Self-driving cars?

[Ozexpatriate]

1 minute ago, Michael Hodgson said:

Driving to visual signs is sub-optimal. 

It is necessary to adapt to the current infrastructure.

 

Obviously a road system where everything (signals, road direction, lanes, other vehicles etc) has telemetry is a far simpler technical problem.

[Nearholmer]

Re: satnavs.

 

Why does anyone buy one?

 

I use google maps, which is free, seems well up to date, has pretty sound real-time traffic info, and gives a screen presentation that I find easier to read than any I’ve seen on a satnav package. It even has a “sort of OK” cycling option, although there is a better freebie app for cycling navigation.

 

When I bought my car, there were two options on the model, with satnav functionality, or for a few hundred pounds less, without, but plug the phone into a usb socket, and use a range of phone-held apps instead. Both had exactly the same screen. You can guess which one I chose!

Edited April 29, 2021 by Nearholmer

[Reorte]

23 minutes ago, Nearholmer said:

Re: satnavs.

 

Why does anyone buy one?

 

I use google maps, which is free, seems well up to date, has pretty sound real-time traffic info, and gives a screen presentation that I find easier to read than any I’ve seen on a satnav package. It even has a “sort of OK” cycling option, although there is a better freebie app for cycling navigation.

 

When I bought my car, there were two options on the model, with satnav functionality, or for a few hundred pounds less, without, but plug the phone into a usb socket, and use a range of phone-held apps instead. Both had exactly the same screen. You can guess which one I chose!

My car came with one built in. Can't say I would've chosen the option though (it was second hand anyway). Although I don't have a mobile phone either, so I'd just make do with paper maps.

[black and decker boy]

44 minutes ago, Nearholmer said:

Re: satnavs.

 

Why does anyone buy one?

 

I use google maps, which is free, seems well up to date, has pretty sound real-time traffic info, and gives a screen presentation that I find easier to read than any I’ve seen on a satnav package. It even has a “sort of OK” cycling option, although there is a better freebie app for cycling navigation.

 

When I bought my car, there were two options on the model, with satnav functionality, or for a few hundred pounds less, without, but plug the phone into a usb socket, and use a range of phone-held apps instead. Both had exactly the same screen. You can guess which one I chose!

My built in satnav  is........ Google Maps :-) The car has 4G wifi built in so is always online (it can also do YouTube, web browsing and play the piano).

 

we own 2 satnavs from before smart phone,  both now live in a drawer and like you we use our phone if we need such a system (eg abroad)

[Canal Digger]

Given that there are people out there who deliberately cause crashes to claim injury with their witnesses ready at the scene, I waiting to see what happens when one of the targets a self-driving car.

[black and decker boy]

7 hours ago, Canal Digger said:

Given that there are people out there who deliberately cause crashes to claim injury with their witnesses ready at the scene, I waiting to see what happens when one of the targets a self-driving car.

The self driving car will have 360 degree dashcam recordings to prove what happened- the Tesla has 6 external cameras that record everything. They also record when parked in Sentry mode as a security measure and images can be downloaded by Tesla should you car go bye bye.

 

mind you, with GPS fitted and 4G connection, Tesla can track and disable a car remotely so the thieves shouldn’t get too far.

[D9020 Nimbus]

I just hope that self-drive cars aren't connected to the internet…

[Nearholmer]

They’ll spend all ruddy day on Facebook, or playing fort nite, and you won’t be able to any get sense out of ‘em.

 

 

Edited April 30, 2021 by Nearholmer

[johnofwessex]

On 29/04/2021 at 17:37, Nearholmer said:

 

I'd agree with 30801, the car owner's inusurer would pay for the loss, but I would also expect them to pursue the manufacturer to recover their costs if they had good evidence of a manufacturing or design defect, which is what "unfit" software would be.

 

Proving that the software was "unfit" would be an interesting exercise though, especially if, as I expect it would have been, it was certified safe to a defined level by an independent safety assessment body.

 

What would happen to your NCB and next premium is another question.

 

We had a slightly similar case, where a fire was caused in our kitchen, causing thousands of pounds-worth of damage, due to a faulty gas fitting. The insurer paid-up without quibble (I had to provide estimates and receipts, but that was reasonable), but they then pursued the builders of the house (it was c8yo at the time), and proved that they had received, and failed to react to, a product defect advisory notice that had been issued by the appliance-maker shortly after the house was built. 

