Dynamic Risk Management for Autonomous Vehicles

The Hazard Analysis and Risk Assessment (HARA) of ISO 26262 evaluates the potential for system malfunctions to cause hazardous events in automotive systems and determines Automotive Safety Integrity Levels (ASILs) for safety measures such as fault prevention, tolerance, or mitigation. However, traditional HARA assumes the presence of a human driver to take control of the vehicle, but autonomous vehicles (AVs) operate without human intervention in uncertain and dynamic environments. Therefore, a Dynamic Risk Management (DRM) framework is proposed to ensure safety in AVs by making risk-dependent decisions for system reconfiguration at runtime. The DRM uses on-board sensor data to perform risk assessment and trigger safety actions as needed. This approach moves from static risk management methods to dynamic risk management to ensure the safe behavior of AVs in uncertain situations.