基于姿态辨识的遥操作液压工程机器人视觉提示  

Visual Cue for Teleoperation Hydraulic Engineering Robot Based on Posture Recognition

作　　者：张文斌 李笑[1] ZHANG Wenbin;LI Xiao(School of Electromechanical Engineering,Guangdong University of Technology,Guangzhou Guangdong 510006,China)

机构地区：[1]广东工业大学机电工程学院,广东广州510006

出　　处：《机床与液压》2021年第3期20-24,共5页Machine Tool & Hydraulics

基　　金：广东省计算机集成制造重点实验室资助项目(2014B030301072);广东省计算机集成制造重点实验室开放基金项目(CIMSOF2016012)。

摘　　要：针对遥操作液压工程机器人视频传输时延问题,提出一种基于姿态辨识的遥操作液压工程机器人视觉提示方法。利用摄像机获取遥操作液压工程机器人图像,通过关键点识别堆叠沙漏网络(KPR-SHN)模型和姿态辨识神经网络(AR-NN)模型,获取遥操作液压工程机器人液压缸的位移,依此驱动虚拟工程机器人实现对操作者的视觉提示。实验结果表明:该方法简单有效,误差小于8%,可为基于虚拟现实的遥操作工程机器人的视觉提示系统设计提供指导。Aiming at the delay of video transmission in the working process of teleoperation hydraulic engineering robot, a visual cue method based on attitude recognition for teleoperation hydraulic engineering robot was proposed. The camera was used to acquire the image of the teleoperated hydraulic engineering robot, and the keypoints identification stacked hourglass(KPR-SHN) model and the attitude recognition neural network(AR-NN) model were used to obtain the displacement of the hydraulic cylinder of the teleoperation hydraulic engineering robot, thereby driving the virtual engineering robot implement a visual cue to the operator. The method has the characteristics of simple structure and high technical robustness.Experimental results show that the method is simple and effective, and the error is less than 8%, which can provide guidance for the design of visual cue system based on virtual reality for teleoperation engineering robots.

关 键 词：遥操作工程机器人 视觉提示 姿态辨识 堆叠沙漏网络 

分 类 号：TH16[机械工程—机械制造及自动化]