Adopt Livox lidar

[qiqzhang]

Thanks for your wonderful job. When I adopt livox lidar with point time relative to first point time, header time use first lidar point time, but traj is drift than undistorted pointcloud. what should i do to adopt livox lidar

[qiqzhang]

I change some code to adopt livox，only lidar + imu run, like this,
success = state->get_interpolated_pose(lidar_inL->time + lidar_inL->pointcloud->points[i].curvature * 1e-6, RGtoIi, pIiinG);

if ((*it)->time + state->lidar_dt.at((*it)->id)->value()(0) + (*it)->pointcloud->points[(*it)->pointcloud->points.size() -1 ].curvature * 1e-6 > state->newest_clone_time()) {
it++;
continue;
}

the gap of lidar first timestamp with last timestamp about 100ms, i also try use get_interpolated_pose_linear, but also misalign.
1、use get_interpolated_pose function, raw_remove_motion_blur set false

2、use get_interpolated_pose_linear function, raw_remove_motion_blur set false

3、raw_remove_motion_blur set true，pointcloud will mismatch；

I try simulation with only lidar and set raw_remove_motion_blur true
roslaunch mins simulation.launch cam_enabled:=false lidar_enabled:=true the result is also mismatch, use get_interpolated_pose_linear is correct

if set raw_remove_motion_blur false,
roslaunch mins simulation.launch cam_enabled:=false lidar_enabled:=true the result is correct

anyone can

[muttistefano]

Hi @qiqzhang ,
Are you using the livox mid360 or another model?
I'm asking because the mid360 gives me some problems.
Thanks

[qiqzhang]

Hi @qiqzhang , Are you using the livox mid360 or another model? I'm asking because the mid360 gives me some problems. Thanks

I use models that same as mid360, undistorted pointclouds need to adopted to mid360, bit original code supposes every point has
fixed time gap, I tried to adopt mid360, but not work well, the bias is too large. @WoosikLee2510 Could you give us advices to get correctly results.

[WoosikLee2510]

Hi, @qiqzhang. Thanks for bringing up this issue. I will check to see what the problem is. Definitely, not working well with simulation is a bug issue.

[jeannotes]

hi, any results