Cybersecurity Test Range for Autonomous Vehicle Shuttles
Andrew D. Roberts, Olaf Maennel, Nikita Snetkov · 2021
Autonomous vehicle (AV) shuttles for public transportation are challenged by a lack of cybersecurity testing and platforms to conduct testing. Real-world, operational vehicles are expensive and operators are reluctant to test cybersecurity test cases that could impart damage to the systems. Furthermore, despite rigorous regression testing methods used within the broader automotive industry, vulnerabilities of automotive systems are still being found by creative edge and corner cases. To enable testing for edge and corner cases for AV shuttles, we propose the integration of cyber-physical test beds and ranges into the testing program of real-world AV shuttles. We evaluated the Massachusetts Institute of Technology (MIT) DuckieBot as the basis for comprehensive cybersecurity vulnerability testing and show how the results can be applied to the iseAuto, a real-world AV shuttle operating in Tallinn, Estonia. The MIT DuckieBot test bed is used to replicate the complexity and interactions of relevant systems of the iseAuto AV shuttle. The practical evaluation, involving cybersecurity edge and corner test cases demonstrated that cyber-physical test beds and ranges can support agile, repeatable cybersecurity testing that is low-cost and is conducted in a controlled and safe environment.