WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
The same way you do, by seeing a solid line and/or the edge of the road?
I took a course on AI in college, and we had an assignment on machine vision where we needed to find circles in a static image.
This was a rather painful process to build, for something that I can do intuitively. And mind you, this was just a static image that wasn't changing, and wasn't having interference injected randomly.
Machine vision is difficult,and is a large part of why autonomous cars have weird fail states. So no, "look for the solid line" is not an answer.
Just to expand on this a bit...
"Look for the solid line" technology already exists, and it's already used in lane assist features on cars on public roads today.
However, if the technology is unable to find a solid line with adequate confidence, it refuses to take control. That's why it's a driver assist feature and not an automation feature.
We might have "look for the solid line" technology that detects the solid line 80% of the time, or 90% of the time. However, humans are very bad at paying attention to semi-autonomous systems. When a system is autonomous for 2 hours but requires intervention for 10 minutes, human attention spaces out.
The danger is that automation levels around 3 on the chart that Jebus helpfully posted can be more dangerous than lower automation levels.
We have to remember that AV safety is not a linear progression, where more automation is safer. It's safest at levels 1 and 2, then has a valley at 3, climbs back up at 4, is safe again at the top of 5.
Yeah, I'm definitely not wild about the middle categories there. If kinda prefer either the driver has all the control, or the car has all the control. Because the car having all the control except WHOOPS THERE MIGHT BE A TRUCK UP AHEAD IN 5 SECONDS, WHO CAN SAY, I'M FUCKING ON OUT OF HERE, YOU TAKE THE WHEEL NOW just seems like a terrible idea.
And I get the instinctive response, here, but it's important to still ask the question of do these Level 2/3s help? It's possible they do reduce accidents and injuries even if it feels more dangerous.
I'm certainly willing to entertain evidence that shows they actually do help - gut instincts shouldn't limit research, because guts are often dumb and wrong. But for some of the reasons cited, it just sounds reeeeeal sketchy, and more motivated by sounding cool than actually improving safety.
I submitted an entry to Lego Ideas, and if 10,000 people support me, it'll be turned into an actual Lego set!If you'd like to see and support my submission, follow this link.
+3
Options
Powerpuppiesdrinking coffee in themountain cabinRegistered Userregular
Doesn't the technology exist to detect if the driver is drunk by smelling the air they're breathing over the steering wheel? I thought that worked like five years ago and people just didn't want it.
Doesn't the technology exist to detect if the driver is drunk by smelling the air they're breathing over the steering wheel? I thought that worked like five years ago and people just didn't want it.
it waa probably faulty
get a few people locked away from their cars unnecessarily and your product isn't going to last
After I entered the car, I was surprised to see another gentleman sitting in the front seat next to the driver. As he drove me to my hotel, my driver explained Okinawa's daiko or "substitute driver" system. When a person has had a drink outside his home, he can call a daiko company who will come in a car with two drivers. While the substitute driver drives the customer home in his car, the daiko car with a second driver follows closely behind. After dropping off the customer and his car at the destination, the daiko car and two drivers drive away. My driver told me that Okinawa's public safety commission carefully regulates the daiko business by making sure drivers and their vehicles are licensed and insured.
Those services do exist in the US, they're just not anything like universal and are as poorly regulated as the rest of the gig-economy services. I can't remember the name of the one I tried to use once but turned out to be just outside of their service area.
I worked for one for awhile. Of course it was run by two idiot brothers who weren't paying taxes and literally fled the city without letting any of the employees know (or even moving out of their office and shutting down the phone line).
tbh if we were serious about cutting down on drunk driving everywhere large enough would use DUI fines towards subsidizing these types of services. At least here the vast majority of drunk people who drive home after the bar do so because they are they need their car in the morning for work. If they can get a ride home, with their car, for say $20-$30+tip that's the route they tend to go.
+1
Options
Munkus BeaverYou don't have to attend every argument you are invited to.Philosophy: Stoicism. Politics: Democratic SocialistRegistered User, ClubPAregular
And I'm not optimistic we ever reach the requirements for the level 5. Extreme winter driving conditions almost feel destined to defeat level 5.
Jesus, I'm gonna trim this tree.
Anyway, I am hesitant to say what can or can't be done by technology, since a lot of the stuff we have nowadays is basically Star Trek level tech. Shit, you change Alexa's name to "Computer" and you're basically on the bridge.
Humor can be dissected as a frog can, but dies in the process.
On the transponder things. My feelings are pretty mixed there.
-On one hand AI is pretty shit at seeing things and that has been a big factor in allowing it to operate. Granted, it's not like this is being limited to just cyclists and pedestrians. The bigger question is how do we prevent this from becoming another thing that fucks over the poor and also individuals with mental impairment. Costs can become a huge issue for the poor. Then even if you make the stuff free, people with mental impairments are more likely to not have those equip do to forgetting. I mean we already issues with poverty being criminalized and I can see how this could easily make that worse, while also hitting certain individuals with disabilities as well, that might not be poor.
-On the other, I'm familiar enough with physics and how shitty a number of drivers are as well. Doesn't matter who fucked up on the road, if you're a pedestrian or a cyclist, you aren't going to be a winner in any collision that happen with you be one of the objects being hit. Hell, if with just assisted stuff, those two groups with transponder would reduce a number of accidents. I've seen plenty of people in the Hampton Roads area that have run out into the street when it's dark and the aren't that visible because even though they have a reflective vest, they are dressed in dark clothing. Even worse because early in the morning you can get fog, while it's still dark, that cuts visibility down quite a bit. A transponder can give the driver a better heads up in a number of scenarios where that could make a difference. Standard caveat of people having to wear it and not do something really fucking stupid, like jump out in front of car that is is moving 40+ MPH.
On the transponder things. My feelings are pretty mixed there.
-On one hand AI is pretty shit at seeing things and that has been a big factor in allowing it to operate. Granted, it's not like this is being limited to just cyclists and pedestrians. The bigger question is how do we prevent this from becoming another thing that fucks over the poor and also individuals with mental impairment. Costs can become a huge issue for the poor. Then even if you make the stuff free, people with mental impairments are more likely to not have those equip do to forgetting. I mean we already issues with poverty being criminalized and I can see how this could easily make that worse, while also hitting certain individuals with disabilities as well, that might not be poor.
