Sustainable Energy Technologies and Assessments 47 (2021) 101473

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments

journal homepage: www.elsevier.com/locate/seta

Bio-inspired algorithms for industrial robot control using deep

learning methods
Jiwen Guan a, Yanzhao Su b, *, Ling Su c, C.B. Sivaparthipan d, BalaAnand Muthu d
a
Guangdong University of Science and Technology, Dongguan City, Guangdong Province, 523083, China
b
State Key Laboratory of Mechanical Transmissions, Chongqing University, Chongqing 400044
c
Chongqing Changan Automobile Co., Ltd., Chongqing 401120, China
d
Department of Computer Science and Engineering, Adhiyamaan College of Engineering, India

A R T I C L E I N F O A B S T R A C T

Keywords: The biologically inspired algorithm is a significant embranchment of sequence on computational intelligence and
Deep learning plays a critical role in industrial robot control. Industrial robots are well suited for conducting manipulation or
Bio-inspired algorithm material handling functions in extremely constrained and predictable environments. By comparison, living or­
Industrial robot control
ganisms are highly resilient and able to perform tasks in evolving environments, such as stable locomotion over
uneven terrain. The robotics challenge is to use inspiration from biology to build devices capable of functioning
in unrestrained or moderately constrained environments. Hence, in this study, a Bio-inspired Intelligent Indus­
trial Robot Control System (BIIRCS) has been suggested using Deep Learning methods. A bio-inspired neural
network is considered to model the complex environment and to guide a team of robots for the coverage task.
The collected data is fed into a Deep Learning Neural Network to comprehend the localization and recognition of
present objects from various classes. With the derived data, appropriate robot actions can be planned and
executed. Usable objects are identified and seized in the robot’s workspace, or that feed is sufficient for unat­
tainable objects. This study confirms the ability to create intelligent systems using existing Deep Learning al­
gorithms and industrial robotics. The simulations’ findings reveal that the new approach achieves a high-
performance ratio of 83.5%, accuracy ratio of 95.4%, less operational time of 7.8%, low RMSE rate of 11.2%,
and increased coverage task rate of 96.7% when compared to other existing approaches.

Introduction to industrial robots Finally, the growing device sophistication creates a problem for process
management and the QoS in particular [6]. It includes path planning and
The advanced industrial robot technology introduced advanced a system for the study of end-to-end functionality [7].
manufacturing networks, physical, and Internet of Things [1]. There Cloud process is the technically independent subnets of a network
would be a significant growth in the amount of accessible physical op­ segment that may have such features as guarantee latency and stability
erations and a constant connection with local cloud providers to behave and exists as a cohesive system for implementations in a conceptual
intelligently and efficiently, bringing various challenges to both the manner that is exclusive of the core technology [8]. It can be developed
traditionally quite stable and segregated industrial networks [2]. Several with a constraining control of access that cannot leave the compartment
new technologies, wired and wireless, are increasingly integrated into in a particular process [9].
the manufacturing lines, leading to a rather diverse network [3]. It could be necessary to create a closed network to deliver low con­
A wide variety of implementations with differing Quality of Service sistency from a community of sensing tools to a network surface re­
demands from conventional private monitoring systems, accessing pository or cloud storage [10,11]. To ensure low latency, connectivity
sensors, and AI interfaces would be needed for the network [4]. In the and buffer services are allotted to one route according to the corporates’
order that the end-to-end delay, for instance, monitoring and warning planned statistical data, and single packet scheduling systems support
service, may need distribution reliability, whereas digital devices the network with high priority while retrieving [11-13]. Simulta­
simultaneously need significant data levels and modest latencies [5]. neously, another section of the network may allocate space on a separate

* Corresponding author.
E-mail address: [email protected] (J. Guan).

