Self Driving Car thread

[chalex]

Then you're demonstrably unfit, by your own admission, to be in charge of a moving vehicle,

Do you live in the real world? Have you seen the half-blind grandmas in your local supermarket parking lot?
I'm all for increasing requirements for drivers but you have to contend with the current reality.
Sounds like maybe you're not in the US where there is literally a percentage of non-licensed drivers on the road just zooming around doing whatever.

 

[Ananke]

I'm all for increasing requirements for drivers but you have to contend with the current reality.

Uh...?

"You're responsible for the safe operation of the vehicle you are driving is not an increase in expectations, nor is it unreasonable - it's a basic expectation across (at minimum, the EU, EEA and UK. I''m not an expert on the USA, but I undeerstand that driving under the influence is a crime there, so probably similar.

That there exist people who abrogate that expectation does not make it unreasonable - it makes those people a social problem to be managed, regardless of whether they are unfit because of fatigue, medical condition, drug use or just being sociopathic to some extent. Differerent countries may spend more or less effort managing that problem, and to greater or lesser effect, but making it easier to be a social problem is rarely a desirable political outcome.

(As I understand the term) level 3 self driving does not change that responsibility - L3 is defined such that the car is capable of all necessary actions to complete a given journey, but the driver must be capable of taking back control on being prompted to do so, with no particular expectation nor guarantee that such a prompt will always provide a minimum safe time for the driver to regain situational awareness.

 

[Shavano]

Essentially correct, which means Level 3 is unsafe by definition and should not be allowed. But here we are.

 

[w00key]

Essentially correct, which means Level 3 is unsafe by definition and should not be allowed. But here we are.

Heh I'm glad you're not the one making the rules. Your opinion does not match mine.


The first and only L3 implementation shifts the liability to the provider, if it fucks up, their insurance pays. Only after getting your attention and you take over, it's you driving again. Enable L3 and pass out in a medical emergency? Car is responsible to turn on the hazards and slowly come to a stop. Not beep boop disable itself 0.5s before a crash a la Tesla.


Mercedes' L3 comes with a ton more redundancy than a standard car. Redundant power, steering, braking and sensors. In a standard car none of that is redundant and a human is supposed to instantly unfuck the situation when ADAS flips out.

 

[sporkman]

Uh...?

"You're responsible for the safe operation of the vehicle you are driving is not an increase in expectations, nor is it unreasonable - it's a basic expectation across (at minimum, the EU, EEA and UK. I''m not an expert on the USA, but I undeerstand that driving under the influence is a crime there, so probably similar.

That there exist people who abrogate that expectation does not make it unreasonable - it makes those people a social problem to be managed, regardless of whether they are unfit because of fatigue, medical condition, drug use or just being sociopathic to some extent. Differerent countries may spend more or less effort managing that problem, and to greater or lesser effect, but making it easier to be a social problem is rarely a desirable political outcome.

(As I understand the term) level 3 self driving does not change that responsibility - L3 is defined such that the car is capable of all necessary actions to complete a given journey, but the driver must be capable of taking back control on being prompted to do so, with no particular expectation nor guarantee that such a prompt will always provide a minimum safe time for the driver to regain situational awareness.

I just find it asinine that if the car detects that you're not paying attention, for which especially on tesla's system there are many false positives, it will intentionally make a crash more likely by allowing the car to drift out of lane. In my modern driving record I haven't had an at-fault accident for close to 20 years. A level 3 system with no restrictions would be perfect for me.

 

[Chuckles]

I just find it asinine that if the car detects that you're not paying attention, for which especially on tesla's system there are many false positives, it will intentionally make a crash more likely by allowing the car to drift out of lane. In my modern driving record I haven't had an at-fault accident for close to 20 years. A level 3 system with no restrictions would be perfect for me.

raises eyebrow

That is a) a disturbingly weak answer against b) a disturbingly weak metric.

The better questions, which you are probably too biased to answer (this is not particularly a condemnation of you personally):
1. How many times in the last year have you gone into microsleep while driving?
2. How many times in the last year have you done other tasks while driving (eating, texting, gotten in a deep conversation with a passenger or third party, etc.)
3. How many times in the last year have you been in extremis while driving (i.e. where immediate drastic action was needed to prevent an accident)

Here’s what we know from literally decades of human factors research, accidents, and near misses.

