Synchronized Control of Robotic Arm based on Virtual Reality and
IMU Sensing
CHIH-JER LIN, TING-YI SIE, YU-SHENG CHANG
Graduate Institute of Automation Technology,
National Taipei University of Technology,
1, Sec. 3, Zhongxiao E. Rd., Taipei 10608,
TAIWAN
Abstract: - This study introduces a robotic control system that combines virtual reality integration and inertial
measurement units (IMU) using mixed reality (MR) devices. The system integrates three main modules: (1)
virtual reality (VR), which simulates remote reality by the Hololens2, (2) the wearable IMUs device, which
captures the operator's hand movements; and (3) the robotic arm UR5, which is controlled by the user and the
VR environment. The virtual and physical systems communicate via a Message Queuing Telemetry Transport
(MQQT) communication architecture to establish communication between modules. To introduce a closed-loop
control system for the robotic arm, model predictive control (MPC) was achieved with precise path planning to
provide a flexible, intuitive, and reliable method to operate the remote-controlled manipulator. To validate the
system integration and functions, two demonstrations were conducted: (a) the offline mode, where the VR
module of the robotic arm was controlled by the IMUs device to check correctness, and (b) the online mode,
where the control command was transferred to UR5 to complete a target mission via artificial potential field
(APF) adjustment. The primary outcome of this study was the development of virtual and real industrial robotic
arms to test the developed model in VR and shop floor labs.
Key-Words: - model predictive control (MPC), MQTT, Hololens2, artificial potential field (APF), virtual
reality, path planning, visual localization, inertial measurement units (IMU).
Received: February 9, 2023. Revised: October 13, 2023. Accepted: November 19, 2023. Published: December 13, 2023.
1 Introduction
With the advancement of technology and artificial
intelligence, the application of remote manipulators
has been gradually growing, not only limited to
industrial automation but also extended to the fields
of healthcare, education, and environmental
monitoring. However, this remote detection type
still needs improvement, such as inaccurate
operation, slow response, and the inability to sense
the environment directly. Researchers have been
working on solving problems related to remote-
controlled robotic arms by integrating various
technologies in recent years. Head-mounted display
helmets and wearables are widely used in virtual
reality simulations and operational interfaces, [1],
[2].
The head-mounted display helmet can provide
the operator with an immersive virtual environment,
which allows the operator to experience the
operation of the robotic arm in an immersive
manner. This improves the intuitiveness and realism
of the operation. The wearable device can accurately
capture the movement information of the operator’s
arm to provide accurate input for remote control, [3].
In addition, a closed-loop control system for model
predictive control is usually introduced to improve
the control accuracy and stability of the remote
control. The control system predicts the movement
trajectory of the robotic arm based on the sensor
data and optimizes the control instructions for
smoother and more accurate path planning, [4]. This
method can reduce the influence of external
interference on operation and improve the system's
control performance and reaction speed.
Remote-control technology is used in various
fields, including manipulator operation and control
of robotic systems. However, traditional remote
controls suffer from communication delays, which
affect real-time performance and accuracy.
Researchers have recently addressed these
challenges by combining various techniques, such
as head-mounted displays and wearable devices, for
realistic simulations and model predictive control,
[5], [6], [7], [8]. Model prediction control utilizes a
mathematical model of a manipulator that predicts
the motion trajectory to generate a smooth and
accurate trajectory, as shown in Figure 1. The model
predictive control is a path-planning method based
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
460
Volume 18, 2023
on a system model that uses a mathematical model
to predict future motion trajectories. By predicting
and optimizing the model, smooth and precise
trajectories can be generated to achieve a higher
motion control performance. The model predictive
control method exhibited good adaptability and
response speed. It can cope with uncertainty and
external interference such that the robotic arm can
be controlled accurately in a complex environment.
This study aims to investigate the feasibility of a
remote-controlled manipulator for performing tasks
using head-mounted display helmets, wearable
devices, model prediction control, and intelligent
path planning. To realize accurate and efficient
remote-controlled manipulator operations, it
promotes applications and development to enhance
efficiency and safety and increase convenience and
possibilities, [9], [10], [11], [12], [13], [14], [15],
[16], [17], [18], [19], [20].
Fig. 1: Model Predictive Control illustration, [2]
2 System Description
The proposed system architecture comprises four
parts: a self-made wearable device to capture the
operator’s gestures, a virtual reality integration
platform to project the operation preview in real-
time, a synchronized manipulator for remote
control, and a transmission and computation
terminal program. As shown in Figure 2, the first
part is the arm pose capture device, which
comprises three inertial measurement units (IMU) to
measure the arm posture of the operator. Through
the application of a Kalman filter, the IMU data can
be calibrated and estimated to obtain accurate
attitude information, which is analyzed by the
program and transmitted to the robotic arm UR5 for
manipulator synchronization control. The second
part is a virtual reality integration platform, which
receives the data from the first part and displays
them on HoloLens 2. A self-developed C# program
receives the operator’s posture to display on the
HoloLens 2. The third part is the terminal program,
which integrates the commands to the actual
hardware and transfers the program to the
manipulator to perform the closed-loop control
algorithm. The fourth part describes the real-time
hardware used in this experiment, including a UR5
collaborative manipulator, Robotiq 2F-85 two-
finger servo gripper, and Intel RealSense D435
depth camera. Table 1 presents the software and
corresponding applications used in this study.
Fig. 2: System Architecture Diagram
Table 1. Software and applications used in the study
Development
environment
Integrated content and subparts
Python
Analyze manipulator posture
Object recognition(D435)
Path planning(MPC&APF)
Inverse Kinematics(IK) calculation,
Data transmission
Arduino
Capturing human arm information
Visual Studio
C# (Unity)
Manipulator model simulation
Human posture visualization
Hololens real-time demonstration
Provide operation UI
SolidWorks
Draw and export 3D object models.
2.1 Wearable Manipulator Instructor and
Gesture Gripper Control
Self-developed wearable devices include battery-
charging, boost-voltage, microcontroller, and
multiplexed I2C circuits. These circuits were
integrated to capture the operator's gestures. Three
BNO055 integrated 9-axis inertial measurement
units (IMUs) are used for the operator’s hand
motion tracking. Each IMU chip contains an
accelerometer and gyroscope sensors and is referred
to as a 6-axis IMU or 6 D.O.F. IMU. The IMU
includes a magnetometer besides the accelerometer
and gyroscope sensors. The IMU data can be
calibrated and estimated by applying a Kalman filter
to obtain accurate attitude information from the
operator. In terms of communication integration, all
the three IMUs were equipped with I2C
communication functions. The problem of
conflicting I2C addresses is solved using
multiplexers that connect multiple I2C devices with
the same address simultaneously. In addition, a pair
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
461
Volume 18, 2023
of data gloves were fabricated in our laboratory, as
shown in Figure 3(a). It uses a microcontroller and
bending sensors to recognize the operator's hand-
paw operation. The hand position and gesture
information can be transmitted to the processing
unit of the wearable device, and the information can
be uploaded to the host computer through radio
frequency (RF) wireless transmission to realize
manipulator arm control based on human gestures.
The overall hardware device usage is shown in
Figure 3(b).
(a) (b)
Fig. 3: (a) Self-made data glove (b) Overall
hardware setup
2.2 Virtual-Reality Integration Platform
The real-time visualization environment of
HoloLen2 is compiled and programmed using Unity
software. In the experiment, we created a human-
machine interface (HMI), as shown in Figure 4, and
the most important aspect was the model. To match
the hardware model used in the experiment, the
UR5 model and gripper were imported into Unity.
