News   Dec 20, 2024
UrbanToronto Year-End Poll: Vote on the Best Buildings of 2024, presented by Sapphire Balconies
 1.1K     5 
News   Dec 20, 2024
UrbanToronto's Social Media Presence Skyrocketed in 2024
 863     2 
News   Dec 20, 2024
Mattamy Homes Records A Year of Accomplishments
 1.7K     0 

Self-Driving Vehicles/Autonomous Vehicle Technology

So far the insurance and liability issues haven't come to the forefront, but I expect they will become very complex.

In Australia, it's already possible to get discounted auto insurance if the vehicle is equipped with the new adaptive technology. The presumption is, the new technology will lower claims costs compared to traditional drivers.

Tha's good for the moment, but as the technology becomes the norm, and human errors disappear, I don't expect insurance companies to remain satisfied with a pure no-fault, actuarially-based claims process. Insurers will want to know if claims are impacted by flaws in the technology, so they can pass costs back to the makers. I predict that they will want the 'black box' data for every accident. Some may not realise, that black box data exists in today's adaptive technology vehicles and when you buy a car with the new technology you are consenting to releasing it if the maker asks for it.

When you look at the airbag and emission scandals, you realise that auto makers go to extreme lengths to conceal flws in their products. Will they own up to failures in the new technology? I doubt it. Proving a fault in deep learning technology will be far more difficult than proving a flaw in an airbag. I can see insurers demanding some relief to level the playing field.

Perhaps the law will change, making the car designers liable for all accidents unless they can prove the flaw that caused the accident came from somewhere else. I can see the car makers issuing very complex instructions (even more than todays' owners manuals) disavowing liability if the vehicle is operated under 'extreme' conditions where the technology is known to be iffy.

- Paul
 
Jonny5 said:
Would you accept paying for that service so it could be transferred between vehicles? Would you accept paying $1,000 when buying a used vehicle and have your old car AI transferred to the new one?
Click to expand...

I think you're confusing Autonomous with Artificial Intelligence.

Cars will never have true AI computers to control their systems, they will never be sentient. Autonomous vehicles make decisions based on programming. If they are given the ability to learn and adapt it would be wasteful if their collective experiences aren't shared with a central computer to which all autonomous vehicles are linked. These systems would not require you to "transfer" AI between vehicles, the only thing worth transferring might be your movement history and preferences, all of the other learned behaviour would be instantly accessible through a connect with a central computer. For instance, your smart phone today has all your downloaded apps saved on it, when you buy a new phone you don't need to transfer these apps to the new phone, you can simply download them again from your app store account. Maybe you were too ambiguous in your use of the term "transfer" if that's what you meant, but as I have stated the most fundamental functions of an autonomous vehicle need not be transferred.
 
Jonny5 said:
Manufacturers, insurers, and governments will embrace car driving AI because it's a gold mine to extract cash from you. Cars with autonomous driving AI will be more expensive than what we have now. Everything will be licensed, leased, rented, and monthly billed by usage. You will not own anything. In the long run, it will cost you more than ever to drive.
Click to expand...
Yes -- and then which will force a percentage of people to send their car for hail service.
(aka your car behaves like an autonomous taxi/Uber/ZipCar while you're not using it).
This would offset costs somewhat.

Whether run by the likes of Uber, ZipCar, TTC, Ford, Blue Line Taxi, etc -- no difference between them if the car can self-chauffer itself to their hail/renter/client...

Especially if autonomous hail capability is a part of all self driving vehicles eventually (e.g. 30 years+ from now). As an optional opt-in. If even just 5% of people enable it, that's a lot of ride hailing capacity. It would have to be when Class 4 autonomous cars becomes practical and when also sufficiently reliable in rain/snow, too.

Perhaps car maker keeps 30% revenue, you keep 70% credited towards your car usage bill (monthly autonomy/AI bill, monthly car lease bill, car cleaning service, maintenance, etc).

Cities would discourage (financially penalize) empty cars going long distances, so it's cheaper to hail a nearby car that's not currently being used by its owner. Policies would penalize congestion, to the point where you've got an equilibrium of roads moving more people using fewer owned vehicles.
___________

More conjecture of what might happen if optional hail autonomy is someday built into all cars:

Opt-in, of course. With dormant hail software built into all cars sold by big name companies, activated by users whenever they feel like reducing costs. Instead of parking your car for the day while at work, you'd send your car off to hail service while you're working. Or you've lost your job or need to pay for your kid's education.

And the cars would be able to chauffer themselves to maintenance & maid services, helping mitigate cleanliness/maintenance/time-management concerns of extra use during hail service. And of course, vomit incidents would automatically be charged to the hailer's bill & cleaned fully before the car returned to you at the end of your workday or weekend (prescribed need of car)...

