Electric Car Prediction - Whatdayathing???

[SemiSalt]

I was wasting an hour in our local library. I do this while the cleaning ladies are at the house. I picked up an issue of Car & Driver which had a pretty long feature on self-driving cars edited by Malcolm Gladwell. (This could be a plus or a minus. There are a lot of Gladwell haters out there.) Gladwell's intro piece is here: https://www.caranddriver.com/features/what-happens-when-we-give-up-control-of-our-cars-feature. There were several different essays on various related issues.

On balance, they were more pessimistic than the average among print journalists about both technical and legal/regulatory issues.

This seems to be the main page:
https://www.caranddriver.com/features/autonomous-addressing-the-totality-of-the-driverless-car-feature

[BeauV]

While I'm sure that Tesla is having a lot of trouble, I also happen to know that they've increased Model 3 production massively in the last two months. A surprise to me was that almost no one wants the base model. It appears that most folks on the "list" would like a version that ends up costing about $45k. BTW, this is typical of the Model S also, where less than 15% buy the "base" model at about $80K. I believe the average selling price of a Model S is hovering around $125k.

If it turns out that the customers like he cars, which everyone I've ever talked to do.

If it turns out that they all migrate up into the higher priced models, which early indications are that they are.

Then, I think they've got a chance. We may be seeing the classic "Valley of Despair" which every new adopter product goes through. Could be a buying opportunity coming up with the stock. (I am a TERRIBLE stock picker.)

[TheOffice]

Beau

I’m not surprised no one wants the base model. No leather sunroof autopilot color selection.
The big battery is 9k. BMW sells cars without leather. Most end up as service loaners.

[kimbottles]

While I'm sure that Tesla is having a lot of trouble, I also happen to know that they've increased Model 3 production massively in the last two months. A surprise to me was that almost no one wants the base model. It appears that most folks on the "list" would like a version that ends up costing about $45k. BTW, this is typical of the Model S also, where less than 15% buy the "base" model at about $80K. I believe the average selling price of a Model S is hovering around $125k.

If it turns out that the customers like he cars, which everyone I've ever talked to do.

If it turns out that they all migrate up into the higher priced models, which early indications are that they are.

Then, I think they've got a chance. We may be seeing the classic "Valley of Despair" which every new adopter product goes through. Could be a buying opportunity coming up with the stock. (I am a TERRIBLE stock picker.)

Only Keith should try stock picking, the rest of us should be happy with good cheap index funds.

[BeauV]

Looks like one professional stock picker thinks this is a buying opportunity, to the tune of $35 million.

HERE

Edit: Sorry about the link. Fixed it.

[SemiSalt]

I had a contract programming gig with Louis Dreyfus Company. They are a commodity trading firm, principally grain from the midwest. (Fun fact: The owner, William LD is the father of the actress Julia LD.) Someone left a primer on commodity trading lying around and I read the first few chapters. The author went to great pains on one particular point: you will lose everything unless you bring outside information to the market. LD conducted an annual, worldwide survey of grain production (i.e. how many acres under cultivation in every significant grain producing region), and monitored rain, drought, storms, political events, etc.

Stock trading is not quite so fierce. Unlike a losing futures contract, a bad stock pick doesn't result in a 100% loss of capital. But I think the same principle applies, if only in a somewhat weaker version.

[TheOffice]

Pretty amazing that CR found a flaw in the Model 3 brakes and within a week a software update resolved it. No other car company can do that.

Saw a Model 3 at the Tesla store in Littleton Co. It was across the street from the high school graduation we attended, so Sue could not resist.

We weren't able to open up the doors, but she looked sleek (if you upgrade the wheels).

There's a new grill on the S. Looks better.

[Orestes Munn]

We weren't able to open up the doors.

Just cycle the power, reboot, and see what happens.

[LarryHoward]

Pretty amazing that CR found a flaw in the Model 3 brakes and within a week a software update resolved it. No other car company can do that.

Saw a Model 3 at the Tesla store in Littleton Co. It was across the street from the high school graduation we attended, so Sue could not resist.

We weren't able to open up the doors, but she looked sleek (if you upgrade the wheels).

There's a new grill on the S. Looks better.

My question is what happened that Tesla needed CR’s help in identifying a brake issue? What else is the public beta testing for Elon?

[LarryHoward]

We weren't able to open up the doors.

