I think we'll be better than human by the end of this year
Is that human Elon himself?
Tesla CEO Elon Musk let an interesting tidbit slip during yesterday's Q2 earnings call: his car company has plans to license its as-yet Full Self-Driving stack to other automakers. Musk made the claim in his opening remarks on the call, saying that Tesla was open to licensing both its FSD software and hardware to other car …
Automaker X in court, being sued: "While there was a heading deviation in the plaintiff's vehicle," [non-lawyer version: One of Automaker X's Tesla-FSD-equipped vehicles made a 60-degree turn in mid-block, smashed into a streetside cafe, and killed/hurt people there] "we relied on the expertise and experience of the people providing our Full Self-Driving software and hardware: the Tesla Ccorporation. This software and hardware was fully approved by the [name-of-country-here] government."
This is the problem - until a car company steps up and says "Our car is the driver, hence our driver is the one insured, hence the human *passenger* doesn't need insurance and all claims and responsibility go through us", then you don't have a self-driving car.
And when that happens... can you IMAGINE how much that's going to cost the manufacturer in an area that they simply don't have to deal with.
And, no, they can't just necessarily pass the buck, because it'll mean every Tesla recall also becomes a Ford recall.
Either Tesla would be bankrupt - from the number of recalls, plus the manufacturers who licenced it suing them into oblivion, or from the sheer amount of claims against them - or they'd have to not accept any responsibility (in which case, we're back to square one... "you can licence the software from us, but it's on you if it kills someone").
Until there is a blanket insurance on a vehicle and it's ACTUALLY the responsibility of the manufacturer as to whether or not it runs over small children, self-driving is just nonsense.
And that's a business proposition that I can't imagine ANY manufacturer wants.
I suspect Tesla has realised if it licensed it to others they could develop it further (and actually get it to work which Tesla seems incapable of), so they can sack some more people which is something that Musk likes to do as much as his mate Trump.
The only problem is that other manufacturers already have or are quite close to what FSD allegedly delivers (Audi, for instance, demonstrated autonomous driving two years before Tesla even started to talk about Autopilot, and Volvo demonstrated it on HGVs), they're not skimping the the sensors required to support it and are just far more careful by not using other road users as guinea pigs..
By many accounts, Tesla's FSD is lagging behind other makers' systems.
The way I see autonomous driving is that it's worthless until it's perfect. I'd rather take a nice train where I can get up, buy some needful supplies from the cafe car (G&T, etc), visit the loo or just go walkabout if I don't want to drive. Brightline in the US seems to have figured out that it's possible to go faster by slowing down. They have promised service from Los Angeles to San Francisco (Emeryville) taking something around 11 hours. The beauty is that it's a sleeper train and traveling around 50mph it should do better on fuel and the route is what Amtrak uses once a day in each direction for the Coast Starlight service. I always see any travel taking over 4 hours as the entire day. One can fly from LA to SF for a very modest price and the time in the air is pretty shot, but all of the inconvenience of flying these days makes it more like 5-6 hours of being groped, asked personal questions and having one's luggage inventoried so any valuable bits can be removed. I'm biased as I sleep like a log on trains. All I need is an adult drink to relax and the promise of a coffee and a bun in the morning when I wake up.
The way I understand Tesla's approach, they want to solve FSD everywhere, not within a defined area limit or limited weather conditions. Musk is betting that since vision is the basis of human driving, other vendors using tools like LIDAR may be temporarily ahead but will cap out lower than a camera-based approach.
That's stupid. Humans would be better drivers if we had radar and LIDAR vision in addition to our eyes.
He's just stubborn, and when everyone told him he was stupid for trying to go with visible light only he feels like he has to double and triple down to prove he was right.
So despite the early lead Tesla will continue to fall further and further behind other automakers. All because of incompetent leadership.
SpaceX's CEO is the reason why it is doing so well. Musk is just head cheerleader and getting credit for her results in the popular press. Give him credit for hiring her I guess, but even before Twitter he couldn't take credit for it as he's not the one driving that bus.
He's just stubborn
Actually, he was being a cheapskate.
He thought in his arrogance that he could get LIDAR for far lower prices by just talking the usual crap about fantastic volumes and all that, and the vendor told him that that was excellent and he'd get a volume discount as soon as he bought in volume, and not before. Given that that added a LOT of extra costs to a Tesla (read: less profit) he decided to be his usual self and sulk, then BS the rest of the planet by saying he didn't need it.
Unfortunately, the Universe doesn't bow to Elon's will as it strictly works with reality, and so Tesla's FSD started off with a massive detection handicap, and the rest is not history as much as now the subject of multiple investigations.
That is actually why I personally will never buy a Tesla: I can see where they shaved to save on manufacturing costs (like the tablet, which is cheaper than physical buttons but far less safe) and it would annoy me on a daily basis.
"Unfortunately, the Universe doesn't bow to Elon's will as it strictly works with reality, and so Tesla's FSD started off with a massive detection handicap, and the rest is not history as much as now the subject of multiple investigations."