1. Humans get bored really easily.
2. Bored humans get distracted or nap.
3. Humans take for bloody ever to get up to speed (reaction time while paying attention is ~0.25 s, while not it’s 2-15 seconds). It gets worse when humans are tired.
4: Humans are really, really bad about normalizing deviance. (As in, based on familiarity we push the envelope and erode safety margins)
5: Humans believe they are much more capable than testing shows.

So while you claim to be, and (giving the unverified benefit of doubt) might be a 6 sigma safest driver, the 200 humans next to you also claim the same thing erroneously. Thus, the human factors engineers tell the systems engineers that the driver alert system is safety critical, and not something for end users to “adjust”.

 

[Scandinavian Film]

I just find it asinine that if the car detects that you're not paying attention, for which especially on tesla's system there are many false positives, it will intentionally make a crash more likely by allowing the car to drift out of lane. In my modern driving record I haven't had an at-fault accident for close to 20 years. A level 3 system with no restrictions would be perfect for me.

I get what you're saying (I agree that maintaining the lane while slowing down with the hazards on would be a less-bad response than potentially drifting off into oncoming traffic), but if the car actually does drift out of lane when the system disengages, the attention detection was not a false positive.

 

[sporkman]

raises eyebrow

That is a) a disturbingly weak answer against b) a disturbingly weak metric.

The better questions, which you are probably too biased to answer (this is not particularly a condemnation of you personally):
1. How many times in the last year have you gone into microsleep while driving?
2. How many times in the last year have you done other tasks while driving (eating, texting, gotten in a deep conversation with a passenger or third party, etc.)
3. How many times in the last year have you been in extremis while driving (i.e. where immediate drastic action was needed to prevent an accident)

Here’s what we know from literally decades of human factors research, accidents, and near misses.

1. Humans get bored really easily.
2. Bored humans get distracted or nap.
3. Humans take for bloody ever to get up to speed (reaction time while paying attention is ~0.25 s, while not it’s 2-15 seconds). It gets worse when humans are tired.
4: Humans are really, really bad about normalizing deviance. (As in, based on familiarity we push the envelope and erode safety margins)
5: Humans believe they are much more capable than testing shows.

So while you claim to be, and (giving the unverified benefit of doubt) might be a 6 sigma safest driver, the 200 humans next to you also claim the same thing erroneously. Thus, the human factors engineers tell the systems engineers that the driver alert system is safety critical, and not something for end users to “adjust”.

There have been times when I have been driving while extremely fatigued and going into microsleeps.

My comma.ai system kept me safe going in the lane perfectly. It was a lifesaver. If I didn't have the system it would undoubtedly have been much less safe. Because I experienced such a circumstance in my previous car which wasn't compatible. That circumstance being fatigued with no self driving system was very dangerous. Driver monitoring would basically disable the system when it is most useful.

The Tesla system in contrast is such bullshit. You can't even look at the car controls or the map on Tesla's own screen without it giving a warning. Tesla designed a car with a central touchscreen. It punishes you for even looking at it or using it.

But uh, the comma system does have driver monitoring though it is way better than Tesla's in that it has far fewer false positives. You can't completely fall asleep while using it but you can adjust the sound system.

 

[smartalco]

There have been times when I have been driving while extremely fatigued and going into microsleeps.

Then you shouldn’t have been driving, whether you were getting an assist from the car or not.

If you have some extreme circumstance where you literally have no choice (driving towards emergency care from somewhere without cell service or something) then that’s one thing. But since you said “times”, meaning this has happened to you more than once, and taken with the rest of your recent posts in this thread, I’m just coming to the conclusion that you’re a terrible driver. And possibly just mad that the Tesla won’t let you be a terrible driver.

 

[sporkman]

Then you shouldn’t have been driving, whether you were getting an assist from the car or not.

