Dynamic Control of Mobile Multirobot Systems: The Cluster Space Formulation

Abstract

The formation control technique called cluster space control promotes simplified specification and monitoring of the motion of mobile multirobot systems of limited size. Previous paper has established the conceptual foundation of this approach and has experimentally verified and validated its use for various systems implementing kinematic controllers. In this paper, we briefly review the definition of the cluster space framework and introduce a new cluster space dynamic model. This model represents the dynamics of the formation as a whole as a function of the dynamics of the member robots. Given this model, generalized cluster space forces can be applied to the formation, and a Jacobian transpose controller can be implemented to transform cluster space compensation forces into robot-level forces to be applied to the robots in the formation. Then, a nonlinear model-based partition controller is proposed. This controller cancels out the formation dynamics and effectively decouples the cluster space variables. Computer simulations and experimental results using three autonomous surface vessels and four land rovers show the effectiveness of the approach. Finally, sensitivity to errors in the estimation of cluster model parameters is analyzed.