4 ms·

I wouldn't be planning any fixes infrastructure transit programs that didn't have an ROI in the next 7 years. It will be hard for anything to compete with the e

by xnx 1y ago

I wouldn't be planning any fixes infrastructure transit programs that didn't have an ROI in the next 7 years. It will be hard for anything to compete with the efficiency of Waymo.

digitalPhonix 1y ago
Each Waymo vehicle is probably close to half a million USD in just hardware cost. I don’t think fixed route transport infrastructure is going to have trouble competing on efficiency.
- cyberax 1y ago
  The hardware costs for Waymo are estimated at $30k.
  - digitalPhonix 1y ago
    Source? They were targeting $7.5k for their in house honeycomb lidars and they have 12 of them - that’s 90k already. https://www.theverge.com/2021/8/27/22644370/waymo-lidar-stop-selling-autonomous-cars-laser-bear https://www.theverge.com/2021/8/27/22644370/waymo-lidar-stop... They also aren’t close to the $7.5k target (there isn’t any public source for that so you’ll have to take my word for it). Also $30k wouldn’t even cover the base vehicle.
    - cyberax 1y ago
      This is an estimation for the self-driving hardware cost (computers, LIDAR, sensors). It does not include the base car price, as it can be easily optimized down to almost nothing (sub $10k). The price is somewhat of a guess, several years ago, the hardware in Waymo was priced at $130k by Munro&Associates. But since then the cost of the LIDAR sensors fell by 90% or so, reducing the main expense. And Baidu has cars on the road that cost $30k for the _entire_ car. So presumably, so even a couple of pricier sensors won't affect the estimate too much.
      - rafram 1y ago
        > include the base car price, as it can be easily optimized down to almost nothing (sub $10k). Waymo uses brand-new electric Jaguars.
        cyberax 1y ago
        Sure. Because why not? The cars are in the experimental stage, so they might as well use nice ones. But there is no _reason_ to use Jaguars and not specially-built smaller and less powerful cars, when Waymo finally starts a real rollout.
      - digitalPhonix 1y ago
        Source? None of those numbers make sense. > The price is somewhat of a guess, several years ago, the hardware in Waymo was priced at $130k by Munro&Associates 10 years ago they had even more sensors dotting the car, instead of one honeycomb on each corner they had 2, so I find $130k hard to believe given what we know about the sensor kit today. > But since then the cost of the LIDAR sensors fell by 90% or so, reducing the main expense. I do not know of any lidar that has done that, and Waymo makes their own and we know their price(-ish) (and quantity). I think they’ve actually gone up in price (but also capability - honeycomb 1 vs gen 2).
  - cycomanic 1y ago
    Are you talking about raw material costs? Or is that one of these extrapolations of if we scale everything to millions of cars and realise no inefficiencies and nobody making any money in the supply chain?
    - cyberax 1y ago
      I'm talking about the current cost of the self-driving system, that is already produced by companies that charge a significant markup. With volume, it will go down more. I'm not including the base vehicle in the cost. It's highly variable, and can be as low as $10k for small personal intra-city taxis. China has already launched a $30k taxi: https://www.forbes.com/sites/bradtempleton/2024/05/14/baidu-launches-new-28000-robotaxi-in-wuhan/ https://www.forbes.com/sites/bradtempleton/2024/05/14/baidu-... This _completely_ blows any transit out of the competition. Literally nothing can come even close in the end-to-end efficiency.
      - cycomanic 1y ago
