my set up is a intel realsense 435i and two luxonis oak-d pro’s (I have a quadruped and the the intel realsesne is facing forward at the head of the robot, and the two luxonis cameras are midway between the body of the robot facing left and right)
In the link I included, it tells you what topics nvblox ros subscribes to, but it doesnt say if the images should all be the same resolution.
can you tell me if we should leave all cameras at full resolution, or if we should downsample them. also does it matter if all images are the same resolution? other frameworks require this. also it states that the value of “I”, for the camera{I}… topics, can only be 0-3, so the maximum amount of cameras is 4? I thought the documentation said many more cameras could be used. so is this limit just on the specific implementation for the nvblox_node for ros?
also, if I have 3 cameras, do I need to sync the data to be published at the same time? or will nvblox handle the data by the cameras time stamps to update the map according to the timestamp and therefore syncing the cameras to publish at the same time is not needed? (in other words, is syncing the camera capture of data an optimization rather than a necessity? as time stamps will allow for the proper placement of the data?)
is the only way to really use the modules is to read the source code and to find out what is going on myself? it seems like the ros nvblox package is very peculiar to the specific examples and not that modular by a few tweaks. even the config files aren’t explained, and if the yaml config files are dependent on the implementation of the ros package, id have to go in the source code and look at this myself. so is the best thing to edit these files, or jus use the python/c++ nvblox package and make my own ros package from it? the documentation doesn’t really tell much about how it works to easily plug and play my own setup.