基于视觉算法的采摘机器人控制优化研究

Study on Control Optimization for Picking Robot Based on Visual Algorithm

为了优化采摘机器人控制,基于视觉技术设计了控制系统。首先,在大津分割法的基础上,计算图像中的角点,将角点从分割后的二值图中去除,降低背景干扰点簇的影响;其次,建立褐菇几何特征参数,以此参数对褐菇进行特征图像提取;再次,采用遗传学算法,对褐菇采摘路径进行优化,与传统往复前进式路径规划相比路程缩短了30.5%;最后,对系统进行实地褐菇采摘测试,结果显示:辨别成功率分布区间为[87%,91.7%],采摘成功率分布区间为[80%,85.75%],单个褐菇平均采摘时间分布区间为[12.4s,13.2s],系统具有良好的采摘性能。

In order to optimizing the control for picking robot,this system was designed based on visual technology.Firstly,on the basis of Otsu division method,angular points were calculated,and then angular points were removed from the segmented binary graph to reduce influence of background interference point cluster.Secondly,geometric characteristic parameters were achieved for brown mushroom to extract characteristic image of brown mushroom.Thirdly,genetic algorithm was used to Optimized brown mushroom picking path.Compared with traditional reciprocating picking path,distance reduced by 30.5%.In the end,the test was taken for system,the result showed that the identification success rate range was[87%,91.7%],success rate of picking range was[80%,85.75%],and average picking time range for single brown mushroom was[12.4s,13.2s],this system had good picking performance.