If anyone was going to focus on making sure their self-driving car could drive in snow, it'd be the Russians.
Mind you, self-driving when snow has coated all your landmarks is definitely impressive, but the title made me think it'd be driving in snow. Which it isn't. The roads are, what we'd call in Chicago, "clear". The car is not driving in/on snow or ice in this scenario, which is something I do regularly every winter.
Canadian perspective: This is very clear. Much more challenging is something like a multi-lane highway where the road markings have completely disappeared under snow. I'll be impressed when it can drive from Hope into the BC interior in winter in conditions that a human driver can handle with winter tires and possibly studs, but no need for chains.
Oh man I've driven through canada to blue mountain during a blizzard. That was scary. I really have to wonder how computer vision will account for that. Or similarly torrential downpours here in the Tropics.
I don't think CV ever will account for that. I always have thought the answer is going to be somewhere between great technology, and infrastructure to support it. We need "smart" roads.
Ultimately I think it'll have to be smart highways. You could embed a coded marker in the centre of each lane which the car can pick up using RF (e.g. an induction loop). This plus other constraints like GPS, visual-inertial odometry, etc should be enough to localise and remain safely within a lane boundary even if it's invisible. This would work fine through a layer of snow - some of the more expensive ski jackets have built in loops for detection in case of avalanches.
It's hard to see how we could get from where we are right now, infrastructure-wise, to there, at least here in the US. We can barely keep our roads driveable in some places, much less modernized with the latest in smart beacons and such.
It wouldn't necessarily need to be in-road, but there are numerous places where you could adapt. For instance, cats-eyes in the UK (retro-reflective markers) are placed on virtually all motorways at lane boundaries and the car could look for a signature from them (I imagine at night, autonomous cars already use them as a feature for lane boundary detection). Nobody bats an eyelid about re-setting cats-eyes every time a bit of road is replaced. You could also put boundary markers on either side of the road, in the central reservation or something similar.
The idea would to put the smart tech in the car. What's going at the roadside could be made very cheaply and integrated into existing features without needing to rip up the entire road.
Other countries have toll highways which are well maintained, that's another option for installation.
I don't see this happening in rural/country lanes, the cost would be obscene (but again, we do liberally use reflectors in the UK). I also don't see this as a panacea, it would simply be a backup in horrendous weather.
I've seen those cat eyes before, but never in locations where plows have to travel. They are also relatively easy to place, just requiring some tar and a manual "pick & place". Making something plow proof would require embedding it in the road, either with a drill or a "rumble strip" like divot.
Volvo has proposed driving magnetized nails into pavements in Sweden as lane markers. Not only are these useful for self-driving cars, they're good for snowplow guidance.
Uh-oh. In other words, there's a smart road ahead, you can 100% trust it. Except when it's not there, because it hasn't been maintained for decades, and it's unclear where the missing pieces are. Can't see what could possibly go wrong.
Plus there's also no incentive for anyone to waylay the smart road off the physical road, because there are no evil people (i.e. with motives that are incompatible with the vehicle's survival).
It's a hint, like lane lines. You're going to be using GPS and LIDAR as well. Yes, some of the lamer self-driving systems put too much faith in lane lines.
Indeed. Which means that we're back to square one, "it's where most people are, which is exactly the spot where it's least needed; whereas in places where more conventional navigation fails, there's minimal incentive to also build and maintain this."
TL;DR: good roads will be getting better, bad roads won't.
If it can take you 90-95% of the way, and seamlessly pass over control for the non-beaconed start and end stretches, that's still a serious convenience boost over a train/tram.
There's some interesting recent work at (IIRC) MIT using ground-penetrating radar techniques to image subsurface features of the roadbed in real time, as the car drives over them. Rocks, rebar, and such are almost completely weather-independent (short of torrential rains that wash out the road itself), and they don't change significantly over time.
Since off-road use for smart cars is comparatively small, we should optimize the roads to be smart-car friendly.
Perhaps we could install systems which would enable smart cars to stay in lanes, perhaps with some hardware preventing the smart cars from going off-road.
And while we're there, we could also have systems providing charge built right into the road! Wouldn't that be neat?
The lane-keeping and power systems could allow regular cars to operate on the same roads, for backwards compatibility reasons.
