AlphaBit - OpenML

Auto Aim with Webcam Only

Vision-Driven Mode

Webcam-only mode is useful for vision-focused testing and correction experiments. Treat it as a controlled mode, not your only match fallback.

Step 1

Configure Vision Stack

AprilTag processor for ID + pose information.

Color blob processor (optional) for artifact context.

VisionPortal as runtime camera pipeline manager.

Step 2

Set Detection Filters

Decode-style blob filtering parameters:

Contour area range: 50 to 20000

Circularity range: 0.6 to 1.0

Use ROI to reduce irrelevant background detections.

Step 3

Bind Vision Output to Aim Decisions

Use tag metadata confidence to decide if correction is allowed.

If confidence drops, hold previous stable target or fall back to IMU mode.

Never feed noisy frame-by-frame angle directly to servos.

Step 4

Test in Match-Like Conditions

Test under bright and dim lighting transitions.

Test with motion blur (sudden turns/acceleration).

Test with temporary occlusions by robot parts and game elements.

Step 5

Keep Safety Layers Active

Keep shooting-area checks active.

Keep manual cancel and fallback mode switching on controller.

Store telemetry logs so you can debug confidence drops quickly.

Webcam-only mode can degrade fast if lighting changes suddenly. Always keep an IMU-based backup.

Continue to IMU + Webcam Fusion for a more robust competition-ready mode.

Choose Language / Alege Limba