Cruise Shows Off Level 4 Skills in SF, Passing Uber and Maybe Waymo | Hacker News Reader

Cruise Shows Off Level 4 Skills in SF, Passing Uber and Maybe Waymo

driverless.id

371 pointspaul143 comments

At the risk of sounding pedantic this is not a Level 4 demo. It is a Level 2 demo where the driver didn't have to intervene during the demonstrated route.

What's the difference? Apart from the legal requirement to have a diver ready to take over in test vehicles (which necessarily makes it Level 2), the fundamental difference is that you'd have to show a lot more than one demo to establish that you've achieved Level 4. Level 4s are supposed to be able to operate without human intervention at all within prescribed domains (e.g. downtown cities). That doesn't mean operate one trip or one day or one month without a disengagement -- that's still Level 2.

I'm super impressed by the demo but Cruise will have to show more data to back up a Level 4 claim.

Thank you! This is not pedantry at all. You are correcting incorrect misleading safety claims.

I agree completely. And the video although not new is quite impressive. But if the claim of the author that San Francisco is a tough city in which to drive is true, than I fear that some parts of Italy (and probably A LOT of cities in the world) won't have self driving cars for a very long time. When I saw the video the first time I was very impressed, but I would have never imagined that for someone that was a VERY challenging urban scenario. From the article:

    when we see the streets of San Fransisco conquered, then we know that self-driving is ready to come of age.

Seriously? And if you conquer the street of Naples what will come of age?

This is moreso true for China and India.

If you've ever been to those countries, the first thing you notice as a westerner is that the streets are chaos, traffic rules are barely obeyed, and aggressiveness is required to get anywhere.

Meanwhile, Beijing and Dehli have extreme pollution problems partially caused by cars.

If any place was in need of point to point autonomous ride sharing, it's Asia. The reduction in cars would reduce congestion from the outset. But in the future, protocol guided right of way negotiation could reduce congestion even further.

I tried to flag a taxi in downtown Shanghai for 45 minutes and finally gave into the broken English speaking motorcycle taxi who had been circling the block every five minutes.

Riding bitch through Hangpu to Pudong with a laptop in one arm while lane splitting with aggressive (and otherwise bad drivers) all around was a rather exciting experience.

Even with a skilled pilot, the adrenaline rush was similar to my first time tandem skydiving.

Your driver must be a real level 4 driver.

> Even with a skilled pilot, the adrenaline rush was similar to my first time tandem skydiving.

following lines from neal-stephenson's cryptonomicon come to mind:

      "like Disneyland with the safety locks taken off"

I could imagine the robot car having an easier time in a place with minimal expectation to obey traffic rules. It simply becomes an entirely self-serving entity solving a constraint maximization problem. Namely, get from A to B, don't hit anything, don't get hit.

There is quite complex social interaction going on road. Rules and commands are communicated to participants and there are only seconds to understand them and obey

I like this video of traffic in Hyderabad, India. Whole different kind of logic applies there: https://m.youtube.com/watch?v=JYjG_S8XJM4

As we get to 100% autonomy, traffic will probably look like this but 100x faster and safer.

Something you often see in simulation. But I wounded if these simulations ever use different decision engines and criteria for different vehicles. The different manufacturers are going to produce their own software (some at least) and not treat an encounter the same way. Are we going to have race conditions or "database" lock-like situations (not sure what to call it), as cars try to interpret each others behaviour?

This example, and most simulations, seem to use an algorithm which is very close to "if there is space for me to drive, drive".

In the medium term, I expect self driving cars will instead start to communicate with each other to find a better approximation of optional traffic flow (by finding the solution that is best for the negotiating vehicles). This has several competitive advantages over the simple algorithm: it enhances safety for participating vehicles by removing prediction errors, and if you can negotiate with the other car instead of guessing what it will do you can do much tighter manoeuvers with smaller seaters margins. And of course better traffic flow makes driving more pleasant, which is good for car sales. Given all those market forces I think it's very likely to happen.

In the medium term I agree, but there are som major standards issues and negotiations, like common implementation against the law in several major markets, to work out first, I think.

So beautiful to watch. Order in chaos

If a machine is able to drive in Bangalore, I'll presume that Artificial General Intelligence has dawned upon us.

https://www.youtube.com/watch?v=WMwJD9NG6NQ

> "If you've ever been to those countries, the first thing you notice as a westerner is that the streets are chaos, traffic rules are barely obeyed, and aggressiveness is required to get anywhere."

I'm going to get down voted for this, but it has to be said.