https://doi.org/10.1016/j.seta.2021.101473
Received 1 April 2021; Received in revised form 30 June 2021; Accepted 10 July 2021
Available online 26 August 2021
2213-1388/© 2021 Published by Elsevier Ltd.
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

physical network for a different application-oriented system or a high- Bio-inspired intelligent industrial robot control system (BIIRCS)
performance best effort operation, such as installing software up­
grades logically separated from essential control traffic from another Robots employed in different industries most commonly perform
computer cloud factory [14,15]. repetitive roles in today’s manufacturing process. The overall task per­
Therefore, Unlike the existing research, an advanced robotics formance of the robots depends on the accuracy of their controls to
involvement and management implementation in the industrial sector is detect predefined motions. The robots have very limited ability to
considered and calculated using end - end properties of the processing handle unorganized, dynamic situations such as the flexible seizure of
network. Along with that, for dynamic object detection, a different al­ previously unseen objects or the assembly of new components. Making
gorithm is proposed using deep learning methods. The proposed BIIRCS more intelligent machines capable of autonomous acquiring skills and
is effectively applied to collect data fed into a deep learning neural generalizing unforeseen circumstances is a game-changer for several
network to comprehend the localization and recognition of objects from industries. Industrial robot’s employment increases contemporary pro­
various classes. duction. Industrial robot assessment takes place in development: their
The leftover of the article is structured as follows: Section.2 Dis­ kinematics and control systems with modernized software applications.
cussed various industrial robot and deep learning application methods. The interest in naturally inspired algorithms has risen significantly in
Section.3 described the proposed BIIRCS and its associated deep recent decades. The popularity of machine learning technology is
learning strategies for object localization and recognition Section.4 growing due to the advancement of neural networks and the consequent
Presents the simulation studies and results along with discussion, and increase in deep learning technologies. This always reflects the
finally, the conclusion with future perspectives is discussed in section.5. increasing amount of computational power that citizens and small or
medium-sized businesses can readily access. The computerized infor­
Background study mation performance and these imposing outcomes from a certain area of
bio-inspired algorithms and computing power obtainability have a
Chattunyakit et al. [16] addressed the problems emerging from ro­ mutual interest in bio-inspired algorithms. In the resolution of engi­
botic network allocation architecture (RCNA). The technical dimensions neering problems, bio-inspired algorithms have evolved into a new
of the coexistence of committed and mutual systems were the emphasis. dimension.
It had explored that in particular, how a control on robot locomotion can The new category of bio-inspired robotics is based on learning from
be applied, with a concentration on the network and its effect on nature and extends to the system of real-world engineering. Bio-mimicry
network architecture. The drawback is that it is not a long-time evalu­ is copying the nature and design learned from nature, making the device
ation to produce a robust empirical result. or processes more effective and easier. In general, these biological sys­
Moazenzadeh et al. [17] presented a generalized Motion assessment tems are multifunctional and are built specifically for particular
robotic application (MAR) methods isolated from the practical appli­ coverage tasks. It is known as the transfer of natural engineering to other
cation of the theoretical underpinnings. The fact to use specified theo­ fields, including fabrication, materials science, design, etc. Artificial
rem to test the end-to-end network properties had demonstrated even intelligence robotics and the other areas that make advanced bio-
though the proposed techniques did not achieve greater efficiency in mixtures have made significant progress in the last few decades. This
implementation. interdisciplinary work led to systems that can interpret facial speech,
Balakrishnan et al. [18] presented a new structure of object detection communication, and locomotion in robust human-like bipedal gaits. In
arm architecture (ODAA) to satisfy the criteria. Different practical the development of intelligent biologically-inspired robots, the biolog­
resource parts can be installed on-demand to link various sites to their ical model and developments in analytic modeling, graphical simula­
extreme edges to provide the needed interoperability and handling tion, and the physical application of related technologies require
characteristics. In three possible use cases, it had confirmed that the understanding.
suggested system helped to demonstrate the anticipated advantages for Fig. 1 shows the Industrial Robotic Control Model. Motion planning
investors. The major limitation is that the run time efficiency and opti­ and path planning are indeed a programming problem to find the ob­
mization are not concentrated. ject’s right configuration from the source to the destination. In computer
Moshayedi et al. [19] demonstrated the deep learning classification geometry, computer entertainment, robots. The key reason we use
of objects in industrial robotics (DLCOR) and describe the air configu­ Motion Planners is that when the robot is in a dynamic environment, it
ration methods, radio access networks, core networks, computer plat­ accelerates the programming process. The road between two points in
forms, and more. This paper was intended to launch a dialogue on the the workspace is free of barriers for many robotics applications. A me­
long-term plan. It promoted the creation of approaches for the static chanical actuator acts to perform the movement, for example, by
network. The computation and the resources of the calculation are not translating one type of motion into another type, like a linear movement.
optimized. Mechanical actuators are based on structural part configurations,
Agrawal et al. [20] developed a mathematical model formalized as a including gears and rails and poles and chains. The ability to navigate
robotic classification security System (RCSS) to solve the offline and prevent obstacles and communicate with or control objects are the
embedment problem network. A responsive aim feature ensures the two common features used in robotics projects. Robotics developers
optimum network use and minimal delay in the connection between the focus on drives, elements that translate electronic controls into true
network. The major disadvantage is that were many uncertainties in the world motions, to incorporate this feature. A controller is a person who
parameter discussion, quality, and availability. is responsible for all accounting operations within a company, including
Chen et al. [21] suggested using diverse handling systems (DHS) high-level accounting, administrative, and financial activities. The
incorporated into the hierarchical communication and data processing controller reports significant variances in budgeting and administration
model in which the central processor, which guarantees an active global spending. The industrial robot controller’s main demand is to provide an
function of the network, manages the dispersed and local management efficient control device in which robot arms can be easily manipulated
decisions. Different operation stages included implementing a variety of and a simple interface between computer and man or sensor systems.
management techniques to satisfy their specific latency and durability The broadest concept is that a sensor is a detector, module, unit, or
criteria. The flexibility and reliability were not well satisfied. subsystem designed to sense events or changes in the environment and
Based on the survey, to resolve the existing system issues, BIIRCS is relay the information to a computer processor to other electronics. For
proposed, and the same has been solved using the deep learning tech­ other electronics, a sensor is still used. Machines require sensors in in­
niques with appropriate robot actions with proper security application dustrial and automation to provide them with the knowledge they need
as follows. to work properly. To improve their adaptability, several sensors can be

