Citation: Hou, L.; Wang, H.; Zou, H.;
Zhou, Y. Robotic Manipulation
Planning for Automatic Peeling of
Glass Substrate Based on Online
Learning Model Predictive Path
Integral. Sensors 2022, 22, 1292.
https://doi.org/10.3390/s22031292
Academic Editors: Yuansong Qiao
and Seamus Gordon
Received: 8 January 2022
Accepted: 5 February 2022
Published: 8 February 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Robotic Manipulation Planning for Automatic Peeling of Glass
Substrate Based on Online Learning Model Predictive
Path Integral
Liwei Hou
1
, Hengsheng Wang
1,2,
*, Haoran Zou
1
and Yalin Zhou
1
1
College of Mechanical and Electrical Engineering, Central South University, Changsha 410083, China;
lwhou1992@gmail.com (L.H.); zouhr1995@163.com (H.Z.); 1012656597@163.com (Y.Z.)
2
State Key Laboratory for High Performance Complex Manufacturing, Central South University,
Changsha 410083, China
* Correspondence: whsheng@csu.edu.cn
Abstract:
Autonomous planning robotic contact-rich manipulation has long been a challenging
problem. Automatic peeling of glass substrates of LCD flat panel displays is a typical contact-rich
manipulation task, which requires extremely high safe handling through the manipulation process.
To this end of peeling glass substrates automatically, the system model is established from data and is
used for the online planning of the robot motion in this paper. A simulation environment is designed
to pretrain the process model with deep learning-based neural network structure to avoid expensive
and time-consuming collection of real-time data. Then, an online learning algorithm is introduced to
tune the pretrained model according to the real-time data from the peeling process experiments to
cover the uncertainties of the real process. Finally, an Online Learning Model Predictive Path Integral
(OL-MPPI) algorithm is proposed for the optimal trajectory planning of the robot. The performance
of our algorithm was validated through glass substrate peeling tasks of experiments.
Keywords:
glass substrate peeling; manipulation planning; system model; deep learning; online
learning; Model Predictive Path Integral
1. Introduction
The glass substrate is the base material of LCD flat panel displays, and its thinning
process is necessary for the end product being lighter, thinner and smoother [
1
]. At
present, the thinning process mainly adopts polishing machines with chemical mechanical
processing. The workpiece of glass substrate, say 0.73 m in width, 0.92 m in length and 2
mm in thickness, needs to be fixed on the surface of a rotating rigid workbench plate, on
which an adsorption pad is attached. The adsorption pad (a polyurethane foam pad about
1 mm thick) is generally used for the fixing task, and the glass substrate is absorbed into the
pad with the vacuum suction force in between after pressing the glass substrate evenly on
the pad. Separating the glass substrate from the pad is needed after the thinning process,
and the process of unloading the glass substrate is usually done by human operators. They
use their fingernails to “peel” the glass substrate off the pad from a corner of the substrate
(Figure 1a). This manual process is labor-consuming, apt to glass damage, and therefore
also expensive. Automating the process of unloading the glass substrate has long been
expected, and we presented a scheme (Figure 1b) in [
2
]. It should be pointed out that
the original intention of introducing Figure 1b here is to better show the task we need to
accomplish. The actual experiment platform used can be seen in Figure 7, which is different
from Figure 1b.
We invented a wedge blade (Figure 1b) as the end-effector of a robotic arm to imitate
the “peeling” motion of human operators, but the action and reaction of human operators
in the “peeling” process is still a challenge to imitate by robots, and this paper reports our
Sensors 2022, 22, 1292. https://doi.org/10.3390/s22031292 https://www.mdpi.com/journal/sensors
