Hazard Analysis for the SOTIF of Automatic Parking Systems Based on STPA and FSM

Background: The automatic parking system (APS) is an advanced intelligent driving assistance system that depends on sensors, software algorithms, controllers, and actuators to complete parking tasks, ensuring safety and efficiency in the parking process. However, in different parking scenarios, the limitations of sensors and software algorithms potentially lead to a safety issue known as the safety of the intended functionality (SOTIF). Objective: To ensure the SOTIF of the APS, a more effective hazard analysis approach should be adopted to identify potential system hazards. Methods: The study proposes a hazard analysis approach based on systems theory process analysis (STPA), finite state machines (FSM), and the analytic network process (ANP). First, potential unsafe control actions (UCAs) and hazardous events are identified from system state transitions using the STPA-FSM method. Second, qualitative and quantitative cause analysis of UCA is performed using the STPA-ANP method to identify key causal factors. Finally, the proposed approach is applied to an APS, and the formal tool UPPAAL is used to verify the causal logic of the identified UCA and causal factors. Results: Take the state transitions of the APS model as an example; six UCAs and five hazardous events were identified. A qualitative cause analysis was conducted on UCA-1, identifying eight causal factors. The quantitative assessments were made on these causal factors to identify three key causal factors. Conclusion: The approach was suitable for the SOTIF hazard analysis of APS, and it could also be applied to other complex high-level intelligent driving assistance systems. It can identify potential UCAs in the dynamic changes of complex systems and determine key causal factors through causal analysis. Formulating corresponding safety strategies for these key causal factors can enhance the system';s SOTIF. This demonstrated the effectiveness of the approach in this patent.