Modern vehicles generate vast amounts of operational data that reflect their health and performance. When processed correctly, this data can guide manufacturers and technicians in performing targeted repairs and predictive maintenance. This not only enhances vehicle safety but also mitigates repair costs through the enabling of timely interventions. Currently, the predominant fault-detection systems are based on static rules formulated by domain experts. These rules, however, can only capture a limited scope of anomaly patterns. This thesis explores the application of machine-learning techniques for anomaly-detection within vehicle data. Beginning with an introduction to anomaly detection and time series analysis, the study then examines various unsupervised anomaly detection algorithms, from both theoretical and practical perspectives.