Simulating
Manufacturing Processes in Education and Training:
Color Sorting Example
|
Richard Johnson |
Samuel Cotton Ph.D. |
|
Department of Industry and Technology |
Department of Industry and Technology |
|
|
|
Abstract
Simulation of industrial operations has
become a well established and effective instructional tool used in both
industrial training and formal educational programs. Duplicating large scale operations is not a
cost effective strategy for training and education of trainees. Since typical operations used in manufacturing
are basically the same regardless of the scale of the equipment, one can educate
learners for specific tasks using smaller and less expensive equipment in a way
that these skills can be easily transferred to larger operations. This article demonstrates one task, color
sorting, which is typical in commercial applications that can be simulated
effectively in a small scale. With
decreasing budgets for education and training it is becoming critical to
educate learners with tools and equipment that are affordable and do not
interfere with normal company operations.
Many variations are possible once the general concept of small scale
industrial process simulation is mastered.
Introduction
Simulating realistic manufacturing, assembly, or
transportation operations used in industry can be a valuable learning
experience for students. In occupational
program areas, students are often taught using applied educational
methods. Accurately replicating
operations, although often on a smaller scale than used in industry, is an
important tool to help learners to recognize and appreciate real-world
applications of new skills or concepts to be learned. To learn to operate many types of equipment,
it is not always necessary to work with equipment or materials on the same
scale used in industry.
Since the skills and
functions used are very similar or identical regardless of actual scale,
schools or adult training centers can often effectively train future workers
for the skills needed on smaller and less expensive equipment, minimizing the
need to use production equipment for training.
This can be attractive to industry because of reduced training expenses
caused by a slow down or interruption of production work that would occur
during training exercises. Schools find
accurate simulations attractive because of the reduced cost of the equipment and
space required for the simulations.
In both industry and
education today, many production educational tasks are accomplished through
computer controlled devices or operations.
This reduces the need for large scale or large quantities of production
equipment for either training or production.
An example of an operation typical to commercial production is shared in
this work that demonstrates how color sorting may be simulated in a classroom
or training center. This is accomplished
by sorting different colors M&M™ candies (see graphic #1). There are many other operations that could be
reproduced using similar strategies (e.g. sorting by size or shape, weight,
destination, or potential defect).
Copies of the program
listings are included in appendix A for reference purposes. References to items in the program listings
are included in body of the report.
For a materials list see appendix B.
Mechanical
Calibration
The candy sorter used in this example will identify
six different colors of candies and deposit them in an appropriate collection
tube (see graphic #2). A rotating
plate on the mechanism is provided to move the candies from a storage location
to a color analysis location and ultimately to collection tubes that hold the
candy sorted by color. One collection
tube is designated for candies that the color can not be determined.
There are ten positions on
the sorting mechanism (see graphic #3) that must be identified in order to
operate the system. Eight positions are
for the collection tubes (only seven are used for this project), one position
is under the TCS230 color sensor, and one position is under the supply
tube. The supply tube position is
defined as the 0° or base reference point.
The color sensor is 30° from the supply tube position, followed by the
first drop tube at 75° from the supply tube.
The other seven drop tubes are spaced at 15° increments, placing the
last drop tube 180° from the supply tube.
To assist the mechanical
calibration process (see graphic #4), two small alignment holes have been
placed in the rotor plate, the first hole at 30° from the carrier hook and the
second hole at 150° from the carrier hook.
The rotor plate alignment holes can be easily detected using the TCS230
color sensor.
A simple hobby servo motor is
used in this project to move rotor plate to the various positions needed to
operate the sorting mechanism. The servo
is controlled by sending five volts electrical signals lasting between one and
two millisecond to the servo. The term
“pulse width” is commonly used to describe the process of sending electrical
signals of a specific duration to a servo motor. For example, the servo used in this project
will rotate to its maximum physical counter-clockwise position (see
graphic #3), when an electrical signal with a pulse width of 2.0
milliseconds is sent to the servo. The
servo will rotate to its maximum clockwise position when an electrical signal
with a pulse width of 1.0 millisecond is sent.
Pulse width signals between 1.0 millisecond and 2.0 milliseconds are
used to accurately position the servo at intermediate locations. It is important to note that the pulse width
values can be associated with the ten physical locations on the sorter
mechanism referred to previously.
The mechanical calibration process begins by
locating the second alignment hole on the rotor plate (see graphic #4);
this initial rotor plate positioning is accomplished by having a human operator
follow directional prompts displayed on the computer screen. Under program control the rotor plate is
slowly moved while values from the color sensor are used to locate the second
alignment hole. The second alignment
hole location is determined when the color sensor detects a lower light level as
the alignment hole moves under the sensor (see graphic #5). The lower light level results when the white
LED (light emitting diode) light passes through the alignment hole on the rotor
plate onto the black base plate that is located below rotor plate. The pulse width value for this alignment hole
location is saved, the value is referred to as “Align_hole_2” in the program
listing (see appendix A, "Program Listings"). The rotor plate is advanced and the same
process is used to determine the location the first alignment hole. The servo pulse width value for the first
alignment hole location is saved, the value is referred to as “Align_hole_1” in
the program listing (appendix A).
