Learning-Based Testing for Safety Critical Automotive Applications.

Learning-based testing (LBT) is an emerging paradigm for fully automated requirements testing. This approach combines machine learning and model-checking techniques for test case generation and verdict construction. LBT is well suited to requirements testing of low-latency safety critical embedded systems, such as can be found in the automotive sector. We evaluate the feasibility and effectiveness of applying LBT to two safety critical industrial automotive applications. We also benchmark our LBT tool against an existing industrial test tool that executes manually written test cases.