On the other hand, will Tesla FSD come to be seen in the public consciousness as the start of self driving in the same way Apple is seen as the inventor or the smartphone? Neither are true, but marketing and the general public think they are both true. The main difference, of course, is that Apple took an existing invention and actually improved it. I'm not sure the same can be said of Tesla and FSD. Although credit where credit is due, Tesla has done a lot to bring EVs and self-driving into the forefront of the public mind, even if then end up at the bottom of the pile or simply consigned to history.
If you want to credit a company for EVs it would be Toyota with their Prius hybrid, that started it all (its new model actually looks like a car and is according to reviews quite impressive - that review pokes some deserved fun at it, though, the proper review starts at 2:38 and updates here). Tesla has since done some sterling work with the Roadster and then Model S (IMHO their best ever) and on motor and battery improvement for which they deserve credit, but I am not convinced that Musk has been a great help as it subsequently turned Tesla into a cult thing, and with Musk more and more disclosing just how much of a horrible human being he is, it has started to switch off more people than actually benefit the company.
This also bites into the company's evolution - I have the impression that Musk is developing some delusions (well, more, he seems to collect them) that he's the Steve Jobs of cars, given the horrific disaster called the cybertruck. Like their HGV, it seems talking to everyday users of this calls of vehicles seems to have never happened during development, and in this case it appears they didn't even talk to designers. Add to that the name (cyber? WTF?) and I wonder who this was meant to please. Customers, or Musk?
Steve Jobs had an eye for the practical and made the exiting easy to use (it's not like we didn't have smartphones before he got involved), Musk seems to drive new ideas that are not even discussed with possible owners. Now I would occasionally scoff at the overuse of focus groups, but in Tesla's case some interaction with future customers would not have gone amiss. Look, for instance, at their HGV - that's an all out impractical disaster according to professional HGV drivers that have tested it. That is presently not a bad thing as grids are by no means ready to haul that much power, but that's a mere coincidence.
" battery improvement"
you do know the battery improvement was making the battery cells hold 50% more charge with the genius method of making the battery cells 50% larger.
at the same time making it almost impossible to do repairs on the battery packs. so an utter musktwat move.
I suspect it's a cost/availability thing. Kind of like how, when there was a shortage of chips for radars, he suddenly "discovered" that cars didn't need radar after all and eliminated it. Now they're starting to put it back in because -- guess what? -- it was important.
I wonder how long before we see a class action suit? Musk has made some very public promises about self driving, including a wild claim back in 2019 or 2020 that in a year every Tesla owner would be able to make $100,000 a year operating it as an autonomous taxi during the time they weren't using it themselves.
Who wouldn't want to buy a car for $50K (including the FSD markup) and have it more than pay back its sticker price every year after that? That would make for a nice change from having it lose value every time you drive it, or every day it sits unused in your garage, like every car does now (aside from weird times like during the pandemic when recent model used cars started selling for more than their sticker price was the year before)
You lead into an interesting question which probably only a Tesla owner can answer: when you pay for FSD, who holds the license? The car, or the payee?
If it's the payee it means it moves with him/her/it/whatever when they buy a new car but then there's a the question of what happens when someone else drives the car, or it's the car but then you've just added, what, a good €12k to the car that you may never get back when selling it.
This is part of the problem I have with Tesla: it's all about new things and new gadgetry, but their whole lifecycle management feels more like a hand-wavy Musk afterthought. Other car brands have a whole infrastructure behind them where even for a 20 year old Nissan I managed to get a new part and some bits from a breaker's yard (I think that's what it is called, sorry, not a native English speaker), whereas Tesla will happily cut you off from their fast charger network if you as much as think about recycling bits from other Teslas. The only argument I can see there is that they may have been structurally damaged by a battery pack fire, but I reckon that's fairly easy to spot.
This is why I'm more interested in the EVs that decent car manufacturers are now starting to bring out, and reluctant to consider cars from rather interesting new kids on the block such as Rivian and Lucid. They're interesting, but as yet lack a mature international support infrastructure.
AFAIK that license is with the car. But if you buy FSD and your car is NEVER able to self drive (because they can't make it work without LIDAR or because self driving turns out to be a much harder problem than expected) then you and your class would have great grounds for a lawsuit. At least for the recovery of what you paid for FSD plus interest, plus whatever additional damages good lawyers can convince a jury to award. How that would be allocated to the original owner versus subsequent buyers would be for the class action attorneys to figure out.
Even if you bought a car with FSD in 2020 and they finally got in working in say 2028 you could easily argue there are damages because Musk promised you autonomous capability in a year and it took eight years. By which time the car is worth much less, may no longer be with the original owner, etc.
Its a mess, and I'm guessing Tesla's lawyers must have PLEADED with Musk to quit tweeting unfounded promises. That's why when he was talking about this now he isn't making promises about dates any longer. The lawyers know there is a lot of accumulated liability, they want to make sure Musk isn't adding more on top. Though the name "full self driving" is a problem in itself. If someone pays for that, even though Tesla is clear it isn't capable of that today they are still promising it will arrive eventually.
It would be interesting to know how many FSD packages Tesla has sold and how that compares to the total amount of profit they've made since they started making a profit. Less the overhanging liability that they may someday have to cough up, it may still be losing money!
