Indoor positioning has historically struggled where GPS fails. Fusion-DHL addresses this by merging wireless signal strength, smartphone motion sensors, and building blueprints — a hybrid method that overcomes the weakness of relying on any single technology alone.
The core problem
Wireless-based approaches depend on extensive databases and consume significant battery power while still producing meter-level errors. Motion sensor navigation works offline but drifts as measurement errors compound with each step.
Technical solution
The system aligns motion sensor trajectories against periodic WiFi position estimates, then a neural network refines the path against architectural features, repeating the process using the improved trajectory as anchoring points.
Real-world results
Testing in a shopping mall over 44 minutes reduced average WiFi positioning error from 12 meters to 5 meters, generating measurements far more frequently than conventional methods, with applications spanning space planning, transportation efficiency, and retail optimization.
