the environmental requirements isn't trivial. (And by environmental, I dont just mean weather, but also things like available power quality, vibration, etc.)
These industrial devices aren't expensive just because the suppliers aren't aware of RasPi, or are looking to gouge you. Its because they put in a lot of work to qualify these devices for their intended conditions. The RasPi, hacked up solution might actually be the WAY more expensive option over it's lifetime, if I have to roll maintenance crews every couple of months to replace SD cards or fix a flaky connector. And god forbid our code hangs in a way that makes all of the lights at an intersection green. "Oh you have a way to prevent that? Prove it." Designed to fail works fine for microservices where I can spin up a new instance rapidly, but it isn't the best match for industrial or infrastructure in most cases.
Apologies if I'm coming across overly snarky here. I genuinely love the RaspPi, and its got a great niche for prototyping, hacking up one-off projects, or even as the base for many types of connected projects. But I've also worked on industrial and infrastructure types of systems, and more than once I've seen someone say that they've got a cheaper solution to X, when really they've got something that may look cheaper if your only looking at upfront costs, but they are completely ignoring the lifetime costs of the solution.
I do some IIoT stuff, so I have a little experience here. Many of the lower cost PLCs and industrial edge devices have no better environmental characteristics than a Raspberry Pi. They just happen to be more expensive by comparison because they're competing in a market with much more expensive devices. In many cases, as you suggested, buying those devices is a false economy.
My point is simply that they're not that bad, and like you said, one should truly consider the lifecycle costs.
Meanwhile, if conversation involves a plausible future where all/most vehicles (inevitably including police cars and other government-owned vehicles) are autonomous and have the same kind of sensors, its hard to imagine the anti-surveillance angle being a decisive factor. I'm not for pervasive surveillance, and perhaps the NSA will be nice and claim really convincingly that they can't access records from traffic lights... but that bird has flown, anyway.
I think that the problem is that often there is too much traffic from all directions and even with all the sensors in the world the traffic light can‘t change that.
It's not always quite so simple as using ML to adjust light timing, however. In big cities, at least, traffic lights are sometimes deliberately intended to slow traffic down in order to reduce volumes and speeds on the street, or to reduce congestion elsewhere on connecting roads.
Missouri Department of Transportation's St Louis District has one of the best FAQs [1] on this tech that shows how these cameras work: exactly as you'd suspect from some OpenCV proof-of-concept.
The cameras are a more robust replacement of the in-pavement induction loops, which are much more widely deployed, but give you a subset of the information at a higher cost.
[1] http://www.modot.org/stlouis/links/signalcameras.htm
https://en.wikipedia.org/wiki/Sydney_Coordinated_Adaptive_Tr...
That being said, I had some interesting conversations with a person who worked for Transport for NSW who said some of the signalling backend was recently improved to streamline future integration with self driving cars, I can't find a reference for this though.
[1] http://www.nyc.gov/html/dot/html/pr2012/pr12_25.shtml
[2] https://en.wikipedia.org/wiki/Sydney_Coordinated_Adaptive_Tr...
Sounds like a profoundly interesting result to me -- strong evidence that we'd need to look elsewhere to increase transit carrying capacity. Maybe AI cars can't reduce traffic that much afterall.
Human drivers preclude such a system. The article actually is about replacing the intersection but you still need the self driving cars to actually perform the optimizations.
It will take time to iterate to get to the 4x-10x improvements we all hope to see.
As long as roads continue to exist in any form, it seems crazy not to aim for efficiency. What is to be gained by passing up opportunities to make them efficient?
So this is not a wrong metric, it is just that both metrics need to be taken into account.
>> With today's traffic volumes, queues would vanish and travel delays would be cut to almost zero.
I’m pretty sure it’s critical in traffic engineering to consider how the design would hold up in the event of larger volume given greater efficiency. There’s a single lane exit on the Mass Pike about 5 miles outside of Boston that’s always backed up onto the highway. I’d always wondered why they hadn’t added a lane, put a divider, or otherwise designed the exit differently. A few years ago a former colleague met the engineer who did the traffic analysis for that exit. Apparently adding capacity to achieve greater throughput would’ve resulted in worse traffic at the exit. According to the models they developed, the more constrained solution was actually optimal, although to the frustrated driver (me!) it still seems poorly designed.
Granted, this is an example of increasing capacity rather than improving efficiency. Regardless, I’d be interested to see how the MIT solution holds up to increased traffic volume.
So it seems like they should be trying to amortize that cost by grouping cars together, both horizontally (side by side) and vertically (front to back).
So in other words, have a cluster of, say, 5 or 10 cars that cross the intersection together.
That can be done by adjusting the speed of cars well in advance of the intersection to group them, so it's not necessary to come to a stop as with stoplights.