-On the other, I'm familiar enough with physics and how shitty a number of drivers are as well. Doesn't matter who fucked up on the road, if you're a pedestrian or a cyclist, you aren't going to be a winner in any collision that happen with you be one of the objects being hit. Hell, if with just assisted stuff, those two groups with transponder would reduce a number of accidents. I've seen plenty of people in the Hampton Roads area that have run out into the street when it's dark and the aren't that visible because even though they have a reflective vest, they are dressed in dark clothing. Even worse because early in the morning you can get fog, while it's still dark, that cuts visibility down quite a bit. A transponder can give the driver a better heads up in a number of scenarios where that could make a difference. Standard caveat of people having to wear it and not do something really fucking stupid, like jump out in front of car that is is moving 40+ MPH.
See, I view this as a symptom of us mixing car and pedestrian traffic more than we need to. It is inappropriate and dangerous to combine 40+ MPH traffic with pedestrians, and adding transponders doesn't really fix that.
+4
Options
Munkus BeaverYou don't have to attend every argument you are invited to.Philosophy: Stoicism. Politics: Democratic SocialistRegistered User, ClubPAregular
This whole transponder thing is idiotic on multiple levels. If the car can't detect pedestrians or bicycles then how am I suppose to believe it will correctly handle other random items in the road, cones, flares, debris, etc. This isn't even getting into the issue of access that will almost certainly get it banned in some cities. SF has rightly banned cashless businesses and I don't think a transponder is going to fly for the same reasons.
+4
Options
SummaryJudgmentGrab the hottest iron you can find, stride in the Tower’s front doorRegistered Userregular
Talking about driver assist features gave me a good example why I support owner customization, and why I'm not fond of solutions to the 'unsafe customization' problem that are along the lines of 'lock the owner out.' (EG, sign the firmware and lock the bootloader.)
EU models of my BMW have a limited lane assist system. It will automatically steer just enough to keep you inside a lane if you're traveling at low speeds on a highway (eg, in heavy traffic). The lane assist system utilizes the power steering to automatically adjust the steering angle, the optical camera (that is also used in assisted cruise control) to detect lane markings, and the proximity sensors (also used to warn you of an obstacle or vehicle when changing lanes) to detect other vehicles. To ensure that the driver doesn't attempt to use it as poor-man's-autopilot, it is tied to touch sensors in the steering wheel. If you take your hands off the wheel, the car beeps at you loudly and then disables the lane assist.
However, in the US, my car is primarily marketed in California, and California has some of the strictest automobile laws in the nation, so every version of my car sold in the US is manufactured to conform to California state law. It's just easier on BMW that way. Build it for California, and it will be legal in other states. I don't live in California, I live in Washington, so I own a car that entirely for commercial reasons is built to conform to the laws of a jurisdiction don't live in.
California state law does not allow this lane assist system. So the US version of my car has this feature disabled in software. It has all the hardware necessary for it, except the touch sensors in the steering wheel. However, there are other states in which it is legal.
When folks in the US want to enable this feature in software, they can either buy a whole new steering wheel with the touch sensors and install it themselves, or they can go a more economical route of buying a small dongle that you plug into the touch sensor cable, which emulates the touch sensor and sends a continuous 'hands on' signal.
The emulated hands-on signal is a step too far. However, I don't have a problem if people import the steering wheel and enable the feature in software. As long as they're in a state that allows it, they aren't doing anything illegal. If the customization (with the steering wheel) is unsafe, then the stock uncustomized version of the vehicle is also unsafe.
I'm kind of surprised re: the CA thing. I'd imagine every manufacturer would be behaving similarly, e.g. standardizing off the strictest common regulatory denominator, and if anything, that luxury manufacturers would bother when others don't.
I've got a 2020 Honda CR-V and it's got ACC and LKAS and will run both simultaneously, at any speed, but gives you haptic feedback through the wheel and the UI barks at you if you're not pantomiming the wheel movements with your hands, with increasing severity.
It will take curves but not turns, it's Android auto compatible but I don't think LKAS will respond to turn way points (although I don't use Android Auto and just run the map on my phone, and I've never tried).
SummaryJudgment on
Some days Blue wonders why anyone ever bothered making numbers so small; other days she supposes even infinity needs to start somewhere.
+1
Options
zepherinRussian warship, go fuck yourselfRegistered Userregular
Talking about driver assist features gave me a good example why I support owner customization, and why I'm not fond of solutions to the 'unsafe customization' problem that are along the lines of 'lock the owner out.' (EG, sign the firmware and lock the bootloader.)
EU models of my BMW have a limited lane assist system. It will automatically steer just enough to keep you inside a lane if you're traveling at low speeds on a highway (eg, in heavy traffic). The lane assist system utilizes the power steering to automatically adjust the steering angle, the optical camera (that is also used in assisted cruise control) to detect lane markings, and the proximity sensors (also used to warn you of an obstacle or vehicle when changing lanes) to detect other vehicles. To ensure that the driver doesn't attempt to use it as poor-man's-autopilot, it is tied to touch sensors in the steering wheel. If you take your hands off the wheel, the car beeps at you loudly and then disables the lane assist.
However, in the US, my car is primarily marketed in California, and California has some of the strictest automobile laws in the nation, so every version of my car sold in the US is manufactured to conform to California state law. It's just easier on BMW that way. Build it for California, and it will be legal in other states. I don't live in California, I live in Washington, so I own a car that entirely for commercial reasons is built to conform to the laws of a jurisdiction don't live in.
California state law does not allow this lane assist system. So the US version of my car has this feature disabled in software. It has all the hardware necessary for it, except the touch sensors in the steering wheel. However, there are other states in which it is legal.
When folks in the US want to enable this feature in software, they can either buy a whole new steering wheel with the touch sensors and install it themselves, or they can go a more economical route of buying a small dongle that you plug into the touch sensor cable, which emulates the touch sensor and sends a continuous 'hands on' signal.
The emulated hands-on signal is a step too far. However, I don't have a problem if people import the steering wheel and enable the feature in software. As long as they're in a state that allows it, they aren't doing anything illegal. If the customization (with the steering wheel) is unsafe, then the stock uncustomized version of the vehicle is also unsafe.
I'm kind of surprised re: the CA thing. I'd imagine every manufacturer would be behaving similarly, e.g. standardizing off the strictest common regulatory denominator, and if anything, that luxury manufacturers would bother when others don't.
