How a billion dollar autonomous vehicle startup lost its way(bloomberg.com)
bloomberg.com
How a billion dollar autonomous vehicle startup lost its way
https://www.bloomberg.com/news/features/2018-08-13/how-a-billion-dollar-autonomous-vehicle-startup-lost-its-way
40 comments
It's entirely possible that self-driving will just be an auto part, sourced from the usual big auto parts makers.
I've always assumed that this is the eventual outcome, even if there is a brief period where SelphDrivin, LLC, et. al., sell their own cars. For decades Bosch would sell a whole fuel injection and ignition system for your new model. Brembo and Bendix will sell you some brakes. I don't know who makes ABS sensors and computers, but I'd guess it's not GM (probably Delco, though) and Ford. MobileEye's lane assist was an option on the Mercedes Sprinter we bought. The next logical step in my mind is self-driving systems.
I've always assumed that this is the eventual outcome, even if there is a brief period where SelphDrivin, LLC, et. al., sell their own cars. For decades Bosch would sell a whole fuel injection and ignition system for your new model. Brembo and Bendix will sell you some brakes. I don't know who makes ABS sensors and computers, but I'd guess it's not GM (probably Delco, though) and Ford. MobileEye's lane assist was an option on the Mercedes Sprinter we bought. The next logical step in my mind is self-driving systems.
Speaking of MobileEye, from what I understand their goal is to slowly work up their parts offerings from driver assist to self-driving.
This makes sense from a branding / insurance / shiriking responsibility perspective also.
“Our newest Toyota uses a state of the art NeverHandsTM system!”
After a few bad car crashes: “We at Toyota have decided to discontinue our relationship with NeverHands following the NTSB review of their product.”
“Our newest Toyota uses a state of the art NeverHandsTM system!”
After a few bad car crashes: “We at Toyota have decided to discontinue our relationship with NeverHands following the NTSB review of their product.”
I suspect this will be a fairly regular story in the next few months/years. Self-driving tech seems to be a lot further off than people think so a lot of companies that were formerly very exciting will not be able to deliver and will need to close up shop. I think there's some possibility that it will go the way of flying cars at least for now.
Fortunately a lot of the tech that does work can find a second life as progressive enhancement for human drivers. Blind spot detection, auto lane following, and smart cruise control are great features that ease some of the burden in driving and I think it's good to see them being more widely distributed in newer cars.
Fortunately a lot of the tech that does work can find a second life as progressive enhancement for human drivers. Blind spot detection, auto lane following, and smart cruise control are great features that ease some of the burden in driving and I think it's good to see them being more widely distributed in newer cars.
I thought Waymo has pretty much cracked that nut. They've covered several million miles of testing in AZ and CA. You might be right, but do you have any links where I could see where your skepticism is coming from.
Waymo vs Quanergy is apples and oranges I think
Waymo vs Quanergy is apples and oranges I think
Not the OP, but I think rain and especially snow and fog don’t happen that often in those two States and afaik driving autonomously under those circumstances is still an unsolved problem with no easy solution in sight.
Now, if you want to extend the self-driving thing to other parts of the world you’ll most probably need to solve the singularity problem, at least for driving in the Eastern European capital city where I’m currently living where you have to actually “guess” what the other traffic participants will do next: is that kid going against the flow of traffic on his bike? (answer: yes) Is that cab going to cut me off in the middle of the intersection and should I take my foot off the acceleration pedal so that we won’t crash? (answer: yes to both questions). Are those pedestrians going to jaywalk without checking for any incoming vehicles? (answer: yes). All these things happened to me during my short city drive half an hour ago.
Now, if you want to extend the self-driving thing to other parts of the world you’ll most probably need to solve the singularity problem, at least for driving in the Eastern European capital city where I’m currently living where you have to actually “guess” what the other traffic participants will do next: is that kid going against the flow of traffic on his bike? (answer: yes) Is that cab going to cut me off in the middle of the intersection and should I take my foot off the acceleration pedal so that we won’t crash? (answer: yes to both questions). Are those pedestrians going to jaywalk without checking for any incoming vehicles? (answer: yes). All these things happened to me during my short city drive half an hour ago.
All of those scenarios are definitely a thing in sunny clear Oakland, CA. And San Francisco, where a lot of the testing has been run, is notorious for both its hostile and unpredictable driver/biker/pedestrian behavior, and for its fog. Rain and snow, which change the observed road surface, seem to be the harder problems.
