AFAIK, Waymo pays far less attention to road markings than Tesla. Tesla FSD drives like a human, looking through the cameras and trying to figure out where the car is supposed to go in the lane. Waymo relies heavily on super precise mapping to determine where the road is. In theory Waymo should be able to drive correctly with no road markings at all.
The reliance on maps means Waymo can drive in Wayless places, but more reliably in those places.
Not extensive Autopilot data, but basic vehicle state attributes (speed, throttle, seatbelts engaged, etc). All vehicles have this black box (event data recorder) in their airbag subsystem.
There have only been a handful of Tesla accidents while autopilot was engaged where the vehicle was catastrophically damaged, so it’s a bit of an edge case.
Very impressive that AP disengaged at the beginning.
I don’t have much experience in SDCs, but did some computer vision/ robotics work previously.
I truly think that Tesla’s approach, if it works, will win in the long run (ie - next 15-20 years, I think the breakthrough for true L5 isn’t here yet - but will be in improvements in one-shot learning over the next decade).
Winter driving is a completely different beast than driving out in Arizona. You have to account for snow banks, slipperiness, lidar obstruction, and control itself - which is much harder.
How many thousands of miles of snow driving data does Tesla possess, compared to the #2 player?
Tesla’s approach of vision based, AP disengagement-led training will scale better, precisely due to what you’re looking at in this article - yeah, this is an unimpressive demo - but winter driving, rainy driving, dirt road driving are all general cases for Tesla, while they’re special cases for other companies (from what’s been revealed; Comma might be an exception).
Most people drive in the place they live, and drive on the highway to visit their friends and parents. It’s easy to be deceived into thinking driving is a small problem space of stop signs and lane changes. But even within just North America, there is enormous variety.
There is massive risk involved - these folks are pushing NNs further than any application I’ve seen, but with the last few releases, it suddenly doesn’t look all that crazy.
Secondly, combining rules based systems and neural networks is difficult. You end up with the N * M problem, and inevitably end up missing cases. It makes sense to do the absolute basic safety features as rules, and Tesla might be doing this, but all it takes for Tesla’s NN approach to pay off is one leap in explainability. There is an entire industry working on this - and Tesla will be the single biggest benefactor.
I don’t know who will win L5 - but I’m putting my bets on Tesla for L4.
But they're driverless. This means a very high confidence in the system. Hopefully, they'll reach total confidence next year and get rid of chase cars.
> How many thousands of miles of snow driving data does Tesla possess, compared to the #2 player?
It depends. If they're not recording the right things, they have the same 0 bytes of data as everyone else.
There's a not insignificant chance that their sensor data is insufficient to perform the tasks, and the recorded data won't be usable for training a system with whatever sensor is needed; in which case, no advantage for logging customer movements for N years.
Let me know when Tesla's cameras can put their arms up to shield their eyes from glare. Or move over a few inches to see around a big truck. Or react to a stopped big red fire truck with flashing lights. /s
Well but Tesla cars don't have brains. Maybe the lack of lidars makes the task 100 times harder, i.e. de facto impossible. We don't know. This is a big risk for Tesla but they didn't have a choice. If lidars were as cheap as cameras, of course they would use them.
On the other hand, Waymo et. al. are not risking anything by using lidars. If they turn out to be redundant, it should be easy to remove them cheaply.
This isn’t true: if you build a system for highly sophisticated inputs, simplifying the inputs can break all sorts of assumptions about the accuracy of the input data and other such invariants. If you start with a more primitive set of sensors and go as far as possible, adding more accurate sensors seems less risky.
Maybe in general, but I don't think you're right in the case of self-driving systems. Waymo says that perception is the easy part and most of the work is elsewhere. So in the absolute worst case scenario, they would have to develop the perception part from scratch. At Tesla, this took 2 years.
BTW, all companies that can, use lidars. Even MobileEye says that their L4 system will use lidars and they've been working on a camera based system for 2 decades. Tesla doesn't use lidars because they can't.
I mean - I have no idea what’s going on at Tesla, but its very likely they’re doing some internal work with Lidar. They surey understand the risk with w vision-only approach. Elon, and Karpathy, and Laettner previously, aren’t fools.
The problem for Tesla is they can’t be too Lidar dependent because nobody mass-produces them
Tesla can't use lidars because: 1. It would make the cars much more expensive. 2. They would have to refund all people who've bough FSD or provide them a free upgrade.
1. Lidar is expensive for everyone. It’s not like Tesla will pay more than other manufacturers. They’d prefer not to use Lidar, but if cameras just don’t cut it, I don't think they would die on that hill.
2. I’m sure there is some fine print around the current FSD being for camera based sensors only. They can easily say the car tops off at L4 for cameras, and L5 needs Lidar, if we come to that bridge
Yeah, it's expensive for everyone but the point was that the FSD product is de facto impossible with lidars. Therefore, they have to pretend that it can be done with cameras.
Why isn't every car body made from carbon fiber? It's lighter and sturdier that alternatives. It's also more expensive so you can put it in $60k car but not in $25k car.
Radar and ultrasonic sensors are cheap, low power and plentiful.
Putting several lidars on a car would be too expensive, would use too much power (less range) and there isn't a supply chain to make 2 million a year just for Tesla (500k cars x 4 lidars per car).
It's just isn't possible to put a lidar today on a mass market car and Tesla wants to solve self-driving before lidar becomes cheap enough.
> Tesla’s approach of vision based, AP disengagement-led training will scale better ... [about weather] they’re special cases for other companies (from what’s been revealed; Comma might be an exception).
> all it takes for Tesla’s NN approach to pay off is one leap in explainability.
It's not vision and disengagement-led training that's conferring the scale advantages you're talking about, because other AV companies actually do train on and attempt to drive in various kinds of weather, and safety drivers do the same thing normal drivers do except that they're more trained. The difference is the approach Tesla has to safety (vs the others). Which is to say: Tesla doesn't worry about training their supervising drivers, at the risk of those folks' lives.
Tesla also doesn't seem to worry about _proving_ the safety of their systems: explainability isn't simply going to be a matter of a new algorithmic breakthrough. How do you prove that your software is safe if you don't sample heavily from different operating domains you want your car to work in, and then either simulate on those miles or actually test your car? Before letting a driver behind the wheel of an autopilot, companies like Cruise/Waymo/Aurora all do manual driving first in a new domain, and then move to safety drivers (even 2 to start, for extra safety). This is not because their software is incapable of doing the same things as Tesla, but because it's very worthwhile to sample disengage rates and assess safety criticality to various issues before allowing the car to act autonomously. And even then there are safety nets in place to further mitigate risk. (You're probably right that L5 is a long ways off, but this is how we'll get to L4.)
Of course it's more "scalable" to avoid all this and put the responsibility of risk on users who, as your system improves, become less and less attentive. And so this is what Tesla does. I suspect solving the long tail of issues is going to be more of a challenge for Tesla because at least in my experience it's easier to {convert a naive rules/simple model --> complex model} than to {debug the complex model --> handle failure modes with logic}.
Very valid points. I think the key is that you’re going to need a complex model for full self driving. No way around that - even if Waymo starts with a rule based model, they’re going to have to work with black boxes and debug them at some point (and they do).
Tesla just decided to jump into the deep end from the beginning.
Fleet data of Tesla's scale probably doesn't give you a substantial advantage. MobileEye seems to be ahead of Tesla with a lidar-less system without such data (but they have partnerships with car makers to collect mapping data). From what I hear, companies like Waymo don't want more data, the bottlenecks are engineering.
In self-driving, you basically have 3 parts - perception, prediction and planning. Perception is the easy part (if you have lidars). Tesla and AFAIK others rely heavily on NNs for perception. The real challenge is prediction and planning. To what extent are NNs involved here in Tesla's system is unknown, probably not that much. Overall, I don't think that Tesla has a fundamentally different approach than others.
That’s not “snowy conditions,” that’s a clear day on a well-plowed road. We don’t have such nice conditions anytime from January to March in parts of the US. =)
Clearly a ways to go, though better than I would have expected.
I would call it well-plowed, yeah. Our streets in town don’t look nearly that nice for a few days after each storm, and we all drive around in normal passenger cars with snow tires.
The parking lot he starts in has like three inches of slush, after that there’s a quarter or half inch of packed snow on the road.
This is fascinating. In short, the vision processing aspect still worked quite well, wrt. drivable area, lane lines, pedestrians, etc. Room to improve, for certain, but I’m not seeing a fundamental flaw here.
What’s left is the driving part, the “how to make the car follow this path” part. I predict this part will be ridiculously good once Tesla gets around to it. Imagine the traction control in your car, now with immediate-response electric motors, and it knows where it wants to go in 3d space (not just “driver is turning wheel to the right”, but actually knowing the arc to follow).
> What’s left is the driving part, the “how to make the car follow this path” part. I predict this part will be ridiculously good once Tesla gets around to it.
So they are almost there with the part others barely even have to do, while the part everyone else is working on will eb ridiculously good once they get round to it?
Yeah, I'm curious after seeing the video how much the control could be corrected by a simple change to the car's software. Do Teslas have a sophisticated-enough physics model that you could reduce some constant for the coefficient of friction enough that it would be able to tell that it can't safely drive at 30mph or take those corners at 15mph?
Or is the software just hardcoded to assume that if it turns the steering wheel at a specific angle that the car will move a specific amount in that direction?
What I’m seeing is a model that has no concept of slippage. I suspect what they’ll do is have the fleet collect data to predict that coefficient of friction.
