Update model card with paper details and authors

README.md

@@ -1,26 +1,47 @@
 ---
+pipeline_tag: robotics
 tags:
 - model_hub_mixin
 - pytorch_model_hub_mixin
-pipeline_tag: robotics
 ---
 
-PointGoal navigation policy trained with the FrodoBots8K dataset.
+# PointGoal Navigation Policy (FrodoBots8K)
 
-Presented in the paper [Data scaling for navigation in unknown environments](https://arxiv.org/abs/2601.09444)
+PointGoal navigation policy trained with the FrodoBots8K dataset, as presented in the paper [Data Scaling for Navigation in Unknown Environments](https://arxiv.org/abs/2601.09444).
 
-[Project page](https://lasuomela.github.io/navigation_scaling/) | [Code](https://github.com/lasuomela/NavigationScaling)
+[**Project Page**](https://lasuomela.github.io/navigation_scaling/) | [**Code**](https://github.com/lasuomela/NavigationScaling)
 
-## Details
+## Authors
+Lauri Suomela, Naoki Takahata, Sasanka Kuruppu Arachchige, Harry Edelman, Joni-Kristian Kämäräinen
 
-Architecture:
+## Model Description
+This model is an end-to-end, map-free visual navigation policy designed for sidewalk robots. It was trained using imitation learning on crowd-sourced data to achieve zero-shot navigation in unknown environments. The research demonstrates that data diversity (geographical variety) is significantly more important than data quantity for real-world generalization.
 
-MLP-BC with Theia encoder + MLP action head
+- **Task:** Zero-shot PointGoal navigation.
+- **Generalization:** Evaluated across 125 km of autonomous driving in four different countries.
+- **Key Finding:** Doubling geographical locations in training decreases navigation errors by ~15%, while adding more data from existing locations leads to rapid performance saturation.
 
-Dataset:
+## Details
 
-1024 total, 32H from 32 distinct geographical locations
+### Architecture
+- **Type:** MLP-BC
+- **Visual Encoder:** [Theia encoder](https://huggingface.co/theaiinstitute/theia-base-patch16-224-cddsv)
+- **Action Head:** MLP action head
 
-## Usage
+### Training Dataset
+- **Dataset:** FrodoBots8K (subset)
+- **Scale:** 1024 total trajectories (~32 hours of data) from 32 distinct geographical locations.
 
-See the Github repo
+## Usage
+The model is built for deployment on EarthRover Zero robots using ROS2. For detailed instructions on environment setup, training, and deployment, please refer to the [official GitHub repository](https://github.com/lasuomela/NavigationScaling).
+
+## Citation
+If you use this model or code in your research, please cite:
+```bibtex
+@misc{suomela2026data_scaling,
+  title={Data Scaling for Navigation in Unknown Environments},
+  author={Suomela, Lauri and Takahata, Naoki and Kuruppu Arachchige, Sasanka and Edelman, Harry and Kämäräinen, Joni-Kristian},
+  journal={arXiv:2601.09444},
+  year={2026},
+}
+```