|
MrYenko posted:Why does a construction zone with 70mph construction-area speed limit signs and no lane closures cause people in Miami to brake down to 39mph in the left lane? Sawgrass? Because it's a terrible road.
|
#
?
Nov 8, 2017 06:37
|
|
|
|
| # ? May 21, 2024 05:50 |
|
|
Volmarias posted:This has probably been covered before, but as my now one lane of traffic crawled along a road with cones up for 2 miles before coming to about 100 feet of actual roadwork, then another mile before the cones stopped and I could rage-pass everyone on the right: Having crews change the lanes all the time puts people into exactly the sort of traffic danger that they are trying to prevent. Extra crew may have to come in to modify the site since current standards seem to require two or three barrier trucks anytime anyone does anything in open lanes. Just as much traffic calming psychology goes into temporary traffic management as goes into permanent roads. If the roadworks don't narrow the lanes, people just go at unsafe speeds. And you have that temp traffic has exploded into an industry in it's own right in the last decade and takes in a lot of warm bodies who aren't very good at the job. A lot of stuff happens because of generic traffic management plans and lazy or poorly trained staff. So sometimes common sense isn't applied and things go way overboard. And finally, as someone pointed out to me in a recent induction, cones are absolutely worthless as a protection device. If a car intrudes into a worksite (or a machine or person strays out of it), then the cone will do absolutely nothing to protect the workers. So if stronger protection isn't available, you don't want the workers near the areas where dopey or distracted drivers encounter the roadworks.
|
|
#
?
Nov 8, 2017 07:27
|
|
|
Thanks for the answers, everyone!
|
|
#
?
Nov 8, 2017 17:36
|
|
|
Varance posted:Sawgrass? Because it's a terrible road. I-75 down near the 826. It gets worse on the 826, of course, but the 826 is expected to be terrible in every way.
|
|
#
?
Nov 11, 2017 04:22
|
|
|
Toronto just changed a major downtown street to there's no more through traffic except for streetcars (and cabs at night so drunk people can get home). As I transit this daily, it should be interesting to see what happens. Cops are currently in "education mode" and giving warnings and pamphlets out to those drivers that travel straight at intersections. https://web.toronto.ca/city-government/planning-development/planning-studies-initiatives/king-street-pilot/ The PDF they are distributing (at above link) has a hugely wide image, so here it is split below:      
|
|
#
?
Nov 13, 2017 19:10
|
|
|
|
|
#
?
Nov 13, 2017 21:00
|
|
|
^^^^ You might actually be able to fit an extra row of autonomous cars in there as they can drive closer together. Still, a good meme. Melbourne did that to one of the central streets in the CBD and the busiest tram corridor in the world with 9 tram routes running along it. It took awhile for people to learn when it happened in 2011 since the change but it now has wide bikelanes and 'super' tram stops yet a few times a year a car will drive down and get stuck in the elevated tramstops, despite the copious signage that the road is car free. drunkill fucked around with this message at 21:05 on Nov 13, 2017 |
|
#
?
Nov 13, 2017 21:03
|
|
|
drunkill posted:^^^^ You might actually be able to fit an extra row of autonomous cars in there as they can drive closer together. Nope. Human drivers frequently drive way too close for any situation, while the self-driving cars will ideally keep a safe distance at all times - which should average out to about the same space.
|
|
#
?
Nov 13, 2017 21:37
|
|
|
fishmech posted:Nope. Human drivers frequently drive way too close for any situation, while the self-driving cars will ideally keep a safe distance at all times - which should average out to about the same space. If vehicle-to-vehicle communication is part of the autonomous systems, they can stay much closer. The safe gap is for dealing with human reaction times.
|
|
#
?
Nov 13, 2017 21:39
|
|
|
I love how people keep talking about the perfect autonomous car system like we are just 5 years from total roll out and replacement of human drivers. E: like when Sunnyvale City council members decided self driving cars would make a dedicated bus lane on El Camino obsolete. CopperHound fucked around with this message at 22:09 on Nov 13, 2017 |
|
#
?
