BioBuzzAutoDog

Vision work for FTC DECODE (2025-2026): finding the green and purple balls with a camera, tracking them in field coordinates, and driving the robot to chase them like a dog.

What's here

docs/

• ftc-auto-dog.pdf — design document for the auto dog: pixel to field projection, per-track Kalman filter, gating, Mahalanobis association, and pose time sync.

python/

Desktop computer vision tests, from FTCDecodeVision.

• vision.py — main AprilTag-anchored green/purple ball detection pipeline
• vision_fixed.py, vision_debug.py, vision2.py — variants and debug views of the pipeline
• ballDetect.py — ball detection plus tracking with an OpenCV tracker
• SimpleBallTracker.py — SimpleBallTracker class: per-frame green/purple detection plus CSRT tracking, meant for robot integration
• LuxonisTracking.py — 3D ball trajectory tracking with a Luxonis OAK-D (depthai 3.x), with CSV logging
• empiricalTrajectory.py — empirical trajectory recorder
• remoteTestingRoom.py — saves frames and debug masks for offline analysis
• capture_calib.py, calibrate.py, calib_imgs/ — chessboard capture and camera intrinsics calibration (the K and distortion values the design doc's pixel to field projection needs)

robot/

On-robot Kotlin code from the tanayVision branch of FTC-SigmaCorns-22377/Decode, kept at its original TeamCode paths. These files depend on the rest of that repo and do not build on their own.

• GlobalShutterVision/eyebrain.kt — EyeBrain, the VisionProcessor port of vision.py for the global shutter camera
• opmode/test/VisionDriveTest.kt — teleop that uses EyeBrain on the globalShutter webcam to turn toward and chase a ball
• opmode/test/ShooterExperimentTest.kt — shooter experiment teleop for camera system integration
• control/localization/VisionTracker.kt, constants/Limelight.kt, res/xml/teamwebcamcalibrations.xml — team localization and camera calibration code the vision work relies on (written by Sigma Corns teammates)

Running the Python tests

cd python
python -m venv .venv && source .venv/bin/activate
pip install opencv-python numpy depthai
python vision.py