Multi-robot Dynamic Task Allocation And Motion Planning Method Under Localization Uncertainty

Multi-robot systems provide innovative solutions for smart factories,rescue,environmental monitoring,etc.Task allocation and motion planning are key technologies of multi-robot systems,while the localization uncertainty of robots restricts the application of task allocation and motion planning methods in practice.Therefore,this paper conduct research on multi-robot task allocation and motion planning methods under robot localization uncertainty.The main contents are as follows.(1)A multi-robot obstacle avoidance model and a collision avoidance model are designed to address the problem of obstacle avoidance and internal collision avoidance in multi-robot systems under localization uncertainty.First,the influence of localization uncertainty on the motion of mobile robots is analyzed and a kinematic robot model under localization uncertainty is constructed.Next,a robot obstacle avoidance model and a multi-robot internal collision avoidance model under localization uncertainty are constructed,taking into account the localization uncertainty factor.Finally,the simulation verifies the validity of the obstacle-avoidance model and the multi-robot internal collision avoidance model.(2)A multi-robot task allocation method based on the parallel chaotic adaptive cuckoo algorithm is studied to address the problem that localization uncertainty introduces additional costly,probabilistic form of constraints to multi-robot task allocation,which leads to infeasible results of the original multi-robot task allocation.First,the impact of localization uncertainty on the multi-robot delivery task allocation problem is analyzed,and a multi-robot delivery task allocation model under localization uncertainty is constructed.Next,considering that multi-robot task allocation is a discrete optimization problem,this paper improves on the cuckoo search algorithm.The adaptive neighborhood search strategy based on Lévy flight,the preferred random wandering strategy based on competition mechanism,and the chaotic perturbation strategy based on tolerance degree are studied to improve the solution accuracy.And then,a parallel strategy based on multi-core processors is designed to speed up the algorithm.Finally,the effectiveness of our method is verified by comparison on the Solomon dataset.The comparison results show that the solution stability of the method in this paper is high and the solution quality is good.(3)A priority-based model predictive control method is proposed to address the problem that it is difficult to plan high-quality trajectories in real-time for multi-robot systems under localization uncertainty.First,a conflict search algorithm is used to plan the initial guidance trajectory of the robot.Next,based on the initial guidance trajectory,the robot obstacle avoidance constraints are reconstructed using a series of convex regions that do not overlap with the obstacles in the environment to simplify the obstacle avoidance constraints.In addition,the multi-robot internal collision avoidance constraint is reconstructed to be applicable to larger scale multi-robot systems,considering the robot operation speed.Next,considering the balanced multi-robot motion planning,a priority-based rolling planning strategy is proposed to solve the multi-robot motion planning in groups,where the robots with fewer passing initial guide trajectory points have higher priority.Finally,the simulation in the robot operating system verifies the effectiveness of this method,and the comparison results show that this method can better balance the quality of motion trajectory and real-time computation.