 

To cut a long story slightly shorter, our insurer did not raise our premiums, and undertook not to have the claim "count against us" if we sought quotes from other insurers.

 

As its gas the HSE should have been advised as well

[Nearholmer]

Really? By whom? Who would know to do that?

[kevinlms]

On 01/05/2021 at 03:26, Nearholmer said:

They’ll spend all ruddy day on Facebook, or playing fort nite, and you won’t be able to any get sense out of ‘em.

 

 

Then they will need more and more data. This will mean more and more 5G mobile bases on each and every street corner. All the better to control us all!

 

There, I'd better put a smilely in, just in case!

 

[PatB]

I would suggest that if an autonomous vehicle can (a) keep within lane markings, (b) maintain a steady speed when it's appropriate to do so, (c) maintain a safe distance between itself and other road users, (d) maintain its awareness of other road users and objects its vicinity and use the information to not collide with them, it is already streets ahead of 95% of West Australian drivers. 

[adb968008]

On 29/04/2021 at 17:42, Michael Hodgson said:

That's where Artificial Intelligence comes into its own.  Just like a human, such technology has to learn what combinations of visual or other clues can indicate a likely outcome.  Yes, there is a colossal learning curve.  A human has a learning curve too, and the reason we don't allow a five year old to drive is that he hasn't learned enough yet.  But once machines have learned, their experience can be passed from machine to machine, rather than each individual having to make the same mistakes as its elders and betters have already made.  And the machine won't doze off at the wheel after a long day at work.  

 

As you say however recognition of cyclists, pedestrians and stray dogs is a major technical challenge, but they do seem to be getting there.  

 

Long term the machines must win, because humans will always be subject to human error.  Autonomous vehicles will become safer than us at the wheel once their statistical error rate is less than ours.   That's not to say that accidents will finish completely - a regulatory body will still need to investigate such accidents as do occur and take remedial action.  As I recall the early OD crossings had to be manned during their first winter because the Lidar arrangement didn't cope adequately with falling snow - but the problem has been fixed.

 

Machines rarely make mistakes - the computer bugs that everybody whinges about are down to human design errors and oversights when we are programming the machines. 

Agree with most, however there is one thing computers will never be able to overcome.. that is human behaviour as an external influence.

 

Humans (and other mammals, fish, birds, viruses etc) are susceptible to once in a lifetime behaviours, which are unique to the individual, at the time at the place..,computers cannot currently overcome this...if the technology existed to predict this,  there would be no covid.

 

So computers can only learn based on past events*, and apply trends..but the one off will always catch it out, and many accidents are one off events, so at best computers can only overcome common errors, which is a step forward, but not a panacea.

 

* remeber when tomtom was new and cars getting stuck in fords, lorries in country lanes...

 

Edited May 16, 2021 by adb968008

[adb968008]

What I would probably trust as a starting point would be centralised management of vehicles on motorways...

 

if all vehicles gave up to be controlled entering the on ramp, gave up on exiting, a management system could plan traffic in block sections of motorways controlling speeds / lanes of all vehicles with in it, and handing off to the next section.

 

I suspect 100% fully autonomous vehicles will be too expensive for individual ownership. Similarly the raw compute power required will require vast amounts of data transfer that could not be handled centrally. A lot of edge networking computing, that would require huge infrastructure would be required (think along the lines of setting up a nationwide network of local sub stations on housing estates) to handle the localised required data.

 

Should technology reach this point I suspect the Ubers of this world will be the winner, and we just rent a vehicle on demand.

 

Edited May 16, 2021 by adb968008

[Nearholmer]

A ‘central brain and slave vehicles’ system would be a nightmare to implement for the reasons you’ve set out, and would be prone to ‘right side’ failures, frequently having to hand control back to people.

 

As to the cost of autonomy, it will be huge up-front, but very low per unit to deploy, and it can be arrived at in baby-steps, so I’d be more optimistic about its prospects.

 

MOT Test requirements around it? Are there any now in respect of the first-step functions that are already, and are about to be implemented?

