Autonomous Vehicles

[Fusion]

Great, sesevans spamming another thread/forum with comments where Google have skin in the game .

Does seem excessive, especially given his other heavily pro-Google threads and posts. Paid advocate?

 

[sesevans]

Does seem excessive, especially given his other heavily pro-Google threads and posts. Paid advocate?

Answer: not. Just a tech enthusiast.

 

[sesevans]

Anyone notice today's announcement between Google and JD.com of China concerning a joint venture, initially with a purchase of $ 550 million of JD shares by Google?

While I expect much discussion on areas that are obvious for a joint venture, namely search ad generation, artificial intelligence and machine learning and Google cloud services, I wonder if it will be viewed as a front-runner for joint collaboration for both Waymo and Google Maps in the areas of delivery and logistics? The geographic areas of focus are South East Asia, Europe and the US, areas that I see Waymo expanding into over time.

Thoughts?

 

[sesevans]

Driverless vehicles could eliminate millions of jobs in future. A number of companies that are leading the AV disruption are joining forces to consider the "human impact" of this technology.

A partnership of leading AV companies say “Concern for the safety of workers and the public is paramount to PTIO,” the group’s executive director Maureen Westphal, said in an email to The Verge, “and safe deployment of [autonomous vehicle] technology is fundamental to securing better job opportunities for workers, so we plan to engage with a variety of concerned stakeholders already having conversations and planning for this transition to an autonomous vehicle future.”

https://www.theverge.com/2018/6/19/17476112/waymo-uber-ford-self-driving-cars-coalition-jobs

 

[sesevans]

Boston, Mass has approved the expansion of self driving vehicles from Nutonomy. The approval shows the stark difference between States.

Unlike Arizona which has approved a ride hailing taxi service with no human driver in the front seats, Boston requires two human safety drivers including one in the drivers seat with a hand near the wheel.

http://www.ttnews.com/articles/boston-oks-citywide-tests-nutonomy-self-driving-cars

 

[sesevans]

The Uber safety driver in the Tempe Arizona fatality was streaming Hulu tv show on her phone just at the time of the crash.

The crash was therefore entirely avoidable. When we first began talking about this crash we speculated whether a distraction was causing the safety driver to look down and away from the road. Now we know.

Uber still has not restarted its self driving car programme since the crash and we also know they have had discussions with Waymo about introducing Waymo self driving vehicles into the Uber fleet.

https://uk.reuters.com/article/uk-u...driving-car-crash-police-report-idUKKBN1JI0L9

 

[VincentHanna]

So once again, it was somebody not doing their job correctly / assuming fully autonomous means they don't even have to keep an eye out.

Sad for the poor women killed, but that's why they employ drivers to avoid situations like this whilst the cars are being tested...

 

[sesevans]

So once again, it was somebody not doing their job correctly / assuming fully autonomous means they don't even have to keep an eye out.

Sad for the poor women killed, but that's why they employ drivers to avoid situations like this whilst the cars are being tested...

I agree that someone was not "doing their job correctly", but it is really debatable who that somebody was.

Yes the "safety" driver was distracted and disregarding Uber policy of not keeping a mobile phone in the car to distract. But try doing a crushingly boring job for hours on end and see how easy it is to become tired and distracted. It becomes extremely easy to "trust" the technology, which one could argue is happening to some extent with Tesla "Autopilot", despite all the warnings not to rely on it. Uber relied on imperfect hardware and software because they assumed their "safety" driver would bail them out of an accident happening. They also disengaged the automated braking systems on the Volvo to ensure a "smoother" ride.

I have mentioned it before but it is worth repeating. About 4 or 5 years ago, Google/Waymo tested their autonomously driven vehicles with company employees in the drivers seat and found the humans became easily distracted. Since then they have never put a human driver in the drivers seat and rely "exclusively" on their AV technology to "drive" the car. Their performance since then on public roads (they have "driven" in autonomous mode more than 7 million miles) is nearly flawless.

Which is why:
1. They plan to launch their commercial ride hailing AV service this year in Arizona.
2. They plan to have no human "driver" anywhere near the steering wheel or brakes.

