Security and Safety in Autonomous Driving - RMGS
DescriptionWith the technology advancement, autonomous vehicles are foreseen to achieve full autonomy and supplant human drivers in the near future. However, a wide adoption will not come before some obstacles are overcome, among which, security and safety are the top ones. Autonomous vehicles are vulnerable to various types of attacks. The driving decision of autonomous driving relies on communication externally with infrastructure and internally among different on-board components, which expose them to cyber-attacks. Moreover, current AI solutions adopted by autonomous vehicles for perception can be fooled by adversarial examples, which may lead to serious accidents. In this project, we aim to 1) comprehensively study the possible attacks that an autonomous vehicle may face, ranging from cyber-attacks hijacking vehicles remotely to physical adversarial attacks confusing deep learning systems by altering the physical environment; 2) evaluate the corresponding risks to the safety of autonomous driving, and 3) design counter-measures to safeguard autonomous vehicles.
|Effective start/end date||1/04/20 → …|