A thread to discuss self-driving cars and other vehicles. No politics, please.
Technology, commercial aspects, legal considerations and marketing are all on-topic.
- Companies (e.g. Tesla Inc.) are only on-topic when discussing their self-driving products and research, not their wider activities. The exception is when those wider activities directly impact (or are impacted by) their other business areas - e.g. if self-driving car development is cut back due to losses in another part of the business.
- Technology that's not directly related to self-driving vehicles is off-topic unless you're discussing how it might be used for them in future.
- If we're talking about individuals here, that should only be because they've said or done something directly relevant to the topic. Specifically, posts about Tesla do not automatically need to mention Elon Musk. And Musk's views, politics and personal life are firmly off-topic unless you can somehow show that they're relevant to self-driving vehicles.
Cadillac plans to release a scaled back, more simple version of similar technology by 2015 - what they call "Super Cruise", which isn't total self-driving, but does let you relax on highways. It positions your car in the exact center of a lane, slows down or speeds up as necessary, and is said to be meant for ideal driving conditions (I'm guessing that means ideal weather, no rain or snow, etc.).
I am looking forward to such tech. If enough people prefer to drive this way, and the technology works reliably, it could result in safer roads with fewer accidents. Another possibility is that, using GPS and maybe the ability to know ahead of time which roads are most clogged, they can find the quickest route from place to place.
On the other hand, hacking could be a real concern, and I hope it doesn't become a serious threat. It's looking like we're living more and more like those sci-fi Everything Is Online worlds depicted in fiction for a long time.
(Mod edited to replace original post)
Edited by Mrph1 on Mar 29th 2024 at 4:19:56 PM
As was mentioned last time it came up, what Musk means when he says "feature complete" does not match what most people think "feature complete" means. When Musk has used the phrase "feature complete", he basically means that it will be able to recognize everything in the world around it. These are lane markers, those are road signs, this is another car, that's a pedestrian, etc. What he does not include is the ability to successfully predict interactions with those elements. Some of them (like road markings and signs) are static and won't change, but others (like cars and pedestrians) are doing things of their own volition, and Musk's "feature complete" self-driving software doesn't take that into account.
Other self-driving developers have been working on the "prediction" problem for years and are still in testing. Musk has suggested that the problem is easy to solve and it won't take very long after the main "feature complete" recognition software is ready. Of course, if it was that easy, then you'd think other people working on it would have already solved it.
The alternatives are either that the problem is more difficult than Musk gives it credit for, or that Musk is a super genius who can solve a complicated, difficult problem far more easily than anyone else working on it can.
Here
's the same source I linked last time for context about Musk's "feature complete" comments.
Right, but I think what Musk means is that the car will recognize and react appropriately to 99+ percent of situations, and they'll keep collecting data and implementing solutions for the edge cases that come from those problem scenarios. What gives Tesla the advantage above and beyond any hardware or software they've developed is data: billions of road miles of people using their software and intervening when it does something wrong.
Edited by Fighteer on Jul 16th 2019 at 11:38:53 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"That's what you're meant to think, because that's what "feature complete" sounds like. But that's not actually what Musk means when he says "feature complete", because Tesla's nowhere near that point yet.
What it does mean is that Tesla vehicles will be able to successfully navigate static environments. They'll be able to stay in their lane, stop for stop signs, avoid running into stationary obstacles, etc. That's legitimately impressive! It's a difficult problem and a huge step toward solving the entire self-driving issue.
But solving the dynamic environment problem that takes into account other drivers, pedestrians, animals crossing the road, that ladder falling off the back of a pickup on the highway, etc etc etc is another equally enormously difficult problem. Worse, it's an entirely different type of difficult problem ("what is this?" recognition vs "what is this likely to do?" prediction), so I remain skeptical that solving the second problem will be as quick and easy as Musk claims after the first problem is solved.
Really from Jupiter, but not an alien.Skepticism is valuable. I've seen videos of "action" tests where Teslas have responded appropriately to those kinds of dynamic environments, so I'm going to offer that experience is the ultimate proof. To be fair, I've also seen videos of Teslas acting wonky in certain situations. There are YT channels dedicated to driving these cars around and deliberately stress testing each software update in both old and new situations. Watching them can be very educational. What you're looking for is: does the car do the right thing before the driver feels the need to intervene?
