self driving cars in nyc

[Jeanne Boyarsky]

General Motors is testing self driving cars in Lower Manhattan. Maybe I'll get to see one "in the wild." I think it is great they are testing in a heavily trafficked environment.

And there's two people in the car - one ready to drive at a moments notice and another collecting data.

[Paul Clapham]

I'm waiting for them to start testing in places where there's sometimes snow on the roads. New York is one of those places, isn't it?

[Pete Letkeman]

I have not been to NYC, but I understand that the traffic can be very bad at times.

A couple of months ago:

Ontario and Michigan launched North America’s first national cross-border automated vehicle...

You can read all about that here: https://www.insauga.com/driverless-cars-being-tested-on-ontario-roads

I think that the problem with self driving cars will be when there are self driving cars and regular cars on the road at the same time.
Computers are great, but they do not always read the situation the same as a human. Not only that not every human read the situation the same way.
I think that once everyone uses a driver-less car then the rules and situations will have a more predictable outcome.

[Jeanne Boyarsky]

Paul: Depends on when in 2018 they start .

Pete: Agreed that the problem is far simpler when only self driving cars are on the road. But we need to get there. So first...

[Campbell Ritchie]

My experience is that the worse the traffic, the easier the driving. The same applies on a bicycle: if you want trouble go somewhere where there is hardly any traffic.

Of course it is also a case of more haste, less speed.

[Pete Letkeman]

Paul Clapham wrote:I'm waiting for them to start testing in places where there's sometimes snow on the roads. New York is one of those places, isn't it?

They are planning on testing self driving cars in Ontario as noted here http://www.cbc.ca/news/technology/driverless-cars-ontario-pilot-project-1.3268641, which was a couple of years ago now.
Ontario is rather large and some times Ontario can get a lot of snow during the winter months.

[Pete Letkeman]

Jeanne Boyarsky wrote:Pete: Agreed that the problem is far simpler when only self driving cars are on the road. But we need to get there. So first...

So how can we get there quickly and safely and how quickly is quickly?
I suspect that one of these days there will be an injury and a law suit.
Insurance companies and laws will have to be adjusted and I'm sure that rates will change as well (my guess is higher rates).

I recently read that one of the biggest impacts on self driving vehicles will be the transportation industry a.k.a. truckers.
Some truckers in North America drive anywhere from three to twenty four hours or even more with a single load.
Not only that some trucker delivery perishable goods/materials, others dangerous goods/materials while others delivery heavy machinery and auto parts or even just UPS/FedEx packages.
I can see this industry becoming even more streamlined and self driving vehicles becoming very disruptive for the industry.

However, we still see cars from the 1960s on the road today and it's 2017. So does that mean in 2067 we will see cars from 2017 on the road?
If we do then we will there be aftermarket kits to make these cars self driving?

[Paul Clapham]

Pete Letkeman wrote:I think that the problem with self driving cars will be when there are self driving cars and regular cars on the road at the same time.
Computers are great, but they do not always read the situation the same as a human. Not only that not every human read the situation the same way.
I think that once everyone uses a driver-less car then the rules and situations will have a more predictable outcome.

You're not the only one who had that idea. Have a look at this WIRED article: To Survive the Streets, Robocars Must Learn to Think Like Humans.

[Pete Letkeman]

I wonder about the no win situations that some people cause and face with cars.

One example of a no win would be:
You are driving down a street and a person decides to walk a cross the street right in front of you.
There is oncoming traffic in the other lane and if you stop too suddenly someone will hit you.
Some people would hit the person walking, some people would stop. What should a computer do?

In a situation like the one listed above, people were asked that exact same question with some surprising results.

The majority said that the car should stop...but only if they were the person walking on the street and they were in danger.

Yet if they were in the car, then maybe the car should not stop.

Some people said they car should stop no matter what if they were in the car or not.

But now we come to the point where we try to evaluate lives and that is not easy.
What if the:

person crossing the street may have a gun and they have just shot at someone (maybe you). Now what?

person is crossing the street because they are talking on their phone and not paying attention?

crossing the street has a child with them. Now it's your passengers lives vs a parent/guardian and a child.

The list goes on and on, so that decision structure could be very intricate.

I'm fairly certain that we have all seen people do stupid things with regards to crossing a street or in the parking lot.
I'm not to sure what the win is, but I know that there is a lot needed for self driving cars.

[Jeanne Boyarsky]

And then there are scenarios were algorithm predictability hurts. Some people are worried about humans "bullying" the self driving cars by repeatedly going in front of it. Because the car would have to stop, it is safer. Whereas with a human driven car, there is the chance the car won't stop so going in front of it is risky.

