【摘要】 Tripod机器人增加一套视觉系统,使其具有视觉处理功能。硬件由摄像头、Tripod、计算机组成,将摄像头采集的Tripod平台图像传输给计算机;软件由Tripod控制软件、Visual Studio组成,设计图像采集与预处理、识别与定位、最优轨迹等算法处理图像。透视变换完成侧拍图矫正后,灰度化、高斯滤波、二值化、Canny边缘提取,获取目标轮廓图;完成几何体上符号与几何体形状的识别,两步定位法实现拾放中心点精确定位;遗传算法获取最优拾放轨迹,完成染色体编码、初始化种群、适应度选择、遗传算子等研究。VS将轨迹数据传送给贝加莱PCC,Tripod运动控制系统将数据运算并产生控制指令,驱动Tripod运动。实验结果证明,此视觉系统在Tripod硬件平台上精准、快速地完成了特定的拾放任务。更多还原

【Abstract】 In this paper, a visual system is added to Tripod robot to make it have the function of visual processing. Hardware is composed of camera, Tripod and a computer, and the Tripod platform image acquired by the camera is transmitted to the computer. Software is composed of Tripod control software and Visual Studio. Three kinds of algorithms are designed to process the image, including image acquisition and preprocessing, recognition and positioning, the optimal path. Perspective transformation obtains the correction chart of lateral shoot images. Gray scale, Gaussian filtering, binarization and edge extraction of Canny operators obtains contour profiles of the target. After completion of distinguishing symbols printed on geometries and shapes recognition of geometries, two step localization method is adopted to achieve precise positioning of pick & place center points. Genetic algorithm is used to obtain optimal grasping scheme, and to complete research on chromosome coding, initial population, fitness selection and genetic operators. Visual studio transmits grasping data to B&R PCC. Tripod motion control system generates data operations and control instructions to drive the robot to move. Experimental results show the visual system realizes specific pick & place task precisely and quickly in the Tripod hardware platform.更多还原