Re: rtabmap does not find loop closures

Posted by acp on
URL: http://official-rtab-map-forum.206.s1.nabble.com/rtabmap-does-not-find-loop-closures-tp6586p6591.html

matlabbe wrote

Based on the shared database, the drone traversed the same area but in different direction. As the point of view is different, loop closures cannot be detected. RTAB-Map  detects loop closures if the camera is close to an already visited location and if the camera point of view is similar (looking at the same direction).

Hi Mathieu, thank you for your answer. I have noticed what you have commented, that when RTAB-Map  detects a loop closure if the camera is close to an already visited location and  if the camera point of view is similar (looking at the same direction).  However, I have few questions, if you do not mind I would appreciate your replay  


1,- What type of local or global optimization to use for a better loop closure detection. I mean I am a bit confuse about them :)


TORO, GTSAM, SURF or g2o


So far I have $rosparam get   /zed/rtabmap/g2o/Optimizer   '0'

Which means: Optimization approach used: 0=TORO, 1=g2o and 2=GTSAM.

How about using the following parameters:

<"RGBD/OptimizeMaxError"     = "0,0"/>
<"g2o/Optimizer"  = "1"/>
<"RGBD/LoopClosureReextractFeatures" = "true"/>

2.- How close the camera field of view must be to a visited location.



3.-What depends to make a good map?


4.- To make a good map is it recommendable  to pass twice in the same direction and close to an existing map?