Autonomous Cars

Breaking news, useful data or technical highlights or vehicles that are not meant to race. You can post commercial vehicle news or developments here.
Please post topics on racing variants in "other racing categories".
Greg Locock
Greg Locock
235
Joined: 30 Jun 2012, 00:48

Re: Autonomous Cars

Post

As usual it depends on definitions. AI has been expanded to include machine learning (ML) and hence Neural Nets (NN). NN are commonly used for target identification in vision systems. So yes, AVs do use AI because they use NN.

roon
roon
412
Joined: 17 Dec 2016, 19:04

Re: Autonomous Cars

Post

Phil wrote:
30 May 2019, 07:09
roon wrote:
29 May 2019, 23:48
No need for hyperbole, Phil. They're claiming to be able roll out full self driving, I believe they mean level 5, by sometime within 2020. Taxiing would be an opt-in. Calculating tire and brake wear would be relatively trivial, all things considered. That would be part of the ride cost, which would comprise, logically: profit to Tesla and profit to vehicle owner which would need to encapsulate consumables. All taxi and rideshare services consider this necessarily.

I miswrote earlier. $0.18 per mile, not per ride, was Tesla's claim at the FSD press conference.

Consider $4k worth of tires consumed within 100k miles. Dwarfed by the potential $18k in profit, shared between Tesla and the vehicle's owner.
I was assuming it was a rate.

Assuming ride costs of 18 cents per mile and Teslas claim that my car could generate me 30k per year... my car would need to travel (paid travel) 166k miles per year (266k kms)... considering servicing costs, maintenance and depreciation, do these numbers make any sense to you?

Just consider for a moment if a Tesla can actually do 166k miles on its originally fitted batteries.

I’ll throw another one in there: cost to recharge the car. Assuming a battery of 100kWh. Here in Europe, we pay around 0.28 Euros per kW. That equates to 28 euros per “tank”. I’m hearing a Tesla uses around 21kW per 100km in the real world.

Lastly; would you want your Tesla driving around servicing complete strangers?
Further factor in that the .18 USD is their cost-to-run estimate. Tack on service fees and the numbers may assuade you.

Regardless, 33k kW, $4k in tires, and .2-.3 USD/kW still renders 4k-7k USD profit per 100k mi. If operaters charge upwards of $.40-$.50/mi, still undercutting valet apps, then you have your $30k/yr, although I don't recall that figure being claimed, despite seemingly being workable.

To the rest of your post, my knee-jerk aesthetic reactions are not relevant. Your skepticism of tech availability and nearness seem somewhat unfounded, which is what I addressed.

User avatar
subcritical71
90
Joined: 17 Jul 2018, 20:04
Location: USA-Florida

Re: Autonomous Cars

Post

in the US, the IRS mileage reimbursable rate has is 0.58c. this is the cost the government has determined is their cost to run their fleet of vehicles and pass that figure on to businesses/individuals for reimbursement of travel costs. i would think a Tesla would be cheaper in both fuel and maintenance costs and charging even 0.58c a mile would be a good profit. those ride-share scooters are like $1 then 0.15c a minute... surely a comfortable car can get a better rate.

joshuagore
joshuagore
0
Joined: 12 Feb 2010, 04:01

Re: Autonomous Cars

Post

I didn't see this in the thread, maybe I missed it.
https://youtu.be/Znz3yelWIBk

I notice a lot of commentary in this thread seems to be missing history. Cars were not better than horses for awhile, but people bought them because they were better for something, but cars unlike horses are just now coming when you call (meaning we gave up some utilities in favor of others, i.e. no eating shitting or training).

Why do we have such an ahistorical understanding about how new wares come to market? You don't solve all existing problems with a new technology or ware, you solve the ones people want to pay to solve with the available technology they are willing to pay for. Right now that is not paying attention on highways. What happens next is still ruled by the above equation.

I sense a lot of resistance in this thread from enthusiasts and drivers who seem to see the goals of future generations not reflecting their own, but keep in mind horse racing still exists, and likely the sporting cars that will exist for 'drivers' may get better as they won't also have to be pawned off to customers who also expect that same car to drive them back and forth to work, cause they don't even own a car for transportation(they ring up their ai fleet)... are horses slower or worse for riding enthusiasts now that they have been relegated to race tracks and trails?

I predict enthusiast cars get lighter, safer, and ultimately more fun as they are built for specific purpose, and don't have the duality required of current vehicle development. I can fantasize, maybe with a bunch of BOT cars running around with sensors all over and crashes going the way of the buffalo, my vintage sports car resto modded with carbon tub(fantasy remember) will be seen like the horse traveling down a country road... annoying to bots but virtually ignored by regulation.

