Re: Errors on icp odometry

Posted by Mikor on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Errors-on-icp-odometry-tp7546p7608.html

Hi Mathieu,

I think I fixed the pointcloud, here is the bag, I attach the full bag this time. It is a scan of a nice building in Barcelona and I did not want it to go to waste. (I merged every two messages of the topic using the point_cloud_assembler.)

Anyway, I am trying to do for starters only icp odometry with it using the following parameters.

$ roslaunch rtabmap_ros rtabmap.launch       use_sim_time:=true    depth:=false    subscribe_scan_cloud:=true    frame_id:=velodyne    scan_cloud_topic:=/pointcloud2    scan_cloud_max_points:=131072     icp_odometry:=true    approx_sync:=false    args:="-d  \
      --RGBD/CreateOccupancyGrid false \
      --Rtabmap/DetectionRate 2 \
      --Odom/ScanKeyFrameThr 0.8 \
      --OdomF2M/ScanMaxSize 10000  \
      --OdomF2M/ScanSubtractRadius 0.5   \
      --Icp/PM true \
      --Icp/VoxelSize 0.5   \
      --Icp/MaxTranslation 2   \
      --Icp/MaxCorrespondenceDistance 1.5 \
      --Icp/PMOutlierRatio 0.7 \
      --Icp/Iterations 10 \
      --Icp/PointToPlane true \
      --Icp/PMMatcherKnn 20 \
      --Icp/PMMatcherEpsilon 5 \
      --Icp/Epsilon 0.0001 \
      --Icp/PointToPlaneK 5 \
      --Icp/PointToPlaneRadius 0 \
      --Icp/CorrespondenceRatio 0.01 \
      --Icp/MaxTranslations 5 \
      --Icp/MaxRotations 1.5"

But it fails miserably, the course of odometry goes wild and finally drifts away. And it makes me thinking how can it go so bad when at most of the kitti datasets I run with the same parameters it worked good enough. Anyway I turned the point to plane off and this odometry is the result.




Also the pointcloud is disappointing. Do you have any suggestions..?