Sounds too expensive to build? Not until you realize that you've re-invented the infrastructure for trams and trolleys put in place a 100 years ago.
Computers can provide near "perfect" speed/traction ratios based on active road conditions... which humans are merely estimating based on "feel", without instant direct feedback from the wheel traction sensors, combined with whatever (potentially limited) winter driving experience they have.
Computers can also "remember" any specific location of the route from hundreds of thousand of other rides in the past and of other cars on the road communicating in real time.
In 5yrs I'd expect to prefer putting my trust in any self-driving car on winter roads over a human any day...
Perhaps self driving cars might never support highways in places with inclement conditions to begin with, which is totally fine. I agree with the other commenter about "smart" highways with embedded markers to let the vehicle know its position despite rain/snow/mud.
It's only "fine" if you don't live in places with inclement conditions, and everyone's talking about how we're gonna ban cars that aren't self-driving. ;)
And if there is self-driving zones, and non-self-driving zones, you better believe I still expect my normal car to work if I drive it from Chicago down to San Francisco.
I'd be impressed if I see a self-driving car making a left turn in a crowded junction without a specific left-turn signal at LA. The only way to turn left in such roads is to make an illegal turn after the signal just becomes red. I wonder how the self-driving car would do in such a situation.
In vancouver the general rule in dense traffic seems to be that two cars can left-turn through a stale orange light, right as the light is changing and before the opposing traffic has a chance to start rolling into the intersection from their newly green light.
Sometimes only one car, if the timing works out that way. Three is almost certainly too many. The ability of a car to judge the light timing requires near 180 degree vision and head pivoting on the part of a driver.
Maybe after all cars become autonomous. Earlier there will be a hybrid period where both manned and unmanned cars drive together I think.
Even all cars become autonomous, the problem will still be hard I believe. Hard coded rules may solve the problem but relying on communication and cooperation will not be easy.
And computer-driven pedestrians, too - the car had to stop multiple times for turning left/right across a pedestrian crossing which was green at the time. (If you have watched the video, you have seen that most potentially dangerous situations did not originate from cars.)
Turning after the light turns red isn’t illegal, it is what you are supposed to do to clear the intersection. Otherwise you’d be sitting there blocking traffic.
In California (and most places I've lived), this is actually legal. To complete the turn, you first enter the intersection while the light is green, then wait for traffic to clear. If it only clears after the light turns red, you're still in the right - because only when you fully entered the intersection (on the green light) is considered. See here: https://patch.com/california/sanbruno/ask-a-cop-should-i-pul...
Of course, this only allows one car (or maybe two, if the intersection is particularly large) per light cycle, which isn't much. People who enter the intersection on the yellow or red light (tailing the person who was in the intersection waiting to turn left) are turning illegally.
In my country there is very specific provision along the lines "it is illegal to enter intersection if an obstacle would force a stop in intersection and block traffic" which means that in a "turn left or go straight" lane you enter intersection and cannot turn left due to traffic and block traffic for cars going straight making that technically illegal. Human drivers, of course, do this all the time.
Are you sure that moving, oncoming traffic is seen as an obstacle under that definition?
The real trick against gridlock is to equate the switch to green not as "go", but as "go, once the previous wave has cleared the intersection". The "don't enter before there is room for your car on the far side" is only an optimization on top of that.
I am not a lawyer so I may not be 100% correct. As far as I remember my driving classes and other resources (cannot quickly find exact provision), anything that is not part of road infrastructure (there are intentional obstacles to guide traffic) and causes vehicle to change speed/direction is an obstacle, so moving traffic kind of is. I understand that this particular provision is specifically there to prevent gridlocks and is never enforced in this particular situation.
The way traffic works is somewhat dependent on region. I can go to another city an hour away and already feel a bit alien traffic wise there. This is relevant in discussion on autonomous vehicles: in any foreseeable future we need them to coexist with human drivers and abide by unwritten "everyone drives like that here" rules.
In situations in which my country has that same provision, it does not treat moving traffic that crosses your path as an "obstacle", so if you're turning across traffic, you're allowed to enter the intersection and wait there. It might be worth double-checking whether your country has a similar subtlety.
