Cartographer is a system that provides real-time simultaneous localization and mapping (SLAM) in 2D and 3D across multiple platforms and sensor configurations.
Apache License 2.0
7.04k
stars
2.24k
forks
source link
Current state of Continuing, Life-Long and Dynamic SLAM features #1814
[This issue is a replica of this one, but maybe this repository is more active]
We have been using the 2D cartographer and cartographer_ros package for our mobile robot for the past few months. Right now we are interested in three main features: Continuing mapping (1st mapping session, save, load, continue 2nd mapping session, refine/remap zones), Life-Long Mapping and Localization (run slam permanently with bounded memory and cpu load) and robustness/adaptability to changing environments (specifically for low/semi-static/slow dynamics).
After reviewing the available documentation and reading the relevant repository's issues, this is my understanding/questions of the current state of these three features:
Continuing Mapping: This tags indicate that this feature is implemented (/cartographer-project/cartographer_ros#1115 and /cartographer-project/cartographer_ros#1507), but it is confusing how this works and documentation is missing. Would you please indicate how to properly use this feature?
Life-Long Mapping and Localization: This tag and the milestone created indicate an initial interest in this feature (/cartographer-project/cartographer#197) back in 2017, but there isn't any documentation and it is unclear if it ever was ready for prime time. Is this currently implemented? If yes, could you please indicate how to use it? If not, what is the work done/left? Also, there are multiple discussions about a trimmer for removing data from overlapping sub-maps, I know this is related but I'm missing the bigger picture of the feature.
Changing environments: The Life-long feature (if implemented) could deal with this, but maybe this functionality is done (or could be) in a separate way. I couldn't find any information about this, so I would appreciate some clarification of the current capabilities.
I am asking all of this because I'm planning to work/improve these features for the following months, so I would like to know their current status.
Thank you very much for your time and consideration @wohe @MichaelGrupp
Hello!
[This issue is a replica of this one, but maybe this repository is more active]
We have been using the 2D cartographer and cartographer_ros package for our mobile robot for the past few months. Right now we are interested in three main features: Continuing mapping (1st mapping session, save, load, continue 2nd mapping session, refine/remap zones), Life-Long Mapping and Localization (run slam permanently with bounded memory and cpu load) and robustness/adaptability to changing environments (specifically for low/semi-static/slow dynamics).
After reviewing the available documentation and reading the relevant repository's issues, this is my understanding/questions of the current state of these three features:
Continuing Mapping: This tags indicate that this feature is implemented (/cartographer-project/cartographer_ros#1115 and /cartographer-project/cartographer_ros#1507), but it is confusing how this works and documentation is missing. Would you please indicate how to properly use this feature?
Life-Long Mapping and Localization: This tag and the milestone created indicate an initial interest in this feature (/cartographer-project/cartographer#197) back in 2017, but there isn't any documentation and it is unclear if it ever was ready for prime time. Is this currently implemented? If yes, could you please indicate how to use it? If not, what is the work done/left? Also, there are multiple discussions about a trimmer for removing data from overlapping sub-maps, I know this is related but I'm missing the bigger picture of the feature.
Changing environments: The Life-long feature (if implemented) could deal with this, but maybe this functionality is done (or could be) in a separate way. I couldn't find any information about this, so I would appreciate some clarification of the current capabilities.
I am asking all of this because I'm planning to work/improve these features for the following months, so I would like to know their current status. Thank you very much for your time and consideration @wohe @MichaelGrupp