Flocking of Linear Parameter Varying Agents: Source Seeking Application with Underwater Vehicles

Abstract

A distributed flocking control scheme is proposed for a network of Autonomous Underwater Vehicles (AUVs) which are modeled as Linear Parameter Varying (LPV) systems. This scheme is applied here to the solution of a source seeking problem, i.e. the vehicles (agents) measure the local values of a scalar field and are required to flock to its maximum (source). It is assumed that agents have the gradient and Hessian information of the scalar field at their current position. The control architecture of each agent is divided into two modules: a flocking filter which receives data from neighbours and generates a reference signal based on a flocking control law, and a feedback loop for tracking this reference. By this approach, a separation in design is achieved by designing a local LPV tracking controller for each agent and a network flocking filter which can be analyzed to guarantee stability in the sense of Lyapunov, i.e. the boundedness of agents’ trajectories. Simulation results illustrate the practicality and benefits of the proposed flocking architecture scheme by applying it to a network of realistic autonomous underwater vehicles.