If you come to LA, you'll see there's no way to avoid it. The majority of lights are unprotected left turn lanes. The cops here even break the laws and no one judges. Sure, everyone could make 3 right turns instead, but I'd imagine if everyone did that the damage from wasted fuel, smog, and extra traffic would outweigh the few accidents that happen due to cheating the left turns.
But CA law also says you can't enter an intersection without having clearance to vacate it. In other words, if the cross street itself is backed up such that its traffic backs up or into that intersection and prevents you clearing the intersection as you complete the turn, you violate the "anti gridlock" law.
That's not relevant to this situation. Waiting for oncoming to traffic to clear, then turning while the light is red is legal, and orthogonal to whether "there is sufficient space on the other side of the intersection". If there isn't sufficient space to turn, whether the light is green, red, or yellow, then you can't turn and are violating the anti gridlock law. If you entered the intersection before it turn red, and there is space to turn, you can turn on green, yellow, or after it turns red.
Most jurisdictions do not consider it blocking the box if you enter the intersection to turn and it turns red after. They consider it blocking the box if you enter with the intention of going straight while it is not clear to do so.
CA Vehicle Code 22526(a): "Notwithstanding any official traffic control signal indication to proceed, a driver of a vehicle shall not enter an intersection or marked crosswalk unless there is sufficient space on the other side of the intersection or marked crosswalk to accommodate the vehicle driven without obstructing the through passage of vehicles from either side."
This does not seem to be relevant to the situation here (waiting for the oncoming traffic to clear): usually, there's plenty of space to turn into, but one has to yield to oncoming traffic before occupying that space.
You are allowed to enter the intersection on yellow in California.
21452. (a) A driver facing a steady circular yellow or yellow arrow signal is, by that signal, warned that the related green movement is ending or that a red indication will be shown immediately thereafter.
Equally impressive would be four self driving cars turning left (in right-hand traffic) at equal priority intersection. At least in my country the rules are muddy in this case and no car has explicit priority. The most general "priority in right" loops around and cars should be stuck there indefinitely :)
I like to imagine to avoid such situations there would some kind of control on the dash that allows a stalemate to be overridden (not the steering wheel :P). It would be great if it were a "give way" button rather than a "I'm going now" button.
In this particular instance yes, that is easy. The core of the problem is that hard rules force traffic into this deadlock. To avoid this deadlock autonomous vehicles must be taught to sometimes bend the rules. Which brings its own can of worms - how do you teach a robot under what specific circumstances rules can be bent.
The rule in California is “first to arrive”, with “yield right” as fallback for simultaneous arrival. Deadlock is unlikely (and a tertiary fallback by compass direction would solve it completely.)
That's assuming a very accurate GPS in very good conditions. You also need to be sure that there aren't any road works that would temporarily change the lines. Seems a bit risky to commit 100% to the GPS for that.
That's easy if you're not, say, in a winding, narrow, deep canyon, such as you might find in a mountain pass. Which is, for related reasons, the same place you'll find the heaviest snow cover. Also for related reasons, "well mapped areas" are around where people live, as opposed to - again - places where you could expect uncleared snow.
"Works great near my house, might accidentally kill you if you try to ride it outside the happy path" is somewhat underwhelming.
Rochester NY is probably the best place in the world for actual snow driving. Daily snowfall during the winter due to lake-effect snow. It's where I trained to drive a rear-wheel drive Camaro. Very useful place.
The difference between what most of New York considers undriveable and Rochester considers "normal Tuesday" is amazing. Three inches on the ground going to get your garbage plate is considered part for the course.
The main challenge here for cars to overcome is that many, like Google's, are heavily based on centimeter-accurate 3D mapping data. (I am not sure where Yandex's sit on this.)
So giant piles of snow often means that all of your reference points on your fancy 3D map of the road are now obscured by piles of snow, and you still have to be able to position the car well enough that you know when to look for stoplights, crosswalks, and the like.
It'd be incredibly complex. You'd not only need to factor in the height of the snow, but how snow plows are likely to push and pile up said snow. The edges of the roads after a snowfall are quite varied in landscape as plows unevenly shove piles of snow off the road.
Yeah, that's nothing, judging by the amount of snow on the parked cars and roofs of houses, there's an inch or two of snow. Maybe "Self Driving car on Moscow streets after a light flurry"
I expect this point of contention is that the snowfall wasn't all that heavy. In Canada we typically don't think of snowfall as being heavy until you're pushing 30cms (12 inches) in a single snowfall event. Indictors suggest that they'd be lucky if they received 5cms in this single event, and have no more than 30cms from all of the snowfall events the city has received.
