|
Administrator
|
Hi,
In the last NodeData contained in MapData, we have the lidar so the server doesn't have to subscribe to lidar directly. However, for inter loop closures between the robot, the visual words would be used first, then lidar can be used to refine the constraint (like done internally to a single rtabmap).
MapData contains also all poses and constraints (intra loop closures). The server could indeed add them all to its global graph. The server would only try to detect loop closures between the robots.
For comparison of the approaches, if you have images to compare it would help as I cannot re-read all the old posts.
cheers,
Mathieu
|
Locked
