Hacker News Reader

Comment by OrwellianChild | Hacker News Reader

Don't fully get the benefit of moving cars on rails vs. moving cars on wheels... Higher fixed costs to power the rail vs. just letting cars auto-drive with internal propulsion. What am I missing?

View on HN

Or how about moving trains on rails?

I am of two minds about self driving cars. From a technology point of view, I think its great, but I don't really see what problem they solve that would not be better solved by ubiquitous public transportation (trains, mostly). Now, Musk wants to carry cars on sleds through tunnels under LA? Why not just build some damn trains?

EDIT: the real news is that he wants to improve boring machines which is not a bad goal. There will be more and more tunnels regardless of what you actually put in them, so faster, more efficient boring is a worthwhile goal.

I thought this too completely inefficient for your transport dollar. I actually thought maybe this was satire.

It's about the last mile connectivity. People don't want to walk/uber from home to the train station and from the train station to work. Having a car on rails solves this problem.

When public transport is ubiquitous, the last mile is really just that, a mile, tops (in urban environments). The difference between public transit somewhere like Japan, where everything is linked and you have intermodal hubs efficiently connecting higher speed and lower speed transit and what we have in US cities is staggering. It's downright easy to cross the entire country in the span of a few hours, only stepping foot outside for the very short walk to the nearest metro. Then you usually just have metro -> shinkansen -> metro, all connected to eachother, to get to where you have to go. You might need to change your metro once, but that's hardly an inconvenience when the connections are in the same building and you just walk a few feet from one platform to the next.

Compare that to Seattle, where my friend got lost trying to get from the light rail to the incredibly slow train up to Vancouver because they're several blocks apart. And then had to do the same on the Vancouver end for the same reason. To make matters worse, the train between the two cities is actually slower than the train between Tokyo and Hiroshima, with roughly 3x the distance.

Shared self-driving cars would essentially be an ultra-efficient bus service. Instead of huge (often half-empty) vehicles driving set routes, you have small custom electric vehicles (remember, the interior could be anything) going exactly where they're required.

Keep the trains, but replace buses, taxis, and probably quite a lot of private cars with an automated fleet. It would be a huge change, a huge improvement in many lives.

It reduces the complexity of the tunnel needed for combustion vehicles (most cars today) by not needing more complex exhaust infrastructure, and thus makes this idea more immediately viable.

If you want cars to be more closely packed, travelling at higher speeds and controlled by computers, you probably don't want individual car owners to be able to bring the whole thing to a screeching halt through sloppy maintenance or shutting the engine down

(See also: Channel Tunnel)

> If you want cars to be more closely packed, travelling at higher speeds and controlled by computers

Isn’t that called “a train”?

It allows the entire system to be controlled by one system - e.g., no need for each car to have its own LIDAR system. ATO is way easier than self-driving cars. You get all the benefits of being able to group vehicles together, have all stop simultaneously in response to hazards, etc.

Plus, like with most rail systems, you get the big benefits of having a custom right-of-way. No need to contend with pedestrians, people parking in the traffic lane, vehicles that aren't as fast as others, etc etc.

Of course... if you want to build a new underground right of way, what's the best way to use it? Is a big open question in my mind. You could go for electric BRT, for instance, or big-dig style freeway.

The rolling friction of rails is really really low. And couple that with removing driver variability out of the equation and it makes some sense.

The fact that the idea is that most cars would be able to use the system, not just ones from a certain company. Additionally, 130mph is pretty sporty to maintain for all vehicles using the system! Most cars probably can't even achieve that speed all out, let alone safely (even if self-driven).

Rails allow the sleds to not require batteries. This is especially important if you're moving things at inefficiently high speeds.

I suppose you can't control (or detect?) if the car fails, and if that is the case, the whole system would be screwed.

These platforms will have to strap the car down somehow to prevent someone from driving off the platform while it's operating.

A crash at that speed would wreck that tunnel and put it out of service for weeks.

Reply on news.ycombinator.com