RGB features without depth for loop closure detection and motion estimation

[alexk1976]

Hi @matlabbe, Our RGBD camera has large FOV in RGB than in Depth. We'd like to utilize RGB features even where Depth is not available. If we disable Mem/DepthAsMask and Vis/DepthAsMask will be features without depth used for loop-closure and motion estimation like in ORB3? Or we need to implement triangulation for such features and calculate depth for them first like it's done in Mono mode? Any recommendation for it? thank you

[matlabbe]

You would also need Mem/StereoFromMotion set to true, so that loop closures can be closed on rtabmap side. For rtabmap's VO, it requires some of the features with depth.