The Failure Of This Self-Driving Truck Company Tells You All You Need To Know About Self-Driving Vehicles

[SethInMI 163]

8 minutes ago, gowlerk said:

Eventually they will have support from sensors and signals built into the roadway.

Maybe. But probably not anytime soon. The problem with these implementing these systems is tax dollars have to pay for them, and there is a chicken and egg issue of them not useful until many cars have ability to process data, but building the tech into cars is not useful until many places have the sensors in the road.

it does remind me of a system i read about where bmw is working with some cities to get access to their traffic light controllers so their cars can "know" when a light will be red or green and do routing and or slow / speed up depending on schedule. 

so some interaction between infrastructure and cars is happening, but i don't think it will affect autonomous vehicle development.

[gowlerk 2,156]

9 minutes ago, ryoder said:

Bwahahaha! I can't show you plenty of places where I ride in the mountains that don't even have cell service yet, let alone smart roadways. 

Yup, not all roadways. Just major ones. And that gets into why I believe that fully autonomous vehicles are both far into the future and will be limited in scope. They will work best and most efficiently when on a support grid. The economics will favour the use of "driverless" vehicles on fixed routes long before full AVs are practical.

[ryoder 1,590]

9 minutes ago, gowlerk said:

Yup, not all roadways. Just major ones. And that gets into why I believe that fully autonomous vehicles are both far into the future and will be limited in scope. They will work best and most efficiently when on a support grid. The economics will favour the use of "driverless" vehicles on fixed routes long before full AVs are practical.

The Cadillac "Super Cruise" uses LIDAR to assist GPS in pinpointing immediate location, and works only on major highways that have been previously mapped by LIDAR. I think that is more economical approach than trying to make the highways smart.

https://techcrunch.com/2019/06/05/gm-is-opening-up-its-super-cruise-hands-free-driving-system-to-more-roads/

[gowlerk 2,156]

31 minutes ago, ryoder said:

The Cadillac "Super Cruise" uses LIDAR to assist GPS in pinpointing immediate location, and works only on major highways that have been previously mapped by LIDAR. I think that is more economical approach than trying to make the highways smart.

https://techcrunch.com/2019/06/05/gm-is-opening-up-its-super-cruise-hands-free-driving-system-to-more-roads/

Super Cruise uses a combination of lidar map data, high-precision GPS, cameras and radar sensors, as well as a driver attention system, which monitors the person behind the wheel to ensure they’re paying attention. Unlike Tesla’s Autopilot semi automation driver assistance system, users of Super Cruise do not need to have their hands on the wheel. However, their eyes must remain directed straight ahead.

Systems that need to monitor drivers to make sure they are paying attention are major accidents waiting to happen. The human mind does not work that way.

 

Edit, I just noticed that the GM map of "another 70,000 miles of compatible divided highways in the United States and Canada." Includes 100s of miles of Canadian highways that I can assure you are neither divided, nor compatible. 

Edited March 29, 2020 by gowlerk

[mbohu 77]

4 hours ago, SethInMI said:

but i submit that swerving to avoid is not required for autonomous cars to be allowed on roads. it would be nice to have, and a "simple swerve" is a lot different than the complex trolley problem decision tree that mbohu is saying is required.

It doesn't really matter what the exact scenarios or rules are. No matter how you slice it, somewhere at the center of the program has to be some sort of decision making algorithm, which is fed the data from the sensors or even from multiple cars and road sensors.
It sounds like billvon was mostly talking about using neural networks in the pattern recognition algorithms. These would be the algorithms feeding the decision making. It's quite possible that the decision making algorithms themselves are extremely limited right now, and operate on extremely simple instructions (always try to break, never drive faster than your ability to break immediately--but as you pointed out with the jaywalker, this is really not realistic--same with the curve, if you drive around a blind corner, ready for your own lane being completely blocked, you'd have to come to a virtual stop)
The more data is being fed into the decision making algorithm though, the more options it will have available, and the more it will have some kind of morally relevant preferences. This would actually be much more the case in a centralized system that is aware of multiple vehicles (via vehicles sending it data and/or road sensors) It now has the ability to consider consequences for ALL vehicles and in situations that have no perfect outcome it will have to prioritize between all vehicles. Again, the more data it has, the less likely such situations will be, but the likelihood will never be zero.

[gowlerk 2,156]

Another win for self driving cars. Safety features defeated?

https://www.cbc.ca/news/canada/edmonton/tesla-driver-autopilot-alberta-ponoka-speeding-dangerous-driving-1.5727828

[turtlespeed 212]

On 9/17/2020 at 2:59 PM, gowlerk said:

Another win for self driving cars. Safety features defeated?

https://www.cbc.ca/news/canada/edmonton/tesla-driver-autopilot-alberta-ponoka-speeding-dangerous-driving-1.5727828

They need more subsidies to make them better!!!!

Edited September 19, 2020 by turtlespeed

[billvon 2,879]

26 minutes ago, turtlespeed said:

They need more subsidies to make them better!!!!

Naah.  Save those subsidies to increase the bonuses for Exxon executives.  Why, these days some of them can't even afford a second yacht.