If you have some extreme circumstance where you literally have no choice (driving towards emergency care from somewhere without cell service or something) then that’s one thing. But since you said “times”, meaning this has happened to you more than once, and taken with the rest of your recent posts in this thread, I’m just coming to the conclusion that you’re a terrible driver. And possibly just mad that the Tesla won’t let you be a terrible driver.

but the thing is that tesla technology can make me a very very good driver in spite of myself. It's fully capable of that, but the company has this overly paranoid driver monitoring system that prevents me and the car from being very very good at driving.

 

[Chuckles]

At L3, the car is designed to fail over to you. The manufacturer knows that it can’t handle a significant number of edge cases. The driver monitoring is to ensure that you are ready to take over as the backup system. Driving fatigued means that you are going to incapable of taking over in a timely manner without the monitoring.

 

[sporkman]

At L3, the car is designed to fail over to you. The manufacturer knows that it can’t handle a significant number of edge cases. The driver monitoring is to ensure that you are ready to take over as the backup system. Driving fatigued means that you are going to incapable of taking over in a timely manner without the monitoring.

hmm, true. that means that what I really need is level 4.

 

[w00key]

At L3, the car is designed to fail over to you. The manufacturer knows that it can’t handle a significant number of edge cases. The driver monitoring is to ensure that you are ready to take over as the backup system. Driving fatigued means that you are going to incapable of taking over in a timely manner without the monitoring.

What L3? Tesla and everyone else is still at level 2 and expect you to pay attention. Level 3 means you can read a book or watch TV, no nagging.

 

[sryan2k1]

What L3? Tesla and everyone else is still at level 2 and expect you to pay attention. Level 3 means you can read a book or watch TV, no nagging.

Mercedes in limited cases.

 

[w00key]

Mercedes in limited cases.

The Q was what L3 system nags you for attention, @Chuckles wrote, as reply on a Tesla can drive me while I'm tired comment chain:

At L3, the car is designed to fail over to you. The manufacturer knows that it can’t handle a significant number of edge cases. The driver monitoring is to ensure that you are ready to take over as the backup system. Driving fatigued means that you are going to incapable of taking over in a timely manner without the monitoring.

That doesn't sound very L3 to me.

 

[sryan2k1]

https://www.sae.org/blog/sae-j3016-update

It's the textbook definition of L3. "When the feature requests you must drive"

 

[MilleniX]

https://www.reuters.com/business/au...h-weather-ahead-commercial-launch-2025-03-25/
Aurora claims that their ODD will include inclement weather and construction zones, and that they're going to operate in Texas, New Mexico, and Arizona this year.

 

[Ananke]

The distinction between L2 and L3 is that while the self-driving mode is engaged, in L2, you remain fully responsible for the car's actions, and you need to be ready to correct for erratic actions in real time, without prompting. In L3, the car is responsible for itself unless/until you are prompted to take over. Sporkman is reading that as "so it's fine if I'm drunk, asleep, or engrossed in a movie, then?"

But you are still responsible as a fallback whenever and wherever the car determines it is no longer capable of acting autonomously. You need to be sufficiently alert and aware of what is happening around you to be ready to take back control given, potentially, very little notice. So you can't be drunk or asleep; or any other reason to be ineligible to drive a car (under age, not passed a driving test, medically unfit, temporarily or permaanently banned from driving, etc), because in any of those casses you're not capable of taking control when the computer prompts you to.

Just how immersed you can be in another activity - reading a book, watching a film, making out in the back seat etc - is going to depend on what sort of minimum handover time is guaranteed - i.e. what liability the manufacturer will accept based on how predictive they believe their model is. Being yanked out of another activity and expected to regain situational awareness while controlling a dangerous amount of energy (heavy car going fast) is a recipie for disaster unless the computer can guarantee long (like 30s+) handover times. Which I personally believe is wildly improbable - if the car can predict that far ahead, it's probably capable of getting itself to safety, i.e. verging on L4.

Per previous comments about human factor research, this is essentially the worst possible level of self-driving, because it relies on something at which humans, on average, are absolutely abysmal.

Sporkman's open acknowledgement of being a dangerous driver aside, what he's asking for is L4.

