2019 Telescoping Arm Robot, Revisited

TL;DR: new telescoping arm robot. download CAD here.
Hi, all, and happy holidays. I bring you tidings of a new telescoping arm robot.

Some of you may remember that back in the summer of 2019, I threw together my very first telescoping arm design and posted it here. In the meantime, I’ve played around with new designs, and I figured that it was time to design a complete robot to show off what’s new. I’ve also used this as a chance to get familiar with the new REV control system, which is the main reason it took me so long to upload this in the first place (I was waiting the whole time for REV to drop their CAD models…).


I heard the cool kids these days were all doing single speed, single reduction gearboxes with their shiny new Falcon motors, so I joined them. It’s very convenient as a design structure and I’ll probably do it again on future designs. I also tried out the “corner horn” approach to supporting the bumpers, which at the cost of increased manufacturing effort (welding!) gives you an extra ~1.5" of wheelbase.


One of the key things that I really didn’t like about my first design was that the manipulator was ridiculously large and heavy, and for all of that weight, I didn’t even get much width to pick up balls. To solve this, the top roller has moved from the manipulator to its own, separate system, which allows it to be wider without having to figure out how to control a 30 pound manipulator. I also got aroud to adding some fancy sensors - a reflection sensor for balls and a limit switch for hatch panels. Otherwise, it’s a pretty standard 1684-esque manipulator - the arms close together to grab hatch panels from the center.


The arm is a total redesign from all of the previous telescoping arms I’ve done, with an emphasis on ease of manufacturing and assembly. This time, there’s only one unique billet part (as opposed to the 12 or so on my first telescoping arm), the arm hub. I’ve included detail views on the bearings and arm hub to show off some clever design tricks there. The bearing replaces the old billet base block with a 3D printed setup, but uses the same extended inner ring bearings as before. The pulley uses a much beefier body and mounting setup than my last few designs, and pivots around a thicker shoulder screw. On the hub, the sprocket is mounted using a set of shoulder screws to take up the side loads from pivoting, while we use the same clamping-block style as before.


I didn’t like that my first telescoping arm robot didn’t have a climber, so I’m here to correct that issue. The intake arm doubles as a climbing arm, and in the back, we have a set of legs which extend down. There’s not much else to this design (it wasn’t very original anyway) but it was a neat addition and it fit well with the rest of the robot.

I’m glad I managed to get this uploaded before Kickoff - that way, if any of you all feel like making a telescoping arm this January, you can feel free to use this design as much as you like. Thanks for reading this all the way through. If you didn’t read this all the way through, what are you doing? Go back and look at my pretty pictures. Feel free to check out the CAD here - same link as on the top of the post.

As always, questions and comments are much appreciated!


This robot has really smart usage of MF Onyx prints to cut down on the amount of billet aluminum sprinkled across every subsystem. I’m also a fan of your use of oil embedded bushing for various pivots!


It looks really great, but looking at the renders, I have a few questions. I’ll try to sort them from befuddled to asking for confirmation to inconsequential:

  • I don’t understand the extension at all. I’m guessing that the gearbox shown at top right of “arm stowed back” powers it, but I don’t see how it controls the first stage, much less the second. At first I thought the cylinder in the foreground of that gearbox was a pair of motors facing each other, but now it looks more like a drum, and the motor might be just above that cylinder.
    • Is/are there some cable/cord/chain/belt/rack I’m missing? If so, please show it/them.
    • Where is/are the motor[s] or other actuators for extension?
    • Does the extension work equally in each stage, or does one stage extend first, and if so which one extends first and why?
  • The arm extension bearings are a similar source of mystery.
    • Why are there [external] bearings only on one face for each stage, and the bearings for the stages opposite each other?
    • As the arm swings both ways, the rack forces should apply each direction at various times - how do these one sided bearings deal with this?
    • Are there similar bearings inside the arm for each stage? If so, are they on the same or opposite side, and mounted on the inner or outer stage?
  • Would you please show some detail on the “wrist”? That is, the hinge between the end of the arm and the roller claw. It seems to have at least 150 degrees of motion, but I can’t make out anything on it.
  • I count 2 degrees of freedom for the drivetrain, 1 for the climber legs, 2 for the intake, 3 for the pink arm (shoulder, extend, wrist), and 2 for the roller claw, making 10. Did I miss any?
  • The intake appears to have six vectored (mecanum) wheels and one omni.
    • Is that correct?
    • Most vectored intake bars have solid or compliant wheels in the middle. Why omni?
  • Am I understanding correctly that the energy chain travels inside the arm, and the yellow bracket is basically a casette for the excess when the arm is down?
  • Why do all of the render titles end in “back”? They appear to be from a variety of angles.
1 Like

Thanks for the questions!

There are in fact 2 cables in extension - the first stage extension rope is visible as a gray cylinder on the very top extending out from the drum, while the second stage rope can just barely be seen on the bottom of the middle stage.

Extension is powered by a single Falcon motor, which you can barely see behind the spool (which you correctly identified).

