Braking distance (without including any decision time) for a 15mph car is 11 ft, for a 30mph is 45 ft. Self driving cars won't change these limits. (well, they may be a little better than humans at maximizing braking power through threshold braking on all 4 wheels, but it won't be dramatically different)
So even with perfect reaction times, it will still be possible for a self-driving car to hit a human who enters its path unexpectedly.
And then he started me on watching for suspicious gaps in the parked cards along the side that could indicate a loading bay or a driveway or an alley or a hidden intersection. And so on though multiple categories of collision hazards, and then verbally indicating them to him while I was driving.
And the reason for that exercise was to drive home the point that if there's a vehicle or a person that could get into my lane, it's my job as a defensive driver to be aware of that and be ready to react. Which includes making sure I could stop or avoid in time if I needed to.
I don't know how driving is taught now, but I would hope a self-driving system could at the very least match what my human driving instructor was capable of.
In fact, self-driving cars may actually improve the situation if cars actually start complying with speed limits en masse.
There's a 30mph city street near me where cars routinely go 45mph -- the fatality rate jumps up to 60% at that speed. So just having cars follow the speed limit would go a long way toward reducing fatalities.
(IMHO) jamitons would dissolve if people would leave a flexible buffer between them and the car in front of them and focusing on minimizing braking, rather than driving up to their bumper, brake, wait for moving, accelerate, brake, repeat. The lag due to reaction time and ac/deceleration exacerbates the "viscosity" of traffic flow. If most drivers focused on "staying fluid" rather than hurry-up-and-wait, traffic ought to improve. Like fluidized beds.
No one is saying reduce speeds everywhere. But in an urban context with lots of pedestrians, these speeds matter, and urban traffic is generally so stop-and-go and congested that drivers rarely sustain the top speed, and reducing it doesn't actually affect travel time by all that much.
However comforting that logic may be, it is unusable in the real world. If you value lives infinitely, then you will never ever do anything that risks your or somebody else's life in order to gain on any other need. You routinely engage in things that are not the safest possible option in order to fulfill other needs, ranging from food or water acquisition through mere entertainment. Therefore you place a finite value on your life. Don't feel bad, so does everybody else. It is possible to determine the value placed on life, I believe there are studies that show the value is more stable than you might think, and balance things appropriately.
It may be uncomfortable thinking, but, again, unless you literally never take even the smallest risk in the pursuit of other goals, you are already thinking this way. You just haven't lifted it up to the conscious level yet.
I'd say you're almost right about nobody asking for infinite value, but I'd say it's more like nobody who has pulled this up to the conscious level is asking for infinite value. People who have not examined the belief are quite prone to speaking as if life's value is infinite... but their own actions inevitably belie that claim. Once examined, it becomes rationally obvious that life is not infinitely valuable (including your own), but, well, if humans automatically accepted and believed all rational things they examine without emotional consequence the world would be a very different place.
http://www.nytimes.com/2011/02/17/business/economy/17regulat...
For evaluating safety regulations relative to the cost of inplementation, NHTSA values the risk of loss of human life in relation to the market value of risk reducing products, and the safety they provide.
They extrapolate from the take rate of airbags and their cost and effectivity, to how much value the average American places on their own life. IIRC on the order of $5 million.
FYI, juries do not look kindly on companies that implement this in liability suits. The data is that if a corporation writes down a $-figure for a (statistical) human life, it anchors punitive fines at a higher level.
Universally adhered to lower speed limits in urban environments would be great.
Conduent offers a turnkey solution for this. They provide and manage speed cameras: https://www.conduent.com/solution/transportation-solutions/r...
Most of them are empty but people unfamliar with the place will usually slow down.
But what about an even lower speed? If 15 mph is good, then 5 mph is better. And if 5 mph is better, 1 mph is superior once more.
I think we can agree there is a point where slow becomes too slow and the 'sufficient planning' becomes an unreasonable burden. So given we aren't operating off the notion that slower is inherently better, then there is some equation giving us our optimal point. What if that point is 45 mph instead of 15 mph?
