机器人无标定手眼协调方法研究  

Research on Robotic Uncalibrated Hand-eye Coordination Method

作　　者：盛荣 孙首群[1] 焦玉格 叶其含 SHENG Rong;SUN Shou-qun;JIAO Yu-ge;YE Qi-han(College of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China;Shanghai Baosight Software Co.,Ltd.,Shanghai 201900,China)

机构地区：[1]上海理工大学机械工程学院,上海200093 [2]上海宝信软件股份有限公司,上海201900

出　　处：《软件导刊》2020年第5期23-28,共6页Software Guide

基　　金：国家科技支撑计划项目(2015BAK16B04)。

摘　　要：无标定情况下的视觉定位是机器人视觉伺服领域的热点和难点。提出一种基于快速自适应滤波的机器人无标定手眼协调方法。首先根据机器人三维基坐标建立运动空间和图像特征空间的微分映射关系,在不预先标定视觉传感器与机器人参数情况下,设计一个直接估计增益矩阵Kk的自适应滤波器,在线辨识图像雅可比矩阵。采用最小二乘法获得状态估计的初始值,在此基础上设计视觉控制器和计算运动控制量。在MAT⁃LAB环境下建立机器人手眼协调控制系统仿真模型,将该方法用于观测噪声统计特性未知情况下的定位反馈,并对机器人空间螺旋运动的跟踪情况进行分析。实验结果表明,该自适应滤波器能够完成不规则环境下视觉引导的追踪任务,定位效果优于常规卡尔曼滤波和Sage-Husa自适应滤波,具有较强的鲁棒性。Visual positioning without calibration is a hot and difficult problem in the field of robot vision serving.A non-calibrated hand-eye coordination method based on fast adaptive filtering is proposed.Firstly,based on the three-dimensional base coordinates of the robot,the differential mapping relationship between the motion space and the image feature space is established.Without pre-cali⁃brating the visual sensor and the robot parameters,an adaptive filter for directly estimating the gain matrix Kk is designed to identify the image Jacobian online.The simulation model of robot hand-eye coordination control system is established in MATLAB environ⁃ment.The method is used to observe the positioning feedback under the condition that the noise statistical characteristics are unknown,and the tracking situation of robot space spiral motion is analyzed.The results show that the designed adaptive filter can complete the vi⁃sion-guided tracking task in irregular environment.The positioning effect is better than the conventional Kalman filter and Sage-Husa adaptive filtering,and it has strong robustness.

关 键 词：视觉定位 无标定手眼协调 增益矩阵 在线辨识 自适应滤波器 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]