Robotaxis don’t cut traffic any more than ride-hailing, study finds

Data shows Waymo's robotaxis are empty for almost half of the miles they drive.

See full article...

 

[raxadian]

"empty for almost half of the miles they drive." tends to apply to regular Taxis too save at peak hours.

 

[Dr Gitlin]

"empty for almost half of the miles they drive." tends to apply to regular Taxis too save at peak hours.

Yes, it does.

 

[mmmancact]

I never understood the "less traffic" argument, unless the service was doing a lot of last-mile to public transit. Private drivers don't (generally) have deadhead miles, so swapping their ride for a hailed one (autonomous or not) is going to increase mileage. The hailed services should reduce parking needs, but they were pretty much always bound to add to traffic.

 

[JohnDeL]

Well obviously, it is because there aren't enough of the robotaxis. If they were to double the number, then there would be a taxi that was closer to where the client/customer/sap was, and so the number of empty miles would be decreased...
/s

 

[gone_fission]

Duh? Replacing a car with a car doesn’t fix the underlying problem. You know what does? Public transit and bike infrastructure.

 

[yakinabe]

How exactly were they supposed to cut traffic? I imagine autonomous cars need less spacing between cars, because they have faster reflexes, but I'm not sure if they actually do that.

Now, if autonomous cars reduce the need for parking (e.g. office buildings don't need one parking space per employee), that would make a big difference. But that hasn't even started to happen.

 

[demonbug]

But safety isn’t the only selling point: Autonomous vehicles are said to cut traffic.

I didn't see much argument to that effect in the linked article, other than a reduction in traffic snarls resulting from accidents (due to lower incident rates). The only way autonomous vehicles would reduce traffic would be if they somehow resulted in a higher ride sharing rate, but there doesn't really seem to be any evidence of that occurring.

 

[_crane]

What if instead of cars we had some sort of pods that could be joined together for greater efficiency?

 

[Tonkaman]

I never understood the "less traffic" argument, unless the service was doing a lot of last-mile to public transit. Private drivers don't (generally) have deadhead miles, so swapping their ride for a hailed one (autonomous or not) is going to increase mileage. The hailed services should reduce parking needs, but they were pretty much always bound to add to traffic.

You're just trading parking for traffic. A vehicle takes up the same amount of space no matter what, so instead of being stationary in a parking lot now it's circling empty on the roads using energy for no reason.

 

[Lexus Lunar Lorry]

I never understood the "less traffic" argument, unless the service was doing a lot of last-mile to public transit. Private drivers don't (generally) have deadhead miles, so swapping their ride for a hailed one (autonomous or not) is going to increase mileage. The hailed services should reduce parking needs, but they were pretty much always bound to add to traffic.

I always felt that the "ride-sharing/robo-taxis will reduce traffic" argument was a disingenuous effort to direct support/funding away from public transit. No need to build out train tracks or bus lanes, just wait for Silicon Valley to save the day!

The problem is that neither solution solves the problem of throughout. 100,000 single occupancy vehicles on the freeway will cause a traffic jam regardless of whether they're being driven by owners, gig workers, or robots.

The only way that I see autonomous vehicles helping here is if we see the rise of robo-buses or robo-jeepneys. But most of the concepts that I've seen only have at most ~4 people per vehicle, presumably because investors (the real customers of these companies) don't like having to sit next to other people.

 

[grommit!]

But public transport doesn’t come cheap. Waymo might have raised $16 billion earlier this year for its robotaxis, and at least $100 billion has been invested in the sector since the 2010s. Meanwhile, the American Public Transport Association called for $268 billion in investment over five years, and a report by Transportation For America puts the price tag for a “world class” transit system at $4.6 trillion over the next 20 years.

The difference being that the latter doesn't need to generate ever increasing returns for investors.

 

[arsorb]

Another failed promise and snake oil from tech bros in Silicon Valley to draw in the public, the government and VC $$$'s. and draw attention and money away from things that do work like mass transit and bike lanes. Honestly how does it make sense that putting MORE cars on the road will decrease traffic? and with autonomous vehicles creating their own traffic issues, accidents and general strangeness only adding to the problem.

*Edited for clarification and specificity

 

[deus01]

How exactly were they supposed to cut traffic? I imagine autonomous cars need less spacing between cars, because they have faster reflexes, but I'm not sure if they actually do that.