 

[moosemaimer]

Coming home from the eclipse a few years back I was in bumper-to-bumper traffic for 11 straight hours. I'm honestly amazed I was still as alert as I was by the end, but having the car do even 75% of the work would have availed me to spend that mental energy on paying attention and not micromanaging the controls.

My Prius has a hands-free mode that only kicks in at <25MPH on a limited-access highway, and I'm kinda itching to get caught in traffic just so I can try it.

 

[demultiplexer]

What I'm reading in this thread now is that like most things, SDCs, if they ever come, won't be used to do anything useful, rather they'll be used to carry overworked, sleeping, black-out drunk or otherwise impaired people who really shouldn't be on the road to whatever place they need to be so they can work, party or care for others even more hours per day...

It's fine to go on a road trip and just take a break every few hours. Have a walk, eat a banana, touch grass. It's not necessary to do everything at peak efficiency all the time.

 

[sryan2k1]

Spring break baby! 14 of those hours were travel day #1, we'll probably do it non stop on the way home.

 

[ramases]

I recently had to do a 1400km round trip as a single driver, in a single day, to fit it into a single work day.

To be honest I wasn't feeling great for most of it. I was tired, and most of my thoughts were elsewhere.

I hencd handed off roughly 1.2k of those km to a somewhat-L2-ish self-driving vehicle able to obtain much greater speeds (in excess of 230 kph on some sections) than regular cars on highways; it literally comes with a kitchen and a sink, and I was able to have some decent meal during the trip without having to stop.

We call those vehicles 'trains'. For the last 100km I used a app-based car rental.

Snark aside, the last couple pages really were full of great exhibits for the 'SDC as a "fix for urban/infra planning without fixing urban/infra planning" chimera.' argument.

Perhaps the solution for having to drive 18 hours for less than a thousand miles isn't to hand off car control to a machine else, and instead try to find a solution so you don't have to spend 18 hours in a tiny metal and plastics box for that distance in the first place.

 

[w00key]

Snark aside, the last couple pages really were full of great exhibits for the 'SDC as a "fix for urban/infra planning without fixing urban/infra planning" chimera.' argument.

We have trains, they run great but is rarely as flexible as a car. Tons of areas have less than ideal coverage in space or in time and my frame of reference is


Hong Kong, pretty much shared #1 spot for best public transportation. Because it's pretty much all built up and city, with great 24/7 coverage by metro, train, light rail, bus, mini bus and a billion affordable taxis.

The Netherlands, #6 on https://www.insidermonkey.com/blog/5-countries-with-most-efficient-public-transportation-1264949/. Great urban public transit, but coverage is lacking quickly outside of main routes. Taxi and Uber is unaffordable for most.

Japan, #5. Great in urban areas like greater Toyko, meh outside, if you are off the main routes, gl with the hourly train. Taxi is very expensive.


The main difference between HK and the other two is that if you need to go somewhere either with less coverage, or shitty connections like Yuen Long district to Sha Tin, you can choose between ~1.5h train to center and back out, or a direct highway between the two via taxi in 40 minutes.

This happens everywhere. In the Netherlands, you ask a friend for a ride or just don't make the the trip. In Japan, we just took JR local and it takes relatively forever. If you're rich enough, fine, grab a taxi for €100/¥100000.


Here is where SDC comes in. As a shared public service, running like a point to point 7 to 16 seat minibus with passenger pooling, even if you have one of the greatest transit systems in the world, you can benefit from it.

Even in HK, when I was just like 10 minutes from a major shopping and business district, I had to give like €/$10 tip in the taxi app to grab someone's attention after 15 minutes of retrying booking a car for a €/$25 trip. Basically had to pay for the first empty leg too. Being even further out like trying to a ride back from the end of a hiking trail is unreliable.

And this is during the day. At night all of the high capacity links are down, weekend is reduced service, SDC busses have no such limitation.

 

[dettociao]

I mean, I live in HK and I'd happily replace the local insane taxicabs with existing SDC (well, maybe not...)

But the point made above stands: if you design things properly, for most metro areas you probably don't need it -- and it's certainly not something that should be a priority in most places, as compared to fixing their public transit.

