Lets flush out this concept a bit:

1. I'm guessing this tech isn't as easy as people think. The sensor array would have to be constantly tuned and calibrated. What happens if you are in this cab and the front sensor goes out? Does the car stop? Do you keep driving at risk? I have tire pressure sensors in my car (lower tech) and they are very finicky.

2. These cars wont be going at very high speeds. Very high speeds need speed rated parts, tires, brakes, etc., high speed is less fuel efficient and will require speed rated roads like the Autoban. All this adds extra costs.

3. Will this vehicle make practical safe decisions? When making a sudden maneuver, will the car know that it is better to slam into a light pole on one side rather than the bicyclist on the other?

4. I've seen 10 year olds zoom around go cart tracks but we don't let them drive in the real world. Will this car be a better driver than a 10 yr old?

Lots of fun tech works in the lab but never sees the real world.

No, there will be 20+ linked sensors installed at trivial cost and they'll be highly redundant. Zero calibration required. Even now, Google cars drive just fine with a just a single revolving laser.

With zero modifications they'll be traveling faster and denser than today due to:

a) far faster reaction times
b) total situational awareness and object tracking that far surpasses what humans can do.
c) networked danger reporting and avoidance.

People see a problem here but this will be trivially easy to solve. To the extent that there could be legal issues, governments will get involved due to the massive public health and safety advantages of removing human drivers from the road.

90-95% of accidents involving death or injury are caused by preventable human driver error (looking at cellphone, blind spot, drinking, etc). An even greater number can be prevented by other things that computers can do. Humans are horrible, horrible drivers with terrible situational awareness. Computers right now are many times safer as drivers.

Humans are like the 10 year olds compared to computers that exist right now.

Driving is a trivially easy problem of following the path and avoiding obstacles. It's easier to solve than computer chess. Computers add several further layers of safety and reliability on top of that which will far surpass humans. I wouldn't be surprised if computer driving becomes mandated at some point - letting humans drive cars is an incredibly dangerous activity.

1. You answered all my questions with laboratory theory not facts and no costs.

2. Chess is played in a controlled environment. Try driving in a city environment like NYC way more variables and obstacles.

3. If this tech was even close to being viable it would be used by the military. Why risk human drivers in a war zone if a supply truck can drive itself?

Google has done over 1 million self-driven miles on California roads. There's plenty of data there. There are other programs around the world including Europe. The data shows that computer driven cars leave humans in the dust.

I'm not sure you've thought through what driving involves, and why accidents happen. Most of the reasons humans crash are caused by humans being utter morons with terrible situational awareness and low reliability. We are extremely poorly designed for car driving. Problems include:

1. Lack of situational awareness - not seeing a kid or another car, for example. Computers solve this completely by having perfect object awareness all around the car, far exceeding humans. Lack of situational awareness includes falling asleep, looking down at a phone, etc

2. Losing control of the car - driving too fast, drifting across lanes, taking corners too fast.

3. Slow or panicked reactions. Time to react to a visual stimuli and begin taking evasive action is about 1.5 seconds, which is shockingly slow, and up to about 50 miles an hour is the greatest component of stopping time. Computers have this down to milliseconds.

4. Unsafe practices - following too closely for human reaction times, driving too fast for the situation, etc.

Humans are extremely dangerous drivers...."the most dangerous part of a car is the nut behind the wheel" is very true.

Computers are already far safer drivers. These are real world facts, borne out by actual autonomous cars driving in actual real-world traffic, not laboratory theories.

A million people a year are killed on the roads, many young and healthy. About 20 million get injuries requiring hospitalization. Nearly all of these accidents due to preventable human error. Estimates are:

Replacing humans as drivers will wipe out a major cause of death, disability, grief and loss, particularly among the young in the West where other causes have been taken care of. For those reasons alone, governments will embrace the technology.
It certainly wouldn't. We can already easily and cheaply remote control supply trucks and we don't do that. Why not?

One thing to note is how different we as human drivers perceive the environment when we are sitting in the car seat and how the computer perceives it. Here is a video of a crash report of a Google car being rear-ended:



As can be seen the computer has a top view with full knowledge about where the edges of the car is located and where other cars are located. The human will have a pretty decent feeling for this but much more uncertainty and increasing the uncertainty in one area decreases it in other. And here it is also clear that some level of automation, such as adaptive cruise control with active braking or a fully self driving car likely would have prevented this accident.

I've read the stories about the Google car and Google is just putting out information not actually answering specific questions about cost, safety, regs, etc. Speed has been limited to 25 mph in all test states. The states that have allowed testing basically want the factories located in their states. There was a mention that in order for this tech to work better street lines have to be better maintained.

