Tesla's "innovation" is being willing to ship "beta" features for better or for worse.
It's not like Chevy couldn't have used the same data it uses for LKA to do lane centering, by as Tesla has shown, doing that can cause "wrong" steering outputs (I.e. ones that would put you in a ditch), and opens up the system for abuse by the driver.
Where as only using it for LKA means when the sensor is "wrong" the worst case is a wrong LKA warning instead of doing what Teslas do on "fail road". And you can't abuse LKA easily. I've noticed my Volt seems to be tuned so that if you try and abuse LKA and let go of the wheel, it'll take longer and longer to apply corrective steering, causing the car to "ping pong". But if you hold the wheel and intentionally try and pass a line with firm, but wrong steering input (as you'd expect with a distracted driver), it will smoothly guide you along the lane as precisely as a human driver would.
To me Super Cruise is a more major innovation than Autopilot. AP works in more situations but the situations it works in, that Super Cruise doesn't, are situations where AP is notorious for having bugs that vary from inconvient to downright dangerous. It reminds me of how people compare AP to other equivalent systems like Distronic + based on disengagements, ignoring the fact that more disengagements can be preferable if it means avoiding events where there isn't a disengagement but the system is wrong
But it looks like Honda is the odd man out, as some of their cars have ACC, but don't have what they call "Low Speed Follow" which is the ability to slow down to a stop, then resume with the push of a button or the accelerator
There are two autopilot systems commercially available today - Tesla Autopilot and Honda Sensing Driver Assistance. Two more are slated to be released - Cadillac SuperCruise and Audi Traffic Jam Pilot. The rest is an all-you-can-read buffet of press releases about some great tech this one company tested in this one lab (I hear Waymo is great, and Volvo’s is even better).
Pending Cadillac’s and Audi’s road tests in 2018, Tesla Autopilot v1 is the best driver assistance suite available commercially in 2017, even with all of the executive departures.
It's an interesting time for Tesla, they feel like the automotive equivalent of AMD - David keeping Goliath honest. How many automotive companies saw Tesla become an actual disruptive threat and then spent tens of billions on self driving and electrification? Tesla isn't going to compete with Honda or Toyota by copying their model (or even their manufacturing process).
Take a minute and look at the bolt and the leaf - they're god-ugly, emotionless, unsexy modes of transportation. Buying a leaf is the result of sitting down with a spreadsheet. Need some anecdotal proof? Think to yourself how many of your coworkers, friends and family you can see driving one. How about a model 3? I see more model S's during my commute than I do leafs, volts and bolts combined.
Another place to look for confirmation is the lithium ion battery line which was built in collaboration with Panasonic. Panasonic is 2/3rd the size of Toyota and are well versed in manufacturing process.
If Tesla wanted to borrow a page from the incumbents they'd be doing so, they have enough brains from those companies already on staff. It's a safe bet that there are very few idiots in upper management at Tesla. Elon isn't the type to ignore a process problem, he's a modeling/simulation junkie and probably figured out something that the others are missing. That's not to say things are perfect, but I don't think they're as dire as the article makes them out to be.
The shareholders have faith, and although Tesla shares are slightly bubbly, they're still doing unbelievably well on most fronts.
0. You have fallen for the "post hoc ergo propter hoc" fallacy. Tesla was not the first company that worked on these things. They were just the first who decided to sell a semi practical implementation and hype it up as if it where completely finished and polished.
Can you elaborate?
I could set the cruise control to a certain speed and it would normally slow down if there was congestion, which was very nice.
But it wouldn't work at bends in the road, because the system was seemingly just directly ahead line-of-sight. I wanted to see if it would slow down itself in that situation before I had to slam the breaks, it would have plowed into the much slower-moving car ahead of me.
So my experience is that while these systems help in bumper-to-bumper traffic on congested highways they're still to dangerous to rely on.
I get Tesla's cars are really good, but my understanding was that it wasn't "miles ahead".
And that's before we get into disengagements per 1000 miles, where Tesla trails most of the industry by far.
No wonder Musk tries to demonize AI. Self-driving is a race to the bottom, with its development slashing car sales by an order of magnitude. That would be bad for an established, mature, car maker, but would be crushing for a company relying purely on future growth.
