Unmanned Aerial Vehicle Formation Robust Control And Reconstruction

Unmanned Aerial Vehicle(UAV)formation is that some UAVs form specified shape to execute special tasks in a coordinated way.The formation technology is widely used in military and civil fields,such as surveillance,fire attack,reconnaissance,networking,air refueling and so on.In recent decades,different methods have been proposed to solve the formation control problems,such as the leader-follower strategy,the virtual structure strategy,the behavior-based strategy,the artificial field strategy and the graph-based strategy.In this thesis,we propose a novel formation method based on the artificial physics to study the UAV formation control and reconstruction.The main work and the innovations are listed as follows:(1)Reconstructing the rules of the artificial physics,and applying them to the UAV formation control.The artificial physics strategy,proposed by William M.Spears in1999,models the Newton’s second law.In fact,it is a kind of direct force control.William M.Spears constructs the force rules,and applies the forces on the omni-directional agents.Based on the artificial forces,the system forms ‘potential well’,and each agent is located in a “potential well”,resulting in a regular formation for agents.Different force rules can form different formation.In this thesis,we reconstruct the traditional artificial physics force rules to drive UAVs to locate in a circumference evenly.We call that kind of formation as the standard formation throughout this thesis.(2)Proposing a kind of bijection to change the standard formation into an arbitrary formation.The control value is the force actually,which is a vector.And it can perform scale transformation,rotation transformation.If we change the value and the direction of the control force,the UAVs formation will break the old equilibrium state,and form a new equilibrium state and ‘potential well’.System will reconstruct a new formation.This thesis explored whether there exists such transformation,which can change the value and the direction of the control force to transform the formation.(3)Mapping the control value obtained by the artificial physics into the real UAV control system via backstepping.The desired heading angle and the velocity of each UAV in formation can be obtained based on the proposed artificial physics,which can be regarded as the control variables in guidance level to the real UAV system.To apply it into practices,we should map the control value obtained by the artificial physics into the real UAV control system.It is known that the dynamics of UAV is a nonlinear system,and backstepping method is suitable for the controller design of such nonlinear system.In this thesis,we map the desired heading angle and the velocity into the load factor of the real UAV control variables.(4)Building a high-fidelity simulation environment using X-plane flight simulator.X-plane is a perfect flight simulation software.The control program is burnt on anARM chip.X-plane sends the messages of the UAV states to the ARM via UDP protocols.After receiving the messages,the ARM will compute the control value,and send it back to the X-plane,which forms a closed-form semi-physical simulation loop.Multiple X-planes and ARMs can simulate the UAV formation,which can be used to verify our formation algorithms.