Now, if autonomous cars reduce the need for parking (e.g. office buildings don't need one parking space per employee), that would make a big difference. But that hasn't even started to happen.

https://www.researchgate.net/publication/313879093_How_Much_Urban_Traffic_is_Searching_for_Parking

Some significant amount of traffic is probably caused just by people looking for parking

 

[Tonkaman]

I always felt that the "ride-sharing/robo-taxis will reduce traffic" argument was a disingenuous effort to direct support/funding away from public transit. No need to build out train tracks or bus lanes, just wait for Silicon Valley to save the day!

The problem is that neither solution solves the problem of throughout. 100,000 single occupancy vehicles on the freeway will cause a traffic jam regardless of whether they're being driven by owners, gig workers, or robots.

The only way that I see robo-taxis helping here is if we see the rise of robo-buses or robo-jeepneys. But most of the concepts that I've seen only have at most ~4 people per vehicle, presumably because VCs don't like being more crowded than that.

Insert joke about Silicon Valley continually reinventing trains. i.e. "self driving truck convoys."

 

[Wheels Of Confusion]

I'm not surprised that displacing single-occupant vehicles with single-occupant-but-on-a-time-basis vehicles doesn't really reduce the number of vehicles in total.
Carpooling them might make a difference... but that would also be true of conventional human-driven vehicles (and especially of mass transit and increasing the viability of bicycles as transport).

 

[Tonkaman]

Operating largely the same as rideshare/taxis but not needing a driver seems to be a positive.

Commuters in single vehicles may not have deadhead miles but their vehicle sits empty for a long time while they're at work or wherever which requires space to store it (and time for them to find parking etc)

Robotaxis make this worse, because you need to overprovision for the "rush hour" peak demand during commuting hours and THOSE vehicles are now sitting or driving empty while the people are at work.

 

[jezra]

single occupancy vehicles don't 'cut' traffic, they are traffic.

 

[NOT_RICK]

Tech bros promising a fantastic utopia whilst hand waving away the shitty reality they're actually manifesting? Color me shocked!

 

[Dr Gitlin]

The difference being that the latter doesn't need to generate ever increasing returns for investors.

No, but consequently also doesn’t attract investment from capital as a result. And voters appear afraid to vote for more public funding for public transport.

 

[Tactical Finesse]

Autonomous vehicles were supposed to cut traffic—what if they don’t?​

If? Of course they won't. The only way to cut traffic is for people to have viable alternatives to driving. If there aren't viable alternatives to cars (autonomous vehicles are still cars), car traffic and congestion will increase without limit--until alternatives become viable.

 

[Wickwick]

For both the ride hailing and automated robotaxis, the miles driven while waiting for a fare obviously add traffic, but need not add congestion. One can preferentially choose slower side streets that aren't particularly good routes for getting from A to B because you're not actually trying to get to B. Partially filling uncongested side streets allows vehicle storage without negatively impacting the rest of drivers.

 

[Dr Gitlin]

How exactly were they supposed to cut traffic? I imagine autonomous cars need less spacing between cars, because they have faster reflexes, but I'm not sure if they actually do that.

Perhaps you should follow the link I included in that sentence and it will tell you.

 

[KrookedRooster]

As others have said, it's still a car, using car roads, with other cars, doing car things, able to get into car accidents which cause car traffic jams.

And now there are more cars out there (Robo or otherwise) trying to make money, driving around because barrier-to-entry is low so more cars that wouldn't be out there otherwise are now out there (what is the paradox of a solution causing the same problem it was supposed to solve again?)

And can they even zipper? Humans can't even do that properly!

 

[cyberfunk]

“Robotaxis Don’t Cut Traffic Any More Than Ride-Hailing” .... except that’s not what the study actually found. The researchers measured how often Waymo cars drive around empty between trips, and found the number lands around 43–45%. That’s an interesting operational data point. It is not a traffic congestion finding. What the article is, at its core, is a single preprint dressed up as a verdict on an entire industry.

The headline comparison of “no better than regular ride-hailing”, is also doing a lot of heavy lifting on behalf of very thin evidence. Uber and Lyft’s deadheading numbers were measured across mature networks with millions of drivers in hundreds of cities. Waymo operates in a handful of geofenced zones and still represents a rounding error of total urban trips. Comparing those two things as if they’re equivalent is like judging whether a restaurant will turn a profit by looking at its receipts from the first week of soft opening.

