pic: SERGE: Shirt-Ejecting Robot for General Entertainment



Jan Pazhayampallil and I (both 2601 alums) have been working for some time now on SERGE, an all-electric swerve-drive t-shirt cannon.

FEATURES
Articulating shooter actuated by belt-driven linear actuator
Revolving cylinder indexed by servo-actuated ratchet and pawl
Servo-actuated rack and pinion to push shirts into counter-rotating wheels
Heavy use of lasercut birch
Custom swerve modules at ~14.9 ft/s max speed

Please offer comments and criticism.

This is super cool, I love it! The fasteners in the CAD really give it a finished look. Have you thought about getting a grant to make this or something similar?

Just [STRIKE]a few things[/STRIKE] many things I have notes about:
[STRIKE]1. Are wheeled shooters good for this sort of thing? I don’t have any experience with shooting t-shirts but I vaguely remember something about marring or unravelling them. Love the way you can adjust the shooter wheels though, although you might want to make them 3/8" thick or heavy, at least.[/STRIKE]
2. What is the linear slide on the bottom for? It appears to have low ground clearance.
EDIT: Is that for articulate the turret up and down?
3. Why rack and pinion instead of just a pneumatic cylinder? More speed? In any case I like your carriage a lot, but I can’t see the side bearing supports- is there a plastic bearing surface on the sides there?
4. Your corners are held together with just a single plate on the swerve modules. Can you add a 2x1 or something there as support?
5. Love the dual battery mounts, very well packaged. You make good use of the way the swerve drive elevates the chassis.
6. Your swerve drive might have a couple extra gears, but apart from that it looks very cheap and easy to make. Why not use timing belts for the first stage instead of a long chain of gears?
7. Love the two sizes of motors and wheels for the shooter.
8. The placement of the of Victors is neat!
9. Are your sensors in CAD yet?

EDIT: apparently you’ve already answered some of my questions in your other photos… whoops. After looking at the swerve module however, are you sure it’s safe to use a window motor to rotate the module at >160rpm without stalling or overheating?

Cool! I love the idea of a swerve drive demonstration bot. You will be able to put on a real show, dancing across the turf before launching shirts.

Those look like Colson 4x2" wheels on the swerve modules? How well do Colson wheels perform on tile/concrete/asphalt/grass/gym floor?

Swerve on asphalt or grass is a little iffy to me. On a gym floor however these will preform fine and colsons are probably the best option for durability.

Yeah, I’d say the wheels the OP chose are by far the best for the situation, but it’s something to consider. Can the swerve capabilities be disabled for driving on certain terrains?

That would effectively turn it into a standard 4 wheel tank drive, which might actually be worse for turning due to the scrub it would induce than just staying with swerve enabled.

There are other possibilities - you could do an automotive or fire truck style steering setup, for example.

Depends a lot on wheelbase considerations… Looks like it’s a square, or nearly so, so turning shouldn’t be as bad as if it was a long-base. And there’s always the possibility of a “spin-in-place” mode that would allow for quick turns.

You called?

Nice looking shirt bot!

Regarding the corners…

I see you’re using one big beefy 1/4" gusset (which is the swerve plate as well). I’m a little leary and would consider running another crossbracing underneath in the corners, to take torsion in the siderails better… Unless you already have that? Hard to tell.

We received a grant from Columbia University to fund this build. The grant program has its flaws, but I think it’s a great step towards moving college education closer towards practical, applicable skills (how could one graduate as an EE without knowing how to solder?).

  1. What is the linear slide on the bottom for? It appears to have low ground clearance.
    EDIT: Is that for articulate the turret up and down?

Right on. The window motor drives an XL belt using a 3D-printed pulley. The linear slide runs on linear bearings and articulates the shooter:
https://camo.githubusercontent.com/5bec8c2dad6a09536dd5c1ca48484e234290f476/68747470733a2f2f692e6779617a6f2e636f6d2f62616433343364643436656234346430363262333339373536313330393863352e676966

  1. Why rack and pinion instead of just a pneumatic cylinder? More speed? In any case I like your carriage a lot, but I can’t see the side bearing supports- is there a plastic bearing surface on the sides there?

