节点文献

视觉引导控制在水轮机修复专用机器人中的应用研究

Research on Vision Guiding Control Application in Turbine Blade Repairing Robot

【作者】 周丽

【导师】 张礼勇;

【作者基本信息】 哈尔滨理工大学 , 测试计量技术及仪器, 2002, 硕士

【摘要】 对水轮机转轮叶片的维护与维修是各国水电部门的一项重要工作,每年都要在其上耗费大量的人力、物力、财力。为了减少此项工作的劳动强度,提高维护、维修的效率,我们引进并自行开发了用于水轮机转轮叶片修复用的专用机器人。由于此种机器人的工作方式是示教再现型的,一旦工作环境或工作对象发生改变,它就无法正常工作。因此,我们为其加入的视觉导引系统就成为提高其智能化水平的一个必不可少的条件。 本文以省科技攻关项目和市重点科技攻关项目为背景,采用基于图像的无标定视觉导引方法来控制系统运行。克服了传统工业机器人只能工作在环境模型已知情况下的局限性。由于我们是直接利用图像控制机器人运动的,因此对图像雅可比矩阵的正确求解是系统得以成功运行的关键。以往对图像雅可比矩阵的求解一般都是采用标定的方式,而标定是一项十分繁琐工作,并且有时在实际操作中是根本不可行的。为避免这一情况的发生,本文给出了一种通过动态估算雅可比矩阵来代替实际雅可比矩阵的方法,并且对采用此策略时存在的对图像雅可比矩阵的估算随时间变坏的问题给出了一些改进措施。其中,通过使用附加的试探性运动方法将提高图像雅可比矩阵估计的精度和速度。 在机器人视觉伺服系统中的图像特征信息提取部分,依据工作环境的特点设计了一套具有较强抗干扰能力的数字图像处理算法。使得将处于自然背景下的标记点信息准确地获取到,并应用于实验中的图像特征信息提取。 本文对机器人视觉导引的控制算法在一个六自由度的机器人系统上进行了仿真实验和实际实验研究,实验结果表明此方法具有完成机器人初始定位任务的能力,并且具有一定的抗干扰性。

【Abstract】 The repair and preservation work of the turbine blades is the international problem for all the hydro power plant. Introducing and exploring a repair robot used for welding and grinding on the turbine blades is very useful for those plants. In order to increase the welding and grinding automatic level, position guiding part is indispensable. This paper takes projects of province and city for tackling key problem as research background and applied vision control method to guide turbine blade repairing robot arm. This work made basis for further exploring vision intelligent technique in turbine blade repairing robots.Due to the teaching and redisplaying property of the Scompi robot, this paper presents an image-based vision control strategy without calibration, which adapt to working spot of turbine blade repairing robot and own practical perspective. In this paper, some efficient schemes are presented to resolve the problem that estimated image Jacobian deteriorates over time in image-based visual servoing control approach without calibration. The most effective method is using "exploratory motion", which considerably improve the precision and velocity of vision guiding. To resolve the task of information extracting of image property in robot visual servo, this paper designed a serial of digital image process algorithms with certain robustness according for the property of work environments. Applying the presented algorithms, the information of markers in natural background is obtained accurately, and successfully extracted target image property information in experiments. To validate presented algorithm, we perform simulation and experiments on a six-DOF robot system. Experimental results illustrate presented method has the function to fulfill the task of robot initial positioning with certain anti-jamming ability.

  • 【分类号】TP242
  • 【被引频次】1
  • 【下载频次】161
节点文献中: