3 hours ago, Bob Scott said:

If we wanted to solve the problem with texting while driving, we could easily require new cars to have an RFID chip embedded in the steering column and an interlock on phones that would disable texts if such a driver proximity chip was sensed and the phone was in motion faster than you could walk.  The same technology is already in use to sense a keyless fob near the door or driver's seat in many cars.

iPhones, and I believe Androids as well, can already automatically enable a Do Not Disturb mode when driving. It will kick in either when you hit a certain speed, or when connecting to the car's Bluetooth connection. However, it can be overridden so passengers in the car can still use their phones.

The only real solution is education. It isn't just phones either as there are still plenty of people messing around with other things in cars as well (eating makeup, shaving, screaming kids etc.), or simply driving when tired. In North America, at least, driving is seen as a right with little thought for the responsibility side of it.

1 hour ago, n4gix said:

"Do not start or stop chain saw if holding it between your legs."

 

 

6 hours ago, martin-w said:

I'd be very wary of trusting full auto drive until its proven itself to be pretty much infallible. The day will come soon though, they are experimenting now with fully autonomous vehicles with no driver.

Why would they need to be pretty much infallible? We already accept poorly trained people operating cars with barely a second thought right now. Aiming for pretty much infallible will be essentially impossible to achieve as there will always be some unexpected edge case. Aiming for significantly better than human operated would be more realistic while still saving thousands of lives on the roads.

Of course autonomous cars will never satisfy those that enjoy the act of driving itself, but for most people driving has become basically just the means to get from point A to B.

2 hours ago, goates said:

Why would they need to be pretty much infallible? We already accept poorly trained people operating cars with barely a second thought right now

 

Fair point, Infallible is probably the wrong word there's no such thing as infallible, more competent than a human would do.  I mean in terms of eliminating the unreliability that's currently present. And addressing  ethical questions like... does the autonomous system in control veer off the road and kill a child, or drive straight ahead toward the brick wall and kill the driver, or passenger in a driverless car. 

 

3 hours ago, goates said:

there are still plenty of people messing around with other things in cars as well (eating makeup, shaving, screaming kids etc.),

Or even this:

https://www.facebook.com/MrBean/videos/driving-and-changing-clothes-mr-bean/321801492809433/

2 hours ago, martin-w said:

Fair point, Infallible is probably the wrong word there's no such thing as infallible, more competent than a human would do.  I mean in terms of eliminating the unreliability that's currently present. And addressing  ethical questions like... does the autonomous system in control veer off the road and kill a child, or drive straight ahead toward the brick wall and kill the driver, or passenger in a driverless car.

To start with, I would hope an autonomous car would be far better at avoiding getting into those situations to start with. And if it does, I would say the goal should be to avoid injury and death of anyone outside the vehicle if there is a choice. By getting in the car I would say you are choosing to accept all risks associated with being in a moving vehicle, while an innocent bystander didn't. However, I also think debates over this are a little overblown as we expect people to make those choices today and don't seem to particularly care, but again seemingly want to hold autonomous cars to an almost impossible to achieve standard.

Watching that video of the Tesla in the winter, I think it's going to be a long time yet before autonomous cars are suitable for driving in winter climates. There are plenty of days here after a snowstorm where you can't see the pavement at all, and many others where there is black ice on the road even though it looks dry, not to mention the ruts that form after a few snowfalls with edges made of ice almost as hard as concrete. I think the biggest problem is that Tesla and the others working on autonomous cars are trying to do it using an entirely self contained system within the car itself, while I think they will need to use external sensors along roads to really make it happen. Sensors or beacons of some kind would make it much easier to provide data to the car, especially for road conditions or location data when you don't have a solid view of GPS satellites, for two examples.

10 hours ago, goates said:

To start with, I would hope an autonomous car would be far better at avoiding getting into those situations to start with.

 

Probably not. If a child jumps in front of the car at the last second, that's something the automation cant predict and avoid, reacting after the fact is the only option. Autonomy, advanced AI, doesn't mean precognition. There are certain scenarios that can be predicted ahead of time yes, but many that can't.

 

10 hours ago, goates said:

I would say the goal should be to avoid injury and death of anyone outside the vehicle if there is a choice.

 

That's your opinion and it might be mine, but there are many that disagree and despite agreeing to the ethical, "needs of the many"  argument, would not favour a vehicle that doesn't prioritise its owner. A vehicle that will kill you if it thinks its appropriate isn't palatable for many, and they are customers, that wouldn't buy the vehicle. So yes, it is a well known and debated issue with fully autonomous vehicles.

 

Quote

 Despite this, the studies also found that participants were more likely to buy an AV programmed to save its own passengers over any number of pedestrians, when offered the hypothetical choice between that and a utilitarian car. 