That said, no North American city comes within an order of magnitude of the complexity of driving in Latin America.
You can forget about traffic rules, and you very often have to check the other driver´s body language to make split-second decisions, and play a game of "chicken" to cross a busy thoroughfare (inch through, then force yourself in front of an incoming car and expect him to stop). Giving way to other cars means you might be stuck in a traffic jam, so drivers can be incredibly aggressive - and I wouldn´t want to drive with an AI trained to try to battering ram their way around.
There are cities where I wouldn´t trust myself to drive (Lima, Peru is one I've experienced firsthand).
I'm hopeful for self-driving, but it's going to be very tough (and might take decades to adapt to the third world)
You can forget about traffic rules, and you very often have to check the other driver´s body language to make split-second decisions, and play a game of "chicken" to cross a busy thoroughfare (inch through, then force yourself in front of an incoming car and expect him to stop). Giving way to other cars means you might be stuck in a traffic jam, so drivers can be incredibly aggressive - and I wouldn´t want to drive with an AI trained to try to battering ram their way around.
There are cities where I wouldn´t trust myself to drive (Lima, Peru is one I've experienced firsthand).
I'm hopeful for self-driving, but it's going to be very tough (and might take decades to adapt to the third world)
>at least for driving in the Eastern European capital city where I’m currently living where you have to actually “guess” what the other traffic participants will do next: is that kid going against the flow of traffic on his bike? (answer: yes) Is that cab going to cut me off in the middle of the intersection and should I take my foot off the acceleration pedal so that we won’t crash? (answer: yes to both questions). Are those pedestrians going to jaywalk without checking for any incoming vehicles? (answer: yes). All these things happened to me during my short city drive half an hour ago.
Those sound very quaint. Try India, China, SEA, most of Latin America and so on...
Those sound very quaint. Try India, China, SEA, most of Latin America and so on...
Is there a reason you have brought up the inferiority of third-world countries in your last several comments?
Blind comparisons (to put things into perspective) aren't useful. You need to have a point of some kind.
Blind comparisons (to put things into perspective) aren't useful. You need to have a point of some kind.
It's not really inferiority: the commonality of the other countries is that they typically follow a different regime of negotiation of the road. In specific, the USA and other anglophone countries are extremely strict about what motions are allowed on the road: the expectation is that everyone stays precisely within the lanes and drives a bit robotically.
This, however, is actually not that common in most of the world, where the road is much more 'negotiated': speeds are slower but the streets more chaotic. I wonder sometimes if it's more like treating the vehicles as horses rather than boxes of death, or something similar: it's a markedly different mode.
I imagine his point is that self-driving cars rely on the robotic driving customs of parts of the West to do as well as they do, and that he's skeptical of these vehicles' ability to interact with more dynamic streetscapes filled with e-scooters, rickshaws, tuk-tuks, people carrying stuff, and so on.
(Source: living in China/Asia)
This, however, is actually not that common in most of the world, where the road is much more 'negotiated': speeds are slower but the streets more chaotic. I wonder sometimes if it's more like treating the vehicles as horses rather than boxes of death, or something similar: it's a markedly different mode.
I imagine his point is that self-driving cars rely on the robotic driving customs of parts of the West to do as well as they do, and that he's skeptical of these vehicles' ability to interact with more dynamic streetscapes filled with e-scooters, rickshaws, tuk-tuks, people carrying stuff, and so on.
(Source: living in China/Asia)
>Is there a reason you have brought up the inferiority of third-world countries in your last several comments?
Is there a reason you have been "stalking" my last several comments across different threads? Checking what this or that commenter wrote in different threads to call them on it is kind of abusive.
For my part, the reason is quite simple: I first wrote about such countries in a thread, and so had the situation there fresh in mind when I encountered a couple of other threads with similar issues. I've also travelled and worked in those places for periods of time, so I also have them in mind.
Also, note that I didn't bring up the "inferiority of third-world countries" (those are your words).
I mentioned how they drive in some of them - which I don't find inferior. If anything it's more lively and less obsessed with cushioning the individual, which I appreciate).
