Kickoff Update
Alongside a few other teams we gathered for kickoff at the NYC First stem center at noon and immediately began brainstorming after the reveal. Our primary goal was to have as many prototypes going on as possible but not just for the sake of ri30h but rather having late season competitiveness in mind. We split off into several 1-3 man cells each working on their own prototype or concept.
Prototypes- Quad Claw
The concept of this quad claw is to be able to dig down into the triangle indent of the sample and clamp outwards like
so as it has the natural tendency to realign itself. Once its up in the air we can grip onto the whole sample with another external gripper as to safely and steadily transport it. Coupled with a downwards facing color sensor, you could significantly reduce cycle times and wasted movements.
Roller Claw
Basically the entire claw design was created with space in mind so the final dimensions of the entire thing (excluding the serve poking out) would be 1.75 x 2.165 x 3. This design is supposed to include both a roller and a claw where the roller is able to kick the game piece up into the claw so that the mechanism is able to hold onto it firmly while also allowing for intaking from every direction. The claw opens and closes with two meshed gears and the roller is supposed to be run by two pulleys that have a 1:4 ratio
Carbon Fiber Arm
This is one of the ideas for an arm mechanism which is simply comprised of unfurling a big spool of carbon fiber that we originally were gonna use for frc last year
. This was cadded and machines by the students overnight and served as a
proof of concept, we like it because of how sturdy and lightweight it is while being able to support a load as heavy as a hammer.
Dual Joint Arm
This arm is inspired by team
971 Spartans 2023
robot and we really like it due to how unique it is and how high its potential is. Using inverse kinematics we can sent both joints of the arm to precise coordinates on a specified plane under custom paramaters and achieve some crazy results. So far a badly tuned pid shows that this arm can get the job
done pretty well so we have high hopes for a fully motion profiled arm control algorithm.
Climber In a Box Arm
We used the Thrifty Bot extrusion kit from
FRC to prototype a telescoping arm tailored to our goals. We customized the design to incorporate string, similar to traditional
FTC mitsumi slides, instead of the rope provided with the kit. The idea is to create a multifunctional mechanism that can serve both as a manipulator and a climber. This design was inspired by
KookyBotz CenterStage bot and
FRC Team 3015 Ranger Robotics 2023 bot.
Software
One of our proudest accomplishments from these sleepless 30 hours is the pathway we’ve laid out for ourselves with vision. The first thing that was worked on was a stable relocalization method that given a detected april tag could return our robots pose relative to the field. This wasn’t without its challenges as while I was implementing this method the apriltaglibrary for the 2025 season wasnt out yet so I had to create our own library that reflected the stem centers field. FInding the XYZ coordinates of each tag is easy enough but the quaternions? Not a fun set of calculations to do at 2 am. We essentially had to reverse engineer the wxyz values of the quaternions for the centerstage tags to find the original frame of reference the sdk uses.
With the help of an amazing teacher we ended up creating a set of references and learning a bunch about how ftc handles tag definitions and metadata. As shown by this
video without any odometry we get a localized pose accurate up to ± 0.5 inches.
Camera On a Servo
Much to my despair, there were no tags on any of the field elements besides the boundary line so I continued down the rabbit hole of looks at tags to relocalize. One of the things my team figured would help is having a camera on a servo that would allow it an adjustable cone of vision.
Using this extra degree of freedom we made it so that not only can the camera track a visible tag but it can recognize when theres a closer tag and snap to that one as shown
here. (Last feature is still WiP as we need to reference our robot pose from our odometers to know where the closest valid tag is).
We had a blast pulling this all nighter challenge and I think we set ourselves up very nicely for the rest of the reason. We will continue to update routinely as we achieve new milestones, Cya!