If just a mere 5% of owned cars are doing autonomous hail service, that would fill lots of nearby parking spots, nearby driveways, and nearby streets with hailable cars -- drastically shortening waits for hail service.

And reducing car-ownership desire by a larger number of people. With proper policies in place, an equilibrium occurs -- enough (but not overloadingly-much) people own cars and are incentivized to opt-in -- to the point where hailing in the suburbs is as quick as waiting for a subway train in a downtown Yonge station. And expensive enough to go solo long distances, that it's cheaper to be dropped off at the nearest mass-transit station, or board a pooled minibus.

After a critical mass of some point, a possible self-feedback effect could occur -- more people find hail service so easy/convenient (e.g. typical sub-2-minute wait times due to huge numbers of hailable cars, even in suburbs). Which may then lead to less desire to own a car, thus reducing number of cars on the road per capita (over time).

And might over the long term (if properly done) -- decongest the streets a little bit assuming algorithmic (software) & governmental (tax) checks-n-balances were included to prevent empty cars from chauffering themselves long distances -- incentivizing nearby hails, carpooled and minibus hails being cheaper, penalizing long empty car trips, etc. Obviously, you need to penalize congestion-generating elements (e.g. empty cars going long distances needs to be made a very expensive proposition), for these benefits to be realized.

In 50-70 years from now, once hailable AI is widespread in nearly every vehicle, this might lead to an equilibrium of:

-- Solo cars that are not hailed (richer owners)
-- Hailable solo cars (costs more)
-- Hailable carpooled cars / vans / minibuses (costs less)
-- Even full size buses (discrete route) automatically creating new bus stops on-the-fly en-route based on demand (surge of people waiting at one location) -- (cost the least)
-- New virtual bus routes created on the fly. (Large numbers of carpools over the same route that doesn't have a bus -- creates analytics for a bus to be reassigned to this route on the fly). Bus stops would be virtual dots on your device.
-- Efficient connections between hails and Subway/LRT/mass transit routes.
-- Semi-discrete public transit where vans / minibuses efficiently deviated up to a few blocks based on demand (e.g. surge of people 2 blocks off) -- (cost even less)
-- Various forms of fare integration that accomodates all the above simultaneously
-- Bus stops may become legal generic hailstops too
(who knows, government would get into the action too, either revenue via the car companies themselves, or revenue from each hail that stopped there, or owning a lot of hailable vehicles, a massive fleet of faster GO DIAL-A-BUS on steroids, etc)
-- Density will determine what form of transit wins. Mass transit (subways, high-capacity LRTs, etc) will still rule supreme for dense corridors/downtowns, while rural public transit will probably more efficiently be served by hailable minibus carpools than by fixed-route buses.
 
Last edited:
sixrings said:
2030's UBER
Click to expand...
Or, maybe, they program the app that is licensed to Ford for preinsfallation into their autonomous cars that they sell.

I think it will be closer to 2050s, to make it smart enough & safe enough in blinding snowstorms, with the 3am bar crowd and keeping that automatically separate from unaccompanied kids.
 
mdrejhon said:
Or, maybe, they program the app that is licensed to Ford for preinsfallation into their autonomous cars that they sell.

I think it will be closer to 2050s, to make it smart enough & safe enough in blinding snowstorms, with the 3am bar crowd and keeping that automatically separate from unaccompanied kids.
Click to expand...
I just want to be alive and have the last car I own be autonomous
 
zenmeister said:
http://www.carbuzz.com/news/2016/7/...Make-Car-Ownership-A-Part-Of-History-7734595/

I'm also interested to see the application of this technology on LRTs
Click to expand...

Self-driving on a regular, well defined, clearly signed route (LRT or busway) is a piece of cake compared to self-driving in "free space".

One major change that will be needed is a uniform, consistent, rigourous approach to roadwork. In the future there will have to be much clearer, computer-recognizable technology and signage for roadway shifts or lane closures. (There is a set of standards for this already, but in practice, construction crews frequently set out their pylons by the seat of their pants. )

Just imagine the average road crew setting up in front of that bus, and moving traffic onto the shoulder for 100 yards. For that matter, think "what would a self driving vehicle use to find the road path?" as you navigate the next few construction zones or lane shifts that you encounter.....

I went to an event last week where the (temporary) parking lot was a farmer's field. There was the usual squad of baton-waving volunteers guiding people into the parking 'spots'. Again, I wondered "how would a self-driving car handle this?"

- Paul
 
How soon do you reckon that the City of Toronto will ban driving? Once road deaths become 100% preventable (thanks to autonomous cars) I can't imagine the City permitting human-driven cars for much longer. In fact, we'll probably ban driving long before we reach our Vison Zero "targets" for pedestrian fatalities.

I say by 2036, Council will at least be seriously considering this. Likely before the DRL is even open :)
 
Last edited:
crs1026 said:
So far the insurance and liability issues haven't come to the forefront, but I expect they will become very complex.

