Wire-crawling robot that lays optical fiber

Submitted by brad on Thu, 2006-10-26 00:13
Topic: 
Internet

In thinking about how to reduce the cost of bringing fiber to everybody (particulaly for block-area-networks built by neighbours) I have started wondering if we could build a robot that is able to traverse utility poles by crawling along wires -- either power, phone or cable-TV wires. The robot would unspool fiber optic cable behind it and deploy wire-ties to keep it attached. Human beings would still have to eventually climb the poles and install taps or junctions and secure these items, but their job would be much easier.

Robots that can crawl along cables already exist. The hard part is traversing the poles. Now it turns out finding live electric wires is something that's very easy for a robot to do. They stick out like a live wire in the EM spectrum. The poles of course have insulators, junctions, tie downs and other obstacles. Crossing them may be hard in certain cases (in which case a human would have to help, either by tele-operation, or by climbing the pole.) It may be possible to have a very small robot that is able to follow the current (easy to tell the lines to the houses from the main lines) and cross a pole like a bug and then, once safely on the other side, pulls the larger robot with a small tether. Again, it won't always work but if you can get it to work enough of the time, you can install fiber with far less time and labour than the manual approach. Fiber of course can be tied to power lies because it is non-conductive material, though it's even better if you can run it along phone or cable lines.

Not that any of these companies will want to give permission to competitors. And you want to pull multiple fibers, not so much for the bandwidth -- we can do terabits in a single fiber if we want to -- but for the backup when one fiber breaks.

If the robots get good enough, they could even string fiber into rural areas, following long chains of power or phone lines with just a single human assistant. Of course overhead wires are going to be more prone to breakage, but with these robots, repairs could be fast and cheap.

There are already robots out there which can crawl storm sewers to install fiber. This is another alternative, though that's good too. Indeed, a robot that can even crawl real sewage lines to put in fiber which comes out your household stack is not out of the question, if it's in a strong enough casing.

Share on:
  • Reddit logo
  • Facebook logo
  • Twitter logo
  • LinkedIn logo

Comments

Alex Goldman

Thu, 2006-10-26 09:10

If we can (try) to design hydro power plants so that salmon can get through them, surely we can design power poles with small portage or ramp areas for the robots.

(A friend of mine claims, only half-joking, that that the Disability Act's ramps are for our future robot masters.)

On a more serious note, as long as power systems are based on standards, we can design the standards for poles to enable robot traversal. A standards team would also want to look at severe weather survival and other aspects of pole design.

Brent Chapman

Thu, 2006-10-26 20:40

Identifying power lines is pretty easy: they're always at the top of the pole, for safety reasons.

The trend in many cities is to get rid of the poles and overhead wiring, for esthetic reasons. Existing overhead wiring is moved underground, and the poles are removed; new neighborhoods are constructed with all wiring underground right from the start. My neighborhoods in both Mountain View and San Francisco have gone through "undergrounding" projects in the past few years.

Having watched installation crews work, I'd say that the tricky and time-consuming part isn't getting the cable from pole to pole; that's usually pretty easy. The tricky and time-consuming part is the attachment work that has to be done at each pole. You've got to install some sort of anchor on the pole, then attach the cable to that, then go on to the next pole. I'm not sure how amenable to automation that part is.

In many cities, the poles are municipal rights of way, and the city government has control over who can put what on them. On the one hand, that means you don't have to negotiate for access with potential "competitors" such as the telcos, cablecos, and power companies. On the other hand, that means that you have to negotiate individually with every separate municipality; that was a big part of Metricom's downfall, I think, and also a big reason why I'm not hopeful about the prospects for widespread public WiFi (from MetroFi and the like).

brad

Fri, 2006-10-27 01:10

I'm talking about a robot that can actually traverse most poles, leaving only a few tricky ones for human assist. It's not out of the question though that you could have remote human assist with a tele-operator who is miles away. The robot is presumably wrapped itself along the wire or has a tether, and probably has multiple parts, so when an obstacle comes, one part crawls around the problem -- being reeled back on a tether to the part still on the wire if it falls off -- possibly making several attempts and falling several times until it figures a way to crawl to the next segment of wire -- after which it attaches and tells the back part to release and pulls the back part along the path it took. (They have to be smooth and the same size.)

