Automated test trajectory for hybrid systems

This paper contains the first steps toward the development of an automated test-input generation algorithm - inspired by test input generators for software systems - for human-in-the-loop embedded systems such as automotive and avionic system as well as an increasing number of smaller consumer devices. The idea is to create an algorithm whose inputs are: a model of the control system being tested, along with initial conditions; and the specification which it is being tested against. The output of the algorithm is a set of open loop test input functions which represent a minimal set of test scenarios required to determine with within some confidence interval if the system meets the specification. The primary motivation behind developing such a method is to avoid the time, expense and inconclusiveness of trial an error testing by using automated rigorous tools to guide the design process. The notions of a test generator and test adequacy criteria, defined for software testing, are formalized for control systems in terms of the maximum principle. A numerical optimal control based technique is presented.