摘要
为克服机器人视觉伺服系统对标定精度的依赖性,提出一种弱标定条件下基于冗余特征的机器人视觉伺服控制方法.该方法无需3D重构,直接以冗余图像特征作为反馈信息建立伺服控制器.鲁棒的特征提取算法确保了伺服系统能够在复杂的环境快速稳定地提取伺服特征.针对图像视觉伺服系统中需具有足够敏感度的要求,结合视觉特征的前期处理,提出一种新的敏感性校验准则,提高了系统的敏感性以及对环境的适应性.试验分析结果表明,该方法对标定误差和深度估计误差具有鲁棒性.
A redundant features approach for robot visual servoing in coarse calibration is presented to overcome system dependency on calibration accuracy. Redundant visual features are directly selected as feedback signal to control robot without 3D reconstruction. Firstly, servoing features are obtained quickly and stably with robust feature recognition algorithm in complex circumstances. Secondly, a novel sensitivity verified rule is proposed to make image-based visual system with enough sensitivity by combining visual feature preprocessing, which improved system sensitivity and environment adaptability. Finally, experiment result shows the approach is robust to calibration and depth estimation error.
出处
《华中师范大学学报（自然科学版）》
CAS
北大核心
2013年第2期167-172,共6页
Journal of Central China Normal University：Natural Sciences
基金
国家自然科学基金项目(61175121)
福建省高等学校杰出青年科研人才培育计划项目(JA10006)
华侨大学基本科研业务费专项基金项目(JB-SJ1003)
关键词
弱标定
冗余特征
图像视觉伺服
条件数
weak calibration
redundant feature
image-based visual servoing
condition number