To comply with the virtual environment, it is
necessary to disassemble UR5's movable joints one
by one and reset the home coordinates of each
component. The gripper model is also disassembled
based on all movable joints. This disassembly
method allows the gripper components to be
operated independently, which better mimics the
real-world behavior of the gripper. After the
assembly is completed, the rotational direction must
be confirmed. Unity has the possibility of a Gimbal
Lock. This is mainly because during the rotation
process, if a specific axis is rotated by 90 °, the
subsequent rotations will lose the rotation
dimension, resulting in an unexpected rotation
result. Finally, all the models were assembled in the
Unity development environment according to the
relative position of the assembly, as shown in Figure
5(a) and Figure 5(b).
Fig. 4: Unity user interface
(a) (b)
Fig. 5: (a) UR5 and Gripper Assembly (b) Total
Components of the Model
2.3 MQTT-based Communication
Architecture
The communication of the proposed system is based
on a message-queuing telemetry transport (MQTT)
communication architecture to integrate each device.
The communication architecture is shown in Figure
6. The local computer is responsible for processing
the hand posture and position returned by the
wearable IMUs device. The IMU data are calculated
and transferred to the joint angle using inverse
kinematics. The values were corrected using model
predictive control (MPC) and artificial potential
field (APF) algorithms. The data are then uploaded
to the MQTT, and Hololens2 subscribes to the UR5
joint angles and gripper information through the
MQTT. It then immediately displays the operation
to local personnel. The remote computer
communicates between the workstation equipment
and the MQTT. The remote computer subscribes to
the UR5 joint information and gripper status and
sends the real UR5 information from the remote
computer to the MQTT. The actual UR5 joint angle
was provided to the local MPC algorithm for the
calculation. The object coordinates of D435 were
provided to the APF algorithm to perform object
attraction for path planning.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
462
Volume 18, 2023
Fig. 6: Communication Architecture Diagram
3 Methodology
3.1 IMU-Based Human Hand Position and
Orientation Measurement
The derivation of the angle of the quaternion
estimation using the quaternion output from sensor
fusion is as follows.
(1) Quaternion Pose Estimation
A quaternion consists of one real and three
imaginary parts, represented by Eq. (1).
(1)
Based on the Euler angle definition for the three
rotation angles, the quaternion can be expressed
using Eq. (2).
Mathematically, owing to the symmetry between the
rotation matrices and quaternions, the Euler angles
can be expressed in terms of the quaternions, as
shown in Eq. (3).
Human Arm Posture
The IMU modules were attached to the upper arm,
forearm, and back of the hand, and the IMU data
were used to compute arm-related information. The
X, Y, and Z values for the hand were derived as
follows.
In Eq. (4), the notation represents rotation in
the order of the z-, y-, and x-axes. The symbols ψ, θ,
and ϕ represent the rotation angles around the z-
axis, y-axis and x-axis, respectively. In Eq. (4), the
IMU X-axis is computed by multiplying the rotation
matrix by
to represent the IMU X-axis vector.
Finally, based on Figure 7 and Eqs. (4), the X, Y,
and Z values of the arm can be obtained by
multiplying with the arm length, as shown in Eq.
(5).
(5)
The X-, Y-, and Z-coordinates of the hand can be
converted from the two IMUs of the upper arm and
forearm. The 3-axis rotation angle of the IMU at the
back of the hand is directly expressed as the roll,
pitch, and yaw values of the wrist in space, which
can be used to express the six degrees of freedom of
the hand in space.
Fig. 7: IMU Vector Relationship Diagram
3.2 Path Planning and Visual Localization
(1) Visual Localization
In visual localization, we must perform several
coordinate system conversions to convert 2D planar
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
463
Volume 18, 2023
coordinates into 3D world coordinates. For object
recognition and localization, we used a RealSense
depth camera to capture the object and obtain its
depth information. The object's position in the
image can be converted through three-dimensional
coordinate conversion into the position in the
coordinate system of the robotic arm base. The
object position is recognized by removing noise and
calculating the center point, and coordinate
conversion is performed to achieve object
recognition and positioning. This process consists of
several steps: the first step is from camera modelling
to depth computation; the second step is object
recognition and coordinate conversion; and the final
step is to achieve stable and accurate object
positioning within 3 mm, as shown in Figure 8.