Nov 13, 2017 22:03
|
|
|
CopperHound posted:I love how people keep talking about the perfect autonomous car system like we are just 5 years from total roll out and replacement of human drivers. The best part will be when fully autonomous cars are finally available and it turns out no one actually wants one.
|
|
#
?
Nov 13, 2017 22:06
|
|
|
Deteriorata posted:If vehicle-to-vehicle communication is part of the autonomous systems, they can stay much closer. The safe gap is for dealing with human reaction times. A vehicle to vehicle communication system doesn't change the fundamental physics of braking and tires. Humans frequently drive much too close for even perfect communication and reaction times to allow for a safe chance at stopping if something suddenly requires the car in front of them to stop - a self-driving car should always stay well back from each other in comparison. I mean you'd have to be really naive to think "the computer can react instantly" means "2 tons of car can stop instantly from cruising speed".
|
|
#
?
Nov 14, 2017 02:32
|
|
|
fishmech posted:A vehicle to vehicle communication system doesn't change the fundamental physics of braking and tires. Humans frequently drive much too close for even perfect communication and reaction times to allow for a safe chance at stopping if something suddenly requires the car in front of them to stop - a self-driving car should always stay well back from each other in comparison. No, that's not what it means. It means that it takes anywhere from 0.5s to 2s or more for your brain to perceive that the stop lights on the car ahead of you are on until you get on the brake. This is called "reaction time." Your car continues to travel at full speed while your brain is figuring out what to do. Part of the safe distance is an allowance for reaction time. That distance can be reduced if reaction time is reduced.  The yellow part is reaction time, which can be mostly eliminated with vehicle-to-vehicle communication. The instant the car in front of you starts braking, your car will, too.
|
|
#
?
Nov 14, 2017 02:49
|
|
|
But there's still going to be latency in any vehicle to vehicle network though, probably on the order of a second at the worst case. I suspect the fastest way would actually be to just use optical sensors to spot the brake lights.  What V2V would actually fix is over braking, but you still need the same following distance.
|
|
#
?
Nov 14, 2017 02:53
|
|
Yes, it's like a lava lamp.
|
Deteriorata posted:No, that's not what it means. It means that it takes anywhere from 0.5s to 2s or more for your brain to perceive that the stop lights on the car ahead of you are on until you get on the brake. This is called "reaction time." Your car continues to travel at full speed while your brain is figuring out what to do. What part, exactly, are you not getting about the fact that humans do not currently leave proper space cushions? And what part aren't you getting that a self-driving car is unable to instantly stop to avoid collisions from driving too close, and so must neccessarily leave itself space to account for things ranging from its own braking abilities to the possibilities that communications are interrupted and it can not safely coordinate with the cars in front of it? Your argument only works if all humans already obeyed the driving laws and obeyed proper following distances. They don't.
|
|
#
?
Nov 14, 2017 04:25
|
|
|
CopperHound posted:I love how people keep talking about the perfect autonomous car system like we are just 5 years from total roll out and replacement of human drivers. And people usually ignore that self-driving cars don't remove most of the problems cars and car-centric development cause, like pollution/inefficient and isolating city planning/huge amounts of space used up on stuff dedicated to cars(highways, parking lots, garages, etc)/major financial burdens on the poor if they want to survive in a car-centric area/etc. Cars are an incredibly inefficient and wasteful way to get around(even though they do have their purposes), and continuing to build things assuming they're the primary(and sometimes only) mode of transportation just propagates that inefficiency and waste. Self-driving cars won't change that.
|
|
#
?
Nov 14, 2017 05:15
|
|
|
Also, given how many cars and trucks and pickups are on the road from the 1980s, it'll be another fifty years to a century before autonomous cars can really take the roads from people-driven cars. (And then you have people who want the option to drive themselves even if they don't do it all the time, kinda like you have people who want to drive manual except more so)
|
|
#
?
Nov 14, 2017 05:46
|
|
|
There will be people driving pre-automation vehicles for as long as they can be kept in operation. Personally, I'd love the option, but unless it becomes available as a retrofit for existing vehicles, it won't be viable for a lot of people.
|
|
#
?
