The Tesla Thread

Tesla made some interesting comments this week about Autopilot, self driving vehicles and the sensors v LIDAR debate. According to Ars Tecnica:

"
A "significant" upgrade to Autopilot, Tesla's advanced driver assistance system (ADAS) suite is due next week. Musk claimed that improvements to the system would increase exponentially in the next 12-18 months. As for "full self-driving," Musk revealed he "was just testing that last night at 1am." He then went on to say that Tesla might be able to release something in a couple of months. According to Musk, there have been two different approaches to developing autonomous driving; one really complicated path "that isn't working great" and a simple one that is. He added that he was able to go from onramp to offramp with the simple system.

Another question from the audience asked when a "showdown" between lidar and optical sensors for autonomous driving might happen. Again, Musk described lidar as a crutch, saying that it "causes companies to go to a local maximum for autonomous capability." He then went off on a tangent about lidar sensors—apparently everyone else is doing it wrong and should be using a 4mm wavelength.

Musk also said that the company might start offering free trials of Autopilot to Tesla owners beginning next month."

Interesting comments about LIDAR and sensors, for sure. Not sure I learned much from them however!
 
the shares bounced following yesterdsays shareholders meeting - could not find an audio of live stream on utube ?
but acknowledged insurance cost
Other big news includes the announcement that starting this month, Tesla stores would offer test drives of Model 3, with the goal of having all Tesla stores offering them by the end of next month. He also mentioned that Tesla was working with insurance companies to try and lower the cost of Model 3 insurance such that it would undercut BMW's 3-series by as much as 30 percent.
...
The most interesting aspect of the Autopilot conversation, though, was Musk saying that Tesla plans to offer a free trial of autopilot to those Tesla owners who have yet to pony up the dough and opt-in fully. Will this see a big spike in adoption or a big spike in Autopilot-related crashes because people are lazy?
great what you can do with online updates - do they disclose detail on autopilot enhancements ? ie improvements related to lessons learned (it's like windows critical updates, how impregnable are they against hacking )
 
I think with autopilot a couple of things are needed:

- Name change, as others have said. I think they shot themselves in the foot with this one. Its led people (americans) thinking that they can text and do whatever they want. Problem is they have a couple of hundred thousand cars and owners there now so its difficult to re-educate them.

- Some form of mandatory training to be able to use it. Yes whilst its great I could go get in an Autopilot equipped Model S right now and enable it and off we go, I am confident it would be ok. I think someone who is less tech/geek oriented were to do so (for example, the wife) they would take it as others do purely by the name alone and don't understand its designed to follow white lines in the road and other things like this.

In Scuba diving, before they let you dive with gas other than normal 21% you are required to take various training to manage your nitrogen levels properly, not doing so would result in waaaay more deaths than what we see now. PADI do this by way of a small book and 45min video. I think some form of training given to owners when they collect their first Tesla would be a good idea. Wouldn't be difficult to implement.
 
the shares bounced following yesterdsays shareholders meeting - could not find an audio of live stream on utube ?
but acknowledged insurance cost

great what you can do with online updates - do they disclose detail on autopilot enhancements ? ie improvements related to lessons learned (it's like windows critical updates, how impregnable are they against hacking )

It is impressive however that Autopilot can be adjusted with a software update with such significant consequences. I assume Waymo will operate the same way. After all, we will soon come to see a car as a mobile phone on wheels!

By the way, did you notice today that Waymo announced that they plan to bring their self driving vehicles to Europe. At the recent Fiat Chrysler investor day in Italy, Waymo demoed their Level 4 Fiat Chrysler SUV. Until now, Waymo has announced plans only for the US. I hope that "Europe" means the UK!

https://www.reuters.com/article/us-...ing-robo-taxi-service-to-europe-idUSKCN1J321T

:):)
 
Last edited:
I think with autopilot a couple of things are needed:

- Name change, as others have said. I think they shot themselves in the foot with this one. Its led people (americans) thinking that they can text and do whatever they want. Problem is they have a couple of hundred thousand cars and owners there now so its difficult to re-educate them.

- Some form of mandatory training to be able to use it. Yes whilst its great I could go get in an Autopilot equipped Model S right now and enable it and off we go, I am confident it would be ok. I think someone who is less tech/geek oriented were to do so (for example, the wife) they would take it as others do purely by the name alone and don't understand its designed to follow white lines in the road and other things like this.