In short, how do we argue that 15 mph is better than 45 mph that can't also be applied to speeds lower than 15 mph?
There is a fundamental inequity between the operator of dangerous equipment comparing hours and years of life, and the pedestrian who suffers the consequence of that comparison. The USA auto industry is built on this inequity, which is why no one ever talks about it.
Lives, dollars, wasted time are all fungible.
Your new, inequity point adds externalities to the discussion. That's fine, but those are also measured in lives, dollars, wasted time.
But I shouldn't pick nits; you were speaking in generalities! So when I said "any transportation goal" and you disagreed, you hadn't actually thought of a particular exception to my universal statement. And you still haven't thought of one. Are your contributions to this discussion offered in good faith?
2. Auto plant needs radiators. Truck delivering them takes longer to get them there. Costs $x more. Car costs more. Car buyer has to work longer to buy car. Car buyer wastes y hours of life.
There are lots of alternatives that don't involve you spending more time driving.
While it's true that goods will cost more to transport if it takes longer, that higher charge is amortized across many products in the truck, so is a very small portion of the finished product.
So if a radiator fits in a box 16x24x6" or 1.3ft^3 and a 40ft truck holds 2400 ft^3 (subtract 20% since it won't be a perfect fix, so 2000 ft^3, so you can fit 1500 of them in the truck.
If a truck+driver costs $100/hour, that means each radiator will cost 13 cents more.
Or, another way at looking at it -- all of the parts that make up a car aren't going to be bigger than a car (sure, some space is lost to packaging, but there's a lot of empty space in a car), and 6 - 10 cars can fit on a car carrier truck, so each car will end up costing around $25 more.
Though since we're talking about urban speed limits, and there aren't many urban car manufacturers, slow urban speed limits won't affect the price of cars.
Often speeding just gets you to the next intersection, or red lighter, quicker, where you then have to wait longer for traffic, or a green light.
(edit: just to be clear, I'm not in support of speeding, however I do value good logic.)
The vast majority of people just go however fast they feel comfortable (considering conditions, etc) regardless of the speed limit.
Mixed traffic speeds decrease safety.
Raising speed limits so that you don't have the % of people who comply with the letter of the law traveling slower than the people who go however fast they're comfortable usually improves safety.
Unless your goal is to increase ticket revenue or appease the "think of the children crowd" there's no point to lowering speed limits. It doesn't do much to affect traffic speed. To do that you have to modify the road or do something to change the traffic flow.
Self driving cars will improve safety because they'll result in political pressure to raise speed limits to match reality and they'll make dynamic speed limits more practical.
Sure, but city streets are already mixed traffic. There are pedestrians, bikes, vehicles parking or turning, etc. It's not reasonable to raise the limit to what people want to drive and just ignore all the other users of the street.
Also, the optical narrowing mentioned in a sibling comment is quite effective. They've done that on a few streets near me via things like sidewalk bulb-outs at intersections, and swapping the parking lane & bike lane (so it goes curb-bike-parking-drive, rather than curb-parking-bike-drive). Everyone drives more slowly on those streets now - myself included.
For more, see https://www.vox.com/the-big-idea/2016/11/30/13784520/roads-d...
The trick is to make people feel like driving fast is unsafe, without actually making driving any unsafer.
Exactly. That is why we should ban all cars anywhere there is people, and they should be limited to highways. Because safety is most important, right? Right???
It’s very simple, put speed cameras on every corner, and fine in terms of day wages. E.g., one week of your income as fine for going x% above.
Several countries are doing parts of this already, or moving towards it.
Apologies for going off topic here, but I'm curious about this. I've tested every car I've ever owned and all of the recent cars with all-round disc brakes have outperformed this statistic, but I've never been able to get agreement from other people (unless I demonstrate it to them in person).
