# Robot Localization Using a Learned Keypoint Detector and Descriptor with a Floor Camera and a Feature Rich Industrial Floor

arXiv: https://arxiv.org/abs/2504.03249 · HTML: https://mlsanity.com/paper/2504.03249/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 8 of its 25 references that are in the map.
- First posted on arXiv 4 Apr 2025.

## Builds on (most cited first, 8 of 8)

- [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://mlsanity.com/paper/1505.04597/) (2015)
- [ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras](https://mlsanity.com/paper/1610.06475/) (2016)
- [Multi-Similarity Loss with General Pair Weighting for Deep Metric Learning](https://mlsanity.com/paper/1904.06627/) (2019)
- [Working hard to know your neighbor's margins: Local descriptor learning loss](https://mlsanity.com/paper/1705.10872/) (2017)
- [SIFT Meets CNN: A Decade Survey of Instance Retrieval](https://mlsanity.com/paper/1608.01807/) (2016)
- [High-Precision Localization Using Ground Texture](https://mlsanity.com/paper/1710.10687/) (2017)
- [Features for Ground Texture Based Localization -- A Survey](https://mlsanity.com/paper/2002.11948/) (2020)
- [Ground Texture Based Localization Using Compact Binary Descriptors](https://mlsanity.com/paper/2002.11061/) (2020)

## Built on by (most cited first, 1 of 1)

- [Graph-based Robot Localization Using a Graph Neural Network with a Floor Camera and a Feature Rich Industrial Floor](https://mlsanity.com/paper/2508.06177/) (2025)

## Abstract (arXiv)

The localization of moving robots depends on the availability of good features from the environment. Sensor systems like Lidar are popular, but unique features can also be extracted from images of the ground. This work presents the Keypoint Localization Framework (KOALA), which utilizes deep neural networks that extract sufficient features from an industrial floor for accurate localization without having readable markers. For this purpose, we use a floor covering that can be produced as cheaply as common industrial floors. Although we do not use any filtering, prior, or temporal information, we can estimate our position in 75.7 % of all images with a mean position error of 2 cm and a rotation error of 2.4 %. Thus, the robot kidnapping problem can be solved with high precision in every frame, even while the robot is moving. Furthermore, we show that our framework with our detector and descriptor combination is able to outperform comparable approaches.