The whole "well it couldn't drive in X country" scenario is ridiculous. Driving in unruly, chaotic, unregulated environments is NOT a problem for self driving cars to solve.

That type of driving environment certainly IS a problem, but one that has already been solved by good governance, including laws, rules, and regulations for improving safety of traffic. Countries like the United States, among others, have developed a set of rules and regulations that make self driving cars possible within their domain. It's foolhardy to assume self driving cars can, or should, work outside of that domain.

I partly agree with you but I feel population density is also a major contributing factor here.

If you're solution to the problem is to not solve it under the constraints offered, then you're not really solving it. Someone could easily use your argument to say we shouldn't have self driving cars in the US either when we can just use automated trains.

> "If you're solution to the problem is to not solve it under the constraints offered, then you're not really solving it."

That is of course, unless, you DO solve it under other constraints. Sure, you haven't solved X problem with Y solution, but you've still solved X problem. Self driving cars may not be the solution to ALL kinds of traffic problems, but it turns out, they don't have to be.

From my experience in my home country, the only rule of the road there is the survival of the strong. I would imagine that designing algorithms that follow only one rule should actually be much easier. An unpleasant and dangerous environment like this would actually benefit much more from self-driving technology.

There's also the issue of technical problems with for example traffic lights. I remember an occasion in Berlin where the traffic lights broke down on a major intersection with long queues. The right of way ended up being solved in a contest between the status quo competed with the impatience of those who were waiting. The only way as a pedestrian to get across was to step out into the street so cars would stop, which was a bit uncomfortable. Worked out alright, if not terribly efficient. Fixed pretty fast, I think, but an AI would still have to handle, uh, dynamic traffic in western countries.

Ideally, the AI should be able to figure out how the dynamics of such a situation work and not create a jam or something.

Arguably, the city should dispatch a traffic guard to deal with the situation.

In California if the traffic light breaks down then you're supposed to stop as if there was a stop sign; all the same rules apply.

In Germany a completely black traffic light has no meaning, rules are as if it wasn't there. Most traffic lights have the usual signs for priority roads for this situation.

But a broken traffic light is a situation that is unfamiliar to drivers, and at peak hours the crossing is guaranteed to be either difficult to navigate or to have suboptimal traffic patterns (otherwise there would be no traffic light in the first place). And of course pedestrians have to improvise (and Germany has a lot of pedestrians compared to the US)

Why would a North American self driving car have to handle Indian traffic?

Naples is a bit extreme. Driving there is insane, but even other more normal European cities like London and Paris are a lot harder than any American cities.

I still haven't seen a driverless car handle a single-lane road (e.g. with parked cars on the side). Something like this:

https://www.google.co.uk/maps/@51.4606316,-2.5562767,3a,75y,...

I was very impressed at the number of fringe cases it handled.

A car forcing a merge right in front of it, an armored car taking a lane and a half where it had to wait to go around (there were numerous instances of similar lane blockages that it navigated around), a pedestrian jay walking in front of it, a bus picking up passengers, a bike cutting in front of it, and car pulling out in front of it in heavy traffic.

I had no idea they could handle that many scenarios already.

How is "downtown city" even defined? Driving in SF seems vastly different from driving in Rome, Istanbul or Ho Chi Min City.

I wondered that too. Pretty ridiculous that they claim "Self-Driving on the Streets of San Francisco Is Really Hard". City driving in the US is pretty much as easy as it gets. These guys are up for a real challenge when they try to make it work in Delhi, Bangkok et al.

Nice. It's annoying that they provide only 10x sped up video. Watching this slowed down is helpful.

Notes:

* There are frequent steering twitches to the left. This may be associated with passing parked cars. There are similar twitches to the right when in the left lane of a one-way street.

* Crosswalk behavior when turning needs some work. The vehicle enters the intersection, then stops in the intersection before the crosswalk with people in it. This is a hard problem, because the system needs to recognize people waiting to cross but not yet in the roadway. When the light turns green, both the pedestrians going straight and the turning vehicle can enter the intersection, the pedestrians having right of way. The pedestrians now block the vehicle, and the vehicle blocks the bike lane.

* Left turns into multi-lane streets are too wide and into the wrong lane.

* On two occasions, the vehicle is stuck behind a doubly-parked vehicle engaged in loading. The options are to wait or to cross a double yellow line. There's a delay of several seconds, then forward movement. Suspect manual intervention.

At one of the red lights it comes to a complete stop then crawls forward. 0:49 Maybe because of the cars turning left. That stood out to me

That happens more than once. Google cars sometimes do that when they don't have a good view of cross traffic due to corner obstructions. It's the right thing to do, but it's resulted in Google cars being rear-ended at very low speed several times.