"Musk is betting that since vision is the basis of human driving, other vendors using tools like LIDAR may be temporarily ahead but will cap out lower than a camera-based approach."
The difference is that in addition to eyes, we have a very complex brain. The psychology and physiology of vision is very fascinating and is way more unusual than people think. To replicate it with some cameras and a computer isn't something that anybody has the slightest inkling on how to do. The classic approach has been to just throw more computer at it, but the increase in accuracy vs. increased power requirements is pants.
I get the feeling that Elon does't like LIDAR/RADAR due to the compromises in design that would be needed to have them on the car. They are also another expensive component that raises the build cost and if they need replacing, the interwebs will blow up with complaints. A new quarter panel isn't too bad, but if a LIDAR also got smashed in an accident, it could add another grand to the repair.
I'm an advocate of Personal Rapid Transit for city centers, trains, trams, etc. Another thing I see in the US is the lack of efforts to join up transportation systems. Why didn't the trolley system in Portland, OR get designed to stop at the train station? It's not that far between them unless, of course, it's hissing down with rain and then the trolly stop is effectively miles away. Why isn't there a subway or trolly line from LAX to Los Angeles Union Station? An express service by preference, but all they have is a bus so if you are on a tight schedule, you may not make it if there is any sort of traffic delay on the streets. There's so much that could be done in transportation that autonomous cars that will work on open streets is the hardest to make work safely.
The other problem with relying on vision is that computer vision will NEVER be perfect. We get fooled sometimes, the sun and shadows hit you just the right way and you think you see a person but there's nothing there. Or there is a person there, but you think you see something else or nothing at all. We'd probably never get fooled like that if we had another sense that didn't rely on light but instead some sort of sonar or radar that could give us a depth map.
If we were bats and saw with sonar we'd be able to be fooled the same way, just in different scenarios. Instead of the sun blinding/fooling us it would be certain types of ambient noise that did it. If we had a combination of the two it would leave us fooled far far less often and even when that happened it would be obvious (i.e. our vision and our sonar might both be fooled at the same time, but they wouldn't be fooled in the same way so we might "see" a person but "hear" a tree)
The hype around autonomous cars was bult by Uber so they wouldn't have to pay humans to drive humans around. Everyone followed their hype without thinking: Is this the real problem?
I think the real benefits of FSD are on long(er) distance, intercity routes (e.g. Motorways/autobahns/autoroutes/etc). And then, probably the first use case would be lorries/trucks: Human drives the truck to motorway, flicks the switch and then goes to sleep until the truck drives unit the exit/off-ramp. (If you've got multiple autonomous trucks, you could get them to talk to each other and slipstream safely)
Why would you need a human in there? For long haul routes it might be on interstates/highways for a day or two in the US. In your scenario there would be a human to drive it to the point where it can take over, and another human in the destination city that would meet it where it pulls over after leaving the highway.
I agree that long haul trucking routes are a place where autonomous vehicles could really shine, but that's by eliminating the need to pay someone. If you still have a "driver" then the economic argument for paying more for a brand new truck is eliminated.
If they already slipstream by computer, why not couple them together with a fixed link? Oh, and now, hear me out, in order to increase fuel economy we should let them roll on special surfaces, lets make those of steel and the wheels as well. This will decrease rolling resistace by quite some margin!
What? We have that already? It is called a 'railway'? How... quaint and very much 18th century! (yes, in my opinion long distance freight belongs on rails - but I am not really sure how good the infrastructure is outside densely populated regions like Europe, or Japan)
Wasn't VW experimenting with automated slipstreaming? I recall seeing their experiments, but there were two challenges:
1) vehicle to vehicle comms must be *perfect* to keep it all in sync, especially when the front car starts braking
2) what wants to be in front? That's the car that will NOT benefit from the slipstream, so some mechanism must exist for everyone to take their turn or it'll be unfair.
Anyway, the idea is far from new.
what wants to be in front? That's the car that will NOT benefit from the slipstream
Actually the car in front benefits some too, just not as much as the car behind, at least in NASCAR (i.e. more traditional looking vehicles that aren't so wildly optimized like F1) because it reduces drag on that car's rear.
Being behind can be a problem for cooling, and in a longer train I'd assume being in the middle is even worse so there might be good reason to split up and reorder the train occasionally.
In the UK, HGV are not allowed in the outside/fast/overtaking lane unless there's only two lanes or unless there is a sign stating "HGVs use all lanes", normally where there is an exit from the outside lane, which is relatively rare on the UK motorway network.
On the other hand, automated trucks running in convoy would need law changes anyway, so if/when that happens, all bets are off as to how sensible and practical any future laws may be. They may just update/adjust the existing laws regarding military convoys, ie first and last vehicle have a green or blue light (or flag, in some case still) and legally you are not allowed to insert your vehicle into the convoy, ie stay back or overtake all of them at once.
Long loading and unloading times mean that rail freight is efficient if your origin and destination are far away from each other and you have no intermediate load/unload points. That’s a good description of freight between the US’s major population centres, but it’s rarely true in Western Europe or South-Eastern Asia with its dense network of short-distance freight routes between industries.