We don't need a Google car to improve road safety. Governments could require all cars or truck to be speed limited to 75mph - easy tech/cheap - why isn't that done? Governments could require all cars be equipped with breathalyzer tech - easy/cheap -why isn't that mandated?

Even if the Google car works at a reasonable price - nobody will want it. I would love to know how much the cruise control feature is actually used by drivers and that's been around for years.

You are completely lost and being a nuisance.

They've driven millions of miles and have had 0 accidents except being hit by other human drivers.

Stop already.

See the Apple link I posted a few pages back?

Looks like we know what their secret car project is all about now. Competing with or supplying Uber rather than an iCar for consumers to purchase is what it seems like.

You know Ford is only 60 billion (80 I guess if you wanted to buy it), and nicely profitable. Once you solve the self driving thing, it makes a lot of sense to own a car maker that guarantees a working installed base for your operating system, which provides directions, entertainment, connectivity, etc. Cars are going to be huge once the driver is removed - they'll be the equivalent of the home entertainment setup/office/computing where people will spend 1+ hours a day and require many services. Something like that has to be their end game. I can't imagine them just offering the software. And it's nearly impossible to set up a car maker at a large scale even with a decade+ in ramping. TSLA is a great example of how you run into major supply chain problems when ramping, even ten years down the road with only 50K cars and even without all the complex components of an internal combustion engine.

You saying AAPL could buy Ford or GM? I guess it is pocket change to them but seems like there would be a lot of issues with that in terms of integration.

Why not just buy TSLA for like 45ish.

Well what the hell is the end game for Apple's research? Think it through:

- Building their own cars is simply not viable this late to the game. Reliable supply chains for the parts to put a car together take 10+ years to put in place, and make building an iPhone look like playing with barbies. Musk has talked about the difficulties.

- Selling or giving away self driving software just seems weird given how integrated it is with the car's systems, and that everyone else is going to get there at the same time with their own software. No one is going to be blindsided by Apple beating them to the first self driving software to be completed. Apple don't have the competency or software ability to beat anyone.

- They'd be mad to buy Tesla when they can get Ford for not much more.

So what the hell are they playing at? They're either going to buy an established carmaker, or they're hedging against Google providing a car operating system and bleeding into everything else, like Google did with Android. Or they see a future with an uber-like fleet management business - transport as a service is the new buzzword, which is really just taxis. This will require very close integration with a car maker. What's the fourth option?

I have to think the most probable is that it's some kind of hedge against Google. Cars running Google operating system - interfacing with your phone that is basically a PC in 2022 that works seamlessly across in-car screens - don't have to interface with iPhones unless Apple has leverage back.

Interesting, what do you think Googles endgame is? Same goal?

They can't ramp up their supply chain either.

Google buys Ford and AAPL buys GM?

I didn't bother with this before since you're obviously an idiot, but since you have such a hard-on about this, even trolling other threads a month later, let me point something out to you. My contention as posted above is that:

a) autonomous driving is about the edge cases, not what we can solve right now
b) the edge cases will require higher resolution cameras, and arrays of cameras. The hardware we have now is not sufficient.

You ridiculed this view repeatedly. You even said this:
So here are what some others in the field think and say as 2015:

Nissan's research division general manager:
NVidia's CEO:

MobileEye's CEO:

I could quote more and more about the need for a) far more cameras and b) higher resolution cameras to solve the edge cases and c)higher bandwidth, which was exactly my contention to begin with.

I mean, your last statement has been proven 100% false. You've just proven yourself to be a total dickhead. Yet when challenged on anything you just slink away into insults.

It's not uncommon for guys in CS, particularly with little real-world experience, to see everything in terms of algorithmic solutions and abstracts (especially when there's a lot of work to do in algorithms such that the sensor data is maximally used), but you're dead wrong. Higher resolutions, higher bandwidth for image processing, and far more cameras will indeed be needed before this space is solved, as Nissan's research chief is telling you. I'd say we'll be there in terms of hardware by 2020.

Even though I think you are a distasteful personality. I did promise to help explain this to you.

Some general comments: I like how you take links out of context in the lay press instead of primary research by innovators. I guess that is how you "inform" yourself.

1) "Doi suggests that highway autonomous functions could be run on current architectures, but more data-intensive functions, such as autonomous urban driving, would require a next-generation architecture before they could be supported."

