How are the results on KITTI

[wzey]

Hi, it is a very helpful work. I wonder how are your results on KITTI. I run mloam on kitti bag, but the number of keyframes is not equal to bag frames? I can't use kitti evaluation tool to plot path. Do you have tested on kitti?

[gogojjh]

Thanks for your interest. I have tested M-LOAM on some KITTI sequences, but have not evaluated it on the KITTI benchmark. I set the mapping frequency as 5Hz, making the output frame numbers are not equal to the groundtruth. You are recommended to modify some parameters and this M-LOAM on the KITTI.

[wzey]

Thanks for your replaying! I find that single_eval command = 'python2 $rpg_path/scripts/analyze_trajectory_single_mloam.py \ maybe you miss to upload that file? analyze_trajectory_single_mloam.py

[gogojjh]

I use this evaluation toolbox, but modify some codes (figure color, printing ...): https://github.com/uzh-rpg/rpg_trajectory_evaluation/blob/master/scripts/analyze_trajectory_single.py . You can download and use it for evaluation. The format is compatible. thanks.

[wzey]

Hi Thanks for uploading oxford ros bag file. But I find this bag is 2730 seconds and not been divided into 5 parts like your paper mentioned. Can you tell the start and end time in ros bag for this 5 parts?

[gogojjh]

You can use this equation to determine the start and end time: from S / 8281 2730 to E / 8281 2730. Five bags: S E ID 1 800 bag1 800 2000 bag2 2000 4900 bag3 4900 6500 bag4 6500 8281 bag5