Seems plausible that they could ask all the cars this winter to report instances of wheel slippage, accompanied by video rewound with the spot the wheel slipped on annotated. Could have a prediction for mu at each point on the road ahead of the car. All the pieces are there, it’s just math and NN work.
Yeah, driving on snow is a completely different beast. Winter tires are a must if you want to have any kind of traction (and no, 4WD / AWD is not a suitable replacement), and you definitely have to avoid sudden moves (my dad taught me to "drive with an egg under your foot"). Adjusting speed to the traction conditions is something I suspect a computer could do much better than I.
Having the lane markings be covered is also a challenge, but we usually find the lanes by looking at the road more broadly. If there is any kind of traffic, you'll get "tracks" pretty quick, too.
I strongly suspect we'll need a lot more AI training before we see a good FSD solution for colder climates, as it's very different from summer driving.
The mistake most people with AWD make can be summed up in one of my favorite phrases: even two wheel drive has four wheel braking. Winter driving is all about grip, having the extra drive wheels does nothing to help your tires maintain mechanical grip on the road.
Winter tires, once you've seen / touched them, are visibly different: different tread, way softer rubber, etc. I also pretty much always use studded tires, as we've been getting a lot of freezing rain in the last few years, and the one accident I've ever had was due to that (fell in the ditch while driving at ~30 km/h. I had a hard time _walking_ on the road without slipping).
The flip side is that they're more likely to skid on pavement, especially in summer. That's why you have both.
> Winter tires, once you've seen / touched them, are visibly different: different tread, way softer rubber, etc.
Exactly, tires are the single most important element of winter driving. If you live in a place with mild winters all season tires can get you far, but if you get a lot of snow or ice seasonal tires are critical.
Cold tires are slick tires that will slide all over hell and back, the tread on winter tires is designed to better move slush and snow out of the way so the contact patch can hopefully reach the road surface; while the softer rubber heats easier so it can get to a nice 'grippy' temperature and grab hold of the road even in conditions where the wet and cold are trying to sap heat from it.
During the warmer months these winter tires will be less effective because they have smaller contact patches (due to more/deeper grooves in the tread to drive moisture away), and will actually get hotter than desired (softer rubber) causing a loss of grip as the rubber will want to start shearing off instead.
Everyone needs to remember no matter what fancy engine, electronics, etc. your car has - your tires are the only thing that actually touches the ground. Acceleration, braking and turning are all highly depending on their ability to grip the ground (and withstand the forces being placed on them) - so make sure you have the right ones and don't put off replacing ones that are nearing the end of their life.
I think the AWD helps a lot more with starting out. My old FWD car (no snow tires) would not be able to stop at a stop light that was on an incline and then start again without spinning the tires forever. The same thing with my Subaru (AT tires) and there’s no slip at all.
FWD cars can't start on an incline in the rain, either. It's just a fundamental problem of that architecture. Electronic traction control can help somewhat.
None? Seen plenty of FWD stuck at the bottom of a hill that an AWD can climb. As you might imagine hanging out at the bottom of a hill is a great way to get hit.
It helps a ton any time you have to touch slush. Which is often.
People aren't stupid. There's a reason why everyone who can swing it buys an AWD to commute and not everyone buys snow tires. The snow tire is still gonna be slush-o-planing any time you go onto highway slush of appreciable depth. Heck a mud tire with big fat grooves is your best bet when the tire is on that. There is no tire (suitable for general use, I'm sure the Florida mud racing crowd has something up their sleeve) that can handle that without planing. There just isn't enough space between the tread blocks for all that viscous fluid to be forced through (assuming you're going like 20+mph) and you end up on top of it.
Being able to change lanes and enter traffic from side streets/driveways with less slip is just a soooooo much better experience than not being able to do those things gracefully but in exchange for a slightly a better worst case stopping distance in fairly rare conditions (ice and fresh powder).
Commuting in snow is nothing like those commercials where they drive on ice. All the traction in the world won't help you when your problem is that the tire isn't making contact with the road. Driving on ice and fresh powder are so rare it's no surprise that consumers don't optimize for that. Sure you can put snow tires on your AWD but the lions share of improvement in the conditions people actually encounter comes from AWD.
You don't actually have to touch the pavement. Deep, aggressive, self-clearing tread seems to be enough. I put an ice spiker pro on my bike and slush goes from ice rink to no big deal, whether or not I touch pavement.
I speculate most snow & slush under compression has some shear strength to offer even though there is little dry friction.
The weight of your vehicle and heat in your tires (plus the heat generated by friction) causes slush, snow and light ice to melt. If your tires have aggressive enough treads they are just pushing the water away and making contact with the road surface.
If conditions are bad enough to cause a FWD or RWD car to hydroplane then there’s nothing AWD can do to help it given the same tires. Grip is all that matters when it comes to staying put on the road, and that comes entirely from your tires on normal road cars (most road cars aerodynamic profiles generate lift rather than downforce, so you aren’t getting any aero grip unless you are driving a sports/supercar with aero elements).
Hydroplaning means you have a layer of water between your tires and the road surface. Regardless of whether you have two driven wheels or four, if they aren't making contact with the road you aren't putting any power down and your wheels are just spinning. The only thing that can help you is better tires that can move more water out from under them, hopefully giving you a contact patch with the surface.
Imagine you're driving highway speed and crossing some slush to change lanes.
As you drive through that slush would your rather be trying to put all the vehicle's motive power through two tires or four?
Do you dispute that doubling the number of tires moving the vehicle forward will drastically reduce the ease with which the vehicle breaks traction crossing over the slush?
> Do you dispute that doubling the number of tires moving the vehicle forward will drastically reduce the ease with which the vehicle breaks traction crossing over the slush?
Grip and traction are two words for the same thing, the only thing that determines grip are the tires plus the weight (and distribution thereof) they are carrying. So, yes, having AWD does nothing to help grip - that's a mechanical force between your tires and the road.
AWD helps utilize grip, by distributing torque to all four wheels. It helps acceleration by avoiding wheelspin where two driven wheels don't have the grip, but nothing else.
Edit: Clearance is more important when trying to cross over piles of shit on the road anyway. It doesn’t matter how many wheels are driven on your car if you bottom out.
AWD helps when you're under load going around a corner because the traction is distributed over four tires not just two. And AWD cars tend to have better weight balance.
I've also noticed when driving AWD cars on sand and dirt they are stable at higher speeds than FWD.
Traction (mechanical grip) is always distributed among four tires according to physics of center of mass, velocity and angle of travel - AWD does not change that. That said, AWD will let you put power down earlier into a corner because the extra drive wheels will help you change your angular momentum without sending you straight into a slide - but you also put extra lateral load onto your tires doing this so it's best to continue to drive like you just have RWD.
Absolutely not, you can’t change direction without grip - angular momentum and weight are what cause lateral loads, after all.
The thing is, AWD makes it easier to get going by letting you put engine torque through all four tires. The same set of tires has the same mechanical grip regardless of how many wheels are driven, so if a FWD car would spin taking a corner at X speed due to lateral load so will your AWD car (and the AWD will have an easier time doing it).
I suppose that’s good advice because many people overdrive their vehicle’s AWD in winter but it’s not explicitly correct. Real life as usual is more complicated.
Braking in AWD scenario and traction are not same as your example and depends on the power split between the front and rear and how the 2 are coupled together. In 4WD with the 50/50 split, the drive shafts are locked together through the differential and braking is much more effective in part due to the fact that the braking is more evenly distributed among the 4 wheels since the drive shafts are locked. Also when you lock up the brakes in 4WD usually 2 wheels on the diagonal keep turning, again due to how the differentials work. This is not the case in FWD.
> braking is more evenly distributed among the 4 wheels since the drive shafts are locked.
There is something seriously wrong about either my understanding of how cars work, or yours.
I have brake disks and brake pads on all 4 wheels of my (FWD) car; pretty much every car does. How would a drive-shaft help "distribute" braking power? The limiting factor here isn't the force each brake can exert on their wheel (they're pretty much always able to lock up), it's the traction between the wheel and the ground (the tire).
In my mind, an "ideal" emergency braking scenario would be all 4 wheels braking at the maximum force they can without slipping, and that can be done much better by the ABS system when they're all disconnected from each other. A driveshaft connecting them would be irrelevant at best, but more likely it's a nuisance.
4wd/awd is pretty complicated. Some depend mostly on hardware (like subaru), others try to use sensors and torque control and/or brake control to put the torque where it's needed.
AWD is complicated with 3 differentials that can be a mix of open, limited slip, or actively controlled (by torque split or brake based system). It's especially tricky since it's not the actual traction that matters as much as the driver perceived traction and predictability. I've drivin AWD systems with snow tires that seemed to have more traction than my subaru with all seasons, but ended up driving more slowly because I couldn't predict what it was going to do. Subaru seems to communicate well with the driver, and making sliding just not a big deal. Other cars make it harder to tell what a given throttle/braking input is going to result in and can trigger heavy front or rear bias based on what the sensors/software tell it to do. It drives me nuts when a car acts completely differently in high traction and low traction conditions, especially when it changes one second to the next. Some Subaru folks make fun of other systems that they call "slip and grip", where there's a heavy FWD bias (minimal torque to the rear), until AFTER you start slipping.
My WRX had a limited slip in the rear, center diff was a viscous coupling (100% mechanical/no software), and a open front diff. It was reassuring, never surprised me, and even when significant slipping I could always put the car exactly where I wanted. Even when crab crawling sideways in a parking lot at 2 mph it never surprised me. Numerous Toyota, Ford, Honda, and BMW cars I tried would try to outsmart me. Limiting throttle, moving the torque around, limiting braking forces, etc. Last thing I want when climbing a hill is the car to cut the throttle because one of 4 wheels slipped for 50ms.