2
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

Fig. 1. Instrial Robot Control Model.

attached to various robots. Generating a trajectory: Construct a trajec­ Learning from Demonstrations, and trajectory learning issues. Repro­
tory for this (path + time scaling). In a given time, the robot can hit a set duction consequences endorse the helpfulness of this method for various
of stages. Roughly smooth trajectory and conform with mutual vari­ dynamic robot control and trajectory. The visual servo control and tra­
ables, speeds, accelerations, or torques constraints. The expected tra­ jectory imitation are collective in a single control task confirmed on
jectory travels from point A to point B and, over time, avoids collisions. robots in an industrial indoor atmosphere. It is common for mobile ro­
This can be measured in a discrete and continuing manner. Trajectory bots to express nonholonomic differential Kinematics,
planning for autonomous vehicles is a significant field for robotics. This
ẏ = f (y, v) (1)
paper proposed a Bio-inspired algorithm for industrial robot control
using deep learning methods such as deep belief networks (DBN). T
As shown in Eq. (1) where y = [ y x φ ] denotes the state vectors
Fig. 2 shows the human motion prediction in human-robot collabo­ (y and x indicates the robot location synchronizes in the plane and φ
ration. The movement observation and transition likelihood matrix are T
denotes the robot alignment), and v = [ u s ] signifies input (control)
then produced using the proposed BIIRCS. Based on the outcome, a vectors (u and s indicate angular and linear velocity, correspondingly).
human movement forecast becomes probable. The human intention is The nonholonomic differential kinematics can be obtainable in the state-
analyzed with the input of anticipated human movement. The forecast space formula as
human intention can be utilized as input for assistive robot movement ⎡ ⎤ ⎡ ⎤
planning. Therefore, industrial robots can cooperate and support human ẏ − sinφ 0 [ ]
⎣ ẋ ⎦ = I(φ)∙v(u, s) = ⎣ cosφ 0 ⎦∙ u (2)
employees based on intentional robot motions. The input to human s
movement prediction the result of human movement detection. φ̇ 0 1
Fig. 3 shows the proposed BIIRCS model. In industrial applications, The equation provides the no-slip condition. (3), which denotes that
the accuracy of robot pose assessment and safety necessities for work­ the robot can only change in the direction regular to the wheel axes
place monitoring in collaboration with the human-robot is required in
robotics. It provides visual and profound data at a considerably low-cost ẋcosθ − ẏsinθ = 0, (3)
and has some shortcomings though it is a major innovation concerning As inferred from Eq. (3), where θ is the robot positioning angle
robotics. Bio-inspired algorithms use depth data to identify objects, and restrained in a counter-clockwise way from the x-axis.
since the sensor creates the cloud point noisy, such information needs to This paper reflects the pinhole camera models, where the calibration
be enhanced. One idea is to integrate motion sensors with HD cameras, matrices are stated as
which can be used for robotic guidance or correction to create a high- ⎡ ⎤
resolution cloud of dense points. The movement sensors are paired βy w v0
with the 3D zone surveillance cameras authorized for safety. Combined
⎣
L = 0 βx u0 ⎦ (4)
virtual 3D robot models for tracking and collision identification with 0 0 1
sensor details (pictures of operators). 3D models used to reflect an As discussed in Eq. (5) where βy and βx are the focal length of the
organized floor surrounding and attached to actual movement sensors cameras articulated in pixel dimensions (in y and x directions corre­
are moved to imitate real-life actions. Motion sensors add foreign objects spondingly), v0 and u0 are the coordinates of the key points, and w is the
which have not been structured in 3D models, like mobile operators. A skew parameter. The viewpoint projection of 3D point Q onto the image
deep belief network has been proposed for object recognition using hydroplane is provided by
cameras. The error rate has been minimized based on the proposed
model in the industrial environment. q = L[R|t ]Q (5)
As derived in Eq. (5), where t and R are the translation and rotation
Model 1: Mathematical modeling among the world and the camera frames, stated in the camera frame
(extrinsic camera variables), correspondingly, and L denotes the
This study brings incorporation of the bio-inspired algorithm,

