摘要
针对足球机器人射门行为中运算的高复杂性和反应延迟的局限,引入一种基于类高斯函数的自适应神经模糊推理系统(ANFIS),用于确定最合适的射门点.系统由前件网络和后件网络构成,结合模糊逻辑理论,建立基于人类语言描述的射门行为模型.采用实际的比赛记录作为训练数据,离线地拟合系统输入与输出之间的映射关系,经训练的系统能够自动地调整前期隶属度函数的形状和后期的自适应权值.仿真结果表明,射门成功率和反应速度都能够达到预期的效果,方法的有效性得到了验证.
In order to solve the limitation of the high computational complexity and delayed reaction in the shooting behavior of soccer robots, an adaptive neuro-fuzzy inference system (ANFIS) was proposed. The proposal invokes the Gaussian-type function technology to determine the optimal shoot point. The entire system was composed of the antecedent network and consequent one. The system integrated the fuzzy logic theory, which, lead to the establishment of the behavior model described by human language. Moreover, the training samples were derived from the shoot data of actual medium competitions, along with the implementation of off-line training methods to describe the mapping relationships between inputs and outputs. Once the training process was completed, the system is able to automatically adjust the shape of antecedent membership functions, as well as the consequent weights adaptively. The simulation results demonstrate that the high shooting success rate and reaction speed can be achieved as expected, proving the effectiveness of the proposed approach.
出处
《智能系统学报》
CSCD
北大核心
2013年第2期143-148,共6页
CAAI Transactions on Intelligent Systems
基金
吉林省教育厅"十一五"科学技术研究计划资助项目(2010075)
黑龙江省自然科学基金资助项目(F200917)
黑龙江省教育厅科学技术研究计划资助项目(11553046)
关键词
类高斯函数
神经模糊推理系统
自适应性
射门点
足球机器人
Gaussian-type function
neuro-fuzzy inference system
self-adaptiveness
shooting point
soccer robot