Shoot on the move resources

[PChild]

We identified shoot on the move as being important for how we want to play the 2026 game. This might be the most complicated part of our robot this year as it requires excellent turret control, pose estimation, and understanding of how the game pieces will behave when we launch them. I thought it might make sense to share some resources for handling this.

• 1690 presentation
• Rusthounds post with video and code for their non-turreted implementation
• GoatSim shoot on the move vizualization

[aidnem]

@PChild I've been working on a PDF that summarizes my findings so far. I'll add it here when I get to the next good checkpoint in progress.

[aidnem]

Here's my first draft of the PDF that includes shoot on the move:

FRC 2026 Math-6.pdf

[godmar]

The "low" shot solution, is it feasible? Wouldn't it make the ball hit the outside of the hexagonal hub funnel walls?
Yes you say that.

[godmar]

Other resources found on CD:

• trajectory calculator with error bounds
• another one, no error bounds
• another one with drag + magnus
• yet another one with drag

[aidnem]

Here's my no drag shot calculator on desmos that I've been testing with.

[PChild]

My only note on the PDF so far is that the hood roller is mechanically coupled to the shooter wheels so we can't independently vary its speed.

[PChild]

A couple more resources:

• A detailed write up with code
• And another one with time of flight stuff
• Fuel sim

[aidnem]

Those writeups are interesting and useful. I hadn't thought of completely eliminating the vertical axis from my calculations. I'll try the "tuning map" strategy that this person suggests and perhaps combine it with fuel sim as soon as I get hood & shooter merged.

[godmar]

That Fuel sim looks really interesting. I'd like to study how this is integrated.

[aidnem]

Justin sent me this docs site (it's a fork of the wpilib docs) that includes Newton's Method for shooting (using the derivative of an error residual to iteratively find the actual time of flight for a shot faster than a standard iteration would due to oscillation). It looks promising, my only concern is that I'm not sure I understand where they get their number for $v_p$. I can look into implementing it if our current first order approximation (finding $\tau = \tau(D)$, shifting our position by that time, and then not re-calculating to account for the new time of flight) isn't accurate enough.

[aidnem]

I've updated my super long PDF with the info from that docs article, and I think I understand it now.

[Samson560915]

@aidnem How are we deriving through these equations?

For the first equation, is the r robot vector not a distance vector? As why are we multiplying it with time (is it supposed to be a velocity vector).
And for the second: Are we deriving that from the first? Or is that a new equation? Because deriving it from the first doesn't seem to add up when you would need to add r robot to both sides.
Thank you for your time!

[aidnem]

It looks like I made a typo there. That should be v_robot * t_1 to find the amount that the velocity of the ball contributed by the robot causes the "effective" goal to "move". PDF is updated now.