I'm talking about optimal conditions here, wet roads would change things obviously but each of these cars was able to stop within it's own car length (around 15 feet) from 30mph, simply by stamping on the brake pedal with maximum force, triggering the ABS until the car stops:
2001 Nissan Primera SE
2003 BMW 325i Touring (E46)
2007 Peugeot 307 1.6 S
2011 Ford S-Max
I can't work out how any modern car, even in the wet, could need 45 feet to stop. In case it's not obvious, this is only considering mechanical stopping distance, human reaction time (or indeed computer reaction time which is the main topic here) would extend this distance, but the usual 45 feet from 30mph statistic doesn't include reaction time either.
http://www.government-fleet.com/content/driver-care-know-you...
http://www.brake.org.uk/facts-resources/15-facts/1255-speed
Car and Driver did a test with sports cars and profesional drivers and came up with 142 - 155 ft from 70mph, while my first reference quotes 245 ft (around 40% less, so extrapolating, their stopping distance from 35mph would be around 20 feet).
https://www.caranddriver.com/features/rocket-sleds-the-best-...
The average car on the road is not a sports car with performance tires and is not stopping on a clean, dry track. So I don't think it's a stretch to assume that an average car on average roads with tires optimized for tread life would be 40% worse than a $100K sports car with $400 tires that are optimized for grip rather than lifetime.
The 45 feet from 30mph is a common figure (the UK government's Highway Code uses it as well).
The cars I tested are normal cars, but I concede I had good tyres and I always choose smooth roads to test on. Once you include:
1. Driver ability
2. Vehicle quality
3. Road quality
4. Prevailing conditions
5. 4 passengers plus luggage
I guess you can explain the difference.
I learned about this feature when reading car forums for my car and finding threads from people who were rear-ended when they accidentally triggered this feature by slamming on the brakes when they didn't intend to come to a complete stop.
Which is utterly stupid. Braking is the natural reflex, but not always the right one. I've been in more than one close call (think left turn on incoming traffic) where the correct response was not to floor the brake, but floor the gas.
> people being taught to pump brakes, which increases braking distance in cars with ABS.
Which is a mechanical turk version of what the ABS is doing under the hood.
How did you measure that? Because plugging these figures into a uniform acceleration calculator, 50km/h to 0 in 4.47m requires a deceleration of 2.2g but "Analysis of emergency braking of a vehicle"[0] experimentally measured very best case deceleration as barely scraping 1g (with ABS at 80km/h, significantly lower at lower speeds or without ABS).
[0] https://www.tandfonline.com/doi/pdf/10.1080/16484142.2007.96...
But they should be willing to walk in front of it in an in-spec performance regime. There's some really good Volvo commercials along that line, with engineers standing in front of a truck.
If a car can't observe any potential hazards that might impact it using different threat models it should drive more slowly. Blowing down a narrow street with parked cars on both sides at precisely the speed limit is not a good plan.
Cameras are not accurate enough though, their dynamic range being terrible. Wonder how humans would fare if forced to wear goggles that approximated a lidar sensors information.
Theoretically self-drivers will always see everything that is relevant, unlike a human driver. And theoretically a robot-driver will always react more quickly than even a hyper-attentive human driver, who has to move meat in order to apply the brake.
JFYI, it is most probably apocryphal: https://skeptics.stackexchange.com/questions/18558/were-roma...
still it's a nice one, I have heard the same about engineers/architects in ancient Babylon and Egypt.
Hammurabi's code:
> If a builder builds a house for a man and does not make its construction firm, and the house which he has built collapses and causes the death of the owner of the house, that builder shall be put to death.
> If it causes the death of the son of the owner of the house, they shall put to death a son of that builder.
https://www.fs.blog/2017/11/hammurabis-code/
Heard about it on Econtalk's recent episode with Nassim Taleb.
http://journaltimes.com/news/local/confronting-a-legacy-of-a...
"In order to gain the trust of the employees he asked his wife to enter the building while the movement was taking place."
It is not possible to extract more punishment on an individual than their death, technically, unless you believe in an afterlife.