Nov 14, 2017 05:58
|
|
|
Haifisch posted:Seriously. Even if we magically got perfect self-driving cars tomorrow, the legal stuff(who's at fault if a self-driving car hits something? how much attention do the occupants of a self-driving car have to pay to their surroundings?) and practical stuff(how long will it take to reach a critical mass of self-driving cars, given that people aren't going to get rid of their old cars overnight? how will self driving cars prioritize protecting their 'driver' versus protecting other people?) will take much longer to sort out. Most of those are being solved right now, actually. We don't need a self driving car to actually exist in order to predict what regulatory issues need to be addressed. Your whole post is 'the perfect is the enemy of the good'. Most self driving cars will be fleet owned, almost certainly directly by the manufacturer, and insured by the manufacturer. Liability will be handled in the usual insurance way. The upside is that if the self driving car encounters an accident situation where it is at fault, the lesson can be applied to all self driving cars very quickly, eliminating it. SAE International defines different levels of self driving, with Level 1 being none, and Level 5 being complete autonomy. Autonomous car manufacturers provide the rating for their cars when they have self driving features. For instance, the lane assist and automatic braking that is becoming common on cars these days is Level 2: driver remains in control at all times and vehicle provides safety features only. The Waymo taxi service that just launched in that Phoenix suburb is Level 4: within a prescribed area the vehicle is completely autonomous, outside that area the vehicle has no autonomy. Self driving cars will replace taxi services and for hire cars first, then probably short haul shipping, then long haul shipping, then certain types of public transit, then all road based public transit. Eventually, people will realise it's easier to subscribe to a 'for hire' car service than to own a car, and car companies will offer different pricing models based on service level, luxury, and need. EG: Your Chrysler car service will be cheaper than your BMW car service. Road sharing with human drivers is expected to persist for about 20 years, as new personal vehicles become luxury only items that have full self drive but also a 'human driver' mode in certain areas, like race tracks. Electrification should help abate most of the pollution issues, and the efficiency of centralized fleet management, car reuse, predictive routing, and several other 'network effect' based niceties should help optimize road use a good bit. Remember that self driving, electric buses will also be a thing, and probably self driving trams, light rail, and subways are also going to become common. If you want to stop sprawl, however, you need to change how cities, towns, and villages zone and plan residential and commercial areas, because that is what drives sprawl, not the existence of cars and roads. If you have anything that isn't whataboutism, go ahead and post, but if you just want to whine about self driving cars not being perfect go somewhere else.
|
|
#
?
Nov 14, 2017 06:01
|
|
|
Moving to cars-as-a-service instead of ownership is exactly how you make people flee far and fast from the idea.
|
|
#
?
Nov 14, 2017 06:38
|
|
|
Javid posted:Moving to cars-as-a-service instead of ownership is exactly how you make people flee far and fast from the idea. Well it sure doesn't solve the "these are the things I like to keep in my car" problem that's for sure. Or "oh my god the last person using this car service vehicle really farted up a shitstorm in here and every surface is covered in bodily fluids and solids." It's a bad enough issue on the bus or in a cab sometimes.
|
|
#
?
Nov 14, 2017 06:45
|
|
|
Javid posted:Moving to cars-as-a-service instead of ownership is exactly how you make people flee far and fast from the idea. Yeah, it's not like taxi services, limo services, hire cars, car rental companies, or car sharing services aren't hugely successful businesses or anything. I do agree America will probably hold out the longest, though. fishmech posted:Well it sure doesn't solve the "these are the things I like to keep in my car" problem that's for sure. Or "oh my god the last person using this car service vehicle really farted up a shitstorm in here and every surface is covered in bodily fluids and solids." Everything in a hire car will be filmed, so you can report something like that in on your phone before even getting in and a new car will get dispatched and the offending person billed for the cleanup. Different hire car companies will offer different cleaning schedules and guarantees as part of their service levels. I've never felt the urge to keep things in my car outside of required items (insurance, safety kit, etc) that wouldn't apply any more, so I'm not sure what you are thinking off, however if you wanted to make a multi stop trip and keep the same car throughout, or a longer trip or time period with a single car, I;m sure there will be a service for you. Somebody company will identify that market and differentiate themselves into it. It's not going to be a day one thing, but it will happen if people want it.
|
|
#
?
