search

KTH-RPL / dufomap

[RA-L'24] DUFOMap: Efficient Dynamic Awareness Mapping
https://kth-rpl.github.io/dufomap
BSD 3-Clause "New" or "Revised" License
162 stars 17 forks source link

static areas remove too much #3

Closed coldbath closed 6 months ago

coldbath commented 6 months ago

Hi Dr.Zhang

First of all, thank you for sharing the data. I have run dufomap with twofloor and our own data and found that static areas such as floors and walls will also be remove, which with configs in assets/.

I did not find this phenomenon on Octomap, but the running time is very slow

Can this phenomenon be optimized by adjusting parameters?

Here is the screenshots, run with twofloor:

_dufomap_output_dynamic.pcd_ lQDPKHz13-zXEF3NA0XNBW6wWrCHyUfKrPQGFCp5rPwoAA_1390_837

_dufomap_outputstatic.pcd lQDPJwrp0MccB13NA0XNBZCwUjSqZzgmMRQGFCqHv8pOAA_1424_837

_octomapfgoutput.pcd lQDPJwolXW7pV13NAz3NBY2wPJGAMn0E0B0GFCqdmLrTAQ_1421_829

Best regards,

Cino

Kin-Zhang commented 6 months ago

could you run the code version in this repo?

coldbath commented 6 months ago

could you run the code version in this repo?

This is the result of this repo, and the phenomenon is similar lQDPJwFo6uiFBv3NAy3NBeGw1r1_i457pD8GFDXCi5STAA_1505_813 1713865928827

Kin-Zhang commented 6 months ago

Don't know why your result is different, but here is the one I run just now through the repo: image image

And here is the paper effect (no quantitative but qualitative): it's actually have some holes (the reason is because localization error and ground could be heavy) image

if possible could you provide your uname -a and lscpu?

coldbath commented 6 months ago

Don't know why your result is different, but here is the one I run just now through the repo: image image

And here is the paper effect (no quantitative but qualitative): it's actually have some holes (the reason is because localization error and ground could be heavy) image

if possible could you provide your uname -a and lscpu?

lQDPJx5akj65fw02zQLWsBkYv8hVvSc3BhRaPi4reQA_726_54 1713875470283

Kin-Zhang commented 6 months ago

didn't see any problem with your system and CPU info.

Could you again git pull, check config and run again with the latest commit? There are holes but as I attached from my side, should that much (the config file) or uncomment this line: std::cout<<config; and screenshot when you run the code.

Let us know if you still have problem with latest commit.

coldbath commented 6 months ago

didn't see any problem with your system and CPU info.

Could you again git pull, check config and run again with the latest commit? There are holes but as I attached from my side, should that much (the config file) or uncomment this line: std::cout<<config; and screenshot when you run the code.

Let us know if you still have problem with latest commit.

Previously, I downloaded the code directly, but this time I git clone the code. Perhaps this is the reason why the code is not the latest version.

I got the same result as you, 56318352 points (before is 44021782 points).

Although there are still exist some holes, maybe the reason is because localization error and ground could be heavy as you say.

I will continue to collect more data by livox-360 and test it. If there are any issues, I will provide feedback to you.

THX.

1713922491403

lQDPJxi2SwaDgWXNAxHNBPSwdY6VPhB8rRQGFRLslOM3AA_1268_785

Kin-Zhang commented 6 months ago

Great, also feel free to pull request if you want to contribute data.

one more word, for why octomap w GF doesn't have many holes mainly because it did ground segmentation as an independent step only on a single frame (The disadvantage of ground seg is that needed tune parameters for a different setup.)

Looks like the problem already was solved. open a new one if you have other questions.