IEEE Systems, Man and Cybernetics Magazine - October 2023 - 74

The Second Step (C)
Combined with artificial scene datasets and actual scene
datasets, various computational experiments are performed
to design and optimize vision algorithms and to evaluate the
performance of vision algorithms in complex environments.
Here, techniques such as machine learning, domain adaptation,
and statistical analysis are mainly used. In conclusion,
extending computational experiments from actual scenes to
artificial scenes not only broadens the breadth of experiments
but also increases the depth of experiments, which
helps to improve the performance of vision algorithms.
The Third Step (P)
The visual model is executed in parallel in the actual scene
and the artificial scene so that the model training and evaluation
are online and long term, and the visual system is continuously
optimized through the virtual-real interaction and
human-machine mixing between the actual and artificial. At
this point, we can further clarify the basic principles of parallel
vision: driven by physical and cyberspace big data; combined
with computer graphics, virtual reality, machine
learning, knowledge automation, and other technologies;
and using artificial scenes; computational experiments; parallel
execution; and other theories and methods to establish
a theoretical and methodological system for visual perception
and understanding in complex environments.
Parallel Loading and Unloading System
The parallel loading and unloading intelligent system is
the popularization and application of ACP theory in the
field of logistics loading and unloading. The parallel loading
and unloading intelligent system is mainly composed
of a virtual loading and unloading system and a real loading
and unloading system, as shown in Figure 5. Under this
framework, the main contents of the parallel loading and
unloading intelligent system are as follows.
Learning and Training
Through learning and training, the model in the physical
system is constructed in the virtual system to make it
approach the real system, and a comprehensive, true, and
accurate description of the real loading and unloading system
is established to form the " descriptive intelligence. "
Experiment and Evaluation
The virtual system analyzes and predicts the actual loading
and unloading process through computational experiments
and provides the best loading and unloading plan to
form the " predictive intelligence. "
Prediction and Correction
The solution generated by the virtual loading and unloading
system calculation experiment induces the actual loading
and unloading system. The actual loading and unloading
system returns the induced results to the virtual loading
and unloading system to form the " guided intelligence. "
Parallel Flexible Manufacturing Triplets
To make the production line in the real world operate better,
clarify and solve production line problems more clearly,
and increase production capacity, based on CPSS, we
apply it to three parallel worlds and propose " parallel flexible
manufacturing triplets. " The so-called " triplets " consist
of the physical world, the microphysical world, and the
artificial world, respectively. The physical world mainly
refers to the actual operation of the manufacturing production
line, and the microphysical world refers to a set of
a flexible parallel manufacturing production line designed
in a laboratory environment. The artificial world is the
digitization of the physical system in the virtual space.
The basic principle of the parallel FMS is to complete the
Parallel
Execution
tasks of the complex production line. The artificial world
simulates the production line situation and interacts with
the microphysical world and the physical world. The historical
data are trained in the virtual world to predict defective
products and modify detailed operations to avoid producing
defective products and reduce the probability of defective
products. The three worlds interact with each other, depend
on each other, provide feedback to each other, and jointly
restrict and improve flexible manufacturing performance.
The production process of the production line greatly
improves the efficiency of the industrial production process.
Figure 6 shows the relationship among the three.
We take a set of flexible parallel manufacturing chess
Artificial System
Virtual Model
Information
and Status
+
Parallel System
Figure 5. Modeling of the parallel loading and
unloading system.
74 IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE October 2023
Physical System
Real Equipment
production line systems as an example. In the laboratory,
the microphysics chess production line includes a
series of hardware equipment, mainly eight desktop
robots, several sensors and cameras, and other auxiliary
equipment. These hardware devices complete the chess
production process in the physical world. In the information
system, the core dimensions in the physical production
line are virtualized through a lightweight digital
twin, including the coordinates of the real-time pick-andplace
operation of the desktop robot, the joint angles of
each robotic arm, and the digital information received
by sensors and cameras.
In the chess intelligent
IEEE Systems, Man and Cybernetics Magazine - October 2023

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - October 2023