        Why does it blow away any other public transit? That can't be true because if you put the same self driving tech into a bus, you already am an order of magnitude cheaper per passenger (likely more). Moreover let's assume robotaxis are cornering the market and make all other forms of transport non viable. Why would the public then maintain the roads? So at that point at least costs are suddenly going to explode.
        cyberax 1y ago
        Sigh. People are WAAAY too accepting of urbanist propaganda. Buses are _barely_ more effective than cars. A regular passenger car with 4 people is more efficient than a city bus. An EV needs 2.5 people (these numbers are for the US). The explanation is simple: 1. Buses have to drive _all_ _the_ _time_, even when there are few passengers. As a result, the average bus load tends to be around 10-20 people. And you can not increase the bus interval to compensate for it because it makes off-rush-hour bus commutes impractical. 2. Buses have INCREDIBLY polluting components: 2-3 drivers for each bus needed to provide the service. They are by far the dirtiest part of the bus. This part can be removed with the self-driving hardware, but... A full self-driving bus also makes no sense. It defeats the main advantage of self-driving: door-to-door transportation. That being said, self-driving mini-buses seating 6-10 people are a good idea for rush hour transit.
johnea 1y ago
It is really efficient, at sitting in traffic with all the other cars 8-/
- cyberax 1y ago
  And yet, still faster. That's the paradox of cars.
  - rsynnott 1y ago
    Cars are very much not faster than trams in high traffic conditions. Nevermind heavy rail.
    - cyberax 1y ago
      Yes, they are. Do an experiment, drop ten points randomly on a city map. Then plot routes between them, using Google for transit and cars. For most cities, cars will be about 2-3x faster. For three simple reasons: no walking, no waiting, no stopping. And that's the paradox. Cars are better than transit, yet modern cities are hell-bent on destroying the car infrastructure. You can also check this site with isochrones for different transport modes: https://www.geoapify.com/isoline-api/ https://www.geoapify.com/isoline-api/
      - xnx 1y ago
        > Cars are better than transit And are even better when parking isn't a concern (e.g. taxi service)
rafram 1y ago
Public transit isn’t supposed to have a financial “ROI.” Are highways expected to turn a profit from tolls?
- bluGill 1y ago
  The libertarian answer is yes highways should. Most self proclaimed libertarians refuse to go that far - if you allow for highways to not make money then transit shouldn't be held to the higher bar.
  - charlieyu1 1y ago
    I always find this a dumb idea. Transportation networks enable things to happen that comes back as taxes, that is. Same for education, or other public services.
- xnx 1y ago
  Fair, but how unprofitable should they be? -$5/passenger-mike? -$12/passenger-mile? I think we can do a lot better than the current US mass transit status quo.
  - rafram 1y ago
    In 2022, the NYC Subway budgeted about $0.75 per passenger-mile (and that was during Covid, when ridership was very low) [1]. You’re really overestimating how much public transit costs to run. Private vehicles are an extremely inefficient way to move people around, hence the cost of Uber/Lyft/taxis. [1]: https://data.transportation.gov/Public-Transit/2022-2023-NTD-Annual-Data-Metrics/ekg5-frzt/explore/query/SELECT%0A%20%20%60agency%60%2C%0A%20%20%60mode%60%2C%0A%20%20sum%28%60total_operating_expenses%60%29%20AS%20%60sum_total_operating_expenses%60%2C%0A%20%20sum%28%60vehicle_revenue_miles%60%29%20AS%20%60sum_vehicle_revenue_miles%60%0AGROUP%20BY%20%60agency%60%2C%20%60mode%60/page/aggregate https://data.transportation.gov/Public-Transit/2022-2023-NTD...