[turtlespeed 212]

7 minutes ago, billvon said:

Naah.  Save those subsidies to increase the bonuses for Exxon executives.  Why, these days some of them can't even afford a second yacht.

Nooo.  Can we just get them aimed at something else?

[turtlespeed 212]

8 hours ago, RobertMBlevins said:

Ask yourself WHY they keep trying to invent self-driving cars, and WHY they keep adding things like automatic braking and lane warning systems to cars?

Almost everything that gets invented in the tech genre is done so to fill a market. And the market for these cars, and these systems, is SIMPLE:

Too many drivers are busy watching videos or playing with their smartphones while they drive. This type of tech is invented to countermand human stupidity and foolishness, not so much to make driving 'safer' or 'more convenient'. 

Yes, instead of making restrictions and forcing people to pay attention, lets make it easier for them to be stupid.

 

[olofscience 473]

8 hours ago, RobertMBlevins said:

Ask yourself WHY they keep trying to invent self-driving cars, and WHY they keep adding things like automatic braking and lane warning systems to cars?

Almost everything that gets invented in the tech genre is done so to fill a market. And the market for these cars, and these systems, is SIMPLE:

Too many drivers are busy watching videos or playing with their smartphones while they drive. This type of tech is invented to countermand human stupidity and foolishness, not so much to make driving 'safer' or 'more convenient'. 

 

21 minutes ago, turtlespeed said:

Yes, instead of making restrictions and forcing people to pay attention, lets make it easier for them to be stupid.

 

 

Both incorrect.

While I would welcome a self-driving car for my own convenience in long trips, the bigger drive is for big employers (large truck shipping fleets, taxi fleets, delivery vehicles) massively reduce their payroll expenses.

With millions of people employed around the world just for driving, the potential savings would be immense. And no, these benefits will *not* trickle down.

Many of them will also be electric, and not require as much maintenance as traditional ICE vehicles, so those newly-unemployed drivers won't find new jobs maintaining these new robots. Even many people in the traditional maintenance industry (mechanics, garages, etc.) will see reduced revenue and workload.

Meanwhile the company that owns the vehicle will benefit from both reduced payroll and reduced maintenance costs. Win-win from a business point of view.

[yoink 321]

AI controlled cars would make traffic orders of magnitude more efficient.

Watch this for a short example: 

 

[airdvr 205]

https://www.wivb.com/news/tesla-network-outage-locks-some-customers-out-of-app-car/

Wonder if outages like this would effect autonomous vehicles.

[billvon 2,879]

48 minutes ago, airdvr said:

https://www.wivb.com/news/tesla-network-outage-locks-some-customers-out-of-app-car/

Wonder if outages like this would effect autonomous vehicles.

Autonomous vehicles would use mesh networking to communicate between each other.  Mesh has so many advantages over star networks for vehicles that they will dominate. 

 

[gowlerk 2,156]

1 hour ago, billvon said:

Autonomous vehicles would use mesh networking to communicate between each other.  Mesh has so many advantages over star networks for vehicles that they will dominate. 

 

It’s easy even for myself to understand the advantages of this so-called mesh network. But would it not also have a potential disadvantage of one faulty node in the mesh bringing traffic to a standstill?

Come to think of it each of us driving in our cars today are currently a meshed network.

Edited September 24, 2020 by gowlerk

[brenthutch 428]

I can’t wait until I can ride in my electric, flying, self driving car.  

[billvon 2,879]

2 hours ago, gowlerk said:

It’s easy even for myself to understand the advantages of this so-called mesh network. But would it not also have a potential disadvantage of one faulty node in the mesh bringing traffic to a standstill?

Well, no.  That one car would have to revert to a simpler mode (i.e. visual or LIDAR separation rather than coordinated separation) and would be unable to participate in driving that required coordination, like intersections without traffic lights - so they would avoid them.  But we are a _long_ way from be able to do that anyway.

 

[turtlespeed 212]

13 hours ago, olofscience said:

 

 

Both incorrect.

While I would welcome a self-driving car for my own convenience in long trips, the bigger drive is for big employers (large truck shipping fleets, taxi fleets, delivery vehicles) massively reduce their payroll expenses.

With millions of people employed around the world just for driving, the potential savings would be immense. And no, these benefits will *not* trickle down.

Many of them will also be electric, and not require as much maintenance as traditional ICE vehicles, so those newly-unemployed drivers won't find new jobs maintaining these new robots. Even many people in the traditional maintenance industry (mechanics, garages, etc.) will see reduced revenue and workload.

Meanwhile the company that owns the vehicle will benefit from both reduced payroll and reduced maintenance costs. Win-win from a business point of view.

Some, but they are actually both only partially incorrect from your point of view.

My assumption was that a self driving vehicles will have a driver in them to start. 

They will have them for a LONG time before they will be completely autonomous.

My thinking is that Commercial applications will come behind the public ones. (Corporate liability and all)

 

[olofscience 473]

7 hours ago, turtlespeed said:

My thinking is that Commercial applications will come behind the public ones. (Corporate liability and all)

No, the commercial benefits are too high for businesses to ignore.

