Electric Cars

[Michael GilliganParticipant @michaelgilligan61133]

Posted by pgk pgk on 14/07/2019 06:01:32:

I think the last few posters are missing the point of computer AI.

1) In the case of issues at the traffic lights the simplified options were to race ahead or to wait and take a gap behind. Racing ahead worked this time (and my EV could have beaten the lot off the lights if I'd wanted to) but was based in male agression and risked causing a road rage incident. Doubtless AI would have learned that it's statistically better to arrive later and less often undamaged than take those risks. Indeed AI might have learned that such extra lanes are better not used and just queued to being with… patiently

 

 

.

Apologies if I am 'missing the point' … but I described briefly [and only from my perspective] a real incident, that occurred at a real, recently built, road junction.

• If I had braked firmly to let the van have his way, someone may have piled into the back of me [there was no-one behind me when we left the lights, but … who knows?]
• Yes, I could [and probably should] have simply taken third position in an orderly queue; but that would have meant not using the road as it was clearly intended
• I was in no hurry, and had nothing to 'prove' [I was driving my Wife to a meeting]
• As for 'male aggression' … I don't know: There was no aggression on my part, it was just a 'survival' choice, and I didn't have the time or inclination to check the gender of the other two drivers

In the future, roads may be designed so as to preclude such incidents; and the vehicles may all have appropriate algorithms to preserve themselves and their passengers … but during the 'implementation period' things will remain complex on our roads.

I was trying to give an example for analysis and discussion; but evidently my first few words were the appropriate ones.

MichaelG.

Edited By Michael Gilligan on 14/07/2019 06:51:44

[Michael HornerParticipant @michaelhorner54327]

Posted by Michael Gilligan on 13/07/2019 23:30:27:

The exit lanes merge into one after a reasonably short distance [the implied rule being that the two lines of vehicles will take their places alternately].

MichaelG.

Hi Michael

You answered your own question.

This would be my take on it, unless they are going to program in male aggression and selfishness.

The cars would agree amongst themselves who would be first away from the lights.

Cheers Michael

[pgk pgkParticipant @pgkpgk17461]

Posted by Michael Gilligan on 14/07/2019 06:51:0

Apologies if I am 'missing the point' … but I described briefly [and only from my perspective] a real incident, that occurred at a real, recently built, road junction.

• If I had braked firmly to let the van have his way, someone may have piled into the back of me [there was no-one behind me when we left the lights, but … who knows?]
• Yes, I could [and probably should] have simply taken third position in an orderly queue; but that would have meant not using the road as it was clearly intended
• I was in no hurry, and had nothing to 'prove' [I was driving my Wife to a meeting]
• As for 'male aggression' … I don't know: There was no aggression on my part, it was just a 'survival' choice, and I didn't have the time or inclination to check the gender of the other two drivers

In the future, roads may be designed so as to preclude such incidents; and the vehicles may all have appropriate algorithms to preserve themselves and their passengers … but during the 'implementation period' things will remain complex on our roads.

I was trying to give an example for analysis and discussion; but evidently my first few words were the appropriate ones.

MichaelG.

Edited By Michael Gilligan on 14/07/2019 06:51:44

I've often wondered why they put that extra lane in. It serves two purposes, I suppose, that it shortens the length of queue at the lights and presumably was originally intended for ocassional heavy slow stuff to use nearside and allow nippy stuff to pass as the lights change. We've all experienced that two lane scenario coming up to a roundabout with an HGV on the inside and having to really keep a look-out that he has enough room for the curve of the roundabout without squeezing us…

If that lane is really intended as a neat and polite alternate car feed-in then it has little benefit to an orderly queue and you'd think it'd be documented in the highway code (haven't looked).

1)You say if you'd braked someone may have hit you from behind. Yes I accept that in a real emergecy stop there may not be time to check mirrors – at least an autonomous vehicle keeps that info updated at all times – but also one hopes/expects that any driver behind you would be traffic aware and paying attention to your situation. If the vehicle behind was autonomous then it'd respond to your braking faster than a human reaction.

2)We have some variance on the possible reasons for that lane. I'm not so clear on why.

3)I wasn't trying to suggest you had anything to 'prove' as you put it but evidently your interpretation of queue rules was different to the van drivers and you felt you had right of way.

4) Apologies if I implied that you were aggressive but you may concede that the whole scenario occurred because primarily the van driver was being pushy while you interpreted the situation as your right of way. Such episodes are the usual initiator of road rage episodes even when it's a simple error by one driver. Hence we should all drive defensively. If id been in that situation then honestly it would be a case of what mood I was in – again something that wouldn't affect AI. If I was in my slow car with poor brakes then I'd have been driving carefully and cautiously.

If i was in my EV with stunning acceleration and hot brakes and genuinely in a hurry or just feeling playful then i admit I might have floored it (my bad) but mostly I drive that with care 'cos it's so sodding expensive and repairs won't be cheap and age has slowed my reflexes – my excuse for buying a car that protects it's occupants that well.