[Jeanne Boyarsky]

Pete Letkeman wrote:However, we still see cars from the 1960s on the road today and it's 2017. So does that mean in 2067 we will see cars from 2017 on the road?
If we do then we will there be aftermarket kits to make these cars self driving?

Aren't cars from the 60's rare? Also, weren't they better made than today's cars?

[Paul Clapham]

Jeanne Boyarsky wrote:Aren't cars from the 60's rare? Also, weren't they better made than today's cars?

Yes, they are. (Except in Cuba.) And no, they weren't.

But Pete has a point. If after some number of years the roads are given over to self-driving cars only, where does that leave poor people who can only afford an old beater?

[Paul Clapham]

Just the other day I read about a study they did, I think, in Portland. The law there is that vehicles have to stop at uncontrolled intersections (no stop sign, no traffic light) if a person is standing there waiting to cross. Of course this rarely happens. However the study showed that drivers stopped more often for white people than for black people, and more often for men than for women. I'd be interested to see if the self-driving cars learned that behaviour from their trainers.

[Jeanne Boyarsky]

Paul Clapham wrote:But Pete has a point. If after some number of years the roads are given over to self-driving cars only, where does that leave poor people who can only afford an old beater?

I think that car ownership will decline once self driving cars are accepted. Because you could rent a zipcar type thing as a self driving car. And it doesn't need to park. It can go to the next appointment or meet you later or whatever.

[Paul Clapham]

Jeanne Boyarsky wrote:I think that car ownership will decline once self driving cars are accepted. Because you could rent a zipcar type thing as a self driving car. And it doesn't need to park. It can go to the next appointment or meet you later or whatever.

You could rent some time in a self-driving car. But the guy living in a trailer outside Chetwynd who has to get to his job in the gas fields which are a two-hour drive away probably can't. But maybe I'm wrong -- maybe the self-driving cars will spread rapidly to the remotest boondocks once they start appearing.

[Campbell Ritchie]

Paul Clapham wrote:. . . the guy living in a trailer outside Chetwynd . . . probably can't. . . .

Maybe because of the long walk to the hire place. For such car hire to catch on, it must be possible for everybody to pick up a car very quickly. If there is a self‑driving car within 3 min walk, he probably will take it.

[Pete Letkeman]

Jeanne Boyarsky wrote:Aren't cars from the 60's rare?

Okay, then let's replace 60's with the 80's. It does not change things too much. I know that some people really like their car, not mater what year it was made and whatever car they have they won't give up on until it's driven to the scrape heap.

Jeanne Boyarsky wrote:Also, weren't they better made than today's cars?

Yes and no. I have no illusions, we live in a disposable society. It's not just cars, washer machines, dryers, refrigerators, etc all appear to last longer from the 'good old days'.
However, those machines from the 'good old days' where not always safe and they were not always efficient.
Aside from that if one of those items breaks down you may have to spend a lot of money to get it fixed.
Plus, many of us choose remember the 'good old days' favorably and choose to forget the 'bad old days'.

Ease of access will most likely be a determining factor for a lot of people with regards to renting a ride.
Some places in Ontario, Canada for instance are a good 30 minute drive or more from a population center of 2,000 people and even further from a population center of 40,000+ people.

I wonder how much self driving vehicles will affect farming of things like corn, wheat, etc.

[Campbell Ritchie]

I suppose a self‑driving car can drive to where you live. That gives a new meaning to the term self‑drive car hire.

When I was little, I found out from reading posters in the shop window that there were places who did car hire, and there were two sorts: self‑drive and radio controlled. I never worked out how on earth a car could drive itself, but it was obvious that the other sort needed somebody with a joystick behind the office for the radio control.

[Jeanne Boyarsky]

Paul: Good point about rural areas having less access.

Campbell: The car could drive to him. But I agree with Paul that it is a long drive for the car so not cost effective.

Pete: Interesting. (I don't currently own a car so didn't know some of this.)

The mayor of NYC is currently fighting against the test. I hope it goes through. Testing is important for progress.

[Pete Letkeman]

Jeanne Boyarsky wrote:Pete: Interesting. (I don't currently own a car so didn't know some of this.)

NYC has mass transit and I suspect that the mass uses that transit, taxis or Ubers for the most part.

It would be great to not have to own a car, but I have no mass transit where I am and I have family as close as a couple of KM away and as far as 300 KM away.
Which means that to see my family I drive anywhere from five minutes to three hours.

[Campbell Ritchie]

Jeanne Boyarsky wrote:. . . I agree with Paul that it is a long drive for the car so not cost effective.

With the current generation of electric cars, it would run out of juice before the whole journey is completed.

. . . I don't currently own a car so didn't know some of this. . . .

