李岳明, 万磊, 孙玉山, 张国成. 一种基于注意力机制的AUV控制层指令理解方法[J]. 机器人, 2012, 34(4): 406-410,417.
引用本文: 李岳明, 万磊, 孙玉山, 张国成. 一种基于注意力机制的AUV控制层指令理解方法[J]. 机器人, 2012, 34(4): 406-410,417.
LI Yueming, WAN Lei, SUN Yushan, ZHANG Guocheng. A Method of Instruction Understanding for AUV Control Layer Based on Attention Mechanism[J]. ROBOT, 2012, 34(4): 406-410,417.
Citation: LI Yueming, WAN Lei, SUN Yushan, ZHANG Guocheng. A Method of Instruction Understanding for AUV Control Layer Based on Attention Mechanism[J]. ROBOT, 2012, 34(4): 406-410,417.

一种基于注意力机制的AUV控制层指令理解方法

A Method of Instruction Understanding for AUV Control Layer Based on Attention Mechanism

  • 摘要: 为了更好地衔接水下机器人(AUV)规划层与控制层,参考Itti视觉注意力模型,提出了基于注意力机制的规划指令理解模型,建立了规划指令理解与再次规划环节.通过使用模糊推理的方法对指令元素特征、指令元素显著性进行分析处理,得到规划指令注意力聚焦;根据注意力聚焦进行再次规划,保证规划层关注的状态得到优先执行.仿真实验表明文章提出的规划指令理解环节实现了对规划层指令意图的获取,使规划指令得到更好地执行,有效地改善运动控制效果,提高了控制层的智能水平.

     

    Abstract: In order to maintain an appropriate relationship between the planning layer and the control layer of autonomous underwater vehicle (AUV), with Itti visual attention model considered as a reference, a planning instruction understanding (PIU) model based on attention mechanism is presented, and the processes of PIU and the secondary planning are established. The attention focus of planning instruction is obtained using fuzzy reasoning for analyzing the feature and significance of instruction element. Furthermore, the secondary planning is implemented adopting attention focus, which can ensure a higher priority execution of the state concerned by the planning layer. The simulation results demonstrate that the proposed PIU process can not only realize the acquisition of the instruction intention of planning layer, and the planning instruction can be executed more effectively, but also the motion control effect is heightened and the intelligence level of control layer is improved.

     

/

返回文章
返回