Research On Multi-agent 3D Collaborative Control Algorithm Under Visual Perception Constraint

With the rapid development of society and technology,more and more fields are no longer satisfied with simple single-agent control.Therefore,more flexible and reliable multi-agent systems have entered the field of vision of scientific researchers and technicians.Multi-agent systems Cooperative control technology has broad development prospects in many fields such as aerospace and intelligent transportation in the future.However,most of the current researchers focus on controlling multi-agent systems in twodimensional space.This is because in three-dimensional space,there are many degrees of freedom in the agents and it is difficult to design control algorithms.This paper aims to propose a general strategy for controlling the cooperation of multiple agents in 3D space.Considering the communication limitation in special cases,this paper uses vision sensor as the means of multi-agent communication.To do this,each agent is required to stay within the field of view of its neighbors’ visual sensors,which is called the visual perception constraint.In order to always maintain the visual perception constraints among multi-agents and complete the formation,cluster and consistency collaborative tasks,this paper proposes multiple cost functions with different functions,and each cooperative task cost function is composed of sub-cost functions related to relative distance and relative angle.Among them,the cost function related to the relative angle can prevent the agent from entering the blind spot of its neighbors,that is,the relative angle between each pair of neighbor nodes in the multi-agent is greater than the visual blind angle.The cost function related to the relative distance can not only maintain the relative distance between agents within the maximum field of view radius,maintain the connectivity of visual perception,but also realize the three collaborative tasks of multiagent formation,cluster and consistency.In this paper,a general multi-agent cooperative controller is designed based on Lyapunov stability theory and generalized gradient theory.In order to verify the versatility and stability of the controller,in addition to using the singular perturbation technique for theoretical proof,this paper also designs three simulation experiments of formation,cluster and consistency,and analyzes the simulation results one by one.Finally,we applied the cooperative controller to the unmanned vehicle experimental platform,proving the practical feasibility of the controller.