I don’t know if this will be much help to anyone, but I figured that it couldn’t hurt to share.
After many cups of tea, I have finished my personal attempt to CAD a full(ish) robot in 24 hours ~3 days. Apologies in advance if this post doesn’t read too smoothly, I’m a bit sleepy as I write this.
As a bit of context, I graduated last year and am now am helping to mentor for our team. I wanted to get in on the kickoff fun though and decided to see how much of a robot I could CAD in 24 hours. Welp, three days later I decided to call this good enough.
I would consider this model to be more inspiration than anything to take too seriously since I designed it more as CAD practice than a genuine robot (it certainly shows). I also designed all of it without even ever touching the game pieces. I’ll break down what I learned from designing it here though in case it can be useful to anyone, but make sure to take it with a shovelful of salt.
CAD is here
Frame & Drivetrain
Nothing much going on here. I opted to use a simple MK4i drivetrain (thanks Logan for putting the awesome looking logos on the covers). I used 1/16 rectangular tube stock because there shouldn’t be too many violent collisions in this year’s game.
The electronics are designed to be mounted from the underside. The space is pretty limited given that there’s a hole going through the center of it for the deep climb, but with a bit of perseverance, it would probably work. The battery would also be mounted from the other side with some type of quick release panel, I just didn’t have time to model it.
Elevator
The elevator is designed as a 2 stage continuous elevator to both lift the manipulator and act as the climb. It uses 2 krakens to lift and retract, which my calculations said should be enough to lift it (thanks JVN calculator and desmos), but instinctually I feel like this is too little reduction. Theoretically, it should be able to go from bottom to top in about a second, and also be able to climb, although the structure (mainly pulleys) would have to be made a little stronger, because I originally intended for the climber to go on the first stage, but later realized that that would not go high enough. Some sort of servo actuated latch/brake would also be needed to keep the robot from dropping when the match ends. Overall, the pulley paths also need to be modifies as I just kinda stuck them in as placeholders without giving them proper mounting or thinking too much about where they would need to be.
Manipulator Joint
Our team has a bit of a phobia of big rotation joints due to our Charged Up bot, so I opted to overbuild the structural side of things going into this. Additionally, I designed this before thinking too much about what manipulator I would use, and I originally thought that the manipulator would have to be mounted on a 12-16in extension which would require much more torque. Because of this it is seriously overbuilt for the job.
It is designed to be powered by one kraken x60 with a 100:1 sport gearbox and a further 15:46 pulley reduction. Knowing now how light the manipulator is, assuming no losses (this is a wild oversimplification, but let’s go for it for the sake of showing my blunder) I would only need 7Nm to hold the manipulator straight out with a Coral (noticeably this is less than the advertised stall torque of a Kraken x60). Changing this gearbox and pulley to have less reduction would give you more room and save weight.
The joint uses two 3x2.5x0.25in bearings that can be bought from Thrifty bot. These two bearings with the 6in between them should give more than enough rigidity (although admittedly I did zero calculations, so this is just based on vibes). The rotating portion is a section of aluminum extrusion with a round flange welded on each end, which is hard to manufacture for most teams although I couldn’t think of a simpler solution.
Manipulator
I unwisely decided to design the manipulator after designing the elevator and joint (a horrible idea) and in the time it took for me to do so, the design I had in mind changed drastically from Coral only end affector on a ~16 in arm to a top and bottom roller setup as close to the joint as possible. This meant that the geometry of the elevator and joint is no longer optimal; however, it should still be able to reach all of the needed positions, it just has to swing out of it’s own way in between them.
It is designed so that the Algae get grabbed by the rollers in the main section. This was inspired by one of the really early Cranberry Alarm prototypes which seemed promising. If I were to continue this design farther I would certainly add some arms that support the Algae on the left and right to keep it centerd.
I somehow only just realized as I was writing this that the manipulator wouldn’t actually be able to grab the Coral because it is ~5in wide which notably doesn’t fit inside a 4" tube (whoops, you don’t really need to interact with the main game piece do you???). For the sake of this write-up, assume that the top arms of the manipulator where 2.5in wide instead of 4.5in (This would be pretty doable, you’d just have to bring the plates closer and use two thin wheels either side of the pulley instead of 1. The Coral are supposed to slide around the upper of the top the two top rollers and and can only be picked up from the Coral Station, which need some pretty precise driving, so I don’t think this is reasonable for a competitive robot.
