good work about kitti-lego-loam

[xuwuzhou]

by the way ,I use the tool "kitti2bag" to transform the raw kitti data to bag

[Mitchell-Lee-93]

Hello xuwuzhou
This is the result of kitti-lego-loam of 00 sequence with only lidar data(no imu, no loop closure)
And I haven't made any changes to the IMU part.
Would you try without IMU and see what you get?

[xuwuzhou]

Hello,Mitchell-Lee-93
Of course I will,but I have a question,if you test the 00 sequence with only lidar data,the IMU part in the lego-loam should be deleted,isn't it?

[Mitchell-Lee-93]

you can play only Velodyne topic with "rosbag play *.bag --clock --topic /kitti/velo/pointcloud"
(It differs by your bag file's topic name)

[xuwuzhou]

Thanks for your reply!
I know what you mean,but I wonder that if there is no IMU data put in, the IMU part in the source code will stay static when running the repo(=if the IMU part infuence the final result without IMU data in).

[xuwuzhou]

Hello, @Mitchell-Lee-93
Just now,I tested the repo with 00 sequence without IMU(use the kittibag which you provided to generate a bag),but the result is worse:

Your result is really good ,just with a little bias;
the tools is kittibag,and directory structrue is as follows:

and I have change the params in the launch file:

After that ,I run the repo,and I play the bag with order "rosbag play 00.bag --clock".
Do you know where I made a mistake?
Best regards!

[Mitchell-Lee-93]

Did you maybe speed up the rosbag play factor with "rosbag play -r" command? I tried to speed up the topic with it, but I got a distorted trajectory like yours. If it's not the problem, I'm not sure what's the problem for now

[xuwuzhou]

Hello @Mitchell-Lee-93
I didn't speed up the rosbag play factor;
And when I play the bag which the kittibag generated,something wrong happens to the rviz:

but in fact it can put out result in result.txt as mentioned;
Just now,the author of lego-loam replied that the comment which he made in issue#12 isn't right,do you know what should be changed?

[xuwuzhou]

Just now,I used the tool "kitti2bag" to generate a bag through raw data,
and use the order"play **.bag --clock --topic --topic /kitti/velo/pointcloud"
the result didn't get better:

it seems that with or without the order"--topic /kitti/velo/pointcloud" don't make any change,
But your result is really good ,do you make any change on the source code?

[Mitchell-Lee-93]

I believe since kittibag doesn't generate TF topic, it's like that only when you use Rviz but result is still same. And for --topic, if you've used kittibag, there is no IMU topic so it's normal that there is no difference. For the difference of the result, would you try my kitti-A-loam repository? Since it's using optimization, I guess the results might be a bit different from computer to computer. If you get different result of my kitti-A-loam result, that might be the problem

[xuwuzhou]

Thanks for your reply;
Yesterday,I tried your kitti-A-loam,the result is same as you ;
I know what happened;
Because I used raw data to generate a bag before,after I used the distorted data,the result gets better

[Mitchell-Lee-93]

It's good to know that you've solved the problem!