Professional Documents
Culture Documents
conveyed Objects
M. Weyrich1, Y. Wang1, J. Winkel1, M. Laurowski1
1
University Siegen, Chair of Automated Manufacturing and Assembly
Paul-Bonatz-Str. 9-11, 57068 Siegen, Germany
Abstract
Under the pressure of cost reduction and productivity improvement, this paper presents a new methodology
which provides a fast inspection of defective objects and generates a real time motion trajectory for
processing objects being conveyed with high speed in an industrial large-scale production. The image data
obtained by a multispectral imaging system is analyzed within image processing algorithms using
classification methods based on support vector machine. These data provide a basis for a path planning
algorithm which considers location, orientation and arrangement of defects on the conveyed objects. Selective
processing tool guided by the planed path is motion controlled.
Keywords:
Design method, Visual inspection, Adaptive control
1 INTRODUCTION
Machine vision has been researched and applied in many
branches in science and engineering. Their applications
allow automating many technological processes, e.g.
industrial quality control.
In the industrial quality control, instead of manual
inspection, conventional vision based automatic
inspection has been applied. The approach aims to sort
objects or products according to vision features, which
are predefined, into variant categories indicating different
quality levels. Quality grading and automatically sorting of
oranges is a typical example [1]. However, the increasing
interests are not only aimed at simply sorting the
defective objects from flawless objects, but also
processing the defective objects directly after defect
detection. This will help to reduce waste during production
under the pressure of cost reduction and productivity
improvement. Therefore, vision based path planning
which can be applied for guiding the manipulator to
process defective objects, is becoming an important issue
in vision based automatic inspection system.
2
2.1 Objectives
This paper presents an approach for an automated
system consisting of a vision unit, path planning and
motion control as well as object processing. The system
provides a fast inspection of objects being transported on
a conveyor belt with high speed. An object can be a 2D or
3D object. According to the inspection result, the target
objects are classified in three categories: (1) flawless, (2)
defective and processable and (3) defective and nonprocessable (see Figure 1). The defective and
processable target objects are further classified according
to user-defined defect types. Meanwhile, a real time
motion trajectory is generated for processing the defective
and processable objects. Figure 1 display the schematic
representation of work flow concept of our proposed
processing system.
Comparing to conventional vision based inspection
systems, our approach allows detecting and classifying
Processing tool
Conveying
direction
Further
processing
Conveyor belt
Rotary encoder signal
(c)
(d)
Figure 3: Experiment with different bands of
electromagnetic radiation.
SensorA
Camerasensor
(a)
SensorB
Twoobjectives
Spectral
lightingA
Spectral
lightingB
Spectral
lightingC
(b)
Treatment
The selection of a suitable processing tool is a major
challenge in order to the given problem formulation. For
the materiel processing, an efficient, fast and food
compatible system have to be found. Within an extensive
application analysis, the systems in Figure 7 have been
compared with the consideration of desired processing
task.
As a result of the high processing speed, the demand of a
linear section orientation per processing unit and the
massive operating costs of using compressed air brought
the selection to a one dimensional linear axis using a
water cutting system. In combination with the moving
conveyer belt, fast and effective cutting path can be
realized.
bbz
conveyor belt
x
y
time to process
d
encoder
signal
distributor
PC-VU
data
communication
PC-PU
(a)
(b)
Figure 9: Water jet motion controlled by a linear axis
manipulator.
Rotary encoder signal from conveyor motor is the
footstone of our control system. Benefits of using encoder
signal are: on the one hand, synchronization of unstable
conveying velocity and camera (allows to capture
conveyed objects precisely without consideration of
unstable velocity), on the other hand, synchronization of
vision system and processing tool (allows guiding
processing tool to reach target position).
The proposed multispectral vision system in Section 4.2 is
used for inspection of potato slice. The image processing
algorithm is implemented with the development platform
LabVIEW. The image processing is performed in the color
space HSL. Figure 10 displays some experimental
results; wherein, (a) displays the defects of peel, (b)
green and (c) decayed spot.
Path planning and processing strategy
Through variant test series, the development of an
overarching processing strategy was focused. These
come to the result, that the direct processing strategy is a
solution for our problem formulation. The planned path for
an online processing has already been taken place in
LabVIEW-based image processing algorithm. Figure 11
shows an example of a direct processing strategy using a
LabVIEW-based path planning. The equivalent ellipse
according to the geometry shape of a defect will be firstly
analyzed. Cutting path is planned to be parallel (with an
offset of s) to the major axis of the equivalent ellipse. If a
potato slice has a big defective fraction, after cutting only
a small non-defective fraction will be left. In this case, this
potato slice will be considered as a non-processable
object. Cutting path will be planned to go around the
potato slice with the avoidance of wrong treatment (e.g.
cutting a non-processable potato is a wrong treatment).
The weight of the cutting head-sets limits the maximum
acceleration of linear axis. Due to this system behavior,
the minimal accessible cutting angle is a key factor for
the processing strategy.
peel
(a)
Figure 11: Path planning based on the result of image
data analysis.
green
(b)
decayed
(c)
Figure 10: Result of image analysis.
5.1 Remove of defects by using water jet
The developed processing tool consists of a water jet
cutting head based on a high speed linear axis with a
maximum acceleration of 15 g. For the holding fixture of
the components and to absorb the processing forces, an
individual food compatible stainless steel frame was
developed and embedded in the system concept. The
linear axis is placed, as illustrated in Figure 9,
perpendicular to the moving direction of the conveyer belt,
so that a two dimensional cutting movement can be
realized.
Thereby the tool center point of the high pressure water
jet is adjusted on the coordinate zero point line (process
center line) of the image processing system to reach
cutting path directly and accurately.
The system behavior of the water jet cutting application
was tested in different experiments, using paperboards to
map the expected cutting path. Furthermore the properly
functioning of the developed data communication was
surveyed by entering the committed point with the tool
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]