Localisation in Dynamic Environments

Question

Localisation in Dynamic Environments

hardesh-peppermint opened this issue 2 years ago · 5 comments

hardesh-peppermint commented 2 years ago

Hi,
Thanks for sharing this project with the open-source community.
I had a question regarding localization in a dynamic environment. Have you had the chance to test this out dynamic environments such as factories and warehouses over time? If yes how was the performance of the solution?

TIA

Answer 1 · 2022-09-26T23:07:12.000Z

I haven't tested this framework in a dynamic environment.
My prior experience is that:

when the ration of dynamic objects is reasonable, just filter them out, the performance is still acceptable;
when ration is huge,it is much challenging

Answer 2 · 2022-09-27T08:18:20.000Z

Hi,
Thanks for the reply. This was helpful.
Have you seen any papers that consider deployment of the SLAM solution in dynamic environment?

Answer 3 · 2022-09-28T06:02:24.000Z

Hi,

just google "Dynamic Environment && SLAM", you will get plenty of papers.

Dynamic object tracking for slam is an example

Answer 4 · 2022-09-28T06:42:21.000Z

Hi,
Thanks for the reply.

I've seen some papers that use object detection using a camera and then remove dynamic objects from the PCL. Some direct pointcloud-based approaches use some kind of clustering to remove the dynamic objects.

Closing the issue.

Answer 5 · 2022-09-29T02:40:29.000Z

Right. You can also check our Mapping Module: https://github.com/PJLab-ADG/Livox-Mapping, where we use one camera for object removing in post-processing. Hope it could help you.