At my job we were recently given access to Codex from OpenAI and asked to test the capabilities. I decided to see if it could be used to make a DXF nesting tool and it worked well enough that I’m releasing the tool publicly under CC BY-SA. My hope is that by releasing this that it will result in a significant reduction in scrap/waste compared to manual nesting of parts
The .zip file above contains everything you need to run the tool locally and the functions have been tested in both Firefox and Chrome. The .dxf files included in the download are the top and bottom panels for the current version of Antix which I used to test the tool. When using the tool, since everything runs in a browser window it can take a bit of time to process complex shapes or a large number of parts and the browser may ask if you want to stop the process/page. If you let it continue to run it should finish nesting just fine.
There are two nesting options, minimum rectangle and minimum “rope” footprint. You also have the option to limit the parts to 90deg rotations for nesting which can be useful for some materials. You can also use the sheet dimensions to manipulate the nesting outcomes if you want to fit things into a specific area. Also, the sheet size you put in will be used for the preview of the nested dxf, so if you want a closer look you may want to re-run the nesting after getting the general size it can fit in.
A while back I came to the conclusion that the existing arenas that were getting used for insect class events were getting to the point where they weren’t going to be able to handle the 3lb bots that were showing up at events. That meant it was time for upgrades or a new arena, and I was sick of messing around with 80/20.
So, time for a new arena.
The brand new arena set up and ready for its first event at Dragon Con.
I’ll jump right to the good bit, I’m not just going to show you the build, I’m also making the files available for anyone that wants to examine, iterate, and evolve on the design. As I went through the build I made adjustments to the CAD to reflect tweaks I made to the real parts during the build to account for the realities of fabrication tolerances, tolerance stack-up, and general ease of use.
The design requirements had a few oddly specific elements that dictated some design decisions, including-
The arena must fit on the existing trailer in whatever form it travels in and can take up no more than half the available width
The arena must be able to roll through a standard door when packed for travel
The arena must have an option for a pit/push-out
The arena must be able to be set up from its packed configuration in under 30 minutes by 4 or more people – Note, the setscrews for the roof side were snug enough that half of them were pulled and replaced by bolts. This allows the other two (on opposing corners) to work for alignment and quick support during setup while making it easier to install/remove the roof, though a large mallet is still quite helpful
Beyond those requirements there were several other design goals that I included based on the prior arenas I’ve worked on, those were-
Replacement of the polycarbonate should be simple and fast
The polycarbonate should be fully floating in its frame
The walls should support double layers with air gaps for dual 3/8″ polycarbonate should it be needed in the future
Each element of the arena should be easy to maintain with minimal other elements needing disassembly to perform the maintenance
Visibility and footprint should be as large as is practical
The hazard area should have options for pits and a flat floor
Setup should use as few tools as possible
Use tube laser cutting for tube parts for easy replacement and fit-up
The arena door and wall should be the same thing for maximum visibility – Note, the large size of these does mean there’s a bit of flex but there’s not much to be done there without adding a lot of weight or blocking sight lines
With all that, on to the build photos
First up, prepping the uprights and the leg mounts.
Next up, the floor and roof frames.
At this point things were getting bulky and hard to work around, so it was time for the transport cart.
On to the doors.
Now for a big batch of printed TPU polycarbonate supports, paint, and making sure stuff fits on the cart. You can also see the polycarbonate retainers that also join the internal and external door uprights. You can remove two from either side to have open access to the polycarbonate for replacement or cleaning.
The keen eyed may notice that the gussets on the cart change a bit between pictures, when designing the cart I kept clearance for the legs. When building the cart I decided to make the gussets larger, forgetting why I had made them a specific size. I then solved the clearance issue with an angle grinder and some more welding.
Also worth noting that the bushings that are eventually installed in half of the uprights (the hinge side, you’ll want one side to be the hinge and the other to be the long pin that’s shown later for locking it closed) are shaved down to fit snugly since they’re a line to line fit in the CAD. This allows you to effectively get a perfectly fit hinge without needing perfect tolerances.
Time for the floor, roof, and lighting.
Now for the finishing touches, long handled pins to lock the doors with easy to identify pull points and an assembled arena ready for combat.
Following the first event and the lessons you learn by doing there are a few adjustments/upgrades in the works-
Detachable handles for the doors so there’s a clear, easy grab point for opening and closing the door
Adding additional D rings and/or handles to the cart for easier handling during load/unload
The floor and roof may get some additional handles to give easier, more consistent grip points for load/unload and setup
I’m not going to say you should build this exact arena since a lot of design decisions were based on very specific wants or requirements, but there’s a lot of ideas here that can translate to another arena build with minor adaptation. Feel free to reach out if you have any specific questions.
Prior to attending Robot Battles 72 I decided that it was time to start on a new 3lb build. With the new build in progress and the damage done at the event being pretty spectacular now seems like the time to retire High Steaks.
With that, I’ve also decided to release the full CAD in STEP format.