3
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

Fig. 2. Human motion prediction in Human-Robot Collaboration.

calibration mentioned above matrix (intrinsic camera variable). It can ⎡ ⎤

cosφ 0 sinφ
be easily demonstrated that a projective transformation G narrates the
R=⎣ 0 1 0 ⎦ (8)
projection of a similar 3D point Q onto the image flat surface of two
− sinφ 0 cosφ
cameras that be affiliated with a horizontal of the scene q.
qd = G∙qb (6) c = [y x 0 ]T (9)
As inferred from Eq. (6), where qb and qd indicates the projection of Given the similar planar motion parameter, homography G is pro­
the points Q in two independent camera pose (in real and desired camera vided by
poses, correspondingly), and G denotes a homography matrices. This ⎡ ⎤
g11 g12 g13
variable G can be connected to the camera movement utilizing the
G= 0⎣ 1 0 ⎦ (10)
subsequent equations g31 g32 g33
( ) ( ) ( )
mT − 1 mT − 1 mT − 1
G = L R− t L = L R + Rc L = LR J + c L (7) As discussed in Eq. (10), these matrices’ elements are determined by
d d d mounting the Eq. (7) utilizing Eqs. (4), (8), and (9) as
As shown in Eq. (7) where m = [ my mx mz ]T indicates ordinary to
the plane q and d is the distance of among the points Q in-plane p and the
camera harmonize model.
Fig. 4 shows the relationship between the current and desired robot
pose. As the wheeled robot utilized in this paper is moving in a plane,
camera translation and rotation matrices are stated as

4
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

Fig. 3. Proposed BIIRCS.

⎧
⎪ my
⎪
⎪ g11 = cosφ + (ycosφ + xsinφ)
⎪
⎪ d
⎪
⎪
⎪
⎪ βy
⎪
⎪ mx
⎪
⎪ g12 = + (ycosφ + xsinφ)
⎪
⎪ βx d
⎪
⎪
⎪
⎪ ( mz )
⎪
⎪ g13 = βy sinφ + (ycosφ + xsinφ)
⎪
⎨ d
( (11)
⎪
⎪ 1 my )
⎪
⎪ g31 = − sinφ + ( − ysinφ + xcosφ)
⎪
⎪ βy d
⎪
⎪
⎪
⎪
⎪
⎪ 1 m x
⎪
⎪ g32 = ( − ysinφ + xcosφ)
⎪
⎪ βx d
⎪
⎪
⎪
⎪
⎪
⎪ mz
⎩ g33 = cosφ + ( − ysinφ + xcosφ)
d
This paper utilizes the shortest path control method, in which the
robot vision-based movement is decoupled into rotation and translation
Fig. 4. Relationship between desired and current robot poses. stages. Robot linear velocity is identified utilizing the obtainable
homography-based control method. It can be perceived that the robot
moves forward in a line when φ = φt . Utilizing the geometrical equation
y = − xtanφt and equation assumed in (11), the homography G is stated
by

