摘要
提出了一种基于递阶行为,并将重点放在对于基本行为的设计以及协调融合上的控制算法结构。基本行为由模糊推理系统设计,较高级的复杂行为由基本行为的融合产生。采用神经网络来模仿人类行为,学习人类复杂的决策技能,从而达到对基本行为进行融合的目的,以减少行为之间的强硬转换给整个系统带来的不良影响。采用不同的任务类型进行了仿真以测试这种基于模糊逻辑以及神经网络的递阶行为控制系统的性能,仿真结果表明该算法具有令人满意的性能。
A new hierarchical fuzzy behavior-based control approach of an autonomous robot was proposed, which focused on the design on coordination and fusion of these elementary fuzzy behaviors.The elementary behaviors were achieved by fuzzy reasoning scheme, and high-level behaviors were composed of these primitive behaviors. Neural networks were used to select and fuse different behaviors like a person, so that the motion speed and rotational velocity of the mobile robot can change smoothly, because the sharp shift of different behavior would exacerbate the absolute position errors. The feasibility of the proposed design is validated by different simulation experiments.
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2005年第4期391-397,共7页
Journal of Jilin University:Engineering and Technology Edition
基金
"863"国家高技术研究发展计划项目(2003AA735084)
关键词
自动控制技术
移动机器人
导航
递阶行为
模糊推理
神经网络
automatic control technology
mobile robot
navigation
hierarchical behavior
fuzzy reasoning
neural network
