A sad day for the team.
Today is a sad day. We lost a trooper in the field. The Spark Max, a motor controller that has been with us since the beginning of the season, has been fried. The cause of death is unknown, but it is suspected that the motor controller was plugged in backwards. The motor controller was found dead on the field, with the smell of burnt electronics filling the air. The Spark Max will be missed, but we will continue to fight on in its memory. It was declared dead at 4:20 pm on October 2nd, 2024 with the cause of 12v gate fault. Rest in peace, Spark Max. You will be missed.
Electrical · Frc · Rev · Spark · Sparkmax · Whoopsie
1 minute
FRC Vision Research Part 1b
It’s been a while. I know this isn’t AprilTag stuff, but the next blog post (coming soon) will be. This is kind of a personal project, but it’s related to vision. So I did more Coral stuff. I tried to reverse engineer and rewrite libedgetpu in Rust and actually learned quite a bit of possibly useful information, even if it feels like a waste of time in hindsight. The README explains why I’m doing this (pardon the tone):
Code · Edgetpu · Frc · Tpu · Vision
7 minutes
FRC simulation in godot to train drivers virtually.
This year, before we were able to finish designing and building our 2024 robot Bothoven, I was busy at home reconstructing the field in Godot from the CAD models provided by First to make a semi-realistic simulation. Building the field To make a good simulation you first have to start with the environment. In this case, it would be the FRC field. I first imported the CAD model from the Onshape page into Godot, however, all the textures and colors were gone due to the file format/Onshape. To fix this I made scripts in Godot to help apply textures I sourced online. I grabbed the carpet texture from the manufacturer’s website, images from first, and textures from various online sources. When I was done it looked something like this:
Code · Frc · Godot · Simulation
6 minutes
FRC Vision Research Part 1
So I’ve been doing a lot of research and work on vision stuff. This is going to be a multi-part series, because there’s a LOT to cover. Edge TPU So what is an “edge TPU”? A made-up thing called a “tensor processing unit”. It’s really just an ASIC1 designed by Google for fast, efficient machine learning in embedded environments. It’s great for FRC, because it’s a relatively low-cost, easy way to do tons of cool ML stuff off of a Pi.
Code · Edgetpu · Frc · Tpu · Vision
4 minutes