It is not punitive and meant to 'correct' the builder; it is meant to 'prevent' the builder from cutting corners.
Not true at all. Not that I'd recommend it, but there's torture, and punishing family. They were far less outrageous in Ancient times.
Although the actual logistics of your proposal might be challenging (child comments point out that some speeds/distances might be impossible to solve) your instinct is a correct one: the people designing and deploying these solutions need to have skin in the game.
I don't think truly autonomous cars are possible to deploy safely with our current level of technology but if I did ... I would want to see their family and their children driving in, and walking around, these cars before we consider wide adoption.
From an ethical standpoint the interesting phase will only start then. It‘s one thing to bring a small fleet of high tech (e.g. having LIDAR) vehicles to the road. It‘s another to bring that technology to a saturated mass market which is primarily cost driven. Yes, I assume self-driving cars will eventually compete with human driven ones.
Will we, as a society, accept some increase in traffic fatalities in return for considerable savings that self-driving cars will bring?
Will you or me as an individual accept a slightly higher risk in exchange for tangible time savings?
I believe the claim is 38 multitudes better than humans, significantly better than marginally.
> accept some increase in traffic fatalities
No. And the question is more about "some" than "some increase"
> accept a slightly higher risk in exchange for tangible time savings?
Texting while driving and even hands free talking were becoming laws in many states before smart phones -- and my experience is that many people readily accept this risk and the legal risk just to communicate faster. The same can be said for the risk of drunk driving -- it's a risk that thousands of Americans take all of the time.
IOW, setting this up as some kind of quality standard gives unjustified cover ("Hey, our own CEO risked his life to prove the car was safe!") if AVs fail on the open road, because the requirements of open and closed tests are so different.
Some airlines do not allow pilots to manually fly above 3k feet, nor allow first mates to land manually[0].
0: https://www.wired.com/story/boeing-autonomous-plane-autopilo...
There is a (prevalent) happy path where autopilot and autoland can get you the whole illusion "but the plane flies itself" - but the whole system is built around humans catching and handling all and any exceptions.
A USAF C-54 made a complete transatlantic flight (takeoff and landing included) on autopilot back in 1947: https://www.flightglobal.com/pdfarchive/view/1947/1947%20-%2...
Modern Cat IIIc (zero-zero) autopilots handle everything except taxiing and takeoff. And I think the takeoff thing is more political than technical.
Sounds silly when compared against old tech.
Accidents happen, best we can do is try to prevent them.
It's impossible to fully remove the issue. Vehicles still need to react to and reduce accidents, but they will absolutely never eliminate them.
also that testing does happen in the case of every av program i know of. closed obstacle courses with crap that pops out forcing the cars to react. look up gomentum station.
† i did it for you. http://gomentumstation.net/
Just like they take the repair mechanics on the first test flight after a major repair of a big airplane.
o At speed we're pretty ok with cars hitting things people drop on the road, examples of cars hitting wagons and babies are already plentiful
o Visual recognition & rapidly updated multi-mode sensor data, backed by hard-core neural networks and multi-'brained' ML solutions, have every reason to be way better at this job than we are given sufficient time... those models will be working with aggregate crash data of every accident real and simulated experienced by their systems and highly accurate mathematical models custom made to chose the least-poor solution to dilemmas human drivers fail routinely
o AIs have multiple vectors of possible improvement over human drivers in stressed situations. Assuming they will bluntly stop ignores how vital relative motion is to safe traffic, not to mention the car-to-car communication options available given a significant number of automated cars on the road -- arguably human drivers can never be as smooth as a fleet of automated cars, and "slamming the brakes" is how people handle uncertainty already
Presently the tech is immature, but the potential for automated transport systems lies well beyond the realms of human capabilities.
Self driving tech is up against the marketing arm of the auto-industry, they've got to be better than perfect to avoid public backlash. If they're a bit slower but far safer then I think they'll do well.
I doubt that's true, but even if it were, I believe the rules of the road are "always yield to pedestrians, even if you have the right-of-way".