Provide pointcloud without depth image is block the path generation in current code.

[tdnet12434]

Hi.

First of all, Thank you for the amazing public work!

I have tried FUEL with the simulation provided in the code. I see it relied on a depth camera and projection to 3d pointcloud with the help of camera_pose/orientation provided by the VIO system.

But I want to try it with my stereo-imu setup. I already have pointcloud which is generated fast enough to feed the system. And odometry msg which have front-z, left-x, up-y. I know that simulation based on ROS front-x, left-y, up-z so I transform it in every Odomcallback. And posemsg I think it's already in z-front, x-left, y-down already, so it should be fine. However, after many tries, I can't make it work as the same when providing a depth image at all.

One thing I can observe is, when I use a depth image along with pointcloud topic (which is a bad depth image), I can see trajectory after trigger a goal topic by RVIZ. But if I provide only pointcloud and pose and disable depth image topic, I can see the map.

Is it possible to provide only pointcloud without depth image? Any advice would be appreciated. Thank you advanced!