Self driving buses, trains, trucks etc

eeguy

plodder wrote: »

I don't think we'll have pilot-less planes any time soon, but one of the reactions to that incident is interesting. The FAA has decreed that planes in the exact same situation in future, must be be landed using the automation rather than manually, which is one more situation where the computers are taking over and humans have to take a back seat.

Planes used to have 3 people in the cockpit.

They now have 2 as the computer does navigation. Soon they'll have one and they'll be little more than busses.

eeguy

Del.Monte wrote: »

I'm intrigued by this whole idea - but riddle me this, surely the changeover to self driving vehicles would have to be done overnight (literally on one night) not phased otherwise the potential for ordinary vehicles to create mayhem while mixing with self drives is mind boggling, or am I just being really thick?

A self driving car is well aware of its surroundings. Otherwise we'd have to microchip every pedi cyclist dog and cat.

Avoiding a car is quite simple now. The problem is seeing a pedi disappear behind a car and predict that he'll pop out on the road in 5 seconds. And AFAIK that's been solved

dr.fuzzenstein

plodder wrote: »

Yeah, maybe though I think fully driverless will be a long way off. This idea sounds quite practical though. I'd imagine that any truck with the technology would be able to join convoys like this randomly on the m-way, ]b]though there would have to be some way to compensate the vehicle at the front for providing the "service" but not directly benefiting.[/b] Keeping trucks together in tight groups of three could help congestion to some extent, as well as fuel economy.

I hope not. Since these mini "trains" would break up and regroup all the time, how would that he calculated? Truck a leads for 20 km and gets a fiver from the trucks behind, except truck c that left after 10 km. Trick d joined the group as lead vehicle and truck a left, but followed d for 5 km, the group breaks up at a major junction and goes separate ways and join different convoys.
What about cars? Same applies.
As with any simple concept, there's always ways to make it needlessly difficult and complicated because someone feels they are owed a few measly bucks. Or maybe it could be integrated with the electronic tolls that are applied to every meter a vehicle drives in the future, all it takes is a computer system that follows the movement if every single vehicle on the road.
I wonder how self driving will work on country boreens.

magicbastarder

dr.fuzzenstein wrote: »

I wonder how self driving will work on country boreens.

possibly easier for them, less likely to be other traffic, it's a simpler environment to model.

magicbastarder

magicbastarder wrote: »

possibly easier for them, less likely to be other traffic, it's a simpler environment to model.

Particularly when the car knows the exact road layout and exactly where it is at all times in relation to the road layout.

plodder

dr.fuzzenstein wrote: »

I hope not. Since these mini "trains" would break up and regroup all the time, how would that he calculated? Truck a leads for 20 km and gets a fiver from the trucks behind, except truck c that left after 10 km. Trick d joined the group as lead vehicle and truck a left, but followed d for 5 km, the group breaks up at a major junction and goes separate ways and join different convoys.

I don't see an issue with any of that. For each km, the leader leads the convoy he gets paid X. For each km (or part thereof ) the followers are in it, they pay Y. By the way, I'm not saying this is how it will pan out, but just one way it might. It could be a system of payment and credits for being the leader, or a multitude of different ways.

Billing would be the easy part of this. Everyone has to sign up in advance, obviously, since you need assurance about technical compliance etc

What about cars? Same applies.

The idea makes most sense at this stage for trucks, but I'd see it extended to cars as well eventually on the same basis.

As with any simple concept, there's always ways to make it needlessly difficult and complicated because someone feels they are owed a few measly bucks. Or maybe it could be integrated with the electronic tolls that are applied to every meter a vehicle drives in the future, all it takes is a computer system that follows the movement if every single vehicle on the road.

If people want this, who is going to pay for it otherwise?

I wonder how self driving will work on country boreens.

I noticed in the Nissan clip earlier, the car wasn't able to overtake on a single carriage way. The driver had to take over manual control when it passed a stopped vehicle. I'd say country boreens could be problematic, particularly Irish ones that aren't wide enough for two cars to pass.

eeguy

plodder wrote: »

I noticed in the Nissan clip earlier, the car wasn't able to overtake on a single carriage way. The driver had to take over manual control when it passed a stopped vehicle. I'd say country boreens could be problematic, particularly Irish ones that aren't wide enough for two cars to pass.

Wonder how they'll program having to reverse into a field to let another car go by:D

dr.fuzzenstein

plodder wrote: »

I don't see an issue with any of that. For each km, the leader leads the convoy he gets paid X. For each km (or part thereof ) the followers are in it, they pay Y. By the way, I'm not saying this is how it will pan out, but just one way it might. It could be a system of payment and credits for being the leader, or a multitude of different ways.