 

[cuke2u]

Was it not the case that if the car's actual pre-sense had been turned on the accident would have been avoided? I seem to remember reading that and know my car would have stopped before hitting the woman thus this suggests they were either not running fully automonous or their systems aren't up to it..

 

[sesevans]

Was it not the case that if the car's actual pre-sense had been turned on the accident would have been avoided? I seem to remember reading that and know my car would have done so before hitting the woman thus this suggests they were either not running fully automonous or their systems aren't up to it..

What the NTSB report found was that Uber took poor engineering decisions that contributed to the crash. Specifically the NTSB said the AV's software "determined that an emergency braking maneuver was needed" 1.3 seconds before the crash. However the Uber AV wasn't programmed to perform emergency braking procedures, nor was it programmed to alert the safety driver. Uber disabled the Volvo's factory installed emergency automatic braking feature.

I am not in any way avoiding focus on the "safety" driver. After all, when she was first interviewed by the NTSB she said her mobile phones, which were in the car, were switched off until she called emergency services after the crash. We now know this was untrue---she lied---because her phone was on and she was watching a Hulu tv show. The new report said she looked down 166 times while the vehicle was in motion, mostly in the direction of her right knee.

However Uber decided not to have a second safety driver in the vehicle, which some other AV companies still have, especially those still dependent on humans in the front seat (unlike Waymo). Perhaps they asked and answered the question by assuming that a non-driving passenger (called a safety assistant) would not be more alert than the safety driver and perhaps they could cut costs too? Questions must also be raised about Uber's hiring policies in general to have an individual that could so flagrantly flout Uber policies about mobile phone usage and conceal it. And here again a second safety assistant reviewing software readouts in the vehicle might mean that it is harder for the other person to flout the rules.

Although not directly related to the Arizona fatality, there is an interesting piece of timing for Uber in London next week. Uber is due to appear before the Magistrates Court in Westminster to argue why it should be allowed to continue to operate its ride hailing service in London. While I assume that Uber will reach a compromise with the court that allows it to continue operating here, one has to wonder if the operating culture at Uber has changed all that much since the appointment of a new CEO? A set-back and loss of operations in London would be a very heavy blow to Uber ahead of its anticipated 2019 initial public offering.

Again my thinking of Uber's AV future under the new CEO leadership: rapid changes need to occur and new thinking needs to take place. If I recall correctly, Uber's early response to the Sheriff's office after the crash was that the safety driver was monitoring diagnostics. Uber needs to figure out how to collaborate with Waymo (and be linked to Waymo's more than 7 million miles of successful AV testing) because the chance the public will forget Uber's poor handling of the safety of its AVs on public roads are slim. Waymo parent Google is a shareholder in Uber so the idea that they might collaborate anyway seems quite possible.

 

[sesevans]

The United States plans to ratchet commercial tensions higher by barring many Chinese companies from investing in U.S. technology firms, and by blocking additional technology exports to Beijing.

As we know, there is a race to succeed in AVs and many Chinese companies, like Pony.ai founded 18 months ago, are actively involved in recruiting US talent to assist them. The online university Udacity also provides advanced degrees in machine learning and data science.

Worth following developments here to see if the attempts to limit tech transfer, a hallmark of Chinese development which can be argued is unfair under WTO rules, affects AVs.

http://www.ecns.cn/news/sci-tech/2018-06-25/detail-ifyvmiee7353919.shtml

 

[dowie]

Elon's big tent seems to have "triggered" automotive industry analysts

http://fortune.com/2018/06/25/elon-musk-tesla-tent-factory-model-3/

Elon Musk has six days to make good on his pledge that Tesla will be pumping out 5,000 Model 3 sedans a week by the end of the month. If he succeeds, it may be thanks to the curious structure outside the company’s factory. It’s a tent the size of two football fields that Musk calls “pretty sweet” and that manufacturing experts deride as, basically, nuts.

“Words fail me. It’s insanity,” said Sanford C. Bernstein & Co.’s Max Warburton, who benchmarked auto-assembly plants around the world before becoming a financial analyst.

[...]

The assembly line inside will start out as fully manual and gradually convert to automation in steps, according to Brian Johnson, an analyst at Barclays who met with Tesla’s investor relations department last week.

