Standalone tools to project NavOCR detections to 3D world coordinates, without modifying the original NavOCR code.

• navocr_slam_cpp: Integrated C++ node that subscribes to NavOCR detections, images, depth, and camera info to compute 3D world coordinates via rtabmap SLAM. Publishes RViz markers and saves CSV + annotated images.

Inputs:

• Image: /camera/infra1/image_rect_raw (sensor_msgs/Image)
• Depth: /camera/depth/image_rect_raw (sensor_msgs/Image, 16UC1 in mm)
• CameraInfo: /camera/infra1/camera_info
• Detections: /navocr/detections (vision_msgs/Detection2DArray)
• Odometry: /odom (nav_msgs/Odometry)
• TF: camera_infra1_optical_frame → camera_link → odom → map (via rtabmap)

Outputs:

• /navocr/markers (visualization_msgs/MarkerArray) - Green cubes and text labels in RViz
• CSV file: results_cpp/detections_YYYYMMDD_HHMMSS.csv with columns:
  • frame, bbox (x,y,w,h), confidence, timestamp (sec, nsec)
  • depth_m, camera (x,y,z), world (x,y,z), has_world_pos
• Images: results_cpp/images/detection_XXXXXX.jpg with drawn bounding boxes

Python deps (example):

pip install ultralytics opencv-python numpy

Make sure your ROS 2 workspace has cv_bridge and image_geometry installed (standard on ROS 2 Humble desktop). If using system Python, prefer a virtualenv/conda and make sure ros2 run resolves to the same environment.

From your workspace root (e.g., /home/sehyeon/ros2_ws):

colcon build --packages-select navocr_projection
source install/setup.bash

Full integrated system with rtabmap for accurate world coordinates:

Terminal 1: Play rosbag with clock

ros2 bag play ~/Downloads/rosbag2_2025_11_18-10_05_53 --clock

Terminal 2: Run NavOCR detector (Python)

cd ~/ros2_ws/src/NavOCR
conda activate navocr  # or your NavOCR environment
python3 run_navocr_ros.py

Terminal 3: Launch integrated rtabmap + projection

source ~/ros2_ws/install/setup.bash
ros2 launch navocr_projection navocr_slam_full.launch.py

Terminal 4: RViz visualization (optional)

rviz2 --ros-args -p use_sim_time:=true

Results:

• CSV: results_cpp/detections_YYYYMMDD_HHMMSS.csv
• Images: results_cpp/images/detection_XXXXXX.jpg
• RViz markers: /navocr/markers topic