Consumer reports tested a Subaru forester, Toyota RAV 4, and Honda CRV in heavy snow conditions and concluded the Subaru was a clear winner despite them all having similar dry road performance and AWD.
I think for normal snowy roads Subaru does it best with a mostly manual system, the ABS and traction control has a light touch that can be mostly avoided. But for heavy offroad the best thing to do (in certain messy conditions) is lock all differentials, which is something the Subaru can't do.
I can certainly appreciate the "doesn't surprise me" aspect, but I still don't see how AWD / 4WD will help you when braking.
It's obvious that it'll help when accelerating, and perhaps even when turning, with a (more) even application of torque on all wheels.
My main gripe with it is that it often gives drivers a false sense of security (as they got going easily), and they'll drive too fast for the conditions. Just look at some parts of this sub-thread, where people say "I don't need winter tires, I have AWD / 4WD".
If you need to brake in an emergency on snow / slush / ice, nothing is going to help you except your tires (and perhaps some smart ABS).
Well, good ABS is hard. Having locked differentials can make the most of a really messy situation. Thus the reason for locked differentials (for going or stopping).
There's a wide variety of software solutions as well, some work better than others. It's a non trivial problem. For instance. In normal conditions less braking when any wheel slips is best. In slippery conditions for stopping or going some slipping is best. In really messy conditions it can be best to not keep applying torque (in the positive or negative direction) without any "smarts". For emergency braking in the snow it can be best to just lock them up and have snow build up in front of the tires.
To see an example of how big a difference this makes, check out
I agree that AWD leads to overconfidence to some, I've been driving at 35 mph on the highway (with a 65 limit) in my Subaru (with good tires) in bad conditions and had giant SUV blast past me at 10-20 mph over what I was doing ... only to find them in a snow bank later.
Doesn't mean that AWD wasn't a huge help. My first AWD was a result of me spinning out in a FWD around lake tahoe in poor conditions, in heavy traffic, at 10 mph or so. My front right tire lost traction, the open front differential put all power to the front right tire, and it spun me into a snow bank (just 18" away") in a blink. I was quite impressed how much rotational acceleration could happen in a blink of an eye. The open differential meant I was basically 1 wheel drive (when slippery), the car I replaced it with (Subaru WRX) had 3 wheel drive (when slippery) and was MUCH more predictable. In fact sliding in the snow would bring a smile to my face... and I changed my weekly commute (6 7200 foot passes) to maximize the snow involved.
And yes I'd MUCH rather have all seasons and AWD instead of snow tires and FWD. It's not the ultimate traction as much as being predictable. Sure all seasons have less traction, but it's way more predictable and gives much better control when sliding. FWD, even with snow tires can be very unpredictable, and can easily spin a car when the traction is low. Subaru + snow tires seemed overkill for the roads I was on. Sure Subaru + all seasons might be safe in some conditions at 35 mph, snow tires might raise that to 40 or 45 mph, but FWD regardless of tires would mean much less, and Caltrans would require the pain of installing and uninstalling chains and worries about chain failures and/or damage to the car.
Even with all seasons, my main fear with a Subaru (I've had two) is hitting enough snow to life the tires off the ground.
I find that a very weird statement. Driving a FWD, especially when climbing, shows how limited the grip of FWD is. AWD distributes the power to twice as many wheels (often 3 times actually) and loses grip less often.
Sure both cars would descend the same, but you can always lower your speed to decrease the amount of traction you need. When climbing there's no such options for FWD.
My Prius can climb a hill like a champ, until the battery is drained and the poor ICE is all that's turning the driveshaft. Torque matters when it comes to hills, not the number of wheels connected to a driveshaft.
Torque is highly relevant, the entire 'benefit' of AWD is being able to distribute torque to all four wheels instead of two. Thing is, with winter tires even a Prius has enough mechanical grip to put the extra torque from the MG's down and many AWD vehicles won't be able to max theirs out with all season tires.
When it comes to winter driving the entire perceived benefit of AWD vehicles for commuter use goes out the window as soon as you need seasonal tires, basically. Yes, it will still better help put the power down given equal tires - but at that point the extra cost of the AWD system isn't giving you too much benefit if you are only buying it for winter driving.
Yes, the distribution of torque is very important and is the reason for AWD. But the amount of torque is not. When it's treacherous out even 5% of a normal cars HP/Torque is plenty.
I've heard the FWD + snow tires beats AWD + all seasons before and personally I don't believe it, but it's a pretty subjective thing. The big win for AWD is predictability and control, not that I want to go around corners sideways at 100 mph rally style. Having control over the car, even when sliding, without any rear or front bias just makes for a MUCH more controllable car.
My Subaru forester with AWD and all seasons handled numerous treacherous conditions, sometimes so bad that every highway around me was closed. I used to do 6 7200 foot passes every weekend, often in the middle of storms dropping 2-3 feet. My main problem on the roads was others that don't know how to drive and block the roads.
I've driven roads so slick that going too slow would cause you to slide off the road. I'd have to apply torque, while sliding, to keep the 10-15 mph required to avoid sliding to the inside of every banked turn. Which would be MUCH harder with FWD or RWD.
Amusingly at chain control points, often the workers would look back in the line, see my Subaru and pull me out of line and let me go through with a smile and a wave. Presumably because there are (or weren't anyways) any non-AWD Subarus (this was before the BRX) and Subarus aren't the typical cars getting stuck and causing problems.
Sure AWD and snow tires is even better, just seems overkill for the highways in the high sierras.
The millisecond scale tiny adjustments to throttle position and correct braking force to apply to a specific wheel in slippery conditions computers are absolutely better than human drivers already. Systems like ABS, stability control and traction control have done wonders to keep people out of ditches already.
The long term, macroscopic decision making involved in driving computers are nowhere near ready to drive better than I can in winter (or even in perfect conditions). I love my model 3, but the autopilot is simply not a good driver. It would fail any minimally rigorous driving test because it makes rookie mistakes and only drives in the exact moment. It cannot plan ahead more than even a handful of seconds. It stays in other driver's blindspots when there is plenty of room to adjust the relative positions of the cars. It accelerates into gaps that are about to not exist because other drivers' small hints scream they are about to change lanes, even if they haven't signalled yet. It gets confused about speed signs, often reading the off-ramp maximum as the speed and slams on the brakes. It plants itself in the dead centre of the lane even when other drivers encroach and then when it decides the situation has become dangerous it swerves violently and applies applies significant braking. Experienced, safe drivers give ground within the lane, preventing the dangerous situation from ever developing, and if the other driver does fully enter their lane are ready to calmly move over without creating an entirely different collision.
The Model S and X both have two screens -- one behind the steering wheel, and one in a portrait orientation to the right of the driver. The Model 3 and Y are the ones that have only a single screen to the right of the driver in a landscape orientation.
Just knowing where to point the car when you're going to smash into something is half the battle with bad winter conditions. I've hit a handful of ditches and snowy fields in my time that could have been much worse if I had veered right instead of left.
My wife lets me drive her Model 3 with FSD when I'm good. I drove it quite a bit this past weekend, in lovely Dallas winter conditions. (No snow, no fog, no rain.) While snow is obviously a challenge, the Tesla is so far from being able to handle routine driving in unchallenging conditions, it would be a big miscalculation to think, for example, that Teslas drive with FSD just fine in normal conditions. I would estimate that I take over about once every 3-5 miles in absolutely routine conditions.
I love my Model 3, but I cannot overstate how little confidence I have that Tesla is anywhere near to full self driving.
Just this weekend I drove a 600 mile round trip. It was cold weather with some snow on the shoulders, but the actual road surfaces had no snow. I was mainly using just the cruise control, and no auto-steer. Even with just cruise, the car gets "spooked" from time to time and will suddenly jerk to a slower speed for absolutely no reason, even during the day and with no precipitation. At one point it did this several times within the span of 10 minutes, which was such an unpleasant experience that it made me want to stop using cruise entirely. Besides being jarring for the people in the car, it makes me worry that anyone tailgating will rear end me.
Besides that, the rain/sleet that was falling from time to time apparently blocked the camera, so at one point the car wouldn't even let me use cruise any more.
The car is tons of fun to drive and I don't regret the purchase at all. But I really don't like the overpromising on full self-driving.
EDIT: bmcahren pointed out to me that the new FSD beta software merges images from multiple angles, which my software does not do. I don't know much about that, but it sounds like the kind of change that could lead to a noticeably better result than what I experienced.
no, there’s just a fascinating phenomenon where Tesla owners appear to shrug off bad quality control, bad service, buggy software and controls, safety issues, etc en masse because they love their car.
I think it happens in just about every realm when you get close to the innovative edge, especially in tech: you can be really excited about some aspects of $SHINY_NEW_THING while recognizing that it has serious drawbacks in other realms--possibly realms that people expect $SHINY_NEW_THING to excel at as a contender in the space.
(inb4 "Sure it goes fast, but the panel gaps!")
Personally I love Tesla-the-company and my Tesla vehicle because of the innovation it's driving in the space--and I'm willing to tolerate the ways it falls short due to how incredible it is in other ways.... but I do try to be up front about those shortcomings when I discuss it, at the risk of being one of those "I love my Tesla, but..." people.