    - xnx 1y ago
      NYC is special in that it's on of the few places that subways make sense in America. That said, operating costs are common but extremely misleading way to measure transit costs when new tunnel costs $2.2B/mile.
Johnny555 1y ago
Single passenger cars still have a problem with density even if there's no need for a driver. Combining multiple people into one trip can help, but also lessons the utility of Waymo if riders have to go out of their way to pick up additional passengers. Getting 1000 people downtown could be up to 2,000 Waymo trips (one trip to drop off the worker, another trip for the car to go back out to pick up another passenger). While one of these 56 passenger very light rail cars can do it in 18 trips. A light rail vehicle like the Siemens trains used in San Francisco can carry up to 200 people at crush loads, so that's 5 trips.
- mmooss 1y ago
  Yes, I always say: Next time you see mass transit pass, imagine all those people in their own private cars on the road.
cycomanic 1y ago
You mean it's very difficult to compete with a company that is massively subsidised by public infrastructure? That's what really killed freight rail in most of Europe, make the train companies pay for track maintance (often the rail companies even want this because it keeps competition out as well), while trucks atpapy very little of the cost they impose on the public (i.e. much higher road usage, causing most of the traffic issues).
jcranmer 1y ago
If this were true, cities would have abandoned mass transit for taxi system decades ago. The requirement for there to be a warm body driving a taxi isn't among the prime causes of its inefficiency.
yyyk 1y ago
The big AV transit efficiency gains* can only happen when nearly all human drivers are removed from the road. Alas, that's at least 20 years ahead of now**, or more (e.g. if tech stalls for some reason, though I consider it unlikely). Otherwise they'd be limited by having to account for human drivers and that limits speed and throughput enough so other solutions are a must. * Think having much higher speed limits (as far as humans are concerned, nonexistent), or mass coordinating movement over the entire traffic. ** We can reasonably estimate the minimum without bothering to ask how fast the tech will improve: Even if the tech were available now, think about fleet replacement costs which no one group would be too eager to pay. Best case, it's the typical 'make a concentrated pressure group lose for societal benefit' and we know how that politics goes. It will happen, but slowly. *** Another thing to account for is that there's no good reason to design an AV car like a normal car, and there'll be some iteration time over that too.
- mitthrowaway2 1y ago
  Higher speeds increases noise and stopping distance, even for autonomous vehicles.
  - yyyk 1y ago
    True, but this can be compensated for. Current vehicle design is based on human-operated gas vehicles - so it better be aerodynamic (to save gas), and a human needs to be in the front (to see) with only a glass to separate, and it needs a particular stopping action (again a consequence of carrying humans without enough separation). This has unfortunate implications for noise and stopping distance. Electric-powered AV can have creative designs to enable much quicker (yet safe) stopping action, an action which AV would also make rarer.
    - mitthrowaway2 1y ago
      Autonomous vehicles can have faster reaction time, but once they hit the brakes, stopping distance will still increase with speed because it's limited by the friction between tires and pavement. Unless you put parachutes or rockets on these vehicles, or thrust spikes into the pavement...

