Seems to be an awful lot of hysteria about about electric cars rebooting mid journey causing untold amounts of carnage and devastation :-)
Commercial aircraft have been using fly by wire systems for decades now, and I'm not aware of too many incidents where the aircraft has crashed mid flight due to a BSOD. The engines are FADEC too meaning they run completely by themselves using their own inbuilt software logic independent from the aircraft avionics, and they tend to run with impeccable reliability.
Whilst I appreciate that automotive is not quite designed to the same standard as aerospace, I see no reason why an electric cars control systems cant be completely sound given enough development. Personally id much prefer it if my car could be fixed by a software update, rather than having to pull half the front out say to replace a clutch.
I wouldn't use fly-by-wire aircraft as a 'safe example', at l;east in the early years of adopting the tech. I'm sure I'm not the only person who remembers the fatal crash of one of the first FBW airbuses at a demonstration doing a low flyby. I also recall many aircraft since (including Boeing) having serious recalls for the electronics and wiring, grounding many for long periods in order to get the changes made.
It's why I fear things are progressing too quickly on the self-drive cars, whwre the working environment is far more varied/complex that aircraft.
My 16yo car experienced its first and (thus far) only recall last year, for a defect in the airbag system. Many software glitches could at one point or another result in a catastrophic error and lead to an accident in a far more 'computer driven' car, whether today's ICE/hybrids or EVs.
So many interdependent systems interracting all the time and seemingly not fully tested before release - in my view because they want to be 'the first' (or not the last) to adopt the new system.
|
Ah yes you would be referring to this, back in 1988, per chance? Air France Flight 296Q - Wikipedia
This was a demonstration flight, where the aircraft was pushed beyond the limits of its normal operating conditions. Airbus blamed the pilot, the pilot blamed the computer (as of course, one doesn't want to carry the can for smashing up a brand new, and very expensive, aircraft), and the whole lot ended up finger pointing.
I was thinking more in lines of the flight control computers in my post rather than the operation of the autopilot which is perhaps analogous to what you mean when you mention self driving. Aircraft flight controls are several times redundant, so if one computer throws up a red flag, the remainder 'vote it out' and everything carries on uninterrupted. Perhaps cars need to be done the same way.
Another thing of note is that aircraft use transponders to broadcast their position, altitude, speed and direction etc, so that all other aircraft in the area know where their neighbours are at and where they're going, helping the autopilot navigate around them. I think for self driving cars, something similar would be needed. I don't think a few cameras dotted around the car relying on visual feedback is good enough. Whether this would prove popular is another matter entirely.
Anyways the crux of my point is that fly/drive by wire system's can be extremely reliable if done right, which leads to,
So many interdependent systems interracting all the time and seemingly not fully tested before release - in my view because they want to be 'the first' (or not the last) to adopt the new system.
As I alluded to before, the bar is much lower for automotive manufacturers, and perhaps more stringent regulation is required to raise it. I certainly don't trust 'self driving' for the time being,
|
I don't think a few cameras dotted around the car relying on visual feedback is good enough.
they don`t just use cameras though, they use, apart from GPS which is too slow, mobile phone masts and Wi-Fi from any place they can to triangulate to nearest metre or so.. they get a better idea of where they are than using cameras, but they can if thats all there is to use ie no GPS or Wi-Fi
|
I don't think a few cameras dotted around the car relying on visual feedback is good enough.
Eh ? :-)
|
The soft squidgy thing behind the wheel only has visual feedback as it lacks radar, heat sensors, and an internet connection streaming data on other hazards in the area..
No great surprise it gets things wrong so often!
|
Indeed, many an aircraft has been lost by CFIT due to poor visibility, hence why I think using cameras on a vehicle as the primary means of navigating for the self driving system is insufficient. Good driving conditions cant be guaranteed all the time.
For self driving cars to work properly, IMO its not enough for an individual vehicle to simply know its own position and heading, it needs to know the same of all the vehicles around it in order to navigate around them. Basically they will need to intercommunicate, whether this means each vehicle having its own transponder or something similar. For this reason, I think true self driving is a long way off yet, as it requires changes to all cars on the road, not just a few new models at a time.
|
Indeed, many an aircraft has been lost by CFIT due to poor visibility, hence why I think using cameras on a vehicle as the primary means of navigating for the self driving system is insufficient. Good driving conditions cant be guaranteed all the time.
For self driving cars to work properly, IMO its not enough for an individual vehicle to simply know its own position and heading, it needs to know the same of all the vehicles around it in order to navigate around them. Basically they will need to intercommunicate, whether this means each vehicle having its own transponder or something similar. For this reason, I think true self driving is a long way off yet, as it requires changes to all cars on the road, not just a few new models at a time.
