Search-based Software Testing for Drone Applications: An Experience with the Simulink Environment

Unmanned aerial vehicles (UAVs) are frequently used in monitoring and inspection of large and isolated areas, and often use line following techniques to guide their movement. The successful execution of this task greatly depends on the correct design of the software controller. Search-based software testing (SBST) is a widely used technique to check for software defects. It iteratively generates test cases until either violations of the system requirements are detected or the time budget is exceeded. However, the effectiveness of SBST strongly depends on the application domain. This empirical evaluation paper assesses the effectiveness of SBST in supporting the design of UAV applications by considering a rigorous case study. It considers three different versions of a drone software controller. It assesses the capability of SBST in generating failure-revealing test cases and the usefulness of the test cases. Our results confirm the effectiveness of SBST and the usefulness of the generated test cases.