Appearance-only SLAM at large scale with FAB-MAP 2.0
We describe a new formulation of appearance-only SLAM suitable for very large scale place recognition. The system navigates in the space of appearance, assigning each new observation to either a new or a previously visited location, without reference to metric position. The system is demonstrated pe...
| Main Authors: | , |
|---|---|
| Format: | Journal article |
| Language: | English |
| Published: |
2011
|
| _version_ | 1797099461648842752 |
|---|---|
| author | Cummins, M Newman, P |
| author_facet | Cummins, M Newman, P |
| author_sort | Cummins, M |
| collection | OXFORD |
| description | We describe a new formulation of appearance-only SLAM suitable for very large scale place recognition. The system navigates in the space of appearance, assigning each new observation to either a new or a previously visited location, without reference to metric position. The system is demonstrated performing reliable online appearance mapping and loop-closure detection over a 1000 km trajectory, with mean filter update times of 14 ms. The scalability of the system is achieved by defining a sparse approximation to the FAB-MAP model suitable for implementation using an inverted index. Our formulation of the problem is fully probabilistic and naturally incorporates robustness against perceptual aliasing. We also demonstrate that the approach substantially outperforms the standard term-frequency inverse-document-frequency (tf-idf) ranking measure. The 1000 km data set comprising almost a terabyte of omni-directional and stereo imagery is available for use, and we hope that it will serve as a benchmark for future systems. © The Author(s) 2011. |
| first_indexed | 2024-03-07T05:24:03Z |
| format | Journal article |
| id | oxford-uuid:dff20825-7689-479e-b869-bed8aee7be0c |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T05:24:03Z |
| publishDate | 2011 |
| record_format | dspace |
| spelling | oxford-uuid:dff20825-7689-479e-b869-bed8aee7be0c2022-03-27T09:43:02ZAppearance-only SLAM at large scale with FAB-MAP 2.0Journal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:dff20825-7689-479e-b869-bed8aee7be0cEnglishSymplectic Elements at Oxford2011Cummins, MNewman, PWe describe a new formulation of appearance-only SLAM suitable for very large scale place recognition. The system navigates in the space of appearance, assigning each new observation to either a new or a previously visited location, without reference to metric position. The system is demonstrated performing reliable online appearance mapping and loop-closure detection over a 1000 km trajectory, with mean filter update times of 14 ms. The scalability of the system is achieved by defining a sparse approximation to the FAB-MAP model suitable for implementation using an inverted index. Our formulation of the problem is fully probabilistic and naturally incorporates robustness against perceptual aliasing. We also demonstrate that the approach substantially outperforms the standard term-frequency inverse-document-frequency (tf-idf) ranking measure. The 1000 km data set comprising almost a terabyte of omni-directional and stereo imagery is available for use, and we hope that it will serve as a benchmark for future systems. © The Author(s) 2011. |
| spellingShingle | Cummins, M Newman, P Appearance-only SLAM at large scale with FAB-MAP 2.0 |
| title | Appearance-only SLAM at large scale with FAB-MAP 2.0 |
| title_full | Appearance-only SLAM at large scale with FAB-MAP 2.0 |
| title_fullStr | Appearance-only SLAM at large scale with FAB-MAP 2.0 |
| title_full_unstemmed | Appearance-only SLAM at large scale with FAB-MAP 2.0 |
| title_short | Appearance-only SLAM at large scale with FAB-MAP 2.0 |
| title_sort | appearance only slam at large scale with fab map 2 0 |
| work_keys_str_mv | AT cumminsm appearanceonlyslamatlargescalewithfabmap20 AT newmanp appearanceonlyslamatlargescalewithfabmap20 |