Autonomous tracking of chemical plumes developed in both diffusive and turbulent airflow environments using Petri nets

This paper presents a Petri net model for autonomous tracking of chemical plumes developed in both diffusive and turbulent airflow environments. It has been challenging to develop a generalized algorithm to effectively trace both types of chemical plumes due to the significant differences of their kinetic and dynamic properties. Our idea is to utilize a Petri net to model the change relationships of chemical concentrations acquired by two sensors mounted on the both sides of a DaNI robot during a tracing process. Because the relationships imply the effects of flow variation on chemical puffs, a flow sensor is eliminated. To express and maintain the knowledge of chemical concentration changes using the Petri net, we design a mapping algorithm for generating the Petri net from production rules. The Petri net model is implemented on the robot using LabVIEW. The chemical plume tracing experiments achieve 93.8% and 87.5% of source localization rates under both turbulent and diffusive airflow environments, respectively.