EdgeLPR: On the Deep Neural Network trade-off between Precision and Performance in LiDAR Place Recognition
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910188145999872 |
|---|---|
| author | Serio, Pierpaolo Wang, Hetian Wei, Zixiang Infantino, Vincenzo Gentilini, Lorenzo Pollini, Lorenzo Donzella, Valentina |
| author_facet | Serio, Pierpaolo Wang, Hetian Wei, Zixiang Infantino, Vincenzo Gentilini, Lorenzo Pollini, Lorenzo Donzella, Valentina |
| contents | Place recognition is essential for long-term autonomous navigation, enabling loop closure and consistent mapping. Although deep learning has improved performance, deploying such models on resource-constrained platforms remains challenging. This work explores efficient LiDAR-based place recognition for EdgeAI by leveraging Bird's Eye View representations to enable lightweight image-based networks. We benchmark representative architectures without aggregation heads using a unified descriptor scheme based on global pooling and linear projection, and evaluate performance under FP32, FP16, and INT8 quantization. Experiments reveal trade-offs between accuracy, robustness, and efficiency: FP16 matches FP32 with lower cost, while INT8 introduces architecture-dependent degradation. Overall, the presented results are a strong basis for future research on 'use-case'-aware quantisation of Neural Networks for Edge deployment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_02275 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | EdgeLPR: On the Deep Neural Network trade-off between Precision and Performance in LiDAR Place Recognition Serio, Pierpaolo Wang, Hetian Wei, Zixiang Infantino, Vincenzo Gentilini, Lorenzo Pollini, Lorenzo Donzella, Valentina Computer Vision and Pattern Recognition Artificial Intelligence Robotics Place recognition is essential for long-term autonomous navigation, enabling loop closure and consistent mapping. Although deep learning has improved performance, deploying such models on resource-constrained platforms remains challenging. This work explores efficient LiDAR-based place recognition for EdgeAI by leveraging Bird's Eye View representations to enable lightweight image-based networks. We benchmark representative architectures without aggregation heads using a unified descriptor scheme based on global pooling and linear projection, and evaluate performance under FP32, FP16, and INT8 quantization. Experiments reveal trade-offs between accuracy, robustness, and efficiency: FP16 matches FP32 with lower cost, while INT8 introduces architecture-dependent degradation. Overall, the presented results are a strong basis for future research on 'use-case'-aware quantisation of Neural Networks for Edge deployment. |
| title | EdgeLPR: On the Deep Neural Network trade-off between Precision and Performance in LiDAR Place Recognition |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2605.02275 |