(a) (b)
Fig. 8: (a)D435 Mounting Schematic (b) Object
Coordinate Calculation
(2) Artificial Potential Field
In the remote control of a robotic arm, an artificial
potential field (APF) can be used to control the
movement of the robotic arm and quickly guide it to
the target position. An attractive potential energy
field (APF) is a common type of potential energy
field. Attractive APFs allow the robotic arm to be
subjected to a force toward the target position,
which assists the remote operator in quickly
acquiring an object. The attractive potential field is
given by Eq. (6).
(6)
where () denotes the current position of the
robotic arm, represents the target position,
represents the joint angles of the robotic arm, and
is a parameter that adjusts the influence of the
attractive potential field.
The gradient of the attractive potential field can be
expressed by Eq. (7).
(7)
where represents the force exerted on the
manipulator and can be used to control the direction
and velocity of the motion of the robotic arm. When
the manipulator enters the potential field,
points toward the target position. Adjusting the
parameter that influences the attractive potential
field can control the magnitude and range of
attraction. Thus, it influences the trajectory of the
manipulator movement, as shown in Figure 9.
(3) Model Predictive Control
The model predictive control (MPC) is a method
used to implement system trajectory planning and
control. MPC uses mathematical models to predict
the system behavior, which makes optimal control
decisions based on future predictions. The MPC
derivation process can be described as follows.
There are four elements in the MPC model:
prediction, rolling optimization, error compensation,
and derivation of the MPC formula, where P
denotes the prediction step, and M denotes the
control step.
Fig. 9: Schematic diagram of APF
3.3 Model Establishment
When modeling predictive control, the system under
control is subjected to step inputs, and step-response
outputs are observed. Assuming that the
corresponding outputs of the model at moments T,
2T, ..., PT are , according to the
superposition principle of linear systems, the
augmented representation can be expressed as
shown in Eq. (8).
(8)
(1) Prediction Model
From Eq. (8) for y(k), an expression can be obtained
to predict the output for n steps, as expressed in Eq.
(9).
(9)
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
464
Volume 18, 2023
If is the predicted output from past inputs, it can
be obtained using Eq. (10).
(10)
In matrix form, this can be represented by Eq.
(11).
(11)
(2) Rolling Optimization
First, the desired trajectory must be determined to
achieve a smooth and stable output. Therefore, a
first-order filter is often used to generate the desired
trajectory, denoted by w. The core of the rolling
optimization is to find the optimal solution of the
objective function to determine the control
increment ∆u. The objective function is essentially a
quadratic polynomial in terms of ∆u, which can be
given by Eq. (12).
(12)
The objective function is primarily separated into
two parts: the first is to minimize the difference
between the actual and desired outputs, and the
second is to approach the target with as little energy
as possible. In Eq. (12), q and r are the weighting
coefficients of the two objectives. The relative
settings of q and r are inversely proportional. For
example, if the ratio of q to r is 10, this indicates a
greater emphasis on minimizing the control effort
(energy consumption). In contrast, if the ratio of q to
r is 0.1, this signifies a greater emphasis on quickly
approaching the target. Subsequently, transforming
it into a matrix form leads to the expression of ,
as shown in Eq. (13).
(13)
(3) Feedback Correction and Error Compensation
Suppose that, at the current time k, we predict the
output of P and at time k+1, the current value of y is
y(k+1). A prediction error exists at this point, as
expressed in Eq. (14).
(14)
Using this error to correct future predictions, Eq.
(15), where H denotes a weighted matrix.
(15)
The prediction process can be simplified as
follows: At time k, the prediction of the forward P
steps is as described in Eq. (16).
(16)
Therefore, by moving to time k+1 and
continuing the prediction forward, Eq. (17) can be
obtained as.
(17)
If we use shift matrix S, the initial prediction
can be expressed as shown in Eq. (18).
