F1technical.net

Just_a_fan wrote: ↑

04 May 2019, 12:15

Zynerji wrote: ↑

03 May 2019, 23:57

We will see when its on the market.

Just getting 3-4 lawsuits a year could wipe out enough profit to sink a manufacturer if they take 100% responsibility for the operation of a machine that they no longer own, and have no compulsion to force dealership maintenance/calibration on any type of schedule.

Doubtless the acceptance of liability will come with a long list of "you must service it this way" type requirements. Usual get-out clauses for any bit of kit, to be honest.

I could see the vehicles having self-check systems that just refuse to start the vehicle if it's not maintained correctly, and which fail safe to park at the side of the road if an error occurs.

I will personally base my stock purchases of auto manufacturers purely from their stance on this topic.

That is your right, of course. Others will do likewise but in the other direction. It'll balance out.

I fixed a BMW dealership a few years ago. I found that 75% of the total dealership profit came from the service department, and only 25% from sales.

Taking this responsibility under contract of dealer- only servicing might be the most financially beneficial for manufacturers going forward.

Time will tell.

strad wrote: ↑

01 May 2019, 20:00

15 minutes of BS.

Why would you say that?
The next three pages of discussion here are about ethical considerations. But an elaboration on the difficulties of ethics, is somehow BS?

I know the 737-Max crashes have previously been brought up already, but having just spent some time reflecting on what happened, I do wonder how confident people are feeling about AVs relying on potentially faulty/dirty/erroneous sensors while driving at 22m+ per second (80kmh) on a 3m wide lane in a highly populated or dense area. While in a plane, the aircraft still had about 6 minutes of flight where the pilots helplessly tried to override the faulty software (and failed), my guess is that a mere consumer (possibly lacking any drivers license because with AVs you no longer need to drive right?) will have less than about 3 seconds to react.

Of course we can also assume that cars will have dozens of fail safe sensors to avoid any potential risk, but at what cost? And how heavily scrutinized will car maintenance have to be to ensure the safety of not only the passengers of that vehicle, but everyone else? Will they be held to similar rigorous safety standards as the aviation industry is, costing millions?

Phil, I think that is why the big boys are making steps slowly. AFAIK most OEMS have an L4 solution, but it is totally ridiculous cost-wise. When they start to lose significant market share because they haven't got L2 or more in the showroom, that's when you'll see some action. Until then Tesla are going to push their L2 system, my guess is they'll be forced to introduce more driver monitoring, and everyone else will mess about with L1-L3, with no great enthusiasm. I rather like GM's Super Cruise, except it demands that you stay alert.

Remember history: The big boys gamed the emissions checks so that their cars "knew" when they were being tested and therefore reduced the emissions they made.
Why - to increase sales and therefore profit.
How - to suggest that their diesel was cleaner than alternatives - it was when mixed with amonia, but at a cost that they believed the consumer didn't want to pay (possibly true)
Impact - real world usage emissions were higher and more toxic, leading to poor air quality and a reduction in life span

Whats my point? Profit and the drive for the bottom line makes people play fast and loose with other peoples lives either directly or indirectly.

Not aimed at anyone just my observations.

Phil, the big difference between aircraft and cars is that cars can just fail-safe stop. Of course, that needs to be considered in the design and implementation of AVs, but it is a lot easier than trying to deal with an aircraft in flight.

markc wrote: ↑

07 May 2019, 15:59

Impact - real world usage emissions were higher and more toxic, leading to poor air quality and a reduction in life span

open road emissions were higher but real world emissions weren't (because diesel cars don't make NOx in city driving)
ie there was no fall of air quality in regions of poor air quality
ie there was no reduction in human life span

Phil
re the Boeing scandal
it's not a software fault
it's a system design fault - and a failure of regulation (seemingly due to Boeing lies)

Tommy Cookers wrote: ↑

07 May 2019, 17:03

Phil
re the Boeing scandal
it's not a software fault
it's a system design fault - and a failure of regulation

True. Without wanting to go into much detail about the 737 scandal, I did find it interesting from the perspective that an AV is expected to do quite a bit more than what aircrafts are currently doing in relative empty skies while adhering to rigorous safety standards (relatively speaking). I wanted to highlight the difference between an aircraft with highly trained Pilots ready to step in and potential time to react vs the tight and narrow tolerance levels in an autonomous vehicle expected to drive safely on highly populated roads.

The second 737-max crash was supposedly due to a faulty sensor reading. My point is that a software driving an AV relies on its sensors to know what is happening around it at every moment.

Just_a_fan wrote:Phil, the big difference between aircraft and cars is that cars can just fail-safe stop. Of course, that needs to be considered in the design and implementation of AVs, but it is a lot easier than trying to deal with an aircraft in flight.