TankMarvin
TankMarvin
3
Joined: 21 Apr 2015, 00:05

Re: Autonomous Cars

Post

AJI wrote:
30 May 2019, 06:37
Nickel wrote:
30 May 2019, 05:34

You stated that it could be argued that they are better based on the ai deciding not to merge. that's clearly your subjective opinion, no? anyways thanks for the downvote.
I said nothing of the sort. In fact, I didn't say anything at all about AI.
There seems to be two different conversations going on here, one about AV's and one about AI.
As far as I know, and I'm happy to be corrected, there is no AI in current AV tech (or anything else for that matter) because it doesn't exist.
Perhaps @Phil can chime-in on this one?
Deep Learning techniques (a subset of Machine Learning which in turn is a subset of the catch-all, often misrepresented "Artificial Intelligence") is a fundamental part of self-driving technology being developed by car manufacturers and technology companies.
It absolutely exists and is in use across a broad range of industries from manufacturing to finance to drug discovery.

An important thing to remember though is that Deep Learning (as part of ML) is a classification/regression capability which helps spot and extract (hopefully useful) features from data - in AV's case this would be images for lower level AV systems which can then warn the driver of an impending situation, or images and appropriate behaviours for higher level AV systems which can deal with such actions themselves.
How you act on those insights with respect to morality is another point entirely, but again the inputs and outputs of certain actions can be used as training data and the system can then infer which actions to take in any given situation using the classifications it has determined from it's understanding of the world around it.

The AVs are not "programmed" in the traditional sense where their capabilities are limited by the expertise, knowledge or talent of the programmers and based on conditional statements such as "if the object walks like a duck and quacks like a duck then it's (probably) a duck"
Instead these system (models) are trained using millions of images and scenarios which represent the world around it. The system's ability to recognise scenarios and objects is improved by iterating over these images, assessing accuracy of the system's recognition and associated behaviour, adjusting the emphasis the system puts on certain aspects of the image and repeating (ad infinitum) to reduce error.

An often used quote on the definition of Machine Learning is
"A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E" - Tom Mitchell, Carnegie Mellon
For humans, the rate of accumulating experience is limited by the time. For these systems, experience can be acquired at the speed of the processor and ultimately just downloaded.

It's very similar in concept to the learning steps any baby/child goes through as certain behaviours are rewarded or encouraged as the system is nudged towards a low-error/high-accuracy state through this repeated training process.

Once the model is at a desired level of accuracy (and repeatability) it is replicated and loaded into many AVs and set loose...

There are obvious risks and issues here
  • The system model is only as good (accurate) as the training process
  • The behaviours and actions associated with recognising certain events still needs to be trained and incentivised in such a way that outcomes are desirable - no point in the AV recognising something in the road and driving into it anyway
  • There are moral judgements (causing the least bad accident) and social conventions (letting cars out of busy junctions... or not) that are very difficult to represent/teach

User avatar
Phil
66
Joined: 25 Sep 2012, 16:22

Re: Autonomous Cars

Post

roon wrote:
30 May 2019, 07:59
Further factor in that the .18 USD is their cost-to-run estimate. Tack on service fees and the numbers may assuade you.

Regardless, 33k kW, $4k in tires, and .2-.3 USD/kW still renders 4k-7k USD profit per 100k mi. If operaters charge upwards of $.40-$.50/mi, still undercutting valet apps, then you have your $30k/yr, although I don't recall that figure being claimed, despite seemingly being workable.

To the rest of your post, my knee-jerk aesthetic reactions are not relevant. Your skepticism of tech availability and nearness seem somewhat unfounded, which is what I addressed.
That's assuming 100% efficiency. E.g. get to work, hit the button to put your car into taxi-mode and off it goes with someone waiting just around the corner. Reality is probably different, as in that your car has to travel first to the not to close pickup point. After the drop off, it will most likely be driving around passenger less, racking up mileage. Last time I used an Uber in Sydney, my driver had to travel 1 mile to get to me to drive me for 3 miles. It's anyone's guess how far he was traveling (or waiting) until his next fare. This will decrease your efficiency substantially. Assuming traveling half a mile to get to a passenger to travel a mile and again a half mile to get to the next one and your efficiency is down to 50%.

Of course, there are a couple of more potential problems: Too much supply to the point your "robotaxi" will just be waiting around, not getting any or little work. Waiting will probably cost in the future too, given parking space is limited. So either your car will just be roaming the streets passenger less racking up miles, adding to congestion and costing you money, or it will be looking for parking spaces ($$$ - doubtful that will be free in the future). Either way, the $30k earnings a year is a pipe dream.