Or the "heavy snowfall" was immediately followed by a 10° heatwave. Usually that's the only way we get black city streets after a snowfall in the northern US.
People often forget that while Moscow is in Russia - this is not Syberia or something. We have a bit more snow than Berlin usually and a bit lower temperature.
I'd love for it to be true, but it's tough to believe that's real. There's one point where a pedestrian is standing in the snow on the road just to the side of where the car would pass:
https://youtu.be/Bx08yRsR9ow?t=35s
Wouldn't any pedestrian detection decide the person in the road has priority and stop and let them cross? It isn't capable of reading a person's intention to stand there vs. cross or not.
The title claims it supports interactions with pedestrians, and the video does seem to show it pausing in the middle of a turn once for a pedestrian in the road, but that could be easily faked by having the driver still using manual breaks.
From what I can tell it looks like the pedestrian’s vector is heading away from the car (right to left) for a few frames before he pauses. Couldn’t a level 5 model infer intentionality in a direction based on the way the pedestrian was walking?
That is the kind of premature optimization humans make that ends up with crashes. It's automated, there is no fatigue, there is no irrational feeling of losing time or being rushed or pressured by drivers behind over what would be a few seconds safely stopped and of no consequence to the overall travel time.
That pedestrian is already in the street and they momentarily stutter then pass the pedestrian with very little room.
That and the way the steering seems super sensitive to various snow remainders on the side of the street. It's almost as if it's controlling for its position relative to the snow instead of picking a mostly straight trajectory.
In much of Europe, and I would assume Russia as well, that's a pretty typical scenario. I suspect the car is set up to handle the situation much like a native driver.
>That and the way the steering seems super sensitive to various snow remainders on the side of the street.
Yes, I also noticed it, it seems like there are continuous micro (and not so micro) corrections on the steering wheel, with such behaviour the passenger (besides the driver that has some more chances, being attentive to the road, to not get it) is likely subject to get car-sickness.
No, please, we need AI, which will make road a bit less dangerous.
We have such an anecdote: when you cross the road in Russia, you have to look not only left and right, but also up.
I also noticed that it was driving 25... I assumed that was 25 km/hr, which is pretty darned slow (15mph) even on snowy city streets. However, it could be that they're running the car in US mph instead?
I think it's 25km/h but those streets are pretty narrow and filled with pedestrians. Normal speed limits on those streets is 30km/h so driving between 20 and 30 seems right.
It was 25-33km\h, yeah. But you can't drive faster than 40 in this kind of area legaly.
And the top cap for the city is 60 (except for things like https://en.wikipedia.org/wiki/Moscow_Ring_Road, you can also go to 80 when overtaking and many people abuse this rule to go ~80 all the time)
I would consider anyone driving more in a single-lane street lined with cars and pedestrians on both sides insane - robot car or not. This is a city, not a racecourse.
It moves at 17 seconds in, at 44 seconds in, 1 min 51 seconds in, etc. I'm guessing the driver was intentionally holding his hand there with effort, ready to grab control of the wheel if necessary.
The actual speed "limit" on those streets is definitely no higher than 30km/h, regardless of what the signs say. If I drove 60 there I hope I'd lose my license for reckless driving.
technically correct title (the best kind!) - but clickbaity because everyone wants to see a serious test of self-driving on snow-covered, icy roads, ideally with actually blinding, falling snow - at night. This is, if anything, even easier than a non-snowy drive because of the high contrast between the completely clear road and the snow-dusted sidewalks.
The dream that the developers, the media, and many of us share is to change the world by making driving a thing people do not do. Matching a human driver in the most treacherous conditions is a major, and unmet milestone. This is a misleading post.
Have you ever seen someone driving in snow anywhere south of Pennsylvania in the US?
This tech, as-is, could probably save millions in damage on minor or one-car accidents in snowy aftermath. Even if it snows once or twice a year, somehow everyone forgets how to drive safely in the snow and Every.Single.Time there is snow the world just stops.
