Re: Tilted 3D map with floam

Posted by nugu_ on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Tilted-3D-map-with-floam-tp10380p10459.html

Hi,

thank you for your response. I also assumed that the slope in the map comes from the lidar not being perfectly aligned with gravity in the beginning. I tried to calculate the starting position with the imu of the ouster, but as you said the imu is very noisy and poorly calibrated. You said the transform between the imu and the lidar frame seems off. I found this github issue. There they say that the pointcloud published by the ouster is in the os_sensor frame, not in the os_lidar frame. The os_sensor frame marks the bottom of the ouster, not the actual lidar that captures the point cloud. May this be an issue here?

When I use your parameters, the localization gets completely lost after some driving and the robot starts drifting across the map. Also, the slope in the map I experience is greater than just an initial orientation offset to gravity I believe. As mentioned, the robot lifts of the ground by more than 60cm at the highest point, when exploring the building.

Another thing I noticed is that when I start my mapping/ localization node and don’t move at all, the output of the icp_odometry is not at 0, 0, 0 or around these values, which it should. It always has an offset of around 4 cm in x direction and around 10cm in z direction. I assume this comes from errors in my urdf model, so the transform between the sensor frame and the base_footprint frame does not match the real-life properties of the robot. However, after remeasuring the robot and adjusting the urdf, I was not able to fully eliminate this initial offset. Is there any other reason you can think of why this happens?

Thank you for your help.