Cascaded Safety Analysis and Test Scenario Generation Techniques for Autonomous Driving: A Case Study with WATonoBus

Area of Science:

• Autonomous Systems Engineering
• Robotics and Control Systems
• Functional Safety Analysis

Background:

• Ensuring the safety and reliability of autonomous driving systems is paramount.
• Understanding system capabilities and operational boundaries is critical for safe deployment.
• Existing verification methods require enhancement for complex autonomous functions.

Purpose of the Study:

• To develop a comprehensive framework for safety verification of autonomous driving function stacks.
• To propose an effective procedure for generating diverse and relevant test cases.
• To demonstrate the practical application and efficacy of the proposed methods in real-world scenarios.

Main Methods:

• Synergistic integration of operational flow-oriented Hazard and Operability Study (HAZOP) with cascaded System-Theoretic Process Analysis (STPA).
• A novel test case generation procedure involving discrete parameter expansion via tree search.
• Heterogeneous sampling in the continuous parameter space for comprehensive test coverage.

Main Results:

• The proposed integrated approach enhances the safety verification process for autonomous driving systems.
• The test case generation method effectively covers a wide range of operational scenarios.
• A real-world case study with WATonoBus validates the practicality and effectiveness of the methods.

Conclusions:

• The developed methods provide a substantial contribution to autonomous vehicle safety.
• The findings offer critical insights for the ongoing research and development in autonomous driving.
• The approach enhances the reliability and secure operation of autonomous vehicles in complex environments.