by xnx 1y ago

I wouldn't be planning any fixes infrastructure transit programs that didn't have an ROI in the next 7 years. It will be hard for anything to compete with the efficiency of Waymo.

digitalPhonix 1y ago

Each Waymo vehicle is probably close to half a million USD in just hardware cost. I don’t think fixed route transport infrastructure is going to have trouble competing on efficiency.

cyberax 1y ago
The hardware costs for Waymo are estimated at $30k.
- digitalPhonix 1y ago
  Source? They were targeting $7.5k for their in house honeycomb lidars and they have 12 of them - that’s 90k already. https://www.theverge.com/2021/8/27/22644370/waymo-lidar-stop-selling-autonomous-cars-laser-bear https://www.theverge.com/2021/8/27/22644370/waymo-lidar-stop... They also aren’t close to the $7.5k target (there isn’t any public source for that so you’ll have to take my word for it). Also $30k wouldn’t even cover the base vehicle.
  - cyberax 1y ago
    This is an estimation for the self-driving hardware cost (computers, LIDAR, sensors). It does not include the base car price, as it can be easily optimized down to almost nothing (sub $10k). The price is somewhat of a guess, several years ago, the hardware in Waymo was priced at $130k by Munro&Associates. But since then the cost of the LIDAR sensors fell by 90% or so, reducing the main expense. And Baidu has cars on the road that cost $30k for the _entire_ car. So presumably, so even a couple of pricier sensors won't affect the estimate too much.
    - rafram 1y ago
      > include the base car price, as it can be easily optimized down to almost nothing (sub $10k). Waymo uses brand-new electric Jaguars.
      - cyberax 1y ago
        Sure. Because why not? The cars are in the experimental stage, so they might as well use nice ones. But there is no _reason_ to use Jaguars and not specially-built smaller and less powerful cars, when Waymo finally starts a real rollout.
    - digitalPhonix 1y ago
      Source? None of those numbers make sense. > The price is somewhat of a guess, several years ago, the hardware in Waymo was priced at $130k by Munro&Associates 10 years ago they had even more sensors dotting the car, instead of one honeycomb on each corner they had 2, so I find $130k hard to believe given what we know about the sensor kit today. > But since then the cost of the LIDAR sensors fell by 90% or so, reducing the main expense. I do not know of any lidar that has done that, and Waymo makes their own and we know their price(-ish) (and quantity). I think they’ve actually gone up in price (but also capability - honeycomb 1 vs gen 2).
- cycomanic 1y ago
  Are you talking about raw material costs? Or is that one of these extrapolations of if we scale everything to millions of cars and realise no inefficiencies and nobody making any money in the supply chain?
  - cyberax 1y ago
    I'm talking about the current cost of the self-driving system, that is already produced by companies that charge a significant markup. With volume, it will go down more. I'm not including the base vehicle in the cost. It's highly variable, and can be as low as $10k for small personal intra-city taxis. China has already launched a $30k taxi: https://www.forbes.com/sites/bradtempleton/2024/05/14/baidu-launches-new-28000-robotaxi-in-wuhan/ https://www.forbes.com/sites/bradtempleton/2024/05/14/baidu-... This _completely_ blows any transit out of the competition. Literally nothing can come even close in the end-to-end efficiency.
    - cycomanic 1y ago
      Why does it blow away any other public transit? That can't be true because if you put the same self driving tech into a bus, you already am an order of magnitude cheaper per passenger (likely more). Moreover let's assume robotaxis are cornering the market and make all other forms of transport non viable. Why would the public then maintain the roads? So at that point at least costs are suddenly going to explode.
      - cyberax 1y ago
        Sigh. People are WAAAY too accepting of urbanist propaganda. Buses are _barely_ more effective than cars. A regular passenger car with 4 people is more efficient than a city bus. An EV needs 2.5 people (these numbers are for the US). The explanation is simple: 1. Buses have to drive _all_ _the_ _time_, even when there are few passengers. As a result, the average bus load tends to be around 10-20 people. And you can not increase the bus interval to compensate for it because it makes off-rush-hour bus commutes impractical. 2. Buses have INCREDIBLY polluting components: 2-3 drivers for each bus needed to provide the service. They are by far the dirtiest part of the bus. This part can be removed with the self-driving hardware, but... A full self-driving bus also makes no sense. It defeats the main advantage of self-driving: door-to-door transportation. That being said, self-driving mini-buses seating 6-10 people are a good idea for rush hour transit.

johnea 1y ago

It is really efficient, at sitting in traffic with all the other cars 8-/

cyberax 1y ago
And yet, still faster. That's the paradox of cars.
- rsynnott 1y ago
  Cars are very much not faster than trams in high traffic conditions. Nevermind heavy rail.
  - cyberax 1y ago
    Yes, they are. Do an experiment, drop ten points randomly on a city map. Then plot routes between them, using Google for transit and cars. For most cities, cars will be about 2-3x faster. For three simple reasons: no walking, no waiting, no stopping. And that's the paradox. Cars are better than transit, yet modern cities are hell-bent on destroying the car infrastructure. You can also check this site with isochrones for different transport modes: https://www.geoapify.com/isoline-api/ https://www.geoapify.com/isoline-api/
    - xnx 1y ago
      > Cars are better than transit And are even better when parking isn't a concern (e.g. taxi service)

rafram 1y ago

Public transit isn’t supposed to have a financial “ROI.” Are highways expected to turn a profit from tolls?

