Hi I'm using the VSLAM as a state estimator in nvblox
with D435i as camera, x86 with nvidia GPU(Ubuntu 20.04.5LTS, ROS2 foxy)
and using realsense vslam examples in launch folder of VSLAM and
https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_nvblox/blob/main/docs/tutorial-nvblox-vslam-realsense.md

but it seems after DP2.0 release update, the VSLAM always show "tracker is lost"

there are very few feature points, but as I use VSLAM release DP1.2, there are many feature point output in rviz

I've followed the trouble shooting suggestions:

Fast motion causing the motion blur in the frames.
Low-lighting conditions.
The wrong camerainfo is being published.

my stereo input is 60fps, and I use realsense D435i and splitted the camera image so there are no light dots in it
As to the camera info, I'm not sure what is the "right" camera info, what I only knew is the camera info can be seen via rostopic echo.

What I want to ask is that is there any difference for DP2.0 to DP1.2 which would affect the performance?

and is there any ways to tune the VSLAM aside from the trouble shooting? Or for the elbrus VSLAM, can I tune configurations such as feature point number like ORBSLAM3?

And is that involving D435’s IMU worth a try?

Thanks.