Multi-robot Formation Based On Gaussian Process Motion Planning And Virtual Structure Method

Multi-robot systems have been widely used in many tasks,such as rescue,transportation,patrol,and so on.Compared with single robot systems,multi-robot systems have many advantages,such as good robustness,high execution efficiency,strong fault tolerance,and good adaptability.In some cases,multi-robot groups need to move in an appropriate formation to cooperatively complete specified tasks.Therefore,as a key issue in cooperative control of multi-robot systems,multi-robot formation has received widespread attention.In the past,most of the research was devoted to the problem of multi-robot formation control,and the motion planning was realized through local reaction,with poor global and optimality.Motion constraints and trajectory optimization were often ignored,which was difficult to apply in actual robot formation.In this paper,the real-time global motion planning method is adopted in the formation of multi-robot,considering the speed and other limitations.Based on the Gaussian process model,factor graph inference is performed on motion constraints and trajectory optimization problems.Aiming at several formation problems,designing corresponding motion planning and controller,building a multi-robot formation system,and conducting experimental verification on multi-mobile robot platforms.The specific contents are as follows:1.Aiming at the problems of large computation amount and poor real-time of motion planning for multi-robot systems in complex obstacle environments,Gaussian process motion planning algorithm(GPMP2)is extended to multi-robot situations.The motion trajectory of each robot in the multi-robot system is represented as a continuous time Gaussian process model,which can be quickly solved through the sparsity of the Gaussian process model,thereby improving the efficiency of motion planning.2.Aiming at problems such as collision avoidance,obstacle avoidance,and kinematic constraints,motion planning is considered as probabilistic reasoning.Then,the trajectory optimization problem is transformed into a factor graph for reasoning,and corresponding prior factors are constructed on the factor graph.Bayesian tree model is used to further achieve fast incremental reasoning for factor graphs.Finally,the effectiveness of the algorithm is verified by the designed multi-manipulator and multi-mobile-robot simulation experiments.3.Aiming at the problem of poor robustness and stability of multi-robot formation,the formation of multi-robot is considered as a rigid structure,and a formation strategy based on virtual structure method is studied.Construct a virtual structure model and establish its kinematic model.Then,a Gaussian process motion planning algorithm is used to plan the motion of the virtual rigid body model.Finally,the feasible path is centrally planned while maintaining the stability of the formation during movement.4.Aiming at the problem of weak formation coordination ability of multi-robot platforms in practical environments,a multi-robot formation system based on the combination of Gaussian process motion planning algorithm and virtual structure method proposed in this paper is deployed on the multi-robot platform.The performance of the proposed method in practical environments is tested,and the algorithm is optimized and adjusted based on the test results,improving the collaborative ability of the multi-robot system.