hyye
commented
4 years ago If there are too many lidar feature points, the processing time would be longer. Maybe consider reducing the window sizes or feature points. For configuration, the outdoor config can be a reference.
msg_time_delay: the time offset between lidar and IMU, for typical Velodyne data, buffer the IMU to the middle timestamp of the point cloud.
cutoff_deskew: turn off deskewing, when using corrected point cloud data, like KITTI.
init_window_factor: a factor makes the local window of initialization longer.
update_laser_imu: update the optimized states, just for debugging.
LIO becomes very slow in my large environment, to a point where I have to pause my bag file to give LIO time to catch up.
Do you have any advice on how to tune the parameters to keep things running smoothly or constrain processing time?
I would also be interested in some documentation on the following parameters:
msg_time_delaycutoff_deskewinit_window_factorupdate_laser_imu