 

[w00key]

I mean, I live in HK and I'd happily replace the local insane taxicabs with existing SDC (well, maybe not...)

But the point made above stands: if you design things properly, for most metro areas you probably don't need it -- and it's certainly not something that should be a priority in most places, as compared to fixing their public transit.

Easier said than done, "just fix public transit". Just how many rail projects succeed vs fail? Every one in the last 20 years cut scope and raised cost massively. Even in well run rail systems people complain non stop about cost, then about quality because lack of budget for maintenance and service. Because that shiny new railway for $$$$$ must be paid back somehow.

SDC is orthogonal to well run transit system, even HK can use cheaper, better taxis and minibusses. A driverless triple bench 9 seater can replace a few taxis (most hailers are 1-2 pax) taxis and run 24/7, increasing quality of life for everyone not on the main routes.


Last time I considered going from Tsuen Wan West to Tsing Yi, they are right next to each other separated by just a bridge, and both are on the main MTR lines. I can see the destination easily. But no, it's 9 min drive / taxi vs 30 min metro because of a V shaped detour to interconnect between lines, plus a bunch of walking. Nah never mind then, just walking over the damned bridge is faster, if I didn't have a baby stroller with me and can take stairs vs epic detour to accessible routes. Plus weather, severe wind and sun of course, this is a 1km long super tall bridge, while it has a separate pedestrian and bike lanes, you rather not cross it on foot.

SDC powered public transit can make all these annoyances go away and run overnight with mininal staffing cost.



And in countries and regions with less demand / developed transit systems it is going to radically change travel for everyone too poor to drive or take an Uber everywhere.

 

[MilleniX]

Every one in the last 20 years cut scope and raised cost massively.

There are occasional exceptions. Brightline in Florida has kept to plans. Metro expansions in LA and Seattle have beaten schedule and budget.

Incidentally, I've just been reading a book about exactly how this plays out.

 

[w00key]

There are occasional exceptions. Brightline in Florida has kept to plans. Metro expansions in LA and Seattle have beaten schedule and budget.

Incidentally, I've just been reading a book about exactly how this plays out.

Sure. There are always exceptions to the rule. Rotterdam also succesfully expanded the metro system on time, within budget.

But generally speaking construction delays and cost overruns are more common than "oh wow they actually managed to do as promised". And overruns are quite epic at multiple times the cost and often there is no way out once you started work, other than raise money / prices, cut scope, and often both.

 

[Chuckstar]

There are occasional exceptions. Brightline in Florida has kept to plans. Metro expansions in LA and Seattle have beaten schedule and budget.

Incidentally, I've just been reading a book about exactly how this plays out.

I remember when LA’s subways first opened. They had gone so over budget that they swore never to build more subways. Then people started actually using them and now…

It’s not like major road projects are immune from being expensive and going over budget, either.

 

[dettociao]

I would support more SD mass transit, sure. I'll likely never favor anything that makes it easier or cheaper to have more personal transit on the road. It's just not worth the tradeoffs imo.

 

[w00key]

I would support more SD mass transit, sure. I'll likely never favor anything that makes it easier or cheaper to have more personal transit on the road. It's just not worth the tradeoffs imo.

Personal self driving car, what Tesla keeps promising and delaying, is probably never going to be a reality. Real world SDC requires a room with operators to unfuck situations and monitor operation, why would you offer that on personal vehicles vs on a fleet designed for commercial use?


Real high volume commercial taxis have radically different requirements than personal cars, from interior to duty cycle / wear and tear on drivetrain, brakes and suspension.

Tesla is attempting SDC on hard mode with cheap sensors to save cost. Waymo throws much more money at each taxi as it can afford to do so, charging $/trip instead of one time.

 

[zeotherm]

Incidentally, I've just been reading a book about exactly how this plays out.

Off topic, but that is a FANTASTIC book

 

[Chuckstar]

Personal self driving car, what Tesla keeps promising and delaying, is probably never going to be a reality. Real world SDC requires a room with operators to unfuck situations and monitor operation, why would you offer that on personal vehicles vs on a fleet designed for commercial use?