The car is comfortable, responsive, stylish, fun, and feels great to drive. The cruise control is fantastic in stop-and-go traffic. I always look forward to driving it. And then every once in a while does something that drives me absolutely crazy, for a minute, and then goes back to being fun again.
the biggest issue is the service model not anything related to the car unless you got a lemon.
they aren’t scaling service centers fast enough to keep up with an extra 200k cars added on the road every quarter, so service meaningfully deteriorates every week as more cars are sold and existing cars age. it simply doesn’t work unless they were committed to building out 2x-3x the SCs they have now but nothing indicates that’s on the roadmap at all.
- increased service locations 12% YoY
- increased mobile service fleet 8% YoY
- increased supercharging stations and connectors 32% YoY
They are clearly building more service.
During quarterly conference call Musk stated that he believes service is the biggest driver of sales and that Tesla will invest in that. So contrary to what you claim it's on the roadmap.
Many people are over the moon with Tesla service given that mobile service comes to you and they can do fixes without you even there. That's something no other car maker does at scale, as far as I know.
Did you have a personal experience with Tesla's bad service or is it just amplifying FUD from other people?
growing service centers 12% YoY when they added 50% of cars they’ve ever sold this year alone and plan to sell close to 1M next year (100% YoY) is the definition of inadequate.
Every car is a game of tradeoffs. Any other EV you can buy today (including the Mach-E) will have you trading off build quality and access to good service (probably; some dealerships suck at servicing EVs) for extremely slow software, low range, and lack of supercharging access. MKBHD has a good video on this that came out yesterday - https://youtu.be/pJlA2J92Fw4
I have owned 2 cars that I bought the first year the model came out.
The Model 3 I have had a few dumb issues like, if you run the paint app sometimes the gps location on the map doesn't update until you close the paint app and your displayed location lerps from the old point to the new one over a minute or so.
The Mini had an eco mode that would occasionally shut off the engine entirely at highway speed and not allow restart until you came to a full stop and put it in park. Eventually, they just disabled eco mode in software Which took a few weeks of me working from the mini dealer. There was display of how many miles you had gained that reset when you filled the gas tank. My record is 42 Miles gained in eco. After the final update to the software, which was a full day in the garage, My best was 8.
One of those was worth running stories in the press about.
I guess what I'm saying here is that there is a deep vein of shadenfreude in the news about Tesla. A single Tesla on fire is a big national and some times international news story. I have stopped even reading the damn things. It seems like a weird combination of them being seen as rich people cars, early adopter cars, and liberal hippy cars. Everyone can hate someone on the spectrum. Typically, the issue in articles about Tesla flaws is so isolated and dumb that it exists solely to generate clicks. The last article about Tesla I clicked through stated that consumer reports had downgraded their durability rating due to the discovery of single human hair in the paint of a model Y. I get that it sucks. But, I don't get why I should care about it.
I own a Model 3 because I needed a four door car because I had kids, electric seemed like a net win, the local Bolt dealer told me not to buy their car, and the range let me drive to the grandparents houses without a charge. I recognize that there is stuff about the model 3 and Tesla that sucks but overall it has been a pretty solid car for the last 2 years for me.
Would I recommend one to you? Yeah. I don't think there is a company that offers the combination of electric car, mature tech, range, and charging network at a similar price tag.
I don't think your anecdote lends any evidence to the claim they're not close. Several reasons:
- You weren't using the actual FSD software, but the live production autopilot, so it's Apples to Oranges.
- It's not clear that the transition from low quality to high quality full self driving will be linear. Most claims around anecdotes come with an unspoken assumption we ought to see steady improvement over time vs a step-function like behavior. The latter seems somewhat possible given the nature of how improvements in AI systems seem to happen.
- Beyond all this, without understanding the underlying cause of the behavior you mention, it's hard to say from a risk perspective if the car made the right choice. Unless you have a clear understanding of the data and mechanisms that led to the speed reduction, it's hard to be sure it's a bug beyond not doing such transitions in a way that are comfortable for passengers. (My understanding is adding smoothing to the control algorithm is somewhat separate from overall decisionmaking, and hence poor control algorithms may mislead you into thinking it's more dangerous.)
Also a Tesla owner, on latest AP hardware: I would be amazed if the FSD update manages to solve all the problems, I have been through a few updates where things noticeably improved but we now have to pay Tesla money because while using autopark it just completely reversed itself (at a pretty high speed) into an object. It was insane. We’ve had weird AP issues too and now we don’t feel confident using any of those features.
I agree Tesla's shipped software that doesn't inspire confidence in the current version. I only use it on highways and remain fully engaged. (If I had to guess, in that scenario it does provide a marginal improvement in safety over fully manual driving.)
However I don't think lack of confidence in the current software, particularly the production autopilot software, is really giving me much signal in helping predict if Tesla will have FSD anytime soon or not. Now that FSD (at least in Phoenix conditions) is known to be possible the prior that it will never be achieved by Tesla to me is near zero, beyond the chance that LIDAR is strictly necessary and they have a doomed hardware stack. I think there is a pretty big range of possible timelines though - I could see it happening in as soon as 18 months or as late as 10 years depending on how poorly things sift out with their hardware stack.
I mentioned this in another comment, but it's relevant here; OP wasn't using Autosteer or Autopilot, they were using cruise control: a miles more simple system--and that system behaved in a way that seriously reduced confidence in it.
(For the record, I've experienced exactly the same behavior in my M3 as well.)
So, if we can't trust something as simple as cruise control, and we've seen similar confidence-eroding issues in the autopilot features Tesla has already deemed "good enough" for a live release, I think it is 100% reasonable to be skeptical that Tesla is anywhere close to "solving self-driving."
They have a history of overblown claims and faulty software in this realm--as much as I adore my car and the disruption Tesla has done to the industry, Tesla does not deserve the benefit of the doubt in this realm, especially not with safety-critical systems.
The transition from low quality to high quality driving will be sublinear, not linear or greater than linear. This has been the case at every other self driving company. You get big gains from working on low hanging fruit early on in the process, then there's a long, arduous march to make improvements once those fruit are gone and avoid regressions. That's why Waymo, Cruise, Zoox, etc. Have been able to do impressive demos for the last decade, but it took Waymo a full decade to actually launch a public ride-sharing service.
Before you bring up Tesla's "Fleet Learning", it's not used at all outside of gathering perception training data for labelers. greentheonly got access to a dev build of the FSD beta - ML is just used to generate a representation of the current scene - all of the driving and planning logic is rules-based and hardcoded by engineers.
Seems unknowable still since Waymo and Cruise are now both deploying fully autonomous vehicles ahead of most peoples' expectations. If graphed, I'd imagine this implies a non-linear curve given these expectations were grounded in linearity imo. So it's under-determined basically.
They're deploying AVs way (1-2 years, at least) behind their own schedules, by their own admission. Ask anyone inside of the industry (including Musk) and they'll tell you that the problem is the long march of nines I mentioned earlier.
This is also why you saw a bunch of AV startups with < 50 employees that got big initial funding rounds but folded after a couple of years - it's pretty easy to grab a few dozen grad students, slap some LIDAR on a car, encode basic road rules, and get a demo working that's compelling to investors. Once you get past that initial pitch though, you get into the weeds of trying to progress without causing regressions elsewhere.
Also, what makes you think those expectations were grounded in linearity? Waymo went from an initial project at Google in early 2009 to driving 10 complex 100 mile routes without intervention in mid 2010 - on the surface, it would have seemed they were 95% of the way there. Improvement has been sublinear for a very long time and everyone in the AV industry is aware of it.
The linear assumption I'm referring to is forwards-looking not backwards-looking. Up until say last year I think the consensus was that a production-deployed FSD vehicle was ~5 years away. It shipped this year for both Waymo and Cruise. Unless you're arguing that the pessimists and optimists have just been arguing over the slope of the line, not the curvature (which may be fair.) If that surprise is any indication it leads me to think a similar consensus-breaking event may happen with Tesla, but time will tell when that is, since as you state they are behind in several ways.
It's the same people but FSD was a rewrite of the system.
The rewrite was to do image recognition and inference not from individual still images (pre-FSD) but from video feed from all cameras. This was enabled by the FSD chip Tesla started putting in cars about a year ago.
Plus the improvements that come simply from gathering more data from cars and using it to improve the training of neural network.
It's documented in Musk's tweets and Karpathy's talks.
Your position here is, essentially, the same as a Q believer. Just trust in Elon, wait for the big reveal. Despite a complete lack of evidence in favor of the idea, and also despite an abundance of evidence to the contrary, FSD will emerge fully formed in the near future, without any intermediate signs of progress.
You have a reading comprehension problem. Saying "what you just said does not confer evidence to your claim" (which is what I said) is not the same thing as saying "I am 100% certain of the counterclaim." Responses like this one are why discourse on the Internet has fallen so far. The analogy to Q lands you squarely in the "lol fag" tier of responses.
Can you fill me in on where the levels of Tesla software are? The version of the OP who you are calling out as "apples" (the old, legacy one), what SAE autonomous level is that one? And the "oranges" one that is new and beta, is that a different level? If so, which level?
Different features of Tesla's software have different levels. As a whole, it's level 2.
If you restrict it to certain parts of driving, eg TACC, that might be level 4 (doesn't conform to the chart though). It can handle everything in it's domain, albeit in a rudimentary way, so a human doesn't always have to shadow it, but they do need to be ready to take over in some cases.
The apples to oranges bit is using the cruise control software to make guesses about FSD. I'm not certain but it would be very surprising if the software that provides basic cruise control shares much of the underlying algorithms and models as FSD, and if it does I'd be surprised if it's using models that aren't quite a bit older.
> Beyond all this, without understanding the underlying cause of the behavior you mention, it's hard to say from a risk perspective if the car made the right choice.
