Dynamic Reconstruction and VSLAM with ZED2i

Autonomous Machines Robotics - Isaac Isaac ROS
, , , ,
1

Hello,

Starting from version 2.1.0 of the ISAAC ROS packages, I have been trying to perform Dynamic Reconstruction (without human recognition) using ISAAC ROS Nvblox, ISAAC ROS VSLAM, and a ZED2i camera. I noticed that there is no dedicated version for these elements, as the ZED2i already provides a pose topic. However, using the pose from ISAAC ROS VSLAM allows for the robot to be relocalized using the /load_map action.

Due to this, I have been attempting to integrate ZED2i, VSLAM, and Nvblox within the .launch file for dynamic reconstruction, initially designed for Realsense. Unfortunately, I have not been successful so far.

Could you advise on the correct way to launch these three nodes together?

PD: does the human reconstruction from Nvblox also recognize other dynamics elements?

Thank you!

2

Hi @jim.fabian

If you want to try the dynamic reconstruction, you can use this demo:

https://nvidia-isaac-ros.github.io/concepts/scene_reconstruction/nvblox/tutorials/tutorial_isaac_sim.html#reconstruction-with-dynamic-scene-elements

This setup available for Isaac Sim or for Realsense enables the dynamic reconstruction mode.

About the ZED camera is currently not supported, but will be live again when the drivers become available for Jetpack 6.0.

https://nvidia-isaac-ros.github.io/concepts/scene_reconstruction/nvblox/technical_details.html#people-reconstruction

You can use and change the model from Isaac ROS Image segmentation node to detect other elements

https://nvidia-isaac-ros.github.io/repositories_and_packages/isaac_ros_image_segmentation/index.html

3

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.