双目立体视觉的羽毛球机器人环境自动感知  

Automatic Environment Perception of Badminton Robot Based on Binocular Stereo Vision

作　　者：严财庆[1] 刘子强[2] YAN Cai-qing;LIU Zi-qiang(Ministry of Physical Education,Jiangxi University of Technology,Jiangxi Nanchang 330098,China;College of Physical Education,Nanchang Hangkong University,Jiangxi Nanchang 330098,China)

机构地区：[1]江西科技学院体育教育部,江西南昌330098 [2]南昌航空大学体育学院,江西南昌330098

出　　处：《机械设计与制造》2023年第6期285-288,295,共5页Machinery Design & Manufacture

基　　金：“教体社”结合羽毛球后备人才培养模式实践研究—对熊国宝羽毛球学院的实证研究(TY19107)。

摘　　要：为了提高羽毛球机器人的接球成功率,提出基于双目立体视觉的羽毛球机器人环境自动感知方法。分析羽毛球机器人外部结构与内部智能控制系统的组成要素,结合气缸工作过程,利用旋转矩阵变换坐标构建机器人运动学模型;采用棋盘格标定算法获取相机内、外部参数,矫正畸变现象,采用双滤波算法过滤图像噪声;匹配不同视角下图像同一特征点,重构三维环境信息,使用核相关滤波器算法对感知区域做循环位移,构建海量目标分类器,实现环境自动感知。仿真实验证明,所提方法能够获得感知区域更多的目标信息,无论在哪种击球方式下,都能准确感知环境,减少接球定位误差,提升接球成功率。In order to improve the catch success rate of badminton robot,an automatic environment perception method of badminton robot based on binocular stereo vision is proposed.The external structure and internal intelligent control system of the badminton robot are analyzed.Combined with the working process of the cylinder,the kinematics model of the robot is constructed by using the rotation matrix to transform the coordinates.The chessboard calibration algorithm is used to obtain the internal and external parameters of the camera,correct the distortion,and the double filter algorithm is used to filter the image noise.Match the same feature point of the image from different perspectives,reconstruct the three-dimensional environment information,use the kernel correlation filter algorithm to make a cyclic displacement of the sensing area,and construct a massive target classifier to realize the automatic sensing of the environment.Simulation experiments show that the proposed method can obtain more target information in the sensing area,and can accurately perceive the environment,reduce the positioning error of receiving the ball and improve the success rate of receiving the ball no matter which way of hitting the ball.

关 键 词：双目立体视觉 羽毛球机器人 环境感知 相机标定 跟踪检测 

分 类 号：TH16[机械工程—机械制造及自动化] TP242[自动化与计算机技术—检测技术与自动化装置]