The debate over whether ride-hailing makes traffic worse or better has been running seriously in the academic literature since at least 2017, with multiple competing datasets, methodologies, and conclusions.

 

[deus01]

Robotaxis make this worse, because you need to overprovision for the "rush hour" peak demand during commuting hours and THOSE vehicles are now sitting or driving empty while the people are at work.

This is the same issue with taxis though. It's still going to cause a problem at peak times that can only be solved with mass transit but if the cost for autonomous can be lowered it could be a good a supplement to mass transit off major arteries

 

[MMarsh]

https://www.researchgate.net/publication/313879093_How_Much_Urban_Traffic_is_Searching_for_Parking

Some significant amount of traffic is probably caused just by people looking for parking

Yes it is. Finding space for a car in a dense downtown or a Costco parking lot often takes nearly as long as getting into that downtown in the first place.

Also a significant amount of traffic is caused by taxis, delivery vans, etc. stopping in live lanes to load and unload. They don't stick around long enough to be ticketed, but they often hold up enough cars behind them to gridlock a traffic light for one cycle.

Robotaxis are fine in a low traffic area where everything's too spread out to walk or to justify the cost of transit.

The moment traffic actually becomes a factor, you are far better off by putting 12 people in one bus than by putting 12 people in 12 robotaxis.

When you start talking about 1000 –10,000 people at a time then you really need to think about replacing cars with trains.

 

[KingKrayola]

How exactly were they supposed to cut traffic? I imagine autonomous cars need less spacing between cars, because they have faster reflexes, but I'm not sure if they actually do that.

Now, if autonomous cars reduce the need for parking (e.g. office buildings don't need one parking space per employee), that would make a big difference. But that hasn't even started to happen.

I imagine the pitch was that they could improve utilisation 'with an algorithm' but that turned out to be a simplistic assumption (insert relevant xkcd here).

Even where I live, in the densest part of London with the more expensive flats swamped with Ubers, there isn't so much demand that drivers can get one fare after another, particularly if riders don't want to wait.

 

[deus01]

Yes it is. Finding space for a car in a dense downtown or a Costco parking lot often takes nearly as long as getting into that downtown in the first place.

Also a significant amount of traffic is caused by taxis, delivery vans, etc. stopping in live lanes to load and unload. They don't stick around long enough to be ticketed, but they often hold up enough cars behind them to gridlock a traffic light for one cycle.

Robotaxis are fine in a low traffic area where everything's too spread out to walk or to justify the cost of transit.

The moment traffic actually becomes a factor, you are far better off by putting 12 people in one bus than by putting 12 people in 12 robotaxis.

When you start talking about 1000 –10,000 people at a time then you really need to think about replacing cars with trains.

I'm not arguing in favor of robotaxis replacing mass transit but rather that they can supplement mass transit and are an improvement over existing taxis and private drivers (even if it's just because they won't park in bike/bus lanes or run over cyclists )

 

[Tonkaman]

This is the same issue with taxis though. It's still going to cause a problem at peak times that can only be solved with mass transit but if the cost for autonomous can be lowered it could be a good a supplement to mass transit off major arteries

The need for a profit motive ensures that private taxies will never be cheaper than mass transit. Most of the costs of running a taxi business are the licensing fees (medalians), costs of the vehicle (buying it, maintaining it like repairs and cleaning it, fueling it), and THEN labor way way down the list. It's cheaper to maintain 1 train than 50 taxis, and they both carry the same amount of people.

 

[jlredford]

They do cut emissions, though, since the robotaxis are all EVs. Taxis and ride-shares usually aren't. That's also better for pollution and noise. The robotaxis also don't mind waiting at chargers as much as human drivers would.

 

[wxfisch]

I never understood the "less traffic" argument, unless the service was doing a lot of last-mile to public transit. Private drivers don't (generally) have deadhead miles, so swapping their ride for a hailed one (autonomous or not) is going to increase mileage. The hailed services should reduce parking needs, but they were pretty much always bound to add to traffic.

In theory ride share services like Uber and Lyft could reduce traffic if it were run more like how Slugging works in the DC area. For those not familiar, around DC there was and I think still is, a practice of single drivers picking up random commuters heading in the same direction to meet HOV requirements. The drivers get a faster trip and the riders get a free ride into or out of the city. The initial idea behind Uber and Lyft were that we would all be drivers and grab a fare nearby on our way to the places we were already going, this would reduce traffic because you are filling more seats in cars that were already going to be on the road.