[Allegheny1600]

Here is an interesting piece in the Guardian today;

https://www.theguardian.com/books/2021/may/16/daniel-kahneman-clearly-ai-is-going-to-win-how-people-are-going-to-adjust-is-a-fascinating-problem-thinking-fast-and-slow

A quote from Daniel Kahneman:

”You don’t talk about driverless cars in your analysis. But that, I guess, is becoming a key arena of this argument, isn’t it? However much safer automated cars might be statistically, every time they cause an accident, it will be excessively magnified?” (Guardian)
“Being a lot safer than people is not going to be enough. The factor by which they have to be more safe than humans is really very high.” (Kahneman)

 

Food for thought?

[Edwin_m]

From the HS2 thread:

6 hours ago, billbedford said:

 

Unless they are used as an ubiquitous 'pay as you go' transport that starts by replacing taxies, Uber etc. 

That is one possibility, and a shared model would reduce vehicle ownership and reduce problems associated with mass production of batteries, as well as making it more likely that people would consider alternatives for a particular journey.  But it raises a whole lot of other questions about whether a shared vehicle is in a fit state to be used by the next person.  Imagine what some of them would be like on a Saturday night!  

5 hours ago, GoingUnderground said:

You've missed the point about car parking. Why should the self-driving car go all the way back to it's normal overnight parking space? They wouldn't. They'd be programmed to park close to the drop-off point, or possibly the pick-up point if that was to be different to the drop off point.

 

The point that I was making is that city car parks do not need to be close to the most popular destinations in towns and cities, but could be on the outskirts or on nearby less busy areas, and that they may need a lot less space than at present, and they certainly won't need ventilation to remove nitrogen oxides or carbon dioxide or monoxide because there won't be any if the cars are all pure electric.  So they could be underground megaparks many levels deeper than at present once the ventilation and access issues are removed. Empty cars driving themselves to and from car parks could generate more congestion, but it might mean less as there would be no need for on-street parking in business, commercial and shopping areas, increasing available road space at stroke. There would be minimal emissions if they're fully electric. 

It depends on property values.  If cars are self-driving then car parks will gravitate to less expensive areas, which still means extra mileage and congestion in the city centre, particularly if reduction in parking charge leads to more people using the self-driving car instead of public transport.  Congestion may ration this effect, but any increase in city centre traffic isn't ideal for a whole range of reasons, even if the vehicles in question are electric.  

[Nearholmer]

While a lot of people will continue to spend a lot money in return for the combination of convenience and status of having a personal car, I think it is imaginable that some people will subscribe to "car clubs", in the same way that they take out membership of gyms, in fact some do already in cities.

 

The "club" aspect could offer a range of options, from an expensive club for "nice" people who are unlikely to leave the car spattered with vomit and smelling like the alleyway behind a nightclub, through to a really cheap club, that practically guarantees those things, plus full ashtrays, and a nasty sticky feeling on the seats. Gyms certainly range from the "exclusive" (where people barely do any exercise, and mostly sip lattes while looking at the lawns), to the pretty bare, basic, and smelling faintly of stale sweat.

 

Early adopters might be younger people, and couples who currently own/lease two cars, and could happily move to one owned/leased, plus call on a club car for times when both need a vehicle simultaneously.

 

Autonomous vehicles make the "club" model viable in suburbia, whereas now it only works in densely-populated areas, where walking to the club car that you've reserved isn't a great hike. 

 

The mental leap from leasing to clubbing isn't all that great, and a lot of people currently lease cars.

[Ozexpatriate]

1 hour ago, Nearholmer said:

I think it is imaginable that some people will subscribe to "car clubs" ...

Or that they will click a button on their telephone and an Uber / Lyft / Hertz vehicle will materialize, take them to their destination and then scuttle off to a recharging point until the next fare.

 

It's easy to imagine never owning a car, or paying for car insurance. No need for big car parks downtown. No parking tickets or towing. No need for a driving license. Mobility for seniors than can no longer drive.

 

This can be very disruptive technology.

 

[Ozexpatriate]

5 hours ago, adb968008 said:

Humans (and other mammals, fish, birds, viruses etc) are susceptible to once in a lifetime behaviours, which are unique to the individual, at the time at the place... computers cannot currently overcome this..

Current autonomous vehicles are trained on databases of historical accidents - so they have built-in learned behaviour for a vastly richer set of experiences than any human driver. Doubtless there will be truly unique, new corner cases, but programming for fail-safe modes will still likely be superior to human reaction.

