Andrew James Roberts, the PhD student of the Department of Software Science, will defend his PhD thesis “Cybersecurity Testing and Attack Propagation Analysis of Autonomous Driving Software” (”Autonoomse sõiduki juhtimistarkvara küberturvalisuse testimine ja rünnakute leviku analüüs”), on June 25, 2025 starting at 16:00 (UTC +3 / Eastern European Time (EET)). The defense will take place in room ICT-701 (ICT building of TalTech, Akadeemia tee 15a, Tallinn) and can be also followed via Zoom.
Autonomous Driving (AD) software relies on input from telemetry of diverse sensors (LiDAR, camera, radar, inertial measurement unit, etc.) to create environmental perception and localisation for navigational planning and motion-control. Integrity and availability of sensing input are critical to ensuring the robustness of autonomous control decisions. As the system architecture underpinning AD is developing, there is a preponderance of challenges for security. Primarily, there is a lack of cybersecurity testing frameworks, methods and toolsets aligned with the considerations of safety validation testing and control system design. In this thesis, these problems are addressed within 3 areas of need for the community, attack models to semantic and system-level components, methods for evaluating and fingerprinting attack behaviour, and tools for cybersecurity testing. New attack models are developed to sensing data communication and physical attack models to sensors. To evaluate these attacks, a testing tool-chain approach is developed, which integrates diverse digital-twin and real-world testbeds. To address the problem of a lack of generalised vulnerability testing methods applicable to diverse AD systems and available tools for testing, the research contributes an overall methodology for vulnerability testing and analysis and tools for fuzzing and penetration testing. The capability of the methodology and tools are illustrated through the discovery of vulnerabilities in popular AD software such as Open-Planner planning and navigation module, NDT-matching based localisation module, Autoware.Universe LiDAR perception module and Apollo camera-based perception module.
The thesis “Cybersecurity Testing and Attack Propagation Analysis of Autonomous Driving Software” is published in the Digital Collection of TalTech Library.
Supervisor: Prof. Olaf Manuel Maennel (TalTech)
Co-supervisors: Dr. Mohammad Hamad (Technical University of Munich, Germany), Prof. Raivo Sell (TalTech)
Opponents:
- Prof. Iain Phillips, Loughborough University, UK;
- Prof. Monowar Hasan, Washington State University, USA.
Meeting ID: 959 7501 0463
Passcode: 061691