A Cognitive Architecture for Naval Mine Countermeasures (MCM) Navy SBIR FY2012.1 Sol No.: Navy SBIR FY2012.1 Topic No.: N121-082 Topic Title: A Cognitive Architecture for Naval Mine Countermeasures (MCM) Proposal No.: N121-082-0266 Firm: Soar Technology, Inc. 3600 Green Court Suite 600 Ann Arbor, Michigan 48105-2588 Contact: Robert Marinier Phone: (734) 887-7615 Web Site: www.soartech.com Abstract: Current Mine Countermeasure (MCM) operations attempt to balance risk to commander's intent and time via processes that are staff and time intensive. Existing tools such as MEDAL have helped to alleviate some of this, but those tools lack an comprehensive understanding of the uncertainties involved, and how those impact commander's intent. Furthermore, existing tools are mostly passive, representing data, but not interpreting it, leaving issues of situation detection, assessment, and responding entirely up to the staff. An ideal tool would be a decision aid that works with users and existing tools to perform situation detection, assessment, and response. Such a tool would need to understand commander's intent and the uncertainties inherent in MCM-relevant situations, apply expert MCM knowledge in understanding situations, relating intent and situations, and generating appropriate recommendations. Furthermore, such a tool needs to build trust with users by explaining how it arrived at those conclusions, including making the uncertainties involved clear. Recent years have seen increased development of cognitive architectures, which are ideal platforms for encoding expert knowledge and interacting with users. SoarTech proposes to leverage prior work in situation detection, assessment, response, and explanation. It will extend this work for the MCM domain, uncertainty, and user interaction. Benefits: A decision aid that transparently helps users detect, assess, and respond to uncertain situations has the potential to revolutionize how MCM operations are conducted. Current MCM operations utilize MEDAL, which eases many forms of data analysis and planning, but still requires extensive manpower and time, resulting in slow and potentially suboptimal responses. Instead, MCM staff should have access to a transparent decision aid that interacts with both MEDAL and the user to perform situation detection, assessment, and response generation in order to best achieve commander's intent in uncertain conditions. Our proposed TRUFAST architecture will allow users to better detect, assess, and respond to situations by explaining the uncertainties involved while simultaneously instilling trust by explaining its reasoning. Beyond MCM applications, TRUFAST can provide a basis for many applications across a variety of domains that require some form of situation detection, assessment, and response, both as a decision aid and an autonomous system. As a knowledge-based system, TRUFAST has a core set of capabilities for situation detection, assessment, and response, but is agnostic to the particular domain in which it is operating; indeed, it is based on a system originally developed for autonomous ground vehicles, both real and simulated. Return