In your utopia, will pedestrians and cyclists be required to fit a transponder - where will it be fitted?
|
In your case, I know exactly where id put it ;-)
|
|
|
The soft squidgy thing behind the wheel only has visual feedback as it lacks radar, heat sensors, and an internet connection streaming data on other hazards in the area.
I don't know about you, Terry - I'll admit to having no radar or internet connection, but my heat sensors aren't too bad :-)
|
|
The soft squidgy thing behind the wheel only has visual feedback as it lacks radar, heat sensors, and an internet connection streaming data on other hazards in the area..
No great surprise it gets things wrong so often!
I think you give self-drive cars too much credit and not enough for people, given the former has yet to get anyweher near mastering what is required in most of the world. The difference is that humans make mistakes because they are lazy or careless, whereas the machine will make mistakes (other than being poorly designed/constructed) because it doesn't know any better.
We aren't anywher near the utopian (or maybe that dystopian) situation regularly depicted in scifi films set in the supposed not-to-distant future. I would also note that many of them predict safe self driving cars for 20XX and thus far none have come remotely true, despite many scientists saying they would years ago when pontificating on Tommorrow's World and the like.
|
Do you think your self driving car could avoid driving into a flood or aquaplaning on a motorway. Could it avoid skidding on black ice or successfully negotiate a snow covered road( where have the white lines gone or why are they suddenly so wide) Can it see perhaps 400yards ahead and foresee a possible problem that might be avoided Can it tell a Sainburys plastic bag from a Morrisons bag that has caused it to ram on the anchors? No. If you want someone else to do the driving take the train
|
Do you think your self driving car could avoid driving into a flood or aquaplaning on a motorway. Could it avoid skidding on black ice or successfully negotiate a snow covered road( where have the white lines gone or why are they suddenly so wide)
It might avoid aquaplaning or black ice by detecting the change in conditions before many human drivers. But I certainly wouldn't expect it to be anywhere near foolproof.
|
|
Do you think your self driving car could avoid driving into a flood or aquaplaning on a motorway. Could it avoid skidding on black ice or successfully negotiate a snow covered road( where have the white lines gone or why are they suddenly so wide) Can it see perhaps 400yards ahead and foresee a possible problem that might be avoided Can it tell a Sainburys plastic bag from a Morrisons bag that has caused it to ram on the anchors? No. If you want someone else to do the driving take the train
It could avoid driving into a flood, it could avoid aquaplaning, it could avoid skidding, it could negotiate a snow covered road, it could see further up the road than you, and I'm sure it could tell the difference between a Morrisons or a Sainsburys bag. It's all things that a self driving car needs to be able to do to be able to self drive.
I will ask you though - can you avoid a flood, avoid aqua planing or avoid skidding on black ice?
|
Do you think your self driving car could avoid driving into a flood or aquaplaning on a motorway. Could it avoid skidding on black ice or successfully negotiate a snow covered road( where have the white lines gone or why are they suddenly so wide) Can it see perhaps 400yards ahead and foresee a possible problem that might be avoided Can it tell a Sainburys plastic bag from a Morrisons bag that has caused it to ram on the anchors? No. If you want someone else to do the driving take the train
It could avoid driving into a flood, it could avoid aquaplaning, it could avoid skidding, it could negotiate a snow covered road, it could see further up the road than you, and I'm sure it could tell the difference between a Morrisons or a Sainsburys bag. It's all things that a self driving car needs to be able to do to be able to self drive.
I will ask you though - can you avoid a flood, avoid aqua planing or avoid skidding on black ice?
Some car companies are using the film already made of road trips/and still gathering videos- all over the countries they are in, idea is, they can write software in order that AI and the computers will know any situation they get into and what to do to avoid problems, it will take years to perfect but it is getting better as time goes on...
|
<< Idea is, they can write software in order that AI and the computers will know any situation they get into and what to do to avoid problems, it will take years to perfect but it is getting better as time goes on. >>
No doubt what you say will eventually come, but I still wonder whether, in the overall scheme of things, it will be much more than a glorious vanity project.
|
but I still wonder whether, in the overall scheme of things, it will be much more than a glorious vanity project.
wouldn`t surprise me as I think Google have stopped its R&D into it, though alphabet, googles parent company, are still researching, but for how long is anyone's guess
Biggest progress is being made by Tesla so far....
|
It's abundantly clear that many human beings are unable to consistently identify flooded roads, black ice, missing road markings, potholes, kerbs etc etc.
All these will be easily managed by a properly programmed autonomous vehicle equipped with the right sensors.
Even though that may be some time in coming, bear in mind that they only need be better than flesh and blood, not infallible. It may be sooner than you think.
|
All these will be easily managed by a properly programmed autonomous vehicle equipped with the right sensors.
