基于STM32的视觉识别导盲机器人设计  

Design of Visual Recognition Guide Robot Based on STM32

作　　者：刘梦禹 田志宏[1] 朱涛[2] LIU Mengyu;TIAN Zhihong;ZHU Tao(School of Electronic Information and Automation,Tianjin University of Science and Technology,Tianjin 300222,China;School of Artificial Intelligence,Tianjin University of Science and Technology,Tianjin 300457,China)

机构地区：[1]天津科技大学电子信息与自动化学院,天津300222 [2]天津科技大学人工智能学院,天津300457

出　　处：《自动化仪表》2024年第2期101-105,共5页Process Automation Instrumentation

基　　金：国家级大学生创新训练计划基金资助项目(202210057014)。

摘　　要：针对现阶段导盲产品功能单一、不易携带等问题,对导盲机器人领域进行了研究。采用传感器与视觉识别技术,设计了一种智能化移动式导盲机器人。在传统避障功能的基础上,创新性地添加了定位与图像识别功能。机器人以STM32为主控制器,采用伺服系统运动设计,在前进途中获取超声波传感器反馈的障碍物信息以实现避障。同时,可通过串口与全球定位系统(GPS)通信获取经纬度信息。搭载的Jetson Nano采用你只看一次(YOLO) v3算法对摄像头扫描得到的图像进行识别,达到GPS定位精度为2.5 m、摄像头平均检测准确率为88.33%的性能指标。以木箱避障与计算机端图像检测为试验案例,验证设计软硬件方案的可行性。该设计能够提高盲人出行的安全系数。Aiming at the problems of single function and not easy to carry of blind guide products at the present stage,the field of blind guide robot has been studied.Using sensor and visual recognition technology,an intelligent mobile guide robot is designed.Based on the traditional obstacle avoidance function,it innovatively adds the localization and image recognition function.The robot uses STM32 as the main controller,adopts servo system motion design,obtains the obstacle information fed back by ultrasonic sensors on the way forward so as to realize obstacle avoidance.At the same time,it can communicate with the global positioning system(GPS)through the serial port to obtain the latitude and longitude information.Equipped with the Jetson Nano adopts you only look once(YOLO)v3 algorithm to recognize the images scanned by the camera,which achieves GPS positioning accuracy of 2.5 m,the camera average detection accuracy of 88.33%of the performance indicators.The wooden box obstacle avoidance and computer-side image detection are used as test cases to verify the feasibility of the designed hardware and software programs.The design can improve the safety factor of blind people traveling.

关 键 词：自动化技术 导盲机器人 STM32 Jetson Nano 避障 视觉识别 

分 类 号：TH139[机械工程—机械制造及自动化]