Research On Multi-UAV Formation Control Based On Information Consensus

Multi-UAV formation control is an important and hot issue in aviation, which hasgood prospects in both military and civil fields. In recent years, the emergingmulti-agent information consensus provides new ideas for multi-UAV formation control,making the objectives of multi-UAV achieve convergence through informationexchange. Though it has unique advantages in dealing with key technology anddifficulties, there are lots of theoretical and technological problems to be explored. Thisdissertation focuses on two important and challenging issues: information consensuscriteria for multi-agent systems under nonideal situations and multi-UAV formationcontrol based on information consensus.1. Considering the high-order, nonlinear and nonideal system with communicationdelays and disturbances, the criteria of information consensus are put forward. Firstly,based on matrix decomposition theory, closed-loop transfer function is constructed forlinear systems. If each weight of different state derivatives satisfies Hurwitz stability,information flow can achieve convergence stability, which is a sufficient and necessarycondition for the system with a directed spanning tree. Then, the nonlinear system withmeasurable disturbances is simplified to higher-order linear integrator form based ondisturbance decoupling. After that, a sufficient condition is deduced for convergence byapplying generalized Nyquist criterion and Gershgorin disc method while time delaysexist in first-order and second-order systems. Further, for high-order systems with timedelays, Euler-Lagrange equations are constructed to obtain the sufficient condition ofasymptotical convergence stability. Moreover, Euler-Lagrange equations are extended tochaotic systems with nonlinear inputs, and distributed sliding-mode law is designed tocompensate for the effects of nonlinearity. Finally, Simulation results show theeffectiveness of the proposed consensus algorithms.2. Multi-agent formation keeping algorithm is proposed, combined withinformation consensus. Firstly, the topology of cluster head is established for formationcommunication. Each UAV can be treated as the formation reference point todemonstrate formation geometry pattern and the desired position for every UAV information. The coordination variables of position and velocity are defined for designingformation controllers. Secondly, cooperative flight control law and tracking control laware designed together to maintain a specified formation configuration, managingposition and attitude respectively. By contrast, cooperative flight controllers combininginformation consensus with nonlinear dynamic inverse or PID are designed differently.Sensor errors in position data are suppressed by synchronization technology to improve control precision. In order to follow three-dimensional flight path, tracking control lawis designed by introducing flight path as a reference state into information consensus,which makes the formation maintain configuration and follow the flight accurate pathsimultaneously. With whole system composed of UAV model, flight control law andtracking control law, the formation keeping strategy is full closed loop and with fullstates. At last, simulation results demonstrate that the formation keeping algorithm hasquick response and high control precision.3. Formation swarm, dissolution and change strategy are put forward based oninformation consensus. Firstly, swarm process is decomposed of three steps, includingchoosing and allocating target assembly points, generating loose formation andgenerating close formation. Using optimization select algorithm in distance space, targetassembly points are assigned to every UAV. Velocity consensus is applied to generateloose formation efficiently with imprecise track control. With gradual compression ofconfiguration, close formation is formed ultimately with velocity and attitude consensusfor precise track control. Secondly, formation dissolution is the inverse process ofswarm, which is realized by gradual expansion from close formation to loose formation.Thirdly, formation change strategy is given out by changing geometric constraints of theformation, switching between two or more fundamental equilibrium states. Finally,Formation swarm, dissolution and change strategy are verified by experiment.4. During close Multi-UAV flight, united collision avoidance algorithm isproposed based on artificial potential field and information consensus. Firstly, theconcept of communication topology and weights of different states are fused in artificialpotential, which makes artificial potential field closer to actual application. Next,artificial potential of obstacle avoidance in the airspace is also given out, consideringnot only the distance between formation and obstacles but also the relative velocitybetween them with repulsive potential increasing along with the relative velocity. Then,combining collision avoidance algorithm based on artificial potential field withformation keeping algorithm based on information consensus, united collisionavoidance algorithm is designed, which is used by null space strategy to project themission of formation keeping with low priority to the mission of collision avoidancewith high priority. Rigid constraints of formation are also reinforced. In the end,simulation results confirm that united algorithm can avoid collision effectively and isfeasible to maintain configuration.