rtabmap VS maplab

I used RTAB Map for my masterthesis as well. You have to use the cloud topics and disable all scan parameters. There is a masterfile with 800 lines of parameters that should help you out. https://github.com/introlab/rtabmap/blob/master/corelib/include/rtabmap/core/Parameters.h

RTABmap project homepage

Website: http://introlab.github.io/rtabmap/

I'd suggest RTABMAP.

For your application rtabmap would be a good starting point I think. (You don't really need ROS either for rtabmap) http://introlab.github.io/rtabmap/

check out http://introlab.github.io/rtabmap/ and https://opensource.google/projects/cartographer and https://docs.nvidia.com/isaac/isaac/doc/tutorials/assemble\_kaya.html

I've ported https://github.com/ethz-asl/maplab/releases/tag/1.3 to the TK-1 and have some OpenCV experience with the Jetson family but no specific camera driver experience. Are you looking for a long term hire or just this effort?

I haven't had a chance to look at the data in detail, but I assume you're talking about the flip-flopping of some clouds from one of the ANYmals? We use a centralized mapping framework called MapLab which tries to optimally align the maps of all robots and generate a single consistent map. Map optimization is totally out of my domain, but I suspect a few bad constraints caused the optimization to oscillate between two local minima. It gave the operator a heart attack at the time, but its a fairly common occurrence before the process is complete, and fortunately it all got worked out.