[aeternum]

Imagine if the subway car you were riding in could detach from the train, take an exit tunnel to the surface streets then drop you off within a block from your house.

AVs in tunnels allow for a wide range of density from 2-seater to high density bus/train-car like vehicles that can convoy through tunnels but still do point-to-point and on-demand.

The main challenge is convincing the train-obsessed that 3d travel is better than 1d travel.

[mikaTheThird]

1. If my vehicle can be detached from the convoy, why even attach in the first place. Just get me to my destination as efficiently as possible. Also being dropped a block from your house doesn't really sell it.

2. Aren't AVs suppose to greatly alleviate the traffic problem on existing roads since AVs can/will coordinate efficiently. I'm actually surprised that the loop doesn't operate autonomously and requires drivers

3. Wouldn't flying cars be the ultimate 3D model, as oppose to digging limited tunnels.

[aeternum]

1) The "attach" would likely be virtual, just meaning a tight following distance. You could have optional shock absorbers if you're really worried about collisions in the tunnels. Most likely not needed though.

2) Yes I'm surprised also. Any theories on why? Too much of an edge case for tesla's focus?

3) I think yes but very few building have heliports, far more have parking garages that could double as tunnel ingress/egress. I think both will be used but tunnels support higher density and have less dependence on weather.

[jraph]

> if you're really worried about collisions in the tunnels

Cars can fail, you have to leave enough room to allow cars behind to stop in time to avoid a (cascade) crash.

[aeternum]

This is mostly due to human attention and how the human vision system works.

[jraph]

There's also inertia, especially for cars that heavy. You can't go from 140 km/h to 0 instantaneously.

(yeah, taking 140 km/h, quite high for a car especially in tunnels like this, but anything slower would be slower than our slow trains and would be quite disappointing)

[jraph]

> The main challenge is convincing the train-obsessed that 3d travel is better than 1d travel.

I suppose you do that by making the system as convenient and efficient (cost effective both for the operator and the users, energy efficient, good capacity, good speed, not more wasteful hardware wise, good longevity in the long run) as, or more than a properly setup train system?

Another commenter has a good point though, in a place that's not willing to invest in good public transport, there's a void to fill I suppose.

[kennyadam]

I have so many questions...

How many exit tunnels will need to exist for the final above-ground stage of the trip to be quick and not increase the total length of the route significantly?

If there are exit tunnels all over the place, how long would it take to build all of them? If there aren't, how do you avoid massive jams as the multiple lanes merge to the single exit tunnel lane?

How do all the exit tunnels connect to the surface streets in the many areas where there's no room for another lane to be built anywhere?

[aeternum]

They would generally be single lane tunnels so no merge issues. Breakdowns would be rare since these are well-maintained but if one did occur, the fleet would be programmed so the car behind simply pushes the AV through to the next exit.

They could terminate to converted underground parking garages (cities already have quite a few) and with AVs the need for personal parking would be greatly reduced. People would have plenty of time to get in and out without blocking the tunnel.

It's actually insane that most trains stations block the track, how quickly would you fire the dev that insisted on holding a perf critical mutex while waiting on user input?

Building the tunnels does need to be automated but that's exactly what Boring company is doing.

Ultimately human driving will be banned within cities. It's just far too dangerous. It's almost as deadly as smoking especially in cities and the risk asymmetry is extreme: pedestrians and those around the car are in far more danger than the driver.

[amluto]

It sounds like the Loop, as it currently exists, is not actually able to do any of the above. It can take you from an unpleasant “station” to another one. This limitation isn’t fundamental, but it is still a limitation of the system as built.

As for the idea of a convoy of magic cars, it’s a nice idea but I don’t think it holds up to actual math. Specifically, area used. There are plenty of variants of this graphic, and here’s a tongue in cheek one:

https://xkcd.com/2684/

The fundamental issue is that individual low-capacity vehicles are large and use lane space inefficiently. Sure, perfect self driving might improve the situation, maybe by a factor of two or, extremely optimistically, a little more. But a factor of two isn’t really good enough.

Consider the 405 freeway in Los Angeles through the Sepulveda pass. It carries over 300k passengers per day (I’m not sure whether this counts Sepulveda Blvd as well, but, either way, we’re talking about 12+ traffic lanes), and traffic sucks. Maybe really advanced self driving could double capacity. Sticking a freeway lane in a narrow tunnel does nothing to speed it up or increase capacity.

In contrast, Wikipedia’s estimate of light rail capacity is about 8x as high for a rail track as compared to a lane of traffic. A single pair of light rail tracks could perform comparably to the entire Sepulveda Pass mess of roads. If actual intelligent, retroactive planning happened, there would be four tracks for one express and one local each way, New York style, and the whole assembly would be quieter, cheaper, far faster (at least during rush hour), and use massively less energy. But self-driving Teslas would not get most of those gains.

(E-bikes and the kind of culture that would get a hundred thousand or so people to ride them over the pass on their commute would also do the trick, but good luck with that. Although… the pass isn’t that long, and maybe Dutch-style levels of dedication to cycling might actually make it work. Los Angeles with real bike infrastructure would be quite a sight!)