I was at Denver Maker Faire for about 30 seconds before I started planning the bot I would bring next year. Cpt. Hook's design was proved to be solid structurally, but fundamentally ill suited to the competition. At Denver Makerfaire I networked with several other CO bot builders, and we are going to attempt to collaborate on a modular combat robot platform. The platform will feature a common drive base and a standard mounting arrangement for attaching weapons.We're planning a roughly 50-50 split of drive train to weapon mass. The initial designs are starting to look good:
I am still really excited by the idea of a crushing robot in the 3lb weight class. However, I'm a little burned by Cpt. Hook's performance so for at least my next bot I'm going for something a little more dangerous:
My collaborators have some other interesting ideas that they are working on too. I'm hoping we'll be able to field roughly 4-5 bots by next maker faire, and that we'll have a prototype ready for Rocky Mountain STEAM Fest in April.
Cpt. Hook was a good first attempt at a combat bot. I'm satisfied with how he turned out, but I now want to change 100% of the things I did on him. On Sunday at AVC I entered him into a few rumbles against some beetle weights and his armor held up really well. At the end of the day I had the chance to see him off in splendor, so this is how Cpt. Hook died:
In the end, his mono-body armor was remarkable strong. The aluminum armor wasn't ideal, but the method of construction proved itself. After 3 hits from poison arrow I was able to reconnect the battery, attached a fresh wheel and drive. Overall, I call him a success and next year's bot is already in the works.
Cpt. Hook Doesn't Do Very Well
Okay, so Saturday was the initial rounds of the main tournament. You can view the tournament bracket here. Captain Hook ended up going 1-2 in the tournament, fighting 3 robots without active weapons. I can't complain too much as Cpt Hook's weapon wasn't exactly "effective." The end result was 3 matches without even scratching the paint, and I was out of the tournament. It was a pretty disappointing result, and I'm definitely going to make sure I have a bit more of an active weapon next year. You can see the fights from Saturday here.
The judging process used is a 3 point system, one point each for Aggression, Control, and Damage. Of the fights I watched, I would say only matches in the first 3 rounds of the tournament regularly ended in knock-out. After the first time builders had their bot's dismantled, the competitive bots just didn't die inside the matches. If you want to make it near the end of the competition you need to play the scoring system and win over the judges. Judging seems highly biased towards mobility and driver skill, good wedge robots are able to win on judges decision by just driving around their opponents. I've already talked about how Cpt. Hook was too high-speed and low-torque to be effective, but this was even more true than I expected at the competition. He was also too responsive on turning, and not controllable enough in straight line. I could/should have resolved this by simply moving the wheels further out from the center of the robot.
Armor vs. Structure is a big deal for most robots. Cpt. Hook was the only "mono-body" robot I saw at the event. By mono-body I mean that the armor and the structure of the robot were the same component. Instead almost all bots featured a core chassis with attached armor panels. Chassis were often plastic at the beetle weight, and armor was typically titanium. UHMW (ultra high molecular weight polyethylene) was featured on a huge number of robots, and seems the plastic of choice. I would have expected to see more Delrin or Poly-carbonate, but UHMW was everywhere. I still haven't come to a strong conclusion about which design is fundamentally better; I suspect that the mono-body tends to fail catastrophically whereas conventional designs tend to be salvageable after damage. Certainly Cpt. Hook's aluminum armor appears to be too soft, and that spinning weapons will be able to bite deeply into it. Titanium and hardened steel armor appears to be able to repulse the S7 Tool steel teeth that most robot use on their weapons.
I've been very busy since the start of September trying to juggle finishing Cpt. Hook while mentoring the local FRC team, so I haven't had time to make updates on the project. However, you can now enjoy seeing the finished product. I made a couple of quick youtube videos which can walk you through the final bot.
This first one is a simple demonstration of it driving and moving the weapon:
In this second video I open up the bot and walk through the internals:
Today, I was able to perform the first full test of the crusher on my bot. I did not go well.
Two separate issues were encountered during testing. The first one is my fault, the second one is only probably my fault.
Firstly, I underestimated (more like failed to estimate) how much force there would be trying to separate the final stage gears of my crusher. At some point I changed my design to make the front forks, and the shaft columns one piece. When I did this I moved the columns closer to the center line of bot, leaving the gear cantilevered at the end of the shaft.
The gears are now pulling apart and slipping. The gears have a pressure angle of 20 degrees, and as it turns out there is almost 120 lbf trying to push them apart. Somehow, I will have to reinforce the end of the shaft.
The second problem I encountered is that the gear-motor I purchased is not capable of producing its advertised torque. Instead if you stall the output shaft and continue to run the motor, you will blow teeth inside of the gear box. I've now been able to reproduce this twice. It may be due to the end of the motor shaft being unsupported, but after seeing the tiny size of the gears inside the gearbox I suspect it is simply under engineered.
Luckily, I purchased a pair of these gear-motors and as each gear stage inside the motors is identical, I can swap gears between the gearboxes to get a functioning set. For now I'm simply limiting the current I'll send the motor until this issue can be resolved.
Based on my research, the majority of beetle-weight battlebots seem to have success with "lite-flite" wheels. I believe these were originally intended for model airplanes, but they have been used by a number of successful bots. Additionally, FingerTech Robotics make an excellent snap-hub that fits them and provides a secure connection. The wheels are a light-weight, resilient, and sticky rubber foam which seems to absorb major hits excellently. I decided to use these wheels and snap-hubs on my robot. Unfortunately, I discovered an interesting issue.
