| jerrym | 21d0433 | 2013-02-18 09:40:08 +0000 | [diff] [blame] | 1 | This is a working computer vision target recognizer for the FRC 2013 game.
|
| 2 |
|
| 3 | Run it on 2013-test-images/*.jpg.
|
| 4 |
|
| 5 | TODO: Capture 640x480 images with the robot's camera (Logitech C210), green LED
|
| 6 | light ring, and the shutter speed turned down as in 2012.
|
| 7 | Also turn off auto whitebalance.
|
| 8 |
|
| 9 | src/org/frc971/DebugCanvas.java -- displays intermediate frames for tuning
|
| 10 | src/org/frc971/Recognizer.java -- the vision recognizer interface
|
| 11 | src/org/frc971/Recognizer2013.java -- the UI independent vision recognizer
|
| 12 | src/org/frc971/VisionTuner.java -- a Java UI for testing and tuning the recognizer
|
| 13 |
|
| 14 | TODO: Connect the recognizer to the robot camera (e.g. using ffmpeg).
|
| 15 | TODO: Get the relevant robot measurements: Camera height, setback, pitch,
|
| 16 | horizontal FOV, and vertical FOV.
|
| 17 | TODO: Compute target distance, elevation, and azimuth, and send them to the
|
| 18 | robot's control loops.
|
| 19 | TODO: Pass through a timestamp with each frame.
|
| 20 | TODO: Wean off of WPIJavaCV and its DaisyExtensions.
|