In another comment I mentioned how countries near the tropics where billions live have lots of mosquitos and lots of heat -- and that was again in a positive spin, to contrast someone complaining about how Houston was unlivable before air-conditioning because of such things.
And that's it, a grand total of 3 comments, 2 about traffic patterns and driving, and one about heat/mosquitos.
Not all places I mentioned are "third world" either. I explicitly included e.g. China which is the 1-2 biggest economies in the world. Do you consider all of those places "third world"?
>Blind comparisons (to put things into perspective) aren't useful.
"Blind" is a weasel word without which there's no argument above.
Those are not "blind comparisons", they are simply "comparisons". Putting things into perspective is extremely useful alone.
In fact you first need to put things into perspective (survey what's going on across countries/domains/etc) and only _then_ form your "point". Else it's just a priori dogma.
Is there a reason you have been "stalking" my last several comments across different threads? Checking what this or that commenter wrote in different threads to call them on it is kind of abusive.
For my part, the reason is quite simple: I first wrote about such countries in a thread, and so had the situation there fresh in mind when I encountered a couple of other threads with similar issues. I've also travelled and worked in those places for periods of time, so I also have them in mind.
Also, note that I didn't bring up the "inferiority of third-world countries" (those are your words).
I mentioned how they drive in some of them - which I don't find inferior. If anything it's more lively and less obsessed with cushioning the individual, which I appreciate).
In another comment I mentioned how countries near the tropics where billions live have lots of mosquitos and lots of heat -- and that was again in a positive spin, to contrast someone complaining about how Houston was unlivable before air-conditioning because of such things.
And that's it, a grand total of 3 comments, 2 about traffic patterns and driving, and one about heat/mosquitos.
Not all places I mentioned are "third world" either. I explicitly included e.g. China which is the 1-2 biggest economies in the world. Do you consider all of those places "third world"?
>Blind comparisons (to put things into perspective) aren't useful.
"Blind" is a weasel word without which there's no argument above.
Those are not "blind comparisons", they are simply "comparisons". Putting things into perspective is extremely useful alone.
In fact you first need to put things into perspective (survey what's going on across countries/domains/etc) and only _then_ form your "point". Else it's just a priori dogma.
That's true. But Bird is a still billion dollar company selling transportation that is compatible with only certain weather and population density.
If you can make 5% of the world's transportation 5x better, you have an enormous success on your hands.
If you can make 5% of the world's transportation 5x better, you have an enormous success on your hands.
> Bird is a still billion dollar company
It's hype at this point, we'll see how that shakes out as well.
It's hype at this point, we'll see how that shakes out as well.
I’ve been in Paris this week and I’ve started to see more of them. I’ve found because they are silent they have an absolutely minimum impact on me.
They appear to be able to slow to walking speed and still remain stable, I’ve walked right along side one without any concern.
I’m thinking safety concerns are overblown. It’s a rad way to get around.
They appear to be able to slow to walking speed and still remain stable, I’ve walked right along side one without any concern.
I’m thinking safety concerns are overblown. It’s a rad way to get around.
Yet I still see a great deal of very irresponsible driving of those scooters. I expect there to be a lot of local pushback as regulators catch up.
It’s always too bad that regulations catch up instead of manners. Things could be so much easier...
To be fair though, there are lots of irresponsible pedestrians. People buried in their phones paying attention to nothing. Stepping into the road, ignoring crossing signals, taking up an entire sidewalk while walking in a group of three. People want to regulate the hell out of scooters, yet some lady with a dog on a long leash can pretty much do whatever she or or dog wants. If San Francisco denizens care so much about sidewalk etiquette, then I suggest prohibiting people from camping and shooting up and shitting on those sidewalks. That’s a bigger danger than anything I’ve faced as a pediatrian sharing space with Lime or Bird.
Someone is going to be seriously injured first.
The easy answer is to have autonomous driving be region-locked. To only work in regions that are tested, and to have those regions slowly grow as automakers slowly gather data from people driving around.
I think this has unethical implications at the system level - it would create an even more unequal distribution of the future, to paraphrase Gibson, and an unequal expectation or burden on labor.
I think this has unethical implications at the system level - it would create an even more unequal distribution of the future, to paraphrase Gibson, and an unequal expectation or burden on labor.
I've had the same thought. Where is the skepticism coming from?
