hyye / lio-mapping

Implementation of Tightly Coupled 3D Lidar Inertial Odometry and Mapping (LIO-mapping)
https://sites.google.com/view/lio-mapping
GNU General Public License v3.0
898 stars 320 forks source link

Is the raw point cloud de-skewing be implemented? #7

Closed daijicheng closed 5 years ago

daijicheng commented 5 years ago

In the paper, the raw point cloud from lidar sensor is de-skewed before feature extraction, but in the code PointProcessor.cc I can not find this part, did I miss something?

hyye commented 5 years ago

In my implementation, the linear de-skewing is done by TransformToEnd in src/point_processor/PointOdometry.cc or src/imu_processor/Estimator.cc.

daijicheng commented 5 years ago

Thanks for your answer. I think extracting feature in the distorted point cloud will have negative effect in the quality of feature points, is this effect serious?

hyye commented 5 years ago

I don't think it will have serious effects on the feature extraction since the timestamps of consecutive points in one scan are very close. Thus, most of the features should still be preserved, except the points between the end and beginning of one scan.

daijicheng commented 5 years ago

For the feature only rely on local points, this makes sense, thank you.