ARGOS

Abstract

Performance of wireless navigation systems degrade sharply in industrial environments dominated by metallic clutter and heavy multipath. The primary cause lies in how anchors are placed or selected. Methods that succeed in open spaces fail under severe non-line-of-sight (nLoS) and frequent layout changes, for instance, induced by moving forklifts or shifting shelves. We introduce Argos, a multimodal wireless digital twin that fuses visual and RF information to optimize anchor selection. Visual imagery reconstructs the 3D layout, updated continuously via existing surveillance cameras, while RF measurements capture material-specific attenuation and reflections. We show that layout priors alone are insufficient; combining them with RF optics yields a material-aware channel model that predicts range errors under severe nLoS. Argos adapts proactively to environmental changes, without the requirement of repeated RF calibration or retraining, sustaining sub-meter localization accuracy in dynamic scenes. We validate Argos in a 120 m² factory testbed spanning over 700 locations, where the digital twin is built from 5K RGB images and 0.4M UWB CIR samples. To our knowledge, this is the first system to exploit visual priors for adaptive orchestration of wireless navigation infrastructure.

Video Demonstration

Results

Impact: 3D reconstruction fidelity shapes localization performance.
Top, Left: Median localization error decreases with finer voxelization as PSNR improves, but accuracy gains saturate beyond 40 cm voxels (≈14 dB).
Across resolutions: Argos outperforms ArgosLO.
Top, Right: Segmentation accuracy across methods.
Bottom: Example reconstructions at 10, 20, 40, and 100 cm voxel resolutions.

Top-left: Effect of RSS threshold with 5 anchors.
At −74 dBm: Coverage drops to ≈40% uncovered while median ∆FMD improves to 7 taps, giving a practical balance.
Top-right: Effect of anchor count at −90 dBm.
Observation: Largest gains occur from 4 to 5 anchors; beyond 5, coverage gains and ∆FMD reduction are marginal while ToA-TWR overhead grows.
Bottom: ∆FMD heatmaps for representative anchor placements.

Overview: Effect of layout dynamics on recomputation latency.
Left: Original voxel map.
Middle: Changed map with displaced robotic arms (green) – 6% voxel change.
Right: ∆FMD cache recomputation latency grows with voxel churn: ≈2 s for 2% updates and ≈7 s for 10%.

Result: Under a 5-anchor budget, Argos yields a median localization error of 0.6 m, compared to 1.8 m (ArgosLO) and 2.97 m (GDoP).

BibTeX

BibTex Will Be Added Soon