多自治水下机器人（Autonomous Underwater Vehicle, AUV）协作目标搜索是实现水下救援、水下侦查、水下攻防等任务的重要环节，是水下机器人研究的核心内容之一。本项目对多AUV协作目标搜索控制进行研究，探索多AUV搜索中的协作任务分配以及搜索路径规划等问题，为形成系统化的多AUV协作目标搜索控制算法和实际有效的控制系统奠定基础。首先，针对多AUV的协作任务分配问题，给出一种栅格信度自组织神经网络，为每一个AUV动态的分配任务，增强AUV之间的协作，避免冲突；其次，引入生物启发神经网络模型，给出生物启发多AUV搜索策略，解决多AUV对未知目标搜索的路径规划问题。最后，考虑到水下环境的特殊性，在生物启发神经网络中嵌入速度矢量合成算法，通过运用向量相关的知识对AUV航行方向进行调整，克服海流对AUV航行的影响。通过系统仿真研究和水池实验验证AUV控制的有效性。

Multiple autonomous underwater vehicles (multi-AUV) cooperative target search is the important element to realize underwater rescue, underwater detection, underwater offensive and defensive. The target search is one of the key content of underwater vehicle research. This project, focusing on cooperative task assignment and search path planning during the search process, aims to lay a sound foundation for a systematic algorithm which may contribute to a real effective control system for the cooperative multi-AUV team. In this study, an improved self-organizing map method is adopted by using the grid belief function. This method makes a dynamic assignment among multi-AUV team members, thus determine responsibilities of each AUV to avoid a ask conflict. For better cooperative search, a biologically inspired neural network model is also taken for path planning. Finally, considering the particularity of underwater environments, an integrated algorithm.is proposed by introducing a velocity vector synthesis algorithm into the biologically inspired neural network. This integrated algorithm is taken to keep AUVs on their planned tracks to offset ocean currents’ influence by adjusting AUVs’ moving directions. Through the system simulation research and pool experiments verified the effectiveness of the control AUV.