Looking for Visual Odometry Resources in Simulation (ROS 2 + Gazebo)

General
1

Hi,

I’m working on a project that involves visual odometry, and I’m specifically looking to implement and test everything in simulation using ROS 2 (Preferably Humble or Jazzy) and Gazebo—no physical hardware or cameras involved.

I’d really appreciate it if anyone could share:

  1. Tutorials or guides for setting up visual odometry pipelines in ROS 2 with Gazebo
  2. Recommended simulation setups (e.g., camera plugins, world files, robot models)
  3. Open-source packages or repos that work well in simulation
  4. Any datasets or tools that help evaluate VO performance in a simulated environment

Kind regards,

Johanes

3 Likes
2

This is a very common question in the forum, you can search around and will find valuable answers, personally I think that this three repositories can give you hints on your journey:

It’s always a good practice to look at the official ROS2 documentation/tutorials as well:

If you’re a visual/video person, the Articulated Robotics have some good materials/tutorials that you can check out.

There are a plenty of materials on the internet, mostly outdated, keep in mind that the more reliable sources are the official repositories on Github. :slight_smile:

1 Like
3

You could try RTAB-Map for VO and SLAM

4

You could check out our open source mobile robot package - GitHub - blackcoffeerobotics/bcr_bot: A Gazebo-ROS AMR robot with differential drive, depth camera, IMU and 2D LiDAR

It supports a range of ROS2 distributions with multiple Gz distros (and ISAAC sim). We’ve used rtabmap for visual odometry in the past