imLabNTU / VIUNet

VIUNet: Deep Visual–Inertial–UWB Fusion for Indoor UAV Localization (IEEE ACCESS'23)
MIT License
14 stars 1 forks source link

VIUNet

This is the official implamentation of VIUNet from VIUNet: Deep Visual–Inertial–UWB Fusion for Indoor UAV Localization (IEEE ACCESS'23)

Quickstart

Enviornment

The enviornment can be installed via conda

conda env create -f environment.yml
conda activate VIUNet

Dataset formating

The dataset should be formated as in kitti format, you can do so with following command

python preprocess.py --dir ../euroc --output_dir ../test  --type Euroc

Then, run the UWB simulation to generate the UWB data

python simulate.py --dir ../test --type Euroc

Training

First Download the pretrained model from here and unzip it into the ./pretrained folder.

cd pretrained
wget 'https://www.dropbox.com/s/7qgncgmwqlfc51t/euroc_test.tar.xz?dl=0'
tar -xvf euroc_test.tar.xz
cd ..

Then run the following command to train the model

python train.py <path/to/dataset>

In default, the result will be saved at ./results/checkpoints folder, tesnorboard log will be saved at ./runs folder.

Testing

Realworld Dataset

(Not yet ready in formating) We present a real world dataset, available here