Products > Other Equipment & Products

OC Robotics snake arm robot teardown.

(1/3) > >>

Thought I'd show you guys something you might not see every day. It's not that much electronics but pretty cool none the less.

This is a very early "snake arm" robot designed and built by OC Robotics in the UK. It was used during a repair project at a nuclear plant in 2004 and has been in storage ever since, recently it was decided to be torn down and I took the opportunity to shoot some photos during the process. I appologize for the somewhat crappy phone photos but that's what I had at hand at the time.

Basically what it is, is a stack of flexible sections, 11 in this case, forming the "snake" - sort of like a human spine.
Three stainless steel cables are then anchored to the "end" of each of these 11 sections and threaded back thru each of the previous sections, thru a system of pulleys and back to a set of actuators. By pulling/releasing on the cables the "curvature" of the snake can be controlled:

The snake is positioned on top of pole which is attached to a vertical slide with which you can manually adjust the height, this was mostly done to get the overall height down for transportation and storage.

To be continued...

Then the other end of all the cables, where they enter the base of the machine...

And distributed...

to one row of actutors:

Lifting the cover exposes all the actuators, each with its dedicated servo drive (more on that later):

To be continued...

Each actuator is hinged to the base around the point where the cable enters the actuator. This allows the actutators to be folded out for access:

Here we can see the cables coming from the actual snake on the left hand side and routed into two rows of actuators. Note how the cable passes thru the shaft for the idler pulley:

These actutors are really quite clever. Just looking at them it took us a couple of minutes to figure out how they worked.
The motor (a MAXON RE40, 150W) drives a 14:1 wormgear, firmly attached to the output shaft of the wormgear are two belt pulleys. The 2100mm long endless AT5 belt is run thru the actuator and attached to the "shuttle" with a pulley arangement providing a further 2:1 reduction ratio. Following the belt path around inside the actuator it looked like the "shuttle" would just never move, it appeared as if the belt would just "circulate" inside the actuator without actually pulling the "shuttle" in either direction.

The clever part, and what allows the "shuttle" to actually move, is that one of the two driving belt pulleys on the wormgear output has one tooth more than the other. This makes one "side" of the belt drive "feed" 5mm more belt than the other side for each revolution which, due the 2:1 pulley ratio, moves the "shuttle" 2.5mm. All in all it takes the motor 28 revolutions to move the "shuttle" 5mm. I've seen a few belt driven linear actuators before but nothing quite like this.

Each actuator has it's own servo drive, also custom designed for the application. The supply for the drive is 48VDC and they are "attached" to the control system via a CAN bus.

To be continued...


Nothing too exciting really. Power in top right corner, fuse, 39uF capacitor across supply. T2/T3 are the two low side switches in the H-bridge, 30mOhm current sense resistors on the low side of each leg. TLC274 for the current sense circuitry. LS7166-S 24bit quadrature counter handles the encoder.

Bottom side, 89C51CC01 with built in CAN controller, MCP2551 CAN tranceiver, 78L05 regulator, TL431 reference, the two top side switches of the bridge and the HIP4081A bridge drive IC.

On the system level there was a big box with a 1.1kW 48VDC switch mode power supply for the motors, a 24VDC auxillary supply, safety relay, contactors etc. A standard Windows XP PC for the userinterface which communicated with an "industrial PC" (a stack of PC104 stuff from Versalogic) via ethernet. The "industrial PC" then handled the CAN-bus and some discrete I/O signals. Unfortunately I don't know anything about how the software was divided between the system, which one handled all the calculations etc.

The robot was controlled using a bog standard gamepad controller. The aproach was (and still is as can be seen in the videos linked to further down) that you steered the "head" of the snake and the body followed wherever you went (within limits of course). The path was recorded so you could easily "retract" and/or play back the path.

As I said, this was some 13-14 years ago now but I'm happy to see that OC Robotics is very much still around (no I don't work for them). Here are a couple of vidoes I found that may give you a better idea of how it works and see some "in action":

The second one is a little bit of a MAXON motor infomercial but look and listen beyond that to see what the robots can do (I remember Andy, the guy in the video, from when we did the project for which the robots were bought):

If you look at 1.15-1.25 in the following video you'll see the very robot I just tore down. The footage is from a mockup but the still images are from the actual repair.

We have a second one exactly like this one and also another shorter but more rigid unit. The principle is exactly the same but the engineering that went into that one is even more impressive due to serious space constraints. If there's any interest I'll shoot some pictures of it too, haven't decided its faith yet.

Anyway, not much electronics but I hope you found it interesting none the less.

That is really neat, thanks for showing this.
I've seen enough hentai to know where this is going.


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version