Welp. Looks like like we don't need higher speed cameras in that quote. But may need "next-generation architecture" for city driving. Whatever that means.
2) They aren't even using the forefront of modern digitial cameras in the device he is talking about. They are using an amount of pixels just over 5k..... except with 12 cameras.

Guess we don't need to wait for a 10x improvement in camera technology when they're over an order of magnitude behind current technology.
3) I know you suck at English, but the writer is juxtaposing higher resolution cameras against cameras with lower resolution and wider fields of view. I'm sorry that you can't read real good.

Anything else you would like me to clarify? (I can see you've started goalpost shifting. Maybe once you move all the way to what I said initially, you'll be correct!)

I'm treating it like this:

When the car was invented, over 2K car companies existed. Today only a couple do, and only a handful would of been good/great investments. And picking which one was a good/great investment, was HARD!

But what was easy to determine, was the horse/buggy was going out, and all the companies dealing with them were in trouble.

The same goes for autonomous vehicles. While I have NO idea which one/s will win (google? tesla? others?) I do know who will lose (insurance companies is one of them)...the problem is everyone and their dog knows the example I've said, and it's no secret that the insurance companies will lose (even Buffett said it about GEICO publicly) so the price is baked into the market already....I might as well just buy VTI or VT and go home.

How I personally plan to benefit from this, is to enjoy the increased life expectancy when they do come around But then again, everyone will be getting that benefit, so not really an "edge"

[quote] Avis up 13% today on this news:
Why are people so bad at thinking? It was always obvious that Avis was going to crush Uber in the autonomous car space.

[QUOTE=ToothSayer;52441135] Don't hold your breath on that one. Decades away.

Why would on-airport location of cars be valuable in a world of autonomous vehicles? There's no reason an autonomous car couldn't be waiting for a passenger at passenger pickup by monitoring flight arrivals, even if it has to drive from an off-airport lot.

Nothing stops the current rental car majors from becoming providers of autonomous fleets (clearly they have valuable experience and structure), but I can't see how their current advantage of airport locations protects them in an autonomous world.


I'll also ask...why in the world is it obvious that Avis will crush Uber? This seems like a total crap shoot at this point to who will dominate fully automated fleets.

Very surprised this thread hasn't been bumped in a year and a half...

Just to create some (hopefully not low/zero content discussion, does anyone like any of the companies exposed in this space?

Personally I still like GOOGL given the data Waymo has collected and the valuation of the parent company. Analysts think it could be worth anywhere from $100b-$250b long term and that's clearly nowhere near priced into the stock. Still a lot of uncertainty about all the different potential business models and how fast level 5 regulations will come. I'd bet that actually won't be much of an issue when they can prove how much safer it is when people don't drive, it's hammering down the exact details and infrastructure that's going to hold it back.

Who is the biggest threat to Waymo or is someone already ahead of them? Also which chip stocks, auto stocks etc are best positioned for this inevitable transition?

It's interesting that this discussion was going on over 3 years ago. People are still talking about AAPL buying TSLA today. Now the prices are different though. TSLA is worth $59 billion while Ford is worth just over half that at $32 billion.

I really can't wrap my head around AAPL buying TSLA. A big part of the reason TSLA's stock price is so high is Elon Musk. Would he really stick around? Also, there is value in the Tesla name, which AAPL doesn't need and wouldn't use. If AAPL is already working on autonomous then that is just overlap. Buying TSLA would remove a direct competitor though, plus include an EV station network, and better EV technology and a product that closer fits what AAPL would want to produce.

Personally, I find the Ford proposition more interesting. It's obviously a much better deal from a valuation standpoint. That being said, there's a lot of value in Ford that AAPL doesn't need, like the truck business (#1 selling vehicle in the US since before we were born) and the Ford brand. I don't even think they need a dealership network. What I'm wondering is if they could do a deal with Berkshire to acquire Ford and spin off most of the Ford business to Berkshire, in particular the truck business which is a cash cow that isn't a fit for AAPL. The Lincoln brand would get axed as would most of the car business which Ford is trying to get out of anyways.

you can argue TSLA is the much better buy because of the supercharger network already nationwide and because the only real reason TSLA is in such a ****ed up position is because of Elon Musk. if there is a way to both keep the value of Musk's presence in the company AND have rational supervision and control over the decisionmaking (basically someone who would know how to run an actual car company and be imaginative enough to keep Musk engaged + ego fed maybe? i dunno), then TSLA can actually grow into their valuation

think about all the things that suck regarding Tesla. AAPL can clean all of that up and TSLA's biggest issue is money. why not buy TSLA?