Metrics for co-evolving autonomous systems

Autonomous system innovations have overrun the test and evaluation capability to find problems before they become expensive to fix --- or lethal. The autonomy paradigm demands that an equivalent test and evaluation system be conceived, architected and engineered, operated and evolved. This in turn demands an autonomous test and evaluation enterprise, staffed with competent systemists, as the enabling agent. This paper outlines the metrics and key capabilities for realizing such an enterprise. It features a game-theoretic basis, a model-based systems engineering approach and a four part strategic framework. This paper focuses on the unclassified situation in the U.S. Dept. of Defense. However, these ideas will apply to other domains of autonomy in both the public and private sectors.