Nov 14, 2017 07:45
|
|
|
Javid posted:Moving to cars-as-a-service instead of ownership is exactly how you make people flee far and fast from the idea. Right, that's a change in mentality that I can imagine will only really take hold in 20-30 years or more. Like the necessity to make manual driving illegal, you'll have enough holdouts to make it politically impossible until a lot of old car owners die, basically. From what I can see, in my country in Europe, the younger generation of drivers are all for these changes, seeing car ownership less as a lifestyle and more as a means of transportation.
|
|
#
?
Nov 14, 2017 12:54
|
|
|
Deteriorata posted:No, that's not what it means. It means that it takes anywhere from 0.5s to 2s or more for your brain to perceive that the stop lights on the car ahead of you are on until you get on the brake. This is called "reaction time." Your car continues to travel at full speed while your brain is figuring out what to do. The reason that a platoon of autonomous vehicles can not only reduce most of that yellow distance, but a lot of that red distance, is that every car in an autonomous platoon does not need to be doing "Is the roadway in front of me safe" checks every moment. For current driver-operated cars, each person has to be able to see enough of the roadway in front of their car to be able to say "There are no debris/obstructions or stopped vehicles here, I can continue driving at my current speed". If you're in a platoon of coordinated autonomous vehicles, the front vehicle is the only one that has to be checking that - the trailing vehicles just have to maintain communication and be prepared to brake on instruction from the lead car. You would still need a little buffer to account for situations like an errant vehicle or deer or crazy person running into the roadway, and allow for some room for the cars to initiate a simultaneous stop, but for access-controlled high speed roadways, a set of close-following autonomous cars does not really have to worry about a stopped car materializing in the center of the pack.
|
|
#
?
Nov 14, 2017 13:12
|
|
|
Devor posted:The reason that a platoon of autonomous vehicles can not only reduce most of that yellow distance, but a lot of that red distance, is that every car in an autonomous platoon does not need to be doing "Is the roadway in front of me safe" checks every moment. This is incorrect. Each car still needs to check for itself against what the platoon is doing, because something might have gone wrong with any vehicle immediately ahead of it at any time. And you can't get rid of most of the "red distance" because there are physical barriers to hoq quickly you can stop 2+ tons of metal and lithium batteries hurtling down the road.
|
|
#
?
Nov 14, 2017 13:22
|
|
|
If your privately-owned Uber auto-taxi joins a Google car-convoy on the publicly closed federal highway, and the fifth car in line blows a tire leading to a fifteen car pileup, who's liable for your health coverage? Until you can answer this kind of question, a lot of these hypothetical efficiency benefits aren't ever going to be possible. Even if you can resolve the myriad legal and contractual issues surrounding this technology, you're still going to run into a practical limit in that 40 self-driving cars are never going to be as efficient, comfortable, or eco-friendly as a bus or metro. Self-driving cars are going to throw a lot of people out of work and marginally reduce the cost of transport, as well as letting more commuters catch up on their social media, that's about it.
Kaal fucked around with this message at 13:30 on Nov 14, 2017 |
|
#
?
Nov 14, 2017 13:26
|
|
|
fishmech posted:This is incorrect. Each car still needs to check for itself against what the platoon is doing, because something might have gone wrong with any vehicle immediately ahead of it at any time. And you can't get rid of most of the "red distance" because there are physical barriers to hoq quickly you can stop 2+ tons of metal and lithium batteries hurtling down the road. Keep in mind that those stopping distances are the distance you need from a non-moving obstruction. If the thing in front of you is also moving at roughly the same pace and braking at roughly the same rate the safe following distance is significantly reduced. Assuming that the lead car in the pack does not actually hit the thing causing it to initiate braking, the safe following distance in a fully computerized system becomes the reaction time of the system plus the difference in braking ability. A light sports car following behind a semi truck could basically sit at the reaction distance and be fine, where obviously the opposite would require a lot of cushion.
|
|
#
?
Nov 14, 2017 16:43
|
|
|
If the cars linked up like a train, wouldn't that help too? Direct electrical connection for communication, and the physical connections and bumpers would help with the braking issue.
|
|
#
?