The arm is cascade-rigged, so for every inch the middle stage moves the end stage will move 2 inches. I’ve included a detail view below, which may elucidate things:

Realistically, what I call a “rope” in this picture is really two separate ropes on the same route, one for pulling up and one for pulling down, but I never actually got around to making them not a complete loop. The S1 rope is tied to the spool and to the S2 moving bearing, which has the effect of pulling the middle stage up as the spool is wound. Meanwhile, the cascade rope is tied to the S1 fixed bearing and the S3 moving bearing, which causes it to pull the 3rd stage up along with the 2nd stage at double the velocity.

I think what you call external bearings are actually the rope redirect pulleys. The bearings are actually sandwiched between the aluminum plate and the nylon block. For each fixed bearing block, there’s also a matching bearing block with outward-facing bearings on the stage above it. I’ve included an exploded view here, which shows how the outward-facing bearings work:

Accordingly, the racking issues mostly come into play when near the maximum extension. However, given the skinny shape of the arm and the high overlap at max extension, I wasn’t very concerned about racking. The only note I would make for someone trying to fabricate this in person is that extrusions are rarely exactly dead-nuts on with their sizing, so the bearings will have to be resized to fit whatever tolerance error there is on your stock.

Of course! The wrist is a NEO 550 driving a 49:1 versaplanetary, with an 18:64 reduction after that.

That’s correct, unless you count the pneumatic cylinder on the manipulator as an extra degree of freedom.

Yes. As far as why-omni, it’s because 100 actually used the same setup (albeit with 4" wheels) in 2019 to great success, and I saw no reason to change it.


The “back” marker is for me - when I make renders, the render that gets saved is a transparent PNG. I keep that transparent image around in case I need to touch it up, but I upload the images with a solid-colored background so that they show up well on all kinds of backgrounds. The versions with backgrounds get the word “back” put on them.



I counted the rollers as one (they seem to be powered by one NEO or NEO 550 motor) and the claw open/close powered by the cylinder as number two. Is there another?

Nope, you’re right. I miscounted.

1 Like

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.

How much to hire you for the season?


A few people asked me for native CAD, so I’ve updated the original GrabCAD post with the native SolidWorks files.

Additionally, here’s the link to my original design calculations for gear ratios.

The moderators have re-opened this thread, so feel free to ask any questions when you have them!


More than you can afford :wink: .

However, since I suspect a few people will try to build this arm or a derivative of it, I’ll include a few notes as to possible gotchas:

  • Assembling this arm will be difficult, especially the rigging part. I intentionally made it so that the mounting holes for the bearings at the bottom of each stage can be accessed through the pocketing holes of the stage surrounding them. The net effect should be that you can assemble the whole thing by shoving the bearing assembly inside of the tube, tying the ropes and rigging on, and then inserting the inside tube and attaching it to the bearing. If this doesn’t make sense, let me know and I will add some pictures :slight_smile:
  • You will need a solution for tensioning the ropes. I would recommend using a turnbuckle on the first stage (and if you can fit it, also the second stage). I would absolutely put a tension spring in both stages to pick up changes - you want it to be ~0.25-0.3" in diameter and as long as you can get it.
  • I’m not 100% happy with the design of the pulleys on the cascade stages. I suggest you experiment a little and figure out if you want to swap them out for some other approach.
1 Like

I think that this design could work well this season it would have to be tweaked a little bit but I think this design could be used well at competitions .

I’ve been thinking about this design recently.

I believe it would be relatively simple to adapt this to use an HTD timing belt instead of a rope. This would likely address my main concerns of the rope sliding off the pulley. Additionally, there’s a decent amount of width available for the belt, so you could use something as wide as a 25mm belt for this.

If people are interested, I can throw together a mockup which makes this more clear.


The prototype system I made in college used HTD belt but I’ve since lost my model of it. It would be great if you could mock it up as a demo

I don’t have a lot of time on my hands these days, but I was able to take a minute to sketch up how the rope could be substituted for a passive 16T GT2 3mm pulley. The belt here is nearly flush with the inside tube, so a single clamp plate on the inner stage would be enough to retain it. On the outside, you can add a tensioner pulley (using whatever your favorite tensioner stackup is).


This looks really sweet. I was wondering if you had any concepts or thoughts on how you would drive the second moving stage with this kind of thing? I appreciate any input!

The belts for the second moving stage would have to be clamped in the same positions as with the rope: namely, at the top of the fixed stage and at the bottom of the second moving stage. There are many solutions out there for clamping a belt, but the easiest approach is to 3D print the negative of the belt tooth pattern and screw it into a backing plate (in fact, you can find exactly that in the climber assembly of the same robot).

This results in four tie-down points: two in the second moving stage and two in the fixed stage (one each for the pull-up and pull-down belts). To tension it, you can use a screw to move two of the tie-down points closer to each other. In my opinion, the easy spot to put a tensioner would be to mount it to the fixed bearing assembly on the fixed stage.

To tension the first moving stage’s belt, I would simply use an idle roller, as you would need one anyway to interface with the driver pulley.

in the rope version, is the rope being clamped somewhere? I opened up the CAD and it doesn’t appear to be clamped anywhere.

There are a few tie-down points for the rope to be knotted in place. I was too lazy to model the knots. Anywhere you see three evenly spaced holes that a rope runs through, that’s where a knot is supposed to be.


Is there a native CAD of the version 3 arm? I am trying to edit it for my team and am struggling to edit it as a .step file. Other than that, it’s an awesome design and better than anything I would be able to make by myself. Thanks for sharing this publicly with all teams!