Billing would be the easy part of this. Everyone has to sign up in advance, obviously, since you need assurance about technical compliance etc

The idea makes most sense at this stage for trucks, but I'd see it extended to cars as well eventually on the same basis.
.

To my mind it's not only a solution looking for a problem, but a solution looking to create far more problems than it's worth.
If it does happen it will be a tribute to the ability of humanity not being happy with a situation that works and having to "improve" into one that doesn't.
OK, just my opinion and we'll see what comes to pass.
It is a bit like truck tolls in Germany. A complicated and sophisticated system of overhead gantries that are networked into a massively powerful computer system to track Tue movement of all trucks (and soon cars) that took years to implemented and cost countless billions, so no truck could drive one centimeter and not pay.
The Swiss and Austrians sell you a paper disk. The argument is that some people manage to get away with not paying, but is it worth billions to save millions? The fact that the answer is yes, says a lot about humankind.

plodder

dr.fuzzenstein wrote: »

To my mind it's not only a solution looking for a problem, but a solution looking to create far more problems than it's worth.

I'm not clear what aspect of the solution you don't like. Is it autonomous driving generally, truck convoys, or this idea of how it might be charged for/funded?

If it does happen it will be a tribute to the ability of humanity not being happy with a situation that works and having to "improve" into one that doesn't.
OK, just my opinion and we'll see what comes to pass.
It is a bit like truck tolls in Germany. A complicated and sophisticated system of overhead gantries that are networked into a massively powerful computer system to track Tue movement of all trucks (and soon cars) that took years to implemented and cost countless billions, so no truck could drive one centimeter and not pay.
The Swiss and Austrians sell you a paper disk. The argument is that some people manage to get away with not paying, but is it worth billions to save millions? The fact that the answer is yes, says a lot about humankind.

I've been caught out by those vignette systems in the past. You end paying the same motorway tax if you drive in the country for one day, or for 365 consecutive days. A different issue though ...

But if you think that this needs a system of overhead gantries every few km, or a massively powerful computer system, like the truck monitoring system you describe, then that's not the case. The state isn't going to be involved in developing these systems either; regulating them maybe, but definitely not building them. All the complicated stuff with this is built into the truck. They would use today's regular mobile networks for billing purposes. They don't need 100% coverage (or any better than today).

seamus

TheQuietFella wrote: »

Completely agree with this and that would be my concern also!

A lot of the companies developing this stuff came to that realisation early.

Google has been working on full autonomy and been ignoring safety-assist features because they realised that the more assistance you give to human drivers, the more likely they were to make a mistake - they became more complacent, spend less time focussing on the road.

This creates a form of negative feedback loop, where you have to keep increasing the number of safety features and improving their sensitivity, just to stand still. You hit a plateau of safety even though you're continually adding safety features. The more safety features you add, the worse drivers people become.

If you remove the human from the equation at the start, then the effort required to make the vehicle safer and get straight to full autonomy was way less than trying to transition from fully manual to semi-autonomy to full autonomy.

magicbastarder wrote: »

something you get from experience of driving - that spidey sense that car X is not to be trusted, there's just something about the driving you don't like, which is often confirmed when they prove they're distracted and do something stupid; detecting that is where these systems earn their money.

We pat ourselves far too much on the back for driving. We tell ourselves that it's a massively complex activity requiring unique intelligence that only a human can do.

In reality it's not. It's a by-the-numbers activity. Human stupidity adds the extra complexity to it, and we convince ourselves that you need human brain power to account for this.
But if you follow a short and simple set of rules, you'll will navigate 99.9999% of journeys without incident. The problem is that humans refuse to follow these rules and improvise when they encounter the unexpected.

My favourite example of this "simple rules for unexpected situations" idea is when Google's Self-Driving car encountered a woman in a wheelchair using a broom to chase a duck in the middle of the road:
https://www.theguardian.com/technology/video/2017/mar/16/google-waymo-self-driving-car-video-woman-bird

Without intervention, the car knew what to do and handled the scenario. I can think of plenty of people who would have done something dumb and ended up killing the bird or hitting a parked car.

Bear in mind that although the Guardian article above is only a couple of months old, the incident itself happened 3 years ago. So we're now 3 years beyond that level of quality in autonomous driving.

People just don't seem to get that autonomous vehicles are already better than people at driving for the most part. They seem to be under the impression that it's still pipe dream of people who were fans of Knight Rider, and that we're no further on than we were in the 1980s.