After determining the location of the two
alignment holes, sufficient information is available to calculate the location
of the ten target positions needed to operate the sorting mechanism. For clarity all of the calculations discussed
below are shown in a text box as equations.
The first calculation will provide the pulse width value equivalent to
15° of rotation, multiples of the 15° rotation value can be used to compute all
the other needed physical positions on the sorting mechanism. The two alignment holes are known to be 120°
apart, therefore the difference between the two alignment hole pulse width
values is equivalent to 120°. Dividing
the difference between the two alignment hole pulse width values by 8 will give
a pulse width value that is equivalent to 15°, this value is referred to as
“Pos_delta” in the program listing (appendix A).
Three pulse width values are
calculated that will be used to position the rotor plate carrier hook at one of
three locations: the supply tube position, the color sensor position, and the
first drop tube position. The relative
positions of the supply tube to the first drop tube and to the color sensor
position are known to be minus 30° and plus 45° respectively. Also, the carrier hook on the rotor plate is
known to be plus 30° from the first alignment hole. Calculation #2 computes the pulse width value
needed to position the carrier hook under the color sensor by subtracting the
pulse width equivalent of 30° from the first alignment hole pulse width
value. Calculation #3 computes the pulse
width value needed to position the carrier hook under the supply tube by
subtracting the pulse width equivalent of 60° from the first alignment hole
pulse width value. The last calculation
(#4) computes the pulse width value of the first drop tube position by adding
the pulse width equivalent of 15° to the first alignment hole pulse width
value.
The pulse width values for
the color sensor position, supply tube position, and first drop tube position
in addition to the “Pos_delta” value are written to the Electrically Erasable
Programmable Read Only Memory (EEPROM) for later use by the operation program
(appendix A). EEPROM is a type of memory
that can be changed, but retains its information when the power is turned off.
The other seven drop tube positions are calculated in the operation program to
save EEPROM space. The first time a sorting
system microcontroller completes the calibration process a calibration flag
value of “99” is written to the EEPROM.
Table #1 shows a partial EEPROM memory map.
Table 1: Partial EEPROM Memory Map
|
Location |
Usage |
|
0 |
Pos_delta, pulse width
equivalent to 15° |
|
2:3 |
First drop tube pulse
width position |
|
4:5 |
Color sensor pulse
width position |
|
6:7 |
Supply tube pulse
width position |
|
8 |
Calibration flag value (99 or 199) |
The program code for the
mechanical calibration begins at the “Arm_Calibration:” label in the
calibration program listing (appendix A).
The TCS230 Color
Sensor
The TCS230 sensor module (see
graphic #6) from Texas Advanced Optoelectronic Solutions (TAOS) has an
array of photodetectors most of which have one of three color (red, green, or
blue) filters. The photodetectors are
evenly distributed across the array to avoid location bias among the
colors. The sensor outputs a square-wave
whose frequency is proportional to the intensity of the selected color of
light. The color filter selection is
accomplished by pulling the S2 and S3 lines on the sensor high or low as 
shown in table #2. A program subroutine was developed
(“Check_color” in the program listings) to activate the sensor, turn on the two
white light emitting diodes, select the appropriate filter and obtain red,
blue, and green color intensity values.
More information about TCS230 sensor can be obtained from the Texas
Advanced Optoelectronic Solutions web site [www.taosinc.com].
Calibrating
the Color Sensor System
This implementation of the
M&M™ sorter system identifies six different colors of candies. The red, green, and blue (RGB) color
intensity values were obtained for two candies of each color and these values
averaged. The highest and lowest
acceptable RGB intensity values for each color were computed and then written
to the EEPROM for later use by the operation program.
The RGB intensity values were obtained experimentally
for twenty candies of each color (brown, yellow, orange, red, green, and
blue). The RGB intensity values were
analyzed for each color to determine an acceptable range of RGB values that
would accurately identify each color of candy.
The guideline used in determining the acceptable RGB ranges assume it is
better to classify a candies as unknown rather than sort it incorrectly. If the average color intensity value was
greater than 100 then an acceptable range was determined to be plus or minus
10% of the average value for each color.
However, if the average color intensity value was less than 100 then,
the red color range was determined to be plus or minus 20% of the average
value, the green color and blue color ranges were determined to be plus or
minus 30% of the average value. These
finding are summarized in table #3.