[not done it yetParticipant @notdoneityet]

Questions:

How do auto-piloted vehicles react to a good covering of snow? Do they simply stop?

How about black ice? Can they see it? Do they attempt the impossible(like some drivers on u-tube videos where the vehicle simply accelerates backwards after losing adhesion attempting to drive up a slope?

Perhaps they will simply take to the air in future?

[pgk pgkParticipant @pgkpgk17461]

Posted by not done it yet on 14/07/2019 07:33:00:

Questions:

How do auto-piloted vehicles react to a good covering of snow? Do they simply stop?

How about black ice? Can they see it? Do they attempt the impossible(like some drivers on u-tube videos where the vehicle simply accelerates backwards after losing adhesion attempting to drive up a slope?

Perhaps they will simply take to the air in future?

At the current level of autonomy of my car it will stay in self-steer if it can determine where the road is and it does that buy assessing edges and markings and making some limited assumptions. I have used self-drive in lousy rainy night conditions and monitored it's display when it's cameras were clearly seeing better than i could but in complelety covered road markings and snow drifts or confusing (to it) road markings or no road markings etc then at the present state of AI it won't self drive.

How do you determine black ice? It's a combination of temperature, colour, reflectance and doubtless other factors AI will learn (really that means be taught by presenting it with multiple sets of data to sort out) what ators affect traction, cargo load and slope angle/camber. Your car at the moment has traction controls built in to assist.

Elon Musk claims that the new Roadster will be available with a spaceX package option to give rocket thrust for acceleration and directional control and possibly short flight. He does make these wild promises and often they come to pass – somewhat later that everyone hoped. The limits of road car acceleration is down to tyre grip; rockets byass that.

[Barnaby WildeParticipant @barnabywilde70941]

Posted by not done it yet on 14/07/2019 07:33:00:

Questions:

How do auto-piloted vehicles react to a good covering of snow? Do they simply stop?

How about black ice? Can they see it? Do they attempt the impossible(like some drivers on u-tube videos where the vehicle simply accelerates backwards after losing adhesion attempting to drive up a slope?

Perhaps they will simply take to the air in future?

A bunch of programmers can sit around a table & define the definitive response to every definitive situation, if X happens the vehicle will do this, if Y happens the vehicle does this, etc. Utilising the sensory capacity they have given the vehicle they can even programme the vehicle to recognise a developing situation i.e. one that is not yet dangerous but soon will be.

What they cannot do & I believe what they can never do is to enable the vehicle to make a decision in the way a human would. Yes, we get it wrong & yes some of us are incredibly poor drivers, but we have the unique ability to handle the exception to the rule.

Subconciously, we process an incredible amount of information, if I toss you a ball & you catch it, do you appreciate the complexity of the math calculation your brain has just processed inbetween your hand / eye co-ordination. You also made a decision on whether or not it was OK to catch the ball, you decided whether or not you actually want to play ball.

What the computer running a programme can NEVER do is to question itself whether it wants to play ball.

[pgk pgkParticipant @pgkpgk17461]

Posted by Barnaby Wilde on 14/07/2019 08:14:18:…

What the computer running a programme can NEVER do is to question itself whether it wants to play ball.

Why not? What are the determinants? Social pressures, time of the month, work-load, sickness, muscle strain? I accept people and animals are hugely complex and the task to create an imitation of our behavious is huge – should one ever actually want to do it (rather pointless when people are cheap to make and easily thrown away) – then i dare say a simulation could be developed…..

[J HancockParticipant @jhancock95746]

Call me 'old-fashioned' but I don't really want to participate in this futuristic world you have described.

Reads utterly boring , like being the Mekon, sitting on that flying saucer thing it used to move around on.

Nope, 1964 ish was nirvana, downhill ever since.

[SillyOldDufferModerator @sillyoldduffer]

Posted by Michael Gilligan on 14/07/2019 06:51:04:

Posted by pgk pgk on 14/07/2019 06:01:32:

I think the last few posters are missing the point of computer AI.

1) In the case of issues at the traffic lights the simplified options were to race ahead or to wait and take a gap behind. Racing ahead worked this time (and my EV could have beaten the lot off the lights if I'd wanted to) but was based in male agression and risked causing a road rage incident. Doubtless AI would have learned that it's statistically better to arrive later and less often undamaged than take those risks. Indeed AI might have learned that such extra lanes are better not used and just queued to being with… patiently

.

Apologies if I am 'missing the point' … but I described briefly [and only from my perspective] a real incident, that occurred at a real, recently built, road junction.