I've got a 2020 Honda CR-V and it's got ACC and LKAS and will run both simultaneously, at any speed, but gives you haptic feedback through the wheel and the UI barks at you if you're not pantomiming the wheel movements with your hands, with increasing severity.
It will take curves but not turns, it's Android auto compatible but I don't think LKAS will respond to turn way points (although I don't use Android Auto and just run the map on my phone, and I've never tried).
The CR-V ACC is the best, it’s like masturbating on a pile of money for long trips.
Someone was asking earlier about the drunk driving detection tech?
At the same time though, the mandate has drawn concerns from safety experts and digital rights groups that warn driver monitoring technology could have knock-on privacy implications. In a letter sent last year by the American Highway Users Alliance, the organization urged support of the NHTSA’s DADSS Research Program but expressed concerns that the technology could potentially infringe on driver’s civil liberties.
“It is also extremely important that a technology designed to control human behavior not be imposed before it is clear that civil liberties are protected and the technology works properly – without false positives where law-abiding drivers can’t start their cars and false negatives where law-breaking drivers over the legal alcohol limit rely on the technology to make the dangerous assumption that they are safe to dive,” the group wrote.
The group also expressed concerns over how the collection and storage of driver data would work and who would have the rights to that data. Others, like Surveillance Technology Oversight Project (STOP) Executive Director Albert Fox Cahn, have expressed concerns over the accuracy of driver monitoring technology and potential risks of bias.
Someone was asking earlier about the drunk driving detection tech?
At the same time though, the mandate has drawn concerns from safety experts and digital rights groups that warn driver monitoring technology could have knock-on privacy implications. In a letter sent last year by the American Highway Users Alliance, the organization urged support of the NHTSA’s DADSS Research Program but expressed concerns that the technology could potentially infringe on driver’s civil liberties.
“It is also extremely important that a technology designed to control human behavior not be imposed before it is clear that civil liberties are protected and the technology works properly – without false positives where law-abiding drivers can’t start their cars and false negatives where law-breaking drivers over the legal alcohol limit rely on the technology to make the dangerous assumption that they are safe to dive,” the group wrote.
The group also expressed concerns over how the collection and storage of driver data would work and who would have the rights to that data. Others, like Surveillance Technology Oversight Project (STOP) Executive Director Albert Fox Cahn, have expressed concerns over the accuracy of driver monitoring technology and potential risks of bias.
It also goes into how attention tracking tech apparently biases against those with disabilities
Not seeing details on the bias against those with disabilities.
I did see something about how this will probably mess with the practical allowable blood alcohol level for driving, though. Having a glass of wine with dinner then driving home is actually legal, but there's no particular requirement for a car equipped with anti-drunk-driving technology to allow this. While I'm willing to listen to arguments about how the allowable blood alcohol level should be changed, maybe this should be done as an actual law instead of being put in the hands of a bunch of random engineers? (And having a different allowable blood alcohol level depending on the brand of car you drive seems a bit yikes.)
This whole transponder thing is idiotic on multiple levels. If the car can't detect pedestrians or bicycles then how am I suppose to believe it will correctly handle other random items in the road, cones, flares, debris, etc. This isn't even getting into the issue of access that will almost certainly get it banned in some cities. SF has rightly banned cashless businesses and I don't think a transponder is going to fly for the same reasons.
If handled intelligently, it could be a reasonably useful tech that just adds an extra level of safety. Attach it as an app on smart phones and an optional thing you stick on your bike as a failsafe. If the car sees a person or bike, don't hit it. If the car sees a transponder signal in a place it didn't find a car or pedestrian, don't hit that, either. In low visibility conditions, where a car (or a human) might have trouble seeing other folks, it's just an added precaution.
Making it mandatory is dumb, and making car AI that doesn't function without it is even dumber, but the idea itself has merit.
I submitted an entry to Lego Ideas, and if 10,000 people support me, it'll be turned into an actual Lego set!If you'd like to see and support my submission, follow this link.
This whole transponder thing is idiotic on multiple levels. If the car can't detect pedestrians or bicycles then how am I suppose to believe it will correctly handle other random items in the road, cones, flares, debris, etc. This isn't even getting into the issue of access that will almost certainly get it banned in some cities. SF has rightly banned cashless businesses and I don't think a transponder is going to fly for the same reasons.
Talking about driver assist features gave me a good example why I support owner customization, and why I'm not fond of solutions to the 'unsafe customization' problem that are along the lines of 'lock the owner out.' (EG, sign the firmware and lock the bootloader.)
EU models of my BMW have a limited lane assist system. It will automatically steer just enough to keep you inside a lane if you're traveling at low speeds on a highway (eg, in heavy traffic). The lane assist system utilizes the power steering to automatically adjust the steering angle, the optical camera (that is also used in assisted cruise control) to detect lane markings, and the proximity sensors (also used to warn you of an obstacle or vehicle when changing lanes) to detect other vehicles. To ensure that the driver doesn't attempt to use it as poor-man's-autopilot, it is tied to touch sensors in the steering wheel. If you take your hands off the wheel, the car beeps at you loudly and then disables the lane assist.
However, in the US, my car is primarily marketed in California, and California has some of the strictest automobile laws in the nation, so every version of my car sold in the US is manufactured to conform to California state law. It's just easier on BMW that way. Build it for California, and it will be legal in other states. I don't live in California, I live in Washington, so I own a car that entirely for commercial reasons is built to conform to the laws of a jurisdiction don't live in.
California state law does not allow this lane assist system. So the US version of my car has this feature disabled in software. It has all the hardware necessary for it, except the touch sensors in the steering wheel. However, there are other states in which it is legal.
When folks in the US want to enable this feature in software, they can either buy a whole new steering wheel with the touch sensors and install it themselves, or they can go a more economical route of buying a small dongle that you plug into the touch sensor cable, which emulates the touch sensor and sends a continuous 'hands on' signal.
The emulated hands-on signal is a step too far. However, I don't have a problem if people import the steering wheel and enable the feature in software. As long as they're in a state that allows it, they aren't doing anything illegal. If the customization (with the steering wheel) is unsafe, then the stock uncustomized version of the vehicle is also unsafe.
I'm kind of surprised re: the CA thing. I'd imagine every manufacturer would be behaving similarly, e.g. standardizing off the strictest common regulatory denominator, and if anything, that luxury manufacturers would bother when others don't.
