Predictive Control Of Multi-agent Formation With Dynamic Constraints In Complex Environment

With the rapid increase in the application requirements of multi-agent formations in disaster relief,military,and industrial fields,multi-agent systems face the problem of achieving formation control under various uncertain and complex environments.At present,many research works and results have been done on the problem of multi-agent formation control in obstacle environments,but mainly for simple obstacle environments.However,problems such as complex environment,limited environmental information,and changing formation mission have led to dynamic changes in obstacle avoidance constraints and formation constraints in multi-agent cooperative formation control.The ability of formation control strategy is a key problem to be solved urgently.In this paper,the collaborative predictive control of multi-agent formation under static complex environment and unknown obstacle environment will be studied based on receding horizon strategy.For the problem of multi-agent formation control in complex obstacle environment,under the virtual leader formation strategy,a control algorithm with offline path planning and online trajectory tracking is proposed.In path planning,a mixed integer obstacle avoidance constraint is designed for polygon obstacles,and a mixed integer quadratic programming problem is formed to solve the reference trajectory of virtual leader.In addition,the virtual leader replacement strategy and the next feasible state search strategy are designed to avoid one-side obstacles and crowded obstacles.In trajectory tracking,a distributed model predictive control algorithm is designed to achieve formation.The stability of the system is guaranteed by applying compatibility constraints and terminal invariant set constraints in the optimization problem.For the problem of multi-agent formation control in unknown environment,the navigation control of single agent in unknown environment is considering firstly,a navigation control strategy based on a rolling window is proposed.For the unknown and variability of the obstacle environment,a hyperplane obstacle avoidance constraint is designed based on the local obstacle information detected by the agent.Considering the local detection ability of agent,a rolling window strategy is proposed for real-time local planning.The local target point optimization based on rolling window and target point tracking based on the model predictive control method realize the movement and obstacle avoidance of agent.Based on the single-agent navigation control,the multi-agent formation problem is further considered,and a multi-agent cooperative formation control method in unknown environment is established.Adopting the formation control strategy of the leader-follower method,the multi-agent formation is realized through the navigation control of the leader and the trajectory tracking control of the follower.In navigation control,based on the environmental information passed by the followers,a local target point considering formation is optimized through the formation prediction method to ensure the obstacle avoidance of the entire formation.In trajectory tracking control,based on the trajectory information passed by the leader,a model predictive control method is used to solve the control input which satisfies various constraints.