Automated initial population generation for genetic assembly planning

Abstract

The initial population quality can affect the performance of a genetic assembly planner. In particular, using a high-quality initial population helps reduce the assembly planner run-time. Prior genetic assembly planners use either a user-supplied initial assembly sequence population or a randomly generated initial population. Creating a user-supplied initial population requires a substantial amount of manpower. Using a randomly generated initial population reduces search efficiency. As a result, we propose a method for automatically generating a high-quality initial population, from an existing assembly model. Our method calculates component assembly complexity and uses both component assembly complexity and component connectivity to generate automatically a valid initial assembly sequence population. Using an automatically generated initial population, we can achieve search efficiencies comparable to the search efficiencies achieved when using a user-supplied initial assembly sequence population. However, we can eliminate the time and manpower required to generate user-supplied assembly sequences.