Adversarially Learned Inference for Countermeasure Exploitation (ALICE)

In the world of autonomous vehicles, agents make intelligent decisions not only when acting to achieve their own goals, but when interpreting the goals of others to inform their own actions. In the case of unmanned air systems (UASs), the goals of others can often be nefarious. Aptima and our partner ASEC are proposing to develop the Adversarially Learned Inference for Countermeasure Exploitation (ALICE) solution. ALICE uses a generative model to find courses of action (COAs) that maximize information gain while minimizing risk to BLUE assets. ALICE will also design and develop a human-machine interface that ensures that the decisions made by the automation are understandable to human operators. ALICE will not only select the best actions for our own UAS, but will also seek to prompt the enemy into action in order to uncover some information about the enemy’s goals or capabilities. Finally, ALICE will leverage the team’s multi-domain expertise in order to better understand how a UAS needs to work in the larger operational system. As a result, ALICE will maintain situational awareness of known threats and real-time threat activity in order to optimize the countermeasure response and successfully maneuver through the threat environment.