Unrelated to this video, I was impressed with my Subaru EyeSight performance in a whiteout rainstorm. I could tell what was going on - barely, with the wipers at full tilt - and my auto-cruise worked without a hitch at > 50MPH for several minutes straight. It's definitely hit-or-miss though, since sometimes it gets confused on a bright overcast day at 25MPH.
Why do people think that driving on snow/ice is going to be a challenge?
Compared to recognizing the cars, lights, pedestrians, and their intention, that's pretty straightforward.
Because it's harder to orientate when you can't see the street markings, see the street signs, when your cameras are snowed over. If you tried parking a new BMW after a snow storm and looked out the rear camera you know what I mean. Basically they have to do all they normally do but just the equipment they need to keep cameras and lidars without snow either costs an extra fortune, OR the passengers will have to start their ride by carefully wiping all sensors. If there is heavy snowfall, you'll probably have to clean them during the trip too.
Roads that are usually 2 lanes become 1 lane when snow accumulates enough. People improvise to get where they want anyway - the 2 lanes at the traffic light which are normally one lane for turning and one for straight - is now just 1 lane for both turns and straight because of accumulated snow. The AI would need to observe other drivers (and pray they aren't all AI drivers) because stored map info will not work.
As a driver you have to be able to guess whether something that looks like a 1ft ridge of snow is actually passable drift powder snow, or a frozen barrier of ice that will damage the car or get you stuck if you drive through it.
> but just the equipment they need to keep cameras and lidars without snow either costs an extra fortune
Sorry if I'm being ignorant (where I live, below 20°C is really cold weather, and the lowest it gets is around 10°C, which I don't recall ever seeing), but wouldn't heaters be enough?
I’m not sure what would be required (currently few manufacturers get this right). Heaters take a long time to clear ice when starting if the car has a lot of snow and ice on it.
VW has a good solution for their rear camera which is actually hidden under a hatch (the VW emblem) so it’s protected when unused.
I’m picturing that future autonomous cars will need this type of solution for many sensors, similar to the “flip up headlights” of some 80s cars. Not least because lidars are also very expensive so you may not want them easy to grab from a parked car.
Not under current paradigm "go to car, start it up, [wait 15 minutes for sensors to unfreeze], go." Might be viable if the cars are 1.shared and 2.at high % of utilization (in other words, the car is never frozen).
Of course, in mountainous winter, I've had wiper fluid freeze when it got to the windshield of a warmed up, running, moving car - had to stop and scrape off what was essentially a rubbing alcohol solution: even heating wouldn't help when local temperature goes to -40 (either scale ;)).
Vision is one problem under snow and ice. But the one I think will be tough to overcome is operational.
In normal conditions, your car stops in a predictable way, at predictable speeds. It doesn't slide. It doesn't follow cut-paths in the road. Acceleration is a simple function.
Think that you've adjusted for these parameters? Well, the ground you've driven to has more snow on it then before. Or you're elevation has decreased so the snow:ice ratio has shifted towards the later.
And the above only describes the situation if you are one of the few people on the road. Add in all the other drivers who are acting _strangely_ in terms of performance and psychology. Driving lanes shift depending on where the road is clear, typically centering (bringing sliding cars closer together).
Never mind people FOOLISHLY walking on the road (the sidewalks are covered in snow).
Driving in cities in summer requires a high degree of attention and quick responses to others. But it is just a matter of interpretation and response. Winter driving is an art...
Well, when driving on snow and ice, none of your control inputs map exactly to the motion of the vehicle. Accelerate? Naw, let's yaw. Brake? Maybe tomorrow. Steering? Let's go straight. Going straight? Let's veer you over into the other lane. Figured out how to handle map inputs to road outputs? It heats up/cools down and changes everything. Again.
In the best case, visibility restriction is limited to the sides of the roads, like in this video. In the worst cases, there is barely any indication of where the road is, and signs are also covered in snow. GPS is being blocked by the storm overhead. Forward visibility is down to being measured in tens of feet, from 10 miles only five minutes earlier.
As for the vehicle itself and the electronics: in winter, they will need to function pretty much perfectly at up to -40°. Cars, if left outside, can be covered in layers of over 6" thick of snow and/or ice - and not always just on the top.