The point of all this is that, by buying a Tesla, you are getting a product that improves over time.
Edited by Fighteer on Jul 17th 2019 at 9:10:51 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"I'd rather wait until all the bugs have been worked out.
I avoid buying stuff like a videogame upon release because I don't want to put up with the initial bugs. So of course I'd definitely be very cautious when it comes to an AV.
Edited by M84 on Jul 17th 2019 at 9:14:21 PM
Disgusted, but not surprisedNot like you can't drive it manually and get the full EV experience. Even if you don't trust FSD, they're still the safest cars you can buy.
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
That’s the kind of double standard I can’t stand. My mom will probably never be comfortable in a self-driving car, yet human driven cars are already at the center of such a large percentage of accidental deaths.
Seriously. After all these years, my mother still doesnt know how to parallel park. When you add together the elderly whose response times are significantly below average, the overconfident young who drive recklessly, long distance drivers who are asleep at the wheel half the time, people driving drunk... if collectively we break an average of 80% error free driving I would be greatly surprised.
I'm done trying to sound smart. "Clear" is the new smart.TBF, AV should be held to a higher standard. Isn't that the whole point? That they are supposed to be far superior to human drivers?
Nobody is expected to be satisfied with a shitty slow-ass computer just because it can still do spreadsheets faster and better than a human.
Edited by M84 on Jul 17th 2019 at 10:12:32 PM
Disgusted, but not surprisedYes, obviously. I've heard four to five times safer than a human as a baseline standard for regulatory acceptance. Tesla FSD is already safer than humans under normal conditions in highway driving; I assume they won't release the city driving update until they're satisfied that it can achieve a similar standard.
What's going to drive the debate is the edge cases — situations where a human would react one way but the software doesn't behave as expected.
Edited by Fighteer on Jul 17th 2019 at 10:14:28 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"Tesla announced its Q2 earnings: 6.3 billion USD with a 200 million operating loss and 400 million net loss. Free cash flow was positive, at 600 million, and the company raised 2.4 billion in additional capital for a 5 billion USD cash position. Gross margin was 14.5%; auto gross margin was 19%. [1]
While the financials aren't as stellar as many people hoped, the company appears stable and solvent, continuing to experience dramatic year-over-year growth. By comparison, Q2 2018 saw 4 billion in revenue and an operating loss of 600 million, and Q2 2017 saw 2.8 and 200, respectively. Not many companies can boast 50 percent annual growth.
Unheralded but very important: Tesla's energy business saw massive growth in Powerwall deliveries at 415 MWh, nearly double its best past quarter with the exception of the Australian project in Q1 2018. Solar is way down, unfortunately, but it's believed that they are simply not pushing it until their Solar Roof is ready.
Edited by Fighteer on Jul 25th 2019 at 7:14:55 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"Yeah, no. They're cash flow positive, so that is literally impossible. I'm not sure what numbers you're looking at.
One thing that came out is that they are deferring revenue on Full Self-Driving purchases. So, when you pay that 6K USD up front, Tesla isn't recording all of that as part of their top line, since they haven't delivered the complete feature yet. This helps explain the difference between cash flow and operating revenue.
Still, even if you assume that their GAAP loss is equal to their cash flow, which it isn't, and that they would make no improvements, which they wouldn't, they'd have a solid 12 months of life left.
Tesla's earnings report is consistent with a technology company investing heavily in growth. Move along, nothing to see here.
Edited by Fighteer on Jul 25th 2019 at 8:01:47 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"Even with the better Q2 earnings, it's not exactly a secret that the company has been struggling during its lifetime. It's not for nothing that Musk decided back in March that Tesla had to trim a lot of fat away to avoid going broke within months.
It doesn't help that analysts are already predicting that Tesla's stock price may drop again after this week.
Tesla's great at innovation, but it just has trouble grasping the business and production side of things. For better or worse, Tesla relies heavily on Musks' ability to get investors on board with him.
Edited by M84 on Jul 25th 2019 at 8:11:20 PM
Disgusted, but not surprisedMmmhmm. I guess the FUD spreads to unexpected places.
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"A reminder that Tesla isn't the only one working on this:
DeepMind is helping Waymo evolve better self-driving AI algorithms
The gist of it is that they're planning to use evolution-based algorithms in their SV. Which is interesting to say the least.
