AntiThesis: Squishing is your best friend when dealing with a bot

TBExtent · 7 August 2024 17:35

The first BBB Subterranean Showdown was two years ago today, marking the second anniversary of my pursuit of the art of the crush! I figure I should use the occasion to give myself a kick up the backside and write about it.

I decided to build a crusher because it seemed (to past me) to strike the best balance between interesting and feasible out of all the ideas that I had. I can’t remember what the other ideas were, but they can’t’ve been good.

I built AntiThesis version 1 with Matt Crees, so it was a bit of a fusion of our very different building styles. The crusher was my CAD, overanalyse, reCAD, prototype, rereCAD, get machined; while the rest of the bot was his preferred hand cut HDPE.

This is the only nice photo I have of AntiThesis V1, and I don’t have it to hand to take more. This post will be low in pictures, sorry

This ran the standard BBB brushed drive kit, but with double thickness foam wheels. It used two 300mAh 2s ant batteries in series, which gave it just about enough battery life to make it through a fight. We had to have two chargers running constantly to get it charged up between fights though.

The crusher mechanism was ~~shamelessly copied from~~ inspired by Unconscious 514. It had a pair of 3mm hardox frames with forks built in, and a 3mm hardox claw driven by a custom made linear actuator.

The linac, minus the hole drilled through the leadscrew. The screw with the hole is currently in V3

The linac used a lead screw with a steel gear grub screwed onto the nut. This was driven by a 30:1 37mm gearmotor mounted to a hardox panel. A pair of aluminium side panels gave the linac gearbox structure, and a place to attach to the hardox frames. The claw bolted into a hole drilled through one end of the lead screw.

Version 1’s performance wasn’t amazing. It couldn’t push anything, couldn’t self right, and a very high centre of mass made it prone to popping wheelies. It managed to do alright at Showdown though, simply because most of our opponents had bots that were just as unreliable.

The main success from V1 was the crusher. The first sign that we were on to something was when it bent the 6mm steel screwdriver that we used for the locking bar. It punctured a couple of top plates during the comp, but hit empty space rather than any internals. The only issue that came up was a tendency for the 3mm claw to flex during crushes. (It would definitely have bent horribly if it was hit too)

Join me next time for AntiThesis V2, where I replace the problems V1 had with a whole new set of problems.

Phoenix · 7 August 2024 19:27

I always love to see crushers, in my opinion they are some of the most interesting robots due to their ingrained pros and cons.

I can’t wait for updates on this very nice looking robot

TBExtent · 4 November 2024 17:47

Thanks Phoenix! Almost three months later and its update time
Edit: ooooh! look! It even adds a little label saying ‘3 months later’

I had three design goals for V2:

Drive better
Self right
Look cooler

Photo by Stuart Camp

Looking cooler was the easiest to achieve. 80% of making a robot look cool is hiding that you’ve built it around a box. Put odd angles anywhere convenient, add a couple drive pods that stick out, and jobs a good one. A more elegant Razer-y claw in 6mm hardox topped the redesign off

I did consider copying the self righting mechanism from Razer too. I decided not to because of how slow it would be. Running it off of the crusher linac would’ve meant self righting would take 10+ seconds. That might’ve been ok 20 years ago, but modern spinners will rip you a new one in half that time.

Instead I went for a dedicated self righter driven by a servo. I was uncomfortably tight on weight and space, so a standard size servo wouldn’t fit. I couldn’t risk cutting down on speed and torque though. The solution was a ludicrously expensive Savox mini servo, with a four bar mechanism to cushion it in case the self righting arms were hit.

V4’s self righter (ooooohhhh spoilers!)
Its close enough, you get the idea
V2s now a sad shell thats been gutted for parts. Them servos ain’t cheap

It self rights!

This worked perfectly, but set a bad precedent for my future self. Now whenever I’m looking at some new part that will save me a tiny amount of weight, there’s a little voice at the back of my head saying “Why not? You bought a £75 servo to save 20g”

The drive was the most complex part of the redesign. It had to be four wheel drive to deal with the weird centre of mass, and I wanted to switch to brushless for more beans.

Again I ran into space constraints. The crusher mech takes up the middle third of AntiThesis, and has a large motor swinging through the middle of it. This means that all the rest of the electronics has to pack down into the sides of the bot

For the drive, this meant that I couldn’t run either set of wheels directly off of the motors. I wanted the front wheels to be internal, and the rear wheels had the battery in the way on one side and a stack of ESCs on the other

V4 again (moar spoilers!)
This part of the internal layout is pretty much unchanged since V2
Again, you get the idea. The battery be in the way unless I make it loooooong

This meant belts, and making the centre of mass even worse. I raised the drive motors up, so they sat above the other electronics. A couple of timing belts and 3D printed pulleys drove the wheels.

Drive unit! Motor, wheels, belts. Everything you need to make a bot go vooom
This one is a horrible mix of bits from V2 and V3
you
get
the
idea

I designed the drive so that it all attached to the side panels, to make repairs easier. Rather than having to disassemble the whole bot in a tangled mess of belts, I could remove the side and drive train together

At this point I was happy. AntiThesis had a new drive setup that should be able to properly push things. Instead, it turned the next year into pain

The latex I was using to add grip to the foam front wheels stripped off repeatedly
Latex on the rear wheels gave them too much grip, so the bot popped wheelies again
PLA-ST drive pulleys melted, so I had to swap to PETG (I hadn’t set my printer up for any other materials at that point)
I swapped the foam front wheels to polyurethane ones, for better grip (woo!)
The GT2 belts couldn’t transfer enough power to move the bot (aww)
Swapping to HTD3 belts made them saw through the pulleys on the TPU hubs for the front wheels

I finally fixed my drive issues with separate PETG pulleys that keyed into the TPU hubs, with a side order of chonk magnets for extra downforce. It was ready for the first MITE. It was retired afterwards

Half the back is gorn
Photo by David Harrison

MITE actually went pretty well. The drive was good enough for me to hold my own (sometimes) against some pretty tough bots. I didn’t take any expensive damage. And all the issues that cropped up were ones I was expecting and/or easy to fix. I already knew V2 was going to get retired after the event. I had too many changes I to make

These were:

Make the rear armour a single piece
More solid mounts for the self righter
Make the drive less twitchy (Half fixed on the TX mid event)
Loctite the magnets down so they don’t back out and clamp me to the floor
Swap the wood screws for bolts
Bolt the gear directly to the lead nut in the linac so it can’t slip
Add extra tryhard forks
Interchangable setups (more tryharding!)
Give the crusher more beans

Next episope: V3 (briefly) and V4, to bring us up to approximately the present day