On the Design of Unscented Kalman Filter-Based Multi-Agent Target Interception Control

Abstract

This paper discusses a distance-based control scheme for the multi-agent target interception problem. Using the concept of graph rigidity, a novel Unscented Kalman Filter-based control law is proposed for the multi-agent target interception with single integrator agent dynamics. The control law is formulated at the velocity level and is built on the rigidity characteristics of the graph that depicts the interactions between the agents' sensing and communication. The proposed control scheme includes a formation acquisition term that depends on the graph rigidity matrix as well as a target interception term to take into consideration the unknown target motion velocity using the estimates from the unscented Kalman filter(UKF). A Lyapunov analysis demonstrates that the desired formation geometry is acquired exponentially while the leader is tracking the target motion. Finally, the efficacy of the proposed control law is validated through the simulation results. The chatter-free control signal further demonstrates the superiority of the UKF-based control law.