I drive a BMW i3, their electric vehicle, if that adds perspective at all.
every person who doesn't like an acquired taste always seems to think everyone who likes it is faking it. it should be an official fallacy.
The idea that you could massively disrupt traffic through a junction or section of road for an unspecified period of time by taping a beacon to the underside of a drain cover or whatever is an intriguing one
Beats physically gluing yourself to the road surface, as is the current trend
This whole transponder thing is idiotic on multiple levels. If the car can't detect pedestrians or bicycles then how am I suppose to believe it will correctly handle other random items in the road, cones, flares, debris, etc. This isn't even getting into the issue of access that will almost certainly get it banned in some cities. SF has rightly banned cashless businesses and I don't think a transponder is going to fly for the same reasons.
If handled intelligently, it could be a reasonably useful tech that just adds an extra level of safety. Attach it as an app on smart phones and an optional thing you stick on your bike as a failsafe. If the car sees a person or bike, don't hit it. If the car sees a transponder signal in a place it didn't find a car or pedestrian, don't hit that, either. In low visibility conditions, where a car (or a human) might have trouble seeing other folks, it's just an added precaution.
Making it mandatory is dumb, and making car AI that doesn't function without it is even dumber, but the idea itself has merit.
Pedestrians are already treated as second class citizens intruding on the space and time of the important people in cars. It's already the default assumption that pedestrians are responsible for their own safety and the people in the multi-tons death machines should not have to care about anyone else.
So no. Instead of that, spend money to make streets safe, and reduce the number of cars by making drivers pay for their infrastructure and by preventing car centric development.
We should not force people outside of cars to do even more for the benefit of people inside cars.
I don't think the onus should be entirely on bicyclists / pedestrians to keep themselves safe while they are using the road.
However, that comes with the major caveat that I'm not opposed to laws that improve safety even if they may require a person to do things they might otherwise be unwilling to do. There's a long list of things - seatbelt / helmet laws, restricted access freeways, side marker / reflector laws, etc that people have to comply with to utilize public infrastructure.
By regulation (Consumer Product Safety Commission) almost all bikes sold in the US for must be equipped with pedal, front / back, and spoke reflectors. Some cities / states have additional laws that forbid / penalize removal of those reflectors or add additional requirements to bikes like lights (when operated on public roadways at night). Some cities / states have bicycle helmet laws as well.
So just like reflectors / helmets, I'm not opposed to requiring manufacturers to equip bicycles sold to the public with beacons that improve rider safety. I'm not opposed to a program to provide retrofit costs at no cost for existing bicycles. I also wouldn't be opposed to considering the removal or destruction of a beacon on a bike that was equipped with one to be a potentially mitigating circumstance in a crash just like a bicycle where the reflectors were removed. When I go running at night I carry a light for my own safety, and if I were running in the road because there weren't sidewalks I'd be happy to carry a beacon as well for additional safety.
However, given the number of bicycles on the road that aren't equipped with beacons and the potential risk of failure of a beacon this should be considered a supplemental safety feature and autonomous vehicles should be designed with the baseline assumption that cyclists / pedestrians are not wearing / carrying a beacon.
Generally speaking the idea that AVs = Less Traffic comes from the school of thought that most traffic jams end up being caused by people traveling at different speeds. This problem exacerbates as time goes on. It's why in a big metro area like Atlanta, you'll often end up stuck in traffic for an hour with seemingly no cause behind it. Because traffic is a natural consequence of some people going fast and some people going slow.
Or simply of enough cars. At X MPH you can handle Y cars per lane per hour if everything goes perfectly. The moment you hit Y+1 things start to break down and the consequences spread through the traffic like a wave through fluid.
Actually, the speed doesn't matter. If you require 3 seconds of driving time between cars, then the maximum number of cars per hour you can have is 1200 (3600 seconds per hour times 1 car per 3 seconds = 1200 cars per hour). It doesn't matter if the cars are going 5 mph or 500 mph, the throughput will be the same.
Generally speaking the idea that AVs = Less Traffic comes from the school of thought that most traffic jams end up being caused by people traveling at different speeds. This problem exacerbates as time goes on. It's why in a big metro area like Atlanta, you'll often end up stuck in traffic for an hour with seemingly no cause behind it. Because traffic is a natural consequence of some people going fast and some people going slow.
Or simply of enough cars. At X MPH you can handle Y cars per lane per hour if everything goes perfectly. The moment you hit Y+1 things start to break down and the consequences spread through the traffic like a wave through fluid.
Actually, the speed doesn't matter. If you require 3 seconds of driving time between cars, then the maximum number of cars per hour you can have is 1200 (3600 seconds per hour times 1 car per 3 seconds = 1200 cars per hour). It doesn't matter if the cars are going 5 mph or 500 mph, the throughput will be the same.
The 3 seconds is velocity dependent. Assuming your car brakes at ~1g (Feral or somebody up-thread said that's the average braking rate for cars) and you have a reaction time of ~ 1/2s between needing to stop and applying the brakes, 3 seconds of driving distance between cars gives you time to come to a stop from ~56mph. If you're going 5mph you need 0.7 seconds to stop (assuming a 1/2 second reaction time, again). If you're going 500mph you need 24 seconds.
Yeah 1G decel is a typical hard stop - not panic slamming the brakes to the floor, but a controlled “shit I really need to slow down” stop.
Heffling might be right though? Thinking it over, if all the cars are moving slowly, they are bunched up, and you get one car past a reference point every 3 seconds. If they’re going fast, they’re much more spaced out, so your throughput relative to road space has to decrease. Regardless of speed you move the same number of cars through if you strictly adhere to 3s following distance.
The 3 second rule is a good idea for human drivers but may not be strictly necessary for AVs. There’s a lot of concepts of cars grouping together in tight convoys for longer distance travel. In that case, the lead car would need a long follow distance, but the rest of the vehicles in the convoy could be much closer together, leading to more efficient use of space. All the lead car has to do is communicate “we’re stopping now” and the individual cars can react without needing the processing time to see brake lights and push a pedal.
You do if you adhere to "3 second spacing" but that's a bad way of doing it. At 10 mph that's 13m spacing, at 100mph it's 130m, but braking distance doesn't work linearly like that, brakes convert your kinetic energy to heat and kinetic energy is based on velocity squared. It works out somewhat close to doubling your speed triples your stopping distance because brakes can only work so fast (and also your constant thinking time covers more distance). Increased speed = decreased throughput, as you add more cars everybody slows down to fit them in, bunches up and keeps going smoothly.
