Air and Load Plan Scheduler Multi-Agent System (ALPSMAS)

The process of creating Air Plans and Load Plans in support of aircraft carrier operations can be challenging and time consuming. Navy staff supporting Strike Ops need to gather information from multiple stakeholders to address tactical requirements, to determine the availability of needed resources, and consider logistics, airwing training requirements, and ship operations and commitments. This results in large amounts of information being processed, requires constant communication with staff from the airwing and multiple ship work centers, while addressing frequent mission changes before air plan execution. The proposed SBIR Phase 2 project seeks to provide workload relief to Strike Ops personnel and expedite the availability of air plans and load plans by prototyping the use of intelligent software agents capable of analyzing mission, airwing, and ship requirements. The agents take into account available ship resources to generate Air and Load Plans based on learned planning rules and user preferences. The Air and Load Plan Scheduler Multi-Agent System (ALPSMAS) project objective is to research and prototype a decision aid for carrier strike planners. The proposed project consists of an SBIR Phase 2 project with a Phase 2 options. The Phase 2 project objectives are: Identify the air and load plan methodology and rules used by Carrier Strike Group (CSG) officers when addressing CSG tactical, logistical, and training requirements. Research state-of-the-art schedule generation algorithms based on heuristic schedule generation methods. Research intelligent agent learning capabilities using deep convolutional neural network architectures. Design and implement an ALPSMAS prototype capable of generating air and load plans based on plan requirements. The Phase 2 option project will further explore the use of the ALPSMAS in the context of naval operations. The project options objectives are: Identify ship integration and naval information systems needs for the introduction of the ALPSMAS capabilities into naval operations. Demonstrate an ALPSMAS integration concept with naval information systems. Probus proposes to address the Air Plan and Load Plan schedule generation problem using a hybrid heuristic method in which the planning problem is modeled as a Resource Constraint Project Scheduling Problem (RCPSP). RCPSP is a well-studied optimization research problem in which a set of projects needs to be scheduled and each project uses one or more of a limited set of resources subject to operational constraints. The general problem supports the consideration of dependencies among project resources. We will approach the RCPSP problem by representing the planning problem as a job-shop scheduling problem (JSSP) with resource allocation extensions.