基于模糊神经元网络的智能车辆个性自动驾驶系统的设计与实现
Design and Implementation of a Neruo-Fuzzy based Control System Intelligent Vehicles
-
摘要: 介绍了一种利用模糊神经元网络实现车辆自动驾驶的设计方案.其基本设计思想 是首先通过模糊逻辑描述驾驶者的驾驶行为,然后利用驾驶者实际驾驶时采集的车辆运行情 况作为训练数据,通过神经元网络的自学习功能修改和改进模糊控制所需的输入/输出信 号的隶属度函数以及模糊推理的运算关系,做到简单控制实现与复杂学习算法的有效结合, 从而实现模糊神经元控制.本方案为智能车辆实现个性化自主或辅助自动驾驶提供了一种非 常有效的机制.Abstract: This paper presents an innovative method for designing and implementing automatic vehicle control system for intelligent vehicles using neuro-fuzzy network. The basic design and implementation procedure is as following: 1) Describing driving behaviors of a human driver under various situations using fuzzy logic-based rules; 2) Improving the driving performance using a neural networks specifically conduction according to the fuzzy driving rules (this type of neural networks are called neuro-fuzzy networks, and can be used to refine the input/output membership functions and fuzzy reasoning operators); 3) Implementing the fuzzy driving rules in the driving control hardware, and the neuro-fuzzy networks in the host control computer. In this way, the low cost control implementation and sophisticated refinement and learning capability can be achieved effectively. The proposed method can be utilized for individually automatic driving for intelligent vehicles. Experimental studies used VISTA vehicles had been conducted to verify the suggested advantages.
计量
- 文章访问数: 2619
- HTML全文浏览量: 101
- PDF下载量: 1729
- 被引次数: 0