What argument could there be that suddenly slowing down on the freeway with no visible obstruction is the right choice?
It does not avoid any danger, but it creates a danger of getting rear-ended by cars behind you.
> It's not clear that the transition from low quality to high quality full self driving will be linear. Most claims around anecdotes come with an unspoken assumption we ought to see steady improvement over time vs a step-function like behavior. The latter seems somewhat possible given the nature of how improvements in AI systems seem to happen.
That is true, but it's actually an argument _against_ the assumption of full self-driving capability being just around the corner. Because that assumption itself is based solely on observed recent progress, which - as you correcly state - can be very misleading when used as a means to extrapolate the future due to the tendency of the AI field to have long stretches of very little to nearly no progress at all, divided by huge leaps followed by stretches of very fast progress in specific sub-areas of the field. We clearly have seen such a stretch recently, triggered by the deep learning idea and enabled by an explosion in parallel computing capability over the course of the last decade.
The problem is: you have to assume "something", because otherwise you simply can't make any predictions. Either assume progress is linear (which seems to be the instinctive default most people use if they don't know much about a particular field) or assume it's non-linear. In case of the latter, you can choose between assuming exponential progress to apply or a step-wise, non-continuous kind of progress. It's a bad idea to not assume one of these to be the case if you want to make a prediction, because it's not possible to predict anything without that assumption.
I mean, you can of course refrain from making any predictions. But I have never met a single person who could resist the instinctive urge to extrapolate current developments into the future. It is as much a core part of being human to do this as it is to breathe air.
It's the assumption of linearity in combination with anecdotes that make the prediction weak, not the assumption itself. If you hold that assumption and use a variety of data to justify it and extrapolate from that's different than what I was criticizing.
> You might as well be comparing your experience driving a BMW or Cadillac ....
But OP isn't--he's comparing the current live Tesla features with paywall'd / betawall'd features on the same car, from the same manufacturer. I cannot think of a more relevant comparison! I think it's completely reasonable to doubt Tesla's (and Elon's) claims in this realm, when the existing, public examples of this technology fall far, far short of the kind of confidence we'd want to have in these systems.
> Your experience on the old tech stack is irrelevant.
I 100% disagree. If Tesla can't even get a basic function like cruise control to work well enough that drivers are confident in it, why would I trust that they can get self drving to work well? Especially following things like the Summon beta, or similar fiascos?
I still get twitchy when my M3 comes up on an off ramp it's not supposed to take on autosteer--I've had the same behavior as OP (sudden unexpected speed decrease) happen far too many times for me to have confidence in that system. In turn, I have even less faith in the more complicated systems, like autosteer or FSD.
That's quite reassuring to hear. It's surprising to me that the software wouldn't have been stereoscopically merging cameras already, but adding that capability does sound like the kind of thing that could actually lead to an entirely different level of performance.
There is a history of overpromising that will be still leave me somewhat skeptical until I see it. But it does sound like a move in the right direction.
Perhaps, but if they can't get an orders of magnitude simpler adaptive cruise control right (not just right, but perfect), it doesn't lend much confidence that they'll get FSD right any time soon.
The protestations to the contrary here reek of defensive fanboy/apologist behavior.
> Your experience on the old tech stack is irrelevant.
It's cruise control. If you couldn't fry an egg without burning it, forgive me for being a little skeptical when you claim that you can make award-winning souffles after having attended some cooking classes.
I think phantom braking with TACC is a function of it being alpha software. With the beta stuff (everything else), the driver is expected to always be alert and intervene where appropriate, so Tesla doesn't have to be as aggressive on the safety side.
As it relates to the features, I agree that they aren't ready for what most drivers expect, either because they aren't reliable enough (beta) it because they aren't confident enough (TACC). With that said, FSD is a feature set, not a level of reliability. As a whole it eventually needs to reach a level of reliability, but there's no specific time horizon.
For Tesla, a feature moving from beta to alpha is what indicates reliability. In addition, that reliability may not come with the degree of confidence most people expect from a driver. Safe FSD could just end up as the AI equivalent to a student driver. It'll get you someplace safely, but there may be phantom braking, excessive waiting, and so on. If that's the case it would be nice if we could enable a beta version with more "normal" driving that we would have to keep an eye on.
> The Tesla Model S in this video is equipped with summer-focused all-season tires, so traction and handling could’ve been better with the winter tires.
This is a bit of a nonstarter for gauging snowy weather performance, in my opinion. While comparisons can still be drawn between human and FSD drivers using the same (incredibly unsuitable) tires, handicapping the system doesn't provide a great measure of its prowess or challenges in "snowy conditions."
A bit of an aside, but few driving-related topics grind my gears more than people who live in snowy/icy climates driving with cheap (or just bald) all-seasons and complaining that their car "isn't good" in the snow so they need a bigger 4x4, which is of course subsequently kept on stock all-seasons for only a marginal improvement.
The problem with FSD is that with humans, if they can’t see the road because a window has ice on it, they would go and clean it till they can make good enough judgement. But FSD just keep on going with less performance, whether you know it or not.
It would be a cool feature if cars using autopilot would have a notification device somewhere so I could see they're currently autopiloting as I drive behind them so I can give them more room (since they won't behave human-like).
Nah, all auto-driving cars perform poorly in stop and go traffic. I can usually maintain average velocity around 2-5 mph so I don't have to stop and go, but auto-driving cars will accelerate and brake hard, which makes it hard for me to judge whether we're out of the present traffic and I'm holding up people behind by not accelerating, or whether I should continue the crawl.
Correct following distance in that traffic is ~3-4 metres (~10-12 feet) which is easy to maintain against a human or against an auto-car you know is auto-driving, but that takes a few seconds to calibrate that it's auto-driving.
I've made many rush hour commutes from my home in Glen Park to SOMA with very few brake-use events¹ and travel team equivalent to original Google Maps prediction.
¹ One of the many metrics I attempt to min-max during all of my drives
IMHO, tesla and others do the autopilot thing wrong and the right way to do it is to predict the trajectory for the next 5 sec and display it on the HUD-like screen projected to the windshield. This way the driver would see where the car is about to go and take control only when the predicted trajectory is wrong. This would solve the trust problem: the driver would know what's going on inside the car's "brains" and wouldn't need to act like an instructor to a epileptic student that can collapse at any moment. This would mostly solve the liability problem as well. The downside is that such autopilot can't leave truck drivers without a job.
157 comments
[ 4.8 ms ] story [ 202 ms ] threadhttps://youtu.be/qj9t5GTeKNE - Left turn view from both B-Pillar And Headlight Camera Perspective
There’s also James Locke which does drives a few routes on each new version - https://youtube.com/c/JamesLocke/videos
The reliance on maps means Waymo can drive in Wayless places, but more reliably in those places.
There have only been a handful of Tesla accidents while autopilot was engaged where the vehicle was catastrophically damaged, so it’s a bit of an edge case.
https://www.edmunds.com/car-technology/car-black-box-recorde...
I don’t have much experience in SDCs, but did some computer vision/ robotics work previously.
I truly think that Tesla’s approach, if it works, will win in the long run (ie - next 15-20 years, I think the breakthrough for true L5 isn’t here yet - but will be in improvements in one-shot learning over the next decade).
Winter driving is a completely different beast than driving out in Arizona. You have to account for snow banks, slipperiness, lidar obstruction, and control itself - which is much harder.
How many thousands of miles of snow driving data does Tesla possess, compared to the #2 player?
Tesla’s approach of vision based, AP disengagement-led training will scale better, precisely due to what you’re looking at in this article - yeah, this is an unimpressive demo - but winter driving, rainy driving, dirt road driving are all general cases for Tesla, while they’re special cases for other companies (from what’s been revealed; Comma might be an exception).
Most people drive in the place they live, and drive on the highway to visit their friends and parents. It’s easy to be deceived into thinking driving is a small problem space of stop signs and lane changes. But even within just North America, there is enormous variety.
There is massive risk involved - these folks are pushing NNs further than any application I’ve seen, but with the last few releases, it suddenly doesn’t look all that crazy.
Secondly, combining rules based systems and neural networks is difficult. You end up with the N * M problem, and inevitably end up missing cases. It makes sense to do the absolute basic safety features as rules, and Tesla might be doing this, but all it takes for Tesla’s NN approach to pay off is one leap in explainability. There is an entire industry working on this - and Tesla will be the single biggest benefactor.
I don’t know who will win L5 - but I’m putting my bets on Tesla for L4.
It depends. If they're not recording the right things, they have the same 0 bytes of data as everyone else.
There's a not insignificant chance that their sensor data is insufficient to perform the tasks, and the recorded data won't be usable for training a system with whatever sensor is needed; in which case, no advantage for logging customer movements for N years.
On the other hand, Waymo et. al. are not risking anything by using lidars. If they turn out to be redundant, it should be easy to remove them cheaply.
BTW, all companies that can, use lidars. Even MobileEye says that their L4 system will use lidars and they've been working on a camera based system for 2 decades. Tesla doesn't use lidars because they can't.
1. Lidar is expensive for everyone. It’s not like Tesla will pay more than other manufacturers. They’d prefer not to use Lidar, but if cameras just don’t cut it, I don't think they would die on that hill.
2. I’m sure there is some fine print around the current FSD being for camera based sensors only. They can easily say the car tops off at L4 for cameras, and L5 needs Lidar, if we come to that bridge
Why isn't every car body made from carbon fiber? It's lighter and sturdier that alternatives. It's also more expensive so you can put it in $60k car but not in $25k car.
Radar and ultrasonic sensors are cheap, low power and plentiful.
Putting several lidars on a car would be too expensive, would use too much power (less range) and there isn't a supply chain to make 2 million a year just for Tesla (500k cars x 4 lidars per car).
It's just isn't possible to put a lidar today on a mass market car and Tesla wants to solve self-driving before lidar becomes cheap enough.
> all it takes for Tesla’s NN approach to pay off is one leap in explainability.
It's not vision and disengagement-led training that's conferring the scale advantages you're talking about, because other AV companies actually do train on and attempt to drive in various kinds of weather, and safety drivers do the same thing normal drivers do except that they're more trained. The difference is the approach Tesla has to safety (vs the others). Which is to say: Tesla doesn't worry about training their supervising drivers, at the risk of those folks' lives.
Tesla also doesn't seem to worry about _proving_ the safety of their systems: explainability isn't simply going to be a matter of a new algorithmic breakthrough. How do you prove that your software is safe if you don't sample heavily from different operating domains you want your car to work in, and then either simulate on those miles or actually test your car? Before letting a driver behind the wheel of an autopilot, companies like Cruise/Waymo/Aurora all do manual driving first in a new domain, and then move to safety drivers (even 2 to start, for extra safety). This is not because their software is incapable of doing the same things as Tesla, but because it's very worthwhile to sample disengage rates and assess safety criticality to various issues before allowing the car to act autonomously. And even then there are safety nets in place to further mitigate risk. (You're probably right that L5 is a long ways off, but this is how we'll get to L4.)
Of course it's more "scalable" to avoid all this and put the responsibility of risk on users who, as your system improves, become less and less attentive. And so this is what Tesla does. I suspect solving the long tail of issues is going to be more of a challenge for Tesla because at least in my experience it's easier to {convert a naive rules/simple model --> complex model} than to {debug the complex model --> handle failure modes with logic}.
Tesla just decided to jump into the deep end from the beginning.
In self-driving, you basically have 3 parts - perception, prediction and planning. Perception is the easy part (if you have lidars). Tesla and AFAIK others rely heavily on NNs for perception. The real challenge is prediction and planning. To what extent are NNs involved here in Tesla's system is unknown, probably not that much. Overall, I don't think that Tesla has a fundamentally different approach than others.
Clearly a ways to go, though better than I would have expected.
Tesla's have 5 inches clearance tops. FSD doesn't need to be able to drive through two foot drifts.
The parking lot he starts in has like three inches of slush, after that there’s a quarter or half inch of packed snow on the road.
What’s left is the driving part, the “how to make the car follow this path” part. I predict this part will be ridiculously good once Tesla gets around to it. Imagine the traction control in your car, now with immediate-response electric motors, and it knows where it wants to go in 3d space (not just “driver is turning wheel to the right”, but actually knowing the arc to follow).
Exciting times.
The moment in the video just after 7m0s seems to contradict this.
So they are almost there with the part others barely even have to do, while the part everyone else is working on will eb ridiculously good once they get round to it?
I'm skeptical.
Or is the software just hardcoded to assume that if it turns the steering wheel at a specific angle that the car will move a specific amount in that direction?
Seems plausible that they could ask all the cars this winter to report instances of wheel slippage, accompanied by video rewound with the spot the wheel slipped on annotated. Could have a prediction for mu at each point on the road ahead of the car. All the pieces are there, it’s just math and NN work.
Having the lane markings be covered is also a challenge, but we usually find the lanes by looking at the road more broadly. If there is any kind of traffic, you'll get "tracks" pretty quick, too.
I strongly suspect we'll need a lot more AI training before we see a good FSD solution for colder climates, as it's very different from summer driving.
Winter tires, once you've seen / touched them, are visibly different: different tread, way softer rubber, etc. I also pretty much always use studded tires, as we've been getting a lot of freezing rain in the last few years, and the one accident I've ever had was due to that (fell in the ditch while driving at ~30 km/h. I had a hard time _walking_ on the road without slipping).
The flip side is that they're more likely to skid on pavement, especially in summer. That's why you have both.
Exactly, tires are the single most important element of winter driving. If you live in a place with mild winters all season tires can get you far, but if you get a lot of snow or ice seasonal tires are critical.
Cold tires are slick tires that will slide all over hell and back, the tread on winter tires is designed to better move slush and snow out of the way so the contact patch can hopefully reach the road surface; while the softer rubber heats easier so it can get to a nice 'grippy' temperature and grab hold of the road even in conditions where the wet and cold are trying to sap heat from it.
During the warmer months these winter tires will be less effective because they have smaller contact patches (due to more/deeper grooves in the tread to drive moisture away), and will actually get hotter than desired (softer rubber) causing a loss of grip as the rubber will want to start shearing off instead.
Everyone needs to remember no matter what fancy engine, electronics, etc. your car has - your tires are the only thing that actually touches the ground. Acceleration, braking and turning are all highly depending on their ability to grip the ground (and withstand the forces being placed on them) - so make sure you have the right ones and don't put off replacing ones that are nearing the end of their life.
In winter situations, it's stopping (and steering) you need to worry about, not going.
(Said as someone who drives a BIG 4x4 with excellent winter tires in a part of Canada where they are required by law)
People aren't stupid. There's a reason why everyone who can swing it buys an AWD to commute and not everyone buys snow tires. The snow tire is still gonna be slush-o-planing any time you go onto highway slush of appreciable depth. Heck a mud tire with big fat grooves is your best bet when the tire is on that. There is no tire (suitable for general use, I'm sure the Florida mud racing crowd has something up their sleeve) that can handle that without planing. There just isn't enough space between the tread blocks for all that viscous fluid to be forced through (assuming you're going like 20+mph) and you end up on top of it.
Being able to change lanes and enter traffic from side streets/driveways with less slip is just a soooooo much better experience than not being able to do those things gracefully but in exchange for a slightly a better worst case stopping distance in fairly rare conditions (ice and fresh powder).
Commuting in snow is nothing like those commercials where they drive on ice. All the traction in the world won't help you when your problem is that the tire isn't making contact with the road. Driving on ice and fresh powder are so rare it's no surprise that consumers don't optimize for that. Sure you can put snow tires on your AWD but the lions share of improvement in the conditions people actually encounter comes from AWD.
I speculate most snow & slush under compression has some shear strength to offer even though there is little dry friction.
If conditions are bad enough to cause a FWD or RWD car to hydroplane then there’s nothing AWD can do to help it given the same tires. Grip is all that matters when it comes to staying put on the road, and that comes entirely from your tires on normal road cars (most road cars aerodynamic profiles generate lift rather than downforce, so you aren’t getting any aero grip unless you are driving a sports/supercar with aero elements).
How can anyone say this with a straight face?
Spreading the motive power out over twice as many tires does a ton to help.
As you drive through that slush would your rather be trying to put all the vehicle's motive power through two tires or four?
Do you dispute that doubling the number of tires moving the vehicle forward will drastically reduce the ease with which the vehicle breaks traction crossing over the slush?
Grip and traction are two words for the same thing, the only thing that determines grip are the tires plus the weight (and distribution thereof) they are carrying. So, yes, having AWD does nothing to help grip - that's a mechanical force between your tires and the road.
AWD helps utilize grip, by distributing torque to all four wheels. It helps acceleration by avoiding wheelspin where two driven wheels don't have the grip, but nothing else.
Edit: Clearance is more important when trying to cross over piles of shit on the road anyway. It doesn’t matter how many wheels are driven on your car if you bottom out.
I've also noticed when driving AWD cars on sand and dirt they are stable at higher speeds than FWD.
The thing is, AWD makes it easier to get going by letting you put engine torque through all four tires. The same set of tires has the same mechanical grip regardless of how many wheels are driven, so if a FWD car would spin taking a corner at X speed due to lateral load so will your AWD car (and the AWD will have an easier time doing it).
Braking in AWD scenario and traction are not same as your example and depends on the power split between the front and rear and how the 2 are coupled together. In 4WD with the 50/50 split, the drive shafts are locked together through the differential and braking is much more effective in part due to the fact that the braking is more evenly distributed among the 4 wheels since the drive shafts are locked. Also when you lock up the brakes in 4WD usually 2 wheels on the diagonal keep turning, again due to how the differentials work. This is not the case in FWD.
There is something seriously wrong about either my understanding of how cars work, or yours.
I have brake disks and brake pads on all 4 wheels of my (FWD) car; pretty much every car does. How would a drive-shaft help "distribute" braking power? The limiting factor here isn't the force each brake can exert on their wheel (they're pretty much always able to lock up), it's the traction between the wheel and the ground (the tire).
In my mind, an "ideal" emergency braking scenario would be all 4 wheels braking at the maximum force they can without slipping, and that can be done much better by the ABS system when they're all disconnected from each other. A driveshaft connecting them would be irrelevant at best, but more likely it's a nuisance.
AWD is complicated with 3 differentials that can be a mix of open, limited slip, or actively controlled (by torque split or brake based system). It's especially tricky since it's not the actual traction that matters as much as the driver perceived traction and predictability. I've drivin AWD systems with snow tires that seemed to have more traction than my subaru with all seasons, but ended up driving more slowly because I couldn't predict what it was going to do. Subaru seems to communicate well with the driver, and making sliding just not a big deal. Other cars make it harder to tell what a given throttle/braking input is going to result in and can trigger heavy front or rear bias based on what the sensors/software tell it to do. It drives me nuts when a car acts completely differently in high traction and low traction conditions, especially when it changes one second to the next. Some Subaru folks make fun of other systems that they call "slip and grip", where there's a heavy FWD bias (minimal torque to the rear), until AFTER you start slipping.
My WRX had a limited slip in the rear, center diff was a viscous coupling (100% mechanical/no software), and a open front diff. It was reassuring, never surprised me, and even when significant slipping I could always put the car exactly where I wanted. Even when crab crawling sideways in a parking lot at 2 mph it never surprised me. Numerous Toyota, Ford, Honda, and BMW cars I tried would try to outsmart me. Limiting throttle, moving the torque around, limiting braking forces, etc. Last thing I want when climbing a hill is the car to cut the throttle because one of 4 wheels slipped for 50ms.
Consumer reports tested a Subaru forester, Toyota RAV 4, and Honda CRV in heavy snow conditions and concluded the Subaru was a clear winner despite them all having similar dry road performance and AWD.
I think for normal snowy roads Subaru does it best with a mostly manual system, the ABS and traction control has a light touch that can be mostly avoided. But for heavy offroad the best thing to do (in certain messy conditions) is lock all differentials, which is something the Subaru can't do.
It's obvious that it'll help when accelerating, and perhaps even when turning, with a (more) even application of torque on all wheels.
My main gripe with it is that it often gives drivers a false sense of security (as they got going easily), and they'll drive too fast for the conditions. Just look at some parts of this sub-thread, where people say "I don't need winter tires, I have AWD / 4WD".
If you need to brake in an emergency on snow / slush / ice, nothing is going to help you except your tires (and perhaps some smart ABS).
There's a wide variety of software solutions as well, some work better than others. It's a non trivial problem. For instance. In normal conditions less braking when any wheel slips is best. In slippery conditions for stopping or going some slipping is best. In really messy conditions it can be best to not keep applying torque (in the positive or negative direction) without any "smarts". For emergency braking in the snow it can be best to just lock them up and have snow build up in front of the tires.
To see an example of how big a difference this makes, check out
https://www.consumerreports.org/video/view/cars/auto-test-tr...
I agree that AWD leads to overconfidence to some, I've been driving at 35 mph on the highway (with a 65 limit) in my Subaru (with good tires) in bad conditions and had giant SUV blast past me at 10-20 mph over what I was doing ... only to find them in a snow bank later.
Doesn't mean that AWD wasn't a huge help. My first AWD was a result of me spinning out in a FWD around lake tahoe in poor conditions, in heavy traffic, at 10 mph or so. My front right tire lost traction, the open front differential put all power to the front right tire, and it spun me into a snow bank (just 18" away") in a blink. I was quite impressed how much rotational acceleration could happen in a blink of an eye. The open differential meant I was basically 1 wheel drive (when slippery), the car I replaced it with (Subaru WRX) had 3 wheel drive (when slippery) and was MUCH more predictable. In fact sliding in the snow would bring a smile to my face... and I changed my weekly commute (6 7200 foot passes) to maximize the snow involved.
And yes I'd MUCH rather have all seasons and AWD instead of snow tires and FWD. It's not the ultimate traction as much as being predictable. Sure all seasons have less traction, but it's way more predictable and gives much better control when sliding. FWD, even with snow tires can be very unpredictable, and can easily spin a car when the traction is low. Subaru + snow tires seemed overkill for the roads I was on. Sure Subaru + all seasons might be safe in some conditions at 35 mph, snow tires might raise that to 40 or 45 mph, but FWD regardless of tires would mean much less, and Caltrans would require the pain of installing and uninstalling chains and worries about chain failures and/or damage to the car.
Even with all seasons, my main fear with a Subaru (I've had two) is hitting enough snow to life the tires off the ground.
Sure both cars would descend the same, but you can always lower your speed to decrease the amount of traction you need. When climbing there's no such options for FWD.
When it comes to winter driving the entire perceived benefit of AWD vehicles for commuter use goes out the window as soon as you need seasonal tires, basically. Yes, it will still better help put the power down given equal tires - but at that point the extra cost of the AWD system isn't giving you too much benefit if you are only buying it for winter driving.
I've heard the FWD + snow tires beats AWD + all seasons before and personally I don't believe it, but it's a pretty subjective thing. The big win for AWD is predictability and control, not that I want to go around corners sideways at 100 mph rally style. Having control over the car, even when sliding, without any rear or front bias just makes for a MUCH more controllable car.
My Subaru forester with AWD and all seasons handled numerous treacherous conditions, sometimes so bad that every highway around me was closed. I used to do 6 7200 foot passes every weekend, often in the middle of storms dropping 2-3 feet. My main problem on the roads was others that don't know how to drive and block the roads.
I've driven roads so slick that going too slow would cause you to slide off the road. I'd have to apply torque, while sliding, to keep the 10-15 mph required to avoid sliding to the inside of every banked turn. Which would be MUCH harder with FWD or RWD.
Amusingly at chain control points, often the workers would look back in the line, see my Subaru and pull me out of line and let me go through with a smile and a wave. Presumably because there are (or weren't anyways) any non-AWD Subarus (this was before the BRX) and Subarus aren't the typical cars getting stuck and causing problems.
Sure AWD and snow tires is even better, just seems overkill for the highways in the high sierras.
The long term, macroscopic decision making involved in driving computers are nowhere near ready to drive better than I can in winter (or even in perfect conditions). I love my model 3, but the autopilot is simply not a good driver. It would fail any minimally rigorous driving test because it makes rookie mistakes and only drives in the exact moment. It cannot plan ahead more than even a handful of seconds. It stays in other driver's blindspots when there is plenty of room to adjust the relative positions of the cars. It accelerates into gaps that are about to not exist because other drivers' small hints scream they are about to change lanes, even if they haven't signalled yet. It gets confused about speed signs, often reading the off-ramp maximum as the speed and slams on the brakes. It plants itself in the dead centre of the lane even when other drivers encroach and then when it decides the situation has become dangerous it swerves violently and applies applies significant braking. Experienced, safe drivers give ground within the lane, preventing the dangerous situation from ever developing, and if the other driver does fully enter their lane are ready to calmly move over without creating an entirely different collision.
I thought Teslas only have one giant screen one the right side of the steering wheel?
The Model 3 and Y have only the center screen in a horizontal layout.
https://tesla-cdn.thron.com/delivery/public/image/tesla/2558...
This does not give me any trust that there's decent QA done on this software.
I'd feel a lot safer with that kind of data added to the current sensors.
Just this weekend I drove a 600 mile round trip. It was cold weather with some snow on the shoulders, but the actual road surfaces had no snow. I was mainly using just the cruise control, and no auto-steer. Even with just cruise, the car gets "spooked" from time to time and will suddenly jerk to a slower speed for absolutely no reason, even during the day and with no precipitation. At one point it did this several times within the span of 10 minutes, which was such an unpleasant experience that it made me want to stop using cruise entirely. Besides being jarring for the people in the car, it makes me worry that anyone tailgating will rear end me.
Besides that, the rain/sleet that was falling from time to time apparently blocked the camera, so at one point the car wouldn't even let me use cruise any more.
The car is tons of fun to drive and I don't regret the purchase at all. But I really don't like the overpromising on full self-driving.
EDIT: bmcahren pointed out to me that the new FSD beta software merges images from multiple angles, which my software does not do. I don't know much about that, but it sounds like the kind of change that could lead to a noticeably better result than what I experienced.
(inb4 "Sure it goes fast, but the panel gaps!")
Personally I love Tesla-the-company and my Tesla vehicle because of the innovation it's driving in the space--and I'm willing to tolerate the ways it falls short due to how incredible it is in other ways.... but I do try to be up front about those shortcomings when I discuss it, at the risk of being one of those "I love my Tesla, but..." people.
they aren’t scaling service centers fast enough to keep up with an extra 200k cars added on the road every quarter, so service meaningfully deteriorates every week as more cars are sold and existing cars age. it simply doesn’t work unless they were committed to building out 2x-3x the SCs they have now but nothing indicates that’s on the roadmap at all.
The numbers:
- increased service locations 12% YoY - increased mobile service fleet 8% YoY - increased supercharging stations and connectors 32% YoY
They are clearly building more service.
During quarterly conference call Musk stated that he believes service is the biggest driver of sales and that Tesla will invest in that. So contrary to what you claim it's on the roadmap.
Many people are over the moon with Tesla service given that mobile service comes to you and they can do fixes without you even there. That's something no other car maker does at scale, as far as I know.
Did you have a personal experience with Tesla's bad service or is it just amplifying FUD from other people?
I have owned 2 cars that I bought the first year the model came out.
The Model 3 I have had a few dumb issues like, if you run the paint app sometimes the gps location on the map doesn't update until you close the paint app and your displayed location lerps from the old point to the new one over a minute or so.
The Mini had an eco mode that would occasionally shut off the engine entirely at highway speed and not allow restart until you came to a full stop and put it in park. Eventually, they just disabled eco mode in software Which took a few weeks of me working from the mini dealer. There was display of how many miles you had gained that reset when you filled the gas tank. My record is 42 Miles gained in eco. After the final update to the software, which was a full day in the garage, My best was 8.
One of those was worth running stories in the press about.
I guess what I'm saying here is that there is a deep vein of shadenfreude in the news about Tesla. A single Tesla on fire is a big national and some times international news story. I have stopped even reading the damn things. It seems like a weird combination of them being seen as rich people cars, early adopter cars, and liberal hippy cars. Everyone can hate someone on the spectrum. Typically, the issue in articles about Tesla flaws is so isolated and dumb that it exists solely to generate clicks. The last article about Tesla I clicked through stated that consumer reports had downgraded their durability rating due to the discovery of single human hair in the paint of a model Y. I get that it sucks. But, I don't get why I should care about it.