While operating on winter roads, they will be exposed to a constant spray of slush and snow, combined with salt and sand. This spray will coat the grills and tails of cars. My vehicle has parking assist and radar assisted cruise control; neither of which work after driving down a road with fresh snow. Even the otherwise protected rear view camera gets coated in muck just from the aerodynamics of the car's tail.
None of this is impossible to overcome, but the typical "it's just an engineering problem" is aggravating to hear.
I'm sorry - this isn't a "heavy snowfall" in Moscow. It's just barely a dusting. That isn't even an inch of accumulation. Being able to self-drive when there is black pavement to be seen isn't impressive in the least. Show me a self-driving car functioning in white-out conditions (which is what Moscow faces on a regular basis, as does a large portion of the US) and I'll be impressed.
Then it should be easy to show a self-driving car working in those conditions. Somehow, despite this argument coming up in virtually every discussion of this kind, nobody's done it yet.
Did they ever release any substantial information in that regard? The video just shows a few cars slowly driving around empty streets with barely any snow.
Fully self driving cars are essentially here, except for needing an alert driver to take over when they steer into oncoming traffic. When you have Tesla drivers who wedge oranges into the steering wheel to trick the car into believing someone is holding it, Tesla are going to want to make their system robust before rolling it out.
The hardest human equivalent in New England in my youth, was a 4 way stop at the peak of a steep hill. Would have loved to seen this scenario with the self driving car.
So this video features a self-driving car developed by Yandex. Does someone have insight into why so many big tech companies seem to all be in agreement that developing self-driving cars is an area they need to be pursuing? Is it mostly everyone looking at what Google is doing and replicating that? Or is it the getting swept up by the Uber hype? Or is working on self-driving car software just such a natural extension to what these tech companies have been doing all along in their main areas of expertise anyway?
(I feel like this question must have been brought up a number of times already, since these projects have been going on for a while by now.)
I live in the neighborhood; these streets are all around Yandex headquarters. It's one of the best districts in the whole city, streets rigorously cleaned, very light traffic and surprisingly cultured drivers. Even after recent record-breaking snowfalls, these streets were very clear short after.
Would love to see them trying to drive anywhere else in the city though.
Wonder what's that thing at 1:40 ;) "at turn, wipers suddenly full tilt for no apparent reason, person gives input to the right paddle" - yes, I know that control normally sends CAN messages for wiper operation in all modern cars, and that it wouldn't make sense to release a video with overt manipulation of the system; I have been conditioned to look for side channels though, and this is technically a human giving a driving-related input.
But seriously, what does the car need the wipers for? All the sensors are outside anyway, no?
The wipers are needed for when a human needs to take control and if a human needs to suddenly take control they need to have the wipers already working recently.
135 comments
[ 5.3 ms ] story [ 187 ms ] threadMind you, self-driving when snow has coated all your landmarks is definitely impressive, but the title made me think it'd be driving in snow. Which it isn't. The roads are, what we'd call in Chicago, "clear". The car is not driving in/on snow or ice in this scenario, which is something I do regularly every winter.
https://www.google.com/search?client=ubuntu&channel=fs&q=coq...
The idea would to put the smart tech in the car. What's going at the roadside could be made very cheaply and integrated into existing features without needing to rip up the entire road.
Other countries have toll highways which are well maintained, that's another option for installation.
I don't see this happening in rural/country lanes, the cost would be obscene (but again, we do liberally use reflectors in the UK). I also don't see this as a panacea, it would simply be a backup in horrendous weather.
Plus there's also no incentive for anyone to waylay the smart road off the physical road, because there are no evil people (i.e. with motives that are incompatible with the vehicle's survival).
TL;DR: good roads will be getting better, bad roads won't.
Since off-road use for smart cars is comparatively small, we should optimize the roads to be smart-car friendly.
Perhaps we could install systems which would enable smart cars to stay in lanes, perhaps with some hardware preventing the smart cars from going off-road.
And while we're there, we could also have systems providing charge built right into the road! Wouldn't that be neat?
The lane-keeping and power systems could allow regular cars to operate on the same roads, for backwards compatibility reasons.
Sounds too expensive to build? Not until you realize that you've re-invented the infrastructure for trams and trolleys put in place a 100 years ago.
https://twitter.com/elonmusk/status/753843823546028040
Radar has been used on snowploughs to avoid rocks/curbs/etc buried in the snow (article from 2004):
http://www.nature.com/news/2004/040302/full/news040301-2.htm...