5
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

⎡ ⎤
cosφt 0 βy sinφt
1
⎢
⎢
⎥
⎥ P(o, k) = exp( − E(o, k) ) (16)
0 1 0 M
G=⎢
⎢ ( ) ⎥
⎥ (12)
⎣1
− sinφt +
x my 1 x mx
cosφt +
x mz ⎦ As inferred from the Eq. (16) where M is stated as normalization
βy cosφt d βx cosφt d cosφt d constant or portion function which is determined utilizing the subse­
quent equation:
Then, this paper defines a linear velocity of the mobile robot utilizing
∑∑
g33 constraints as M= exp( − E(o, k) ) (17)
(13)
o k
u = lu (g33 − cosφt )
The likelihood allocated to visible units is determined by marginal­
As derived in Eq. (13) where lu > 0 denotes control gain. Remember
izing hidden units,
that the homography matrices among the target and current picture can
be calculated in every control reiteration utilizing the plane parallel P(o) =
∑ 1
P(o, k) = exp( − E(o, k) ) (18)
notion. In conclusion, utilizing the expression for g11 from Eq. (12), this k
M
paper obtains the expression for velocity u as
The objects can be predictable in various positions, orientations, and
u = lu (g33 − g11 ) = − lu (g11 − g33 ) (14) lighting conditions by improving the number of images utilized during
the training process. The proposed method enhances the coverage task
Model 2: Deep belief networks (DBN) rate, accuracy, and performance, reducing operational time and RMSE
value compared to other existing methods.
This article defines the object in the area by the robot manipulator
using a USB camera. Clicking on the object’s image in the graphical user Results and discussion
interface can choose an object (GUI). The snapshot of a USB camera is
taken after clicking on the image object. Objects are localized and In this section, the proposed strategy of industrial robotics has been
divided into pixel sizes in the captured image. Per image extracted is validated through simulation. Several parameters have fixed for the
used as a DBNN input. DBNN outputs are each object’s recog­ simulation validation as in Table 1:
nition likelihood value. This attribute identifies the requested object or It has been found several robotic devices have been configured and
shows a message which does not occur the clicked object. The direction increased. They are geographically dispersed within a field of testing.
of the object is determined if the requested object is in the visual field. First, the design of the BIIRCS is introduced. It would then enforce the
For object recognition, this paper proposes a deep belief network model Static Setup of deep learning-based object classification and recognition
(DBN). The captured camera image is utilized as the input of the DBN. to reduce human involvement. The above has the same criteria for the
The DBN extracts the object features in the intermediate layers. The bias simulation and does not have the QoS targets. The aim is to research the
and weight connections among hidden and visible layers are stated by a QoS viability for classification using robotics strategy outcomes and
probability distribution utilizing the energy function. The joint config­ equate it with the conventional retention, coverage, RMSE in recogni­
uration energy among hidden and visible units in a restricted Boltzmann tion, and proper access rate with accuracy at the proper time of
machine is stated as follows operation.
∑ nk
no ∑ ∑
no nk
∑
E(o, k) = − ωji oj ki − aj oj − bi ki (15)
j=1 i=1 j=1 i=1 Coverage task rate
As shown in Eq. (15) where ωji signifies the symmetric interaction
The performance of classification and detection with the proposed
among hidden and visible units, o denotes the visible unit with
BIIRCS is analyzed with the above-said parametric simulation setup as in
o ∈ {0, 1}no , k indicates the hidden unit with k ∈ {0, 1}nk , no and nk de­
Fig. 6. The analysis confirmed that the proposed technique is well suited
notes the number of hidden and visible units, and aj and bi represents the
for the object detection industrial usage in the robotic application’s
bias for hidden and visible units, correspondingly. The network allocates
managing sector. The coverage’s efficiency reaches a peak, and it is too
a likelihood value in every state of hidden and visible units via the en­
compared with the existing robotics involved in industrial sectors. In
ergy function.
comparison, BIIRCS remains first, and it is an emerging technique for the
Fig. 5 shows the deep belief network. The joint probability distri­
evolution-based industrial robots. Figure 0.6 shows the better coverage
bution between hidden and visible units is stated as
rate of the BIIRCS and the foremost manner in techniques.Fig. 7.Fig. 8.

Accuracy rate

Figure 0.7 shows the accuracy of the introduced BIIRCS method. It

achieves high precision ineffective resource allocation for classification
and recognition using BIIRCS. The model is analytically analyzed for
evaluating the better result with varying numbers of nodes. It produces
excellent results than the pre-available industrial robotic approaches.

Table 1
Simulation Parameters.
Parameters Values

No. of conducts 10
Limit 8 min
No. of Nodes 10
No. of mediums 5
No. of objects 6
Fig. 5. Deep Belief Network.

6
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

RMSE ratio

Figure 0.8 shows the overall RMSE rate of the BIIRCS process in in­
dustrial robotics applications with security caution in classification and
recognition. It shows a minimum error rate of the process than the
existing strategies. From the graph, it is visible that there is a very
minimal overall error rate in the BIIRCS process. It is evident due to the
effective usage of the perfect pathfinding solutions under the concerned
direction distribution and stabilizes locomotion using the deep learning
classification process.

Performance ratio

Increased the performance of distributive clouds using current ap­

proaches over the industrial Robot application. BIIRCS is suggested and
implemented using simulation to reduce inappropriate access. Fig. 9
shows that performance strategies are likely to be boosted using the
suggested approach from simulated experiments. Due to the rise in ap­
plications, the access to recognition grew. So, with the growing number
of applications, the suggested procedure would reduce the undesired
Fig. 6. Coverage Task Rate.
delay and prevent a higher level of performance than the techniques
existing.

Operational time

As in Fig. 10, the operating time in the industrial application is lower

than in previous techniques, the operating time being decreased doubly
as the good locomotion, categorization, and identification of objects and
data is increased. A less working time rate than traditional approaches is
achieved with the suggested technique. With a decrease in working time,
the favourable characteristics of the model are increased and the rec­
ommended strategy to future industrial application creation is more
appropriate

Conclusion

The growing usage of intelligent techniques with expanded capa­

bility contributes to a revolutionary change in industrial robotics ap­
plications. To face modern security challenges, classification and
locomotion in recognition with low latency, high reliability, and per­
formance are required. The classification advantages should be applied
to meet the demands for excellent results. Throughout this regard, it has
Fig. 7. Accuracy Rate. been suggested a system for industrial robotics through BIIRCS to

Fig. 8. RMSE Ratio.

Fig. 9. Performance Ratio.

