Universal Vehicle Self-Diagnosis: Predictive Component Monitoring

This program develops a universal, data-driven, real-time predictive self-diagnosis system for automotive components — applying the same fault-detection / integrity-monitoring methodology that secures GNSS-based navigation to the broader vehicle-system health monitoring problem. Vehicles produce dense time-series data streams from dozens of components; the goal is to forecast component failures before they occur, across vehicle makes and operational profiles.

Why it matters. The lab’s largest new external grant in the AV-safety direction. Conceptually it extends the integrity arc one layer outward — from “can we trust this position fix?” to “can we trust this vehicle’s components?” — using the same statistical-modelling tools (fault detection, error bounding, probabilistic graphical models).

Funding.

• HK Smart Traffic Fund PSRI/102/2410/PR — HK$6.19M external award plus PolyU matching, Li-Ta Hsu as Rep. Co-PI / Project Coordinator, 03/2026–02/2028 (largest single new external grant)