So, it seems that this ethical dilemma is simultaneously a social one: every person wants those around them to behave in a way that would lead to the best global outcome, but they are not motivated to practice that behavior themselves [3]. And while the utilitarian approach to accidents would theoretically save the maximum number of people, it seems that consumers would be less likely to actually buy AVs programmed with utilitarianism in mind.

 

https://vce.usc.edu/volume-4-issue-2/autonomous-accidents-the-ethics-of-self-driving-car-crashes/#:~:text=At its core%2C the main,using them) [4].

 

10 hours ago, goates said:

Watching that video of the Tesla in the winter, I think it's going to be a long time yet before autonomous cars are suitable for driving in winter climates.

 

They are suitable for driving in winter climates now, you mean "extreme winter environments".  

 

Quote

There are plenty of days here after a snowstorm where you can't see the pavement at all, and many others where there is black ice on the road even though it looks dry, not to mention the ruts that form after a few snowfalls with edges made of ice almost as hard as concrete.

 

I see no reason why an autonomous vehicles couldn't have the capability of a human in extreme winter conditions. And I would say better than a humans capabilities. It will come, and given the advancements that are currently being made, sooner than you think. 

Edited April 2, 20224 yr by martin-w

6 minutes ago, martin-w said:

Probably not. If a child jumps in front of the car at the last second, that's something the automation cant predict and avoid, reacting after the fact is the only option. Autonomy, advanced AI, doesn't mean precognition. There are certain scenarios that can be predicted ahead of time yes, but many that can't.

Like I was saying before, an autonomous car won't be perfect, but they should be far better at not getting into avoidable situations that people do (distracted driving etc.). This would then minimize the cases where a car has to make a choice. And even in the case of a kid running onto the road unexpectedly, the car could likely react faster than a person. My Rav4 likely saved me from rear ending a car on the highway that cut me off by hitting the brakes faster than I could, for example.

10 minutes ago, martin-w said:

That's your opinion and it might be mine, but there are many that disagree and despite agreeing to the ethical, "needs of the many"  argument, would not favour a vehicle that doesn't prioritise its owner. A vehicle that will kill you if it thinks its appropriate isn't palatable for many, and they are customers, that wouldn't buy the vehicle. So yes, it is a well known and debated issue with fully autonomous vehicles.

The day an autonomous car runs over a group of kids to save the rich owner inside their fancy autonomous car is the day this changes. The decision isn't going to be made just by those that are buying the cars.

12 minutes ago, martin-w said:

They are suitable for driving in winter climates now, it happens all over the world. But its not full autonomy yet of course. But yes, when we have full autonomy, like now, they will be able to do that.

How much winter driving experience do you have? The lane departure assist and adaptive cruise control features in my Rav4, and similar features in my friends' vehicles, work great in summer, but are almost useless, and sometimes downright dangerous (especially the lane departure assist), in the winter. There are days where the roads are completely covered in snow and people have to make a guess as to where the lanes are, and sometimes a 3 lane road essentially becomes a 2 lane road until the plows can clear it off. How will an autonomous car handle this when it can't get a solid GPS signal due to tall buildings in the way and snowdrifts along the side of the road covering the curbs? How will the car identify how dense or solid the snow on the ground is to judge whether the car can drive through it or not? Right now a person can get a pretty good idea simply from walking to their car.

Some engineers at the University of Waterloo estimate it could be around a decade away before Autonomous vehicles can properly handle winter driving.

https://www.cbc.ca/news/canada/kitchener-waterloo/self-driving-car-university-waterloo-winter-driving-1.5459331

Will they get there? Eventually, though as I said before I have my doubts about doing it with the fully self-contained model they are aiming for right now.

5 hours ago, goates said:

Like I was saying before, an autonomous car won't be perfect, but they should be far better at not getting into avoidable situations that people do (distracted driving etc.). This would then minimize the cases where a car has to make a choice. And even in the case of a kid running onto the road unexpectedly, the car could likely react faster than a person. My Rav4 likely saved me from rear ending a car on the highway that cut me off by hitting the brakes faster than I could, for example.

 

Not getting into avoidable situations... of course. But the ethical dilemmas are related to the example I gave, in which case no, they haven't got precognition and wouldn't be able to predict and unpredictable occurrence. The automatics most definitely would react faster, thus why the dilemma occurs, the autonomous sytem has a choice, which choice it makes is the question.

 

5 hours ago, goates said:

The day an autonomous car runs over a group of kids to save the rich owner inside their fancy autonomous car is the day this changes. The decision isn't going to be made just by those that are buying the cars.

 

Not the point. Autonomous systems ruining into brick walls and killing their owners instead of changing direction would result in loss of sales. An autonomous vehicle that you know is designed to favour the lives of pedestrians instead of its passengers, according to the research I quoted, is less likely to sell, if its less likely to sell its not a viable product. As I mentioned, the dilemma isn't a thing yet, but one day soon it will be. This isn't just me saying this, much consideration and research is being conducted in this respect. The decision is going to be made by those buying the cars. They wont buy them. A rival car maker codes the autonomous system differently and they will sell more vehicles, as the research predicts. Now if the authorities legally demand that all cars favour the lives of pedestrians and kill their owners and scrap all older cars that don't (not going to happen) then I would think there would be a sizable group of people that object.

I'm not claiming to know the answer, that's still to be determined, but its a dilemma that experts are researching and not as simple as just kill the passengers. 

 

5 hours ago, goates said:

How much winter driving experience do you have?

 

Not sure why that's relevant but 40 years. Snow, roads like glass, black ice flooded roads etc. 

 

Quote

and similar features in my friends' vehicles, work great in summer, but are almost useless, and sometimes downright dangerous (especially the lane departure assist), in the winter.

 

Not relevant. We were talking about a vehicle in the future with full autonomy or at least very sophisticated self drive. As I said, Tesla's get over the air updates on a very frequent basis, and the self drive function is progressing pretty rapidly. The capability for a vehicle to handle such conditions as well as a human will come. And as you have pointed out, the automation will have better reflexes than a human. Will it be perfect? Probably not. Will be it be infallible? no. But it most definitely will be better in winter conditions than a human being with a human beings limitations. If we suggest not, then we are making the same mistake that has been made before when it comes to what technology can and will achieve. 

 

5 hours ago, goates said:

How will the car identify how dense or solid the snow on the ground is to judge whether the car can drive through it or not? Right now a person can get a pretty good idea simply from walking to their car.

 

I can think of a number of technological solutions, sensors that could do that. 

 

5 hours ago, goates said:

Some engineers at the University of Waterloo estimate it could be around a decade away before Autonomous vehicles can properly handle winter driving

 

Of course, as I said, it will come, and as I said, sooner than we think, and 10 years is sooner than I thought. 😁

Edited April 2, 20224 yr by martin-w

5 hours ago, goates said:

How will an autonomous car handle this when it can't get a solid GPS signal due to tall buildings in the way

 

That technology already exists. Oxbotica, system ingests data from lidar, radar, cameras  and numerous sensors. Its not dependant on any one modality, like GPS. Never underestimate what technology can achieve. 

45 minutes ago, martin-w said:

Not the point. Autonomous systems ruining into brick walls and killing their owners instead of changing direction would result in loss of sales. An autonomous vehicle that you know is designed to favour the lives of pedestrians instead of its passengers, according to the research I quoted, is less likely to sell, if its less likely to sell its not a viable product. As I mentioned, the dilemma isn't a thing yet, but one day soon it will be. This isn't just me saying this, much consideration and research is being conducted in this respect. The decision is going to be made by those buying the cars. They wont buy them. A rival car maker codes the autonomous system differently and they will sell more vehicles, as the research predicts. Now if the authorities legally demand that all cars favour the lives of pedestrians and kill their owners and scrap all older cars that don't (not going to happen) then I would think there would be a sizable group of people that object.

I'm not claiming to know the answer, that's still to be determined, but its a dilemma that experts are researching and not as simple as just kill the passengers. 

Asking people to think about it in an abstract fashion when it doesn't really apply right now and likely isn't taking other factors into account, I don't think is going to give the final answer. Of course people aren't going to want a car that sounds like it is going to deliberately kill themselves. Did those researchers ask them what they would prefer if the person getting run over was their own child or family member? People are terrible at assessing risk, and I suspect the part that will win people over is the convenience (being able to do all the things that you shouldn't be doing while driving today) combined with far better safety (people once upon a time swore that elevators needed an operator to work safely). And society as a whole will have just as much, if not more say than he people buying the cars. The manufacturers will certainly follow the laws that get set out as not doing so means no sales at all.

52 minutes ago, martin-w said:

Not sure why that's relevant but 40 years. Snow, roads like glass, black ice flooded roads etc.

I ask because often when this debate about winter driving comes up, it's from someone who has little or no experience with proper winter conditions.

As for the technologies to make autonomous winter driving possible, yes, I agree it will come, though I think that ten year estimate is still a little optimistic. The winter conditions in Canada between the Maritimes, southern Ontario, Prairies, foothills of the Rocky Mountains and Northern communities are all different, and isn't just visibility that's the problem. Your Oxbotica example doesn't seem to talk about winter driving much, especially accounting for deeper snow cover and different types of snow.

