摘要
针对传统输油臂管口对接过程效率低、自动化程度低等问题,设计了基于视觉伺服的输油臂机器人智能对接系统。建立了机器人运动学模型,设计了多层次开放式的机器人实时控制系统,采用多条件约束校验SVM分类器,组成了在线自学习双目立体视觉系统,提高了复杂环境下视觉定位系统的泛化能力及稳定性。实验表明所提出的校验SVM分类器识别准确率为97.80%,定位准确率为92.86%;所建立的机器人智能对接系统对接成功率为91.43%,系统故障率仅为1.40%,说明校验SVM分类器具备良好的准确性和稳定性,整个输油臂机器人智能对接系统稳定性良好,能够满足自动对接要求。
To overcome the problems of inefficiency and low-level automation of traditional loading arm nozzle docking process,a smart docking system of loading arm robot based on the visual servo was designed.The robot kinematics model was established,and the multi-level and open real-time control system of the robot was designed.The SVM classifier with multi-condition constraint checking(C-SVM)was used to form the binocular stereo vision system with online self-learning ability,which improved the generalization ability and stability of the visual measurement positioning system in complex environments.Experiments show that the proposed C-SVM classifier target nozzle identification accuracy is 97.80%,the positioning accuracy is 92.86%,the docking success rate is 91.43%,and the system failure rate is only 1.40%.It illustrates that the C-SVM classifier has good accuracy and stability,and the entire robot arm system has good stability and can meet the requirements of automatic docking.
作者
白元明
孔令成
赵江海
张强
方世辉
BAI Yuan-ming;KONG Ling-cheng;ZHAO Jiang-hai;ZHANG Qiang;FANG Shi-hui(School of Information Science & Engineering,Changzhou University,Changzhou 213164,China;Institute of Advanced Manufacturing Technology,Hefei Institutes of Physical Science,Chinese Academy of Sciences,Hefei 230031,China;Department of Automation,University of Science and Technology of ChinaHefei 230026,China)
出处
《仪表技术与传感器》
CSCD
北大核心
2019年第10期88-95,116,共9页
Instrument Technique and Sensor
基金
国家自然科学基金项目(61703390)
江苏省重点研发计划项目(BE2017007-1)
青年科学基金项目(61503364)
关键词
输油臂机器人
SVM分类器
视觉识别
定位算法
目标检测
视觉伺服
loading arm robot
SVM classifier
visual identification
location algorithm
object detection
visual servo