Generally, as the number of “towns” per thousand square km increases, freight shifts from rail to road as rail’s lack of reach and flexibility overcomes its low per-km cost. Conversely, as the population density of those settlements increases, rail becomes the most efficient mode of personal transportation.
There are a few hybrid systems to allow road freight to hop onto rail for long-distance or vehicle-restricted sections of their route: the Brenner RoLa is one example, where trucks drive on to rail flat-cars and are then carried through the Brenner Base Tunnel railway line under the Alps while the drivers take a little nap: on the other side, the process is reversed, and the truck drivers set off on their various ways again. And, more familiar to El Reg readers, there’s the Channel Tunnel, a rail service that also operates a roll-on, roll-off crossings for freight drivers.
There are similar proposals for transcontinental routes in the USA (here the main win would be to still be able to move freight while the drivers are on their mandated 14 hour daily breaks), but as ever in American infrastructure, politics intervenes and things get gummed up.
"Long loading and unloading times mean that rail freight is efficient if your origin and destination are far away from each other and you have no intermediate load/unload points."
You've come across the real problem and that's loading/unloading times, not the cost of drivers. The classic view on economies of scale suggests very large cargo handling centers. With the right sort of planning and infrastructure, it might make more sense to have more smaller depots with automated shipping container cranes and a penalty to the companies picking up freight if they don't have a lorry and trailer ready to accept the container as it comes off the train. A 'loading' depot can be on the opposite side of the tracks so as one container is removed, another one can be loaded immediately. What gets left is freight that doesn't lend itself to loading into a container and needs special handling and that can go on trucks from larger depots on the line.
I was just envisioning train cars used for lumber in the US and they could be redesigned to a standard sea container format for more uniform handling. For some sizes it could look space inefficient, but the difference in handling times could make up for that.
"With the right sort of planning and infrastructure, it might make more sense to have more smaller depots with automated shipping container cranes and a penalty to the companies picking up freight if they don't have a lorry and trailer ready to accept the container as it comes off the train. A 'loading' depot can be on the opposite side of the tracks so as one container is removed, another one can be loaded immediately."
You just described what the UK Freightliner Terminal system could be if it hadn't ended up has just a few large terminals, ie back when there was a terminal in most major cities or large population centres. Historic road transport pricing and "economies of scale" reduced the network to just a few very large hubs and dropped all the smaller regional depots.
The US moves more freight by rail than any of those countries, but these days it's mostly confined to large unit trains of bulk materials -- e.g., coal, grain, taconite -- for heavy industry, and intermodal freight between major ports. (It's often cheaper for a shipper to unload a ship on the West Coast, move the containers by rail to the Gulf, and load on a different ship there, instead of going through the Panama Canal. Railroads don't want to mess with switching individual cars to smaller businesses anymore; it's too slow and labor intensive.
I agree that long haul trucking routes are a place where autonomous vehicles could really shine, but that's by eliminating the need to pay someone. If you still have a "driver" then the economic argument for paying more for a brand new truck is eliminated.
In the UK, and I believe in the EU as well, an HGV driver can only drive for 9/10 hours a day, 56 hours a week, 90 hours every rolling fortnight, and has to rest for 11 hours a day, 9 of which must be uninterrupted. If they can be resting or asleep in their cab, the cab could be active for significantly longer periods.
1) Security. An unaccompanied truck would be a very low-risk robbery job. Running automatic truck convoys with armed security has already been mooted for dealing with this.
2) Drivers a have time limits, and trucks stop while they rest & sleep. If the trucks could do interstate driving while the drivers slept, that would keep the trucks on the road, and also make the drivers more valuable.
Doing away with the meatsack in the cab would be great 99.999% of the time. The problem is with millions of miles covered every day 99.999% is not too hot.
As long as the trucks are maintained perfectly and the roads are kept clear of any debris that could damage a tyre then it could work but trucks are not all perfectly maintained and some never will be and the roads are littered with crap and will always be.
People talk about trucks electronically linking in a "road train" which again would be great most of the time but one tyre blowout in the chain of trucks could cause a disaster. Judging by the amount tyre debris seen on the sides of motorways such disasters would be practically guaranteed.
"Doing away with the meatsack in the cab would be great 99.999% of the time. "
That's too generous. The driving portion of the job is the easiest part to automate but there's always paperwork. There's always nuances at each end that have to be dealt with . Ask any driver in the US that moves freight in/out of a Wally world distribution center. It doesn't matter if there is no dock to back into or space on the lot, the driver/company will be penalized for a late delivery if they aren't at the dock with the doors open at the specified time so it's exceedingly important to have everything documented to dispute the penalties. Drivers are also providing a level of security for the load, able to navigate around blockages and keep checks on special needs such as refrigeration. It could be expensive is the fuel gauge for the reefer unit was stuck on full. A driver might pick up on that and put fuel in where an automated system may not leading to the loss of an expensive load.
My point is that 99.999% safe is rubbish when it comes to life critical safety.
Nobody would ever get in an aircraft if they were only 99.999% safe as with about 100,000 commercial flights a day worldwide 99.999% would mean on average one incident every day.
With road vehicles with millions of trips every day 99.999% safe is lethal.
Numbers can be misleading.
Same here! Iam ok with trains, the kids love the adventure. I commute by train - by car would be faster, in theory, but looking at the traffic patterns I would be either stuck in traffic on the way in or the way out - or both :( By bike takes a bit longer than by train, I do that in the summer months (take the bike on the train on the way in, bike home).
Last holiday trip was by train, went very well, it was relaxing. The next one will be by first half by car, second half by train (leaving the car at relatives' home). We would have to 1. put the car on the train for a segment of the second half and 2. buy a parking ticket for the weeks we are up on the mountains 3. the village we stay at is car free anyway.
I took the overnight train from Bergen to Oslo once, that was not too bad as well.
I like trains for long distance travel as well. Unfortunately both my last planned holiday trips fell foul of Mick Lynch and his merry men, and I ended up driving.
Fuck 'em. I'll just go for standard human driving until we get FSD trains, which should be easier to build than FSD cars.
>>I'll just go for standard human driving until we get FSD trains, which should be easier to build than FSD cars.
Had them for ages dude. Look up the Victoria Line and DLR. The former is effectively FSD (Driver is there for show and emergency brake application), the latter is FSD except when it breaks down.
Sounds like all of the FSD train systems are single operator, closed network systems. It'll me interesting to see if anyone has or is trying to run FSD trains on a network shared with other operators who are using human drivers or a different FSD system. That would a lot more interesting, and probably a lot more difficult.
Sorry I have to ask, but you really need 11 hours to do LA to SF by train??? Goggle maps tells me that's ~560km.
Munich to Hamburg is ~615km. I can catch a train from Munich to Hamburg with a lot more stops in between than in the US in only 6,5 hours!
Hell in 11 hours I can get from Munich to London by rail (and that includes switching trains 4 times and having the associated waiting times in between!)...
You Americans really need to fix your Rail system!
"Sorry I have to ask, but you really need 11 hours to do LA to SF by train???"
That could be a good amount of time. You get on at 9pm and the train arrives at 8am the next morning. Since you would be traveling while asleep, you aren't burning up productive daytime looking out of a window. Effectively, you could have saved a day by going slower. I wouldn't want to get on a train at 5pm and arrive at my destination at midnight. By the time I'd get to a hotel and get my head down, it would be 1:30-2am. I'd only get a few hours of sleep before needing to get up, refresh for the day and get some breakfast. Instead, I have dinner before the train, a drink or two on the train and get a full 8 hours of sleep, a continental breakfast (coffee, tea, juice, danish, fruit, cereal) and I'm read to do whatever I'm there to do. If there is a shower on the train, I could return that same night and be all cleaned up and ready to go again the next morning back at point A.
Just look at the abomination of a High Speed Rail system that California is still wasting money on. They began before they had a full plan and had acquired all of the land and it's now truncated to go between two middling agricultural cities in the central valley nowhere near the biggest cities. There's already service up the middle of the state that is admittedly sort of slow, but it would have been much less money to make upgrades to that to bring the top speeds up to something more useful the same way as was done with the Intercity 125. That would knock a couple of hours off of a trip between LA and SF if they also spent the big money they would have to so they could get the train out of the LA basin. There are coastal tracks that Amtrak uses, but they are frequently out of service due to landslides and the ocean eating away the shore. They are also stuck going through places where it would be nearly impossible to upgrade to higher speeds and fewer/zero level crossings. A flight is a short hop, but all of the requirements that aren't the actual flying make it a whole day as far as I look at it. Driving is hours and hours and while a train might not shave too much time off of taking the car due to needing to make intervening stops, it's still not driving and possible to get useful work done or just relax for the trip.
11 hours if you're lucky. Last time I tried to do part of that route (Santa Barbara to San Jose), the train was 8 hours late getting to the Santa Barbara station. I gave up after four hours and drove, eating the cost of the ticket. I haven't attempted to take Amtrak since.
"11 hours if you're lucky. Last time I tried to do part of that route (Santa Barbara to San Jose), the train was 8 hours late getting to the Santa Barbara station."
The Coast Starlight route isn't very fast and only once per day between those two points. It's more of a tourist run than transportation. I've done that route once on a round trip and from Portland, OR down to LA and they didn't have any significant delays. With a new company on the tracks doubling the usage, maybe chances are good for upgrades on the track. I'm not completely sure, but I don't think that freight shares much of that route. A friend and I were at the station near Vandenberg AFB and I can't recall any trains going by at all. Service from LA all the way to San Luis Obispo is 2x daily via the Surfliner service.
Certainly not any car maker who's just spent years R&Ding Level 3 self-driving and is about to release it or has already done it and in either case beats Tesla which is stuck on Level 2 and just not improving.
Perhaps he can sell it to Lego or make it available in an Arduino kit for schoolchildren.
Who'd want it? Nobody that doesn't want to be critically dependent upon an asset owned and maintained by a competitor, which was designed and built around that competitor's physical product. On the other hand, any car maker that's sat on its backside in the belief that self driving cars were never going to happen, but now changed their mind might feel they have no choice.
Didn't downvote, but with respect to your final sentence, I can't help but feel that no product might be a better choice than integrating the product of a competitor.
Once the hype dies down and people realise that we're fairly far from having a vehicle that can completely drive itself, and that regardless of fancy names like "autopilot" and "full self driving" that imply things they aren't, the technology is more a fancy sort of driving assistance. One that not only still requires the meatsack, but said meatsack will also spend additional brain power on second guessing the assistance for when it does things like slamming on the brakes on a busy motorway because it thought it saw a puddy tat.
With that in mind, maybe being the car brand that doesn't have all this "smart" crap will become a selling point? Insurers can give you lower premiums because your car won't do dumb things. You might, but that's always been the case. One source of mishaps is better than two (especially if one of those sources is wrapped in legal mumble that disclaims liability).
> maybe being the car brand that doesn't have all this "smart" crap will become a selling point?
Yes, yes, a thousand times yes.
On our last three cars, one control stick was/is totally unused, covered in dust: the cruise control. Even commuting for years, never found it to be useful 
We fear the same would be true of lane keeping - what lane markings? Those faded ones that only appear near the T junctions on our leafy country roads? And so on. The automatic headlights are annoying as well: you start to rely on them, then find you don't have the reflex action on the knob when you need to override (and why the hell can't the headlights be flashed when they are dipped? Push & lock for high beam on, pull and spring release for flash - it worked on the previous models of the same brand of car!)
Heck, motorised seats - no need, never had any problems grabbing the manual release every time we swap in the driving seat. All these widgets and "features" are just more things to go wrong (and always at the worst moment). Even the bluetooth to the console - take it away, just more pointless fluff as far as we're concerned: plug in a USB of MP3 files and a radio, that is all we need.
Parking sensors: good (and pretty reliable after - how many decades have these been available now?). Parking camera? Fun when we tried it, but really haven't missed having it day to day.
At some point we're going to have to change the car again and it'll probably be an EV. But so very many of the current models are stuffed with things we just do not want to bother with and don't need to pay for.
Please, please, p!ease, let there be nice, basic but functional models for us to get when the time comes (OR suddenly improved public transport at sane prices with transferable tickets - bring back the good bits of British Rail!).
 for *our* driving needs! You may well feel that cruise control has been a godsend. Ditto for any and every gadget out there. Good for you, still irrelevances for us.
"Please, please, p!ease, let there be nice, basic but functional models for us to get when the time comes "
The UK/EUR gets to see many more models than the US where it's nearly all luxury priced vehicles due to government interference. The last review I saw for an Ora was showing a load of those useless features on it rather than like with their R1 that eschews them other than a reversing camera which I wish I had. The R1 is(was) selling for about US$12,000 in China.
Nobody that doesn't want to be critically dependent upon an asset owned and maintained by a competitor, which was designed and built around that competitor's physical product.
The fact that everyone is switching to Tesla's chargers suggests this isn't a concern.
Do you mean "Tesla's design of charger is becoming a de facto standard as other brands fit compatible ports and run their own compatible charge points"? Because either the industry will chose one (and it matters not one jot in the grand scheme of things which one) or fed up populace will get the government to dictate one (and it matters not ...).
Which is very different from licencing software that has been trained - not designed, with easy to locate and tweak parameters such as length, height, wheel base, but blindly trained - on only the physical designs of vehicle that Tesla build. Only expecting the Tesla sensor suite (so no accurate direct-measurement parking sensors, let alone any fancy stuff, like fog-penetrating radar).
"The fact that everyone is switching to Tesla's chargers suggests this isn't a concern."
In China, Tesla has had to conform to the China national standard for EV charging. In many parts of the world, Teslas have been fitted with the CCS standard port. In the US, what has been Tesla's proprietary plug seems to have been submitted to become a standard through SAE International, formerly named the Society of Automotive Engineers. I don't think it can be accepted as a standard unless Tesla relinquishes sole control over it. Tesla may have figured out that in order to be allowed to get in line for government money, they have to do that. It would allow them to better monetize the chargers they have, add more AND be able to collect vast amounts of data on more people, not just Tesla drivers. Many cars now will transmit their VIN so people can plug-and-charge automatically. You can do a VIN lookup and it will tell you many things about the car. Not who owns it, but that's a given since the person has to be signed up to use that billing method.
A true standard means that everybody will be able to offer it..... it's a standard. All of the established US charging providers will start adding the 'formerly known as Tesla plug' to their stations which means more places for Tesla owners to charger up. If Chargepoint is closer to someplace I want to get some food, they'll use Chargepoint over a Tesla Supercharger. I think that's going to be a bigger metric, location and proximity to shops. The costs in a given area aren't likely to vary much from supplier to supplier and an extra dollar to not have to cross a major road to get to a restaurant isn't a big deal.
As the earnings reveal Tesla's margins reverting to the industry mean, the fact that the company that was once going to replace legacy automakers with a fleet of robotaxis is now planning to flog them their noddy lane assist software says it all.
