[outoftheice]

Came across this post and thought it was interesting. We've had spirited discussions on here before about grade separating Green Line so that they could have gone with driverless trains like Vancouver's Skytrain.

I would argue that driverless buses traveling in mixed traffic have a far higher risk profile than driverless trains operating on tracks that aren't completely grade seperated. I wouldn't be surprised if we see a movement within the next 10 years to start to modify existing LRT fleets to support driverless operations.

https://twitter.com/x/status/2044872939921649893

 

[CalgaryTiger]

Came across this post and thought it was interesting. We've had spirited discussions on here before about grade separating Green Line so that they could have gone with driverless trains like Vancouver's Skytrain.

I would argue that driverless buses traveling in mixed traffic have a far higher risk profile than driverless trains operating on tracks that aren't completely grade seperated. I wouldn't be surprised if we see a movement within the next 10 years to start to modify existing LRT fleets to support driverless operations.

https://twitter.com/x/status/2044872939921649893

Norway... a place with winter... how is this going to work on a snow day like yesterday. You can't just hire a thousand drivers for snow days.

 

[MichaelS]

Why do you think a human can drive in winter better than an automated vehicle? Cameras can be cleaned off (heated, or placed inside with windshield wipers clearing the view). If a human can see the road, a camera should be able to as well (and have the benefit of not being distracted and reacting much faster).

Besides, from what I could tell, the human bus drivers (through no fault of their own) didn't really "work" yesterday very well either.

 

[CalgaryTiger]

Why do you think a human can drive in winter better than an automated vehicle? Cameras can be cleaned off (heated, or placed inside with windshield wipers clearing the view). If a human can see the road, a camera should be able to as well (and have the benefit of not being distracted and reacting much faster).

Besides, from what I could tell, the human bus drivers (through no fault of their own) didn't really "work" yesterday very well either.

The technology is not there. In theory, sure it removes human error. In reality, at this point these systems are good in controlled and straightforward environments where they can react immediately compared to a distracted human, but they fail a lot of the times too. Someone I know has a Tesla and cannot even use the Cruise Control because it frequently phantom breaks at ghosts and sounds alarms at simple lane changes because the lines are not painted how it expects them to be painted.

 

[lemongrab]

Came across this post and thought it was interesting. We've had spirited discussions on here before about grade separating Green Line so that they could have gone with driverless trains like Vancouver's Skytrain.

I would argue that driverless buses traveling in mixed traffic have a far higher risk profile than driverless trains operating on tracks that aren't completely grade seperated. I wouldn't be surprised if we see a movement within the next 10 years to start to modify existing LRT fleets to support driverless operations.

https://twitter.com/x/status/2044872939921649893

Red line north and blue line west (and green line SE) are pretty darn close to full grade separation - at least from cars. I wonder how much savings could be found if operators handed over to autopilot at Sunnyside and Kerby Stations (and then take over the next inbound automated train)? Perhaps remote monitoring for the few crossing spots the rest of the way

 

[lemongrab]

Why do you think a human can drive in winter better than an automated vehicle? Cameras can be cleaned off (heated, or placed inside with windshield wipers clearing the view). If a human can see the road, a camera should be able to as well (and have the benefit of not being distracted and reacting much faster).

Besides, from what I could tell, the human bus drivers (through no fault of their own) didn't really "work" yesterday very well either.

I think a lot of winter driving by humans happens at speeds way higher than a computer would calculate to be within a reasonable risk tolerance. Hard to articulate, but days like yesterday most of the time it feels like you're in control while out of control...not sure how a computer would handle that amongst human drivers

 

[jhappy77]

Red line north and blue line west (and green line SE) are pretty darn close to full grade separation - at least from cars. I wonder how much savings could be found if operators handed over to autopilot at Sunnyside and Kerby Stations (and then take over the next inbound automated train)? Perhaps remote monitoring for the few crossing spots the rest of the way

It's definitely possible. I've ridden in self-driving cars in SF, the technology is good enough to safely handle navigating complex street environments with crazy drivers and pedestrians, it would certainly be 100x easier for a train where you basically only need to decide whether to stop or go.

The main question is would there be a vendor able to do a partial retrofit to an existing LRT system? I imagine most companies are focused on designing systems and lines to be automated from the ground-up. I wonder if other cities have done something like this yet.

 

[MichaelS]

The technology is not there. In theory, sure it removes human error. In reality, at this point these systems are good in controlled and straightforward environments where they can react immediately compared to a distracted human, but they fail a lot of the times too. Someone I know has a Tesla and cannot even use the Cruise Control because it frequently phantom breaks at ghosts and sounds alarms at simple lane changes because the lines are not painted how it expects them to be painted.

Tesla's FSDv14 (I think that is the latest version) seems pretty much there now. Maybe not quite for edge cases such as yesterday, but for 99% of the time, I would say it is good.

But, a technology like that raises a bigger question of how we plan transit. Would we need large buses running on fixed routes through all neighbourhoods? Or, a fleet of smaller ride-shared vehicles that operate on demand, taking you from your house to key transit hubs like our C-Train and BRT stations, and vice versa when departing these trunk lines.

No more expensive, mostly empty busses running through areas in the evening, and much more convenient for riders as it is not fixed to infrequent schedules and unreliable transfers.

 

[Red Star]

Came across this post and thought it was interesting. We've had spirited discussions on here before about grade separating Green Line so that they could have gone with driverless trains like Vancouver's Skytrain.

I would argue that driverless buses traveling in mixed traffic have a far higher risk profile than driverless trains operating on tracks that aren't completely grade seperated. I wouldn't be surprised if we see a movement within the next 10 years to start to modify existing LRT fleets to support driverless operations.

https://twitter.com/x/status/2044872939921649893

I could see the use of autonomous buses on some routes. The routes would have to be:

1. have low vehicle traffic.
2. Low speed.
3. Low ridership routes.
4. Short distances.