True, but braking isn't instantaneous. The vehicle will still require a decent stretch of road (distance), how much depends on the weight and speed and state of tires/brakes. Sometimes taking avoiding measures would be better, though I am doubtful that a piece of software could do that safely (safer than simply stopping) without endangering others. What the car will do, depends on the complexity of the software and the sensors for the software to correctly identify the situation and take the right measures. Incorrectly assessing the situation can/will have consequences.

Don't get me wrong; I'm both a nerd and a petrolhead. I'm an IT engineer as much as I love tracking my car. The nerd in me would love to be working on such AV software, program it and see it out there in the real world, dealing with the stuff I programmed it to do. I'm participating in this topic foremost because I see big issues and problems in regards to market viability, from a cost & benefit perspective. I am also rather doubtful that the tech will mature enough to do what is envisioned. Not while there are so many unpredictabilities waiting around every corner.

The technology is undoubtedly exciting, but is it good - for mankind? Is this what progress is? Giving up freedom and handing it over to a 'machine'? Where does it start and where does it end?

Just_a_fan wrote: ↑

07 May 2019, 16:05

Phil, the big difference between aircraft and cars is that cars can just fail-safe stop. Of course, that needs to be considered in the design and implementation of AVs, but it is a lot easier than trying to deal with an aircraft in flight.

Well, they also can't. Aircraft can glide hundreds and hundreds of kilometers while a problem is worked on, with zero input, or risk of obstacles. Of course that is if there isn't a structural problem with the airframe, or software isn't adding control input that drags it out of the sky.

Cars on the other hand are constantly only a few meters from any given obstacle, many of which the vehicle might be impacted BY, if it were to suddenly stop. Nevermind the fact that a sudden stop of a car could kill a motor or bicyclist, or even the fact that an erratic machine-generated pulling over to the shoulder in smooth running highway traffic could end up being a fatal cutoff for a following vehichle.

I would envisage the equivalent of a 'fail safe' where the vehicle transmits 'panic' to all vehicles around it and to some fixed point off the carriageway and this system then takes over and guides it to the set location where it stops.

The problem with a human controlled vehicle is even if the driver knows what they are doing (and they often do not and do the wrong thing in panic) the vehicles around it do not know how to react in accordance with it.

A command the AV equivalent of " emergency, stay away. I am stopping hard/slowly then veering left/right to the designated position in which I know no other vehicle will be " would remove any variables other than the damage to that one vehicle and remove panic or bad choices.

Sensor maintenance and calibration is the single point failure in this, I believe.

The perfect system will have needs like regular car washes, sensor replacement and dealership software upgrades.

It seems that the cost of ownership will rise severely. The best method, I believe, is moving to a 12 month lease-only system, so the car is guaranteed to get proper maintenance at least once per year.

Phil wrote: ↑

07 May 2019, 17:40

...

Giving up freedom and handing it over to a 'machine'? Where does it start and where does it end?

There are many “freedoms” in play. Pedestrians and cyclists may want the freedom to share the travel network with other transport users. Those who can’t drive, through infirmity or age or whatever, might want the freedom that those who can enjoy. None of these groups are easy to monetise, not at the rate that would make them attractive to investors in AVs any way.

Maybe cyclists and pedestrians might need to carry a beacon if they want to share space with cars. Mobile phones are sometimes used for this now. Not carrying a beacon might carry the same penalties as jaywalking today. In addition to injury or death that is. A restriction on their freedom.

This issue of freedoms comes up all the time. One persons freedom, to enjoy their loud exhaust say, is another’s restriction, to enjoy a quiet environment.

@Big Tea’s suggestion of a panic mode might restrict peoples’ freedom to drive a car not equipped to respond. Etc.

Zynerji wrote: ↑

07 May 2019, 20:19

It seems that the cost of ownership will rise severely. The best method, I believe, is moving to a 12 month lease-only system, so the car is guaranteed to get proper maintenance at least once per year.

I think the key is ownership. We are used to owning our car and being more or less responsible for it.

I would expect AVs would be leased on a per-trip basis, not on a time period. The manufacturer (or some other holding company) would own and be liable for the AV. The user then just rents it on an agreed basis.

We will have to change our ideas / expectations of the car and our ownership / use of it as population numbers and densities increase.

A typical commercial airliner will glide at a ratio of about 3:1. So if I were at 10000 feet and lost all my engines, I could expect to glide about 30 miles.
Quite a bit shorter than a small plane at 17:1
What I keep coming back to is... Who the heck asked for this? No one I know of. It's something they came up with and now thru advertising convince people they want.
It's like the sensors to stop the car...I've never known anyone to say.. "I'm such a poor driver and don't pay attention I need the car to stop it's self".

I for one ask for this. Driving is mega-stressful for me.