Automated Model-in-the-Loop Testing of Continuous Controllers Using Search

Abstract

The number and the complexity of software components embedded in today’s vehicles is rapidly increasing. A large group of these components monitor and control the operating conditions of physical devices (e.g., components controlling engines, brakes, and airbags). These controllers are known as continuous controllers. In this paper, we study testing of continuous controllers at the Model-in-Loop (MiL) level where both the controller and the environment are represented by models and connected in a closed feedback loop system. We identify a set of common requirements characterizing the desired behavior of continuous controllers, and develop a search-based technique to automatically generate test cases for these requirements. We evaluated our approach by applying it to a real automotive air compressor module. Our experience shows that our approach automatically generates several test cases for which the MiL level simulations indicate potential violations of the system requirements. Further, not only do our approach generates better test cases faster than random test case generation, but we also achieve better results than test scenarios devised by domain experts.