bluGill 1y ago
The libertarian answer is yes highways should. Most self proclaimed libertarians refuse to go that far - if you allow for highways to not make money then transit shouldn't be held to the higher bar.
- charlieyu1 1y ago
  I always find this a dumb idea. Transportation networks enable things to happen that comes back as taxes, that is. Same for education, or other public services.
xnx 1y ago
Fair, but how unprofitable should they be? -$5/passenger-mike? -$12/passenger-mile? I think we can do a lot better than the current US mass transit status quo.
- rafram 1y ago
  In 2022, the NYC Subway budgeted about $0.75 per passenger-mile (and that was during Covid, when ridership was very low) [1]. You’re really overestimating how much public transit costs to run. Private vehicles are an extremely inefficient way to move people around, hence the cost of Uber/Lyft/taxis. [1]: https://data.transportation.gov/Public-Transit/2022-2023-NTD-Annual-Data-Metrics/ekg5-frzt/explore/query/SELECT%0A%20%20%60agency%60%2C%0A%20%20%60mode%60%2C%0A%20%20sum%28%60total_operating_expenses%60%29%20AS%20%60sum_total_operating_expenses%60%2C%0A%20%20sum%28%60vehicle_revenue_miles%60%29%20AS%20%60sum_vehicle_revenue_miles%60%0AGROUP%20BY%20%60agency%60%2C%20%60mode%60/page/aggregate https://data.transportation.gov/Public-Transit/2022-2023-NTD...
  - xnx 1y ago
    NYC is special in that it's on of the few places that subways make sense in America. That said, operating costs are common but extremely misleading way to measure transit costs when new tunnel costs $2.2B/mile.

Johnny555 1y ago

Single passenger cars still have a problem with density even if there's no need for a driver. Combining multiple people into one trip can help, but also lessons the utility of Waymo if riders have to go out of their way to pick up additional passengers. Getting 1000 people downtown could be up to 2,000 Waymo trips (one trip to drop off the worker, another trip for the car to go back out to pick up another passenger). While one of these 56 passenger very light rail cars can do it in 18 trips. A light rail vehicle like the Siemens trains used in San Francisco can carry up to 200 people at crush loads, so that's 5 trips.

mmooss 1y ago
Yes, I always say: Next time you see mass transit pass, imagine all those people in their own private cars on the road.

cycomanic 1y ago

You mean it's very difficult to compete with a company that is massively subsidised by public infrastructure? That's what really killed freight rail in most of Europe, make the train companies pay for track maintance (often the rail companies even want this because it keeps competition out as well), while trucks atpapy very little of the cost they impose on the public (i.e. much higher road usage, causing most of the traffic issues).

jcranmer 1y ago

If this were true, cities would have abandoned mass transit for taxi system decades ago. The requirement for there to be a warm body driving a taxi isn't among the prime causes of its inefficiency.

yyyk 1y ago

The big AV transit efficiency gains* can only happen when nearly all human drivers are removed from the road. Alas, that's at least 20 years ahead of now**, or more (e.g. if tech stalls for some reason, though I consider it unlikely). Otherwise they'd be limited by having to account for human drivers and that limits speed and throughput enough so other solutions are a must. * Think having much higher speed limits (as far as humans are concerned, nonexistent), or mass coordinating movement over the entire traffic. ** We can reasonably estimate the minimum without bothering to ask how fast the tech will improve: Even if the tech were available now, think about fleet replacement costs which no one group would be too eager to pay. Best case, it's the typical 'make a concentrated pressure group lose for societal benefit' and we know how that politics goes. It will happen, but slowly. *** Another thing to account for is that there's no good reason to design an AV car like a normal car, and there'll be some iteration time over that too.

mitthrowaway2 1y ago
Higher speeds increases noise and stopping distance, even for autonomous vehicles.
- yyyk 1y ago
  True, but this can be compensated for. Current vehicle design is based on human-operated gas vehicles - so it better be aerodynamic (to save gas), and a human needs to be in the front (to see) with only a glass to separate, and it needs a particular stopping action (again a consequence of carrying humans without enough separation). This has unfortunate implications for noise and stopping distance. Electric-powered AV can have creative designs to enable much quicker (yet safe) stopping action, an action which AV would also make rarer.
  - mitthrowaway2 1y ago
    Autonomous vehicles can have faster reaction time, but once they hit the brakes, stopping distance will still increase with speed because it's limited by the friction between tires and pavement. Unless you put parachutes or rockets on these vehicles, or thrust spikes into the pavement...