I counted 30 seconds of being stuck behind the unloading van with no way to pass without crossing double yellow. I didn't notice the autonomous green light turn off though. Possibly the algorithm is, if you're stuck then certain rules become flexible.

Worth mentioning, 30 seconds would feel like an eternity inside a car.

I assumed the "autonomous green light" was added in post-production, along with the timer and the company logo.

Google had another crash last month.[1] As usual, it was probably the other driver's fault. Other driver apparently botched a left turn in a two-lane left turn intersection at Rengsdorf and El Camino. Google drives in traffic on the SF peninsula every day, and we know exactly how many times they've crashed. It's reassuring seeing all those miles with only the occasional fender-bumper. It's not like Tesla slamming into something on a freeway at full speed. Three times.

Still waiting for the CA DMV to post the 2016 autonomous vehicle disconnect reports. Those cover December through November and are due Jan 1st, so DMV should have them up by now.

[1] https://www.dmv.ca.gov/portal/wcm/connect/3d358211-3f0c-430e...

> It's not like Tesla slamming into something on a freeway at full speed.

To be fair, Tesla is not selling a fully autonomous vehicle, yet, people are treating it like one.

And Tesla had encouraged that perception.

Tesla appears to be extremely clear to users as to the current state of the system. Have you actually used Tesla autopilot? What did you think of it?

It's called "autopilot".

Too many people, myself included, that's all that matters.

They need to stop calling it "autopilot" until you can get in and fall asleep in the back seat.

And you're a Tesla owner? Or you've driven a Tesla and used autopilot? Or you're considering buying a Tesla?

Airplane autopilot isn't designed to be used in the way you described. Should aviation rename that as well?

Pilots know that, but I bet a sizable portion of the passengers don't.

Can't you go theoretically go to the toilet on autopilot? It won't land for you, but while turned on it will keep the plane aloft until fuel runs out, no?

I don't know the answer to this, but it's my guess that at least the pilot or copilot must be at the controls at all times.

should be, but doesn't actually have to I guess. (still relatively empty up there, so not so much traffic to avoid)

Tesla was selling something which is marketed as being able to handle freeway driving under good weather conditions. It can't be trusted to do even that right. The "rams into vehicle stopped at left of roadway" feature is unacceptable. That's an expected possibility on a freeway. Users would legitimately expect auto-braking to activate for that.

Tesla's system has an order of magnitude or two of more miles driven than anyone else though, so I think it's unfair to speak of a higher risk based on current data.

wow, this document is redacted quite poorly

>> Worth mentioning, 30 seconds would feel like an eternity inside a car.

Would it, when the car is driving for you and you're free to putter about with a phone or laptop or book?

I always figured that this would be one of the greatest benefits of autonomous vehicles - perhaps the car isn't as aggressive as the human it's driving it would be, or as quick to get "unstuck", but the human won't care too much because they're too busy on their phone. In most cases, what's a few extra minutes when you're no longer actually driving the car?

I think you're completely right from a rational perspective, but I've definitely sat in the back of an Uber wishing the guy would drive more aggressively since I was running late.

The solution is simple. Screen a drag race across the interior windows of a self-driving car.

> The pedestrians now block the vehicle, and the vehicle blocks the bike lane

Entering the bike lane appears to be legally-required [1] behavior for a driver turning right across a bike lane in California. Entering the intersection, however, i'm not sure about.

Illustrated here: https://www.sfbike.org/news/bike-lanes-and-right-turns/

[1] https://leginfo.legislature.ca.gov/faces/codes_displaySectio...

What is scary is how few drivers in California know about this rule or the reason behind it, which is for the safety of the bicyclists.

Near my home there are several intersections with bike lanes, and I estimate that perhaps one driver in 10 enters the bike lane to make the right turn as they are required to.

In fact, I sometimes get surprised or even dirty looks from drivers making illegal right turns when I do move into the bike lane to make the turn. I guess they think I'm trying to get ahead of them and cut them off, but I'm not, I'm simply following the law and improving bicycle safety.

When only 10% of drivers apparently know the correct way, the problem is with the system not the drivers.

Better signage and more consistent (state to state) rules that make common sense would go a long way.

Keeping in mind what is common sense to a biker is many times counter-intuitive to a driver. I hate the unprotected bike lanes on the right - as a driver you simply are not expecting people to be passing you on the right, on what effectively "feels" like a shoulder. It amazes me there are not more deaths due to this, to be honest.

