Generating Optimized RSP Loading Plans using modern AI-based Constraint Solvers

The U.S. Air Force maintains readiness in part by pre-positioning Readiness Spares Packages for rapid deployment to a forward location in the case of a crisis or contingency. RSPs are typically stored in containers such as ISU-90’s as well as on 463L pallets. When assembling RSPs, an airman takes a list of required spare parts and develops a loading plan that specifies how many containers are required and how the spare parts should be arranged into the containers. This process is time consuming and error-prone because airmen often partially develop a loading plan, try executing the loading plan, and then must backtrack in the event the plan is unachievable. Virtualitics can quickly configure a constraint-solving model to encode requirements and constraints of RSP load-planning – e.g. by using a highly expressive and flexible logic, Satisfiability Modulo Theories (SMT) – and then utilize a modern, high-performance AI-based constraint solver (e.g. an SMT-solver) to efficiently identify optimized solutions for the RSP load-planning model. Second, Virtualitics will leverage its expertise in developing AI-based optimization models to ensure the prototype can: (i) generate load-plans optimized with respect to multiple quantitative metrics and (ii) incorporate custom heuristics (designed specifically for RSP Optimization use case) that accelerate the AI-based optimization process, thereby decreasing the web-application’s runtime to align with user operational requirements and expectations.