I've noticed a lot of articles lately that talk about how far off self-driving cars are. That said the articles all seem to rest on one of two arguments. Some (like this article) point to the failure of some startup that most sane people didn't have any faith in. The others point to the small handful of accidents we've had: the one from Uber (which was the result of borderline criminal irresponsibility), and the couple from Tesla (which were the result of drivers not paying attention or and with their hands off the wheels of their emphatically not-self-driving cars). These sheer number of these articles creates the impression that there's a lot more going bad in the field than I think there's any evidence for.
When I try to dig deeper and figure out why these articles are all appearing I can't find anything.
Probably it's just that after all the stories about the amazing future of self-driving cars, this is the next natural story to tell since we don't actually have the self-driving cars yet, but there's definitely a paranoid part of me that thinks some industry group is sponsoring these stories for some reason.
I've noticed a lot of articles lately that talk about how far off self-driving cars are. That said the articles all seem to rest on one of two arguments. Some (like this article) point to the failure of some startup that most sane people didn't have any faith in. The others point to the small handful of accidents we've had: the one from Uber (which was the result of borderline criminal irresponsibility), and the couple from Tesla (which were the result of drivers not paying attention or and with their hands off the wheels of their emphatically not-self-driving cars). These sheer number of these articles creates the impression that there's a lot more going bad in the field than I think there's any evidence for.
When I try to dig deeper and figure out why these articles are all appearing I can't find anything.
Probably it's just that after all the stories about the amazing future of self-driving cars, this is the next natural story to tell since we don't actually have the self-driving cars yet, but there's definitely a paranoid part of me that thinks some industry group is sponsoring these stories for some reason.
Waymo has the best tech for sure and even they are far off from full autonomy. Full autonomy in Phoenix, sure, they're the closest, but not full global autonomy.
The other issue being those cars prolly cost somewhere in the $500,000 range at the moment. Not exactly a consumer product.
Check comma.ai if you're looking for cheap, accessible, self-driving capabilities (still a long ways off, but you can get hands free today on 15+ cars for under 1k in plenty of common situations).
The other issue being those cars prolly cost somewhere in the $500,000 range at the moment. Not exactly a consumer product.
Check comma.ai if you're looking for cheap, accessible, self-driving capabilities (still a long ways off, but you can get hands free today on 15+ cars for under 1k in plenty of common situations).
They have the best publicized tech. There are a lot more players in the space that don’t paint logos on the sides of their cars.
Why would a company with the best self driving car tech in the world be secretive about it? Sounds like a scam company to me. Google is releasing their disengagement statistics at least. If you could beat theres it seems like you would too.
This is kind of what I thought when I was thinking about entering the space. Especially when I started looking around and interviewing at different places, they all seemed to be doing the same thing in the same way. They were just repeating the successes of the DARPA Urban challenge, and making the systems more polished and capable. They made impressive demos and then got bought up for impressive amounts. But like most things, this follows the 80/20 rule, so most of the work is still ahead of us. We're probably 80% of the way there, but that last 20% is going to take decades and probably some fundamental breakthroughs in sensors and AI. It's going to take a while.
It just seems like every automaker heard that autonomous is the next thing, so they went out shopping for autonomous startups without a clear sense of what they were after and how hard it was going to be to get there. I think a lot of them are going to be feeling buyer's remorse.
It just seems like every automaker heard that autonomous is the next thing, so they went out shopping for autonomous startups without a clear sense of what they were after and how hard it was going to be to get there. I think a lot of them are going to be feeling buyer's remorse.
I've been enjoying Chevrolet's "adaptive cruise control" and "lane-keep assist". They both have a lot of room for improvement but are a good step into progressive enhancement, as you put it. I've been imagining different tools that we might start seeing in cars as this kind of product develops: adaptive high beams, rear view and side mirrors that auto-adjust to your gaze, automatic windshield wipers connected to a weather API and an internal camera system that measures visual noise from the windshield, a radar view in the dash so you know where the surrounding cars are... lots of little opportunities. Make the human driver redundant piece by piece.
Adaptive high beams are on Audi now, they are awesome. Weather API connected wipers? Bad idea, rain sensing already works, why use an API when you never turn on wipers before it rains anyway? Radar view for surrounding cars might be a nice gimmick but other than blind spot detection or parking assist, I’m not sure what benefit a radar display would have, unless it’s for parking, and an equivalent exists (Audi for example has 360 degree parking cameras and some cars have full autonomous parking already.)