Ford developed a LIDAR algorithm that can "detect" snow flakes and still be effective, this video shows driving in real snow:
https://qz.com/637509/driverless-cars-have-a-new-way-to-navi...
Computers can provide near "perfect" speed/traction ratios based on active road conditions... which humans are merely estimating based on "feel", without instant direct feedback from the wheel traction sensors, combined with whatever (potentially limited) winter driving experience they have.
Computers can also "remember" any specific location of the route from hundreds of thousand of other rides in the past and of other cars on the road communicating in real time.
In 5yrs I'd expect to prefer putting my trust in any self-driving car on winter roads over a human any day...
And if there is self-driving zones, and non-self-driving zones, you better believe I still expect my normal car to work if I drive it from Chicago down to San Francisco.
Sometimes only one car, if the timing works out that way. Three is almost certainly too many. The ability of a car to judge the light timing requires near 180 degree vision and head pivoting on the part of a driver.
Even all cars become autonomous, the problem will still be hard I believe. Hard coded rules may solve the problem but relying on communication and cooperation will not be easy.
Of course, this only allows one car (or maybe two, if the intersection is particularly large) per light cycle, which isn't much. People who enter the intersection on the yellow or red light (tailing the person who was in the intersection waiting to turn left) are turning illegally.
The real trick against gridlock is to equate the switch to green not as "go", but as "go, once the previous wave has cleared the intersection". The "don't enter before there is room for your car on the far side" is only an optimization on top of that.
The way traffic works is somewhat dependent on region. I can go to another city an hour away and already feel a bit alien traffic wise there. This is relevant in discussion on autonomous vehicles: in any foreseeable future we need them to coexist with human drivers and abide by unwritten "everyone drives like that here" rules.
This is the case for NY state, see VTL 1175: http://www.safeny.ny.gov/rowa-vt.htm
https://leginfo.legislature.ca.gov/faces/codes_displaySectio...
21452. (a) A driver facing a steady circular yellow or yellow arrow signal is, by that signal, warned that the related green movement is ending or that a red indication will be shown immediately thereafter.
(GP: That's actually easy: random exponential binary backoff. It works for other domains quite well.)
"Works great near my house, might accidentally kill you if you try to ride it outside the happy path" is somewhat underwhelming.
Which may make the auto driving easier.
So giant piles of snow often means that all of your reference points on your fancy 3D map of the road are now obscured by piles of snow, and you still have to be able to position the car well enough that you know when to look for stoplights, crosswalks, and the like.
For snow-covered roads we'll have to wait for reports from, say, Chelyabinsk. Or Tomsk, or Novosibirsk...
Wouldn't any pedestrian detection decide the person in the road has priority and stop and let them cross? It isn't capable of reading a person's intention to stand there vs. cross or not.
The title claims it supports interactions with pedestrians, and the video does seem to show it pausing in the middle of a turn once for a pedestrian in the road, but that could be easily faked by having the driver still using manual breaks.
https://www.youtube.com/watch?v=Bx08yRsR9ow&feature=youtu.be...
That pedestrian is already in the street and they momentarily stutter then pass the pedestrian with very little room.
That and the way the steering seems super sensitive to various snow remainders on the side of the street. It's almost as if it's controlling for its position relative to the snow instead of picking a mostly straight trajectory.
Considering the amount of traffic-related deaths in Russia... They'd better handle it like non-native driver.
Yes, I also noticed it, it seems like there are continuous micro (and not so micro) corrections on the steering wheel, with such behaviour the passenger (besides the driver that has some more chances, being attentive to the road, to not get it) is likely subject to get car-sickness.
If self-driving cars are to use the current infrastructure and drive as a regular car on regular roads then better be able to handle this.
And the top cap for the city is 60 (except for things like https://en.wikipedia.org/wiki/Moscow_Ring_Road, you can also go to 80 when overtaking and many people abuse this rule to go ~80 all the time)
Only when you are outside of apartments' territory.
http://pddmaster.ru/pdd/pravila-dorozhnogo-dvizheniya-dlya-z...