Now this is something a bug can do easily enough. We don't yet have robots as smart as bugs but I think we're getting closer. And a teleoperated robot -- we are trailing a small bundle of fiber optic cables after all, so we've got bandwidth -- might be able to do this today. You could imagine robots crawling the entire city, stopping at tough junctions and a pool of tele-operations getting them past the problems. Then a few problems need an onsite human. If a robot has to wait it's not so expensive.

Of course if you are buying wires, then bury a conduit full of fiber. Everybody's doing that I hope. But we need alternative high speed data pipes to our homes today, and cheap. So we can get real competition in the first mile.

Anonymous

Wed, 2006-11-01 17:15

So there's a robot crawling along a phone wire from pole to pole, stringing a fiber cable. Where does the fiber cable come from? The robot can't pull it along from behind -- it's anchoring it as it goes. It has to pull it from ahead, which means there needs to be something on the ground that's following along and feeding the cable to the robot.

You don't want to splice fiber cables (it's really labor intensive, and
reduces the quality of the fiber) so you're going to have a big spool of cable there -- not like a garden hose reel, but like those cable-spool tables. This means you'll have a truck (or at least a trailer with
the spool on it -- and something to pull the trailer) below.

So the job of the robot can be simplified significantly -- since there'll be a truck and a human below it. Indeed there are some similar devices already in existence, which run fiber along high-tension tower ground wires (the top wire), winding a third cable around the two to keep them together despite the wind. This robot requires a human to climb each tower -- but high tension towers are much further apart than telephone poles.

By the way, the easiest way for such a robot to traverse a telephone pole is for the robot to be longer than the pole is wide. Then it can just reach around the pole, grab the cable on the other side, and do the traverse. No need for the complexity of a microbot hitching a ride on a bigger robot.

brad

Wed, 2006-11-01 18:17

I was presuming the robot had a spool on it that was a few hundred meters long, enough to do lots of poles. It would need to be re-spooled when it runs out (ideally at junctions where you would splice anyways.) There a human would bring the respool. Now, to get really clever, the new spools would be designed in a way the robot can load itself, so all the human has to do is drive along the poles and hang a fresh spool on the side of every 10th pole (or whatever the right numbers are.) The spool would have a transmitter to be easy to find. Or it might even be a robot itself, able to climb the pole and come back down for collection and redeploy.

This is for urban. If you want the fibers to be longer than the robot can carry on its back, yes, you are going to need ground assist to carry lots of fiber.

John Kasunich

Tue, 2006-11-07 12:45

That spool is gonna be a lot bigger than you think. Take a look at http://www.blackbox.com/files/productdetails/18760.PDF to see what real world fiber optic cables look like. These aren't the thin single fibers that are used as jumpers between indoor systems in a computer room. Weather resistant cables suitable for outdoor use have a diameter of approximately 10mm, bend radius of 10 to 20cm, and a weight of 80kg/km and up.

Splicing is very labor intensive, and is usually avoided whenever possible, so 100 meter long spools are out of the question. If you DID have a 100 meter spool, it would be about 40 cm (16 inches) core diameter diameter, 20cm (8 inches) long, and 50 cm (20 inches) outside diameter. The cable alone would weigh at least 8Kg (18 lbs), the loaded spool probably at least 20 lbs. Not something that a wire crawling robot is going to be able to easily carry.

A more realistic spool length is a kilometer or more. At that point, you are talking 200 pounds or more and the cable spool simply MUST be on the ground.

Regards,

John Kasunich

brad

Tue, 2006-11-07 19:43

Well, needs some reworking then. Still has application in the area I was first thinking about -- neighbourhood fiber, where you are only sending the fiber short distances, across streets to connect block-area-networks with gigabit connections (at least until FSO gets cheaper.)

Bob

Tue, 2006-11-21 22:13

I like the idea but there are a few other things to consider:
The fiber either contains a supporting strand (called figure 8 fiber because the cross section looks like an 8) or the fiber is lashed to a metal supporting strand (or incorrectly lashed to an existing cable). In the latter case, the strand is installed first, then the fiber is lashed to the strand. It really is a 3 or 4 step process. Believe me, if there was an easier way (robot etc.) my company would have done it.

Add new comment