Journal international d'intelligence d'essaim et de calcul évolutif

Abstrait

An Artificial Swan Formation Using the Finsler Measure in the Dynamic Window Control

Jun S Huang*, Siqing Ma, Gao Li, Oliver W Yang, Chang Shao

This paper studies the swarm formation control for a large number of unmanned robots (artificial swans) to expel invading objects (geese) in applications such as airport bird control. This control algorithm extends the existing DWA (Dynamic Window Algorithm) control by inserting a new term using the Finsler measure in the objective function during the optimization process. As demonstrated and proven with the simulation section and derivation in the appendix, the Finsler measure can gauge the degree of local collaboration among unmanned robots and therefore allow us to form a statistical collaboration strategy to control a swarm formation to expel invading objects. It also allows fast optimization and therefore real-time formation control of a large number of robots without seeing the robots to collide with each other or with obstacles along their moving path, because the new term in the objective function allows the trade-off between swan group moving speed and its group watch-out coverage. Our performance evaluation and analysis of different performance metrics via simulations demonstrate the feasibility of this control algorithm.