Dravokivich

eeguy wrote: »

Hype? You're looking at as vid from 2 years back showing tech that's not currently available.

There's a weird tendency for people to say " sure it's not perfect now so it'll never be perfect"

Here's one from this year of a Tesla casually driving some lad to work
https://youtu.be/6Vsi1Glf__A

Looks like it entered the car park and let him out on the wrong side.

magicbastarder

to be fair, you pulled that quote out of the context it was made in - it was a comment on how easy it is for a car to park itself in a well ordered car park, vs the greater demands of the open road.

MGWR

seamus wrote: »

My favourite example of this "simple rules for unexpected situations" idea is when Google's Self-Driving car encountered a woman in a wheelchair using a broom to chase a duck in the middle of the road:
Guardian link

Heh, the Guardian. They change their online headlines just a little too quickly for my comfort; that bespeaks mendacity, never mind their political stance. And that's to say nothing about the trustworthiness (or lack thereof) of Google itself.

seamus wrote: »

People just don't seem to get that autonomous vehicles are already better than people at driving for the most part. They seem to be under the impression that it's still pipe dream of people who were fans of Knight Rider, and that we're no further on than we were in the 1980s

Sure, because the left wing press and government say so.

It's a great way to make one stay home. So long as home isn't too automated itself.

As for KITT, this stuff that is being pushed on us is anything but something like that.

eeguy

MGWR wrote: »

Heh, the Guardian. They change their online headlines just a little too quickly for my comfort; that bespeaks mendacity, never mind their political stance. And that's to say nothing about the trustworthiness (or lack thereof) of Google itself.Sure, because the left wing press and government say so.

It's a great way to make one stay home. So long as home isn't too automated itself.

As for KITT, this stuff that is being pushed on us is anything but something like that.

ALL online news change as the story develops. It's a huge peeve of mine, but no one is free of guilt on that front.
What does liberal or government have to do with private technology companies developing technology?
You should take your tinfoil hat off for a day.

prinzeugen

People dont have a problem with companies developing technology. Its when this technology (most of the recent stuff being solutions to problems that dont exist) gets forced on people that it becomes a problem.

I cant see there ever being fully autonomous cars etc.

You still need humans. You just have to watch a few episodes of Air Crash Investigation to see that.

Even the "driverless" Docklands railway and 3 Tube lines still have human drivers.

What about sort notice lane closures etc due to accidents? Is one of these driverless trucks going to plow into fire crews while they cut someone out a car?

Stephen15

prinzeugen wrote: »

What about sort notice lane closures etc due to accidents? Is one of these driverless trucks going to plow into fire crews while they cut someone out a car?

But the driverless cars will be so safe that fire crews will never have to cut someone out of a car as there will be no accidents involving driverless cars

eeguy

prinzeugen wrote: »

People dont have a problem with companies developing technology. Its when this technology (most of the recent stuff being solutions to problems that dont exist) gets forced on people that it becomes a problem.

I cant see there ever being fully autonomous cars etc.

You still need humans. You just have to watch a few episodes of Air Crash Investigation to see that.

Even the "driverless" Docklands railway and 3 Tube lines still have human drivers.

What about sort notice lane closures etc due to accidents? Is one of these driverless trucks going to plow into fire crews while they cut someone out a car?

If you did any research you'd know all of your concerns were baseless.
Also air crash investigates usually has the cause of pilot error.

Stephen15

eeguy wrote: »

If you did any research you'd know all of your concerns were baseless.
Also air crash investigates usually has the cause of pilot error.

What about what happened in 2010 on the Hudson river would a driverless plane have the intelligence or even common sense to land a plane in those conditions or even crosswinds for example. How will driverless vehicles adapt to changes in weather conditions which can change suddenly.

eeguy

Stephen15 wrote: »

What about what happened in 2010 on the Hudson river would a driverless plane have the intelligence or even common sense to land a plane in those conditions or even crosswinds for example. How will driverless vehicles adapt to changes in weather conditions which can change suddenly.

Would all 216 passengers on the Air France flight be alive if an autopilot was installed on the plane instead of the pilot?

Picking one freak example as an argument against autopilot is false. Pilot error is causes something like 3/4 of plane crashes. I'd hazard that driver error is probably much higher.

Weather conditions change suddenly? Like it rains? Autonomous cars can see far better than humans. Most weather doesn't even affect them.

prinzeugen

eeguy wrote: »

If you did any research you'd know all of your concerns were baseless.
Also air crash investigates usually has the cause of pilot error.

