This presentation aims to shed light on the gaps in AI perception systems, which are the main components of autonomous driving. It explores the lessons learned from cyberattacks and emphasizes the critical need to address these gaps. By examining practical examples and discussing key challenges, this presentation provides valuable insights to enhance the robustness of AI perception systems in software-defined vehicles.

In this session, you will get a better understanding about:

• How vulnerabilities in AI perception systems can be exploited by attackers to manipulate the behavior of the systems and compromise the safety and functionality of SDV
• Why robustness is enhanced by the implementation of detection mechanisms, and proactive and reactive prevention mechanisms, e.g., adversarial training
• What challenges occur with effective cybersecurity measures for AI perception systems