“It’s preposterous,” Bernstein’s Warburton said. “I don’t think anyone’s seen anything like this outside of the military trying to service vehicles in a war zone. I pity any customer taking delivery of one of these cars. The quality will be shocking.”

 

[sesevans]

Elon's big tent seems to have "triggered" automotive industry analysts

http://fortune.com/2018/06/25/elon-musk-tesla-tent-factory-model-3/

An interesting read. Not exactly related to autonomous vehicle thread but certainly interesting background on build quality, etc.

 

[sesevans]

By the way, Uber has just won its appeal to overturn the ban on their London operations. This will be well received by Uber ahead of their planned 2019 IPO and might help their image overall.

I still view a deal with Waymo as a reasonable probability.

 

[jpaul]

By the way, Uber has just won its appeal to overturn the ban on their London operations. This will be well received by Uber ahead of their planned 2019 IPO and might help their image overall.

they had never been off the road.. but afaik they still need to show how they will address the vetting concerns raised 2016

Uber drivers accused of 32 rapes and sex attacks on London passengers over the past year
or more recently usa
CNN investigation: 103 Uber drivers accused of sexual assault or abuse

Sadiq Khan will be back on their backs

 

[Nasher]

If it's still going on I can't see why it was overturned

I guess it's ok to be a bit rapey if the prices are cheap?

 

[dowie]

An interesting read. Not exactly related to autonomous vehicle thread but certainly interesting background on build quality, etc.

true, the model concerned isn't autonomous though the company itself is heavily invested in autonomous vehicle related research and this sort of thing does pose some risk for them

 

[jpaul]

There is a halo effect of the Tesla built on Friday afternoon in a tent strategy, investors see a failure to execute in the segment by the elected king,
the legislative environment may also tighten for AV's on the back of autopilot problems ... so king's no clothes scenario ? ROI pushed out 5 years.
Investors may want the confidence of close (waymo/tesla) R&D/production shared with the regular/proven auto manufacturers.

Equally if Uber has problems to sort out, then pre-orders for waymo a/v's maybe premature (even if the drivers are retired)

 

[sesevans]

There is a halo effect of the Tesla built on Friday afternoon in a tent strategy, investors see a failure to execute in the segment by the elected king,
the legislative environment may also tighten for AV's on the back of autopilot problems ... so king's no clothes scenario ? ROI pushed out 5 years.
Investors may want the confidence of close (waymo/tesla) R&D/production shared with the regular/proven auto manufacturers.

Equally if Uber has problems to sort out, then pre-orders for waymo a/v's maybe premature (even if the drivers are retired)

Yes, while I often suggest that Waymo has a lead in AVs, I am quick to point out that the OEMs have expertise that should not be discounted. Supporting my view is the recent massive investment made by Softbank in GM Cruise AVs. Softbank clearly believes that the road ahead for AVs will be longer to develop than some believe (due to legal, regulatory and some infrastructure issues) and therefore the OEMs will remain
in a coveted position to collect, analyze and monetize vast amounts of data.

Data is the most valuable asset in training what I see coming: a "swarm" of mobile robots. This vast pool of data is needed to train a neural network. How this data is optimized will shape hardware design which itself impacts how data is collected and communicated between vehicles, in turn influencing software design.

To be successful with AVs, I believe it will involve collecting and analyzing and incorporating data from hundreds of billions of miles of real world driving. I see no reason why the OEMs cannot manage large fleets of vehicles. With their dealer network they are able to buy, finance, maintain and dispose of vehicles. Their advantage must surely be their ability to physically collect data and their ability to physically distribute content. But there are areas where their skills and the skills of Waymo and other tech firms are complimentary too. For example, while auto companies own the data and manage the fleet, tech firms want to grown subscriber bases, reduce logistics costs and collect/monetize data.

In a world where vehicles are data-capturing and autonomously operated, changing the thinking around data privacy and national security will be important considerations.

So perhaps there will be room for a few winners rather than winner take all?

 

[sesevans]

And don't count Apple out of the AV race. Or so it seems.

"YouTube Star secretly develops VR for self driving cars for Apple"

VR for on-board entertainment in Apple self driving cars--for all those passengers.

https://variety.com/2018/digital/news/youtube-maerk-rober-apple-vr-1202858396/