Very few.

You just have to look at the 100s of "miracles" that have been preformed by commercial aviation pilots over the years.

You can never program a computer with a humans knowledge or 1000's of hours experience in the air.

I recall one where the both artificial horizons went. Human solution? Glass of water with a line on it so they knew if they were going up/down or left/right.

AI would not think of that never mind be able to do it.

The guy that went on a rampage outside Buckingham palace had planned to go to Kensington palace but google maps directed him to a pub with the same name..

Not promising is it.

fly_agaric

Stephen15 wrote: »

How will driverless vehicles adapt to changes in weather conditions which can change suddenly...

Unfortunately there's many humans that don't adapt too well. They may carry on driving when perhaps they should have stopped and cause an accident!
The ai just has to be as good as a human (well, realistically it has to be measurably better imo to allay fears). It does not have to be infallible.

Once it gets there, the days of human controlled vehicles on the roads will be numbered. Once the human driven vehicles are gone (...a long way off yet) things should get safer as the environment will be somewhat more predictable for the ais (dealing with just counterpart ai controlling other vehicles, + the like of pedestrians, cyclists etc instead of all of this + some crazy human drivers as well).

May be dismissed as propaganda or advertorial for Google/Alphabet but thought this was interesting so will leave it here (assuming was not posted by someone already as did not read entire thread, if it was apologies).

https://www.theatlantic.com/technology/archive/2017/08/inside-waymos-secret-testing-and-simulation-facilities/537648/

edit: There seems to be vast amounts of money, time & intellect going into efforts to crack this. Our "corporate overlords" (many of of the biggest, wealthiest companies on earth) have a hard-on to make this work and I would not bet against them succeeding.

prinzeugen

eeguy wrote: »

Would all 216 passengers on the Air France flight be alive if an autopilot was installed on the plane instead of the pilot?

Picking one freak example as an argument against autopilot is false. Pilot error is causes something like 3/4 of plane crashes. I'd hazard that driver error is probably much higher.

Weather conditions change suddenly? Like it rains? Autonomous cars can see far better than humans. Most weather doesn't even affect them.

The big problem is that pilots "pilot" an aircraft now rather than "fly" it.

THE pilot of the Air France jet which plunged into the Atlantic in 2009 killing 228 people – including three Irish doctors - pushed the nose upward instead of downward during a stall because of false data from sensors

Over reliance on technology and flying skills being lost is a contributing factor in some accidents.

eeguy

prinzeugen wrote: »

The big problem is that pilots "pilot" an aircraft now rather than "fly" it.
Over reliance on technology and flying skills being lost is a contributing factor in some accidents.

"Neither weather nor malfunction doomed AF447, nor a complex chain of error, but a simple but persistent mistake on the part of one of the pilots."
http://www.popularmechanics.com/flight/a3115/what-really-happened-aboard-air-france-447-6611877/

Modern airplanes basically couldn't fly without all this technology. To take it away would kill our ability to fly.
You can argue against it all you want but the fact is that people consistently make mistakes, and usually the same basic mistakes over and over.

Machines may make a mistake once, usually in a controlled simulated environment and they won't do it again.

Stephen15

Thinking of an example a little closer to home remember the 2004 Wellington Quay bus crash where a 5 people were killed in when bus ploughed into pedestrians waiting a stop well it was proven in a court of law that it was caused by a power surge in the bus in question and the driver was proven not guilty. I would hate to what would happen if it was an autonomous and how would a driverless would react to a power surge.

devnull

prinzeugen wrote: »

Over reliance on technology and flying skills being lost is a contributing factor in some accidents.

The number one cause of airplane disasters is pilot error by quite some distance and most of them are controlled flights into terrain where the pilot has made a mistake, is not paying attention or is suffering from spatial disorientation or lack of situational awareness and has flown a perfectly serviceable aircraft into terrain.

In a very high number of these cases that have been investigated the flight data recorders and cockpit voice recorders have shown that the pilot or the copilot were given correct information by their systems and their instruments were working properly but they disregarded the information these systems were giving them or were not correctly monitoring it with fatal consequences.

THE pilot of the Air France jet which plunged into the Atlantic in 2009 killing 228 people – including three Irish doctors - pushed the nose upward instead of downward during a stall because of false data from sensors

Over reliance on technology and flying skills being lost is a contributing factor in some accidents.

