融合眼电及头部姿态的智能轮椅控制

Intelligent-Fusion Wheelchair Control Based on EOG and Head Posture

摘要:

目前已有的轮椅融合控制系统，为肢体残疾人提供了多种操纵策略。多数研究只从提升控制成功率的角度开展，少有从轮椅使用的安全性和舒适性角度开展的研究。为满足特殊群体对轮椅安全性和舒适性的需求，提出了一种新式轮椅控制策略，结合眼电信号和头部姿态变化进行融合控制。头部姿态角度变化适合用于轮椅的实际导向，符合人们的使用习惯；眼电分眨眼和眼动两种，眨眼行为快速且动作细微，适合作为需频繁使用的控制确认信号，而眼动行为不易发生误操作，因此适合用于发送求救信号。设计了一个集成眼电采集和头部姿态检测的护目镜，位于轮椅处的树莓派通过小波变换特征提取和随机森林算法分类，实现不易疲劳、操纵性好且安全的轮椅控制系统。实验表明，该种控制策略能在识别成功率达到92%的基础上，最大程度满足对安全性和舒适性的要求。

Abstract:

At present, the existing wheelchair fusion control system provides a variety of control strategies for the disabled. Most of the studies were only carried out from the perspective of improving the success rate of control, and few were carried out from the perspective of the safety and comfort of wheelchair use. In order to meet the needs of the special group for the safety and comfort of the wheelchair, a new wheelchair control strategy is proposed, which combines the eye electrical signal and the head posture change for the fusion control. The change of head posture Angle is suitable for the practical guidance of wheelchair and is in line with people's habits. Electro-Oculogram (EOG) can be divided into two types: blink and eye movement. The blink behavior is rapid and subtle, which is suitable for frequent use as a control confirmation signal, while the eye movement behavior is not prone to misoperation, so it is suitable for sending distress signals. The raspberry pie located in the wheelchair is classified by wavelet transform feature extraction and random forest algorithm to realize the wheelchair control system that is not easy to fatigue, has good maneuverability and is safe. The experiment shows that this control strategy satisfies the requirements of safety and comfort to the greatest extent on the basis of the recognition success rate of 92%.