Autonomous Vehicles

Fr0dders said:
What's also in the article is that tests by safety tech company Mobileye revealed the existing volvo systems would have been able to detect the pedestrian one second before the crash based solely on the dash cam video.

The video from uber clearly shows the car did nothing at all, meaning their tech isn't even on par with stuff already out there on the market. Hugely damning of the Uber programme.
Click to expand...

Many of us have commented along the same lines about the current state of Uber's computer vision system and the apparent inability to avoid the fatal crash. Many industry players have expressed strongly that either their systems would have handled the situation differently with a better outcome (Waymo) or deep concern about Uber technology decision making (Mobileye, Velodyne). I have argued in this thread that Uber should strongly consider doing more than temporarily closing down their programme while the investigations are pending---and this could be for a while because NTSB investigations often take many months. In fact I have argued for Uber to partner with a stronger player (Waymo is a shareholder of Uber).

Uber's CEO has stated that they wish to become a publicly traded company (IPO) in 2019 and therefore they are under considerable pressure to show a profit in their business. Some might argue that racing to bring their self driving vehicle to commercial reality was caused by this need to show a profit and cut costs. Recently they sold their South East Asian operations to a competitor in exchange for an equity stake in this competitor----they were losing money in that market region. According to Reuters, they have held talks with Ola, their competitor in India about a merger. They are doing deals in many parts of the world to accelerate their path to profitability.

In my view, the after-shock for Uber after this crash is still being felt. They have pulled their entire fleet from the road. They have been suspended in Arizona, their main testing site and home to 400 employees and dozens of self driving vehicles. Their partners who have either supplied hardware or agreed to test Uber's software, namely Nvidia and Toyota, have suspended their self driving vehicle testing. They have withdrawn from the California market. In other words, while the investigations continue, they are racking up huge costs and getting further away from any profit potential in their self driving vehicles.

Surely their entire self driving vehicle strategy is under active review. The CEO, soon after settling their recent lawsuit with Waymo, expressed his interest in a partnership with Waymo.

Stay tuned.
 
Waymo management discussing its testing for the weirdest possible situations, esp with people running out in front of their AVs while they are in motion. They have "hyper well trained safety drivers". For thousands of the 5 million miles that Waymo has driven autonomously, they carefully constructed edge cases. Take a listen.....video is less than 2 minutes.

https://www.technologyreview.com/video/610675/waymo-tests-for-the-weirdest-possible-situations/

And why they are so confident in the safety of their AV's:

https://www.kxly.com/news/national-news/waymo-ceo-our-selfdriving-cars-are-safe/721831340
 
Last edited:
Reuters: "
Uber’s use of fewer safety sensors prompts questions after Arizona crash"

According to former employees of Uber's self driving programme, they scaled back on the use of LIDAR to a single one. Compare this to the six on Waymo's vehicles. They say that this has created blind spots!

"
Uber referred questions on the blind spot to Velodyne. Velodyne acknowledged that with the rooftop lidar there is a roughly three meter blind spot around a vehicle, saying that more sensors are necessary.

“If you’re going to avoid pedestrians, you’re going to need to have a side lidar to see those pedestrians and avoid them, especially at night,” Marta Hall, president and chief business development officer at Velodyne, told Reuters."

https://uk.reuters.com/article/us-u...s-questions-after-arizona-crash-idUKKBN1H337Q


 
Although Waymo and Uber settled their lawsuit recently, arbitration proceedings are still being pursued by Waymo against two ex-employees who founded Otto Trucking that was sold to Uber. One of the founders was Anthony Lewandowski who was fired by Uber last year. The other co-founder, Lior Ron, resigned from Uber today. Until today, he was CEO of Uber Freight, Uber's self driving truck programme.

With all AVs from Uber now firmly grounded, it is widely expected that other employees at Uber's self driving efforts are now looking for alternative employment.

https://www.cnbc.com/2018/03/28/ube...ior-ron-is-leaving-the-company-dow-jones.html
 
sesevans said:
Many of us have commented along the same lines about the current state of Uber's computer vision system and the apparent inability to avoid the fatal crash. Many industry players have expressed strongly that either their systems would have handled the situation differently with a better outcome (Waymo) or deep concern about Uber technology decision making (Mobileye, Velodyne). I have argued in this thread that Uber should strongly consider doing more than temporarily closing down their programme while the investigations are pending---and this could be for a while because NTSB investigations often take many months. In fact I have argued for Uber to partner with a stronger player (Waymo is a shareholder of Uber).

Uber's CEO has stated that they wish to become a publicly traded company (IPO) in 2019 and therefore they are under considerable pressure to show a profit in their business. Some might argue that racing to bring their self driving vehicle to commercial reality was caused by this need to show a profit and cut costs. Recently they sold their South East Asian operations to a competitor in exchange for an equity stake in this competitor----they were losing money in that market region. According to Reuters, they have held talks with Ola, their competitor in India about a merger. They are doing deals in many parts of the world to accelerate their path to profitability.

In my view, the after-shock for Uber after this crash is still being felt. They have pulled their entire fleet from the road. They have been suspended in Arizona, their main testing site and home to 400 employees and dozens of self driving vehicles. Their partners who have either supplied hardware or agreed to test Uber's software, namely Nvidia and Toyota, have suspended their self driving vehicle testing. They have withdrawn from the California market. In other words, while the investigations continue, they are racking up huge costs and getting further away from any profit potential in their self driving vehicles.

