Kettering Combat Robotics is proud to present Mad Dog! Us students of Kettering University have been hard at work designing a 250 pound combat robot over the course of the last year and a half. As almost all of our members are FIRST alumni, we were heavily inspired by the robots that shaped us, which is why Mad Dog is going to be a pneumatic flipper built on a swerve drivetrain. We just finished applying to BattleBots for the 2021 season; check out our application video here!
We’ve started manufacturing our robot, and we will continue onward with this learning opportunity regardless of whether we are accepted to compete on the show or not. Check out our progress on Instagram or Twitter to follow along and see Mad Dog come to life!
A swerve battlebot?
Shatter’s gonna start feeling inadequate over here…
A swerve drive battlebot, it was only a matter of time…
This makes me happy to see. I’ve joked around with my team mates about building a swerve battle bot for ever. So glad someone is finally making one.
But isn’t that half the fun?!?!
Darn! You beat me to it!
I’ve been designing a swerve battlebot for a few years now (although I never had any resources to do anything other than design). I’m curious as to how you are “battle hardening” the drive modules. I’ve felt that swerve the way its designed for FRC is to fragile for battlebots, so i’ve been redesigning one to be more robust.
I absolutely love your robot tho! And I will be cheering for you assuming you make it onto the show!
Someone clip the Aren Hill interview where he mentioned FRC people trying to do swerve battlebots
What are you guys thinking about for control systems? While you obviously would have access to FRC-type controllers like the RoboRIO, most of the competition in the combat robotics space seems to favor standard RC controls for the simplicity and reliability; though obviously, if you’re doing a swerve, you’d still need some programing capabilities to be able to control it adequately.
I also seem to recall seeing one robot on Battlebots a few years back that connected using some sort of WiFi control and ended up having connectivity issues.
I vote 4 drivers for 4 corners. Weapon operator stands in the middle and shouts what the plan is.
wait, this isn’t how you normally do swerve?
We are controlling it quite a bit differently than FRC robots. The main constraint that we had for the control system was to have as much redundancy as possible so that we won’t lose any controls from any single failure on the robot.
We are running a 32 bit teensy 4.1 microcontroller for our swerve calculations (doing all of the vector additions and trig functions). We are also using a teensy lc microcontroller on each of our individual swerve modules that run the PID loops and encoder feedback so that we don’t have fragile encoder cables running through our entire robot. The teensy LC’s receive commands that tell them the target vector for their wheel over dual I2C buses (similar to CAN). We are running these communication buses in a dual ring topology so that we can lose at least 3 connections without having any swerve modules come off line.
We are using an FRSky radio transmitter and receiver system that is pretty similar to most battlebots teams, since it has practically zero latency and connection delay. The one that we are using has a 16 channel Sbus output, which is a digital signal that is read by the teensy 4.1 and used to send commands to each subsystem controller.
(Another Edit) We are also using REV through bore encoders on our swerves to read the Azimuth. We chose these because they actually provide a digital SPI output as well as a quadrature/absolute output. We are using the SPI output because it continues to track position and provide an absolute signal even if connection is lost, so there isn’t as much of a chance of a wheel losing it’s encoder position if something unexpected happens.
For battle hardening, we are using a third coast style swerve (thanks strykeforce!) that is made from laser cut 4130 chromoly steel and some pretty beefy 6061 T6 aluminum plates. We did a lot of FEA work on the swerve to ensure that it can handle pretty heavy amounts of loading in all directions. The biggest concern was designing it to handle a pretty large impact with the ground from an 8 foot drop, so we did some nonlinear simulations to come up with a geometry that would be able to distribute the energy without damaging any components.
The biggest concern with the swerve right now is that the bevel gears may shear off some teeth if the bot lands perfectly on the wheels. We are planning on doing some destructive testing with some of the prototypes soon to see exactly what the effects are and we have a few ideas for ways to mitigate damage to the bevels.
I have actually joked with an FRC team about doing this as a team building exercise…
From the little combat robotics experience and the large destructive testing experience I have, I have one suggestion. Go to the hardware store, buy the biggest, baddest sledgehammer you can find. Then wail on your modules and anything else you’re worried about.
That is what I have done with a few of my 3 pound robots, but I don’t think that a hammer will hold a candle to the force that a 250 pound robot can dish out. We have some ideas for destructive testing, mainly strapping a bunch of weights to one of the modules and dropping it from 8 feet onto a big steel plate.
This sounds like a robotics equivalent to The Campaign for North Africa.
Those are rookie numbers in this racket, you gotta like double that and do it once every 5 seconds.
Nooo a swerve battle bot was always my dream. lol but good luck anyways
Hydra did seem to be hitting a bit higher than 8 feet this season, so that is definitely something to consider. The good news with our robot is that it is very unlikely that we would hit wheels first after a drop like that, so most of the force shouldn’t be absorbed by our swerve modules. If we do get unlucky and land perfectly on one of the wheels and it breaks, we should have enough drivetrain power to continue moving on three wheels without too much trouble. My personal preference would be to just not let them flip us, but that is easier said than done (Although a fast swerve can hopefully get to their side).
We are planning on bringing enough spare parts to build a few extra robots, so if we lose a swerve it shouldn’t be a big deal.
I was thinking about the same as well. Some napkin math shows it’s be difficult to force landing in a certain orientation via intentional drag (something like the nascar roof flaps) or weird weight distribution. There might be some interesting ways to passively “retract” the modules, but that’s unnecessary complexity when you can add something similar to the printed wheel wedges teams used in 2020 and cut the chance of a direct wheel impact to almost nothing. There’s a whole geometry rabbit hole you can go down there, especially with the twist of the ground game.
And plz get your driver “utilizing swerve to its fullest potential”. Shatter still drives too much like a tank bot at times imo.