ROS 2 project builds live 3D maps on TurtleBot3
ROS 2 + RealSense D435i combo creates real-time OctoMap SLAM on a TurtleBot3 Burger
A community project by Prakash Aryan demonstrates real-time 3D SLAM (Simultaneous Localization and Mapping) on a TurtleBot3 Burger robot equipped with an Intel RealSense D435i depth camera. The setup streams depth, color, and odometry data over WiFi to a laptop, where ROS 2's octomap_server converts point clouds into a live 3D occupancy map (OctoMap) visualized in RViz. Every mapping session is recorded as a ROS bag for offline replay.
Key challenges included resolving camera_info Quality of Service (QoS) mismatches—where RealSense's VOLATILE QoS clashed with depth_image_proc's TRANSIENT_LOCAL requirement—and optimizing thermal throttling. The Pi's CPU throttled at 80°C during 640x480 depth capture, dropping to 2.5 FPS, but dropped to 6 FPS at 424x240 resolution. The project's repo provides troubleshooting nodes, launch files, STL mounts for the D435i, and a video renderer for offline analysis.
- Real-time 3D OctoMap SLAM on TurtleBot3 Burger with Intel RealSense D435i, using ROS 2 pipelines
- OctoMap voxel resolution set to 2.5cm; thermal throttling fixed by reducing depth resolution to 424x240
- Repo includes launch files, STL mounts, QoS fix nodes, and ROS bag recording for offline replay
Why It Matters
Advances ROS 2-based autonomous navigation with practical hardware integration and performance tuning for real-world robotics deployments.