Welcome to the NERD Spark 2025 Build Thread.
Links to past resources we have developed and will be using again this season:
Swerve Base Reconfigurable Chassis CAD:
Featuring under belly electronics and quick change battery system. Auto-regenerates laser cut pieces based on frame size.
NERDSwerve Training Platform
website houses links to CAD, tutorials and programming to build your own swerve enabled 3D printed mini-bot.
We have a dozen or so programmers and have built 6 NERD Swerves. We will be using them, as we did this summer, to do training and give everyone as much hands-on time as possible. As the build season progresses we will likely need to harvest Krakens for the actual comp-bot.
On to this season.
One of our highest priorities has always been to support other local teams. We have assisted a few teams in past seasons peripherally but this season we are integrating one of them directly into our own. 8522, Springport Robotics, is just down the street from us, the next town over. They are a rebooted team that we’re excited to call our sister team now. We will be sharing build and programming resources. Some of our team members and mentors will be hopping onto their team to directly assist and help develop their program.
The goal is to build 2 almost identical yet highly complimentary robots with some very big differences.
NERD Spark’s robot will be programmed in JAVA, use Orange PI’s/Photovision, and have a Kraken base MK4i drive train. This will allow us to take full advantages of WPILib, Path Planner and CTRE’s TunerX Swerve generator.
Springport’s robot will be programmed in LabVIEW, use Limelights 3G’s and have a NEO Vortex based MK4i drive train. We have a pretty competent swerve code base developed in LabVIEW which we have been running for the past 2 season (reminder, yes, we’re the team that runs both LabVIEW and JAVA). Through our experimentation over the summer, we believe we have drive wheel swerve odometry working and are able to read Path Planner JSON files to perform complex path commands.
demo video:
The only thing we are missing is vision and full field localization. Pre-season work has shown that we can read NT variables from Limelight that provides pose-estimations. Should just be a couple weeks of integration, testing and proving out. We are hopeful to have similar full field localization functionality as the JAVA robot.
This is a giant experiment and challenge for our team members. As we learn from our successes and failures we’ll post it here.
As always comments and questions are welcomed.