Palm vision and servo control strategy of tomato picking robot based on global positioning

文献类型： 外文期刊

第一作者： Liang, Zhi

作者： Liang, Zhi;Li, Xiaojuan;Zou, Xiangjun;Wang, Guoqiang;Wu, Fengyun

作者机构：

关键词： Tomato picking robot; Palm vision; Spatial asynchronous positioning; Servo tolerance; Dynamic picking strategy

期刊名称：COMPUTERS AND ELECTRONICS IN AGRICULTURE （ 影响因子：8.9； 五年影响因子：9.3 ）

ISSN： 0168-1699

年卷期： 2025 年 237 卷

页码：

收录情况： SCI

摘要： This study aims to address the challenges faced by existing tomato picking robots in non-enclosed environments, including inadequate visual system positioning accuracy, poor visual servo capability, and incomplete picking strategies. Additionally, the spatial constraints caused by crop height restrict the picking space of a 6 degrees of freedom (DOF) robotic arm, making continuous and efficient picking operations difficult. To overcome these limitations, this research proposes a humanoid perception and decision-making picking system based on visual servo control. A 6 + 1 DOF picking robot experimental platform was constructed, featuring global and palm positioning for multi-visual perception. The study developed a field of view relationship model between global vision and palm vision, establishing a spatial asynchronous visual positioning system. An embodied perceptionbased visual servo control strategy was designed, and a lifting column was added to provide an extra degree of freedom in the crop height direction, thereby expanding the picking workspace. The system dynamically adjusts the picking space based on global vision feedback, while palm vision servo compensates for camera positioning errors in real time to enhance error tolerance and improve the overall picking performance. Moreover, a dynamic picking strategy was incorporated, optimizing operational behavior through real-time analysis of fruit growth scenarios, which improved the overall picking success rate. To comprehensively evaluate the performance of the proposed picking system, autonomous continuous picking experiments were conducted in both indoor and outdoor environments. The experimental results demonstrated average picking success rates of 85.7 % indoors and 80.5 % outdoors. The visual positioning system achieved a depth-direction positioning error of 6.72 mm and a Euclidean distance error of 11.94 mm. The average picking times were 12.28 s per item indoors and 16.55 s per item outdoors. These results demonstrate the robot's reliability in complex environment.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序