名优茶采摘机器人的系统设计与试验

针对名优茶采摘劳动力短缺的现状，设计了一种适用于自然环境下作业的名优茶采摘机器人。系统由采摘机械手、3P-Delta机械臂、视觉系统和主控系统构成。机器人通过深度相机获取茶蓬图像，利用深度学习和骨架法计算采摘点。从名优茶的生长特征出发，建立了机械臂采摘轨迹模型，采用Bézier曲线对茶叶采摘路径进行优化，降低了机械臂在快速运动过程中的加速度突变，提高了采摘运动的平滑性。利用虚功原理建立机械臂动力学模型，基于滑模控制设计了机械臂采摘控制策略，优化了控制算法中的指数趋近律，有效地抑制了滑模面在快速趋近时的抖振现象。实地测试结果表明，采摘机器人能够实现名优茶的采摘，采摘率和完整率分别为75.53%和54.68%，平均采摘速度为0.451颗/s。

Design and Experiment of Intelligent Picking Robot for Famous Tea

In view of the current situation of labor force shortage in famous tea picking, this paper designs a robot for famous tea picking, which is suitable for natural environment. The robot is composed of picking module, 3P-Delta manipulator, vision system and controlling system. The robot captures the image of tea canopy using RGB-D camera, and then calculates the picking points by deep learning model and the skeleton method. Based on the growth characteristics of famous tea, the picking trajectory model of the manipulator was established. The Bézier curve was used to optimize the picking path of tea, which effectively alleviated the acceleration mutation caused by the rapid motion of the manipulator and improved the smoothness of the picking movement. The dynamic model of the manipulator was established by using the principle of virtual work, and the picking control algorithm of the manipulator is designed based sliding surface variable structure control. The exponential reaching law in the control algorithm is optimized, so that the chattering phenomenon in the sliding surface control is effectively suppressed in the process of rapid reaching. The field test results show that the picking robot can realize the picking of famous tea in the natural environment, and the picking rate and integrity rate are 75.53% and 54.68%, respectively. The average picking speed of the robot is 0.451/s.