UWB Indoor Positioning and Tracking (UWB-Indoor)

UWB positioning and tracking dataset with channel impulse response (CIR) measurements across four indoor environments (8 anchors + 1 mobile tag) for range-based localization. WavesFM task id: uwb-indoor (dataset class UWBIndoor).

Official links

• Raw data: https://zenodo.org/records/7629141
• DOI: https://doi.org/10.5281/zenodo.7629141
• Code: https://github.com/KlemenBr/uwb_positioning.git

Tasks supported

• Localization / tracking (Mean Localization Error in meters)

Targets

• Continuous (x, y, z) coordinates (regression); no class labels.

Split

• 80/20 random split (train/val created during training unless a validation set is provided).

Environment used

• Benchmarks on the website use environment0.

Preprocessing

1. Select an environment folder (e.g., environment0) and load data.json measurements.
2. Filter to anchors A1-A8 and channels ch1, ch2, ch3, ch4, ch5, ch7 with complete samples.
3. For each location/channel/index, stack anchor CIRs into (2, A, L) real/imag tensors and record (x, y, z) tag coordinates (the loader uses (x, y) only).
4. Write HDF5 datasets (cir, location, channel) and store global mean/std for real/imag plus loc_min/loc_max.

Script: preprocess_uwb_loc.py

python preprocessing/preprocess_uwb_loc.py \
  --data-path <UWB_ROOT> \
  --environment environment0 \
  --output <UWB_ROOT>/environment0.h5

Metric

• Mean Localization Error (meters) on the test split.

Citation

@dataset{bregar2023uwb,
  title = {UWB Positioning and Tracking Data Set},
  author = {Bregar, Klemen},
  publisher = {Zenodo},
  year = {2023},
  doi = {10.5281/zenodo.7629141}
}