Safety-Critical Requirements Research - Autonomous Vehicles
Research project applying risk-based requirement analysis (FTA, FMEA, STPA) to improve safety assurance in self-driving cars.
Overview
A research project for Deakin's Software Requirements Analysis and Modelling unit, investigating how risk-based requirement techniques can strengthen safety assurance for autonomous vehicles operating in unpredictable, real-world environments.
The problem
Traditional requirement analysis techniques often fall short for safety-critical systems like self-driving cars, where sensor failures, software bugs or unanticipated conditions can lead to disastrous outcomes; existing approaches don't sufficiently address risk-aware validation of dynamic, real-time decision-making.
My approach
Analysed a Melbourne CBD autonomous-vehicle case study using use-case and activity diagrams to model core functions (driving, route planning, traffic-light recognition, collision detection, lane-keeping); applied Fault Tree Analysis, Failure Mode and Effects Analysis and System-Theoretic Process Analysis to identify and mitigate safety risks early in the design process; and translated the findings into actionable, safety-focused requirements aligned with the ISO 26262 functional-safety standard.