What an AV can do here is not get frustrated and switch lanes and cause actual stoppages, if everybody is doing 40 mph and has zero reaction time then you can have something like a 20m spacing (1.1s spacing) and your throughput is ~3200 cars/lane/hour. With human drivers you want closer to 30m (1.7s) for 2100 cars/lane/hour, if you have human drivers with fully safe spacing doing constant 70 you need nearly 100m (3.2s) between them for a throughput of 1100 cars/lane/hour
However, you don't always have to maintain a distance such that you can stop as if a wall had suddenly replaced the car in front of you, at highway speeds almost nothing short of a head-on collision with someone going the wrong way will actually instantaneously halt the car in front of you, what you want to be able to do is stop faster than it can reasonably stop in case of a collision
Ok so some assumptions:
1) Cars are points traveling along a line. This allows us to not consider the fact that a car had a front and back, may be partially through our travel point, etc.
2) Deceleration is constant.
The simple relationship is then:
V + A*ts = 0, where
V is the speed at start of breaking;
ts is the time spent decelerating;
A is the deceleration, which is negatively signed (e.g. -10m/s).
Then;
ts = -V/A; which is the time to stop from first application of the brakes.
The distance equation is:
dt = di + V*ts + A*(ts^2), where;
dt is the total distance traveled;
dr is the distance traveled while reacting;
dr is related to the initial velocity as:
dr = V * tr, where
tr is the reaction time of the driver.
We can then rewrite the distance equation as:
dt = V*tr + V*ts + A*(ts^2)
Throughput is a measure of the frequency of the cars, and is;
T = 1 / (tr + ts); where T is the number of cars that pass through a point per second.
And this is where I realize that throughput is a really, really bad metric. Because to increase throughput, you would decrease the speed that cars travel. But that will result in more time spent on the road, which will decrease the quality of a driving experience. Throughput is probably a good measure for road design consideration (maximizing utilization and road-wear calculations), but not I think for the purposes of our discussions.
You do if you adhere to "3 second spacing" but that's a bad way of doing it. At 10 mph that's 13m spacing, at 100mph it's 130m, but braking distance doesn't work linearly like that, brakes convert your kinetic energy to heat and kinetic energy is based on velocity squared. It works out somewhat close to doubling your speed triples your stopping distance because brakes can only work so fast (and also your constant thinking time covers more distance). Increased speed = decreased throughput, as you add more cars everybody slows down to fit them in, bunches up and keeps going smoothly.
What an AV can do here is not get frustrated and switch lanes and cause actual stoppages, if everybody is doing 40 mph and has zero reaction time then you can have something like a 20m spacing (1.1s spacing) and your throughput is ~3200 cars/lane/hour. With human drivers you want closer to 30m (1.7s) for 2100 cars/lane/hour, if you have human drivers with fully safe spacing doing constant 70 you need nearly 100m (3.2s) between them for a throughput of 1100 cars/lane/hour
However, you don't always have to maintain a distance such that you can stop as if a wall had suddenly replaced the car in front of you, at highway speeds almost nothing short of a head-on collision with someone going the wrong way will actually instantaneously halt the car in front of you, what you want to be able to do is stop faster than it can reasonably stop in case of a collision
Actually, the bolded is wrongentirely correct, I just can't read. Newton's third law states that when two bodies interact, they apply forces to one another that are equal in magnitude and opposite in direction. If a car hits you traveling the opposite direction and has the same speed and mass you do, it is exactly the same as running into a concrete wall because in the concrete wall case, your car is still undergoing a rapid deceleration. Here's the Mythbuster's episode on this:
Ok so some assumptions:
1) Cars are points traveling along a line. This allows us to not consider the fact that a car had a front and back, may be partially through our travel point, etc.
2) Deceleration is constant.
The simple relationship is then:
V + A*ts = 0, where
V is the speed at start of breaking;
ts is the time spent decelerating;
A is the deceleration, which is negatively signed (e.g. -10m/s).
Then;
ts = -V/A; which is the time to stop from first application of the brakes.
The distance equation is:
dt = di + V*ts + A*(ts^2), where;
dt is the total distance traveled;
dr is the distance traveled while reacting;
dr is related to the initial velocity as:
dr = V * tr, where
tr is the reaction time of the driver.
We can then rewrite the distance equation as:
dt = V*tr + V*ts + A*(ts^2)
Throughput is a measure of the frequency of the cars, and is;
T = 1 / (tr + ts); where T is the number of cars that pass through a point per second.
And this is where I realize that throughput is a really, really bad metric. Because to increase throughput, you would decrease the speed that cars travel. But that will result in more time spent on the road, which will decrease the quality of a driving experience. Throughput is probably a good measure for road design consideration (maximizing utilization and road-wear calculations), but not I think for the purposes of our discussions.
Throughput is a good consideration, but your equations only apply if we assume maximum car density. If you're not looking at rush hour conditions on a busy freeway, then your road probably isn't solid with cars.
I imagine you'd want to balance throughput and speed in most normal circumstances. Also, your assumption of the cars being points is only accurate at very high speeds. The effective amount of space a car takes up is the length of the car plus the safe following distance, where the safe following distance is itself a function of velocity. If the length of the car is short compared to the stopping distance, you can ignore it. Otherwise, it's significant.
If you assume that the safe following time is 3 seconds, then the throughput is given as:
Throughput = v/(L + 3v), where
L is the length of the car in meters, and v is the velocity in m/s. If L is around 3 or 4m, it's a pretty big consideration.
Obviously it gets a bit more complicated if the following time is a function of velocity. If we go with the time required to stop with an acceleration of 1g, we get:
T = v / (L + v^2/g)
Here, interestingly, you actually have a maximum throughput at somewhere around 27 mph, if you have an average car length of 4m, and it's about 3 times the throughput you get assuming a constant 3s following distance.
Which... I dunno where I'm going here, I just kinda fell into a math hole and nerded out for a bit.
Where am I? What day is it?
ElJeffe on
I submitted an entry to Lego Ideas, and if 10,000 people support me, it'll be turned into an actual Lego set!If you'd like to see and support my submission, follow this link.
That's the throughput if you want the cars spaced t seconds apart, yes. v*t is then just how far a car travels in t seconds when going at speed v. And L + vt is the distance from front bumper to front bumper.