Structurally, the manipulator would need aluminum reinforcement in some places along the sides for sure as well as something to stop it from wobbling left to right.
Electrically, it uses a neo 550 and a kraken x40 to drive the two rollers as I couldn’t easily use only one given the joint that I had already designed. Additionally, two Pololu Lidar sensors would be used to detect when the Coral and Algae were picked up (these little sensors are fairly inexpensive and really handy).
Climb
Early on when I was thinking about design a Deep Cage climber I thought that one would have to pull it through the middle of your robot, somewhere where there usually tend to be some expensive electronics. And then I thought, yeah, but it’d be funny. So here we are.
I think this design could do both a Deep and Shallow climb. The “hook” mounts onto the second stage of the elevator and has a fairly flat contact patch against the bottom plate of the Cage to hopefully keep it somewhat stable.
The idea behind the process would be to drive slowly into the cage until it pops up over the bumper and falls into the hole in the middle. Then, the elevator lowers and pushes it all the way through the robot.
Initially I thought the CG would be way too far forward, however my model claims that it’s actually pretty centered over the Cage which I’m hoping means you wouldn’t need anything else to latch onto the cage.
As I learned while I was getting the screenshots for further down, The climber as it is also causes a collision with shallow climbs, a quick redesign would fix that though (more on that later).
Above: Contact Patch (Highlighted) of climber & CG of Robot in climb position seen from underneath the belly pan through the cutout
In Action
Apologies for the lack of images of it actually interacting with the game elements. I ran out of time while I was putting this together.
Algae
Picking up from ground (it’s on you to imagine the bumpers there)
Stowed: It’s designed for ~1in of compression, which makes it look like it’s intersecting all sorts of stuff in the model. A redesign of the joint with a smaller sprocket would fix that collision. I haven’t checked the geometry, but my hopes were that it could eject it into the processor from this position.
The hopes are that it could extract algae from the reef in this position although I never actually checked that well. I imagine it could certainly get them out, they just may or may not end up on the ground instead of in the manipulator.
Coral
The ideas was to have it load from the Coral Station in a position like this. My early sketches made this seem feasible although certainly not optimally.
Scoring L2 and L3 would be done something like this
L4 would look like this. As a bonus, if you modified the joint to be above the top bar of the elevator, you could maybe spin the manipulator back passed 90 degrees to shoot into the net.
Climb
I got my “hook” shape a bit wrong, but this is about what a shallow climb would look like
You can barely tell, but it is ~1.57 in off the ground like this. With small design changes, I could probably get another 0.75 inches of travel. This seems pretty unreasonable for actual matches because your CG would probably not be exactly right and you would probably swing into the ground. If you needed more height, you could make it so that the elevator home position is a couple inches off the ground and then the bottom of the “hook” would be below the wheels when the elevator was actually all the way down. I believe the elevator as designed has enough travel for that.
It looks like the upper elevator axle intersects the cage when climbing the Shallow cage (you can really tell how much testing I’ve done thus far). A quick fix for this would be to have a second hook above the Deep Cage “hook”. I wouldn’t consider this a significant design flaw.
General CAD Notes
In the model almost all of the joints are just 3/16in holes that I figured you’d use some sort of fastener to put together. For all of the gussets in 1/16 tube stock, I would probably replace the few 3/16 holes with a bunch of 3/32 holes for pop rivets to distribute the load a bit more.
Also, as I realized while I was working on this changing the CAD visual style to “Wireframe with Visible Edges Only” (Ctr + 9) in Fusion 360 makes everything run a bit faster, since it isn’t dealing with lighting at all. I feel like I probably should have figured this out earlier.
Weight
Altogether, the model clocks in at 92 lubs (no bumpers, no battery). I think that this is a pretty workable weight, it would just be pretty close to the 115 lb limit if it were actually built. The model excludes all wires and electronics so that would add back most of the remaining weight budget. It is also over simplified in a number of ways, but the frame/elevator could also be pocketed/modified to save enough weight to cancel out any missing weight from mechanical components I think.
Conclusion
This year it seems like it should be fairly possible to build a bot that can complete all of the components of the game (even if this isn’t it). That being said I would like to add a reminder not to bite off more than you can chew, just pick what you think is most important and make sure you can do that well (it seems like this bot can almost do a lot of things, but that’s not very helpful in a real bot.
Cheers, and good luck this season!