What the report actually said

- temporary inconsistency between the measured speeds, likely as a result of the obstruction of the pitot tubes by ice crystals, causing autopilot disconnection and reconfiguration to alternate law;
- the crew made inappropriate control inputs that destabilized the flight path;
- the crew failed to follow appropriate procedure for loss of displayed airspeed information;
- the crew were late in identifying and correcting the deviation from the flight path;
- the crew lacked understanding of the approach to stall;
- the crew failed to recognize that the aircraft had stalled and consequently did not make inputs that would have made it possible to recover from the stall

I'm not saying that malfunctions do not happen, but at the end of the day the failure was not so big that it doomed the plane or anything near it, it was something that should have been recoverable but a catalog of wrong decisions by the crew ultimately doomed them and all on board.

magicbastarder

prinzeugen wrote: »

You still need humans. You just have to watch a few episodes of Air Crash Investigation to see that.

i keep hearing this style of defence. it's a complete and utter false equivalence. any crashes mentioned on that show are not a failure of a system designed to be fully autonomous. you are condemning a technology which was not a factor in any of those incidents

and the funny thing is, in probably the majority of those programs, the plane crashes with multiple or total fatalities. and it's regularly human error to blame.

an aside; the episode i found most interesting of air crash investigation was one where a pilot who was trained in the eastern bloc, on planes where the artificial horizon works in the opposite way to western bloc planes. in short, human error.

prinzeugen

eeguy wrote: »

"Neither weather nor malfunction doomed AF447, nor a complex chain of error, but a simple but persistent mistake on the part of one of the pilots."
http://www.popularmechanics.com/flight/a3115/what-really-happened-aboard-air-france-447-6611877/

Modern airplanes basically couldn't fly without all this technology. To take it away would kill our ability to fly.
You can argue against it all you want but the fact is that people consistently make mistakes, and usually the same basic mistakes over and over.

Machines may make a mistake once, usually in a controlled simulated environment and they won't do it again.

As I said, Over reliance on technology, not pilot error was to blame.

Last bit from the link you posted

But the crash raises the disturbing possibility that aviation may well long be plagued by a subtler menace, one that ironically springs from the never-ending quest to make flying safer. Over the decades, airliners have been built with increasingly automated flight-control functions. These have the potential to remove a great deal of uncertainty and danger from aviation. But they also remove important information from the attention of the flight crew. While the airplane's avionics track crucial parameters such as location, speed, and heading, the human beings can pay attention to something else. But when trouble suddenly springs up and the computer decides that it can no longer cope—on a dark night, perhaps, in turbulence, far from land—the humans might find themselves with a very incomplete notion of what's going on. They'll wonder: What instruments are reliable, and which can't be trusted? What's the most pressing threat? What's going on? Unfortunately, the vast majority of pilots will have little experience in finding the answers.

Also it was a french plane built by a french company. They have a history of altering facts to protect themselves and french industry.

One airline is starting to train its pilots on how to use sextants again for navigation at night. Using centuries old technology says tons about the confidence they have in modern stuff.

devnull

prinzeugen wrote: »

As I said, Over reliance on technology, not pilot error was to blame.

That's not what the official report says though, that is your opinion but it is not one that I would share and not one that the BEA would agree with either and they know far more about these things than any of us I think.

One airline is starting to train its pilots on how to use sextants again for navigation at night. Using centuries old technology says tons about the confidence they have in modern stuff.

Pilots are also trained in what to do in emergency landings or how to land in water, does it mean that they have no confidence in their planes and expect to be landing in water?

Ben D Bus

prinzeugen wrote: »

The big problem is that pilots "pilot" an aircraft now rather than "fly" it.



Over reliance on technology and flying skills being lost is a contributing factor in some accidents.

If this is your case then you need to show that increasing automation has reduced flight safety. Not by anecdote, but by cold hard statistics.

Honestly, most of the "it'll never happen" posts on this thread seem driven purely by fear of change and not by any real evidence.

Breakdown of air accidents by cause: (59% human error)
http://www.independent.co.uk/travel/news-and-advice/the-one-chart-that-shows-what-causes-fatal-plane-crashes-10494952.html

prinzeugen

devnull wrote: »

That's not what the official report says though, that is your opinion but it is not one that I would share and not one that the BEA would agree with either and they know far more about these things than any of us I think.



Pilots are also trained in what to do in emergency landings or how to land in water, does it mean that they have no confidence in their planes and expect to be landing in water?

Navigation goes over water. They end up going round in circles untill they run out of fuel.

A sextant would give them an idea of position. It was in one of last months aviation magazines (sorry I am one of the people that use Easons as a library. I can not remember which mag as I skimmed through a few)