After I spent some time in the Netherlands and Belgium I realized this was not just me. Those systems actually work and are designed with "road sharing" in mind. Everyone gets where they need to go, and even a dumb foreign tourist who knew nothing of local traffic laws could operate a vehicle safely amongst cyclists.

> It amazes me there are not more deaths due to this, to be honest.

Intersections are the most frequent place for cyclist/driver collisions, and "right hooks" (where a driver fails to merge or yield and turns right into a cyclist trying to continue straight through an intersection) are the most common kind of intersection collision, so your instinct is spot on. Your perception that there aren't that many fatalities is probably partly because traffic deaths don't get reported on much, and partly because collisions involving turns tend to take place at relatively low speed, so fewer of these collisions are fatal than might be the case under other collision circumstances, as speed of the car is by far the strongest predictor of death in a cyclist/driver collision.

> I hate the unprotected bike lanes on the right - as a driver you simply are not expecting people to be passing you on the right, on what effectively "feels" like a shoulder

Having driven for a bit in California I'm now used to expecting people to be passing me on the right, on the left, between lanes - everywhere - thanks to a combination of legal and illegal lane splitting by motorcycles and mopeds.

All the points you make are heavily characteristic of NYC as well as double parking, rampant cabbie pickups and drop offs, midtown crosswalk blockade by heavy pedestrian traffic, etc

I would love to see how the self driving contraptions handle this city at rush hour.

Fuck. I can't handle your city at rush hour.

Some of these issues sound more like issues with the traffic in SF than with the car. Maybe the cars highlight that issues to us and we can fix it instead of work around it. Bicycle lanes turning into turning lanes leads to accidents and aggression between cars and bikes. Trucks shouldn't be allowed to double park. Maybe self driving cars are like there be dev on the team who points out all the shitty code that you've gotten used to.

I agree with you, but in order for a car to be truly self-driving, it must be able to handle these situations. Just because they're illegal doesn't mean it won't happen

Totally agree. I just am somewhat hopeful that something positive might come from it.

Traffic in SF is traffic that works imho. I love the nuance and judgment that drivers are forced to use. I feel safer walking/bicycling/motorcycling here than in any other city, and that speaks volumes about the city's livability. Density creates these problems, and density is good for other reasons. It's only in suburbs and rural areas that every store can have a loading zone, but then no one can walk anywhere.

Ha, I interpreted the yellow line crossing as a timed fallback.

Dealing with pedestrian crosswalk behavior is, in some ways, in the same class as left hand turns across unregulated traffic in the sense that both are situational. What I mean by that is that with few pedestrians/light traffic, it's appropriate to just wait it out. But there are definitely circumstances in cities where a bit more aggression is called for. I don't mean cutting right in front of a car or sending pedestrians scattering but you do sometimes need to take your space and go or you'll sit there for a very long time with horns blaring behind you.

"Level 4" means that no human intervention is required and in case of conditions going from good to bad, the car can autonomously put itself in a safe state.

That was nice and clean city driving in the video clips but nothing that distinguishes it from "Level 3" (human intervention may be required within ~15 seconds or so) or even "Level 2" (human intervention may be required within seconds, current state of the art).

I imagine a space defined by visibility, traction and road flatness. A large space of Level 0 conditions, i.e. the conditions within most humans can drive, encapsulates Level 1, i.e. cruise control, which mostly encapsulates successive levels.

The benefit of this is you notice where levels "poke through," e.g. Level 4 may work on a sunny day, but a small change in rain or road conditions could downgrade the system to Level 1. As time goes forward, the inner levels can be expected to radiate out.

EDIT: Never mind. The point of "Level 4" is it is competent in all reasonable operating domains.

> Level 4 may work on a sunny day, but a small change in rain or road conditions could downgrade the system to Level 1.

This isn't the way it works. If a car says it can do "Level 4 on a sunny day" it means when you sit in the car and engage the autopilot (and it says it's safe to engage), then no human intervention will not be required during the course of the trip. If conditions change, the car will be parked without human assistance and wait for help (what happens next is outside of the scope). You could be sitting in the back seat, or your kids could be in the car alone on their way to school.

"Level 4 on a sunny day on some pre-certified highways" is ok. "Level 4 in San Francisco traffic" is also ok, and much harder. "Level 4 unless it starts raining and we'll deteriorate to Level 3" is Level 3, not 4.

This is the definition of autonomy levels from SAE, and they're pretty strictly defined.

