Roadmap needed: Upgrading from 2D Nav2 to 3D mapping and navigation with RTAB-Map

[ajmalshariff]

Hi everyone,

I have successfully generated a 2D occupancy grid map and configured the Nav2 stack for autonomous navigation, and it works great. Now, I want to upgrade my setup to generate a 3D map and navigate within a 3D environment.

However, I am having trouble finding comprehensive documentation or a step by step guide to follow for this specific transition. Could anyone point me toward a reliable path, workflow, or tutorial to get this running?

My Setup Details:

SBC: Raspberry Pi 4B (8GB)
Odometry Source: Wheel encoders
Vision Source: OAK-D Lite
Framework: ROS 2 Jazzy

Any guidance, recommended parameters, or reference links would be greatly appreciated!Thanks in advance.

[matlabbe]

Hi,

generate a 3D map and navigate within a 3D environment

as you are on RaspberryPi 4B, I would recommend to revise if you really need to see the environment in 3D. If the robot is a ground robot, you should be able to approximate most environments in 2D and use Nav2. The trajectory can still be estimated in 6DoF (terrain not flat), but the occupancy grid / costmap could be projected in 2D. Note that the 2D occupancy grid can indirectly include 3D obstacles. See turtlebot + RGB-D camera obstacle detection and 2d map generation demo: https://github.com/introlab/rtabmap_ros/tree/ros2/rtabmap_demos#turtlebot3-nav2-and-rgb-d-slam

cheers,
Mathieu

[ajmalshariff]

Okay, Thank you Mathieu, I will definitely check that, however I am planning to upgrade to Raspberry pi 5.