The dream that the developers, the media, and many of us share is to change the world by making driving a thing people do not do. Matching a human driver in the most treacherous conditions is a major, and unmet milestone. This is a misleading post.
https://m.youtube.com/watch?v=7hUut7Hsgys
This tech, as-is, could probably save millions in damage on minor or one-car accidents in snowy aftermath. Even if it snows once or twice a year, somehow everyone forgets how to drive safely in the snow and Every.Single.Time there is snow the world just stops.
Unrelated to this video, I was impressed with my Subaru EyeSight performance in a whiteout rainstorm. I could tell what was going on - barely, with the wipers at full tilt - and my auto-cruise worked without a hitch at > 50MPH for several minutes straight. It's definitely hit-or-miss though, since sometimes it gets confused on a bright overcast day at 25MPH.
Roads that are usually 2 lanes become 1 lane when snow accumulates enough. People improvise to get where they want anyway - the 2 lanes at the traffic light which are normally one lane for turning and one for straight - is now just 1 lane for both turns and straight because of accumulated snow. The AI would need to observe other drivers (and pray they aren't all AI drivers) because stored map info will not work.
As a driver you have to be able to guess whether something that looks like a 1ft ridge of snow is actually passable drift powder snow, or a frozen barrier of ice that will damage the car or get you stuck if you drive through it.
Sorry if I'm being ignorant (where I live, below 20°C is really cold weather, and the lowest it gets is around 10°C, which I don't recall ever seeing), but wouldn't heaters be enough?
VW has a good solution for their rear camera which is actually hidden under a hatch (the VW emblem) so it’s protected when unused.
I’m picturing that future autonomous cars will need this type of solution for many sensors, similar to the “flip up headlights” of some 80s cars. Not least because lidars are also very expensive so you may not want them easy to grab from a parked car.
Of course, in mountainous winter, I've had wiper fluid freeze when it got to the windshield of a warmed up, running, moving car - had to stop and scrape off what was essentially a rubbing alcohol solution: even heating wouldn't help when local temperature goes to -40 (either scale ;)).
In normal conditions, your car stops in a predictable way, at predictable speeds. It doesn't slide. It doesn't follow cut-paths in the road. Acceleration is a simple function.
Think that you've adjusted for these parameters? Well, the ground you've driven to has more snow on it then before. Or you're elevation has decreased so the snow:ice ratio has shifted towards the later.
And the above only describes the situation if you are one of the few people on the road. Add in all the other drivers who are acting _strangely_ in terms of performance and psychology. Driving lanes shift depending on where the road is clear, typically centering (bringing sliding cars closer together).
Never mind people FOOLISHLY walking on the road (the sidewalks are covered in snow).
Driving in cities in summer requires a high degree of attention and quick responses to others. But it is just a matter of interpretation and response. Winter driving is an art...
In the best case, visibility restriction is limited to the sides of the roads, like in this video. In the worst cases, there is barely any indication of where the road is, and signs are also covered in snow. GPS is being blocked by the storm overhead. Forward visibility is down to being measured in tens of feet, from 10 miles only five minutes earlier.
As for the vehicle itself and the electronics: in winter, they will need to function pretty much perfectly at up to -40°. Cars, if left outside, can be covered in layers of over 6" thick of snow and/or ice - and not always just on the top.
While operating on winter roads, they will be exposed to a constant spray of slush and snow, combined with salt and sand. This spray will coat the grills and tails of cars. My vehicle has parking assist and radar assisted cruise control; neither of which work after driving down a road with fresh snow. Even the otherwise protected rear view camera gets coated in muck just from the aerodynamics of the car's tail.
None of this is impossible to overcome, but the typical "it's just an engineering problem" is aggravating to hear.
This: http://bgr.com/2018/02/06/moscow-snow-russia-winter-storm/
Is what it looked like ~two weeks ago. Show me your self-driving car in THAT.
https://www.youtube.com/watch?v=vShi-xx6ze8
This isn’t as impressive as I was led to believe. Those roads are completely exposed!
The video deserves a second view, with a proper BSO :-)
https://www.youtube.com/watch?v=9Bx20cflBBk
(I feel like this question must have been brought up a number of times already, since these projects have been going on for a while by now.)
https://yandex.ru/blog/company/130717
https://www.uber.com/ru/blog/future-uber-yandex/
Would love to see them trying to drive anywhere else in the city though.
But seriously, what does the car need the wipers for? All the sensors are outside anyway, no?