> This is the definition of autonomy levels from SAE

It's still domain dependent, right? If the rules say we're flattening precipitation and visibility within normal bounds, fine, but sometimes you have abnormal weather and badly-maintained roads. It is useful to compare hypothetical cars that autonomously navigate conditions no humans would dare.

EDIT: Never mind. The point of "Level 4" is it is competent in all reasonable operating domains.

> It's still domain dependent, right? If the rules say we're flattening precipitation and visibility within normal bounds, fine, but sometimes you have abnormal weather and badly-maintained roads

No, it's not.

In Level 4, if there's "abnormal weather and badly-maintained roads", the car must be able to deal with the situation and enter a safe state. It can say "please sit down and wait for assistance" but it may not say, "take over the wheel".

In practice you'd probably disengage the autonomy and drive yourself, you take control and it does not give you control. If you can't sit on the back seat drinking beer or send the car to drive your kids to school (we're not considering legislation issues here), it's not Level 4.

This is an important distinction. "Level 3 in most conditions" is good enough to pass as "full autonomy" for most people, but Level 4 is a requirement for not absolving the humans on board for any legal responsibility.

Got it. That's helpful. So Cruise/GM are 4G'ing a Level 3 vehicle?

It's impossible to tell the level of autonomy from this article. What was shown in this article could qualify as Level 2, 3, 4 or 5.

To provide proof of Level 4 autonomy with video (in certain conditions), it would need to show adverse and exceptional conditions, such as terrible weather, accidents ahead, all routes to destination blocked or any other situation short of a force majeure disaster and then provide a safe contingency for that.

In San Francisco, that would probably mean finding a parking lot where it is safe to wait for assistance from your Transport Service Provide(tm). In rural northern Europe where I'm from, it would mean parking on the side of the road and calling your wife/mom/friend to pick you up :)

Level 0: Automated system has no vehicle control, but may issue warnings.

Level 1: Driver must be ready to take control at any time. Automated system may include features such as Adaptive Cruise Control (ACC), Parking Assistance with automated steering, and Lane Keeping Assistance (LKA) Type II in any combination.

Level 2: The driver is obliged to detect objects and events and respond if the automated system fails to respond properly. The automated system executes accelerating, braking, and steering. The automated system can deactivate immediately upon takeover by the driver.

Level 3: Within known, limited environments (such as freeways), the driver can safely turn their attention away from driving tasks, but must still be prepared to take control when needed.

Level 4: The automated system can control the vehicle in all but a few environments such as severe weather. The driver must enable the automated system only when it is safe to do so. When enabled, driver attention is not required.

Level 5: Other than setting the destination and starting the system, no human intervention is required. The automatic system can drive to any location where it is legal to drive and make its own decision.

From the Society of Automotive Engineers

https://en.wikipedia.org/wiki/Autonomous_car#Classification

Watching this video what struck me most was how overbuilt almost all the streets were - they're designed for high speeds and devote essentially all the space to cars. We know how to get at least a 3x reduction in fatalities, as the Netherlands has done, with engineering changes that make human drivers less likely to make mistakes and make their mistakes less costly.

Really? The narrow downtown San Francisco streets? You should check out a suburb sometime.

Very cool. But am I mistaken that they didn't attempt to go through a 4-way stop with massive pedestrians, like in PacHeights or Marina? I would be curious to see how they fared with the obnoxious drivers skipping their turn as well as the constant flow of pedestrians.

I'm also curious how it would react to going up Mason and California, where there's a traffic light at the top of a steep hill. Last time I had to physically pull myself up via the steering wheel to see anything, and as a seasoned driver I was a bit worried.

I'm guessing that will be part of Level 5 or at least we will be very close to level 5(full) autonomy once complex situations like 4-way stops with lots of pedestrians have been mastered.

So you’re saying until we reach Level 5, autonomous cars will be entirely useless almost everywhere except a handful of US cities?

No.... They would work perfect on a highway WAY before we have cities figured out. If autonomous cars and trucks work on highways, then millions of jobs go poof.

Fully agree with you until the last phrase.

Highway autonomy will be a nice feature but you still need a competent human driver to be present for end to end transportation. So I still need to order a car with a driver to get to the airport.

I assume that you're referring to trucking. It's possible that trucking companies will set up depots right off the highway and will reduce drivers as a result. But a lot of long distance container transport is already done by trains so it's not immediately obvious that eliminating drivers for only the highway portion of trucking is necessarily a big enough win to make it happen. (Especially if autonomous driving systems allow for the driver to sleep in his cab while underway.)

