OmniPath Audits Wheelchair Accessibility Using AI and LiDAR.

This paper introduces OmniPath, an innovative agentic framework designed to move beyond passive mapping to actively audit wheelchair accessibility in pedestrian environments. The system integrates network topology data from OpenStreetMap (OSM) with submeter precision aerial LiDAR data from USGS 3DEP, constructing a detailed 3D model of pathways. This allows OmniPath to virtually traverse routes and analyze surface characteristics at fine granularities, specifically measuring running slope, cross slope, and vertical discontinuities. OmniPath rigorously quantifies these physical friction points against ADA compliance standards, assigning a weighted severity score to categorize hazards from "Mild" to "Critical." The framework's reliability was validated through 200 physical ground truth field surveys, demonstrating strong diagnostic accuracy for high-severity hazards. By automating this micro-scale inspection, OmniPath effectively transforms static mapping data into a dynamic accessibility resource, anticipating challenges for wheelchair users before they encounter them in the real world.