The same applies in London; because of congestion, better public transport and the great difficulty of parking anywhere, many people never buy cars in the first place. Is there anywhere else where it is common to avoid buying cars?

[Pete Letkeman]

Campbell Ritchie wrote:With the current generation of electric cars, it would run out of juice before the whole journey is completed.

Did someone say electric cars? I have the same hesitation as some of you. How far can you go before the battery dies?
I would not like to be on a road going 110 km/h (~69 miles/h) and have the car stop working due to battery failure.

Campbell Ritchie wrote:The same applies in London; because of congestion, better public transport and the great difficulty of parking anywhere, many people never buy cars in the first place. Is there anywhere else where it is common to avoid buying cars?

Yes, I have found that this is somewhat common with the big cities like Toronto Ontario and Ottawa Ontario. These big cities can afford to put in subways/tubes and the ridership is there.

Jeanne Boyarsky wrote:The mayor of NYC is currently fighting against the test. I hope it goes through. Testing is important for progress.

Usually testing is leads to change. However not everyone is in favor of change or progress.

[Jeanne Boyarsky]

He's not worried about progress. He's worried about the self driving cars hurting someone.

[Paul Clapham]

Pete Letkeman wrote:I wonder how much self driving vehicles will affect farming of things like corn, wheat, etc.

Farmers already have equipment which uses GPS to produce precise rows for planting corn etc, so it doesn't seem like a big step for that equipment to drive itself. Likewise driving a loaded truck full of grain to the grain elevator is a repetitive task which could be automated.

[Campbell Ritchie]

It's the pedestrians who walk out in front of you that can't be automated. I have seen the phenomenon before, so I knew what to expect, but my passengers were really frightened. Nobody was struck. Nobody was hurt. Not like a week ago round here where one person was slightly injured and somebody else very severely injured like that. A driver has been arrested and been up in court about that incident.

[Randy Maddocks]

Not to rain on Google's or Tesla's self-driving car "parade", but with advantages come some serious disadvantages:

I came across this tidbit on wired.com:

There’s a Whole New Way to Confuse Self-Driving Cars
Car and Driver reports that security researchers at the University of Washington confused autonomous cars into misidentifying road signs, and they did it with simple stickers they made a home computer. The researchers put stickers on road signs and managed to convince the car’s image-detecting alogorithms that they were seeing, say, a speed limit sign instead of a stop sign. Such misidentification could be disastrous in a real-world driving environment. source

Some would even argue that hackers are the real roadblock for self-driving cars: hackersandselfdrivingcars

Knowing at any moment a self-driving car I am in could be hacked and have who-knows-what be done to it does not give me the warm and fuzzies.

Human error when driving is an obvious risk we know is all too prevalent, and has been since the first vehicle rolled off the assembly line at Henry Ford's plant at the turn of the last century. I wouldn't argue that, but I guess having that element of human control when driving (even with it's less than stellar track record) is still my preference by far.

One advantage humans have over computers, or machines in general, in my opinion, is that humans have that innate ability to reason. We generally don't think in terms of "black and white", or binary 1s and 0s like computers do. It's one thing to apply absolute logic in an application, like a bank app, for example, but to try and apply it in a real-world driving scenario seems daunting at best. There is still no technology out there that comes close to the power or capability of the brain (not to take away from the impressive technology out there, for example like the systems that help pilots navigate a huge airplane).

My 2 cents worth...

[Campbell Ritchie]

That article you quoted wrote:Why anyone would want to hack a self-driving car, knowing that it could result in a death?

I am quite sure there are lots of people who are bad enough to do something like that just to see what happens.

Road signs shou‍ld be the least of their problems, since it is possible to link a road sign database to GPS. Of course, that can doubtless be tampered with, too.
Since this is MD: maybe somebody has been round with a felt pen and changed all the 30 signs round our way to read 80. At least that seems to be the speed some people travel at.

[Randy Maddocks]

Campbell Ritchie wrote:...changed all the 30 signs round our way to read 80. At least that seems to be the speed some people travel at.

I can say that not a day goes by that someone has passed me on the road like I am sitting still (and I am not one to drive slow myself...).

To me it's so sad that there are people out there capable of something as sadistic as hacking into a self-driving car and sending the innocent passengers plummeting down some embankment, or into the path of an oncoming 18-wheeler. That, to me, is what can take all the excitement and promise out of something as innovative as self-driving technology.

[Paul Clapham]

Campbell Ritchie wrote:Road signs shou‍ld be the least of their problems, since it is possible to link a road sign database to GPS. Of course, that can doubtless be tampered with, too.

Yes. There's GPS jamming (deliberate blocking of GPS signals) and GPS spoofing (broadcasting of incorrect GPS signals). Both are already in use for various activities which are not entirely illegitimate.

