A behavior-based architecture for multi-robot chemical source localization and its implementation with evolutionary computation

Chemical source localization (CSL) is significant for the protection of public security. In typical CSL problems, a collaborative group of multiple robots are expected to perform better than a single robot. This paper proposes a behavior-based multi-robot chemical source localization (BbMR-CSL) architecture. It includes three group behaviors, i.e., the chemical finding, chemical source tracing (CST), and chemical source declaration behaviors, among which the CST behavior is further divided into two individual behaviors. A hierarchical behavior selection method is designed for coordinating the behaviors in the BbMR-CSL architecture. Then, a method framework is also proposed for combining evolutionary computation (EC) with the BbMR-CSL architecture. Based on the proposed framework, three typical EC algorithms (i.e., the genetic algorithm, the artificial bee colony algorithm (ABC), and the global-best-guided ABC) are implemented in the BbMR-CSL architecture. The resulting behavior-based CSL methods were tested and compared in a simulated filamentous plume environment.