Just cycle the power, reboot, and see what happens.

Friend of mine bought one of the first F Type Jags imported to the US. It has some electrical glitch that drained the battery several times. Like the Tesla recharging doo that requires battery power to open it, the Jag required power to open the trunk and access the battery compartment. Took the dealer 3 days to access the battery the first time it happened and it got towed in for the same problem something like 4 times before they fixed it and I think Jag had to invent a procedure to access the battery when there was no battery power.

[Chris Chesley]

We weren't able to open up the doors.

Just cycle the power, reboot, and see what happens.

Friend of mine bought one of the first F Type Jags imported to the US. It has some electrical glitch that drained the battery several times. Like the Tesla recharging doo that requires battery power to open it, the Jag required power to open the trunk and access the battery compartment. Took the dealer 3 days to access the battery the first time it happened and it got towed in for the same problem something like 4 times before they fixed it and I think Jag had to invent a procedure to access the battery when there was no battery power.

Zis iss yust goot Ghermaan engginneerin! Zee betterees yust nefer go flat. Dey air dat goot! Vee safe de vate dis vay.

[LarryHoward]

We weren't able to open up the doors.

Just cycle the power, reboot, and see what happens.

Friend of mine bought one of the first F Type Jags imported to the US. It has some electrical glitch that drained the battery several times. Like the Tesla recharging doo that requires battery power to open it, the Jag required power to open the trunk and access the battery compartment. Took the dealer 3 days to access the battery the first time it happened and it got towed in for the same problem something like 4 times before they fixed it and I think Jag had to invent a procedure to access the battery when there was no battery power.

Zis iss yust goot Ghermaan engginneerin! Zee betterees yust nefer go flat. Dey air dat goot! Vee safe de vate dis vay.

Many but I think Jag hired Lucas to manufacture and install the system. That’s never good.

[Olaf Hart]

A little creativity from our NZ cousins

https://www.news.com.au/lifestyle/real- ... 5d282ca528

[SemiSalt]

I was thinking some time in the last couple week that I'd never identified a Tesla on the road. I'm not the kind of guy that IDs all the cars I see. As mentioned elsewhere, I drove a good bit around New England this week, and this morning I happened to notice that a car a little ways ahead of me had the license plate "TESLA." It wasn't convenient for me to catch up to it for a good look, but I kept an eye on it. It was going about 67 mph to my 66 mph, so it pulled slowly ahead. Then a silver-colored car of unknown make caught up to the TESLA, and stayed in trail about 1 car length behind. The two of them stayed paired up like that until they were nearly out of sight. I think it the end, they caught up to some slow traffic in their lane and the silver car went around.

I would have been quite uncomfortable in either car staying that close for that long. I got to wondering if the Tesla (if it was Tesla) was on auto-pilot, and if this was a foretaste of future motoring: conveys of cars in close formation.

[BeauV]

Semi,

Without a doubt we’re going to see cars of all types running in convoys. It’s one of the key benefits of have semi-autonomous or fully-autonomous cars/trucks. I know that the folks at Tesla are working on running trucks with only 3 or 4 foot gap between them, as wind resistance is a “big deal” for long haul trucks and the massive shortage of drivers is motivating a great deal of research in this area.

A funny thing happens as one is learning to trust the following/autopilot mode (both my Ford SUV and my Teslas have a way to follow the car in front, but I have to steer the Ford). At first, you think that the autopilot is following too far back, and you realize that we humans have been tail-gating far too closely for far too long. This is particularly noticeable in the rain when the Tesla backs way off and most drivers do not.

Second, other drivers notice the “gap” and think you’re a wussy-baby for leaving so much space so they pass on the right and stuff their car into the gap, leaving less than one car length between them and the car that was in front of you. This causes the Tesla and Ford to slow down and open the gap, then it happens all over again. What the humans are doing is learning to take advantage of the fact that the autopilot actually drives in a safe manner and treat any car that is driving like that as a Boston cab driver would.

Finally, there has been a great deal of discussion about dedicated lanes for automated driving cars. These lanes would be exclusively for cars that are self driving and when in one of these lanes the convoy would make space for a car ahead that wanted to merge in, presumably through some sort of radio connected to nearby cars. The convoy would open a gap as it approached the new entrant, and the new entrant would automatically merge in. The gaps would then be closed up and the column would carry on. The various proposals I’ve read have suggested that on current interstate highways with current auto technology cars/trucks operating in this mode could sustain speeds of 80 to 90 MPH where appropriate and only slow due to known construction, road damage, etc....