Disgusted, but not surprisedInteresting, to be sure, and I'll be very curious to see where it goes. The problem for Waymo, Google, and other companies doing this work is that, while they have good ideas, they don't have hundreds of thousands of cars on the road running their software and providing data to feed them.
Edited by Fighteer on Jul 25th 2019 at 10:46:18 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"Tesla's finances work basically like a tech startup. They bleed cash to develop product, earn a ton of revenue when that product releases, and then rinse and repeat. This is basically how the product cycle of any company works, but major corporations have lots of products in various parts of the cycle at any given time, so their finances don't look so crazy.
Tesla probably has enough going on now that they could slow down and start working like a more traditional company, but it's pretty clear that Musk has no interest in doing so. That's probably a good thing, as it means faster innovation and faster growth... as long as he can keep all the plates spinning. The downside of this approach is that if a product flops, the whole company is SOL since there's no margin to fall back on while you recover.
Really from Jupiter, but not an alien.In the wake of Tesla's latest stock drop this week (almost 14%), investors seem concerned. Though the bears seem to have made good out of it.
From Bloomberg: Tesla’s Biggest Decline in 10 Months Rains Profits on Shorts
The robo-taxi thing may indeed be their best bet.
Op-ed from Financial Times: Why Tesla’s best hope may lie in robotaxis
GM Cruise is also being more cautious about the whole self-driving car thing, having walked back their earlier promise to deliver by the end of the year.
From Reuters: GM Cruise to delay commercial launch of self-driving cars to beyond 2019
Of all things, Walmart seems to be getting a leg up in this field.
Tech Crunch: Gatik's self-driving vans have started shuttling groceries for Walmart
Edited by M84 on Jul 28th 2019 at 6:27:06 AM
Disgusted, but not surprisedAutomated delivery seems like a huge market for self-driving vehicles, but how are they solving the problem at the receiving end? Does someone have to leave their home to retrieve their packages? Is there a little robot inside the vehicle that delivers to the front door? How do they ensure that each customer gets only and exactly what they ordered? How do they deal with theft and/or vandalism?
It seems that these problems are solvable in the long-term, but in the short term they are critically difficult.
Tesla is supposedly very close to releasing a major update to Summon, the feature that lets the car drive itself out of a parking space and find you. This may coincide with a increase in the list price of Full Self-Driving, which is an addon that can be purchased with your vehicle or after.
The current beta version of Summon is very conservative, can easily get stuck on unexpected conditions, and doesn't do a great job with curbs.
Tesla has said in the past that FSD is a feature that will become more expensive over time, an incentive for buyers to get in early since they will get all the features as they are released regardless of when they bought it.
Edited by Fighteer on Jul 28th 2019 at 2:15:16 PM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"I would envision deliveries being shipped in standardized formfactors that can be handled by a single armature or equivalent. The vehicle drives up to the curb (teaching it how to identify the proper drop-off point will be fun), extracts the package, deposits it on the curb, then leaves. It's not going to go ring the doorbell, obviously, so the recipient will need some kind of automatic instant notification that their package was delivered.
For apartment buildings and other situations where there's no driveway or curb, or townhomes without clarity as to which sidewalk/entryway belongs to whom, the challenges multiply.
Edited by Fighteer on Jul 28th 2019 at 9:31:51 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"

That ties into my post earlier. As the manufacturer and owner of a robo-taxi fleet, Tesla keeps all the profit. A consumer who buys a FSD car for 50,000 and earns all the money from the taxi service is a major loss to the company. It only makes sense for the cars to become more expensive over time.
However, for that to come to fruition, Tesla needs the capital to produce and deploy that fleet. Ergo, they need to sell cars until they have reached a critical mass. This is one reason why Tesla would love people to take the lease option on their Model 3 and Model Y. You lease the car from Tesla for private use, Tesla takes the car back when the lease expires and turns it into a fleet vehicle, keeping all of that sweet profit.
Waymo is making progress in autonomy but it's not there yet. Elon has said that their technology approach is a dead-end, but time will tell whether he's right. He's said that Tesla FSD will be "feature-complete" by the end of 2019. That's a huge promise, but if true, they'll have won the war before most of the participants even enter the field.
I can also see Tesla partnering with Uber and Lyft to let their AVs work inside those companies' fleets. It depends a lot on how financially stable Tesla can stay on its own.
Edited by Fighteer on Jul 16th 2019 at 9:45:57 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"