基于CNN-GRU-Attention的自动扶正导盲杖设计  

Design of an Automatic Self-righting Smart Cane Based on CNN-GRU-Attention

作　　者：刘红燕 于鑫 李为梁 修霞 赵玺博 LIU Hongyan;YU Xin;LI Weiliang;XIU Xia;ZHAO Xibo(School of Mechanical and Automotive Engineering,HENG XING University,Qingdao 266100,China;School of Law,Inner Mongolia University of Finance and Economics,Hohhot 010051,China;School of Data Science and Applications,Inner Mongolia University of Technology,Hohhot 010051,China)

机构地区：[1]青岛恒星科技学院机械与汽车工程学院,山东青岛266100 [2]内蒙古财经大学法学院,内蒙古呼和浩特010051 [3]内蒙古工业大学数据科学与应用学院,内蒙古呼和浩特010051

出　　处：《青岛大学学报(工程技术版)》2025年第1期72-78,共7页Journal of Qingdao University(Engineering & Technology Edition)

摘　　要：针对传统导盲杖功能单一、安全性不足等问题,开发了基于CNN-GRU-Attention的自动扶正导盲杖,系统以STM32微控制器为核心,选用了ESP8266 WiFi模块、GPS、GSM模块、K210摄像头、MP160型生理仪和EMG100C放大器等实时采集处理视觉、生理和运动数据等模块实现。系统采用CNN-GRU-Attention混合架构模型,通过图像识别技术自动调整杖体姿态,提升导航精度和安全性。实验结果表明,在自动扶正处理过程中,准确率高、收敛迅速和适应性强。与传统导盲杖相比,实现了自动避障和路径规划,通过实时监测和分析使用者的行走意图及步态,提供个性化辅助。In response to the issues of traditional guide canes having limited functionality and insufficient safety,an automatic self-correcting guide cane based on CNN-GRU-Attention is developed.The system is centered around an STM32 microcontroller and integrates real-time data acquisition and processing modules such as the ESP8266 WiFi module,GPS,GSM module,K210 camera,MP160 physiological instrument,and EMG100C amplifier to collect visual,physiological,and motion data.The system employs a CNN-GRU-Attention hybrid architecture model,using image recognition technology to automatically adjust the cane’s posture,enhancing navigation accuracy and safety.Experimental results show that when processing cane angle data,the system features high accuracy,fast convergence,and strong adaptability.Compared to traditional guide canes,this system realizes automatic obstacle avoidance and path planning,offering personalized assistance by continuously monitoring and analyzing the user’s walking intentions and gait.

关 键 词：自动扶正导盲杖 跌倒预防系统 意图识别算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]