But the fly-by-wire concept could be a good one — if reliability is ultra high.
But the fly-by-wire concept could be a good one — if reliability is ultra high.
Radar view (or some kind of value-add sensing view) gets consumers to start trusting autonomous systems. And I trust your judgement on the weather API - more generally, windshield wipers should be able to operate themselves :)
>Adaptive high beams are on Audi now, they are awesome.
I can't wait for BMW to have adaptive blinkers based on predicted driver behavior.
>Weather API connected wipers? Bad idea
Especially because the radar map will often see rain that's not making it to the ground
I can't wait for BMW to have adaptive blinkers based on predicted driver behavior.
>Weather API connected wipers? Bad idea
Especially because the radar map will often see rain that's not making it to the ground
I'm not saying this is as extreme a case, but I got a whiff of similarity to Theranos, in the aspect of an emphasis on loyalty making it difficult for the CEO to listen to debate. Of course, that could also be because the journalists covering tech businesses nowadays are looking for the next Theranos story.
“I hate the word ‘wall’. We are anti-wall. We have a technology solution,” - Oh, the amount of self-unconsciousness there.
pony.ai
> In Eldada’s vision, lidar sensors could be deployed along the border, monitoring remote regions for movement, then determining whether there were people attempting unauthorized crossings [...] Eldada acknowledged that a contract for a virtual wall would be controversial, but he described it as a kind of protest against Trump’s version of border security. “I hate the word ‘wall’. We are anti-wall. We have a technology solution,” he said.
This physically nauseates me. Where are our ethics? Just because we can engineer something does not mean we should.
It makes me glad that this company seems to be nothing but vaporware.
This physically nauseates me. Where are our ethics? Just because we can engineer something does not mean we should.
It makes me glad that this company seems to be nothing but vaporware.
i dont think the arguments about a border wall were ever about morality, but more about it being a huge waste of money
Would the border wall be a huge waste of money?
It’s a weird game where the stricter you enforce the less people try to come in the first place. So it’s all about signaling.
And if a wall is to be built, should it not be a virtual wall to reduce cost and improve the likelihood that one day it can be merely switched off?
Wouldn’t broad sensors of this type have massive secondary applications?
It’s a weird game where the stricter you enforce the less people try to come in the first place. So it’s all about signaling.
And if a wall is to be built, should it not be a virtual wall to reduce cost and improve the likelihood that one day it can be merely switched off?
Wouldn’t broad sensors of this type have massive secondary applications?
This is by far the best solution.
Plus by charging a small border crossing fee, Mexico could pay for the wall within a few years.
No massive environmental damage, no massive cost, and we could even help slow down the flow of guns back into Mexico.
Plus with a reliable and profitable entry program we could set up a WAY more reliable temporary work visas that would allow people to live and work freely without breaking US laws, stealing identities, or placing anyone at risk on either side!
Plus, police, tax officials and family services will get to know who everyone is and where they are working!
Everybody is a winner! No immigrant communities in fear. Prosperity for all. And did I mention Mexico pays for the wall?
Plus by charging a small border crossing fee, Mexico could pay for the wall within a few years.
No massive environmental damage, no massive cost, and we could even help slow down the flow of guns back into Mexico.
Plus with a reliable and profitable entry program we could set up a WAY more reliable temporary work visas that would allow people to live and work freely without breaking US laws, stealing identities, or placing anyone at risk on either side!
Plus, police, tax officials and family services will get to know who everyone is and where they are working!
Everybody is a winner! No immigrant communities in fear. Prosperity for all. And did I mention Mexico pays for the wall?
There already is a legal entry program.
I expect that Continental, the German auto parts company, will be the winner here. They bought Advanced Scientific Concepts, which makes good but expensive flash LIDAR units for military and space applications. Now they have to make them cheaper and in volume, which is what auto parts companies do. They already have some shorter-range, lower resolution LIDAR products on the market for anti-collision systems, so they have experience with the automotive LIDAR environment.
Delco, the US auto parts maker, is also working on self-driving components, including LIDAR. They're not saying much, but they have a California autonomous vehicle testing license and some cars on the road. Bosch also has a self driving car in test. It's entirely possible that self-driving will just be an auto part, sourced from the usual big auto parts makers.