Accessible Breakpoint Recognition of Tactile Pavement Based on Graph Neural Networks
A graph-neural-network method for detecting breaks in tactile paving networks and supporting safer assisted travel.
Abstract
This study frames tactile paving as a connected spatial network and uses graph neural networks to identify accessibility breakpoints. The method supports systematic diagnosis of interrupted routes, helping translate scattered street-level defects into an interpretable network for planning, maintenance, and assisted-travel services.
Tactile paving is modeled as a connected accessibility network.