Nov 14, 2017 16:49
|
|
|
Guy Axlerod posted:If the cars linked up like a train, wouldn't that help too? Direct electrical connection for communication, and the physical connections and bumpers would help with the braking issue. I'd imagine that could work fairly well if all the vehicles had similar braking capabilities, but I get the feeling that if you had a heavier vehicle towards the back and lighter vehicles towards the front you could really easily end up with the "road train" jackknifing. You'd also need heavy connector infrastructure at the front which wouldn't necessarily be good looking or aerodynamic. It might not be a bad idea in specific niches though like semi trucks.
|
|
#
?
Nov 14, 2017 17:21
|
|
|
drunkill posted:Melbourne did that to one of the central streets in the CBD and the busiest tram corridor in the world with 9 tram routes running along it. It took awhile for people to learn when it happened in 2011 since the change but it now has wide bikelanes and 'super' tram stops yet a few times a year a car will drive down and get stuck in the elevated tramstops, despite the copious signage that the road is car free.  Sadly, we didn't have the guts to totally pedestrianise King St - the excuse is a lot of driveways and parkades exit onto it so car access must be maintained. But even the compromise has achieved good results in its first week. I mean, besides the driver who hit a pedestrian on a parallel route and blamed the project...
|
|
#
?
Nov 14, 2017 17:29
|
|
|
In regard to automated transit... Please. People act like animals when it comes to public transit in the States. The only thing you save on is being able to turn vehicles around faster at end of line. You're still going to need a ton of transit supervision to deal with day to day issues, and Big Brother isn't going to be much of a deterrent (but an authority figure in the driver seat is). Rail isn't as bad because you're dead to rights if the situation is noticed between stations. In an emergency, a self driving bus is going to pull over and open all doors, then the offender bolts into the neighborhood. Happens all the time. There's also the high tech learning curve associated with all self driving vehicles. There is a massive shortage of mechanics that can actually repair today's high tech cars properly. In terms of current self-driving transit systems, ATO variants are easier to manage, as you have that centralized control in a closed system. Dispatching automated bus service is going to be a whole different animal. One of these days I should just bite the bullet and start an a/t for transit. Sooner or later. Varance fucked around with this message at 18:32 on Nov 14, 2017 |
|
#
?
Nov 14, 2017 18:18
|
|
|
The fleet of self driving cars as a service model would also suffer from many of the disadvantages of current mass transit and transit-as-a-service systems that make them more suitable to European environments and urban environments. Look at the trouble bus and Uber have breaking into rural US environments. If your population density is really low, you just can’t support transit as a service with reasonable wait times and revenue margins. Instead you get farmers holding onto pickup trucks for 50 years.
|
|
#
?
Nov 14, 2017 19:00
|
|
|
Guy Axlerod posted:If the cars linked up like a train, wouldn't that help too? Direct electrical connection for communication, and the physical connections and bumpers would help with the braking issue. It's hardly a trivial issue to build a common connector that can handle automatic connection in a situation like that. Just think of where you would need to mount those things on the back and front of say sedans, such, and large pickups to ensure they can all hook up even with manual guidance. Trains can be hooked up "automatically" because rail standards ensure each railcar has the connectors at the same height over the rail and that they're all right on the same track where you connect them all up.
|
|
#
?
Nov 14, 2017 19:27
|
|
|
These two ladies vastly improve an intersection by "painting" some turn lanes. https://www.youtube.com/watch?v=0ziqlGgc8_k
|
|
#
?
