运动学约束条件下多AUV任务分配算法

Task assignment for a multi-AUV system under kinematic constraint

针对运动学约束的自治水下机器人(AUV)任务分配与路径规划问题,本文以多个AUV组成的系统为研究对象,将Dubins Path算法与改进的自组织映射(SOM)神经网络算法相结合,提出一种在运动学约束条件下多AUV任务分配与路径规划算法。通过SOM神经网络方法对多AUV进行任务分配后,若存在运动学约束或障碍物而导致无法进行Dubins路径规划时,则重新进行任务分配,直到所有目标点都有AUV到达。仿真结果表明该算法能够有效完成运动学约束条件下多AUV的任务分配。

To solve the task assignment and path planning problems of a system with multiple autonomous under wa-ter vehicles （AUVs）, a multi-AUV task assignment algorithm under kinematic constraints is proposed, which com-bines the Dubins Path algorithm with an improved SOM（self-organizing map） neural network algorithm. The tasks were first assigned by the SOM neural network. If there were kinematic constraints or obstacles that led to the failure of Dubins path-planning, task re-assignment was implemented until the AUVs reached all the target points. Simula-tion results show that the algorithm can effectively accomplish task assignments for a multi-AUV system under kine-matic constraints.