• If I had braked firmly to let the van have his way, someone may have piled into the back of me [there was no-one behind me when we left the lights, but … who knows?]
• Yes, I could [and probably should] have simply taken third position in an orderly queue; but that would have meant not using the road as it was clearly intended
• I was in no hurry, and had nothing to 'prove' [I was driving my Wife to a meeting]
• As for 'male aggression' … I don't know: There was no aggression on my part, it was just a 'survival' choice, and I didn't have the time or inclination to check the gender of the other two drivers

In the future, roads may be designed so as to preclude such incidents; and the vehicles may all have appropriate algorithms to preserve themselves and their passengers … but during the 'implementation period' things will remain complex on our roads.

I was trying to give an example for analysis and discussion; but evidently my first few words were the appropriate ones.

MichaelG.

…

It's very useful to explore scenarios like this and score what humans do against the autonomous vehicle. In the example, I think autonomous has the advantage in this particular example:

• It understands the rule of the road (merge) when real drivers may not, and would try to cooperate with other traffic.
• It has no emotional response; it won't tailgate because it's busting for a pee, or behave aggressively due to what happened 5 minutes ago, or deliberately ram a police car.
• It knows exactly what is in front, behind and to the sides with millisecond accuracy. It also knows where it is within a few metres, the temperature, risk of ice, and can detect almost instantly when the wheels slip for whatever cause. Humans are at least 0.2 seconds behind reality, and much worse if the check involves looking in a mirror, during which time they go blind in front. A sneeze loses about 2 seconds. Most drivers are bad on ice unless they've been trained, and even good ones react slowly.
• In court, after the accident, the human driver has to explain himself. He is an unreliable witness compared with the scrupulously accurate record kept by the autonomous vehicle. Quite likely I think, the autonomous vehicle's data would normally prove the human driver was at fault, much as Black Box recorders show most air crashes to be caused by pilot error. I don't think pleading innocence because the car's AI system can't resolve marriage disputes or analyse Flaubert will impress judges or insurance companies!

Note that, apart from outright pilot error being number one, pilot error appears again in 'Pilot Error (Weather Related) and 'Pilot Error (Mechanical related)

Here's one where a human might do better. You're driving at speed up an empty road where temporary traffic lights are protecting the movements of a digger just out-of-sight round the corner. The lights have failed. A human might spot the hazard by recognising the in-position lights are not Red, Green or Amber. He should slow down. As the car's map is wrong it has to rely on its sensors and might be travelling too fast to stop in time. But, as I noted earlier, once a hazard is detected, the car will react much faster than a human. We can rely on them to do super-human emergency stops. Maybe even in this situation human and car might score about the same.

On the other hand, autonomous cars wouldn't deliberately or accidentally jump lights as most people do at one time or another. In my experience bad drivers are more common than broken traffic lights!

At root I think autonomous cars are an engineering problem with engineering answers. It's not essential for drivers to be human.

Dave

[Mike PooleParticipant @mikepoole82104]

I look forward to trucks being under autonomous control, my trips to Italy last year and this year have both encountered trucks that wander out of their lane, last year I was very close to being the meat in a truck and Armco sandwich hard braking kept me out of a very quickly diminishing gap, the truck driver apologised with a wave and lights flash but it seems there are too many tired or bored drivers in charge of big trucks.

Mike

[Howard LewisParticipant @howardlewis46836]

Maybe the truck drivers have been taught by many of the UK car drivers, who seem to be unaware of what is around them. In the last 12 months, three rimes nearly run off the road by the car alongside changing lanes without looking!

Being a control freak, and having been used to commercial vehicles, don't like the thought of my driving itself, but autonomous vehicles may be safer than some on the roads.

Howard

[Neil WyattModerator @neilwyatt]

Posted by Bill Phinn on 14/07/2019 04:43:22:

"In what ways is Madame Bovary a realistic novel?"

I just asked my phone that. It cheated and looked up the answer which it read to me, before cutting off mid-sentence.

Now in the real world you could argue looking up the answer to something you don't know is the intelligent response to such a question!

Neil

[Colin WhittakerParticipant @colinwhittaker20544]

What standards should we use for a driving AI?

Must it never make a mistake?

How about ten times safer than the average human?

What about 10% safer than the average human?

What about 10% safer than the average 80 year old driver?

I'm not sure I would go for the last option but the penultimate one looks a good enough starting point as the safety level would only rise with time. Instead we seem to be fixated on total safety.

Colin

[JasonBModerator @jasonb]

Posted by SillyOldDuffer on 13/07/2019 19:22:07:

PS is Barnaby Wilde really Mick Charity in disguise?

One and the same.

[ronan walshParticipant @ronanwalsh98054]

Posted by J Hancock on 14/07/2019 09:51:43:

Call me 'old-fashioned' but I don't really want to participate in this futuristic world you have described.

Reads utterly boring , like being the Mekon, sitting on that flying saucer thing it used to move around on.

Nope, 1964 ish was nirvana, downhill ever since.

I tend to agree. The modern world that everyone in power seems to be striving for, seems a boring, clinical place.

[VicParticipant @vic]

Oh dear.

**LINK**

[Author]

Posts