Shooter speed calculation from vision tracking

[Ether]

Originally Posted by Alon

by physical calculation i mean calculating the position and height of the ball at a given point with physical equation that include air drag and magnus effect

That will be very precise but not as accurate as empirical data.

Besides, there is no closed-form solution to the differential equations of motion when you introduce air drag and magnus. You would have to do a numerical simulation as discussed in the paper I linked. You probably don't want to be doing that in the roboRIO.

[Alon]

Originally Posted by Ether

That will be very precise but not as accurate as empirical data.

Besides, there is no closed-form solution to the differential equations of motion when you introduce air drag and magnus. You would have to do a numerical simulation as discussed in the paper I linked. You probably don't want to be doing that in the roboRIO.

by empirical data you mean like a table of values?

[Ether]

Originally Posted by Alon

by empirical data you mean like a table of values?

By "empirical data" I mean the test data you collected by actually firing your shooter from different distances.

You can put that empirical data into a table and do a table lookup in your code.

Or you can fit a polynomial to that data and compute the polynomial in your code.

Or you can construct a piecewise-linear function from that empirical data, and compute that function in your code.

Or you can construct a piecewise-cubic-spline function from that empirical data, and compute that function in your code.

[euhlmann]

We use a catapult-like pneumatic arm instead of a shooter wheel, but we have found that as other people have stated on this thread, the values we set to make shots look really arbitrary. They're most likely a combination of multiple physical effects and very hard to predict.
We used linear interpolation between values in a lookup table we got from test data. It's been very successful so far, and I suggest you do something similar instead of trying to simulate the physics of the shot. Retaking values for the table took us about 15 minutes on the practice field before our competition last weekend, so it shouldn't be too hard after you've done it a few times.

[Ether]

Originally Posted by euhlmann

We used linear interpolation between values in a lookup table

A piecewise linear function is essentially the same thing as an interpolated lookup table with unevenly-spaced x values.