This paper investigates the effectiveness of designed random behavior in cooperative formation control of multiple mobile agents. A method based on artificial potential functions provides a framework for decentralized control of their formation. However, it implies heavy communication costs. The communication requirement can be replaced by onboard sensors. The onboard sensors have limited range and provide only local information, and may result in the formation of isolated clusters. This paper proposes to introduce a component representing random motion in the artificial potential function formulation of the formation control problem. The introduction of the random behavior component results in a better chance of global cluster formation. The paper uses an agent model that includes both position and orientation, and formulates the dynamic equations to incorporate that model in artificial potential function approach. The effectiveness of the proposed method is verified via extensive simulations performed on a group of mobile agents and leaders.
Convergence and Collision Avoidance in Formation Control: A Survey of the Artificial Potential Functions Approach