In Scuba diving, before they let you dive with gas other than normal 21% you are required to take various training to manage your nitrogen levels properly, not doing so would result in waaaay more deaths than what we see now. PADI do this by way of a small book and 45min video. I think some form of training given to owners when they collect their first Tesla would be a good idea. Wouldn't be difficult to implement.

Using a sledge-hammer to crack a nut I believe. Who needs to be trained when the world and his wife will sooner or later (hopefully sooner) be able to use self driving vehicles that will also be either EVs or hybrid EVs of very good quality. Please see my post above about Waymo coming to Europe.

And this link: http://europe.autonews.com/article/...pand-self-driving-services-to-europe-ceo-says
 
Last edited:
The Pacifica is a people carrier/minivan, not an SUV. Personally I'm more interested in the announcement that FCA are in talks with Waymo for a retail vehicle, although I guess that'll be a few years away. I don't forsee myself taking taxis much more than I do now.

De9_Wxlh_Uw_AAj3_Px.jpg


Interesting sales figures. I wonder if BMW's sales are related or not, but the Model 3 is certainly pitched at this market, if so the German manufacturers are going to have a tough couple of years ahead of them as only BMW have a direct competitor in the works and that's at least a couple of years away. The mainstream manufacturers are insulated by the high cost until at least next year, by which time most will have at least one longer range EV on the market in more profitable/desirable segments.

It will be interesting to see how the Leaf holds up, especially if the 2019 has a 60kWh battery for $35k coming out at around the same time as Musk has indicated the $35k 3 may be available. Currently the Leaf is selling extremely well (their best selling car I believe) and they have preorders of 37,000 in Europe alone and are delivering a few thousand a month.

It will also be interesting to see how the sales figures change when Teslas backlog disappears. Once the reservations are fulfilled (presumably some time next year), will their market share stabilise or drop?
 
The Pacifica is a people carrier/minivan, not an SUV. Personally I'm more interested in the announcement that FCA are in talks with Waymo for a retail vehicle, although I guess that'll be a few years away. I don't forsee myself taking taxis much more than I do now....

Perhaps you might want to wait for the Kitty Hawk, the flying car? You can put your name down on the buyer list already.

https://www.yahoo.com/news/flying-car-startup-backed-google-founder-offers-test-210639148.html
 
That makes more sense.

And while that may be partly true, it’s a £40-60k saloon, so there aren’t going to be many that wouldn’t normally consider a 3/5 series or equivalent even though it’s not a direct competitor.
 
De9_Wxlh_Uw_AAj3_Px.jpg


Interesting sales figures. I wonder if BMW's sales are related or not, but the Model 3 is certainly pitched at this market, if so the German manufacturers are going to have a tough couple of years ahead of them as only BMW have a direct competitor in the works and that's at least a couple of years away. The mainstream manufacturers are insulated by the high cost until at least next year, by which time most will have at least one longer range EV on the market in more profitable/desirable segments.

http://uk.businessinsider.com/tesla-model-3-market-share-claims-are-flawed-2018-6

http://capitolzero.com/2018/06/06/t...-competitors-like-mercedes-bmw-and-audi-tsla/
 
This makes for shocking reading

https://www.theregister.co.uk/2018/06/07/tesla_crash_report/

"Four seconds before impact, however, the Autopilot in the Tesla was no longer following the lead vehicle and was in the triangular "gore area" that separates the two lanes – one going to the 85, one continuing on the 101.

Then, three seconds before impact and while still in Autopilot mode, the car accelerated – from 62 to 70.8 MPH – before slamming into a partially damaged crash barrier, ultimately killing Huang and causing the front end of his car to be torn off. The car did not recognize any barriers nor carry out any emergency maneuvers or braking.

The Tesla then collided with another two vehicles – a Mazda and an Audi – and the ensuing wreckage brought traffic to a halt. The highway wasn't opened for another six hours while the authorities put out a fire from the Tesla's battery, cleared debris, and towed the car away.

It's also worth noting that the towed Tesla's battery fire required 200 gallons of water and foam to put out. And then five days later – that's five days later – it reignited and had to be put out again."
 