Retail investors may not be quite ready to understand what's happening, but every sign is that Tesla is actually 'just' a car manufacturer, and one that only effectively has one volume model at that. When your PE multiple is over 70, a margin in the teens (and dropping) is not a good indicator. Though of course the 'experts' on YouTube have been spending the last week hyping the stock to the sky...
..and if you're a mainstream manufacturer that has witnessed the stream of errors, system failures, inexplicable bugs and failure to respond from Twitter under Musk's leadership, or has read the widely shared online stories of disastrous engineering practises behind the Tesla software stack, why on earth would you trust your reputation with an FSD license? For now, the reputation Tesla has is clearly worth something to the brands that were caught napping, so I can understand them wanting (a little) to look like they're involved, but if they're serious about competition, they should be putting a large gap between themselves and any of the technology coming out of Tesla.
Commenter Jellied Eel on this article:
pointed out that Tesla's core business is financial trading based on Carbon Credits. The auto business allows them into this market, and all the complex financial products you can create based on them.
I remember reading a long time ago that Ford made the most money from its loan business. The cars were just a way to get customers to take a loan out with them.
"As one of our biggest local dealership groups teaches their salespeople, "The car is just a vehicle to sell a loan"."
This is why you never let a dealer know if you are going to pay cash until you have a firm quote from them. If they think you are going to work with them on the financing, they'll make you a better deal on the car to get more money as a kickback on the loan. A private seller with a used car can be motivated with soft folding money. If you pay partially in currency, you might get them to record a lower price which translates to less money going to the Big Guy.
"pointed out that Tesla's core business is financial trading based on Carbon Credits."
I just saw a story that showed the money from Carbon Credits took a big hit the last quarter for Tesla. The more the competition introduces their own EV's, the less credits they'll need to buy from Tesla. Rumor is that the Chevy Bolt lost money at least initially, but the free government money in the form of Carbon Credits saved them from having to buy them to sell large pickups and SUV's. Net, selling the Bolt at a loss was a money maker (due to government meddling).
Net, selling the Bolt at a loss was a money maker (due to government meddling).
Depending on your point of view, isn't that "meddling" result exactly why governments did/do that? Chevy would not have produced the Bolt at all once they saw the very slow, if ever ROI on it without "government meddling". If you believe switching to EVs is good and worthwhile, then that "meddling" is exactly the right thing to do. If you thinks EVs are expensive, a waste of time and believe the end to end lifetime pollution, much caused by rare earths mining, is a bad thing, then obviously that "government meddling" is the absolute worst waste of money.
You should see the nitwits being enthusiastic about the cybertruck.
First off, adding 'cyber' to a vehicle name is a red flag that tells you marketing had *way* too much input. Secondly, we know from DeLorean that unpainted stainless steel panelling are a sod to maintain - the flipside being that you can at least maintain and repair them, unlike aluminium. Thirdly, that shape has (to borrow a Jeremy Clarkson expression) the aerodynamics of a cathedral which, coupled with the high weight of the useless armoured windows will kill available range something fierce.
Tesla has become Microsoft in testing just how much they can abuse their fan base which is not a good move for a company that wants growth. That said, having a mildly insane CEO who cannot keep his trap shut isn't IMHO helping either..
Small self correction: you *can* fix aluminium (even aluminum :) ). Especially if you can fill and paint afterwards.
It gets a tad more complicated with structural parts, but I assume someone has worked out the correct allowable tolerances and welding parameters for that by now.
"stainless steel panelling are a sod to maintain - the flipside being that you can at least maintain and repair them, unlike aluminium. "
It's not that easy to repair SS unless you plan on painting it after. It would take a real master to weld in a patch and finish it to seamlessly blend in.
OK, but it's not a mirror finish so is it not normally a case of sanding it down until you match the brush finish of the rest?
I've looked at some DeLorean restauration footage (the sheer volume of which actually astonished me), and they seem to do just that after a panel repair.
"I've looked at some DeLorean restauration footage "
It CAN be done and I've seen some really good work, but the level of competence to get good results is high. Much higher than many working in body shops right now so those shops can assign that work to the first available person, but to a specialist which is more $$. If they lose that person/people by paying crap wages, they could wind up with repairs to do that they have to allow people not up to it to do or just tell the customer they need to take their car someplace else. It could be smart to do that latter in that situation as a load of back and forth on mediocre repairs is costly in money and reputation.
"the company that was once going to replace legacy automakers with a fleet of robotaxis is now planning to flog them their noddy lane assist software says it all."
I just watched a video that explained how the whole robotaxi thing was complete BS. Something I knew intuitively, but couldn't articulate. The numbers that Elon was tossing about casually as being conservative made no sense. If they did, Tesla would need to immediately stop selling cars to the public and concentrate on deploying robotaxis themselves as quickly as possible. It would be worth Trillions or more. They'd kill off Uber, Lyft and all taxi companies in just a few years. That they didn't do that confirms he was talking out of his.....
They maxed out the loan to get the flashest BMW they could - then realised that they could not afford to keep replacing the amber-coloured bulbs: amber is expensive and really doesn't last long as a bulb filament. Engineering meets marketing.
FSD is software. Improvements to software do not happen on a logarithmic curve, ever : the more code you generate, the harder it is to improve. You also generate more bugs. The more complex and interdependent the software gets, the worse the problem gets as it grows.