At the completion of the
color sensor calibration process a calibration flag value of “199” is written
to the EEPROM. The operation program
checks for this value in the EEPROM to confirm that the sorting system has completed
the required calibration steps. The
program code for the color sensor calibration begins at the
“Color_Calibration:” label in the calibration program listing (appendix A).
Operating
the M&M™ Sorter
The first task performed by
the operation program is to confirm that the calibration routines have been
successfully completed by verifying that the calibration flag value stored the
EEPROM has been set to ‘199’. Next, a
window is displayed on the computer instructing the operator to load the sorter
supply tube with candy and then press ‘1’ to begin the sorting process.
The sorting process begins by
moving the rotor plate with the candy of unknown color in the carrier hook from
the supply tube position to the color sensor position. After positioning the candy under the TCS230
color sensor, the sensor is activated and the red, green, and blue color
intensity values for the unknown candy are obtained. Imbedded in a controlled, “FOR/NEXT” loop are
statements used to read through RGB color intensity values of known candy
colors previously saved in the EEPROM and compare those values to the unknown
candy RGB color intensity values. A six
condition compound “IF” statement is used to compare the unknown candy RGB
intensity values to the acceptable high and low RGB intensity values computed
in the calibration program. If a known
candy color is found the program variable “mmColor” is assigned a value from
zero to five otherwise the “mmColor” variable is assigned a value of six
indicating the candy color can not be identified.
The last major task that the
operation program performs is to move a candy from the sensor position to the
appropriate drop tube position based on its color. The first drop tube is designated to hold
candies of undetermined color, drop tubes two through seven are designated to
hold one of the six known candy colors that are found. Recall that the calibration program determined
the pulse width position value of the first drop tube (Align_hole_1) and the
pulse width value equivalent to 15° (Pos_delta) and saved those values to the
EEPROM. The drop tubes are known to be
15° apart, therefore the position of drop tubes two through seven can be
computed by adding multiples of the “Pos_delta” value to the first drop tube
position value. For example the position
value of the third drop tube would be equal to the first drop tube position
value plus two times the “Pos_delta” value.
Occasionally the supply tube
becomes jammed by an oversized or irregularly shaped candy creating a
processing exception condition. The
system detects this condition when the color sensor is activated and the
resulting RGB color intensity values are all very low (less than 10 in most
cases). If a candy is not found under
the sensor, the operation program moves the carrier hook to the supply tube
position one more time in an attempt to “bump” a candy loose from the supply
tube. If after two attempts at obtaining
a candy, a candy is not moved under the color sensor it is assumed that the
supply tube is jammed or the supply tube is empty. In either case a message is displayed on the
computer screen requesting intervention by the operator.
Summary
Using the color sorting
simulation will help learners understand the processes needed to prepare
equipment to analyze and execute an operation based on color identification. By
understanding this process/procedure a learner can easily transfer much of this
skill into new situations that use color as a factor in an operation. Many other types of sensors require similar
procedures for calibration, so the process used to make these preparations are
transferable to other situations.
This type of system may be
enhanced by increasing the number of alternative storage locations, using
multiple sorting units, sending sorted items through additional processes
instead of storage containers, etc.
Color or other types of sorters are often only stations that are part of
a much larger processing sequence. For
simulations on smaller budgets, equipment used may be modified to replicate
each step of a process instead of sequencing multiple stations in a single
operation. The disadvantage of this
approach is the lag time required to modify a process, but the advantages can
be a dramatic reduction in expense or the ability to concentrate on individual
steps in a process one at a time.
With experience and
additional training, learners can begin to develop more and more sophisticated
systems and can begin to combine a number of different tasks together into a
larger coordinated process. By using
smaller simulations, errors constitute a smaller financial impact and a broader
variety of tasks can be experimented with in a shorter timeframe than might be
required for large scale operations.
Small simulations also require fewer personnel to operate and maintain
equipment.
An enhanced version of the
sorter has been developed that incorporates a small LCD (liquid crystal
display) in place of the computer monitor and pushbuttons mounted on the unit replacing
the computer keyboard. These
enhancements provide greater portability by reducing dependence on local
computer-based control. Future options
under development include the integration of a board level web server, radio
frequency (RF) communication devices, and streaming video. These enhancements will allow the sorting
device to be monitored and to some extent remotely controlled through the World
Wide Web. This will allow effective
instruction for learners who are not able to be visit the classroom and operate
the equipment directly. Methods and strategies for web-based control of
manufacturing equipment is explored in the article “Interfacing with
Manufacturing System in both Education and Industry Using Microcontrollers through
the World Wide Web” (Cotton & Johnson) in this issue. [1].
Acknowledgement
The authors would like to
thank Mr. Christopher A. Hileman for his technical contributions and dedication
to the development of this project.
References
[1] Cotton, S., Johnson, R. (2006). Interfacing with Manufacturing System in both Education and Industry Using Microcontrollers through the World Wide Web, the Technology Interface.