Abstract. This paper presents CBRetaliate, an agent that combines Case-Based Reasoning (CBR) and Reinforcement Learning (RL) algorithms. Unlike most previous work where RL is used to improve accuracy in the action selection process, CBRetaliate uses CBR to allow RL to respond more quickly to changing conditions. CBRetaliate combines two key features: it uses a time window to compute similarity and stores and reuses complete Q-tables for continuous problem solving. We demonstrate CBRetaliate on a team-based first-person shooter game, where our combined CBR+RL approach adapts quicker to changing tactics by an opponent than standalone RL.

No abstract

No abstract

We describe our progress on instrumenting a Navy software simulator for use in the context of intelligent agent research. The Tactical Action Officer (TAO) Sandbox, which is being developed at the University of Southern California, is used by officers to practice tactical decision making in the context of Navy surface fleet missions. NRL and Knexus Research Corporation have integrated this simulator with intelligent agents using the Lightweight Integration and Evaluation Testbed (LIET), thus permitting the agent to play the role of a trainee. This will permit us to use the TAO Sandbox in our artificial intelligence research, where we are currently focusing on algorithms for continuous planning that can dynamically reason about what goal should be pursued at any time during a mission. This paper briefly descibes our motivation for this integration, project status involving this simulator, and future goals.

No abstract

Modern fleet security requires accurate threat analysis in real-time, which relies on a range of contextual information (e.g., vessel size, speed, heading, etc.). Rich contextualization may be possible using imaging systems if the images can be used to detect and classify maritime vessels and track their movements. In this work, the effectiveness of the ensemble of Exemplar-SVMs (E-SVM) object detection scheme is evaluated for maritime data where targets are small and have low inter-class variation due to its scalability and ability to learn from limited training examples. Experimental evaluation shows average precision for Annapolis Harbor vessel data is lower than the general 20-category PASCAL VOC challenge due to confusion between boat types.

No abstract

Mine countermeasures (MCM) missions entail planning and operations in very dynamic and uncertain operating environments, which pose considerable risk to personnel and equipment. Frequent schedule repairs are needed that consider the latest operating conditions to keep mission on target. Presently no decision support tools are available for the challenging task of MCM mission rescheduling. To address this capability gap, we have developed the CARPE system to assist operation planners. CARPE constantly monitors the operational environment for changes and recommends alternative repaired schedules in response. It includes a novel schedule repair algorithm called Case-Based Local Schedule Repair (CLOSR) that automatically repairs broken schedules while satisfying the requirement of minimal operational disruption. It uses a case-based approach to represent repair strategies and apply them to new situations. Evaluation of CLOSR on simulated MCM operations demonstrates the effectiveness of casebased strategy. Schedule repairs are generated rapidly, ensure the elimination of all mines, and achieve required levels of clearance. MOTIVATION & BACKGROUNDThe location, identification, and neutralization of enemy explosive ordnance are key to naval power projection and sea control, two core capabilities of U.S. maritime power, as characterized by A Cooperative Strategy for 21st Century Seapower (2010). Missions that accomplish these tasks are referred to as mine countermeasures (MCM) missions, and they are more important than ever due to the increasing prevalence of asymmetric warfare, and the relatively low expense and high impact of mine threats. A key efficiency roadblock to these missions is the complexity and uncertainty of their schedules, which are meticulously constructed and revised frequently once a mission has started. The complexity of these missions arise from a large number of potential combinations that must be considered to coordinate personnel, equipment, and autonomous vehicles; each resource has its own set of capabilities and operational constraints, as well as characteristic failure points. Achieving mission success requires combining sensor data from all resources to reduce high levels of uncertainty, and mistakes can result in loss of personnel and expensive highly specialized equipment. Frequent disruptions in MCM operations occur due to changes in sea state, visibility, weather, resource and communication bandwidth in the area of operations, and equipment failure, among other factors, all of which interfere with resource availability and/or readiness. Therefore, schedules for MCM operations require frequent changes and updates. Current practice calls for manually creating schedules, which is manpower intensive, error prone and prevents thorough exploration of efficient alternatives. Technological solutions for assisting with these problems are, as yet, nonexistent. In particular, there is no support for exploiting the most up to date situation information to reschedule tasks so as to maximize mission per...