They won't be fully autonomous until we reach level 5, that is what level 5 is by definition. Whether or not a car that isn't fully autonomous but can handle a variety of common situations is completely useless to someone is an individual assessment. I personally look forward to the extra safety.

Pigeons and human have a pact. You continue and they will fly away in the last second. Cars don't need to stop for pigeons. https://youtu.be/xPCZtrac-Ss

What about squirrels? :)

I hope you're kidding. In an area where they had just opened a freeway segment for my commute, my car had a couple of "reverse bird strikes" -- the car was safe but the fate of the bird was unknown -- at the beginning of the new segment. It's possible the birds weren't pigeons.

He was kidding. It's from Seinfeld.

Dumb question: how do these systems distinguish a traffic light? Here in New England at least, there's quite a variety of lights, many different modes (red, yellow, green, flashing red, flashing yellow, no-turn-left red arrow, no-turn-right red arrow, and different light technologies: old-fashioned style, Fresnel lens, slotted shade.

Add to this complexity the weather conditions. Suppose the sun is shining straight at you and you need to squint and shade your eyes just to make out what the light is -- this happens to me frequently -- can the camera see the traffic light and distinguish its color clearly under such conditions?

What about when it's raining, misting or drizzling, snowing heavily, etc. and the traffic lights are these fragmented outlines that you, the human, can heuristically distinguish but a machine might not?

One last thought: suppose it's right turn on red and first car in line is a self-driving vehicle. Can it really look left and safely determine there's enough time to beat the cross traffic? If it's highly conservative and just waits until green, there could be ten irate motorists behind it and guaranteed to honk and curse.

It's exciting technology but there are some very difficult problems to solve. I worry that if these machines can't demonstrate 110% of a human's ability to drive, they simply won't be implemented in many places except some very well defined rigid routes that are free of problematical challenges and variations.

    red, yellow, green, flashing red, flashing yellow,
    no-turn-left red arrow, no-turn-right red arrow

You forgot flashing green, which does actually exist, at least in MA. You treat it like a green light, but it could go yellow and then red if a pedestrian presses the walk button.

Yeah, that, and also a traffic cop standing in random spots waving cars around an accident etc.

In some parts of San Francisco there's a stoplight with a green X. Took me a long time to figure out it means "one way street, don't enter"

> Dumb question: how do these systems distinguish a traffic light?

Well, how do you distinguish / identify a traffic light?

For me it's a combination of knowing the area, knowing what a traffic light looks like, and observing the behaviour of traffic around (mostly in front) of me (if the intersection isn't visible).

For an autonomous vehicle, they'd use the same methods plus they'd have the non-trivial added benefit of colleague robots feeding them updated information.

> Suppose the sun is shining straight at you and you need to squint ...

I'm sure the human eye + sunglasses + visor comes a poor second compared to CCD + mechanical + IR + (etc) can do.

> What about when it's raining, misting or drizzling, snowing heavily, etc. and the traffic lights are these fragmented outlines that you, the human, can heuristically distinguish but a machine might not?

Interesting use of the word heuristic there. If you can determine the heuristics you are using, then an autonomous vehicle can use the same.

> One last thought: suppose it's right turn on red and first car in line is a self-driving vehicle. Can it really look left and safely determine there's enough time to beat the cross traffic? If it's highly conservative and just waits until green, there could be ten irate motorists behind it and guaranteed to honk and curse.

This, and variations, frequently come up in discussions on AV. There's an implicit expectation that no one building these things has considered this problem (this is clearly false). There's two explicit expectations that once a tipping point of AV's are out there, a) there'll be a regulatory push to massively accelerate the adoption close to 100%, and b) inter-vehicle communication means this problem won't arise. In the short term these problems may occur, but to address your question, yes, I'd suggest a computer would be better able to predict if there's enough time to safely turn than most humans.

> I worry that if these machines can't demonstrate 110% of a human's ability ...

Which human would you pick?

Anyway, I wouldn't worry - none of these problems are intractable.

OK the inter-vehicle communication and crowd sourcing and AI sound like they will manage it. I'm assuming though that we'll eventually want AV-aware sensors attached to every light and every stop sign and every intersection. UHF RFID or some such that is cheap to deploy.

I think there'd be massive cost in retro-fitting lights and intersections - probably easier to agree on a mechanism for new infrastructure, and rely upon current approaches for extant infrastructure.

My concern is more about the regulatory side -- before any of this stuff gets widely deployed, various authorities are going to have to come up with some standards, and for the most part nation states are spectacularly poor at doing this sensibly in isolation, and horrendously woeful at doing it in cooperation. I suspect we'll end up with cars & infrastructure that are not 100% interoperable across borders.

