Problems with other dataset about lio-sam HOT 14 CLOSED

I have the similar problem.The bag(casual_wakl.bag) can perform well when running LIO-SAM, but failed when i want to compare the result when running LEGO-LOAM. I have tried to use just /points_raw when running LEGO but also failed.

from lio-sam.

@zwmisdu What I am showing now is that it cannot be run in casual_wake.bag, and it fails. May I ask how you set it up and what parameters need to be changed

from lio-sam.

@zwmisdu What I am showing now is that it cannot be run in casual_wake.bag, and it fails. May I ask how you set it up and what parameters need to be changed

lio-sam performs well with casual_walk.bag. Use "git pull" to make sure your code is newest. You don't need to change anything.

from lio-sam.

@zwmisdu What I am showing now is that it cannot be run in casual_wake.bag, and it fails. May I ask how you set it up and what parameters need to be changed

just like the guy above said, i did not change anything

from lio-sam.

I have the similar problem.The bag(casual_wakl.bag) can perform well when running LIO-SAM, but failed when i want to compare the result when running LEGO-LOAM. I have tried to use just /points_raw when running LEGO but also failed.

@zwmisdu
LEGO won’t work with all the dataset provided here. It needs the lidar to be placed horizontally and a ground in the scan all the time.

from lio-sam.

Thank your for your great job! Since LIO-SAM and LIO-mapping both use lidar and imu data to run slam. I tested your code with dataset(slow2.bag) provided by LIO-mapping and tested LIO-mapping with your dataset(casual_walk.bag). Though I modified extrinsic rotation and translation, both of two tests failed(generating meaningless map). What's the problem? Is it because your imu data is used in a different way?

@getupgetup It’s highly likely an IMU extrinsic problem.

from lio-sam.

@getupgetup
Found the problem. The bags from LIO-mapping uses a 6-axis IMU. As mentioned in the readme, LIO-SAM only works with a 9-axis IMU currently.

from lio-sam.

@zwmisdu What I am showing now is that it cannot be run in casual_wake.bag, and it fails. May I ask how you set it up and what parameters need to be changed

Maybe your GTSAM library was not install correctly, I try to rebuild GTSAM with author's method described in README, then work great on casual_wakl.bag.

from lio-sam.

@zwmisdu Yes, it has been solved

from lio-sam.

@getupgetup
Found the problem. The bags from LIO-mapping uses a 6-axis IMU. As mentioned in the readme, LIO-SAM only works with a 9-axis IMU currently.

It might be part of reason. Since LIO-mapping can't run well with casual_walk.bag, I think there'are some other reasons(extrinsic calibration maybe). Would you please have a look at LIO-mapping and find the difference between them if you have time? Thank you!

from lio-sam.

@getupgetup
Most of my dataset features aggressive motion, without compensation from gimbal or car. So other methods may struggle with it. I think part of the reason is because of the optimization pipeline of lio-mapping, which is adapted from VINS-Mono. I ran a dataset with calibrated extrinsic, the result of VINS-Mono is poor also.

from lio-sam.

@getupgetup
Most of my dataset features aggressive motion, without compensation from gimbal or car. So other methods may struggle with it. I think part of the reason is because of the optimization pipeline of lio-mapping, which is adapted from VINS-Mono. I ran a dataset with calibrated extrinsic, the result of VINS-Mono is poor also.

Thanks! By the way, is there anyway to generate IMU orientation data manually if my IMU data is only 6-axis? What did you do to get the orientation data?

from lio-sam.

@getupgetup

I don't think there is such a way to do that. My IMU outputs orientation data by default.

from lio-sam.

@getupgetup

I don't think there is such a way to do that. My IMU outputs orientation data by default.

Thanks~

from lio-sam.