Real high volume commercial taxis have radically different requirements than personal cars, from interior to duty cycle / wear and tear on drivetrain, brakes and suspension.

Tesla is attempting SDC on hard mode with cheap sensors to save cost. Waymo throws much more money at each taxi as it can afford to do so, charging $/trip instead of one time.

You would provide such a room as a subscription service. Unless you somehow believe companies don't like offering subscription services. But I think the current tech landscape would tend to argue against such a position.

 

[ramases]

We have trains, they run great but is rarely as flexible as a car. Tons of areas have less than ideal coverage in space or in time and my frame of reference is


Hong Kong, pretty much shared #1 spot for best public transportation. Because it's pretty much all built up and city, with great 24/7 coverage by metro, train, light rail, bus, mini bus and a billion affordable taxis.

The Netherlands, #6 on https://www.insidermonkey.com/blog/5-countries-with-most-efficient-public-transportation-1264949/. Great urban public transit, but coverage is lacking quickly outside of main routes. Taxi and Uber is unaffordable for most.

Japan, #5. Great in urban areas like greater Toyko, meh outside, if you are off the main routes, gl with the hourly train. Taxi is very expensive.


The main difference between HK and the other two is that if you need to go somewhere either with less coverage, or shitty connections like Yuen Long district to Sha Tin, you can choose between ~1.5h train to center and back out, or a direct highway between the two via taxi in 40 minutes.

This happens everywhere. In the Netherlands, you ask a friend for a ride or just don't make the the trip. In Japan, we just took JR local and it takes relatively forever. If you're rich enough, fine, grab a taxi for €100/¥100000.


Here is where SDC comes in. As a shared public service, running like a point to point 7 to 16 seat minibus with passenger pooling, even if you have one of the greatest transit systems in the world, you can benefit from it.

Even in HK, when I was just like 10 minutes from a major shopping and business district, I had to give like €/$10 tip in the taxi app to grab someone's attention after 15 minutes of retrying booking a car for a €/$25 trip. Basically had to pay for the first empty leg too. Being even further out like trying to a ride back from the end of a hiking trail is unreliable.

And this is during the day. At night all of the high capacity links are down, weekend is reduced service, SDC busses have no such limitation.

You can approximate true labor costs quite well by looking at "centrally-owned&maintained cabs with drivers renting them from the operating company" models. In those cases, the split is usually around 1/3 of revenue going to the driver.

(Lyft/Uber splits are woeful to use as a model for labor costs, because yes the driver takes like 3/4th, but they are also responsible for upkeep and purchase; not a good proxy for labor costs)

In other words, replacing the labor component of current cab models with SDC will only, at most and under the assumption that your costs for the SDC kit itself is 0, yield savings of roughly 33%

That's likely not going to be good enough to do all the things you want it to do; that'd require using vehicles more efficiently, and whether there is a human in it or not running empty for longer stretches to pick someone up isn't efficient. You are still putting wear&tear on the equipment, and you still have the opportunity cost.

(SDCs need maintenance and cleaning, too; perhaps more cleaning than a regular cab, because some people just cannot behave themselves when no one is watching. The 'low tide' during the night would be the best time to schedule that, at least if you want as much revenue-km per day as possible)

Hence you'd still need to either wait longer until more people show up, or a route that minimizes deadheading, or pay extra.

Likewise, the 7-to-16 seat microbus with (semi-) flex routes and passenger pooling doesn't require SDCs. It exists here and now, and it is ubiquitous in South America and parts of Asia. The SA ones are called "Collectivos".

There have been many attempts to do them in Europe and North America; they overall don't stick around because peope just don't like using them.

 

[w00key]

You can approximate true labor costs quite well by looking at "centrally-owned&maintained cabs with drivers renting them from the operating company" models. In those cases, the split is usually around 1/3 of revenue going to the driver.

No that's for a rent extracting setup with limited availability of licenses, or where drivers lack the capital to purchase a vehicle. Vehicles are not that expensive compared to labor cost in most of the world, nor is operating cost ex driver very high. You see that in public transit, huge full size bus running mostly empty route. Downsizing saves nothing compared to the fixed driver labor requirement.