Here's an article from a Finnish team that I think is on a more realistic track with suggesting external data from other vehicles may be required. I still not convinced that some combination of cameras, lidar, radar and/or other sensors on the car alone will be able to provide all of the data required.

https://techcrunch.com/2017/12/15/finnish-autonomous-car-goes-for-a-leisurely-cruise-in-the-driving-snow/?guccounter=1&guce_referrer=aHR0cHM6Ly9kdWNrZHVja2dvLmNvbS8&guce_referrer_sig=AQAAADCQEq27afvQ2Vzcnm7JttdaNItjqyi5lIbvLlKaSyLhuyJFNkGDTsZXNYERdGKlT8FrgqWKq60K5e-cXHvtOfZIMZVC7_c8fxYROr4l8GjZfqNVZX5EAYvN3YZFzWnLjR6x8HbHbgoORnU7FnFcMJ5QyVc6TtF7iyysLJ4c8Q91

17 hours ago, goates said:

Did those researchers ask them what they would prefer if the person getting run over was their own child or family member?

 

Its not about the opinion of an individual that's had a tragedy or contemplates a tragedy. Obviously in that situation they would make a decision based on emotion. It's bit like asking someone if they agree with the death penalty and they say no, then after a loved one is murdered they, in an emotional state, say yes. Its about what purchasing choices people make, on average, as a population. As the research suggested... people say they agree in principle humankind favouring the needs of the many, but are not motivated to practice that behaviour themselves when it comes to purchasing a vehicle. 

"but they are not motivated to practice that behavior themselves [3]. And while the utilitarian approach to accidents would theoretically save the maximum number of people, it seems that consumers would be less likely to actually buy AVs programmed with utilitarianism in mind."

 

Quote

Of course people aren't going to want a car that sounds like it is going to deliberately kill themselves.

 

Precisely!

 

17 hours ago, goates said:

And society as a whole will have just as much, if not more say than he people buying the cars.

 

Most people drive a car. Society as a whole are the people that buy cars. Its the authorities that could dictate not favouring the occupants of the car. And that's the point, its something that is being researched, contemplated, and its not as easy as favouring "the needs of the many". If it was that easy they wouldn't be doing the research. 

 

Quote

As for the technologies to make autonomous winter driving possible, yes, I agree it will come, though I think that ten year estimate is still a little optimistic. 

 

No possible way to say. Its dependant on technological innovation and we aren't experts in this field, thus I would consider the experts prediction of 10 years to be a time frame worthy of consideration. We are entitled to our own opinions, and what we can base those opinions on is rate of technological advancement we have seen to date. 

 

Quote

The winter conditions in Canada between the Maritimes, southern Ontario, Prairies, foothills of the Rocky Mountains and Northern communities are all different,

 

Might not be a good idea to cherry pick particular locations with particularly challenging weather. Nobody said  autonomous vehicles will be ubiquitous and all over the planet and replace all vehicles in all situations. 

 

Quote

 Your Oxbotica example doesn't seem to talk about winter driving much, especially accounting for deeper snow cover and different types of snow.

 

You asked what would be done if building blocked a GPS signal. As I said, technology can accommodate that now. Now project forward 10 years, or 15. 

I'm wondering what purpose autonomous vehicles will serve.

Is it just more automation to replace human beings with machines in order to increase profit and not have to deal with those pesky humans?

Is it for safety purposes?  Do we know for sure that autonomous vehicles would be safer?  What studies prove this?

There's no question that in some circumstances, like dangerous environments, autonomous vehicles can be useful.  I'm just not so sure about millions of them sharing the road with humans.

Dave

22 hours ago, goates said:

(people once upon a time swore that elevators needed an operator to work safely).

At that time this was a reasonable belief, since elevators had to be manually controlled by a hand wheel, and the operator had to stop movement at each floor such that the elevator's cage floor was reasonably even with the destination's floor!

Having used a manually controlled freight elevator in an old warehouse, I can attest to the difficult challenge of achieving that goal... 😉

4 hours ago, dave2013 said:

Is it just more automation to replace human beings with machines in order to increase profit and not have to deal with those pesky humans?

Is it for safety purposes?  Do we know for sure that autonomous vehicles would be safer?  What studies prove this?

 

We were referring to autonomous self driving cars. So its more of a safety thing and I guess better traffic management. There are 1.3 million deaths and 50 million injures on the roads each year, mostly due to human fallibility, so if the autonomous self driving tech works as its supposed to that figure should drop hugely.

Research has suggested such vehicles could save up to 94% of all fatalities. You cant really prove such a thing until it happens in the real world though, until cars are autonomous. Until then its estimates and predictions. Would seem logical though if the tech works as its supposed to. 

 

https://mindmatters.ai/2019/12/would-selling-self-driving-cars-sooner-save-lives/#:~:text=“DoT researchers estimate that fully,to the World Health Organization.