Simulation Platform for the Evaluation of Robotic Swarm Algorithms

Abstract

One major problem in the development of robotic swarms is the slow process of testing. Testing different algorithms or variations of a single one using physical robots requires reprogramming every robot in the swarm before every run. Hence, the speed at which a robotic swarm can be tested is highly dependent on the time taken to reprogram the entire swarm and the physical speed at which the swarm operates. This paper details the development of a computer-based simulation platform for rapid development and testing of swarm-intelligence algorithms in an effort to mitigate the current bottleneck imposed by testing. The simulator uses an object-oriented programming environment to facilitate the implementation and modification of swarm algorithms. Simulation of food foraging in an ant-hill scenario is used to demonstrate the effectiveness of the simulator.