The difficulty with this last step is that the same Bostonian Cab Driver who will bluff his way in front of a law abiding self-driving car, will also attack a lane like that so he can go fast too. One can imagine such a person feinting that they will merge illegally, the cars braking to avoid a collision, and the bad-actor merging in without the benefit of the active cooperation of the automated cars. Because of this, for this last step to really work, we’ll probably need physical barriers to keep the thugs out of the lane. We have precisely this problem in Car Pool Lanes, Express Lanes, and Toll Roads around where I live. About once a month the police set up a trap and haul dozens of cheaters out of the lane and write them $375 tickets, but is doesn’t seem to deter the offenders very much.

[JoeP]

Interesting problems had to be worked out for this "platooning" (as I have heard it called) concept Beau. Each platoon of x number of cars would be limited by the cornering, acceleration, and braking capabilities of the least competent car and platoons would be limited by the least competent platoon. I understand the self driving networking systems are close to being ready. Interesting stuff.

[BeauV]

Interesting problems had to be worked out for this "platooning" (as I have heard it called) concept Beau. Each platoon of x number of cars would be limited by the cornering, acceleration, and braking capabilities of the least competent car and platoons would be limited by the least competent platoon. I understand the self driving networking systems are close to being ready. Interesting stuff.

Joe, you're absolutely right. The platoons are limited by the lowest performing member. Interestingly, the total through-put of the system, even accepting that they're all moving at the slowest member, is quite a bit higher than a "normal" highway. Imagine removing all the braking, accelerating, lane changing, etc... that goes on. Imagine being able to read or watch TV while in a platoon! Or, for that matter, canoodle with your sweetheart.

What shocked me was that by tripling the bandwidth (total number of vehicles) by packing them close together, the resulting improvement is amazing even if the the platoon only travels at 80% of the peak speed the folks achieve now days.

[JoeP]

Interesting problems had to be worked out for this "platooning" (as I have heard it called) concept Beau. Each platoon of x number of cars would be limited by the cornering, acceleration, and braking capabilities of the least competent car and platoons would be limited by the least competent platoon. I understand the self driving networking systems are close to being ready. Interesting stuff.

Joe, you're absolutely right. The platoons are limited by the lowest performing member. Interestingly, the total through-put of the system, even accepting that they're all moving at the slowest member, is quite a bit higher than a "normal" highway. Imagine removing all the braking, accelerating, lane changing, etc... that goes on. Imagine being able to read or watch TV while in a platoon! Or, for that matter, canoodle with your sweetheart.

What shocked me was that by tripling the bandwidth (total number of vehicles) by packing them close together, the resulting improvement is amazing even if the the platoon only travels at 80% of the peak speed the folks achieve now days.

I can imagine the increased efficiency of the interstates in congested urban areas if all vehicles were capable of networked operation and could travel at higher speeds close together. While I love to drive my commute has gotten pretty stale after 32 years and it would be nice to be able to relax a bit, but that won't happen before I retire.

[BeauV]

Joe, you raise a really interesting question: When will all this happen?

As someone who worked deeply in AI 25 years ago, I had become discouraged until about 5 years ago when all of a sudden a bunch of kids did things I didn't think were possible. What that did was confirm my standing premise:

I always know WHAT is going to happen, I just don't know WHEN.

What I'm struggling with in my later years is trying to get better at the "WHEN" vs the "WHAT". Clearly platooned vehicles will happen because the massive benefits are obvious. But WHEN it will happen - to quote the King And I - "Is a puzzlement". It could happen while we're both alive, but then again, it might not.

[LarryHoward]

Beau,

I think the when is still a ways off. Afterall, in 1962, the Jetsons showed me a future with a self piloted flying car. Still don’t see it fronting up each morning.

I see several steps, all hard before self directing convoys are the norm.
- Autonomous driving has to be accepted and generally considered to be a 6 sigma process. Maybe even that's not enough but we are not there. It can't just be better than the average clown behind the wheel.

- Networking software has to be developed, work seamlessly cross domain. Alphabet and Apple, in particular seem wedded to proprietary hardware and software at the moment. Elon offers to share his but I don't see anybody taking him up on it ATM.

