Research On The Method For Forming Complex Swarm-robot Transport Formations

Cooperative transport is a challenging problem in swarm robotics, and two steps are often used for the problem. Firstly, all handling-points are found according to the shape of the transported object and the number of the robots in systems and each robot finds a handling-point which is different from other robots’ and moves to the point so that a handling formation can be achieved. Secondly, the object is transported from the initial position to the desired one by the transport formation. This paper focuses on the first step of the cooperative transport problem, namely, to form a complex transport formation. The main contributions are as follows.Firstly, a zero element row augmenting path algorithm is proposed for linear assignment problems, because for forming transport formations, it is a good idea to determine each robot’s handling-point by the solutions of linear assignments.Secondly, a method based on bottleneck-linear assignment is proposed for forming complex swarm-robot transport formations, which works as follows. Firstly,the optimal paths from a robot to all handling-points, the basis of handling-points assignment, are calculated by an optimal path two-selection method developed in this paper. Then, a different handling-point is assigned to a different robot according to the bottleneck-linear assignment proposed, which can optimize both the sum of the traveled paths of all robots and the time to form the handling formation. The method for attractive segments and points in the existing artificial moment method is also improved. As a result, each robot can almost move along the optimal path to its handling-point under the control of its artificial moment motion controller.Thirdly, in the method based on bottle-linear assignment, because of the needs of a master-robot and relatively lager computational burden at the initial stage, a method based on artificial social position is also proposed. For each robot Ri at each sampling time, the approach works as follows. Firstly, Ri’s handling-point is determined by an artificial social position method presented. Then, Ri’s artificial moment motion controller makes Ri move one step towards its handling-point. Because a new method for attractive segments and attractive points is also presented, the computational burden is decreased while the controller can still make each robot move safely and reach its handling-point finally. At the next time, the above steps are repeated until the desired formation is achieved. The idea of the proposed method is easy to be understood and its computational burden is low. Furthermore, no matter how complex the transport formation is, it can be always formed rapidly in a self-organizing way.Finally, extensive simulations indicate that the proposed methods are feasible and efficient.