When my robot got up to speed and then tried to turn, the side load on the wheel would cause the wheel to deform underneath the edge of my robot, where it would become stuck. This could completely immobilize my robot. Unfortunately I couldn't get a good photo of this phenomenon, but I hope this diagram makes sense:
The fix to this turned out to be relatively easy. I fabricated some wheel stiffeners out of some 1/8" poly-carbonate and placed them inboard of the wheels. They worked great at resolving the pinching problem. I'm waiting to see if they start causing other issues thou. You can see them in this photo of the partially assembled bot:
Today, I finished fabrication on all of the custom components for Cpt. Hook. Originally, I had intended to fabricate the majority of the parts using the CnC mini-mill at the hacker space in town. Unfortunately this machine was down, and I ended up fabricating all the parts by hand using a band saw and drill press. At the end of the day I managed to hit a 1/16" tolerance on every part, and I doubt for this machine there will be any issues. Here are the finished parts in all their glory:
I regret not documenting the fabrication process more. For next year, I am seriously considering a full build video detailing all the construction steps.
First off, I decided to move my CAD work from Autodesk Inventor to OnShape. This will allow me to open source the design, and I plan to use it as a teaching aid for the FRC team I mentor. The design now has a permanent link here, so you can check it out at any time. I also picked a name Cpt. Hook. I honestly don't have a real reason, I just liked the name.
I purchased most of the COTS components for the bot and after weighing them, I realized that I was going to be overweight. The gear-motors are denser than I estimated, and I decided to switch from a 2S LiPo to a 3S which added some weight. Additionally, after trying several times to come up with a double-action weapon design, I decided there was not an easy enough way to pull it off.
I decided to drop the double action design, and I simplified the front end geometry by eliminating the back-slope and wheel guards. Now the design looks like:
Through my membership at the local hackerspace, I learned that Sparkfun puts on an annual Battlebot competition. They have 1lb and 3lb classes, and the 2016 competition attracted 30+ bots. Information about the competition can be found here: https://avc.sparkfun.com/ . I've decided to try to build a 3lb battle bot.
Doing Some Research
I did a fairly quick "literature" review of Battlebot designs, and tried to compile the core features of a good design. My research was by no means comprehensive (watching 20+ hours of youtube footage), but I came to a few conclusions.
Firstly, at the lower weight classes bots get knocked everywhere and every-which-way around the arena. Robot's which have some way they can land that prevents them from driving will inevitably land on that side and be immobilized. Robots that can't drive upside down, but that have some self righting mechanism can spend large amounts of the match doing nothing but flipping them selves right way up. I decided in my design I want to to be able to drive or quickly recover from any orientation.
Secondly, spinning weapons are significantly less effective at lower weight classes. From the footage I reviewed it appears much more difficult to do significant damage to a robot with a spinning weapon at these weight classes. Spinning weapons get in some incredible hits that send both bots simply flying around the arena, but almost never is a bot disabled as a result of those hits. I believe that this is due to the significantly lower inertia and friction of these bots. When a spinner weapon lands a hit, both bots are simply pushed apart and the weapons cannot bite into the metal.
Lastly, asymmetric armor leads to bad things happening. I saw many bots that had extremely strong front and top armor, with much weaker (often plastic) sides and back panels. These bots clearly attempted to keep damage localized to their more armor sides, but the chaotic nature of the combat often made that impossible.
I looked around and I saw a lot of hammer weapons, spinner weapons, and wedge bots, none of which really seemed to be able to do damage to each other. I didn't see good examples of crusher type weapons at the 3 pound weight class. So after some soul searching, I decided I would try to build a crusher robot and see if it was a feasible concept in the 3 lb weight class.
Weight is by far the most restrictive rule for a battle bot design, so I set about trying to optimize my design around the weight limit. I started playing around with different combinations of drive motor, weapon motor, and armor thickness to try to find the best balance of weight between them. My conclusion was that with 1/8" aluminum armor on all sides, and a "standard" combination of drive motor and wheel, I could devote about 1 pound to the weapon. Here is what my first pass at a design looked like:
I do not have easy access to a welder for this project, so I opted for a bolted design. I decided that I wanted an easy way to service any part of the robot, so I came up with this "exo-skeleton" structure. Instead of a central chassis that carries the armor and components, the armor is the chassis and all components bolt to the armor panels. To make assembly easy, I came up with these "super-nuts" that will connect all of the armor together.
The super-nuts are .25" aluminum columns that are tapped for 4-40 bolts, and serve as captive nuts for the armor panels. The super-nuts are mostly identical and I should be able to fabricate a lot of them very quickly.
For the weapon, I am going to attempt to build a double-action crusher. A small gear-motor will drive a drop cam that will cock and release a torsion spring on the main weapon spike. The spring is designed to clamp with ~8 lbs of force, which should be more than enough to hold an enemy bot. Once the bot is held by the spring, the primary weapon motor can be engage. This will drive forward the carriage holding the drop cam until a rear stop engages a pin on weapon spike resulting in the full torque of the weapon motor being transmitted into the spike. This double-action crusher allows for an extremely fast spike movement of 1", followed by a force which ramps up from 8 to ~120 lbf.