The Concept of Remote Diagnostics of the Technical Condition of Vehicles During their Operation

Abstract

The technical condition of the vehicle directly affects the safety of people. With the increase in the number of cars, this problem becomes even more relevant. Today, checking the condition of the vehicle mostly takes place in the physical presence of a mechanic who reads data from the car's sensors with the help of scanners. The possibility of automating this process and providing the ability of remote access to data would significantly reduce the cost and speed up the detection of malfunctions.

This paper shows a concept of approach to the remote real time vehicle diagnostics. Data collection is possible using OBD2 protocol that allows performing a real time communication with vehicle ECUs. Even though a vehicle diagnostic system already has diagnostic trouble codes (DTCc) on its own they don’t necessarily catch all the edge cases when malfunctions occur. For this purpose we suggest using all available vehicle signals to then analyze and find potential anomalies using different methods for anomaly detection. Reconstruction-based anomaly detection includes training a model using a known normal (healthy) data to then recreate input data. Group based anomaly detection implies analyzing a group of different signals instead of analyzing them separately. Detected anomalies are then saved to a database on a remote server where users can always check them using a web page or a mobile application.

The proposed concept provides a modern approach to detect vehicle malfunctions. Even though car vendors don’t always follow the same standards, the future of vehicle diagnostics is looking bright. Most of the modern cars especially in the luxury segment are equipped with a sim card. This opens a window of other approaches to remote vehicle diagnostics where the scanner is not needed anymore. Vendors can have a custom OEM backend that receives all vehicle signals directly from a car.