 

As to CoViD, it was both predictable and predicted. Governments failed in very human ways.

[adb968008]

26 minutes ago, Ozexpatriate said:

Current autonomous vehicles are trained on databases of historical accidents - so they have built-in learned behaviour for a vastly richer set of experiences than any human driver. Doubtless there will be truly unique, new corner cases, but programming for fail-safe modes will still likely be superior to human reaction.

 

As to CoViD, it was both predictable and predicted. Governments failed in very human ways.

Still disagree, the movements of a small child tripping up on a street corner on Oregan, are different from another near identical child tripping up on a street corner in Peckham. The vehicle involved may be different, or differently configured, to the database archive, the weather, time of day different etc etc.

 

The action might be the same, the variables will be different, which may lead to a totally different outcome.

 

Not even a Cray computer can predict that (Ive worked with one).

 

I’ll put it another way..

Fill a glass with water, then slowly tip and pour it out.

no matter how many times you try, even with a computer, variables will dictate every flow rate of water poured out each time will be unique.

 

somethings science cannot predict... it can model, it can build millions of probabilities, but it cannot be accurate every time.

 

its the reaction that counts, and a computer will at best pick a predetermined action, not a human emotional one.. that could be the difference between life or death, for the driver, the child or both, depending on a programmed choice...

 

will the car be programmed to save the child, or the driver if it had a choice of saving one or the other ?

 

what choice would the car make, if the child might not fall on the road, but if it did take preventative action, 3 people in the car may die as a result of that action?

 

Do we want to give a car that choice, effectively being its own judge and jury, in what usually follows legal proceedings/inquests etc when such variable incidents occur ? - And would the law, the grieving party be bound by the cars decision ?

 

Finally, US1549... the pilot made a decision, that was contrary to the computer modelling, which saved everyones life.

 

 

Edited May 16, 2021 by adb968008

[Nearholmer]

"This can be very disruptive technology."

 

Yes indeed, it could all go off in multiple directions, and probably will until one(or a few) model(s) comes to dominate. 

 

To me, "user not owning" models seem particularly natural for EVs, and especially autonomous EVs, because the major parts (monocoque; batteries; traction package; suspension and steering; interior; software etc.) look almost certain to have different life-spans, so the natural owner is a fleet-operator, who would be well-placed to run cyclic replace/refurb of major parts - a given monocoque could be kept in service almost indefinitely, yet the vehicle presented to the user could be "permanently almost new".

 

But, cars are about a lot more than practicality. They are partly about self-projection (status and/or life-style symbols), otherwise we'd all be driving some sort of genuinely utilitarian (not SUV "utilitarian") vehicles, all the same colour, and that fact will take a long time to work its way out of the system .......... unless self-projection can be wrapped into the dial-a-ride offering. People will spend an awful lot of money to feel one-up on their fellow beings, and a fair bit of money just to feel like an individual!

 

 

Edited May 16, 2021 by Nearholmer

[Ozexpatriate]

16 minutes ago, Nearholmer said:

People will spend an awful lot of money to feel one-up on their fellow beings, and a fair bit of money just to feel like an individual!

Of course there's room for both rideshare and ownership modes.

 

There are generational things at work too. Does a GenZ individual care more about social media 'likes' than what car they drive, for example?

[Ozexpatriate]

22 minutes ago, adb968008 said:

will the car be programmed to save the child, or the driver if it had a choice of saving one or the other ?

 

what choice would the car make, if the child might not fall on the road, but if it did take preventative action, 3 people in the car may die as a result of that action?

 

Do we want to give a car that choice, effectively being its own judge and jury, in what usually follows legal proceedings/inquests etc when such variable incidents occur ? - And would the law, the grieving party be bound by the cars decision ?

Already being considered. Your last comment is part of the regulatory component of autonomy, not the technology.

 

It is why object recognition is so important in the sensor fusion. Emergency braking (that might injure the passengers) might not be deployed for a squirrel or a cat, but you would want it for a baby in the road. Surprisingly, (or perhaps unsurprisingly) different nations have different liability regulations for human driver behaviour. This is being factored in too.

[Michael Hodgson]

43 minutes ago, Ozexpatriate said:

Emergency braking (that might injure the passengers) might not be deployed for a squirrel or a cat