While Uber's self-driving car efforts are questionable, they managed to kill one pedestrian and yet less than 2 years later they're back on the roads continuing development. Liability won't slow them down that much. There is a LOT of money on this.

[wolfriverjoe 1,495]

3 hours ago, olofscience said:

No, the commercial benefits are too high for businesses to ignore.

While Uber's self-driving car efforts are questionable, they managed to kill one pedestrian and yet less than 2 years later they're back on the roads continuing development. Liability won't slow them down that much. There is a LOT of money on this.

Bingo. Driver pay is the 2nd highest expense for trucking companies. (fuel is first)

Stuff like line haul, the doubles & triples up and down the turnpike, are probably going to be the first on public roads. 
I'm not super sure how well the public will take to the idea of unoccupied semi trucks, but right now, the technology could probably do it under good weather.

Some of the big pit mines out west are using autonomous right now (private land). The big (really big) dump trucks just have to go from where they get loaded to where they dump. The dump location is fixed, and they can set up markers for the sensors to read for loading. My understanding is that traffic cones with specific reflectors are used.

 

Also, keep in mind that the Uber fatality was not the car's fault. The person killed was jaywalking.  And the car detected the person, but the 'emergency braking' capability had been disabled for passenger comfort (as is often the case, this is what I understand and I welcome correction if wrong).

[ryoder 1,590]

59 minutes ago, wolfriverjoe said:

Also, keep in mind that the Uber fatality was not the car's fault. The person killed was jaywalking.  And the car detected the person, but the 'emergency braking' capability had been disabled for passenger comfort (as is often the case, this is what I understand and I welcome correction if wrong).

Another factor I rarely see mentioned is that some fool in city planning put what looked like sidewalks in the median between the lanes, which would have led any pedestrian to think they were supposed to cross there. But they weren't sidewalks; They were just landscaping decoration.  I just now looked again at the location, and I see they removed the (non)sidewalks, but you can still see where they were. 

Note the faint Y-shaped rough path in this photo.

Location: https://www.google.com/maps/place/E+Curry+Rd,+Tempe,+AZ+85281/@33.4363769,-111.942976,69m/data=!3m1!1e3!4m5!3m4!1s0x872b09182728f1b3:0xc0fa33b5ec29c865!8m2!3d33.4402456!4d-111.9263235

 

Edited September 25, 2020 by ryoder

[turtlespeed 212]

7 hours ago, olofscience said:

No, the commercial benefits are too high for businesses to ignore.

While Uber's self-driving car efforts are questionable, they managed to kill one pedestrian and yet less than 2 years later they're back on the roads continuing development. Liability won't slow them down that much. There is a LOT of money on this.

We agree at that point.  I just think that commercial will lag behind personal.

It's just my opinion.

[airdvr 205]

On 9/25/2020 at 1:24 PM, turtlespeed said:

We agree at that point.  I just think that commercial will lag behind personal.

It's just my opinion.

I think it's the exact opposite.  As stated before truck drivers are the single biggest liability on the books at most larger trucking companies.  Ever been to a truck stop and taken a look at who's eating there (my apologies to Gowlerk).

Most deaths in large truck crashes are passenger vehicle occupants. The main problem is the vulnerability of people traveling in smaller vehicles. Trucks often weigh 20-30 times as much as passenger cars and are taller with greater ground clearance, which can result in smaller vehicles underriding trucks in crashes. 

Every one of those passenger vehicle occupant fatalities is soon followed by an ambulance chasing attorney. 

 

[gowlerk 2,156]

9 hours ago, airdvr said:

Ever been to a truck stop and taken a look at who's eating there (my apologies to Gowlerk).

I never eat at truck stops I have a fridge, a 12 volt oven, a coffee maker, and a large inverter. I shop for groceries and eat healthy. But I sure do see quite a few "extra large" drivers waddling into the restaurant when I stop for fuel! So you are not wrong.

Edited September 29, 2020 by gowlerk

[wolfriverjoe 1,495]

10 hours ago, airdvr said:

I think it's the exact opposite.  As stated before truck drivers are the single biggest liability on the books at most larger trucking companies.  Ever been to a truck stop and taken a look at who's eating there (my apologies to Gowlerk).

Most deaths in large truck crashes are passenger vehicle occupants. The main problem is the vulnerability of people traveling in smaller vehicles. Trucks often weigh 20-30 times as much as passenger cars and are taller with greater ground clearance, which can result in smaller vehicles underriding trucks in crashes. 

Every one of those passenger vehicle occupant fatalities is soon followed by an ambulance chasing attorney. 

 

Interesting take on it and one I hadn't thought of.

 

My point was purely from an operating cost standpoint. That is, driver salaries are the number two expense behind fuel.

Liability for injuries & deaths from crashes goes to the insurance companies. No idea where 'insurance costs' falls on the balance sheet. 

However, while 'car vs truck' accidents are always harder on the car (physics rules), the fault lies with the driver of the car in the majority of the cases. Depending on where (and when) you look, it's between 70% and 85% the fault of the car driver.

As a result, dash cams are becoming very common in big trucks. Some of them even include a 'blind spot' cam facing the rear on the passenger side. It takes a lot of the ambiguity out of a crash investigation.