This makes for shocking reading

https://www.theregister.co.uk/2018/06/07/tesla_crash_report/

"Four seconds before impact, however, the Autopilot in the Tesla was no longer following the lead vehicle and was in the triangular "gore area" that separates the two lanes – one going to the 85, one continuing on the 101.

Then, three seconds before impact and while still in Autopilot mode, the car accelerated – from 62 to 70.8 MPH – before slamming into a partially damaged crash barrier, ultimately killing Huang and causing the front end of his car to be torn off. The car did not recognize any barriers nor carry out any emergency maneuvers or braking.

The Tesla then collided with another two vehicles – a Mazda and an Audi – and the ensuing wreckage brought traffic to a halt. The highway wasn't opened for another six hours while the authorities put out a fire from the Tesla's battery, cleared debris, and towed the car away.

It's also worth noting that the towed Tesla's battery fire required 200 gallons of water and foam to put out. And then five days later – that's five days later – it reignited and had to be put out again."

What? Shocking that the driver repeatedly wasn’t holding the steering wheel despite the car warning him?
 
It isn't the best story for them, but looks like a lot of user error as well.
The trend with these accidents does seem to be:

- Tesla with autopilot crashes
- We find out Autopilot is enabled
- Internet & News goes nuts, autopilot is dangerous, its killing people blah blah blah
- We find out the user wasn't paying attention/was looking at their phone or otherwise
- The story goes away

Now personally I (clearly) am a Tesla fan however I feel I am also pretty rational in these scenarios. And after some digging it does seem to come down to the above

- You You Xue crashing his model 3, was looking at his phone
- This Model X, guy didn't have his hands on the steering wheel, was alerted multiple times before the crash
- The Model S which hit the firetruck, woman was on the phone (if I remember correctly)

Autopilot is far from perfect, it is pretty clear you cannot just leave it to do its own thing. Again, the name is misleading.

Part of me thinks the internal cameras should be recording for insurance purposes (ie if you're on your phone, then you're at fault etc)
 
The trend with these accidents does seem to be:

- Tesla with autopilot crashes
- We find out Autopilot is enabled
- Internet & News goes nuts, autopilot is dangerous, its killing people blah blah blah
- We find out the user wasn't paying attention/was looking at their phone or otherwise
- The story goes away

Now personally I (clearly) am a Tesla fan however I feel I am also pretty rational in these scenarios. And after some digging it does seem to come down to the above

- You You Xue crashing his model 3, was looking at his phone
- This Model X, guy didn't have his hands on the steering wheel, was alerted multiple times before the crash
- The Model S which hit the firetruck, woman was on the phone (if I remember correctly)

Autopilot is far from perfect, it is pretty clear you cannot just leave it to do its own thing. Again, the name is misleading.

Part of me thinks the internal cameras should be recording for insurance purposes (ie if you're on your phone, then you're at fault etc)

Yup, agree with this. Maybe they were thinking too far ahead when they called it AutoPilot :)
 
But the car is on autopilot? It is driving? How can you have a system where you can remove your hands, do something else entirely, and still expect someone to regain control if it goes wrong, in a manner of seconds.
 
But the car is on autopilot? It is driving? How can you have a system where you can remove your hands, do something else entirely, and still expect someone to regain control if it goes wrong, in a manner of seconds.

You're making the same mistake everyone else does. Tesla's "Autopilot" is a brand name for their driver assist systems. It is not meant to drive the car by itself. Giving it the same name as the systems used in aeroplanes to fly themselves with no input from the pilot is misleading people into thinking Tesla's Autopilot will do the same, when it isn't designed for that at all. If Tesla called the system something like "Drive Buddy", people would stop thinking the car drives itself.
 
Last edited:
But the car is on autopilot? It is driving? How can you have a system where you can remove your hands, do something else entirely, and still expect someone to regain control if it goes wrong, in a manner of seconds.

You're making the same mistake everyone else does. Tesla's "Autopilot" is a brand name for their driver assist systems. It is not meant to drive the car by itself. Giving it the same name as they systems used in aeroplanes to fly themselves with no input from the pilot is misleading people into thinking Tesla's Autopilot will do the same, when it isn't designed for that at all. If Tesla called the system something like "Drive Buddy", people would stop thinking the car drives itself.

That sums up the naming problem perfect I think!
 
Back
Top Bottom