[Pete Letkeman]

Interesting, this was just published https://www.theverge.com/platform/amp/2017/10/23/16510696/self-driving-cars-map-testing-bloomberg-aspen

A new digital tool tracking how autonomous vehicles are being deployed and tested across the world went online today, and it’s an interesting — and honest — snapshot of where we are right now with this new technology. The Global Atlas of Autonomous Vehicles in Cities, a joint effort between Bloomberg Philanthropies and the Aspen Institute, shows which city governments are testing AVs, and more importantly, it shows how few cities are preparing for the onslaught of self-driving cars that is expected in the next decade.

[salvin francis]

Here's an interesting point ...
What if a hotel/shop at an intersection/ or a T-junction displays a sign that's similar to a Stop sign in it's window. If the car is not able to differentiate between a stop sign besides the road vs a similar sticker on someone's window pane, it can cause the vehicle to stop for a couple of seconds at that junction. The hotel owner gets a lot of potential customers subliminally reading his advertisement since they all stop at his junction for a couple of seconds. Even billboards can misuse this. An advertiser can design a bill board with a picture of a stop sign or similar.

Better yet ... a fake signal with Red light always 'on'. It just needs a couple of leds and a convincing picture. A stop can cause a 3 second pause, a red signal can actually cause a minute or more (given that the owner is kind enough to include a green light too)

[Jeanne Boyarsky]

Well, that's scary. And way worse than someone "bullying" the car by walking in front of it!

[Campbell Ritchie]

And dangerous too. There are some countries where it is illegal to display a road sign the same colour as official road signs.

[Pete Letkeman]

Paul Clapham wrote:I'm waiting for them to start testing in places where there's sometimes snow on the roads.

I think that your wait may be over Paul

Waymo plans to test its self-driving car technology on the cold, icy roads of the greater Detroit region this winter (2017), the company said Thursday (October 26, 2017).
...
The Alphabet-owned company has been conducting cold weather testing since 2012. To achieve Level 5 automation -- fully autonomous driving -- vehicles need to be able to handle all environmental conditions that can be managed by a human driver.

http://www.zdnet.com/article/waymo-to-test-self-driving-cars-in-michigans-winter-conditions/

[Paul Clapham]

Yeah, I just saw that article this morning. Good on them.

[Pete Letkeman]

And now

Waymo, the Alphabet self-driving car company, now has cars driving on public roads in the Phoenix metropolitan area with no one in the driver's seat. Waymo CEO John Krafcik plans to announce the news today in a speech at the Web Summit in Lisbon, Portugal.

You can read the full article here
https://arstechnica.com/cars/2017/11/fully-driverless-cars-are-here/?amp=1

[Randy Maddocks]

Jeanne Boyarsky wrote:...Also, weren't they better made than today's cars?

To your question Jeanne I would argue that, in one sense, yes, cars from the 60's were better made. Instead of being made of mostly plastic, alloy and other "cost saving" [read cheap] materials, they were made of actual steel.  I would say I'd feel safer driving the '67 Chevy Impala my dad used to have than my '17 Toyota Highlander SUV (as nice as it is), even with all the sensors cars nowadays have that serve to help us avoid potentially dangerous situations on the road.

My 2 cents on that...   

[Campbell Ritchie]

I think Ralph Nader would have things to say about that. There are all sorts of things like shapes (interior and exterior) which make a car much more (or less) dangerous in the event of an accident.
The 1960s vehicles almost all used crossply tyres (which are only really suitable for two‑wheeled vehicles), which by the 1970s had been replaced by radial ply, giving much better grip whilst cornering.
Controls, including steering and brakes, work much better nowadays giving better handling and less risk of skidding, etc.
The “cheaper” materials are lighter, so there is less kinetic energy to be dissipated in the event of a collision, and their greater deformability absorbs some of that kinetic energy. Even so, recent rules about cages in the bodywork of cars have made them weigh nearly double what they weighed forty years ago.
The 1960s car needed lots of steel because it would rust away to nothing at the first sign of salt for de-icing the tarmac.
The number of road deaths in this country hovered just below the 8,000 mark for decades and declined to just under 2,000 annually. I suspect a large part of that was making seatbelt wearing compulsory in 1983. I may be wrong on that last point.

I think I am going to disagree with RM on that point.

[Bear Bibeault]

Randy Maddocks wrote: in one sense, yes, cars from the 60's were better made.

In the sense of structure, yes. Very much so. But that doesn't necessarily equate to safety. Todays cars, even if made of flimsier materials are much safer (ABS, seat belts, airbags, crumple zones, etc).

And when it comes to engines, today's cars are far more reliable.