The above .zip file contains the .dxf’s for the flange, side plate, polycarb holddown polycarb panel (if you want to route it out or similar) and the .step file for the polycarb pull handle. The steel for this build was all laser cut by SendCutSend.
While it’s certainly possible to build a bot, show up to an event, and run it without taking the time to test it’s not the best idea. Similarly, if you’ve got a weapon capable of damaging your opponent, you’ve got a weapon capable of doing real, lasting damage to a person. If you want to test it, you need somewhere safe to do it, and a test box is a great way to protect yourself, protect the other people around you, and test your bot.
With that in mind, I wanted to replace my old bulky test box with something that hit a nice balance of usable space and portability.
The core goals of this design are:
Small enough to easily go through a door
Able to be mounted to a wheeled platform
Big enough for almost any 3lb robot to be tested safely
Easy to build
Easy to repair
CAD Model of Test Box
Goal 1: Small enough to go through a door. The main structure of the test box is a 26.5″ square frame that can be built to effectively any height. It’s rare to see a door narrower than 30″ so this should fit with room to spare.
Goal 2: Able to be mounted to a wheeled platform. The hole patterns on the exterior of the flanges will allow the test box to be securely bolted to a frame using a simple hole pattern and #10 hardware.
Goal 3: Big enough for almost any 3lb robot to be tested safely. With 1/2″ plywood walls the internal usable floorspace is just over 23″ square. Only the largest of the large in the 3lb class can’t fit that footprint.
Goal 4: Easy to build. The main frame design uses two main parts, a flange and a side plate at qty. 8 each, made from laser cut mild steel. These components key together to aid in fixturing for welding and provide easy attachment points for wall and floor panels. Additionally, the top polycarbonate panel is retained by bolt on flanges and a pin lock to allow a simple rectangle of polycarb to used without any drilling required. The dimensions also allow for 2′ square 15/32″ thick plywood project panels from any local hardware store to be used for the walls and floor with little to no modification required. Similarly, the polycarbonate retainer height can be easily adjusted via 5/16″ OD spacers sized for #10 bolts. For mine I added adhesive backed felt pads to help with sliding the panel in and out. With sufficiently stiff polycarb panels you likely can slide the panel straight back with no issue. If you notice sagging then a small bonded tab that lifts the edge of the panel as it slides in will make closing the test box easy.
Goal 5: Easy to repair The mild steel frame, easy to swap hardware, and use of commercially available plywood panels means that there’s typically a quick, easy repair option for almost any kind of damage.
Fresh cut parts slotted together
Clamped for welding
Off the shelf walls and floor
Welded and ground
Welded and ground
For my build I opted to paint much of the plywood, while it’s not necessary it does add a nice finishing touch to the whole thing.
Floor fitting
Checking alignment
Walls going in
Mostly assembled
Polycarb retainers and felt installed
So, what’s left to do? At this point the test box is fully usable. Most of the box is held together using some fairly short #10 wood screws and the pin to lock the polycarb panel in place is McMaster #98320A125 if you want to track down the same part.
For a long time, I’ve wanted a ground style transmitter with a third channel that wasn’t a basic switch or slow scroll through an input range. Not long ago RadioMaster released the MT12 which is a ground style radio running EdgeTX.
Before I get too deep into what I did, here’s the finished mod:
Finished mod being bonded to the removable base plate
Finished mod being bonded to the removable base plate
With that out of the way, here’s how this was made:
After the MT12 was delivered and RadioMaster told me there wasn’t a CAD file available I looked into free 3D scanning apps, eventually settling on Polycam. I took the MT12, set it on a flat metal plate, scanned it, then exported the scan, used a converter to get it into STL format, imported it into Solidworks, then made the first version of the main housing.
3D scan of the MT12 done with Polycam
Raw STL imported to Solidworks
Raw STL being used to mock up first draft of the joystick mod
First draft mod being used to validate scan geometry
Overall, it was ok, but a bit bulky. It also meant that I couldn’t grip the transmitter the way I wanted to. Enter the realm of near impossible to machine parts that are effectively trivial thanks to 3D printing. The second version of the housing dramatically changed the shape and added mounting features. It also made it clear that I’d need at least a short cable extension. V3 quickly followed with fine tuned mounting and a cover for the back of the joystick.
V3 geometry to determine fits/location and cable length
V3 geometry with rear cover installed
V3 with a test for a removable mounting strap
I didn’t happen to have the right connectors on hand, so a quick Amazon order later and I was ready to build an extension cable. For this step the big thing is making sure you don’t swap around the order of the wires from cable to cable since that could cause input issues or damage a board.
Extension cable installed
With the mod tested and the housing painted it was time for the final step, bonding the housing to the removable base plate with one of my favorite adhesives, Shoe Goo.
Finished mod being bonded to the removable base plate
Finished mod being bonded to the removable base plate
The paint’s a bit glossy, so I may give it a matte clearcoat at some point, but beyond that I’m very happy with the final outcome.
Have you ever thought of a terrible pun? Has that terrible pun resulted in dozens of hours of work to transform it into a tangible object?
