Separate names with a comma.
Discussion in 'Autonomous Vehicles' started by davidc18, Sep 11, 2017.
Headline over at tech crunch
I'd wait for a second source.
GM built 50 cars, mostly Bolts, but some Leafs too, one-by-one by hand.
Then they built a batch of 130 Bolt AVs:
GM is ready to send 130 more self-driving Chevy Bolts onto US streets
So there are 180± on the road today.
GM and Cruise announce first mass-production self-driving car
These articles appear to use data from before the 130 run.
Notice they say 50 cars, and will ramp up now. This is what was said when the 130 units were being mass-produced.
BTW - Lyft in San Francisco will soon have some AV cars. The ride will be free. There will be a safety operator present.
Lyft cars with self-driving AI will hit San Francisco streets
This is the 3rd gen version.
How we built the first real self-driving car (really)
I'll still wait until I hear more from Mary Barra. She rarely plays the 'marketing hype' game.
If Chevrolet and Cruise were making 500 AV Bolts a month, she would probably have mentioned it.
First cars were hose clamp and duct tape Nissans. Then pre-production Bolts, not all exactly alike. Then they produced 130 identical units from production retail parts, not retrofits, from Body-In-White forward to a finished Bolt AV using an assy line. G1/G2/G3.
But we will see. When Lyft offers AV taxi service in SF, we might know more, but probably not.
There is quite an interesting discussion about Cruise on YCombinator which seems to suggest it's all still vaporware: Cruise's 3rd generation self-driving car | Hacker News
Some of the comments:
Hacker News....a great source for reliable information I'm sure. Lol
"Until then, expect many more controlled environment (eg. airports or freeways where traffic is mostly smooth flowing, or roads with perfectly marked stop signs and pavement paint)."
The streets of San Francisco with bikers (and even raccoons) qualify as "controlled environments" with perfectly marked stop signs and pavement paint?
We shall see. Google / Waymo are the most advanced and are at one disengagement every 5,000 miles or so, which is not good enough. I doubt anybody has self driving software. And by that I mean software that works 99.999% of the time (even that may not be enough). The problem is not normal day to day driving, it's the edge cases. I am fine with people disagreeing with me, but I bet the January 2018 DMV disengagements report will reveal a lot. My opinion is that the last 1% of FSD development will take 90% of the time.
Unlike many automakers, GM's boss is an electrical and automotive engineer with high levels of experience in running factories at the floor level, managing said factories, human resources, supply chains, design, and of course being a CEO. She is not a stupid as most auto CEO's when it comes to the nuts and bolts of technology and how to get it in customer's driveways.
"Big promises being made to GM by Kyle which is leading to smoke and mirrors to distract/buy time from more critical problems "under the hood."" Bull.
There was a Jump Off The Cliff day. Mary reviewed what Cruise could do from both the floor, and from inside the test cars. But only after a lot of research did she show up, and nearly by ambush. Mary knew very well what the V2V and Geomapping systems GM had running that day were doing in separate GM division.
She was going to make the decision to jump off the AV cliff that day by focusing 50% of their engineers on EV/AV tech, or putting it on a steady diet for future use or focus entirely on Cadillac. At the end of the day? "Let's do it. Let's make AVs now."
When asked by the press when AV is coming? "Sooner than you think."
Believe it or not. Your choice.
Exaggeration and outright lying is very common at the CEO level in the industry. Mary doesn't have time for the "men beating on their chests or bragging about the fish they caught", she measures results like an engineer, not a CEO trying to get investors.
Waymo has nothing to go coast to coast. In fact, most their 'miles' are virtual or closed course at under 75mph. Actually under 45. Waymo cannot yet produce or even select a decent automobile yet. That is a WHOLE lot of the puzzle. An AV nobody wants isn't as good as one buyers want.
If we are using demo videos as a measuring stick then Tesla is also very advanced in FSD. That is not really the case.
Below is the reality in California as of last year. None of the results are very good.
This is a touchy subject. As much as I love technology and I want self-driving ASAP, I do not want to end up a statistic.
A lot has changed in 2017. The GM results you see were only 26 cars, 5 were Nissans. Nov 2016 logged a GM record 2284mi street miles with 6 disconnects. That's 380 miles per disconnect. There are now at least 150 more AVs on the road and it's been 9 more months of development.
However, California was only one of three states the testing is occurring on city streets, not including track testing.
Nobody really knows how Tesla is testing full autonomy at this point. It's quite possible that they have people testing FSD in standard looking Tesla cars with a driver behind the wheel and nobody would know the difference.
Have you yet seen Tesla hide an advanced feature from the public that is in operation?
How would you know it if you saw it? All current Teslas have the cameras and sensors in place, so the only thing that would need to be added would be software. More than once I have seen Teslas with California manufacturer plates at Superchargers in North and South Carolina.
Tesla Motors tends to be a 'forward looking' company. It is normal for them to announce future features, plans, and dates.
Their website shows a November video titled 'the driver is there for legal reasons only' for their Full Self Driving feature. We know that is not a 100% accurate description of the status yet.
There is nothing wrong with that, it's just a reminder that they are not shy about releasing info on features in progress.
If you're Rocky Raccoon in the video above, just be thankful that was a Cruise coming at you rather than a Tesla.