This is the vision module for Tinker 2026 based on ROS2. Current action servers available: object detection based on YOLO-seg model person tracking with ReID