THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments
Visual object recognition in unseen and cluttered indoor environments is a challenging problem for mobile robots. This study presents a 3D shape and color‐based descriptor, TOPS2, for point clouds generated from red green blue‐depth (RGB‐D) images and an accompanying recognition framework, THOR2. Th...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-04-01
|
| Series: | Advanced Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/aisy.202400539 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849763447158341632 |
|---|---|
| author | Ekta U. Samani Ashis G. Banerjee |
| author_facet | Ekta U. Samani Ashis G. Banerjee |
| author_sort | Ekta U. Samani |
| collection | DOAJ |
| description | Visual object recognition in unseen and cluttered indoor environments is a challenging problem for mobile robots. This study presents a 3D shape and color‐based descriptor, TOPS2, for point clouds generated from red green blue‐depth (RGB‐D) images and an accompanying recognition framework, THOR2. The TOPS2 descriptor embodies object unity, a human cognition mechanism, by retaining the slicing‐based topological representation of 3D shape from the TOPS descriptor (IEEE Trans. Robot. 2024, 40, 886) while capturing object color information through slicing‐based color embeddings computed using a network of coarse color regions. These color regions, analogous to the MacAdam ellipses identified in human color perception, are obtained using the Mapper algorithm, a topological soft‐clustering technique. THOR2, trained using synthetic data, demonstrates markedly improved recognition accuracy compared to THOR, its 3D shape‐based predecessor, on two benchmark real‐world datasets: the OCID dataset capturing cluttered scenes from different viewpoints and the UW‐IS Occluded dataset reflecting different environmental conditions and degrees of object occlusion recorded using commodity hardware. THOR2 also outperforms baseline deep learning networks and a widely used Vision Transformer adapted for RGB‐D inputs trained using synthetic and limited real‐world data on both the datasets. Therefore, THOR2 is a promising step toward achieving robust recognition in low‐cost robots. |
| format | Article |
| id | doaj-art-e1c14cc6a9be449f8a05fd4ec7551d38 |
| institution | DOAJ |
| issn | 2640-4567 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Wiley |
| record_format | Article |
| series | Advanced Intelligent Systems |
| spelling | doaj-art-e1c14cc6a9be449f8a05fd4ec7551d382025-08-20T03:05:24ZengWileyAdvanced Intelligent Systems2640-45672025-04-0174n/an/a10.1002/aisy.202400539THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen EnvironmentsEkta U. Samani0Ashis G. Banerjee1Department of Mechanical Engineering University of Washington Seattle WA 98195 USADepartment of Mechanical Engineering University of Washington Seattle WA 98195 USAVisual object recognition in unseen and cluttered indoor environments is a challenging problem for mobile robots. This study presents a 3D shape and color‐based descriptor, TOPS2, for point clouds generated from red green blue‐depth (RGB‐D) images and an accompanying recognition framework, THOR2. The TOPS2 descriptor embodies object unity, a human cognition mechanism, by retaining the slicing‐based topological representation of 3D shape from the TOPS descriptor (IEEE Trans. Robot. 2024, 40, 886) while capturing object color information through slicing‐based color embeddings computed using a network of coarse color regions. These color regions, analogous to the MacAdam ellipses identified in human color perception, are obtained using the Mapper algorithm, a topological soft‐clustering technique. THOR2, trained using synthetic data, demonstrates markedly improved recognition accuracy compared to THOR, its 3D shape‐based predecessor, on two benchmark real‐world datasets: the OCID dataset capturing cluttered scenes from different viewpoints and the UW‐IS Occluded dataset reflecting different environmental conditions and degrees of object occlusion recorded using commodity hardware. THOR2 also outperforms baseline deep learning networks and a widely used Vision Transformer adapted for RGB‐D inputs trained using synthetic and limited real‐world data on both the datasets. Therefore, THOR2 is a promising step toward achieving robust recognition in low‐cost robots.https://doi.org/10.1002/aisy.202400539human‐inspired perceptionsmobile robotsRGB‐D object recognitionstopological learning |
| spellingShingle | Ekta U. Samani Ashis G. Banerjee THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments Advanced Intelligent Systems human‐inspired perceptions mobile robots RGB‐D object recognitions topological learning |
| title | THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments |
| title_full | THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments |
| title_fullStr | THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments |
| title_full_unstemmed | THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments |
| title_short | THOR2: Topological Analysis for 3D Shape and Color‐Based Human‐Inspired Object Recognition in Unseen Environments |
| title_sort | thor2 topological analysis for 3d shape and color based human inspired object recognition in unseen environments |
| topic | human‐inspired perceptions mobile robots RGB‐D object recognitions topological learning |
| url | https://doi.org/10.1002/aisy.202400539 |
| work_keys_str_mv | AT ektausamani thor2topologicalanalysisfor3dshapeandcolorbasedhumaninspiredobjectrecognitioninunseenenvironments AT ashisgbanerjee thor2topologicalanalysisfor3dshapeandcolorbasedhumaninspiredobjectrecognitioninunseenenvironments |