7
J. Guan et al. Sustainable Energy Technologies and Assessments 47 (2021) 101473

[2] Lv Z, Han Y, Singh AK, Manogaran G, Lv H. Trustworthiness in Industrial IoT

Systems Based on Artificial Intelligence. IEEE Trans Ind Inf 2021;17(2):1496–504.
[3] Manogaran G, Alazab M, Saravanan V, Rawal BS, Sundarasekar R, Nagarajan SM,
et al. Machine Learning Assisted Information Management Scheme in Service
Concentrated IoT. IEEE Trans Ind Inf 2020.
[4] Aladwan MN, Awaysheh FM, Alawadi S, Alazab M, Pena TF, Cabaleiro JC. TrustE-
VC: Trustworthy Evaluation Framework for Industrial Connected Vehicles in the
Cloud. IEEE Trans Ind Inf 2020;16(9):6203–13.
[5] Kulakov FM, Kadry S, Alferov GV, Efimova PA. Remote Control of Space Robots
Change-Adaptive in its External Environment. Int J Online Biomed Eng (iJOE)
2019;15(07):84–98.
[6] Kadry S, Alferov G, Kondratyuk A, Kurochkin V, Zhao S. Modeling the motion of a
space manipulation robot using position control. In: AIP Conference Proceedings.
AIP Publishing LLC; 2019. p. 080005.
[7] Xiao H, Muthu B, Kadry SN. Artificial intelligence with robotics for advanced
manufacturing industry using robot-assisted mixed-integer programming model.
Intel Serv Robot 2020:1–10.
[8] Sivaparthipan CB, Muthu BA, Manogaran G, Maram B, Sundarasekar R,
Krishnamoorthy S, et al. Innovative and efficient method of robotics for helping the
Parkinson’s disease patient using IoT in big data analytics. Trans Emerg
Telecommun Technol 2020;31(12). https://doi.org/10.1002/ett.v31.1210.1002/
ett.3838.
[9] Liu J-S, Pan W-H, Ku W-Y, Tsao Y-H, Chang Y-Z. Simulation-based fast collision
detection for scaled polyhedral objects in motion by exploiting analytical contact
equations. Robotica 2016;34(1):118–34.
[10] Zhang D, Wan J, Hsu CH, Rayes A. Industrial technologies and applications for the
Internet of Things. Comput Networks: Int J Comput Telecommun Network 2016;
Fig. 10. Operational Time. 101(C):1–4.
[11] Al-Turjman F, Alturjman S. Context-sensitive access in industrial internet of things
(IIoT) healthcare applications. IEEE Trans Ind Inf 2018;14(6):2736–44.
observe practically and dynamically with the efficiency criteria through [12] Al-Turjman F, Alturjman S. 5G/IoT-enabled UAVs for multimedia delivery in
the industrial networks. First of all, BIIRCS has proposed distributing the industry-oriented applications. Multimedia Tools Appl 2020;79(13-14):8627–48.
required divisions to satisfy the requirements and determine the speci­ [13] Sakthivel RK, Nagasubramanian G, Al-Turjman F, Sankayya M. Core-level
cybersecurity assurance using cloud-based adaptive machine learning techniques
fications for average throughput divisions in the deep learning-based
for manufacturing industry. Trans Emerg Telecommun Technol 2020;e3947.
classification of objects in recognition. Second, steps have been taken [14] Liao H, Zhou Z, Zhao X, Zhang L, Mumtaz S, Jolfaei A, et al. Learning-Based
to assign locomotion to the representatives of the path effectively. The Context-Aware Resource Allocation for Edge-Computing-Empowered Industrial
simulations’ findings reveal that the new approach achieves a high- IoT. IEEE Internet Things J 2020;7(5):4260–77.
[15] Mistry I, Tanwar S, Tyagi S, Kumar N. Blockchain for 5G-enabled IoT for industrial
performance ratio of 83.5%, accuracy ratio of 95.4%, less operational automation: A systematic review, solutions, and challenges. Mech Syst Sig Process
time of 7.8%, low RMSE rate of 11.2%, and increased coverage task rate 2020;135:106382. https://doi.org/10.1016/j.ymssp.2019.106382.
of 96.7% when compared to other existing methods. [16] Chattunyakit S, Kobayashi Y, Emaru T, Ravankar AA. Bio-inspired structure and
behavior of self-recovery quadruped robot with a limited number of functional
legs. Appl Sci 2019;9(4):799.
CRediT authorship contribution statement [17] Moazenzadeh R, Mohammadi B. Assessment of bio-inspired metaheuristic
optimisation algorithms for estimating soil temperature. Geoderma 2019;353:
152–71.
Jiwen Guan: Funding acquisition, Data curation, Formal analysis, [18] Balakrishnan N, Rajendran A. Computing WHERE-WHAT Classification Through
Investigation, Writing - original draft. Yanzhao Su: Writing - review & FLIKM and Deep Learning Algorithms. In: Advances in Electrical and Computer
editing. Ling Su: Methodology. C.B. Sivaparthipan: Project adminis­ Technologies. Singapore: Springer; 2020. p. 593–608.
[19] Moshayedi AJ, Roy AS, Liao L. PID Tuning Method on AGV (automated guided
tration. BalaAnand Muthu: Resources. vehicle) Industrial Robot. J Simul Anal Novel Technol Mech Eng 2020;12(4):
53–66.
Declaration of Competing Interest [20] Agrawal S, Simasinghe C, Jafari A, Etoundi A, Chong JJ. A De-risked Bio-inspired
Condylar Prosthetic Knee Joint for a Robotic Leg Test Rig. In: 2020 20th
International Conference on Control, Automation and Systems (ICCAS). IEEE;
The authors declare that they have no known competing financial 2020. p. 592–7.
interests or personal relationships that could have appeared to influence [21] Chen X, Huang J. Combining particle filter algorithm with bio-inspired anemotaxis
behavior: A smoke plume tracking method and its robotic experiment validation.
the work reported in this paper.
Measurement 2020;154:107482. https://doi.org/10.1016/j.
measurement.2020.107482.
References

[1] Cao Y, Jia F, Manogaran G. Efficient traceability systems of steel products using
blockchain-based industrial Internet of Things. IEEE Trans Ind Inf 2020;16(9):
6004–12.