The funny thing is those irate drivers would be honking at a piece of metal.. no driver with whom you could be angry.

Random side thought: How do I signal in an ambiguous situation to the car on who's going first. Like Super high traffic and I want to merge in front of them or let them go first? Currently I do that with a wave. Which is super effective on a motorcycle as people almost always say "sure we can squeeze you in here" which lets me get places way faster.

LED signs attached to the car.

Yes, I absolutely do think that some kind of car-to-car communication is required for these situations.

It's not an impossible problem to solve, but I have not heard of any efforts to create a car-to-car or car-to-road communications protocol and infrastructure that would be cross-manufacturer and/or internationally approved.

In my opinion, before such a mechanism exists, autonomous cars will only work as long as the majority of cars are driven by humans. I will change my opinion when I see a demonstration of a city (or city-like test site) traffic where all or the majority of cars are autonomous.

There's a lot of academic work on Vehicle Adhoc Networks (VANETs) - https://en.m.wikipedia.org/wiki/Vehicular_ad_hoc_network has some sources. We discussed it a little at university in a module about mobile/sensor nerworks, it's definitely an interesting research area.

I've read that insurance people strongly suggest to never wave: it's one of the top things that seems to occur during accidents. It's hard to distinguish from someone "waving you on" versus someone "waving to say thanks for letting me go". Trusting anyone else to check for oncoming traffic is a terrible idea. It's really easy to focus on the wave-giver and not check the rest of the traffic. And even if you are good at these, all waves are inherently involving at least 2 drivers, and you can't really count on them being any good.

I couldn't find any real information on this though. I think I picked it up from Reddit comments. Some info does suggest it's legally complicated: http://www.claimsjournal.com/news/national/2014/03/06/244965...

Pre-cell phone while driving ban, my roommate turned left to a woman who waved him through, and was hit by her. She told the cop she didn't wave him through, but was gesturing to the person on the other end while talking on her phone. He was ruled at fault.

I was talking about at a 4 way stop or in 2 mph gridlock... Where is more like a construction site than traffic. I almost got smeared once by believing a blinker. Never again, I won't pull out until they are obviously in a turn.

Yeah there's not much else you can do at a 4 way stop if the timing is bad.

Good we need competition in this space. I love Tesla but if this is going to end up everywhere, we need a few good players.

It looks like the car doesn't travel in a straight line (it wobbles) because it's trying avoid parked cars.

Probably for the best. Someone could always open up a car door unexpectedly. I imagine it would be very hard for a driving algorithm to read social cues and notice drivers getting in/out of cars.

Cruise was a sponsor of ROSCon in 2014, but had no booth or presenters: http://roscon.ros.org/2014/#program

Nor do they have any presence I'm aware of on Github. This is in contrast to BMW, for example, who have made a number of contributions: https://github.com/bmwcarit

Anyway, just curious to what extent Cruise used (or still uses) ROS and open source software in their stack.

Funny to see the it slow down for the pigeon in the road at 14:50.

Hah, nice catch, I remember seeing it do that weird jig but didn't look rewind or look close enough to figure out why it did that.

There is a paragraph and video fragment describing this in TFA.

>within 5 to 10 years we'll see these vehicles driving safely around San Francisco

Pretty ironic in consideration of "red lights don't apply to me" Ubers and more sane but extremely slow grandma Teslas

how dare you emit an utterance that amounts to anything other than deification of our future self-driving overlords and the peter-theil-worshipping engineers who unleash them upon us!

I said it before and I'll say it again:

You might not be able to turn an aircraft carrier on a dime but when you do you've got an aircraft carrier.

GM (and the other automotive manufacturers for that matter) decided they wanted in on self driving and electric vehicles, had a few meetings, wrote a few checks and a few years later look at the result. Are Google and Tesla going to reply with similar videos?

> You might not be able to turn an aircraft carrier on a dime but when you do you've got an aircraft carrier.

I'm not sure what this means.

I interpret it as GM/Ford etc, the big ones, means serious competetion when they start focusing on this, but they are slow to start the focus since they are big.

He's saying that while it may be more difficult to change strategic direction of a $50B company (Ford/Gm/etc.), once you do, you have a huge company and giant production line behind you.

Extremely cool feat. Next try Chicago or Detroit in January guys.

Thats one of the reasons why Uber chose Pittsburgh for development. We might not get the same amount of snow as those cities but we get enough and have a variety of terrain for them to use as test cases.

