Formation,Collision Avoidance,and Multi-objective Control In Multi-agent System Cooperation

Multi-agent system has been effectively utilized in various areas such as reconnaissance,surveillance,exploration,industrial manipulation,intelligent transportation and so on.Distributed cooperative control strategies have drawn increasing attention owing to the fact that such strategies provide attractive solutions to deal with large-scale robots.The main idea of coordinating a group of mobile robots is to design a safe and efficient control strategy which accomplishes several task-based objectives.The high applicability of multi-agent system relies on efficient information exchange and collision free cooperative motion among agents.Therefore,maintenance of information exchange as well as collision avoidance are basic requirements for cooperative control of multi-agent system.The main content of this thesis can be summarized as follows:1.A distributed optimal formation control strategy has been proposed for secondorder linear system using the estimated position information other than the absolute position of neighbors.The energy cost of each agent indicates the minimal travelling distance during the formation process which is minimized by solving the algebraic equation.The estimated position information based control strategy is able to drive the agent to the designated formation,and is optimal if the position estimator is stable.Effectiveness of the formation control strategy is proven by simulation.2.Observed from the simulation result of afore mentioned formation control strategy,minimization of cost function brings about trajectory overlap indicating collision among agents.For this reason,a distributed consensus algorithm is presented based on the position estimation law and an extra collision avoidance potential field.The proposed consensus law realizes collision avoidance and the convergence to the desired static formation in a cooperative manner.Numerical simulation results are presented to demonstrate the effectiveness of the proposed consensus algorithm.3.Restricted by nonholonomic constraint,nonholonomic system has an equilibrium manifold other than an isolated equilibrium point,which makes the control of such system difficult.For this reason,coordinating a group of nonholonomic agents to achieve desired formation,heading angle alignment and obstacle avoidance objectives is considered.The formation is constructed under l-? configuration,and the proposed distributed controller is designed as multiplying gradient vector of Lyapunov function by a skillful designed coefficient matrix.A modified potential field method is further provided to guarantee collision-free motion,and to overcome local minima.Numerical simulation results demonstrate the effectiveness and advantages of the proposed method.4.A single controller is designed for nonholonomic robot system to synthesize formation,connectivity,and collision avoidance objectives,yielding a distributed gradientbased control solution using only local information.A novel class of Lyapunov-like barrier functions which encode the inequality constraints of collision avoidance is introduced into the controller providing continues change in velocity.The robot achieves collision avoidance and connectivity maintenance objectives when satisfying the corresponding barrier inequality constraints.Simulation results are included to verify the feasibility of the formation,connectivity maintenance,and collision avoidance performance.5.Experimental results are presented to demonstrate the practicality of the proposed cooperative control algorithms in preceding three chapters.The experiments are implemented on E-puck robot platform with computer vision providing position information through wireless local network.The positioning data is import into MATLAB to derived robot trajectories in experimental environment,and the video snapshots are provided for better demonstration.In the end,main results of this thesis are summarized,and the prospects for the future research are presented.