Application of Outlier Detection for Volvo Trucks Safety Scoring: Classification of drivers based on driving behaviour and assign safety score using unsupervised learning and object detection.

Abstract

Improving safety and preventing accidents is a pressing concern in today’s growing demand for increasing transportation services. To ensure road safety protocols and minimize traffic-related incidents, Volvo Trucks has committed to evaluating and improving driver behaviour through advanced data analysis and machine learning techniques. This thesis explores the application of outlier detection methods to evaluate and improve safety scoring of Volvo truck drivers based on the driver’s behavior, braking and acceleration patterns, and contextual traffic conditions. The data from the truck’s Advanced Driver Assistance Systems, including brake pedal position, longitudinal acceleration, and longitudinal velocity, is used to examine the braking behaviour of the drivers during Pre-brake and Full-brake events from the CW-EB. These behaviors are then clustered using unsupervised learning and vector quantization techniques to classify them into different driving risk levels and assign safety scores. Additionally, YOLOv8, an object detection model, is introduced to determine whether the event was caused by the driver or the surrounding environment.