Comparing swarm algorithms for large scale multi-source localization

We propose a common set of validation benchmarks and a reference algorithm that provide ground-truth for comparative analysis of multi-source robot localization algorithms for large scale applications. The benchmarks capture the primary first-order attributes of the general problem: source characterization and distribution, initial robot distributions, and dead space. The biased random walk (BRW) reference algorithm represents a simple approach without robot communication. We demonstrate how the benchmarks are used, in combination with sensitivity analysis, to provide insights into the relative performance of algorithms for applications with potentially large numbers of sources. The Glowworm swarm optimization (GSO) algorithm and a new GSO/BRW hybrid algorithm are evaluated in an attempt to improve upon the baseline BRW performance. Unfortunately, none of the algorithms presented are able to locate all sources on the large scale benchmarks. Moreover, the algorithms perform worse on these large scale experiments than they did on smaller examples.