Big wheeled bots aren’t new. Bots like Huge, Gabriel, Starchild, and plenty of others have been out there fighting for years and are inspiring new designs all the time. While thinking about the pros and cons of this style of bot the thought came up: What would happen if Team Food Fight built a big wheel style bot? While I can’t say what the bot would look like, a name immediately came to mind.
With that pun, the CAD effort began and Hotkoin was contacted for the logo you see above.
So, what are the goals of High Steaks?
Re-use the hubmotor and electronics from M-80
Test viability of PLA+ for semi-protected structural elements in a 3lb bot
Test segmented cleats as a means of increasing grip and adding structural stability to UHMW wheels
Lean hard into the theme
With those goals the parts list and profile started taking shape. High Steaks would recycle the drive and weapon esc’s from M-80 along with the hubmotor. Some left over FingerTech Mega Sparks would be the drive solution with an extra 2:1 reduction to the wheels.
Cleated UHMW WheelsWelded 4130 Chassis Being Tempered12lber For Scale
If you’re gonna go all in on a steak theme, you can’t be generic on your weapon disk. This lead to the question of just how steak-like a disk can be made to look. After some digging around I settled on two base colors and the Montana Cans white marble effect paint.
This Was a Very Clean ProcessRaw and Ruined
With the electronics pulled mostly from an existing bot the wiring process was fast with the main challenge being routing wires between the two halves of the chassis. The rear channel was snug, but it wasn’t too difficult to pull the wires through with the assistance of some forceps and a bit of patience.
Electronics Overview
Power Switch with TPU Mounting Spacer
Thanks go out to SendCutSend for making the weapon disks, steel chassis components, and spring steel cleats.
Top View
Bottom View
I Swear It’s Not On Fire
Weapon Lock On Theme
With all the pre-event goals met that leaves the final question: How did it do?
Overall I’m very happy with the performance. The one serious design issue that popped up didn’t really impact the results, but with the width of the tips on the side forks it would require the axle to be removed to replace a wheel, which isn’t ideal if there’s a time crunch. Because of that some new side forks have been designed that keep the aesthetic while allowing the hex bore on the wheels to slip over them for easy replacement. Beyond that, some extra UHMW wheels and cleats have been made to ensure enough spares are on hand and that there’s a heavier duty wheel option for bots where wheel damage is particularly likely.
Outlier is the teams newest 1lb bot and is the end result of trying out a bunch of new stuff and new ideas.
Going into the design the core goals were: 1. Build a 1lb electric lifter that’s competitive 2. All brushless (this goal came in after initial concepts used some brushed parts) 3. Run 4s lipo effectively 4. Test viability of TPU lifter arms 5. Try new wheel construction methods
Early concept with a brushled lifter motor
Layout is starting to take shape
Design as built for Robot Battles 71
CAD from the early concept stages through the design that was made
Over the course of a few design revisions and a few product releases the bot went from a semi-direct brushless drive, brushed lifter with all of the components integrated into a single chassis to an all brushless lifter with replaceable weapon modules and indirect brushless drive.
In addition to the above, the chassis, forks, and plow on outlier were made from 0.050″ laser cut 4130 steel from SendCutSend, the baseplate was 1mm carbon fiber, and there were a lot of PLA+ and TPU parts printed in house.
Weapon module test fit
Weapon module test fit
Lifter gear and wheel hub
Chassis test fit
Plow test fit
Silicone to TPU bonding test
Completed wheels
Assorted photos from the build including test fits and some adhesive testing
Rolling chassis
Weapon module test fit
Test install of the drive wheels & weapon module
With that done and the event rapidly approaching a lot of things happened off camera with the end result being a working 1lb robot.
First drive and arm wigglesA bit of lifter testing
Outlier is a pretty dense little bot and was probably one of the more challenging wiring jobs I’ve done.
Outlier (left) along side Firecracker (middle) and Quark, (right) two of our 150g bots.
Ready to fight
So, how’d the event go?
Can’t really complain.
The good: The drive power was insane, even with the fairly grippy silicone it was easy to spin the tires. The chassis and drive mounting held up great, even with some major hits repairs were easy. The TPU lifter arms took crazy hits without any trouble. The PLA+ – TPU – Silicone wheel combo held together well even when taking some nasty hits
The bad: The D shaft I was using had a very small flat, so between the Delrin gear and printed TPU adapter it was struggling to transfer torque, so the lifter often struggled to lift. Changing configurations took longer than I’d like which meant I ran a config that wasn’t well suited to my opponent and took some massive damage to the weapon module & plow during the fight After a few fights in close succession the printed PLA+ motor guards deformed, jamming the drive motors in the final, luckily late enough that it didn’t cost Outlier the match.
Damaged but still usable wheel
Post event teardown showing deformed PLA+ motor guards
So, what’s next? 1. New homemade titanium D shaft for the lifter module 2. Modified, lighter, 1095 steel plow design that will allow weight for wheel guards so changeover time is reduced between configs 3. Explore additional wheel configurations 4. Connectorized weapon motor to allow easy swapping of entire weapon module
