Risk Assessments for Evasive Emergency Maneuvers in Autonomous Vehicles

Researchers Aliasghar Arab, Milad Khaleghi, and Koorosh Aslansefat have introduced a novel verification and validation (V&V) framework for Evasive Minimum Risk Maneuvers (EMRM) in autonomous vehicles. Published on arXiv, their work unifies Hazard Analysis and Risk Assessment (HARA), System-Theoretic Process Analysis (STPA), and Finite State Machine (FSM) modeling into a single traceable pipeline. This integrated approach addresses a critical gap in existing safety assessment methods by combining hazard identification, unsafe control action analysis, and state transition modeling. The framework drives automated scenario generation with measurable parameter-space coverage, enabling high-resolution testing that no single method can achieve alone.

Applied to a T-junction EMRM case study, the framework guided 1,880 RRT-based simulations varying ego speed, time-to-collision (TTC), and road friction. Key findings reveal that T-junction geometry makes stopping and navigating nearly equally difficult, with the intermediate mitigation mode occupying only 1.9% of the feasible parameter space. EMRM steering strategies achieved an 81% collision-avoidance rate and reduced mean residual impact speed from 18.9 km/h to 9.0 km/h compared with emergency braking alone. The framework also attained 100% coverage for hazards, unsafe control actions, and parameter space, versus ≤1% for traditional methods, demonstrating a significant leap in autonomous vehicle safety verification.