Consensus Control Of Multi-agent Systems In Complex Environments And Its Applications In Cooperative Flight Control Of Unmanned Aerial Vehicles

The agents in a networked system can achieve an amazing overall effect,rather than simple collection of their individual functions,by interacting with each other via a cooperative control scheme.The multi-agent cooperation is a cutting-edge technology of improving efficiency and resolving the conflicts between the requirements for highfunctionality in specific tasks and the limited capability of an individual device.Starting with the consensus problems of multi-agent systems in complex environments,the thesis analyzes and designs cooperative control algorithms to provide system stability and enhance environmental adaptivity.The main innovative results of the essay are listed below:(1)The effect of nonuniform delays on the rectilinear and rotational consensus motion of second-order multi-agent systems is investigated,and the maximal upper bounds of the tolerated time delays(termed “delay margin”)are obtained.A novel frequency domain method is adopted to analyze the stability of first-degree second-order systems with either directed or undirected topology and nonuniform time delays: their characteristic equations are transformed into quadratic polynomials regarding their imaginary eigenvalues to calculate the delay margin.Results show that each system achieves consensus when all the delays are bounded by the delay margin while diverging with all the delays exceeding the delay margin.By extending this method,and studying the monotonicity of high-degree polynomials transformed from characteristic equations of delayed highorder systems,consensus conditions are derived for first-to-sixth-order multi-agent systems in the form of delay upper bounds with inequalities respecting the parameters,and general solutions for arbitrary-order systems are presented in the form of set-valued conditions.Compared with the existing literatures where Lyapunov approaches are chosen for consensus analysis of high-order multi-agent systems with nonuniform time delays,the results in this work are less conservative.(2)By analyzing the rotational consensus motion of second-order systems with nonuniform time delays in two-dimensional plane,each of which is figured out to have different stability conditions from its corresponding system presented in complex plane,the connection between them is proven: the smaller delay margin of the complex-valued system and its complex-conjugate system's,is also the one for the corresponding system discussed in two-dimensional plane,whose eigenvalue set is the union of the two conjugate systems'.However,existing papers have not observed this phenomenon while directly applying simulations on the two-dimensional systems(which doesn't contain complex state variables)to verify results derived upon complex systems;this action may lead to incorrect results if time delays are involved.The discovery of this fact gives theoretical support to the dimensionality reduction treatment of two-dimensional systems in complex plan,which helps to simplify the stability analysis.On the other hand,the thesis also discusses the effect of input delays on the rectilinear consensus motion,and figures out that the delayed systems cannot stay in motion in the absence of input delays,which is illustrated via simulations.(3)A cooperative control method for agents to sweep-cover a priori unknown corridorshaped areas with curved borders has been designed based on consensus principles.The method guarantees flexible formation of the agents by deploying them with a virtual spring model,and resolves the issue in the existing literatures that their algorithms don't apply to areas with dramatically varying width due to the fixed formation of the agents.A second-order continuous-time and a non-holonomic discrete-time sweep coverage control protocols are proposed respectively.The latter produces smoother motion path than other coverage control policies because the limitation of the agents' velocities and turning angles are considered,and the agents are not supposed to turn back.And this is beneficial to its applications on unmanned aerial vehicles(UAVs).Based on these algorithms,further solutions to the coverage problems of U-shaped,non-smooth corridors and extremely wide ones with inadequate UAVs are provided by applying the algorithms of navigation,obstacle avoidance and virtual borders construction respectively.Simulation results show that these solutions can effectively augment the adaptivity of the proposed algorithms while preserving the smoothness of the motion paths.(4)Based on the aforementioned achievements,the dissertation comes up with fine control schemes for autonomous UAV cooperation.By developing a fuzzy controller with information extracted from training patterns,and employing consensus formation algorithms,two multi-rotor UAVs achieves fine inspection of high-voltage transmission lines with role-switch and cooperative obstacle avoidance mechanisms designed upon the both-side-flight strategy,and experiments are carried out to testify the efficacy of the method.With shared-information-passing algorithm,position-offset estimation formula and virtual-border-division coverage strategy,the proposed sweep-coverage method has been improved by avoiding remote communication,and applied on autonomous multiUAV cooperation in forest past monitoring and accurate pesticide spraying;simulation experiments are conducted to display the effectiveness of the methods.The demonstrated applications can reduce labor cost and improve work efficiency by solving the existing problems of inefficiency or potential safety risk in manual operation when tasked with single or non-autonomous UAVs.