Surely their entire self driving vehicle strategy is under active review. The CEO, soon after settling their recent lawsuit with Waymo, expressed his interest in a partnership with Waymo.

Stay tuned.
Click to expand...

And now I have just seen this blog post which confirms my view of a possible partnership between Waymo and Uber:

https://www.theringer.com/tech/2018/3/28/17174548/uber-waymo-self-driving-cars
 
Crashed Tesla Model X had Autopilot turned on

According to Tesla this morning, the driver of the crashed Model X had Autopilot turned on. This crash, coming 5 days after the fatal Uber crash has some same but many different characteristics to the Uber crash. For one, the X uses Autopilot, a Level 2 feature which is really a glorified ADAS system....assistance to the driver, while the Uber car was an autonomous vehicle that apparently failed. Yet it seemed the driver of the Tesla used his vehicle like a robocar, according to this interpretation.

http://ideas.4brad.com/tesla-model-x-fatality-silicon-valley-had-autopilot-turned
 
If you are going to read one article about autonomous vehicles and some of the really intriguing questions they raise about the future of our cities and our lives, here is a good place to start, namely an article about Waymo in The Atlantic (28 March 2018). In it the author raises some intriguing questions and comments including:

1. Within 2 years the US will have entered an entirely new phase in robotics and technology.

2. Until now, maybe 10,000-20,000 people will have had a ride in a self driving car. By 2020, up to 1 million people could have that experience every single day. Autonomous driving at scale will take place within two years.

3. Until now the main question for self driving cars has been: Do they work? And a number of subsidiary questions flow from that:

"Which sensors create the best data? How is that data integrated into a model of the world? What algorithms are best for choosing a route? What kinds of tests generate the most useful new data? What simulation software can accelerate real-world learning the fastest? What are the modes of failure?"

4. Do Waymo think their cars are substantially safer than human drivers? Yes.

"Waymo is not spending a billion dollars on cars for a fleet because they are not sure if their technology works. It works well enough to begin rolling it out to a million people a day."

5. The article raises the possibility of rethinking "the role of wheeled transportation and urban life." Waymo just gave us a timeline.

Comments?

https://www.theatlantic.com/technol...ant-self-driving-car-announcement-yet/556712/
 
"Former Uber back-up driver: We saw this coming"/ Citilab

"Yet the crash raises another question that is perhaps harder to answer by any one investigation: whether humans can be expected to perform safely as backup drivers for semi-autonomous cars at this stage of testing, by Uber or any other company. Research shows that humans will always be fallible, distractable, and, perhaps, easily seduced by machines that function safely—most of the time."

https://www.citylab.com/transportation/2018/03/former-uber-backup-driver-we-saw-this-coming/556427/
 
the Tesla crash is very interesting. The lines for the barrier are not hatched and could be confused as lane markings, as media and other people have mentioned and filmed. But how did it miss a concrete barrier? The other thing which is confusing is other people have driven that stretch and have had no issues with auto pilot. Something wrong with the car? some external factor that changed the situation compared to other cars on that stretch or what?

Also a darwin award seeing as his family have said he's complained about that barrier before, then why would you leave AP on and not pay attention. Although Tesla says no records of a complaint. So maybe complained to family and I should report this but never did.

Who knows but the final report is going to be interesting.

Also @sesevans I've just finished MAchines that think: everything you need to know about the coming age of AI by NewScientists, it does have a bit about Driverless cars but it covers the entire range of ai, thought you might be interested in it.

the technical solution to back up drivers and for autopilot like systems, is driver facing cameras with eye tracking, until we reach true level 5. But I would be surprised if that was passed in law.
 
Glaucus said:
the Tesla crash is very interesting. The lines for the barrier are not hatched and could be confused as lane markings, as media and other people have mentioned and filmed. But how did it miss a concrete barrier? .
Click to expand...

Quite simply and the very reason many doubt this autopilot claim with level 2 tech. It has no LIDAR - interpretation of light is quite different to actually bouncing light off physical objects.
 
Except Tesla's are equipped with radar, and lidar is not a solve all solution like you and many others thing, it is poor at seeing through weather.
 
You and many others? Don’t gauge my opinion on this. I’m not pretending to know everything about it.

Don’t get all bolshy with me, I was answering your concrete block question without donning my Captain Obvious cape.

Sensible approach has both RADAR and LIDAR. I’ll just wait until I’m working in an OEM that’s doing 20k autonomous cars rather than sitting behind a master armchair keyboard...
 
Yes it's obviuse you don't have an idea. And use radar isn't light hence why it kind of knocks your reply out of the water.
Sensible approach is what works and what is economically viable, lidar isn't economically viable yet and has many of the same draw backs as other systems, fixing those draw backs is very similar for all systems so it then becomes redundant.

Lol how was that balshy

Your usual nonsense and overaction post. Think I'll go back to not replying to you.
 
It’s just not tech appropriate for a feature called “autopilot” then people get surprised these incidents happen.

I’m so flattered you even think you have enough of a clue to try retorts. Keep that armchair going sn0wflake.
 
You must log in or register to reply here.
Back
Top Bottom