In Australia, it's already possible to get discounted auto insurance if the vehicle is equipped with the new adaptive technology. The presumption is, the new technology will lower claims costs compared to traditional drivers.

Tha's good for the moment, but as the technology becomes the norm, and human errors disappear, I don't expect insurance companies to remain satisfied with a pure no-fault, actuarially-based claims process. Insurers will want to know if claims are impacted by flaws in the technology, so they can pass costs back to the makers. I predict that they will want the 'black box' data for every accident. Some may not realise, that black box data exists in today's adaptive technology vehicles and when you buy a car with the new technology you are consenting to releasing it if the maker asks for it.
Click to expand...

For insurance costs there are three benefits for full automation.
1. The first is that 90% of accidents are avoidable. Even if the AI predicts 90% of these avoidable accidents that means an 81% decrease in the number.
2 In the GTA it is estimated that there is $1.2 billion of insurance fraud. Camera's on cars will become ubiquitous making fraud less of a concern
3. For those accidents that no one is truly at fault the AI can manage the collision to reduce the amount of damage done to people and the car. The remaining accidents become less costly.
There are estimates that the cost of insurance will drop to below $250 (some people say less than $100 but I'm not betting on that)

In fact, Ford & GM are indirectly predicting the death of auto ownership with automated cars and ride sharing. They predict that they will own the cars, not anyone else. They will take the risk of AI failures. We will just hail a car and pay about $0.60 a km (the cost to build and maintain the car). Ford and GM make a profit per km. No car dealers inflating the cost of the car. Uber not skimming 20% off the top. And no money to the Uber/cab driver.

It costs us around $0.60 a km right now for our own car. Ford and GM just want to make money off the entire value chain. Sounds good to me. i get a car for the same cost and I don't have to pay a fixed amount, maintain it nor pay for parking. The car will just find the next person that needs a ride.
 
muller877 said:
It costs us around $0.60 a km right now for our own car. Ford and GM just want to make money off the entire value chain. Sounds good to me. i get a car for the same cost and I don't have to pay a fixed amount, maintain it nor pay for parking. The car will just find the next person that needs a ride.
Click to expand...

This is a great model for urban areas where the auto is functioning as "transit". I wonder how it will play out in suburban or rural travel. There are too many occasions where I take my car somewhere for days at a time. And where I want to stash things in the trunk so I don't take them with me each trip..... extra shopping bags, my work boots, my camera tripod, a dozen water bottles, a change of socks.

Of course, the price differential may convince me to suck it up and not own a car. But not all car travel is a pick me up, drop me off, go on to the next user proposition.

- Paul
 
I can see that you can still own a vehicle for your exclusive use - you will just have to pay more for that experience (relative to a time/use-based ownership system). Also I can't see driving fully go away - but insurance rates can be stratified by the level of automation required (fully automated, partial automation, no automation) - the more autonomy you want, the higher the insurance you will pay.

AoD
 
crs1026 said:
This is a great model for urban areas where the auto is functioning as "transit". I wonder how it will play out in suburban or rural travel. There are too many occasions where I take my car somewhere for days at a time. And where I want to stash things in the trunk so I don't take them with me each trip..... extra shopping bags, my work boots, my camera tripod, a dozen water bottles, a change of socks.

Of course, the price differential may convince me to suck it up and not own a car. But not all car travel is a pick me up, drop me off, go on to the next user proposition.

- Paul
Click to expand...

You are correct. The cars would be idle for longer and would have to drive further to pick up the next ride. Both mean the cost will be higher the further you get out of urban areas. Also it will take longer to hail the ride (you will have to reserve far in advnace).

However, there should be some sort of discount for longer drives at highway speeds (better mileage and less wear and tear per km), It won't offset but at least something.

Think of Toronto for transit (and you can apply to the suburbs). Buses (back in 2013 where I found some numbers) drove 129M km. There were 239M bus passengers.

Assuming the cost is $1.50 per km (bus driver, fuel, maintenance, depreciation and financing) the cost to run the fleet is $200M. If we switched to a van pooling system at$0.60 per car (max 8 people per van, average of 3 people so $0.20 per person per km) and assumed 4 km to get to high order transit it turns into a cost of $190M. Direct to front door service with the same cost!

Basically we create a true fast grid network (every 2 km) for transit and rely on automated vehicles to feed it.

transit will also benefit since the LRT can communicate with the cars and lights. True transit priority.
 
Liscensing rates amount young drivers are in a free fall: http://www.theatlantic.com/technology/archive/2016/01/the-decline-of-the-drivers-license/425169/

That data is for the United States. I don't imagine it's young people here in Toronto are any more enthusiastic about driving. I'm curious to see how quickly it plummets here, once young people figure out they can text, sleep and eat while "driving" in an autonomous uber, all while costing a fraction of the price of private car ownership.
 
You must log in or register to post here.

Back
Top