Testing the Intelligence of Unmanned Autonomous Systems

There is a common misconception in the testing industry that all unmanned autonomous systems can be tested using methodologies developed to test manned systems. A tester can incorrectly state that driving a manned vehicle is conceptually identical to driving an unmanned vehicle; the only difference lies in the position of the operator. In actuality, the main difference lies in the unmanned autonomous system’s role in the decision process. Because testers cannot make assumptions about the decision process within an unmanned autonomous system, there is a need for a methodology that completely tests this decision process without biasing the system into a default ‘‘human’’ solution. This article presents a comprehensive methodology for testing the intelligence of an unmanned autonomous system within a given set of requirements.