I think Albaby the cat is out of the bag. We are heading for driverless cars and cars without pedals and steering wheels, it is only a matter of time. I will always want them in my cars but I can see future generations not wanting them.
1 Like
As Goofy points out, it depends on how much it costs. If a self-driving car is significantly more expensive than driving a conventional car yourself, adoption of self-driving cars will reach some portion of the market and then stop. It might simply be a luxury product. Unless mandated.
2 Likes
Possibly. But it might be a while before the federal government willingly gives up their ability to make those who want to offer control-less cars get a regulatory approval from the feds, not just the state. They certainly have the ability to keep them in a bit of a liminal state, where they can be owned by a fleet operator but not sold to the general public.
You think the government is going to mandate self driving cars?
Geez, people are still fighting about having a $30 seat belt added, you believe obsoleting half (or more) the driving fleet is going to be popular?
1 Like
No, not directly and not anytime soon. But they will design and build roads that can only handle automated vehicles. At first it’ll be a trickle of roads, but later it’ll be more and more such roads. At some point, driving yourself will be a pain and will take much longer than using the quick and efficient automated roadways. Kind of like how cars squeezed horses off the roads over the early decades of the 1900s.
Who is fighting over the $30 seatbelt? Heck, I don’t even see fighting over the rearview camera and display that was JUST mandated in 2018!
This is why it’ll take many decades. You can’t obsolete half the vehicles on the road!
I think cars will fly before that happens.
While it is possible, I don’t think that’ll happen because while the amount of energy needed to move a vehicle on the surface is relatively high, the amount of energy needed to overcome gravity AND move the vehicle through the air is substantially higher. In the end, cost overcomes nearly everything. So there may someday be flying private, or semi-private vehicles, but they will still be quite expensive and relegated only to members of the upper few percent of wealth/income.
About the same price as a Dodge Ram 3500
Pretty cool
Noted.
This argument would be valid if the autonomy system were truly awful. One needs to take into account the potential failure rate at the worst time, not assume that the whole fleet will crash at the same time.
A straw-man argument.
Why? He likes over the air updates? The remote drivers are occasional, exceptional, to handle unexpected conditions. They might decide to call a tow truck, ambulance, the police, or the fire brigade as the case might warrant instead of driving the failing vehicle.
Miss it I did. Perfection is the enemy of the good.
The Captain
1 Like
I’m not sure what you are arguing exactly. It seemed like you were defending the 99.9% number (which is “truly awful”). If it is really 99.9%, then 1 out of 1000 cars might hit a scenario they can’t handle. If you have 1,000,000 cars driving in the morning to ferry people around (to work, to school, to shopping, to airports, from …) then if 1 out of 1000 could have an issue, you need up to 1,000 remote drivers, and the infrastructure to handle all that. That would be very costly and since those 1000 remote drivers are doing nothing most of the time, it would be a tedious and annoying job … often resulting in low performance.
I’m not sure what OTA updates have to do with remote drivers? The beauty of OTA updates is that you can develop and test new software and then rapidly deploy it to all your vehicles on the road. If you require each vehicle to come to the dealer for updates, as many auto companies do, it would be terrible. You might get 2% of them the first month. Then when word spreads, you might get 30% of them by a year. But most of the folks won’t ever get it because they choose to go to independent shops for their annual or biannual service (because the dealers charge too much, or are too far away, or are inconvenient, etc).
The perfect word for OTA updates is “scale”, you can scale it across the entire fleet relatively easily. Remote drivers (at 99.9%) don’t scale, roughly for every 1000 cars you sell, you need to add a remote driver. It’s a linear relationship. But if you can improve to 99.99% (ironically via an OTA update), you can cut your remote drivers roughly by a factor of 10!
1 Like
I did not pay attention to the number of nines required. I’ll leave that to practitioners and providers to figure out. My interest is not in the details but in the overall logic of the technology.
The Captain
1 Like
But in reality can you? Would not a company have to allow for 1,2 or even 3 standard deviations from the average to be safe?
And, since the number of cars in use is constantly changing, the best a company could do would be the estimate the number of people they needed. I am sure AI could help with that but there is no way it could accurately predict the future need.
Your math only works if you have a specific definition of what 99.9% means. In one such definition, where 99.9% means one in a thousand will have an issue once a day it isn’t too bad…each car, on average, has an issue about every 3 years.
In another definition of what 99.9% means, a car makes a thousand decisions and gets it wrong 1 out of a thousand and needs help. This would be quite bad if 10 decisions are made every second, so in100 seconds you’d have a problem for each car.
If 99.9% means 1 error in 1000 miles and a car goes 100,000 miles in a year (probably 2x too big), you have an issue 100 times per year or once every 3 days or so.
Let’s look at that last one. If a remote operator can handle most issues in 5 minutes, but one complex issue takes an hour, they could handle about 60 minor issue and one major issue per day with some breaks. So for 1000 cars you’d need about 5 remote operators.
(Of course you’d need more remote operators during peak driving times and less in off hours)
Mike
1 Like
I don’t think a company can ever use remote drivers for safety. If an AI is going to actually be a Level 4 or 5 driver without someone in the driver’s seat, it has to at least be smart enough to maneuver the car to a safe stop on its own in every scenario. The remote drivers can virtually take over to help the car manage after that, but I don’t think any human could consistently jump into a perilous driving situation and figure out what to do in the space of a few seconds.
6 Likes
My free Tesla FSD subscription expired yesterday, so I’m now on standard Autopilot with fewer features.
I didn’t see much difference driving on the Interstate, with the exception that I now have to make the turn at my exit, while FSD made the turn for me. (The navigation system still announced the turn, both by voice command and a blinking arrow on the screen.)
intercst
1 Like
That’s actually how the trucking FSD systems work. If the truck sees something ahead it doesn’t understand, it pulls into the highway breakdown lane and waits for the human operator to sort out the image on the camera.
I’m pretty sure Waymo works the same way, as long as it doesn’t hit a human body in the street left by another vehicle.
intercst
To be fair to Waymo your link was to a GM Cruise accident
Mike
There are A LOT of differences. Don’t be fooled. It’s an entirely different stack. I have a road that I drive on at least once a day, and the behavior in autopilot (aka adaptive cruise control with lane keeping) is WAY different than the behavior in FSD. There is a kind of split where an additional lane for turning gets created on the left side. In autopilot, the car briefly aims for that lane, but midway it corrects back to the proper lane, and it only does this sometimes, and the severity varies. In FSD, it remains smoothly in the proper lane every single time (even with FSD versions from a few years ago). Also autopilot will not switch lanes on the highway to pass the slow truck in front of you. And autopilot probably won’t swerve into another lane to avoid a car that suddenly intrudes into your lane. And autopilot won’t even attempt to avoid potholes. And autopilot won’t hug the opposite side of the lane if a large truck is next to you. And many other differences (other than the lack of following the NAV route of course).
Yes, I’m pretty that NAV is a completely separate system when running autopilot (fancy “cruise control”).
1 Like
They can pry my steering wheel from my cold, dead hands.
By they, do you mean the government? We don’t even have enough money to maintain the infrastructure we currently have, how will they build new infrastructure?
Agreed, and that is a long way out.
And…it takes a lot more energy if the passenger is obese. Could work for rich and thin people with short commutes.
Pretty cool indeed. But, they seem like toys, not reliable transportation. Over the next 100 years, maybe technology will improve, and prices will come down.
I’m thinking a Mad Max scenario is far more likely.
1 Like
