夹剪式采摘成串果实的激振特性与防振动脱落研究

Stem-gripping/cutting is regular way for robotic cluster harvesting, but fruit drop by vibration has become a key issue to perplex high-speed operation. So grape and tomato clusters will be select as research objects,and excitation characteristics in fruit detachment and tranportation will be analysed , which refers to: (1)deviation excitation caused by impact force,deviation between fingers and stem and discordance between gripping and cutting motion;(2)controllable excatation caused by unsmooth trail and motor vibration of manipulator;(3)compensable excitation caused by joint clearance and flexibility of manipulator. And then fruit-stem system will be modelled as multirod-elastico-viscous joint-mass system, whose response to the above different excition and probability of fruit drop will be found. At last, high-speed and less- fruit-drop robotic harvesting will be achieved by taking different measures to reduce the vibration of clusters, which includes: (1)accurate positioning of stem and gripper based on machine vision and laser scan together with gripping/cutting coordination control to reduce the deviation excitation in gripping/cutting;(2) motion plan to guarantee Jerk continuity to reduce controllable excitation in transportation;(3)negative-pulse time varying input shaping control together with tradtional input shaping control to reduce compensable excitation in transportation.This research will discover the vibration mechanism of fruit-stem system under aperiodic multisource excitation and present solution for high-speed less-fruit-drop robotic harvesting.It is valuable in both theoretical research and practical operating.

果梗夹剪是机器人采摘成串果实的通用方式，但果实的振荡脱落成为实现快速采摘作业的瓶颈难题。拟以葡萄与番茄为对象，对取果阶段源于高速冲击/指-梗偏差/夹-剪不协调的偏差性激励，移送阶段源于机械臂各电机运动/关节间隙及柔性/末端运动不平滑的可控性和可补偿性激励进行分析辨识；建立果-梗系统的多连杆-黏弹铰链-质量块模型，探明其在上述激振下的动力学响应和果实脱落概率关系；以高效率、低脱落为目标，针对取果偏差性激励进行激光扫描-机器视觉结合的穗梗-手爪精确定位与夹剪协调控制，针对移送的可控性和可补偿性激励分别进行急动度连续的运动规划与负脉冲非时变输入整形-传统输入整形的结合控制，从而有效削减振动激励，解决快速夹剪采摘成串果实的振荡脱落问题。本课题将揭示多源非周期性激励下的果穗系统振荡机制并提出快速防脱落采摘方案，从而为推动果实机器人采摘技术走向实际应用提供技术基础，具有重要的理论价值和显著的实际意义。

果梗夹剪是机器人采摘成串果实的通用方式，但果实的振荡脱落成为实现快速采摘作业的瓶颈难题。围绕串果在机器人采摘中的振荡脱落问题，本项目获得了葡萄串果主穗轴、多级关节、果蒂连接的大样本拉/压/弯/剪黏弹力学特性数据，并构建了“压剪弯耦合杆-黏弹铰链-刚性杆-质量球”复合葡萄果穗模型；提出了串果摘取中的非对称与非平行偏差和“单侧接触-推离-双侧夹剪”的非同步振动激励特征，以及移送与放果的快速加减速和接触性冲击等关键激励特征；建立了串果振动仿真和末端-串果耦合模型；进行了葡萄摘取、水平/竖直移送的多参数试验与仿真，掌握了加减速、加载方向、激励幅度等对果穗振动与脱落的影响；开发成功小型藤架和大型篱架葡萄采摘机器人，提出了架栽葡萄收获的深度传感器-末端配置、穗-梗识别方法与远近景协调的手眼伺服控制策略，完成了机械臂的平滑减振运动规划、关键部件开发与整机集成试验。. 试验验证表明该串果振动仿真模型的平均果粒摆角和脱落率平均误差分别为6.09%和8.71%，能够良好满足研究要求。在移送环节大加速度（12 m/s2）的突然启停冲击下，果穗中果粒的平均摆角可达15°以上，而在摘取阶段的严重偏差情况下，快速作业的果粒脱落概率也仍可能达到10%甚至更高。为实现穗轴的精准手-眼定位，所提出的两种葡萄穗-梗识别方法的识别成功率和速度均在90%和1.4s以内，所开发基于RealSense深度视觉的小型就近放果采摘机器人整机重量小于80Kg。. 该果穗振动仿真模型将大大推进果穗的机器人低脱落采摘研究，同时亦为机械振动收获等领域提供了仿真分析工具；所探明不同因素等对果穗振动与果粒脱落的影响，为采摘全过程减振防脱的“摘-送-放”策略优化设计与控制实施提供了依据；创新提出的葡萄收获的手眼配置、穗-梗识别与伺服控制方法和整机开发，为加速葡萄采摘机器人技术的发展和推广应用提供了支撑。