3D Frontier Explorer — Development Environment

ROS 2 Humble + Gazebo Classic + sjtu_drone + OctoMap, packaged as a single Docker environment for the 3D Frontier Explorer project (disaster-response UAV exploration in GPS-denied 3D interiors).

---

Quick start

1. First-time setup

docker compose build

2. Start the environment

docker compose up

Leave this terminal running. Open a browser to:

http://localhost:8080/vnc.html

You will see a Linux desktop. Gazebo and RViz2 will appear here.

---

3. Apply Velodyne patch and build

Open a container shell:

docker compose exec ros bash

Inside the container:

cd /root/ros2_ws/src
./frontier_setup.sh

This clones dependencies, patches the sjtu_drone URDF to add a VLP-16 LiDAR, and builds all packages. Safe to re-run if the container is recreated.

---

4. Add collision avoidance module

In a container shell:

cd /root/ros2_ws/src
./collision_avoidance.sh

---

5. Add frontier extractor module

In a container shell:

cd /root/ros2_ws
colcon build --packages-select frontier_explorer_py --symlink-install

---

6. Add NBV + RRT* module

In a container shell:

cd /root/ros2_ws
colcon build --packages-select frontier_explorer_py --symlink-install

---

6. Launch the simulator

In a container shell:

source /root/ros2_ws/install/setup.bash
ros2 launch sjtu_drone_bringup sjtu_drone_bringup.launch.py

Gazebo and RViz2 should appear in the browser tab.

---

7. Takeoff + static TF

In a container shell:

source /root/ros2_ws/install/setup.bash

# Takeoff (You can directly enter the command to takeoff)
# ros2 topic pub /simple_drone/takeoff std_msgs/msg/Empty "{}" --once

# Publish the velodyne_link → base_footprint static transform
ros2 run tf2_ros static_transform_publisher \
  0 0 0.10 0 0 0 simple_drone/base_footprint velodyne_link

---

8. Run octomap_server

In a container shell:

source /root/ros2_ws/install/setup.bash
ros2 run octomap_server octomap_server_node --ros-args \
  -r cloud_in:=/simple_drone/velodyne_points \
  -p frame_id:=simple_drone/odom \
  -p resolution:=0.1 \
  -p sensor_model.max_range:=25.0 \
  -p use_sim_time:=true \
  -p publish_free_space:=true

publish_free_space:=true is required for the frontier extractor.

---

9. Run frontier extractor

In a container shell:

source /root/ros2_ws/install/setup.bash
ros2 launch frontier_explorer_py frontier_extractor.launch.py

---

10. Run NBV + RRT* Navigator

In a container shell:

source /root/ros2_ws/install/setup.bash
pkill -9 -f teleop 2>/dev/null
ros2 launch frontier_explorer_3d explore.launch.py

teleop might intercept the navigator command.

---

11. Visualize in RViz2

In RViz2 (visible in the VNC browser tab):

1. Set Global Options → Fixed Frame to simple_drone/odom
2. Add displays:

Display type	Topic	Notes
OctoMap	/octomap_full	requires octomap_rviz_plugins; shows 3D occupancy
PointCloud2	/frontier_extractor/frontier_cloud	colour by cluster_id field
MarkerArray	/frontier_extractor/cluster_markers	coloured spheres at cluster centroids
PoseArray	/frontier_extractor/cluster_centroids	viewpoint candidates for NBV

Monitor frontier counts:

ros2 topic echo /frontier_extractor/status
# {"num_frontiers": N, "num_clusters": K}

---

Next steps (project roadmap)

1. ✅ Environment up
2. ✅ Velodyne VLP-16 LiDAR added to sjtu_drone URDF
3. ✅ LiDAR → octomap_server → 3D occupancy in RViz2
4. ✅ frontier_explorer_py: incremental frontier extraction + clustering
5. ✅ frontier_nbv: information-gain next-best-view selector
6. ✅ frontier_rrt: RRT* collision-free path planner
7. ⬜ Collapsed-building Gazebo world for evaluation