For free for all areas where Ubers and Lyfts aren't licensed the split is more reasonable. Everyone can get a Prius and start earning revenue, but as the app controls demand, they can extract more rent from the situation. There is no reason why a taxi center costs you 1/4 to 1/3 of the revenue.

Upkeep, maintenance and fuel is ~0.25/km unless you overspend. Priuses are common here as taxi, now more often EVs like MG, Tesla 3 SR as they get to use bus / EV taxi only lanes. Less than Uber's cut.


For an end to end controlled scheme like SDC operator with own scheduling app and owned vehicles, it's completely different. As long as there is demand to minimally staff the control room, they can do so and run tens of vehicles concurrently.

Likewise, the 7-to-16 seat microbus with (semi-) flex routes and passenger pooling doesn't require SDCs. It exists here and now, and it is ubiquitous in South America and parts of Asia. The SA ones are called "Collectivos".

They have them here in HK too, for further out areas and as only non taxi public transit at night. But what they won't do is run door to door service like a SDC taxi can, or slightly slower, pooled door to door transport. Minibusses now have a big connection cost if you want to go a km off the route, or 12 minutes walking at 5km/h pace. Do that on both ends and you have a 24 minutes of walk time added to the trip.

Triple bench 9 seaters can pick up three groups and have each sit in their own row.

 

[sporkman]

I've never thought that the Uber self driving initiative would be successful because of how the rideshare business actually is. But it appears that a lot of investors would only invest for the promise of self-driving. It's one of those things where the stated objective is never actually achieved but something else can be found out along the way. Kind of like how Slack came out as someone else's internal development tool for another project.

The best use of self driving tech is a better cruise control that lets me eat noodles while driving.

 

[ramases]

No that's for a rent extracting setup with limited availability of licenses, or where drivers lack the capital to purchase a vehicle. Vehicles are not that expensive compared to labor cost in most of the world, nor is operating cost ex driver very high. You see that in public transit, huge full size bus running mostly empty route. Downsizing saves nothing compared to the fixed driver labor requirement.

Average mileage for a single-shifted cab in most of Europe is roughly 100.000km. To approximate running costs including fuel, wear&tear, amortization and basic maintenace and repair, we can start with the tax deduction for business travel in a privately owned car, because that tax deduction is meant to cover all of that for first-hand (ie not used) cars. Rates differ, but most of it is between 30 to 50 Eurocent/km; lets split the difference and go with 40 Cent.

Using this we get an approximated running cost for the cab at roughly 40k/year. Note that this is almost certainly going to underamortize the purchase costs considerably, because it assumes a usage-pattern of roughly 15-30k km per year, not 100k.

Meanwhile, the median pre-tax income of a cab driver in Germany is roughly 28k.

Labor costs are, quite comfortably, eclipsed by capital outlay and running costs of a heavy-duty vehicle.

For free for all areas where Ubers and Lyfts aren't licensed the split is more reasonable. Everyone can get a Prius and start earning revenue, but as the app controls demand, they can extract more rent from the situation.

The rent extraction argument never made any sense to be honest. Since SDCs would not be operated by the public, and instead are operated by private companies, why would those private companies not succeed in extracting rent when cab companies and ridesharing app companies can?

If there are ways and means to prevent this extraction for SDCs, why cannot they also be applied to cab/ride sharing companies?

And to take a shot at rent extraction as premise itself, wouldn't this all boil down to trying to solve a societal problem with a technological solution, which we know often fails?

They have them here in HK too, for further out areas and as only non taxi public transit at night. But what they won't do is run door to door service like a SDC taxi can, or slightly slower, pooled door to door transport.

A SDC's ability to do so is predicated on the economic utility of it. Given that so far it hasn't been substantiated, and the claim that labor costs being the dominant factors in cab solutions has not been shown, [citation needed]

Minibusses now have a big connection cost if you want to go a km off the route, or 12 minutes walking at 5km/h pace. Do that on both ends and you have a 24 minutes of walk time added to the trip.

A problem for those slow on foot, the disabled, small kids, and similar; for everyone else, assuming there is decent infrastructure like a sidewalk to walk safely:

Oh noes.