CMU being there doesn't hurt, either

Yup. Plus the mayor was trying to remove as much red tape for them as he could to bring the notoriety to the city.

Chicago and Detroit don't get all that much snow.

(Southern Michigan gets quite a lot of snow. Prevailing winds are westerly so lots of lake effect. Grand Rapids gets almost 2x the snow that Detroit gets. Cleveland also gets more snow than Detroit, lake effect from Erie)

Detroit may have a lot more complex situations than Grand Rapids, and even thought it gets much less snow it does get storms with several inches which is enough to cover all the lane marking and making things very messy with traction control coming on a lot. I would say they need to deploy to as many cities in as diverse conditions as possible, but that's after they master stuff like in this article.

Once they can traverse Chicago or Detroit in the winter safely then they can try Grand Rapids or Buffalo.

For the record in East Lansing we get a lot more snow than Detroit but we're pikers compared to say Holland or Grand Haven.

I drove in Grand Rapids and Chicago today. I'd take bad weather in GR than here at home any time. More snow more often doesn't mean more difficult to drive, more population and smaller streets here make it way riskier. Ice and sliding are easy problems to account for in code, bad manners are not.

What I'd liks to see with level 4 is millions of miles driven by a car with a SOFT outer shell. That way collisions don't hurt other cars or even pedestrians. Make the car's top entirely out of foam or something, when driving without a driver.

It's interesting how the autonomous car evolution is to a very large extent a result from the private sector. Companies could have been asking the federal government to install sensors in all stop signs, and under the street to support this evolution. Rather, they look at the cities as they are and build something that works with it. Sometimes I wonder if government should take a more active role and ease the adoption by adapting cities to autonomus vehicle (rather than the other way around as it is today).

Each time the Cruise starts from a stop light, it's too slow.

Every time ... cars pass it, and dive into its lane (a typical reaction to slow moving vehicles).

Still, nice accomplishment.

It's hard to tell for sure what's happening in a sped up video like this, but starting slower than other cars at the light isn't necessarily a bad thing, depending on just how slow they're going. Going hard off the line is extremely inefficient for very little benefit. Autonomous cars correcting that habit could be good.

This is a pro, not a con. Slamming on the accelerator is a bad behavior that self driving cars will be correcting.

Example edge case: Waymo cars know that police cars often stop behind other cars and to expect people to be walking nearby.

Thought: Once we have 99% self driving cars it will be quite easy to convert a portion of roads to pedestrian only at times when traffic is light: bollards go up, lighting changes, cars informed to reroute.

This looks very amateurish, no roundabouts, no give-way intersection. It's really easy to stop at red lights or where then car in front of you breaks and it emits another red light: if ( light == red) stop() ....

This video was posted quite awhile ago. Are we sure its not just Nvidia's reference app configured by the Cruise people? Most of this stuff was in the Tesla and other Nvidia Drive demos.

Now that tesla has started to pretend selling autonomous cars now is making sense, the field went from research ( where researcher honestly reports shortcomings themselves) to the horrendous world of SV start up, where everything should be considered a lie until the day someone can actually buy the stuff and test for himself.

And so you end up with posts like this trying to analyse a video frame by frame to assess the reality of the technology, and yet everyone including the author tries to guess where's the catch ( is the green light really trustworthy ? Why is the video accelerated ? Etc..).

"...until the day someone can actually buy the stuff"

I'm reasonably certain we can stop right there. It's a heavily regulated field and these things can't be sold until they've been rigorously tested. I know SV isn't known for its coziness with regulatory agencies, but in the field of safety, they'll have to play ball or take on all the liability the lawsuits will throw at them.

Plus, many of the players do not have the intention of ever letting you "buy" the hardware. You will be buying transportation from A -> B, and it might just come in the form of an autonomous car (or a human driver).

Cruise is now owned by GM, which I really doubt is going to throw something out into the market without testing very rigorously first.

I get GM cars as rentals pretty often and while they're better than Chrysler I don't have a lot of faith in their product.

My point was that they don't take a cavalier SV-style approach to testing things.

Many Silicon Valley startups don't take a "cavalier SV-style approach" to testing things, either. There are a ton of medical device startups, for example.

Researchers have egos and do not honestly report shortcomings.

5-10 years still seems roughly accurate or optimistic. A tech demo does not a consumer-purchasable licensed product make.

Would be more impressed with a drive from Lombard and Van Ness to Telegraph Hill.

Anybody else think this was about scientology for a moment?

What is a level four skill?

Good news for Lyft?

Reply on news.ycombinator.com