首页|基于目标检测的室内自动视觉寻物系统设计

基于目标检测的室内自动视觉寻物系统设计

扫码查看
针对人们日常的寻物需求,文章设计了一款基于视觉目标检测的室内自动寻物系统.该系统主要由视觉信息采集、目标检测算法和云台运动控制组成.其中,ESP32 读取OV2640 摄像头拍摄的室内场景视频图像,PC服务器接收ESP32 上传的图像.随后,文章采用YOLOv5 目标检测算法确定目标物的空间位置,并在图像中检测红外激光发射器投射的光斑,通过增量式PID算法计算两者的位移偏差,由串口向STM32F103C8T6 主控芯片发送运动指令,驱动 2 台舵机调整二自由度云台位姿,最终使得光斑能够对准目标物,形成寻物引导信号.实际测试结果验证了系统的有效性,该设计为室内自动视觉寻物提供了一种新的解决方案.
Design of automatic system for indoor searching lost items based on visual object detection
Aiming at the daily demand of searching lost items,an automatic system for indoor searching lost items based on visual object detection is designed in this paper.The system is made up of the collection of vision information,algorithm of object detection and the motion control of pan-tilt.Specially,video images of indoor scene are read from OV2640 camera by ESP32.PC server receives images uploaded by ESP32.Spatial locations of a target are determined by YOLOv5 object detection algorithm.Laser spots cast by an infrared laser emitter are detected in images.Position offsets between the locations of the target and the laser spot are computed through incremental PID algorithm,and then the motion commands are send from PC server to STM32F103C8T6 chip through serial port.Two servos are driven so as to regulate the posture of the two-degree-of-freedom pan-tilt.Eventually,the laser spot are thrown to the target,and it turns out to the guiding signals for finding lost items.The effectiveness of the system is validated by the results in the real world.It offers a new solution for indoor searching lost items based on visual object detection.

indoor searching lost itemsobject detectiontwo-degree-of-freedom pan-tilt

曹国锋、殷文慧、陈雅婷、郝孝颉、韩腾昊、罗丰瑞、钱诚

展开 >

常州工学院,江苏 常州 213032

室内寻物 目标检测 二自由度云台

国家自然科学基金江苏省大学生创新创业训练计划

61602063202311055067Y

2024

无线互联科技
江苏省科学技术情报研究所

无线互联科技

影响因子:0.263
ISSN:1672-6944
年,卷(期):2024.21(10)