- Those dedicated roads need to be build. Right now, we don't see regular to HOV conversions in much of the East Coast based on "you can build new HOV lanes but don't dare take away the "free road" I paid for”. We are also seeing an increasing push to move away from gas taxes paying for roads, given CAFE increases and the rise of hybrids and electrics. There is no longer a strong connection between fuel use and road use. Infrastructure funding isn't easy. Just because it has obvious advantages doesn't drive public funding or all of our major cities would have superb mass transit systems and freight rail would push long haul trucks off the roads outside of last mile logistics.

There may be a few "demo roads" built somewhere but integrating the "dedicated roads" into already massively underfunded infrastructure will be a very hard sell. How high a toll will folks pay to access that road?

[TheOffice]

I agree it is not close.
Look at something as simple as EZ-Pass adoption. Its cheaper, faster, easier, yet thousands of people line up to pay cash at tolls unless, like the Golden Gate, you ban cash. Maryland gives you the EZPass now, and a discount for using it. Yet there is only one high speed toll lane at the Bay Bridge and probably 10 that are not dedicated.

You'l need fleets of trucks that are talking to each other and a campaign to educate the public that having a dedicated lane eases congestion. And a new way to pay for all of this. Maybe a kilowatt tax??

[kdh]

I always know WHAT is going to happen, I just don't know WHEN.

What I'm struggling with in my later years is trying to get better at the "WHEN" vs the "WHAT". Clearly platooned vehicles will happen because the massive benefits are obvious. But WHEN it will happen - to quote the King And I - "Is a puzzlement". It could happen while we're both alive, but then again, it might not.

We seem to be focused on something else entirely, autonomous vehicles in the wild, which as far as I can tell won't happen for a long time.

[BeauV]

Like I said, I (and everyone else I know) is pretty bad at predicting the "when". So, I agree I'm probably wrong.

That said, the folks who say: "It'll be decades." have been just as wrong as the ones who say: "It'll be next year" about many many technologies.

Larry, I hear you on the modifications to the infrastructure. I'd like to point out, however, that the ability to drive a car by looking out the front with a camera and LIDAR was considered "impossible" less than 10 years ago. Back then there were actual test highways built with wires in the cement to let the car travel in its lane. Within an 18 month period folks figured out how to lane follow and BANG there wasn't any need for the wires in the road. The folks suggesting wires in the road weren't idiots, they just didn't realize how fast things were going to change when a car has a small supercomputer built into it.

Keith, the difference between long-haul trucking and personal cars is the source of the cash to buy the trucks. The individual trucker is going to have a hard time upgrading because this will be expensive. It'll be a LOT cheaper than having drivers etc.... but that doesn't do the individual trucker any good. As a result, I'm guessing that the majority of trucks will be owned by fleet operators which will raise money from guys like you. They'll offer a reasonable IRR and re-capitalize the industry into a low-labor model. This has happened everywhere that automation has allowed for the elimination of labor, especially where there are literally tens of thousands of open jobs for which owners can't find drivers. Truck driving sucks as a job, as a result there aren't enough drivers. With the boarders being closed to out cheapest and best source of new drivers, Mexicans, we're going to see even more pressure. In CA we already are.

Larry, I agree that building a dedicated lane is a problem in congested places, but that's not really what's needed. We need the dedicated lane along the 3,500 miles across the country and along the interstates. Those are neither congested or defended by "Get Off My Freeway" whack jobs who think they paid for the interstate highway system. Humans will be doing the last-mile stuff for a long time.

As to rail vs trucks, Rail transport of containers out of the Port of LA is growing at about 4 times the rate of truck transport. They're building new lines to get out of So. Cal. the new rail lines will run alongside the 5 rail lines they already have. Things are changing, it's just that they're not changing where we can see them.

[SemiSalt]

I'd like to point out, however, that the ability to drive a car by looking out the front with a camera and LIDAR was considered "impossible" less than 10 years ago.

One break-through was understanding that it was feasible to collect data for every inch of road and not have to depend so much on cameras and AI to "understand" what it was seeing in real time. I think of it in statistical terms as using the population instead of just a sample. One of the first cases of that approach that I know of was the address list in the Post Office. When they scan your envelope to check the address, they aren't just looking at format; they are looking a database of all active addresses in the US.