I think a circular feeder route in Scenic Acres to Crowfoot LRT station could probably work pretty good. Autonomous buses could probably deliver a more frequent service than a manned bus. I also think maintenance, charging infrastructure and cleaning would be a challenge. You'd have to get these things to home back to a bus depot somehow.

 

[trtcttc]

It's definitely possible. I've ridden in self-driving cars in SF, the technology is good enough to safely handle navigating complex street environments with crazy drivers and pedestrians, it would certainly be 100x easier for a train where you basically only need to decide whether to stop or go.

The main question is would there be a vendor able to do a partial retrofit to an existing LRT system? I imagine most companies are focused on designing systems and lines to be automated from the ground-up. I wonder if other cities have done something like this yet.

Absolutely possible to retrofit. I’m pretty sure our LRT run on fixed block signalling. Replacing that with a CBTC moving block system would be very expensive. Toronto refit their line at a cost of $800M per line. And with level crossings, I don’t think you can run at level 4 automation (no driver), you’d need a supervising driver even if the train is driving itself.

 

[DougB]

If the downtown LRT was built as a subway from day 1, I 100% guarantee we wouldn't have any problems with the SE LRT today.

Because we wouldn't be anywhere near building a fifth line, we'd be building the extensions to McKnight and Dalhousie, and doing engagement on the West LRT.


It doesn't matter how many colours are on the map. BART lines have much lower frequency; three of the four lines are 20 minute headways and the yellow is 10 minutes. Put all together, the Transbay Tunnel handles about 15 trains per hour per direction at peak.

Running three lines each "only operating at 10 minute headways" is 18 trains per hour. Running two at five and one at 10 is 30 trains per hour, which is theoretically possible but would be a terrible operating condition -- no room for error and no ability to expand.

BART is really over-engineered commuter rail. The overly long trains require overly long stations. It primarily functions to shuttle commuters into SF and Oakland. No one uses BART to get around within SF or within Oakland.

 

[DougB]

If you do a Stephen Ave cut and cover you don’t need to go deep at all. Stephen Ave itself would act as a pretty good crossover between lines. Too bad they’re redoing all the utilities under it now but at least they know where the utilities are. It’s too bad we took so long to get the Green Line to this point… if the Green Line were open now, at least part of it, I think planners would’ve been looking at a Stephen Ave subway as a real possibility. Instead, we’re still figuring out the Green Line and Stephen Ave is being redone and likely won’t be touched for 30 years because everything was just redone. Who knows, maybe the placement of the new utilities makes a shallow bore realistic?

The Weklund Center expansion and Olympic Plaza revamps are a missed opportunity to build a cut and cover tunnel.

 

[DougB]

It depends how you design the stations. no need for a mezzanine if you take a bit more surface land.

Totally possible to build a station this deep as an 'underground' station. You do end up in an optimization problem though: if you go deeper you can build the station box far less strong with some nice arches. But then your access solution is likely more expensive. Which is better? Don't know.
View attachment 723229

Shallow stations like many of those in NYC are far more convenient as users can enter and exit more quickly. Mezzanines are completely useless

 

[lemongrab]

I think a circular feeder route in Scenic Acres to Crowfoot LRT station could probably work pretty good. Autonomous buses could probably deliver a more frequent service than a manned bus. I also think maintenance, charging infrastructure and cleaning would be a challenge. You'd have to get these things to home back to a bus depot somehow.

I think you'd take the depot out to the transit hubs - for charging and basic cleaning.

I wonder about thinking even more radically - could a city enter into a monopolistic partnership with a company like Waymo to handle everything (all rideshare and low volume transit).

You still have LRT and BRT and high volume bus routes (automated where possible). But then your waymo fleet handles everything else: I open my waymo app and I can choose between:

- door to door service at the variable rate (ie. same as current Lyft/Uber)
- bus stop (but really anywhere on prescribed streets) to transit hub service: regular transit fare, potential to pick up passengers en route

Waymo is providing the tech - for both the app and the vehicles. City Transpo dept. services the fleet.

Plenty of ways you can slice up the revenue/expenses/liabilities, but the key thing is that you are combining profitable custom trips with highly subsidized low volume transit service to achieve peak scale/efficiency for both. Win-win deal - Waymo earns more and Transit spends/subsidizes a lot less than operating separately in a competitive environment.

It feels a bit dystopian, but I think you'd want to compartmentalize the vehicles - like in a limo where both sides have to opt in to rolling down the screen.

Obviously the idea of rideshare replacing/complementing transit is not new, but it's usually presented as public mass transit and private low-volume trips being separate operations.

The Weklund Center expansion and Olympic Plaza revamps are a missed opportunity to build a cut and cover tunnel.

1000%. They should have at least designed and cut out the station box.

 

[jhappy77]

In any case, as driverless cars become more of a thing, transit will need to recalibrate to focus on what only transit does best: Moving large volumes of people to high demand destinations quickly and efficiently

Crappy coverage routes will eventually be completely outcompeted by autonomous rideshare vehicles - hell, even now, most coverage bus routes are slower than biking (and even walking in some cases like Scenic Acres, once you consider the average wait time). But driverless cars cannot overcome the basic geometry problem that most roads can only fit so many vehicles.

If we're smart about designing transit, it could even lead to a transit renaissance. If autonomous vehicles solve the last mile problem, we can concentrate transit service on the fastest, highest demand routes, and make them super high frequency & have bus priority measures - that means less cars on busy roads, way less money spent on upgrading roads and interchanges, and fast high quality service that is competitive with driving.