Fast radar motion estimation with a learnt focus of attention using weak supervision
This paper is about fast motion estimation with scanning radar. We use weak supervision to train a focus of attention policy which actively down-samples the measurement stream before data association steps are undertaken. At training, we avoid laborious manual labelling by exploiting short-term sens...
| Main Authors: | , , , |
|---|---|
| 格式: | Conference item |
| 出版: |
IEEE
2019
|
| _version_ | 1826305075851558912 |
|---|---|
| author | Aldera, R De Martini, D Gadd, M Newman, p |
| author_facet | Aldera, R De Martini, D Gadd, M Newman, p |
| author_sort | Aldera, R |
| collection | OXFORD |
| description | This paper is about fast motion estimation with scanning radar. We use weak supervision to train a focus of attention policy which actively down-samples the measurement stream before data association steps are undertaken. At training, we avoid laborious manual labelling by exploiting short-term sensor coherence from multiple poses in the presence of an external ego-motion estimator (for example, wheel odometry). In this way, we generate copious annotated measurements which can be used for training a learning algorithm in a weakly-supervised fashion. We demonstrate the validity of the approach in the context of a Radar Odometry (RO) task, pre-filtering raw data with a popular image segmentation network trained as presented. We evaluate our system against 26 km of data collected in Central Oxford and show consistent motion estimation with greatly reduced radar processing times (by a factor of 2.36). |
| first_indexed | 2024-03-07T06:27:22Z |
| format | Conference item |
| id | oxford-uuid:f4c7ab3c-6b49-4b8f-b4dc-a83008552986 |
| institution | University of Oxford |
| last_indexed | 2024-03-07T06:27:22Z |
| publishDate | 2019 |
| publisher | IEEE |
| record_format | dspace |
| spelling | oxford-uuid:f4c7ab3c-6b49-4b8f-b4dc-a830085529862022-03-27T12:22:16ZFast radar motion estimation with a learnt focus of attention using weak supervisionConference itemhttp://purl.org/coar/resource_type/c_5794uuid:f4c7ab3c-6b49-4b8f-b4dc-a83008552986Symplectic Elements at OxfordIEEE2019Aldera, RDe Martini, DGadd, MNewman, pThis paper is about fast motion estimation with scanning radar. We use weak supervision to train a focus of attention policy which actively down-samples the measurement stream before data association steps are undertaken. At training, we avoid laborious manual labelling by exploiting short-term sensor coherence from multiple poses in the presence of an external ego-motion estimator (for example, wheel odometry). In this way, we generate copious annotated measurements which can be used for training a learning algorithm in a weakly-supervised fashion. We demonstrate the validity of the approach in the context of a Radar Odometry (RO) task, pre-filtering raw data with a popular image segmentation network trained as presented. We evaluate our system against 26 km of data collected in Central Oxford and show consistent motion estimation with greatly reduced radar processing times (by a factor of 2.36). |
| spellingShingle | Aldera, R De Martini, D Gadd, M Newman, p Fast radar motion estimation with a learnt focus of attention using weak supervision |
| title | Fast radar motion estimation with a learnt focus of attention using weak supervision |
| title_full | Fast radar motion estimation with a learnt focus of attention using weak supervision |
| title_fullStr | Fast radar motion estimation with a learnt focus of attention using weak supervision |
| title_full_unstemmed | Fast radar motion estimation with a learnt focus of attention using weak supervision |
| title_short | Fast radar motion estimation with a learnt focus of attention using weak supervision |
| title_sort | fast radar motion estimation with a learnt focus of attention using weak supervision |
| work_keys_str_mv | AT alderar fastradarmotionestimationwithalearntfocusofattentionusingweaksupervision AT demartinid fastradarmotionestimationwithalearntfocusofattentionusingweaksupervision AT gaddm fastradarmotionestimationwithalearntfocusofattentionusingweaksupervision AT newmanp fastradarmotionestimationwithalearntfocusofattentionusingweaksupervision |