I can see prototyping on a highway across (e.g.) Nebraska happening pretty soon. A lot sooner than on the NJ Turnpike.

[BeauV]

Semi, you raise a terrific point. For many many years "Moore's Law" was batted about as the paragon of rapid growth/improvement. During that same time the disk drive industry out performed Moore's Law by 400%. Very very few "experts" really understood what that meant. Those same experts didn't realize that putting video on your phone would require both storage and bandwidth that was many decimal orders of magnitude more that any enterprise applications.

I distinctly remember a chat with senior AT&T folks 20 years ago when I was explaining that once "voice" was digitized, it was just data. Basically that phone calls would rapidly become a tiny percentage of what folks loaded onto their network. They thought I was crazy. Now "unlimited voice and txt" are pretty much normal because they have been made irrelevant by the demand for real-time video.

Very very soon, every car will either just carry the latest database of all the roads (including a lot of detail) or they'll have access to it via the network. This won't just be the local roads, it'll be every road.

I just rode in a prototype car that literally knew when there was a pot hole. It moved over in the lane because earlier cars had reported the pot hole, not because some camera captured it in realtime.

[kdh]

Just don't try to cross the road unless you're in the database.

[LarryHoward]

I think the testing for “flyover country” May be done in small convoys or perhaps 6-8 vehicles as we still are not going to provide dedicated lanes. (Take away 1 of 2) and I’ve not seen more than about 300 miles without repairs/construction dropping that to one in a long time. Last time I shipped a boat, it took 3 days to get a permit across Pennsylvania due to uncoordinated construction projects.

Add in less than 3g data in a large percentage of the country (the less dense portions) and its a bit premature to plan for a coast to coast immersive data infrastructure with all roads updated real time. Not that long ago, it was easy to fly off the map in rural areas.

As to learning, it’ll have to be better than WAZE or google or we will send convoys into forest fires. I’m a little reluctant to trust unfiltered crowd sourcing of routing for unoccupied 80,000 trucks in convoy.

[SemiSalt]

Very very soon, every car will either just carry the latest database of all the roads (including a lot of detail) or they'll have access to it via the network. This won't just be the local roads, it'll be every road.

Almost true already. When I checked Google Maps before my recent trip to Maine, it asked to download pieces of map in areas with poor cell phone coverage where we might otherwise experience a loss of service.

[BeauV]

Larry, a few things. First, Tesla (for example) already gathers data on road conditions etc.... Yes, folks can change things by putting in new construction etc... by there are literally thousands of cars actively gathering this data passively and sending it to a central database. In the near future, that will probably be most cars, I believe Ford does it too now. (Yes, it's stripped of identifying features for the privacy of the car owner. ) So, the data isn't really "crowd sourced", which gives the impression that humans are entering it, it's "car sourced" which is probably vastly more reliable and happens without the driver either knowing it or doing anything about it.

I'm sure you realize that there is a massive concentration of truck traffic on just a few highways. Yes, there are short hauls from end points, and there are a few trucks headed up Cottonwood Canyon to deliver stuff to Alta and Snowbird. But that's why literally no one with any experience in the problem is saying that trucks will be fully autonomous. (note the qualifier) A driver will do the final delivery. What will be knocked off first will be the major routes like I-80, I-95, I-5 etc.... From data I've read, that will cover well over 80% of the truck miles driven.

Eric, it is really sad that an Uber prototype hit that woman walking her bike. It is far sadder that literally hundreds of people are killed by sleepy truck drivers and drunks. We've become immune to the latter, assuming it's just a social cost to get our stuff delivered from Amazon and Walmart, we are still startled by the death of someone at the hands of an unproven prototype. Perhaps the thugs at Uber should stop trying to test prototypes on the road. There was a driver who was supposed to be watching what was going on, so that event is precisely the same as the idiots hitting an island off of San Diego when they turned on the boat's autopilot and fell asleep.

Like I said before, all we rational thinkers have vastly underestimated the rate of improvement in autonomous systems. How many would have guessed that Waze would exist back in 1990? I almost always know "what" I just don't know "when".

[BeauV]

I just read this article, which uses a computer to watch the driver and nudge them when/if they start to pick up their phone and read it. Something that is illegal in every place I know.

It'll be interesting to see if anyone actually allows that into their car.