We wanted to run everything via electricity - if we had a pneumatic system onboard, we would’ve gone down the more traditional (and likely longer-range) compressed air cannon road. An HS645MG servo drives the entire mechanism. The rack rides on 3D-printed bushings. The dark plastic surface you see is a 1/4" acrylic plate - the material was chosen just for appearance.

  1. Your corners are held together with just a single plate on the swerve modules. Can you add a 2x1 or something there as support?

Absolutely - we’ll be adding additional gussets to reinforce those rails. The rear corner of the robot (as a triangular chassis) is supported by a section of liberally riveted 2x1, so we’re not concerned about the rigidity of that end.

  1. Love the dual battery mounts, very well packaged. You make good use of the way the swerve drive elevates the chassis.

Thanks! As someone who usually spends more time in an IDE than SolidWorks, that means a lot to me.

  1. Your swerve drive might have a couple extra gears, but apart from that it looks very cheap and easy to make. Why not use timing belts for the first stage instead of a long chain of gears?

We were looking to source as many parts as possible from extras, bits, and pieces people were willing to donate. Those gears were far more common than timing belts and pulleys - so this boils down to just a monetary cost decision.

  1. Love the two sizes of motors and wheels for the shooter.

Thank you. We may choose to remove the first ‘stage’ (1:3 BAG motors) to conserve weight, depending on whether the linear actuator can handle it.

  1. The placement of the of Victors is neat!

Thank you. Here’s an image of the shooter from about a month ago, just to prove we’re serious about realizing this concept.

  1. Are your sensors in CAD yet?

We’ve got a rotary encoder and limit switch to provide feedback for the linear actuator - those sensors are in CAD (and real life). The indexing mechanism for the shooter is actuated by a hefty quarter-scale servo, which has builtin feedback. We may choose to add an extra hall effect sensor onto the rotating shirt cylinder, just to make sure those shirts line up properly.

(…) After looking at the swerve module however, are you sure it’s safe to use a window motor to rotate the module at >160rpm without stalling or overheating?

We’ve got one of the swerve modules built, and we’re testing it to see what kind of conditions it can withstand. Hope I can provide a concrete answer soon.

This is very much so the goal! We’re looking to use this robot, once fully built and tested, for promotional purposes on Columbia University’s campus and during competitions to excite crowds and promote interest in the maker community at Columbia.

It’s actually a triangle! Spinning in place is our easiest way of approaching turning right now.

Thank you!

We will definitely need to reinforce how those rails are braced together. The frontmost rail is support entirely by two swerve modules and the 1/4" ABS bellypan, but since we also expect it to take the brunt of impacts from terrain, we’ll be adding additional gussets to strengthen that structure.

Thanks for the answers! Love the picture of the prototype shooter and can’t wait to see this thing finished! I have access to a laser cutter at the college I’m going to, so your work with it is very relevant to stuff that I could make in college.
You may want to add an extra set of mounting holes for the linear actuator (not 100% sure on that) but I’m sure testing will clear up any issues. Worst case you can always stick a DART in anyway.

Great concept and design! any chance to get a closer look at the swerve modules?

In addition to the corner bracing, you’ll likely need to make some of the parts of the swerve module more rigid. Wood is a fantastic building material, but it doesn’t work well with the usual plate/standoff method for building gearboxes (especially ones that involve large forces). Sheets of plywood are bendy, and things flex a lot! If you use the wood in a way that forms boxes, you will end up with more rigid parts.

The Bit Buckets have been playing with plywood a lot over the last few years and that seems to be the big rule of thumb: build boxes. Here’s their rendition of a wooden swerve module, though it hasn’t been implemented yet.

Do you guys have free access to a laser cutter that you can spend time on? If so, just cut some parts, test, redesign, cut, test… Laser cutters and plywood make for fast cheap parts, so they’re great for making lots of design iterations. I love them!

Thank you. The design of the swerve modules has not changed significantly since my last image.

Here’s a clip of the indexing/inserting mechanism, powered by an HS805BB with a ratchet and pawl and an HS645MG on the rack and pinion assembly. The indexing mechanism uses 5 lasercut birch parts, and a 3D-printed ratchet and pawl, to rotate the shirt cylinder 60 degrees at a time.