A swarm of robots built this tunnel [video] | Hacker News Reader

A swarm of robots built this tunnel [video]

youtube.com

85 pointscjg38 comments

Now that's an interesting idea. Dig a ring of small holes around the perimeter of the tunnel and use them to build the tunnel liner. Then remove the dirt and rock in the middle.

Microtunneling has been around for years. It's done with small tunnel boring machines, to install pipes of various sizes.[1] The drives are usually not that long, because these are pushed from the starting point, not self-propelled like the big TBMs.

That's what they seem to be doing here. There's some hand-waving around the "special borepipes". Those are apparently drilled by existing techniques. Only when all those tubes are in place around the perimeter of the tunnel volume do the robots move in and grout. Trying to do all that pumped concrete grouting work with those small tubular robot vehicles is a neat trick. The animations show them drilling small crosswise holes longer than the tube diameter, which is a bit suspicious.

All this would seem to be limited by how far you can drive the "special borepipes" using standard drilling techniques. That can be quite a distance, though; the current microtunneling record is 2.2 km.[2] That's more than enough for most urban projects. You're going to want stations and access points along the route, so the drive can be done in sections.

(When watching the original post video, start at 3:30 to skip a long ad.)

[1] https://www.youtube.com/watch?v=teVrJs6CSe0

[2] https://www.napipelines.com/building-big-microtunneling-terr...

What is the advantage? You have to remove the dirt in the middle either way. And with a bigger drill you can get through tougher obstacles.

In soft ground, you have more protection against the roof caving in and water incursion.

For harder ground/solid rock it seems like you could slice off sections of the inner core material and not have to drill as much total material.

This could make blasting the core easier because there is now room for the explosion to expand into (assuming portion of the tunnel liner is removable) and the shape of the tunnel can be optimized to be non circular.

A somehow similar approach has been done some 30+ years ago for an underground train station in Milan, Italy:

https://www.rocksoil.com/index.php/it/ricerca-e-sviluppo/22-...

The bore pipes in that project were much larger though, and became part of the tunnel supporting structure.

Besides the small test they showed in this video, the animations look a lot like "pure theory in ideal terrain", very likely this technique, if it will ever be proved actually viable, will be suited to only a few, and very homogenuous, kinds of geology/terrains.

That was a nice job in Milan. It's a very common case - a shallow tunnel in soft soil. Usually, those are made by cut and cover. This approach has less disruption to surface activities.

The content about the new method starts roughly here: https://youtu.be/bfJY0syocfU?t=227

The Wadsworth constant strikes again (:

This is definitely an interesting take on tunnelling, but I struggle a little bit to see how this would be either faster or cheaper than conventional methods.

First off you have to buy a load of strong and mechanically complicated robots (expensive), and then many small tunnels to put them in. Drilling these small tunnels won't be that much faster, and certainly no cheaper than one big one, assuming they're using conventional microtunneling equipment (required for the accuracy).

Also, what happens if a robot breaks down mid operation? Are you going to go fishing, or try and send in a human to dig it out? Worst case, you can't do anything and have to revert back to manual digging the whole way. You really don't want a timeline to blow up like that in a real project!

This system has the big advantage that everything is modular, so a bigger tunnel just requires more robot mini-tunnels. Economies of scale should make everything a bit cheaper, especially since the robots and mini-TBMs would be reusable.

Getting geology before you tunnel is also a benefit, but if this was so good, surely tunneling contractors would already do this by sending just one micro-TBM ahead of their main one! What can this system actually do if bad geology is detected? Not much...

Fundamentally the robots don't really solve anything except overcomplicating tunneling. You still have pretty much all the expensive parts (materials, spoil removal, planning, investigatory drilling, sinking shafts, etc) but now with the additional costs and uncertainty of having loads of robots too.

If you asked me to design a lining-first tunneling method, I'd completely ignore robots and try and use existing technology first. Take horizontal directional drilling equipment, make it cheaper and just directionally drill a ring of wells. Pierce the casing at a regular interval, and then fill the whole thing with high pressure grout. No robots needed and this happens everyday, just vertically rather than horizontally.

So the robots are building the shell of the tunnel before the material within the tunnel (spoil) is removed.

The robots do not seem specifically involved with removing spoil or even carving it into smaller pieces (at least pieces smaller than the tunnel diameter).

Does this mean building tunnel shells is the hardest part of tunnel building, not breaking up and removing spoil?

Or does this approach somehow make it easier to fragment and remove the spoil since it's now in tunnel-sized chunks?

(Edit: Yes, stabilising the tunnel is the hard part, see excellent and thorough answer by rmccue)

Stabilising the tunnel is the hard part; anyone can dig a hole, but keeping that hole stable is what makes it a tunnel. The big innovation from Brunel and the ancestor of most modern tunnels was the tunneling shield [1], which provided an area to safely work ahead of the supporting structure.

