Fast, Appearance-Invariant Navigation Transformer (FAINT) is a learned policy for vision-based topological navigation.
The
FAINT-Real model uses
Theia-Tiny-CDDSV as backbone, and was trained for 30 epochs with the ~1.2M samples of the datasets used in the GNM/ViNT papers.
This repo contains two versions of the trained model weights.
See the main GitHub
repo for details regarding input preprocessing, deployment with ROS2, and training.
Only dependency is Pytorch.
1import torch
2ckpt_path = 'FAINT-Real/model_torchscript.pt'
3model = torch.jit.load(ckpt_path)
Need to have the FAINT library installed.
1import torch
2from faint.common.models.faint import FAINT
3
4ckpt_path = 'FAINT-Real/model_pytorch.pt'
5state_dict = torch.load(ckpt_path)
6
7model = FAINT() # The weights in this repo correspond to FAINT initialized with the default arguments
8model.load_state_dict(state_dict)
1@article{suomela2025synthetic,
2 title={Synthetic vs. Real Training Data for Visual Navigation},
3 author={Suomela, Lauri and Kuruppu Arachchige, Sasanka and Torres, German F. and Edelman, Harry and Kämäräinen, Joni-Kristian},
4 journal={arXiv:2509.11791},
5 year={2025}
6}
