WEEK 3 AND 4 RECAP
How to start this. We’ve been extremely busy with everything including CAD, prototyping, fabrication, ordering parts and hardware, t-shirt making, making a new robot cart and our impact award submission among many other things. I’ll do my best to summarize.
Apart from the manipulator which is at the end of the prototyping phase, the rest of our CAD is well advanced, enough to start fabrication and send parts that we can’t do in-house to be fabricated. You can view our CAD here: Onshape And here are some pictures of our progress:
Our drivetrain is a 4 Falcon WCD with 5" wheels on a 6:1 gearbox. We are betting strongly on speed and high maneuverability and driver practice to best the swerve drivetrains we’ll play with or against. We also did a hex-patterned bellypan again this season cause it looks very cool.
Our arm is a very neat subsystem. It uses the upper axle as a dead axle and a live axle at the same time. The goal with this was to keep both gearboxes as low and as centered to the robot as we could to help with our center of gravity. The arm itself should be pretty light, with the heaviest part being the manipulator at the end of it which is still to come.
Our turret is based on the WCP GreyT turret. We used it last year on our robot, but we are remaking every plate to adjust them to the mounting on this year’s robot and some of the cuts we made last year made the turret hard to reuse as it was. We are also buying new bearings and replacement hardware. The motor also now features a MAX 90 degree gearbox to allow the turret to be as low as we could mount it within our capabilities.
We’ve done a bit of prototyping so far, but we decided to try a version of the everybot intake. This design is very versatile and quite a lot of variables can be changed like the type and size of wheels used, the space between the axles, and the angle of each roller pair. We prototyped with 2 different models and a few combinations of wheels. Our goal is that our next design makes it onto the robot, keeping the possibility that we might want to iterate it. Here are some images of our prototyping.
WE HAVE SUCCESSFULLY FAILED
Our goal was to focus on the 1st and 2nd level of the grid to place cones and cubes. Well our design reaches the highest level as well. We still want to focus on the first two levels but… But having the option to aim for more later is still very nice to have, so we’ll keep our design that way for now. We also have the possibility of adding a second backup configuration of arm lengths since the only change is just the cuts on the tubes which might be helpful if we ever need to swap things.
FABRICATION AND ASSEMBLY
We are almost done already with the fabrication of our drivetrain and arm already. The other parts like the turret plates, based on the WCP turret, and our Bellypan are sent to be machined by a company since we don’t have the capacity in-house.
Our cut MAXtubes
Everything we send are 0.125" aluminium plates to be waterjet. Having every part made the same way keeps costs reasonable for us. It’s the 3rd time we outsource parts this way and it gives us really well made parts.
The cuts on our 2x1 and 1x1 MAXtubes are all done and were pretty easy to make on our cirular saw. However, the bearing holes gave us more trouble than we thought. First of all, our stepdrill gave and stopped drilling due to excessive ware through many seasons, so we had to buy a new one. But still, it was hard to make really well centered bearing holes, even though we used the 10 holes already on the stock to guide the drill. A few holes were up to 0.018" off-center, which worries us about alignment when we’ll do the assembly. We have the capacity to do them again, but we would like to avoid wasting new MAXtubes if possible. We might also try other solutions, like usign bearing blocks if the problem persists. Here are what the off center holes look like. One of our mentors made a 1" hole using better equipement to test the MAXtubes.
Drilling all the way in one go
The 10 holes make it easy to see how the 1" hole is a bit off center. And in this case, it’s a skilled mentor that made the test using better equipment we usually use at the school.
Drilling both holes separately
Even when drilling both holes separately, there is still a slight offset.
2 main conclusions to those tests
The first conclusion is that we may be a lot less well equipped than we thought to make centered holes. We are using a small drill press that might not allow us to make such big holes with precision.
The second conclusion is that the 10 hole grid on the MAXtubes are kind of… oval. We haven’t measured precisely but some of the holes on the grid seem kinda off. There are some other slight manufacturing issues on the tubes we have that we noticed. We don’t know if that’s a common issue or if another team brought that up, but we could share more about it if you are curious about it.
This part was written by our Lead Programmer.
Bonjour everyone! I’m Charles, the lead programmer of team 3544. This week, we’ve achieved a ton of progress on both a vision subsystem and just general programming stuff.
As of now, we’ve setup a basic 3D Apriltag tracking solution using a Limelight 2, although we are planning to get a Limelight 3 soon. I’ve observed an average FPS of about 20 at a 640x480 resolution, which is generally sufficient for basic vision. You can get higher FPS with a lower resolution, although tracking range is somewhat hindered.
We have also started writing some basic code, even if we don’t have the full robot yet. We’ve chosen to use command-based programming this year, which allows us to have more “modular” code compared to a traditional project. Lots of subsystem code is already in place, but we’re waiting for the robot to be a little bit more complete to start writing more in-depth commands.
Drivetrain, arm and turret subsystems already have their basic code written. All that’s left is to test it on the robot and to tune some gains. Since we mostly use Falcon 500s, we’ve opted to use Motion Magic to accurately control the position of our arm and turret. It also makes their movement much S M O O T H E R compared to a traditional PID controller. Some vision code is here as well.
As always, if you want to see what we’re cooking up in more detail, don’t hesitate to check out our GitHub repository, where you can always consult our most recent code
T-SHIRT MAKING FOR OUR FTC TEAM
Unexpectedly, we found ourselves making t-shirts the day before our FTC team’s competition. It was the first time we tried making shirts on our own and after a disastrous first attempt, we were able to make something that looks very nice. We liked doing that so much that now we decided to also make the t-shirts for our FLL teams as well in the following weeks.
The first failed attempt
The end result
I hope you learned something interesting in this recap and hope that things are rolling with your teams as well!
Our robot right now