Unless the logarithmic curve he's talking about is a downwards one for quality. That's achievable all too easily, especially for bosses who set unrealistic targets and work their staff into the ground.
It's worse than that. He's talking nonsense. What he describes is not a logarithmic curve. Those plateau to a flat line.
He might *MEAN* exponential, I guess (those grow faster and faster and faster and spike upwards). Or he might mean that progress will constantly become even slower over time (logarithmic).
But worse than even that... This is "AI".
In which the gains after initial training plateau to nothing.
One million data points and it (erroneously) still think it's appropriate to turn left here? It's gonna take one-million-and-one contrary but similar data points before the algorithm starts modifying its behaviour of its own accord. Repeat ad infinitum. Or add in a ton of human heuristics at which point, you're basically just programming it yourself.
AI plateaus, which means the curve is logarithmic. Which means improvements are impossible beyond a given point, because all progress stagnates and flatlines no matter how much time, energy or money you throw at it.
One million data points and it (erroneously) still think it's appropriate to turn left here?
This afternoon, I turned into a B road from an A road, following a motorcycle. A bit further on, the headlights flashed on a van that was travelling in the opposite direction. I wondered why, as I could not see a hazard immediately - a couple of seconds later, the motorcycle slowed down, and I could also see a large bird in front of the motorcycle. The angles were such that at times, the bird disappeared from view as it was running along the road. About a 50m later, it took off into the trees on the side of the road - it was a young peacock!
Now, can the image training teach an AI system into understanding the scene as I viewed it, with what was developing in front of the motorcycle that I was following?
Please correct me if I am wrong, but the impression I currently have is that:
Tesla is relying on the latest fads surrounding Neural Net machine learning (as seen, so brilliantly, at work inside ChatGPT et al) whilst the other automakers are making more use of techniques that *used* to be the subject of AI research but are now just well-worn everyday methods that aren't AI at all.
For example, taking just one of the simpler parts of the whole FD experience, parallel parking (as that is offered by a number of brands already) with examples from (among others) these two videos:
Teslas are continually feeding inputs into the NN and adjusting what they are doing, which may mean dithering and changing its mind partway. Others are putting together a plan, showing it to the driver and then acting upon it (with appropriate continual feedback from parking sensors, pinging away, just in case the world changes).
Tesla is also *removing* parking sensors and relying entirely on the cameras - which not only seems against common sense (measure twice...) but means that instead of an iterative change in the software you are being subjected to new software, with its own set of fascinating behaviours.
BTW yes, one of those was a 2019 model Tesla, but they still sold it as working!
 Planning, machine vision - even limiting the scope of the "vision" or abandoning visible for other sensors to fit the application - were the subject of AI back in the day whilst it was all experimental: I remember one presentation of a successful task - using AI & machine vision to verify the quality of robotic welding in real-time - which started with full-colour video and reduced the visual sensor down and down until it was the simplest mechanism that would do the job (and made the software component simpler and more robut), and it did it well.
PS: I was taught to always try to avoid using the power steering to grind the tyres round whilst the vehicle isn't moving. Yet all of those expensive cars with their clever parking were happily grinding away. Perhaps I'm just being too precious with my tyres (they ain't cheap!)
I dunno. My daily "commute" is often many miles, and there are many others in a similar situation that could quite happily spend an extra few hours asleep, or catching up on some reading or, even work while "driving" to work/office/next customer etc. I do enjoy driving, but there are times I'd happily get into a genuinely self driving car and have a kip while it does the work :-)
Maybe top a problem as such, just a "nice to have" :-D
"could quite happily spend an extra few hours asleep, or catching up on some reading or, even work while "driving" to work/office/next customer etc."
Transfer Booths would be pretty cool too, but there isn't even an inkling of a hypothesis how that could work.
There are plenty of videos where Tesla drivers have nodded off on the freeway and while people honk their horns and there should be a beeper going off in the car, they just sit there and drool on themselves. The highway patrol are willing to get in front of the car and attempt to get it to slow down and stop, but that's not a good solution. Let's say the car does come to a stop, in traffic with a sleeping driver that is so used to the alert chime that they don't register it anymore, how dangerous is that? I'd so want to walk up to the car and bang loudly on the roof while somebody rocks it back and forth. With any luck, they'd soil themselves.
This post has been deleted by its author
LOL. On the other hand, the Bond "submersible" was multiple different "wet" and "dry" versions, including small models and Musk has said he plans to retrofit his to work as advertised. It'll be fun to see if he pulls it off. On the other hand, he doesn't have a good track record with submersibles and PR :-)
Electric traction is ideally suited for this application, and really, he's had more than enough time since he acquired it to have made it happen. He just needs to commission a specialist.
The traction motors could be housed within the wheel, which means the retraction mechanism does not have to transfer the drive. There was an early adoption of a MINI for electric drive that had such motors. Of course that adds unsprung mass, but for this limited use case it is fine. Added complication here of the need to prevent water ingress on submersion.
I read somewhere that in the scene where the vehicle drives up beach was done with a winch pulling it up with the cable buried in the sand