It has now been a year since I last built a 3lb bot, so here we go again! Looking back on the history of the bots that I have built, it seems that once I find a new tool to add to my building arsenal I wind up getting inspired enough to use it to break out of my comfort zone and build a new design. I started out with Angry Erector Set and Enraged Erector Set where all I needed was a Dremel to cut some aluminum extrusions and some screws to put them together. Then I got a 3D printer, and it led me to create my 3D printed antweight, Someone Else’s Problem. Shortly after SendCutSend came onto the scene and made it extremely simple and affordable to get laser cut metal parts I created Portable Apocalypse using titanium plates cut by them. I discovered CNCMadness to make carbon fiber parts, and then went on to create Pocket Apocalypse (as well as working on an update to Someone Else’s Problem in the plate and standoff style I used for Portable Apocalypse). Now I have a new tool to play with again, and it has led me to create a new design as well.
Allow me to introduce Apex Predator.
After getting a small Chinese CNC router and getting it dialed in I found that I could use it to make 2.5D parts out of UHMW, opening up new possibilities for robot parts. I feel like I have found a decent amount of success with my horizontal spinner designs for Portable Apocalypse and Someone Else’s Problem (and I’m hoping for even greater success in the future with their newest iterations), but I was never happy with the performance I got out of my one attempt at a vertical spinner, Enraged Erector Set. Part of these performance issues stemmed from the drivability of the design due to the center of mass and point of ground contact being so far from the wheels. After seeing it work well in other bots I figured that switching to a 4WD setup could help to solve that issue.
So that established the 3 main design goals I was looking to accomplish in a new design. Make use of my new router, make a better vertical spinner, work out a 4wd system for better drive.
While it’s usually cited as not being the most critical part of a successful design, I started by picking out the weapon system. I have made several attempts at designing a vertical weapon in CAD over the last few years since I retired Enraged Erector Set, but for one reason or another I was never satisfied enough with any of them to make them worth the effort of building when I was already putting resources into my current bots. I like the look of some of the fancy machined beater bars I have seen like the one on Voxel, but I’m not ready to go for a design where the weapon bar would cost as much to make as the rest of the bot put together. So rather than try to reinvent the wheel yet again, I decided to simply go with the off-the-shelf solution and design around the Fingertech beater bar. It will be one element that I won’t have the satisfaction of having designed or fabricated myself, but it works well for bots like Shreddit Bro, so it should be worth a try.
I knew that using 4 motors for drive would chew into my weight budget, so I wanted to go for belt drive to save weight and potentially make a more durable system. I keep a folder full of pictures of other bots and parts that catch my attention as particularly cool, innovative, or successful to give me inspiration for future designs. Since belt driving wheels was not something that I have done before, I started by digging into the design inspiration pictures to see what had worked for other builders.
My first attempt at implementing this was modeled after the drive train of Lynx, with bearings of some sort mounted in the frame and using two belts to drive two wheels on live axles. This initially appealed to me, as it would allow the drive system to be made entirely from off-the-shelf parts by simply attaching a timing pulley and Fingertech hub to a D shaft. After a round of prototyping, I realized that I was less happy with the implementation. Even small imperfections in alignment of the bearings that the drive shafts spin in could significantly bog down the wheels, and I was less than perfectly confident in pressed-in bearings staying in place after a few good hits.
My second attempt was inspired by the drive train of Boop The Snoot, which uses shoulder bolts as dead axles and integrates the pulleys into the wheel hubs. I remembered an old post from the build log on Neon a few years back that mentioned tapping inserts intended for wood that could be used to put strong metal threads into UHMW, so that would be perfect to thread a shoulder bolt into. I liked that this design could be bolted rigidly to the frame, wasn’t fully reliant on wheel guards to retain the axle, and looked to be more easy to service if one of the wheels got damaged. The one downside was that I would have to design custom hubs to drive it all.
The Fingertech twist hubs have been my favorite wheel solution to date, so I decided to try to re-implement that in a more printable form as the base of the hubs. Adding a timing belt tooth profile and flanges to the other end serves to integrate it with the belt drive, and turning all of the overhangs into slopes should make it reasonably easy to print. Then all I have to do is press a couple of sleeve bearings into the middle, and I have a belt-driven dead shaft hub all ready to go. Finally, I had initially planned on simply using Fingertech foam wheels for simplicity, but since I’m building this with the new router I realized that I could cut wheels out of EVA foam and give them a tread pattern for extra traction, so I think that I will give that a try.
Once the weapon system and drive system were sorted, the rest of the frame largely fell into place around it. The Fingertech DIY beetle kit includes 3/8″ thick UHMW, so that is where I started for the material for the walls, and then 2mm carbon fiber for the top and bottom plates. (I could cut the carbon fiber panels on my little CNC router myself in theory, but between the great pricing and the many warnings I have seen about the hazards of carbon fiber dust I see little reason not to just let the experts at CNCMadness handle it.) The mounting block for the weapon motor is unfortunately taller than what I wanted for the frame for the rest of the electronics compartment, so I had to add a sloped section at the front to make space for it. The combination of the beater bar weapon, 4WD, UHMW frame, and sloped front section made it look strikingly similar to Lynx, even if I had not intended to arrive at such a similar look. Yay for convergent design!
While the shoulder bolt dead axles are not fully bound to the wheel guard rails, the hole in the outer rails is sized for a loose slip fit with the bolt heads which I’m hoping will enable them to be quasi-supported by the outer rails during impacts. I’m placing the battery under the front sloped plate, which should require less work and time to access for charging between matches than having to take the whole top plate off, a detail I have learned the hard way to keep in mind. The mounts for the wedgelets I plan to print from either nylon or TPU for durability, and then press in two threaded standoffs to mount them to the frame. This should make them easy to produce, and easier to replace after damage than if I was mounting the wedgelets directly into the frame rails. For the AR steel wedgelets themselves I did take an intentional page out of Lynx’s book by making them different lengths. My thought is that the shorter wedgelets could serve to give me a second chance to get under an opponent’s wedge if I get unlucky and the longer wedges don’t get under. I haven’t fully finished designing them yet, but I’m thinking that I will also have a pair of sloped mini wedges printed from TPU for when I face horizontals.
As I’m still learning the ins and outs of CAM and routing my own parts, I didn’t want to go too fancy with the designs and multiple setups or fixturing, but I did add a couple of useful features. The vertical rails have slots cut into the top and bottom that match the top and bottom plates, keying the parts together and helping to handle shear forces. I also included small indents in the rails corresponding to the location of the screw holes, which will make it easy to come back with a hand drill and drill the holes for the mounting screws. Finally, with the exception of a couple countersunk holes the parts for the left side and right side are identical, which means that there are less spares that I would need to make and bring with me to be prepared for possible damage.
In case the 3/8″ UHMW doesn’t wind up being tough enough, I already made an attempt at a version of the frame design with 1/2″ thick UHMW with lots of pockets to maintain the same weight. This approach loses the advantage of interchangeable left and right parts though, so I’m holding off on going this route until it becomes necessary.
So that is my design for Apex Predator. I have started ordering parts to build it, so hopefully once the pandemic clears up we’ll see it in action later this year!