I own a Model 3 because I needed a four door car because I had kids, electric seemed like a net win, the local Bolt dealer told me not to buy their car, and the range let me drive to the grandparents houses without a charge. I recognize that there is stuff about the model 3 and Tesla that sucks but overall it has been a pretty solid car for the last 2 years for me.
Would I recommend one to you? Yeah. I don't think there is a company that offers the combination of electric car, mature tech, range, and charging network at a similar price tag.
- You weren't using the actual FSD software, but the live production autopilot, so it's Apples to Oranges.
- It's not clear that the transition from low quality to high quality full self driving will be linear. Most claims around anecdotes come with an unspoken assumption we ought to see steady improvement over time vs a step-function like behavior. The latter seems somewhat possible given the nature of how improvements in AI systems seem to happen.
- Beyond all this, without understanding the underlying cause of the behavior you mention, it's hard to say from a risk perspective if the car made the right choice. Unless you have a clear understanding of the data and mechanisms that led to the speed reduction, it's hard to be sure it's a bug beyond not doing such transitions in a way that are comfortable for passengers. (My understanding is adding smoothing to the control algorithm is somewhat separate from overall decisionmaking, and hence poor control algorithms may mislead you into thinking it's more dangerous.)
However I don't think lack of confidence in the current software, particularly the production autopilot software, is really giving me much signal in helping predict if Tesla will have FSD anytime soon or not. Now that FSD (at least in Phoenix conditions) is known to be possible the prior that it will never be achieved by Tesla to me is near zero, beyond the chance that LIDAR is strictly necessary and they have a doomed hardware stack. I think there is a pretty big range of possible timelines though - I could see it happening in as soon as 18 months or as late as 10 years depending on how poorly things sift out with their hardware stack.
https://jperla.medium.com/why-tesla-autopilot-ought-to-be-aw...
I mentioned this in another comment, but it's relevant here; OP wasn't using Autosteer or Autopilot, they were using cruise control: a miles more simple system--and that system behaved in a way that seriously reduced confidence in it.
(For the record, I've experienced exactly the same behavior in my M3 as well.)
So, if we can't trust something as simple as cruise control, and we've seen similar confidence-eroding issues in the autopilot features Tesla has already deemed "good enough" for a live release, I think it is 100% reasonable to be skeptical that Tesla is anywhere close to "solving self-driving."
They have a history of overblown claims and faulty software in this realm--as much as I adore my car and the disruption Tesla has done to the industry, Tesla does not deserve the benefit of the doubt in this realm, especially not with safety-critical systems.
FSD is a completely separate system with different tolerances. One does not have any relation to the other
Before you bring up Tesla's "Fleet Learning", it's not used at all outside of gathering perception training data for labelers. greentheonly got access to a dev build of the FSD beta - ML is just used to generate a representation of the current scene - all of the driving and planning logic is rules-based and hardcoded by engineers.
https://jperla.medium.com/tesla-saves-lives-waymo-gambles-pa...
This is also why you saw a bunch of AV startups with < 50 employees that got big initial funding rounds but folded after a couple of years - it's pretty easy to grab a few dozen grad students, slap some LIDAR on a car, encode basic road rules, and get a demo working that's compelling to investors. Once you get past that initial pitch though, you get into the weeds of trying to progress without causing regressions elsewhere.
Also, what makes you think those expectations were grounded in linearity? Waymo went from an initial project at Google in early 2009 to driving 10 complex 100 mile routes without intervention in mid 2010 - on the surface, it would have seemed they were 95% of the way there. Improvement has been sublinear for a very long time and everyone in the AV industry is aware of it.
Do you think Tesla has completely separate groups of engineers working on this? That doesn't seem like Tesla's style at all.
The rewrite was to do image recognition and inference not from individual still images (pre-FSD) but from video feed from all cameras. This was enabled by the FSD chip Tesla started putting in cars about a year ago.
Plus the improvements that come simply from gathering more data from cars and using it to improve the training of neural network.
It's documented in Musk's tweets and Karpathy's talks.
* Level 0 (No Driving Automation)
* Level 1 (Driver Assistance)
* Level 2 (Partial Driving Automation)
* Level 3 (Conditional Driving Automation)
* Level 4 (High Driving Automation)
* Level 5 (Full Driving Automation)
If you restrict it to certain parts of driving, eg TACC, that might be level 4 (doesn't conform to the chart though). It can handle everything in it's domain, albeit in a rudimentary way, so a human doesn't always have to shadow it, but they do need to be ready to take over in some cases.
What argument could there be that suddenly slowing down on the freeway with no visible obstruction is the right choice?
It does not avoid any danger, but it creates a danger of getting rear-ended by cars behind you.
I'm not saying that's what happened, just that it could have happened and therefore would be an explanation.
That is true, but it's actually an argument _against_ the assumption of full self-driving capability being just around the corner. Because that assumption itself is based solely on observed recent progress, which - as you correcly state - can be very misleading when used as a means to extrapolate the future due to the tendency of the AI field to have long stretches of very little to nearly no progress at all, divided by huge leaps followed by stretches of very fast progress in specific sub-areas of the field. We clearly have seen such a stretch recently, triggered by the deep learning idea and enabled by an explosion in parallel computing capability over the course of the last decade.
I mean, you can of course refrain from making any predictions. But I have never met a single person who could resist the instinctive urge to extrapolate current developments into the future. It is as much a core part of being human to do this as it is to breathe air.
You might as well be comparing your experience driving a BMW or Cadillac with SuperCruise. Your experience on the old tech stack is irrelevant.
> You might as well be comparing your experience driving a BMW or Cadillac ....
But OP isn't--he's comparing the current live Tesla features with paywall'd / betawall'd features on the same car, from the same manufacturer. I cannot think of a more relevant comparison! I think it's completely reasonable to doubt Tesla's (and Elon's) claims in this realm, when the existing, public examples of this technology fall far, far short of the kind of confidence we'd want to have in these systems.
> Your experience on the old tech stack is irrelevant.
I 100% disagree. If Tesla can't even get a basic function like cruise control to work well enough that drivers are confident in it, why would I trust that they can get self drving to work well? Especially following things like the Summon beta, or similar fiascos?
I still get twitchy when my M3 comes up on an off ramp it's not supposed to take on autosteer--I've had the same behavior as OP (sudden unexpected speed decrease) happen far too many times for me to have confidence in that system. In turn, I have even less faith in the more complicated systems, like autosteer or FSD.
There is a history of overpromising that will be still leave me somewhat skeptical until I see it. But it does sound like a move in the right direction.
https://twitter.com/greentheonly/status/1343670544529575936?...
https://twitter.com/greentheonly/status/1336467014727110656?...
It's well documented that FSD is a architectural rewrite of the code.
Pre-FSD the input to NN were individual still pictures taken by the cameras.
New architecture feeds NN with video feed from all cameras.
New architecture is possible thanks to new FSD chip that is 20x better that was Tesla used before.
I don't see what definition of NN would consider them the same system.
On YouTube you can find talks by Karpathy where he goes into detail about this stuff.
The protestations to the contrary here reek of defensive fanboy/apologist behavior.
It's clearly order of magnitude more advanced and capable than pre-FSD AutoPilot.
Calling anyone who disagrees with you a "fanboy/apologist" does the opposite of strengthening your argument.
It's cruise control. If you couldn't fry an egg without burning it, forgive me for being a little skeptical when you claim that you can make award-winning souffles after having attended some cooking classes.
As it relates to the features, I agree that they aren't ready for what most drivers expect, either because they aren't reliable enough (beta) it because they aren't confident enough (TACC). With that said, FSD is a feature set, not a level of reliability. As a whole it eventually needs to reach a level of reliability, but there's no specific time horizon.
For Tesla, a feature moving from beta to alpha is what indicates reliability. In addition, that reliability may not come with the degree of confidence most people expect from a driver. Safe FSD could just end up as the AI equivalent to a student driver. It'll get you someplace safely, but there may be phantom braking, excessive waiting, and so on. If that's the case it would be nice if we could enable a beta version with more "normal" driving that we would have to keep an eye on.
https://jperla.medium.com/why-tesla-autopilot-ought-to-be-aw...
https://jperla.medium.com/why-tesla-autopilot-ought-to-be-aw...
This is a bit of a nonstarter for gauging snowy weather performance, in my opinion. While comparisons can still be drawn between human and FSD drivers using the same (incredibly unsuitable) tires, handicapping the system doesn't provide a great measure of its prowess or challenges in "snowy conditions."
A bit of an aside, but few driving-related topics grind my gears more than people who live in snowy/icy climates driving with cheap (or just bald) all-seasons and complaining that their car "isn't good" in the snow so they need a bigger 4x4, which is of course subsequently kept on stock all-seasons for only a marginal improvement.
They need to clean themselves reliably
It was scary.
I have Zero confidence in Tesla self-driving.
Correct following distance in that traffic is ~3-4 metres (~10-12 feet) which is easy to maintain against a human or against an auto-car you know is auto-driving, but that takes a few seconds to calibrate that it's auto-driving.
I've made many rush hour commutes from my home in Glen Park to SOMA with very few brake-use events¹ and travel team equivalent to original Google Maps prediction.
¹ One of the many metrics I attempt to min-max during all of my drives
This is at most a sophisticated driver assistant. The PR on this is just disgustingly disengenious.
https://jperla.medium.com/why-tesla-autopilot-ought-to-be-aw...