Modern Tunnel Boring Machines (TBMs) incorporate a tunneling shield, with the addition of a large machine behind the cutting head which builds the tunnel walls as it goes along.

The innovation here is to build the stabilisation first and tunnel out the rest; it seems more akin to pipe/box jacking [2] in that sense, where you basically force the outer walls through the material using hydraulics, then excavate the spoil. Their demonstrations so far are at the scale where box jacking would probably make more sense, but if they can deliver on the promise of having effectively an arbitrary long tunnel then that'd make the technique more broadly applicable.

[1]: https://en.wikipedia.org/wiki/Tunnelling_shield [2]: https://en.wikipedia.org/wiki/Pipe_ramming

Seems very "Popular Science" to me. I expect to see that when I see the Moller Sky Car in production.

I forgot about that but it looks like it's officially dead. That used to pop up all the time. I'd love to hear a background story of what really happened there.

IIRC, it was mostly a scam to get investors to spend money. "The car of the future"... since the late 80's.

Looks much more interesting than the Boring Company. They look large enough to fit something other than Tesla's through them.

Definitely. Also seems like a much more scalable and flexible strategy than the Boring Co's mini-boring machine.

I'm not a construction engineer by any means but there are a few things I can't warp my head around. Why are they using robots at all? It seems like a pretty big hassle to construct those robots that deposit large amounts of material with high pressure through a hole that needs to be drilled in their own tunnel. Why not excavate the volume between the small boring holes with the same technique they used to create them in the first place? Bore those smaller holes with equal spacing, fill them with concrete, and then repeat for the volume in between. What about steel? Not needed for in tunnels?

When you tunnel in very loose material, it is a long established technique to drill steel reinforced micropiles (in jargon often called "umbrellas") like in this sketch:

https://www.rodio.ch/rodio-geotechnik-ag/pipe-umbrella-rodio...

Typically these are no longer than 18 meters and have a 3 meters overlap, there are limits in the precision of the drilling and to allow the "next" umbrella to start at the right height/width holes need to have a slightly open angle.

Even if it doesn't seem so to the casual viewer, drilling a (usually) 120-150 mm diameter hole 18 m long results in a non-straight hole due to the flexibility of the drilling rods and the possibility of deviations caused by the presence of different materials, it is not at all an "exact science" and there can be easily errors at the end of the 18 m hole of up to 50 cm or so.

There are of course techniques that allow to drill much straighter holes using external casings, like the type that seem to be used in the video, but they are much more complex (and costly), and anyway the precision of these is far from being "high", in case of long holes.

Seems like a cool concept but also quite early stage with unsure viability. Curious how transportation would change if they succeed in making tunnels x2 cheaper.

That still leaves American tunnels ridiculously more expensive than elsewhere in the West. There are more reliable wins elsewhere in the overall process.

I figured the Boring company had a similar approach, but when I researched their technique ... I was very disappointed. This is really cool. Can't wait to see it go into production.

How does Boring's technique differ to conventional current approaches? (i.e. I'm too lazy to repeat your research..)

Can't wait for Adam Something to tear this apart.

While this method can create new tunnels, I see it’s first best use at enlarging existing tunnels by setting the “concrete” type stuff surrounding an existing tunnel but at a larger diameter (well, also need not be round!). Then demolition of the existing tunnel walls and some cave in of other materials and you have a new larger tunnel after cleanup.

You'd still lose use of the existing tunnel. Why not build a new second tunnel?

If you want to increase the loading gauge of a railway, it may be expensive/difficult to reroute the track to go through a different tunnel. Also, making the track less straight may impair the maximum speed of the line.

@6m Could build tunnels up to 10 times faster and at half the cost.

A lot of US cities could be retrofitted with subways.

Even NYC metro area badly needs a few more tunnels:

https://en.m.wikipedia.org/wiki/Gateway_Program_(Northeast_C...

"Could build tunnels up to 10 times faster and at half the cost."

Maybe. So far they have only drilled through wood and not stone.

Seems to be missing a first set of robots that digs the pipes, heck creates the pipes, for the second set that creates the tunnel, before the third set clears the hole. Whew, that is a lot of bots.

Just ctrl-c and ctrl-v, that is how you do it in Factorio at least.

This is 100% how the first permanent (i.e. not tin cans) lunar habitation will be built, possibly Mars too.

I think AI will leapfrog or replace us before we habitate Mars. And besides, we're overfit to the biosphere we evolved on.

Current AI can't think for itself and has no desires, certainly not a desire to inhabit other planets or to replace us.

What gives you such confidence in your statement?

using a throwaway account gives one all sorts of confidence.

I come from the future

Hate these types of videos that don't get to the point

Reply on news.ycombinator.com