Researchers Kushal Khemani and Anjum Nazir Qureshi have introduced an artificial intelligence architecture that predicts connected vehicle maintenance needs by fusing external environmental signals with internal vehicular diagnostics.
Telemetry is collected from the vehicle controller area network via standard on-board diagnostics interfaces, capturing nine internal mechanical metrics including engine coolant temperature, manifold absolute pressure, vehicle speed, fuel trim, tire pressure, battery state, and brake wear indicators.
Operating exclusively on internal mechanical telemetry yielded an F1-score of 0.807, representing a 0.049-point deficit compared to full contextual fusion.
Conversely, removing internal mechanical signals caused performance to plummet by 0.201 to 0.655, confirming that internal mechanical measurements remain the indispensable core of failure prediction.
AI-driven predictive maintenance for connected vehicles using environmental context integration evaluated through simulation benchmarking and field validation.