摘要
针对传统的基于行为的智能轮椅的路径规划方法在室外非结构环境下的路径规划效果差的问题,提出一种新的智能轮椅的路径规划算法.该算法利用模糊逻辑设计了基本控制行为,并在此基础上结合大量实际经验使用神经网络设计了行为协调控制器.改进的算法将仲裁机制和命令融合机制2种行为协调方法有效结合起来,并吸收了这2种行为协调方法的优点,从而改善了系统的反应速度,极大提高了控制精确;另一方面,该算法还可以识别陷阱区域并通过自主改变行为的权重方法控制轮椅逃出陷阱区域,因而具备了较强的人工智能特征.仿真和实物实验验证了该算法智能性高且实现简单,适用于室外非结构化环境下的机器人路径规划.
In order to solve the poor effects of traditional behavior-based path planning of intelligent wheelchairs in an outdoor unstructured environment, a new path planning method was proposed in this paper. The new algorithm uses fuzzy logic to design basic control behavior, and on this basis applies a neural network to design behavior coor- dination by combining a large amount of practical experience. The improved algorithm can combine arbitration mechanisms with fusion mechanisms successfully; it absorbs the major advantage of these two original algorithms and improves response speed of the system while enhancing the control accuracy significantly. On the other hand, the method can identify trap area and control the wheelchair escape from the trap by changing the behavior weights independently, therefore displaying strong artificial intelligence characteristics. The simulation and real experimental results verify that the algorithm is capable of advanced intelligence and can be implemented easily. Additionally, it can be used in an outdoor unstructured environment for robot path planning.
出处
《智能系统学报》
2011年第5期456-463,共8页
CAAI Transactions on Intelligent Systems
基金
科技部国际合作资助项目(2010DFA12160)
重庆市科委资助项目(CSCT
2010AA2055)
重庆邮电大学青年基金资助项目(A2009-50)
关键词
机器人
智能轮椅
非结构化环境
路径规划
神经网络
行为协调控制器
robot
intelligent wheelchair
unstructured environment
path planning
neural network
behavior coordination control
