Application of Bellman's Equation in Ant-Like Robotic Device Path Decisions

Abstract

Swarm Intelligence is about emergency of collective intelligence from groups of homogeneous robotic devices deployed for a purpose. Ant Colony Systems, in particular, are inspiring. They commonly have drawn inspiration from the behaviors of real ants in nature in order to construct routes between the food sources and the nest. There are still gaps in alternative options for path decision in ant agents. Bellman's equation has been successfully used to solve path decision problems in machine learning. We proposed to investigate impact of a Bellman's equation inspired algorithm for path decision on stigmergic ant agent robotic devices. A design science research paradigm was used to design our research experiment in which a simulated environment was designed to simulate the behavior of ant agents when using a Bellman's equation inspired algorithm for path decision. We introduced a reward function to the orientation process of ant agents. Reward function rewards a decision made when an ant moves from one point to an adjacent cell. The Bellman's inspired algorithm for ant orientation led to convergence of ant agents even though there was reduced quality of convergence. Evaluation of results show that Bellman's equation can be used in path decision processes for ant agent robotic devices. Our results contributed to adding an alternative way of implementing path decision for ant agents. This will help in growing the knowledge around ant agents and finding better ways to implementing path decisions for ant agents.