Task Partitioning via Ant Colony Optimization for Distributed Assembly

We address the distributed assembly of a structure by a team of homogeneous robots. We present an ant-colony-optimization (ACO) based algorithm to partition general 2- and 3-D assembly tasks into N separate subtasks. The objective is to determine an allocation or partitioning strategy that minimizes the workload imbalance between the robots that allow for maximum assembly parallelization. This objective is achieved by extending ACO to apply to a team of ants dividing a set of tasks, with pheromone marking connections between tasks guiding decisions on task allocation. We present simulation results for various 2-D and 3-D structures and discuss the advantages of the ACO formulation in the context of other existing approaches.