And that's before we get to depreciation. How much do you think your Tesla will be worth of its original price after it has racked up 166k miles in a single year? Would you trade that for the potential 4-7k of earnings?

Question: How expensive is it to replace the batteries in a Tesla? This article suggests a price of about 230USD per kW or 23'000 USD for a full 100kW replacement. Outch!

Having said that, my best guess is that your car will probably just be sitting around anyway with negligible demand for robo-rides. Either there will be simply a huge supply of them that statistically, your car will not get many, or people will realize that they don't really want to have their car drive around complete strangers without any form of control. I know I wouldn't.
Not for nothing, Rosberg's Championship is the only thing that lends credibility to Hamilton's recent success. Otherwise, he'd just be the guy who's had the best car. — bhall II
#Team44 supporter

roon
roon
412
Joined: 17 Dec 2016, 19:04

Re: Autonomous Cars

Post

Phil wrote:
31 May 2019, 15:43
That's assuming 100% efficiency...
Fareless transit could be factored into that figure already. If not charge the difference. Still less than a dollar/mi going by my off the cuff calc. Which is the point--undercut ride sharing apps significantly enough to become quickly competitive.

Note I've only written of the per mile costs and profits.

Tesla are expecting battery pack life 300k-500k mi. Model 3 is built with four sub packs with the intention of only needing to replace one of the four.

They could be wrong but since we're only burdening ourselves with simple multiplication and division here, I doubt it. Consider you are omitting factors. They don't have a reason to tarnish their reputation with this.

User avatar
strad
117
Joined: 02 Jan 2010, 01:57

Re: Autonomous Cars

Post

are horses slower or worse for riding enthusiasts now that they have been relegated to race tracks and trails?
They are not talking of relegating us to track days and such. They are planning on Banning us.
To achieve anything, you must be prepared to dabble on the boundary of disaster.”
Sir Stirling Moss

roon
roon
412
Joined: 17 Dec 2016, 19:04

Re: Autonomous Cars

Post

Until retrofitting becomes cheap that supposed ban may be a ways off. Cars with full by-wire controls may be easiest--add sonar and camera array, AI computer. Older cars would need servos and solenoids spliced in, or perhaps more simply, a roughly humanoid robot to operate the controls. Tesla Model Jeeves.

User avatar
strad
117
Joined: 02 Jan 2010, 01:57

Re: Autonomous Cars

Post

:roll: Not likely mate.. not likely
The cost of retro fitting would in most cases far exceed the value of the car.
To achieve anything, you must be prepared to dabble on the boundary of disaster.”
Sir Stirling Moss

TankMarvin
TankMarvin
3
Joined: 21 Apr 2015, 00:05

Re: Autonomous Cars

Post

And the power required to run the CPUs/GPUs used to drive the AI means it can really only be deployed in electric vehicles.

I know some of the early development cars carried 8-10 servers and probably had around 8-10kW sustained power draw

roon
roon
412
Joined: 17 Dec 2016, 19:04

Re: Autonomous Cars

Post

TankMarvin wrote:
01 Jun 2019, 20:25
And the power required to run the CPUs/GPUs used to drive the AI means it can really only be deployed in electric vehicles.

I know some of the early development cars carried 8-10 servers and probably had around 8-10kW sustained power draw
Tesla claim 72w for theirs.

Image

strad wrote:
01 Jun 2019, 19:44
The cost of retro fitting would in most cases far exceed the value of the car.
Cost of by-wire equipped vehicle retrofit: ______
Cost of manual control vehicle retrofit: ______
Cost of a mass production Jeeves bot: ______
Distribution of all registered road vehicle values at relevant time in the future: ______

You didn't fill out any of the form, strad.

User avatar
strad
117
Joined: 02 Jan 2010, 01:57

Re: Autonomous Cars

Post

All those figures would be astronomical and you know it as well as I do.
Minimum of 40-50,000 dollars.
To achieve anything, you must be prepared to dabble on the boundary of disaster.”
Sir Stirling Moss

User avatar
strad
117
Joined: 02 Jan 2010, 01:57

Re: Autonomous Cars

Post

It's not just the equipment but the labor costs to install it all.
To achieve anything, you must be prepared to dabble on the boundary of disaster.”
Sir Stirling Moss

Greg Locock
Greg Locock
235
Joined: 30 Jun 2012, 00:48

Re: Autonomous Cars

Post

It took them 3 hours to fit a new windscreen to my Escape because they had to recalibrate the lane following camera.