番茄粘弹性参数机器人抓取在线估计

发布时间：2018-05-27 06:36

  本文选题：机器人抓取 + 番茄　； 参考：《农业机械学报》2017年08期

【摘要】：为了使采摘机器人在抓取过程中能够对被抓果蔬的粘弹性力学参数进行快速估计,实时优化抓取过程,减少末端执行器对被抓取对象造成机械损伤,以抓取力、变形量、作用时间为输入,建立了番茄粘弹性参数估计的人工神经网络模型。运用质构仪蠕变试验所测的力、变形和时间,以及粘弹性参数E_1、E_2、η_1、η_2作为训练数据集,确定了人工神经网络的拓扑结构和参数,并测试了网络模型的粘弹性参数估计性能。利用二指机器人末端执行器对随机番茄样本进行抓取试验,并在抓取过程中用此模型来在线估计粘弹性参数。通过与质构仪的实测值进行对比发现,当时间t≥0.2 s时,各参数的估计值与实测值之间的相对误差均在25%以内,并根据0.2 s时得到的粘弹性参数对机器人抓取力范围进行了估计。结果表明,利用此方法在机器人抓取过程中可以对被抓番茄粘弹性特性进行估计,为在线优化抓取力提供了依据。
[Abstract]:In order to estimate the viscoelastic mechanical parameters of fruits and vegetables quickly, optimize the grabbing process in real time, and reduce the mechanical damage caused by the end actuators, the grabbing force and deformation can be obtained. An artificial neural network model for the estimation of viscoelastic parameters of tomato was established. The topological structure and parameters of artificial neural network are determined by using the force, deformation and time, and viscoelastic parameters E _ 1C _ 2, 畏 _ 1 and 畏 _ 2 as training data set, and the viscoelastic parameter estimation performance of the network model is tested. A two-finger robot end effector was used to grab random tomato samples and the model was used to estimate the viscoelastic parameters online. By comparing with the measured value of the qualitative structure instrument, it is found that the relative error between the estimated value and the measured value of each parameter is within 25% when the time t 鈮，

本文编号：1940874