System and Safety Analysis for Autonomous Center-line Tracking with SYSAI

View Video Presentation: https://doi.org/10.2514/6.2022-2399.vid Machine learning (ML) applications and Artificial Intelligence (AI) components have become increasingly important in the Aerospace domain. However, traditional assurance processes and tools are not well suited for the Verification and Validation (V&V) of safety-critical systems with AI components. In this paper, we present SYSAI (System Analysis using Statistical AI), a flexible statistical learning framework for the V&V and analysis of complex and high-dimensional Aerospace systems with Neural Network components. SYSAI provides functionality for numerous analysis and V&V tasks, including statistical analysis of training data, safety-envelope and time-series analysis, property checking, and intelligent test-case generation. We demonstrate SYSAI with our industrial partner’s ACT (Autonomous Centerline Tracking) component.