Map and Its Impact on the Functional Safety of Automated Driving Vehicles

The increasing progress of Automated Driving (AD) technologies emphasises the significance of maps in ensuring the safety of these AD systems. While research has been conducted on the safety of AD systems themselves, the role of maps has not been thoroughly explored. In this article, we aim to address this gap by conducting an analysis to quantify the impact of maps on the functional safety of AD systems. We employ System Theoretic Process Analysis (STPA) to study an SAE Level 2 automated driving vehicle that relies on maps. Through this approach, we estimate and identify various unsafe scenarios that may arise due to map data. Furthermore, we conduct simulations using CARLA to measure the influence of safety-critical map features (identified based on the outcomes of STPA). To account for uncertainties in these safety-critical map features, we introduce a Gaussian noise signal into the model. To evaluate the vehicle's safety, we establish Key Performance Indicators and record their values across various test cases. Through this research, we successfully identified unsafe scenarios along with their corresponding map features. Leveraging simulations, we also showcased the admissible error margins in the map for the selected map feature, ensuring the secure operation of an AD system.