I submitted an entry to Lego Ideas, and if 10,000 people support me, it'll be turned into an actual Lego set!If you'd like to see and support my submission, follow this link.
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
What if that guy is an EPIC jackass and can eat my whole ass.
How does that affect the math.
I submitted an entry to Lego Ideas, and if 10,000 people support me, it'll be turned into an actual Lego set!If you'd like to see and support my submission, follow this link.
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Wait, aren't you supposed to merge at the last moment though?
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Traffic flow also shows waves and such like a fluid, which is fun. A single slowdown propagates backwards through traffic for ages before it dissipates.
It's not actually that surprising a comparison when you think about how drivers behave and how molecules behave.
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Wait, aren't you supposed to merge at the last moment though?
you're supposed to merge like a zipper
some people believe they are special though and do whatever they want and they can eat my whole ass
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Wait, aren't you supposed to merge at the last moment though?
you're supposed to merge like a zipper
some people believe they are special though and do whatever they want and they can eat my whole ass
Right, but the zipper merge is supposed to be at the end of the merging lane. If someone's going past that yeah they're being an idiot.
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Wait, aren't you supposed to merge at the last moment though?
you're supposed to merge like a zipper
some people believe they are special though and do whatever they want and they can eat my whole ass
Right, but the zipper merge is supposed to be at the end of the merging lane. If someone's going past that yeah they're being an idiot.
No, as soon as the 'lane closed ahead' sign is posted you're supposed to merge out of that lane as soon as it is safe to do so. You're not supposed to wait until the very last minute and merge in when the lane ends.
But I guess this is getting a bit off the point of autonomous cars - I think there is a GST for design of roads thread that might be more appropriate?
Either way, as autonomy increases traffic will get significantly better just because there will be less people being petty assholes. Fuck the guy riding up to the last minute, but also fuck the guy in the open lane who rides the bumper of the car in front of them to block you out.
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Wait, aren't you supposed to merge at the last moment though?
you're supposed to merge like a zipper
some people believe they are special though and do whatever they want and they can eat my whole ass
Right, but the zipper merge is supposed to be at the end of the merging lane. If someone's going past that yeah they're being an idiot.
No, as soon as the 'lane closed ahead' sign is posted you're supposed to merge out of that lane as soon as it is safe to do so. You're not supposed to wait until the very last minute and merge in when the lane ends.
But I guess this is getting a bit off the point of autonomous cars - I think there is a GST for design of roads thread that might be more appropriate?
Either way, as autonomy increases traffic will get significantly better just because there will be less people being petty assholes. Fuck the guy riding up to the last minute, but also fuck the guy in the open lane who rides the bumper of the car in front of them to block you out.
I think there is room to talk here however- one of the promises of AV is that it can remove traffic. The question is - can it remove traffic without global communication, do cars need to talk to nearby cars to coördinate such that traffic is removed, or they all start moving from a green light at the same time. What level of AV penetration is required for this to work? What privacy impacts are there? what social impacts (police telling your car to shut down etc)
In this moment, I am euphoric. Not because of any phony god’s blessing. But because, I am enlightened by my intelligence.
Simply calculating the number of cars that pass through a certain stretch of road is only a small component of traffic congestion.
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
Wait, aren't you supposed to merge at the last moment though?
you're supposed to merge like a zipper
some people believe they are special though and do whatever they want and they can eat my whole ass
Right, but the zipper merge is supposed to be at the end of the merging lane. If someone's going past that yeah they're being an idiot.
No, as soon as the 'lane closed ahead' sign is posted you're supposed to merge out of that lane as soon as it is safe to do so. You're not supposed to wait until the very last minute and merge in when the lane ends.
But I guess this is getting a bit off the point of autonomous cars - I think there is a GST for design of roads thread that might be more appropriate?
Either way, as autonomy increases traffic will get significantly better just because there will be less people being petty assholes. Fuck the guy riding up to the last minute, but also fuck the guy in the open lane who rides the bumper of the car in front of them to block you out.
I think there is room to talk here however- one of the promises of AV is that it can remove traffic. The question is - can it remove traffic without global communication, do cars need to talk to nearby cars to coördinate such that traffic is removed, or they all start moving from a green light at the same time. What level of AV penetration is required for this to work? What privacy impacts are there? what social impacts (police telling your car to shut down etc)
Oh fuck no.
It's a good question on the inter-car communication. There's clearly gains to be had, but also risks, and the gains seem to require a large degree of penetration to materialize but the risks don't.
Posts
I'm certainly willing to entertain evidence that shows they actually do help - gut instincts shouldn't limit research, because guts are often dumb and wrong. But for some of the reasons cited, it just sounds reeeeeal sketchy, and more motivated by sounding cool than actually improving safety.
it waa probably faulty
get a few people locked away from their cars unnecessarily and your product isn't going to last
edit: wouldn't that also kill the idea of a DD?
I worked for one for awhile. Of course it was run by two idiot brothers who weren't paying taxes and literally fled the city without letting any of the employees know (or even moving out of their office and shutting down the phone line).
tbh if we were serious about cutting down on drunk driving everywhere large enough would use DUI fines towards subsidizing these types of services. At least here the vast majority of drunk people who drive home after the bar do so because they are they need their car in the morning for work. If they can get a ride home, with their car, for say $20-$30+tip that's the route they tend to go.
Jesus, I'm gonna trim this tree.
Anyway, I am hesitant to say what can or can't be done by technology, since a lot of the stuff we have nowadays is basically Star Trek level tech. Shit, you change Alexa's name to "Computer" and you're basically on the bridge.
though if I could get a fist full of datas driving my car I might come around on the idea
-On one hand AI is pretty shit at seeing things and that has been a big factor in allowing it to operate. Granted, it's not like this is being limited to just cyclists and pedestrians. The bigger question is how do we prevent this from becoming another thing that fucks over the poor and also individuals with mental impairment. Costs can become a huge issue for the poor. Then even if you make the stuff free, people with mental impairments are more likely to not have those equip do to forgetting. I mean we already issues with poverty being criminalized and I can see how this could easily make that worse, while also hitting certain individuals with disabilities as well, that might not be poor.
