Greenhouse Inspection Robot

ROS2/RK3588 greenhouse inspection robot project for tomato recognition, LiDAR-IMU mapping, RViz goal navigation, obstacle avoidance, VNC remote viewing, chassis control, and spray control.

System Features

• FAST-LIO2 LiDAR-IMU fusion for real-time localization and point-cloud mapping.
• RViz 2D Goal Pose target setting with autonomous navigation and obstacle avoidance.
• 2D planner and path follower that publish /visual_global_path, /visual_local_trajectory, and /cmd_vel.
• RK3588 NPU YOLOv8/RKNN tomato detection for ripe and unripe tomato recognition.
• VNC remote desktop demos for RViz, camera preview, and YOLO visualization.
• C30D chassis bridge for /cmd_vel to serial motion control.
• Spray controller node with /spray_cmd manual control and optional /tomato_summary auto trigger.

Main Demo Commands

Autonomous navigation + RViz + plain camera preview:

/home/elf/stm32/inspection_demo/start_greenhouse_vnc_navigation.sh

Manual VNC driving + RViz + YOLO tomato recognition:

/home/elf/stm32/inspection_demo/start_greenhouse_vnc_manual.sh

Only show YOLO tomato recognition in VNC:

/home/elf/stm32/inspection_demo/start_vnc_tomato_yolo_only.sh

Spray controller:

/home/elf/stm32/inspection_demo/start_spray_controller.sh
ros2 topic pub --once /spray_cmd std_msgs/msg/Bool "{data: true}"

Stop demo processes:

/home/elf/stm32/inspection_demo/pause_all_robot_demo.sh

TigerVNC connection:

<robot-ip>:5901
password: 11111111

Repository Layout

• stm32/inspection_demo: one-key demo scripts, VNC scripts, dashboards, camera viewers, spray control, and report code snippets.
• stm32/tomato_rk3588_deploy: RKNN YOLOv8 tomato inference code, classes, RKNN model, and C++/Python camera demos.
• stm32/ego_planner_ros2_ws/src/Ego-Planner-2D-ROS2: 2D planning, obstacle extraction, RViz goal handling, and path follower.
• stm32/fastlio2_ros2_ws/src/fast-lio-ros2: FAST-LIO2 mapping source and RViz configurations.
• stm32/rslidar_ros2_ws/src: Robosense LiDAR ROS2 driver source.
• stm32/tools: C30D chassis bridge and sensor/debug helper scripts.
• stm32/SENSOR_COMMANDS.md: sensor and demo startup command reference.

Important ROS Topics

/rslidar_points
/imu/data
/Odometry
/cloud_registered
/Laser_map
/goal_pose
/visual_obstacles
/visual_global_path
/visual_local_trajectory
/cmd_vel
/tomato_summary
/spray_cmd

Notes

• Generated build/, install/, log/, .git/, and Python cache directories are intentionally excluded.
• The RKNN tomato model is included because the demo depends on it.
• Third-party projects keep their original license files in their own directories.