Not only Tesla is behind most in self-driving, it is the most exposed to the downsides of it.
This isn't bad if you're the only one who can ship self-driving cars with acceptable quality and volume as you'd be disrupting the rest of the industry to fuel massive growth.
Steve Jobs was also famous for denouncing something right up until Apple had "the best on the market" of that same thing. It seems like standard CEO playbook.
Fixing it:
This isn't too bad if you're the only one who can ship self-driving cars with acceptable quality and volume as you'd be disrupting the rest of the industry to fuel massive growth.
But you won't be the only one for long. And then the entire industry will downsize.
And Tesla is (1) the most susceptible to it, since most of the enterprise value comes from expectation of future growth and (2) it lags behind other makers in self-driving tech.
Ockham's razor suggests it's more likely Elon Musk simply got convinced by the Yudkowsky/Bostrom view on the problem of Unfriendly AI (I particularly recall Musk starting to talk about AI only after reading Bostrom's Superintelligence, but that might have been just a coincidence).
Also note that Musk is "demonizing" advanced AI that doesn't exist yet, but is huge on self-driving.
7-series and 5-series are simply too expensive. If BMW can get it in the 3-series at a Tesla Model 3 price, things will get interesting.
The learning systems need to be fed data to improve their models, so that they can understand bad lane markup, merging lanes, diverging lanes, lane splits and highway off-ramps, construction cones and barriers, motorcycles and trucks being obstacles and flying plastic bags and garbage not being an obstacle.
Unless they’re buying this data some place else, it needs to come from millions of miles driven by the current owners.
You might be able to DIY.
Coincidentally, we're in the market for a 2018 Odyssey as our family car too.
When I took the Tesla test drive in a Model X with my parents, the rep took pains to educate us about the limitations of Autopilot. Lane following in the highway, below 50 mph around curves, and doesn't recognize traffic lights in the highway. Can't avoid collisions below 7 mph. Can't lane follow in city streets. Can't self-drive on main city thoroughfares. I'm probably missing the message, but I don't see how this is that much more revolutionary than a Honda Sensing suite that I could purchase in a 2018 Odyssey [1]. For all the Reality Distortion Field is putting out from Tesla, I was expecting more.
As it was, I can see value if I sit in rush hour traffic on the highway for an hour or more each day, each way. But unfortunately, that is not my use case. And I can get the same tech on other more mature platforms to address the bulk of the value for that use case.
A lot is riding on Tesla's promise to bring full autonomy In The Future. I fail to see how they are besting Waymo at productizing full autonomy or even taking the next incremental steps towards that goal. If I wasn't in software and following what Waymo et al were doing to move towards full autonomy, then I would probably be dazzled, but I'm struggling to figure out how Tesla's driver assistive tech is getting us to full autonomy faster than Waymo.
I'm on the Model 3 waiting list, but that test drive has me seriously re-considering a 2018 Honda Odyssey instead. I'd welcome feedback on what I missed.
My Odyssey can't do that.
Now I'm confused why Tesla would sound so limited, unless they are trying to cover their ass legally on how you SHOULD use autopilot versus how you CAN use it?
[1] https://www.quora.com/Does-Tesla-Autopilot-work-on-city-subu...
Our 2012 Sienna is doing well and doesn't do too many miles, or I'd have considered upgrading.
Personally, we don't share much - she drives hers, I drive mine and only share duties in the weekend.
With our separate commutes, and a PHEV and and EV we could literally not visit gas stations except on road trips.
I also like the 2018 Odyssey's visual refresh. Sienna hasn't changed in eons it feels like. Inside, I wasn't a fan of the second row seat tracks on the Sienna.
I am on the Model 3 waiting list for the Autopilot. Heck, it could be gas powered and I'd be on the waiting list for the Autopilot. I don't need full autonomous. I just want the best system, and currently Tesla has it. So hopefully they figure out their production issues.
I also just bought a 2018 Honda Odyssey as our main family car and it can't even do adaptive cruise control under 20 mph. I think people underrate how far ahead Tesla is in these areas. Their competition is currently Google and Uber, not Chevy, Nissan, etc.