It turns out that model was neither realistic nor profitable for the companies or people that were driving. Most folks don't want to pickup random people on their way to somewhere else (I certainly don't if I am say going to to the airport, and especially not after a flight when I just want to get home). So instead we got taxis, but through an app and not as regulated.

This of course increased traffic, and taking the driver out of that equation in my mind would only make the matter worse since a robo-taxi isn't "normally" going anywhere specific, at best it would let you have more cars without the need for drivers which would reduce deadheading since it is more likely there is a car nearby ready to service the ride.

To really get the benefit though I would expect companies like Waymo to invest in "pools" that they can park their cars in while waiting for fares to reduce the amound of deadheading in place waiting for someone to request a ride. A bunch of small lots or garages, or even some leased spaces in exisiting lots and garages would help with this a lot (though I assume Waymo still sees value in the data from drivining around empty to improve the service, so they may not care just yet).

Edit: formatting

 

[Tonkaman]

They do cut emissions, though, since the robotaxis are all EVs. Taxis and ride-shares usually aren't. That's also better for pollution and noise. The robotaxis also don't mind waiting at chargers as much as human drivers would.

That has nothing to do with robotaxis though. Cities could just as easily mandate that all taxis are EVs.

 

[deus01]

The need for a profit motive ensures that private taxies will never be cheaper than mass transit. Most of the costs of running a taxi business are the licensing fees (medalians), costs of the vehicle (buying it, maintaining it like repairs and cleaning it, fueling it), and THEN labor way way down the list. It's cheaper to maintain 1 train than 50 taxis, and they both carry the same amount of people.

Again I'm not arguing that it will cheaper than mass transit just that it can be cheaper than existing taxis and possibly than owning your own vehicle. Labor is not an insignificant cost especially if you want people to be paid decent wages (this is also why we should be automating transit )

 

[sword_9mm]

Do they have to just keep driving around or can they park somewhere?

An Uber driver can just park and wait for app pings. Can these not do that or is it there's no space to park them either?

 

[Tactical Finesse]

I'm not arguing in favor of robotaxis replacing mass transit but rather that they can supplement mass transit and are an improvement over existing taxis and private drivers (even if it's just because they won't park in bike/bus lanes or run over cyclists )

Well, that ends up being a very severe assumption.

https://news.ycombinator.com/item?id=47912645

https://abc7news.com/post/video-sho...king-bus-lane-laws-in-san-francisco/14777190/

Autonomous vehicles work extremely well, and have been in use for years. So long as they are confined to operating on steel-rails with steel wheels, with of course predefined paths, and have tight time tables, and are carefully managed. This car AV thing---is precisely none of those things. WHich is why even in cities where it is mild and sunny almost all year around--AVs struggle with traffic fundamentals--like bike or bus lanes or even lights.

Roads are chaotic places where meat brains operating cars actually do a stunningly good job at being able to handle sensory overload most of the time.

 

[compagnied]

The same number of people on a bus take up much less room on the road than if they were spread out in passenger cars, and the numbers get even better for trains and subways.

Not if that number is 1, 2 or 3, which are the only numbers of people (counting the bus driver) in buses outside of rush hour here in the Chicago suburbs (Pace buses).

 

[Tonkaman]

Again I'm not arguing that it will cheaper than mass transit just that it can be cheaper than existing taxis and possibly than owning your own vehicle. Labor is not an insignificant cost especially if you want people to be paid decent wages (this is also why we should be automating transit )

Eliminating drivers just increases the profit margin of the conglomarate. It doesn't make anything cheaper for you or me.

 

[Incarnate]

I didn't see much argument to that effect in the linked article, other than a reduction in traffic snarls resulting from accidents (due to lower incident rates). The only way autonomous vehicles would reduce traffic would be if they somehow resulted in a higher ride sharing rate, but there doesn't really seem to be any evidence of that occurring.

I suppose it depends on how you are defining "traffic". I think it can help prevent traffic slowdowns as long as AI doesn't do some of the stupid shit that human drivers do. Reduced accidents will lower "traffic snarls" and reduce commute time. Same with people not merging nicely, blocking lanes to prevent people from merging, people cutting over lanes of traffic so they don't miss their exit, people not paying attention when lights change, staring at their phones, etc.

In theory if all/most cars were autonomous the cars should be able to follow closer together in traffic safely, and know when other vehicles plan to change lanes and add space to let vehicles in without everyone slowing to a dead stop.