Towards Online Partitioning of Agent Sets based on Local Information.

Abstract

This paper deals with the autonomous and uniform partitioning of very simple, homogeneous agents onto a fixed number of targets in a geometric space. A set of algorithms will be presented and compared according to the quality of the partitioning and the runtime of simulations with different starting conditions. Random and heuristic approaches will be discussed. The algorithm with the most promising experimental results, the Border Switch Strategy (BSS), will be explained and formalized in a detailed way.