Design And Implementation Of Tracking And Formation Predictive Control For Networked Multi-agents

As a research branch of Networked Multi-Agents distributed coordination control theory,trajectory tracking control and formation control of Networked Multi-Agents have been paid more and more attention by researchers in recent years with its wide application prospect both in military and civil fields.Considering the structural characteristics of Networked Multi-Agents system,that is,the controller,the sensor,the actuator and the agent are in the spatial distribution state,thus the control system needs to adopt wireless communication network to realize the transmission of various control information and the state information.At the same time,the introduction of wireless communication network will inevitably lead to communication delay and packet loss into the control loop,these communication problems will obviously affect control performance of the system.Therefore,this thesis studies the problem of trajectory tracking control of networked agents and formation control of networked multi-agents in the case of communication delay and data packet loss.Firstly,a local tracking control law for ideal communication network is proposed for a class of networked agents described by nonlinear system model.When constant time-delays exist both in the forward and feedback channels of the networked agent,the local tracking control law is extended to the remote predictive tracking control law based on the idea of networked predictive control to actively compensate the bilateral timedelays and consecutive data packet losses.The simulation and experimental results show that the proposed remote predictive tracking control scheme can effectively compensate the communication delays exist in bilateral communication channels of agent.In order to verify the compensation ability of the remote prediction tracking control method for consecutive data packet dropouts and to make the experimental results more intuitive,the experiment of crossing the tunnel is designed.When the networked agent traverses the tunnel area,consecutive data packet losses happens in the feedback channel of agent.The remote predictive tracking control scheme makes the networked agent pass through the tunnel area smoothly,which indicates that the proposed control method can actively compensate for consecutive data packet losses.Secondly,the master-slave predictive formation control problem of a group of networked multi-agents is studied.When the communication capability of some agents is limited,and different constant time-delays exist in the feedback channels of each agents and in communication channels between the master and the slave agents,a master-slave predictive formation control method is proposed.In the master-slave predictive formation control problem,there are different communication delays in the feedback channel of each individual agents and in the communication channels between the master and slave agents,and only the master agent has access to the formation reference states.Each agent actively compensates the time-delay in its own feedback channel by means of iterative prediction.At the same time,the master agent sends its predicted future states to all the slave agents to actively compensate for the time-delay exist in communication channels between the master and slave agents.Furthermore,when the data communication capability is strong for each agents,this thesis studies the motion coordination predictive formation control problem of a group of networked multi-agents.In this problem,there are different communication delays in the feedback channel of each agents and in the communication channels between agents.Each agent uses the delayed state of its neighbor agents' to predict its own state in current time,and use predictive formation control law to derive its own control inputs.Since the prediction state of the other agents is required in the calculation of the control inputs of each agent,the anti-disturbance of the whole formation is strong.The feasibility and effectiveness of the two predictive formation control methods are verified by numerical simulations and experiments respectively.Finally,in order to realize the experimental verification of the control method proposed in this thesis,a networked control experimental system is designed and established.The experimental system uses Simulink in Matlab as the design platform of the control algorithm to improve the design efficiency of the control algorithms and minimum the error probability while designing the control algorithms.Design and development of a micronetworked controller,named as Net Con-STM32,and which is used as the implementation platform for Simulink control block diagram.At the same time,a networked monitoring configuration software Net Con Top is used as a monitoring platform for control algorithms to realize real-time monitoring for running status and online parameter configuration for Simulink control block diagram.In addition,the networked control experiment system uses Vicon visual positioning system as a global sensor,and the communication data structure between Vicon server and Net Con-STM32 networked controller is designed to realize the acquisition and reception of each agent's states information.The wireless local area network is built to realize the data transmission of various functional units in the networked control experiment system,then the whole experimental system is realized.