AlphaBit - OpenML

Auto Aim with IMU + Webcam Fusion

Competition Mode

Fusion mode combines IMU stability with vision correction. The key rule is simple: IMU runs continuously, camera corrections are applied only in trusted windows.

Step 1

Define Source Roles

IMU: continuous heading baseline (fast, stable).

Webcam/AprilTag: absolute correction when tags are valid.

Localizer: keeps translation estimate between corrections.

Step 2

Implement a Safe Correction Gate

Use correction only when robot is in a stable state:

if (allowedToShoot && !manualControl && isRobotStationary) {
    aprilTagIdentification.getRobotPose();
    if (aprilTagIdentification.locTagFound) {
        calculatedRobotPose_X = aprilTagIdentification.robotPose_x;
        calculatedRobotPose_Y = aprilTagIdentification.robotPose_y;
        robotAngleAprilTag = aprilTagIdentification.bearingAngle;

        gyroscope.resetHeading();
        gyroscope.setAngleOffset(36.5 - robotAngleAprilTag);
        drive.setPose(new Pose(calculatedRobotPose_X, calculatedRobotPose_Y,
                Math.toRadians(126.5 - robotAngleAprilTag)));
    }
}

Step 3

Handle Alliance Variants Explicitly

Use separate angle offsets for red/blue.

Do not merge alliance math into one undocumented expression.

Validate each alliance independently on field.

Step 4

Add Fallback Behavior

If tag confidence is low, continue with IMU-only aiming.

Do not force correction when detections flicker.

Expose a one-button fallback toggle to drivers.

Step 5

Validate Fusion Robustness

Parked correction test (X/Y/bearing convergence).

Short motion segments + correction windows.

Full-cycle stress test with burst shooting active.

Match Rule

Run fusion as primary only after IMU-only and webcam-only modes are independently validated.

Fusion should improve reliability, not hide unstable base subsystems.

Support -> Discord

Choose Language / Alege Limba

AlphaBit AI Assistant