This is the official implamentation of VIUNet from VIUNet: Deep Visual–Inertial–UWB Fusion for Indoor UAV Localization (IEEE ACCESS'23)
The enviornment can be installed via conda
conda env create -f environment.yml
conda activate VIUNet
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
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.
(Not yet ready in formating) We present a real world dataset, available here