search

RobustFieldAutonomyLab / LeGO-LOAM

LeGO-LOAM: Lightweight and Ground-Optimized Lidar Odometry and Mapping on Variable Terrain
BSD 3-Clause "New" or "Revised" License
2.4k stars 1.12k forks source link

Terrible performance on the dataset you shared using LOAM #110

Closed LiShuaixin closed 5 years ago

LiShuaixin commented 5 years ago

Hi, @TixiaoShan . I want to test LOAM with the dataset you shared on your Google Cloud to compare its performance with LeGO-LOAM as you did in your paper.

I found that LOAM works terribly with your dataset as long as it makes a large turn. But I also found that LOAM generated acceptable results in your paper. Would you please give me a hint to test LOAM with your dataset? I just run LOAM and bag file in two different ternimals using roslaunch and rosbag tool.

Scrumantic commented 5 years ago

I've also done some benchmarking and used LOAM for a while before discovering LeGO-LOAM. The only way I managed to process large and complex bags was to decrease the rate which rosbag plays. I don't know if that helps but if you add -r 0.10 (10% speed) to your rosbag play command your results might improve a lot. (sorry if this isn't helpful)

LiShuaixin commented 5 years ago

@Scrumantic Hi, I tried another version of LOAM named A-LOAM, which uses some open source library like eigen and ceres to replace cv::Mat to improve system operation efficiency. I seems much better.

stale[bot] commented 5 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.