Nov 14, 2017 20:32
|
|
|
I found a paper on cooperative adaptive cruise control and minimum safe distances. Conclusion: with both cars communicating, then the following distance is about a quarter second (8m at 30m/s). This parameter is dynamically calculated, and results in <.01% chance of a crash when something goes wrong. From our yellow-and-red bars graph, that bottom graph is all yellow bar now, and is 26 feet long. This is because both cars are stopping at the same rate. .01% of the time, there should have been some red bar in there, and the cars impact, usually with a difference <1m/s when that happens. As far as cars-as-a-service, I see the first implementations as being a subscription option. You get a car for a month at a time and you turn it in when you don't think you'll need it. The car can also let you know when it's not going to be available to you because it needs maintenance. The cars can swap themselves out while they get oil changes or whatever, but "your" car will come back to you. At no point do you have to mess with actually owning the thing. Better cars cost more, but they're all still self-driving. This will let the vehicle remain as a status symbol to those for whom that's important. In this way, people who don't want the hassle of car ownership don't have to have it. Those that want to own, maintain, and operate their cars can. Eventually, there won't be any need to actually legislate "no human-operated vehicles on interstates" because social pressure will have moved the owner/operators over to the places where that's still fun. Timescale for something like this is probably 50 years, but could be quicker in places that don't have a strong car culture, but will probably be slower in places where people still enjoy driving. Other thoughts: There doesn't necessarily have to be all that much or particularly great communication. Front-and-rear mounted radar can let the car know when something's going on, even if active communication isn't sent or received. If there's nothing going on in front of you, then you don't have to slow down as rapidly if someone behind you isn't slowing down enough. Alternatively, you can have these systems decide to have a bit of a fender bender; if they're not privately owned then the insurance is basically a wash. If the car behind you isn't slowing down enough and the car in front is at max braking and you're not, then maybe it's a good idea to let the car behind you touch bumpers at <1m/s difference and let your braking help them slow down to prevent everyone from getting squished. Fleets of cars can be altruistic in ways that people probably never will be.
|
|
#
?
Nov 14, 2017 23:48
|
|
|
I mean, self driving on every road will work in a utopia where everyone can afford this stuff. In reality, only a percent can afford this stuff, even on a rental level. That's my biggest gripe. There are a ton of people who can't afford a $200/month car payment. Assuming you can get the cost of a self driver down to $25k, you'll get 5 years/100k miles out of the thing and have to retire it. $416/month for the car alone, assuming that $25k includes overhead, maintenance and development costs, plus your profit margin. You'd have to drive the cost per car down to $12000 to hit that $200/month mark, and probably auction vehicles after a couple of years to turn a profit. At that low of a price point in terms of materials, how good is that car going to be after five years? You're talking an at cost Chevy Spark or Ford Fiesta here, with no options other than self driving and auto trans. Nobody buys that at market, because you're talking a vehicle without air conditioning. You'll see HOV lanes and BRT corridors converted to work with self driving, as that's the kind of audience that can afford the tech, but general traffic won't be automated for a loooong time. Varance fucked around with this message at 03:41 on Nov 15, 2017 |
|
#
?
Nov 15, 2017 03:28
|
|
|
Look at the price of a Zipcar in most major cities - if you want to reserve one for a whole day (i.e. past like 8 hours at a time) it's usually like $60-$80 and if you just want to use one for up to an hour it's usually like $15 and that's a car that's already parked right near you and has the driving handled by you and the maintenence handled mostly by some guy who drives by the lot every week to vacuum and check basic things, and then another employee to drive it over to an auto shop if something major broke. A really efficient self-driving car service that's got the costs of actually buying the cars down probably isn't going to charge that much less of a price for use. Tons of people can't really afford prices like that as an everyday thing.
|
|
#
?
Nov 15, 2017 05:26
|
|
|
Let's put it his way: if transit agencies charged cost to their riders without subsidy, a monthly pass for basic service would be around $500/month, and that's with massive economies of scale (the cost of a bus split by hundreds of people). The math just doesn't work unless the public subsidizes the hell out of it, 2-3x beyond what we already spend for roads and transit combined. Varance fucked around with this message at 08:01 on Nov 15, 2017 |
|
#
?
Nov 15, 2017 07:59
|
|
|
https://www.youtube.com/watch?v=g_ILtWzH3Ko Space/volume efficiency of various modes
|
|
#
?
Nov 15, 2017 20:34
|
|
|
|
| # ? May 21, 2024 05:50 |
|
|
Varance posted:In reality, only a percent can afford this stuff, even on a rental level. How, exactly, does this differ from regular car prices, today? Millions of people in the US are poor, and will never be able to afford a new car, even the cheapo ones like the Chevrolet Spark ($13,875) or Nissan Versa S ($12,855). So they ride a bus to work, or carpool, or buy used cars. I guess I'm not sure where you're going with your argument that autonomous cars will cost a little more than regular cars.
|
|
#
?
Nov 15, 2017 22:44
|
|