A swarm intelligence based approach to the mine detection problem

摘要

This research focuses on the application of swarm intelligence to the problem of mine detection. Swarm Intelligence concepts have captivated the interests of researchers mainly in collective robotics, optimization problems (traveling salesman problem (TSP), quadratic assignment problem, graph coloring etc.), and communication networks (routing) etc [1]. In the mine detection problem we are faced with sub problems such as searching for the mines over the minefield, defusing them effectively, and assuring that the field is clear of mines within the least possible time. In the problem, we assume that the mines can be diffused by the collective action of the robots for which a model based on ant colonies is given. In the first part of the project we study the ant colony system applied to the mine detection problem. The theoretical aspects such as the ant's behavior (reaction of the ants to various circumstances that it faces), their motion over the minefield, and their process of defusing the mines are investigated. In the second section we highlight a certain formulation that the ants may be given for doing the task effectively. The ants do the task effectively when they are able to assure that the minefield is clear of the mines within the least possible time. A compilation of the results obtained by the various studies is tabulated. In the third and final section we talk about our emulations conducted on the Multi Agent Biorobotics Lab-built groundscout robots, which were used for the demonstration of our swarm intelligence-based algorithms at a practical basis. The various projects thus far conducted were a part of the Multi Agent Biorobotics Lab at Rochester Institute of Technology.