-On the other, I'm familiar enough with physics and how shitty a number of drivers are as well. Doesn't matter who fucked up on the road, if you're a pedestrian or a cyclist, you aren't going to be a winner in any collision that happen with you be one of the objects being hit. Hell, if with just assisted stuff, those two groups with transponder would reduce a number of accidents. I've seen plenty of people in the Hampton Roads area that have run out into the street when it's dark and the aren't that visible because even though they have a reflective vest, they are dressed in dark clothing. Even worse because early in the morning you can get fog, while it's still dark, that cuts visibility down quite a bit. A transponder can give the driver a better heads up in a number of scenarios where that could make a difference. Standard caveat of people having to wear it and not do something really fucking stupid, like jump out in front of car that is is moving 40+ MPH.
battletag: Millin#1360
Nice chart to figure out how honest a news source is.
See, I view this as a symptom of us mixing car and pedestrian traffic more than we need to. It is inappropriate and dangerous to combine 40+ MPH traffic with pedestrians, and adding transponders doesn't really fix that.
If Star Trek taught me anything
It’s that you never develop holodeck technology.
I'm kind of surprised re: the CA thing. I'd imagine every manufacturer would be behaving similarly, e.g. standardizing off the strictest common regulatory denominator, and if anything, that luxury manufacturers would bother when others don't.
I've got a 2020 Honda CR-V and it's got ACC and LKAS and will run both simultaneously, at any speed, but gives you haptic feedback through the wheel and the UI barks at you if you're not pantomiming the wheel movements with your hands, with increasing severity.
It will take curves but not turns, it's Android auto compatible but I don't think LKAS will respond to turn way points (although I don't use Android Auto and just run the map on my phone, and I've never tried).
The CR-V ACC is the best, it’s like masturbating on a pile of money for long trips.
https://gizmodo.com/infrastructure-bills-drunk-driving-tech-mandate-leaves-1848026588
It also goes into how attention tracking tech apparently biases against those with disabilities
Not seeing details on the bias against those with disabilities.
I did see something about how this will probably mess with the practical allowable blood alcohol level for driving, though. Having a glass of wine with dinner then driving home is actually legal, but there's no particular requirement for a car equipped with anti-drunk-driving technology to allow this. While I'm willing to listen to arguments about how the allowable blood alcohol level should be changed, maybe this should be done as an actual law instead of being put in the hands of a bunch of random engineers? (And having a different allowable blood alcohol level depending on the brand of car you drive seems a bit yikes.)
If handled intelligently, it could be a reasonably useful tech that just adds an extra level of safety. Attach it as an app on smart phones and an optional thing you stick on your bike as a failsafe. If the car sees a person or bike, don't hit it. If the car sees a transponder signal in a place it didn't find a car or pedestrian, don't hit that, either. In low visibility conditions, where a car (or a human) might have trouble seeing other folks, it's just an added precaution.
Making it mandatory is dumb, and making car AI that doesn't function without it is even dumber, but the idea itself has merit.
bender-serious.gif
I drive a BMW i3, their electric vehicle, if that adds perspective at all.
the "no true scotch man" fallacy.
Beats physically gluing yourself to the road surface, as is the current trend
Pedestrians are already treated as second class citizens intruding on the space and time of the important people in cars. It's already the default assumption that pedestrians are responsible for their own safety and the people in the multi-tons death machines should not have to care about anyone else.
So no. Instead of that, spend money to make streets safe, and reduce the number of cars by making drivers pay for their infrastructure and by preventing car centric development.
We should not force people outside of cars to do even more for the benefit of people inside cars.
However, that comes with the major caveat that I'm not opposed to laws that improve safety even if they may require a person to do things they might otherwise be unwilling to do. There's a long list of things - seatbelt / helmet laws, restricted access freeways, side marker / reflector laws, etc that people have to comply with to utilize public infrastructure.
By regulation (Consumer Product Safety Commission) almost all bikes sold in the US for must be equipped with pedal, front / back, and spoke reflectors. Some cities / states have additional laws that forbid / penalize removal of those reflectors or add additional requirements to bikes like lights (when operated on public roadways at night). Some cities / states have bicycle helmet laws as well.
So just like reflectors / helmets, I'm not opposed to requiring manufacturers to equip bicycles sold to the public with beacons that improve rider safety. I'm not opposed to a program to provide retrofit costs at no cost for existing bicycles. I also wouldn't be opposed to considering the removal or destruction of a beacon on a bike that was equipped with one to be a potentially mitigating circumstance in a crash just like a bicycle where the reflectors were removed. When I go running at night I carry a light for my own safety, and if I were running in the road because there weren't sidewalks I'd be happy to carry a beacon as well for additional safety.
However, given the number of bicycles on the road that aren't equipped with beacons and the potential risk of failure of a beacon this should be considered a supplemental safety feature and autonomous vehicles should be designed with the baseline assumption that cyclists / pedestrians are not wearing / carrying a beacon.
Actually, the speed doesn't matter. If you require 3 seconds of driving time between cars, then the maximum number of cars per hour you can have is 1200 (3600 seconds per hour times 1 car per 3 seconds = 1200 cars per hour). It doesn't matter if the cars are going 5 mph or 500 mph, the throughput will be the same.
The 3 seconds is velocity dependent. Assuming your car brakes at ~1g (Feral or somebody up-thread said that's the average braking rate for cars) and you have a reaction time of ~ 1/2s between needing to stop and applying the brakes, 3 seconds of driving distance between cars gives you time to come to a stop from ~56mph. If you're going 5mph you need 0.7 seconds to stop (assuming a 1/2 second reaction time, again). If you're going 500mph you need 24 seconds.
Heffling might be right though? Thinking it over, if all the cars are moving slowly, they are bunched up, and you get one car past a reference point every 3 seconds. If they’re going fast, they’re much more spaced out, so your throughput relative to road space has to decrease. Regardless of speed you move the same number of cars through if you strictly adhere to 3s following distance.
The 3 second rule is a good idea for human drivers but may not be strictly necessary for AVs. There’s a lot of concepts of cars grouping together in tight convoys for longer distance travel. In that case, the lead car would need a long follow distance, but the rest of the vehicles in the convoy could be much closer together, leading to more efficient use of space. All the lead car has to do is communicate “we’re stopping now” and the individual cars can react without needing the processing time to see brake lights and push a pedal.
You can't give someone a pirate ship in one game, and then take it back in the next game. It's rude.