Slightly less facetious, in urban areas you can have pickup locations with a shorter distance between them than a km. And outside of urban areas you have wait-times, because vehicles (self-driving or not) will not circle of keep standby in areas where they cannot earn their keep because of too low a rider density.

You're again stuck with either longer waits, higher prices, or a hybrid thing where you walk 5-10 minutes to a pickup-location, in return for a rider density that keeping vehicles near stations. By the by, picking up/dropping off will likely take 30-60 seconds in the best case, and may take several minutes in case of people that cannot move quickly and/or have large luggage.

 

[Shavano]

++ @ramases As you explain, a lot of the enthusiasm regarding self driving cars gets less exciting when you look closely at what problem are you trying to solve, exactly.

The obvious one of reducing motor accidents seems on its face like it ought to be readily doable. In the US, there were 40,990 deaths in 2023, 289,310 were injured. But honestly that's a remarkable level of safety considering how much Americans drive. In 2023, they drove 3.1 trillion miles. So that's about 1 death per 75.6 million miles, and about 1 injury per 10.7 million miles. Statistically we're pretty good. So while the carnage is large in aggregate, it really would take some pretty good self driving cars to be better than people are. More to the point, it doesn't seem that important. There's a lot of lower-hanging fruit to be had if you want to reduce motor accidents - better intersection design, better margin design, lower speed limits, Driver assist features haven't really made that much of an impact if I read what this graph is telling me rightly:



SDC's aren't going to relieve congestion because they don't reduce the number of cars on the road. If anything, they'll increase those numbers, with cars transporting humans plus cars deadheading. But probably not by much. They'll just replace taxis, so a car with human is directly replaced by a car without a human in it. Now they could reduce congestion, but only if the "ride share" cars are actually shared by two or more people going from about the same place to about the same place on the other end. But we don't see a lot of that with ride share cars now so I don't expect to see it happen due to SDC's.

The best argument for them I'm hearing is it will be able to drive you home from the bar when you're drunk or tired, or to the hospital when you're sick. We have taxis and ambulances for that now, so it's solving an already-solved problem there, isn't it?

Cheap delivery service? Nope. Not going to be able to equal a human to solve the pick the right package from the truck and carry it to the doorstep problem as fast or as efficiently as a human.

I think their niche is suburbanites willing to pay a few or quite a few thousand extra to have more time to look around or play games or work on the way to work. Who are you kidding? Nobody wants to work on the way to work.

 

[ChaoticUnreal]

I think their niche is suburbanites willing to pay a few or quite a few thousand extra to have more time to look around or play games or work on the way to work. Who are you kidding? Nobody wants to work on the way to work.

I don't know if the option is sit in traffic driving myself for an hour and then sit at the office for 8 hours and do another hour of traffic vs work while in traffic for an hour sit in the office for 6 and do another hour on the way home I'd take the second option.

 

[Shavano]

It's an option that's only available to people of a certain class, but maybe. I mean if people work on the subway or a bus, they might work in a SDC.

 

[w00key]

It's an option that's only available to people of a certain class, but maybe. I mean if people work on the subway or a bus, they might work in a SDC.

Have you tried working on a subway or a bus?

In HK, JP and NL, places ranking excellent on public transit, you have to be glad you can find a seat during peak times and they are not so roomy that you can do any sort of work. Even cattle class flight is better, at least you have a tray table there.

Bus, better but still a maybe. Good enough for reading a book, but I see zero laptops on the average trip even on express routes from New Territories to Central lasting 40+ minutes.


But SDC + working comfortably is a thing. Maybe one day affordable here in Europe (if/when Mercedes dares to launch L4 service, or when Waymo launches), definitely not in HK or Japan, they don't have space for replacing a full metro with individual cabs. It's going to be priced like a premium option.

SDC for the masses are more like bus 500 here, https://en.m.wikipedia.org/wiki/ParkShuttle. The major improvement would be that new ones won't be stuck on a virtual rail and need dedicated lanes and protected crossings (gen 1, 1999 tech is rather simple). The next upgrade would be able to join normal traffic and extend the line past its current end point.