All this theorising forgets about the adaptability of humans (which is what the whole idea is about). Drivers who have learnt their limitations, and those of their car (some maybe haven't), may drive near those limits. If they change to a car with different limitations some of them adapt their styles to suit. How will those drivers adapt to a vehicle which they expect to avoid any difficulties ?
A 'smart' car may be very good at detecting surface water or black ice, but that won't give the vehicle any magical way of sticking to the road.
|
In a truly autonomous vehicle there will be no driver, and their limitations will be completely irrelevant.
A smart car (one assumes) will react when it detects surface water or black ice, unlike the numpty behind the wheel who won't even know it's there until too late.
In fact it is all but a certainty there will be some who having skidded on black ice will get out of the car to inspect the damage and wonder why they fall over!!
|
A 'smart' car may be very good at detecting surface water or black ice, but that won't give the vehicle any magical way of sticking to the road.
Detecting deep water and black ice maybe the things that will be a problem for sensors, unless the programming can tell the AI the difference between a dry road and wet/icy road which imo may be difficult to do.
|
Sensors will certainly be able to detect road surface temperatures, and possibly standing water either by appearance, reflectivity or sonar.
It will be very aware of external temperatures, humidity, and with a data feed, accumulated rainfall, forecast weather, etc.
Strangely I think the problem autonomous vehicles may have compared to human drivers is the utterly trivial - eg:
- put in a postcode, at the destination does it park in the drive or the road,
- how does it park if you need to put a large load in the boot
- at an event or camp site will it respond to a traffic marshal
- etc etc
There may be an argument for simple joystick control below say 5 mph.
|
An autonomous car will struggle on single track roads.
Anyone who has had to go onto the verge to allow another car past will know why.
|
An autonomous car will struggle on single track roads.
Anyone who has had to go onto the verge to allow another car past will know why.
The donkey equidistance from two identical food sources argument - each AI won't know what to do or who to give priority to, unless that is they are linked via some kind of wireless system.
I keep think about that scene in the film I, Robot where Will Smith's character is being attacked by killer robots whilst driving: "You are having an accident" they say (or words to that effect).
Either that or visions of the HAL 9000 going bonkers... :-)
|
How do two human drivers resolve the problem of who moves first at (say) a mini roundabout.
Both move forward one stops, the other stops, move forward again, stop, repeat, repeat, one then usually remains stationary whilst the other moves off.
It is a kind of random chaos which usually resolves itself - but occasionally leads to bent metal or plastic. Autonomous systems could hardly do worse. Perhaps there will be a default setting as for ships at sea and aircraft - eg: vehicle on a northern setting gets priority!!
|
How do two human drivers resolve the problem of who moves first at (say) a mini roundabout.
Both move forward one stops, the other stops, move forward again, stop, repeat, repeat, one then usually remains stationary whilst the other moves off.
It is a kind of random chaos which usually resolves itself - but occasionally leads to bent metal or plastic. Autonomous systems could hardly do worse. Perhaps there will be a default setting as for ships at sea and aircraft - eg: vehicle on a northern setting gets priority!!
There are a couple of narrow bridges I cross regularly during the course of my work, one direction of travel has priority over the other. Doesn't cause any problems AFAIK.
|
I saw this photograph trending, but assume it's a fake. Generator probably wouldn't be up to the job?
scontent-lcy1-2.xx.fbcdn.net/v/t39.30808-6/2754982...7
Edited by De Sisti on 09/03/2022 at 09:18
|
I saw this photograph trending, but assume it's a fake. Generator probably wouldn't be up to the job?
scontent-lcy1-2.xx.fbcdn.net/v/t39.30808-6/2754982...7
As far as I am aware, I Tesla can be charged on a household plug, its just that it will take about 3 days from empty to full!. So the question wouldn't be whether the genny was up to the job, but what the circumstances were. They may only need enough juice to get half a mile to a charger?
|
I saw this photograph trending, but assume it's a fake. Generator probably wouldn't be up to the job?
scontent-lcy1-2.xx.fbcdn.net/v/t39.30808-6/2754982...7
As far as I am aware, I Tesla can be charged on a household plug, its just that it will take about 3 days from empty to full!. So the question wouldn't be whether the genny was up to the job, but what the circumstances were. They may only need enough juice to get half a mile to a charger?
What would be useful would be a portable power pack (c/w universal connectors) to give just enough power to reach the nearest charge point in an emergency when going on a longer journey. I know such a device exists for 12V ICE car batteries, not sure about for EVs as it's not used in the same way.
|
You can buy a portable power pack for EV’s.
There was a conversation on here about them a little while ago - they’re about the same size as a cabin baggage sized suitcase.
I’ve not seen an EV that won’t charge from a 3 pin plug, but it’s only going to charge at about 2.5kW. And, yes, it takes days rather than hours to go from discharged to ‘full’…
|
|
|
|
|
|