What an AV can do here is not get frustrated and switch lanes and cause actual stoppages, if everybody is doing 40 mph and has zero reaction time then you can have something like a 20m spacing (1.1s spacing) and your throughput is ~3200 cars/lane/hour. With human drivers you want closer to 30m (1.7s) for 2100 cars/lane/hour, if you have human drivers with fully safe spacing doing constant 70 you need nearly 100m (3.2s) between them for a throughput of 1100 cars/lane/hour
However, you don't always have to maintain a distance such that you can stop as if a wall had suddenly replaced the car in front of you, at highway speeds almost nothing short of a head-on collision with someone going the wrong way will actually instantaneously halt the car in front of you, what you want to be able to do is stop faster than it can reasonably stop in case of a collision
1) Cars are points traveling along a line. This allows us to not consider the fact that a car had a front and back, may be partially through our travel point, etc.
2) Deceleration is constant.
The simple relationship is then:
V + A*ts = 0, where
V is the speed at start of breaking;
ts is the time spent decelerating;
A is the deceleration, which is negatively signed (e.g. -10m/s).
Then;
ts = -V/A; which is the time to stop from first application of the brakes.
The distance equation is:
dt = di + V*ts + A*(ts^2), where;
dt is the total distance traveled;
dr is the distance traveled while reacting;
dr is related to the initial velocity as:
dr = V * tr, where
tr is the reaction time of the driver.
We can then rewrite the distance equation as:
dt = V*tr + V*ts + A*(ts^2)
Throughput is a measure of the frequency of the cars, and is;
T = 1 / (tr + ts); where T is the number of cars that pass through a point per second.
And this is where I realize that throughput is a really, really bad metric. Because to increase throughput, you would decrease the speed that cars travel. But that will result in more time spent on the road, which will decrease the quality of a driving experience. Throughput is probably a good measure for road design consideration (maximizing utilization and road-wear calculations), but not I think for the purposes of our discussions.
Actually, the bolded is wrong entirely correct, I just can't read. Newton's third law states that when two bodies interact, they apply forces to one another that are equal in magnitude and opposite in direction. If a car hits you traveling the opposite direction and has the same speed and mass you do, it is exactly the same as running into a concrete wall because in the concrete wall case, your car is still undergoing a rapid deceleration. Here's the Mythbuster's episode on this:
*edit* Fixed
Throughput is a good consideration, but your equations only apply if we assume maximum car density. If you're not looking at rush hour conditions on a busy freeway, then your road probably isn't solid with cars.
I imagine you'd want to balance throughput and speed in most normal circumstances. Also, your assumption of the cars being points is only accurate at very high speeds. The effective amount of space a car takes up is the length of the car plus the safe following distance, where the safe following distance is itself a function of velocity. If the length of the car is short compared to the stopping distance, you can ignore it. Otherwise, it's significant.
If you assume that the safe following time is 3 seconds, then the throughput is given as:
Throughput = v/(L + 3v), where
L is the length of the car in meters, and v is the velocity in m/s. If L is around 3 or 4m, it's a pretty big consideration.
Obviously it gets a bit more complicated if the following time is a function of velocity. If we go with the time required to stop with an acceleration of 1g, we get:
T = v / (L + v^2/g)
Here, interestingly, you actually have a maximum throughput at somewhere around 27 mph, if you have an average car length of 4m, and it's about 3 times the throughput you get assuming a constant 3s following distance.
Which... I dunno where I'm going here, I just kinda fell into a math hole and nerded out for a bit.
Where am I? What day is it?
Cars and drivers can be modeled as a viscous fluid (somewhat) because if a driver's preferred route is congested some drivers will choose other less congested routes. Road / traffic control design is a system and there needs to be considerations both for average travel time (which does generally decrease as average velocity increases) and for knock-on effects like people choosing to drive / not drive based on levels of congestion. All other things being the same, the faster drivers (safely) drive the sooner their vehicle is off the road and not contributing to congestion.
Since its unusual for traffic to be both free flowing and also fully utilizing the full capacity of the road, the reduction in accidents that block lanes / cause people to stop, braking (which leads to chain effect traffic jams) and allowing smooth merges mean that even though the theoretical maximum capacity of the road may be decreased, a traffic system in many / most cases would have more capacity if people increased following distances significantly.
Traffic engineering is very interesting, and creates a lot of perverse incentives - things like instinctually blocking out the asshole who runs up the shoulder and waits until the last second to merge does far more harm to the flow of traffic than that jackass simply being allowed to merge in.
What if that guy is an EPIC jackass and can eat my whole ass.
How does that affect the math.
Wait, aren't you supposed to merge at the last moment though?
3DS: 0473-8507-2652
Switch: SW-5185-4991-5118
PSN: AbEntropy
Traffic flow also shows waves and such like a fluid, which is fun. A single slowdown propagates backwards through traffic for ages before it dissipates.
It's not actually that surprising a comparison when you think about how drivers behave and how molecules behave.
you're supposed to merge like a zipper
some people believe they are special though and do whatever they want and they can eat my whole ass
Right, but the zipper merge is supposed to be at the end of the merging lane. If someone's going past that yeah they're being an idiot.
3DS: 0473-8507-2652
Switch: SW-5185-4991-5118
PSN: AbEntropy
No, as soon as the 'lane closed ahead' sign is posted you're supposed to merge out of that lane as soon as it is safe to do so. You're not supposed to wait until the very last minute and merge in when the lane ends.
But I guess this is getting a bit off the point of autonomous cars - I think there is a GST for design of roads thread that might be more appropriate?
Either way, as autonomy increases traffic will get significantly better just because there will be less people being petty assholes. Fuck the guy riding up to the last minute, but also fuck the guy in the open lane who rides the bumper of the car in front of them to block you out.
We have that thread - https://forums.penny-arcade.com/discussion/240321/a-gst-on-the-design-of-our-roads#latest
I think there is room to talk here however- one of the promises of AV is that it can remove traffic. The question is - can it remove traffic without global communication, do cars need to talk to nearby cars to coördinate such that traffic is removed, or they all start moving from a green light at the same time. What level of AV penetration is required for this to work? What privacy impacts are there? what social impacts (police telling your car to shut down etc)
Oh fuck no.
It's a good question on the inter-car communication. There's clearly gains to be had, but also risks, and the gains seem to require a large degree of penetration to materialize but the risks don't.
3DS: 0473-8507-2652
Switch: SW-5185-4991-5118
PSN: AbEntropy