(18)
In summary, MPC is characterized by allowing
some error in the predicted model, which allows
optimized control to be achieved within a limited
period of time. At each step of the control execution,
the next step is recalculated and reoptimized (rolling
optimization). It has strong tuning capability and
can be used for calculations when the control system
is not precisely modeled. Figure 10 shows the test
results of the experimental control for robot UR5,
where the movements are prerecorded trajectories.
Fig. 10: Joint Responses with and without
Controller
4 Result and Discussion
In this study, two experimental modes were
implemented: offline and online. In offline mode,
trajectory planning and smooth control are achieved
through curve fitting to enable the execution of
predetermined arm movements. In online mode, a
feedback-controlled system was used to compensate
for disturbances and ensure accurate movements.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
465
Volume 18, 2023
Figure 11 illustrates (a) a flowchart of the object
update, (b) offline teaching, and (c) the online mode.
(1) Offline mode control
In assembly lines, it is necessary to perform the
same action repeatedly. We introduced the "offline
mode" based on this architecture to cope with this
situation. In this mode, operators interact with
virtual objects through their actions. Hololens2
provides a local, fully simulated, and visually
interactive environment in these operations. After
simulating the system, the script for the entire action
process was provided to the remote workstation for
execution. The overall flow is shown in Figure 12.
The real-time visualization environment of
HoloLen2 is compiled and programmed using Unity
software. In the experiment, users controlled the
virtual model in a unity development environment
according to the relative position of the assembly, as
shown in Figure 4.
(2) Online mode control
In addition to the applications mentioned above, in
which repeated playback is required for offline
mode control, real-time remote control is required
when dealing with more complex problems.
Compared to storing an entire batch of scripts and
transmitting them all at once, the requirements for
the online mode are more stringent for controllers.
The online mode is equivalent to a closed-loop
controller across the network in terms of
architecture, as shown in Figure 13, where the local
operation of the user is sent directly to the remote
manipulator via the MQTT. The visualization part
of the remote environment provides positional
updates of the interactive model in the Hololens2
through a D435 camera. Local and remote network
interference is inevitable, and packet loss can cause
problems such as unsynchronized operation and
discontinuous tracking angles of the joints.
(3) Anti-Interference Test
Because network interference is unavoidable, we
approximated the synchronization effect of remote
UR5 when the connection was poor by adding two
degrees of simulated noise to the joint
synchronization control. The original UR5 tracking
results are presented in Figure 14. It can be seen that
the original UR5 controller tries to track these
bouncing disturbances, which cause abnormal
vibrations during the UR5 synchronization
operation and increase the difficulty of remote
control. Figure 15 shows the tracking response of
the MPC control strategy used, and it can be seen
that after the MPC adaptation, the response of the
joints has been improved to a great extent, and the
interference resistance is more favorable compared
to Figure 14.
(a)
(b)
(c)
Fig. 11: (a)Object update, (b) Offline Mode Control,
(c) Online Mode Control
Fig. 12: Offline-mode diagram
Fig. 13: Online-mode diagram
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
466
Volume 18, 2023
Fig. 14: Joint control response with interference
(original)
Fig. 15: Joint control response with interference
(MPC optimized)
5 Conclusions
A VR (Virtual Reality) integrated controller was
implemented to test the system integration and
functions. Two control architectures were
investigated in this study: offline and online. The
offline mode uses a curve-fitting method for
trajectory planning and implements the VR model
using the HoloLen2 model to check the control
curve to achieve a predefined arm motion. The
online mode considers the need for remote control
and introduces an APF controller with feedback for
trajectory planning and control. This allows the real-
time controller to compensate for the robotic
trajectory, and even if disturbances are encountered
during remote control, it ensures the accuracy and
stability of the robotic motion. In summary, the
proposed controllers provide a reliable control
strategy for remote-controlled systems and offer an
effective solution for achieving accurate motion.
Acknowledgement:
The authors would like to thank the National
Science Council of the Republic of China, Taiwan,
for financial support of this research under Contract
No. NSTC 112-2218-E-027-007.
References:
[1] J. Guhl, S. Tung, and J. Kruger, "Concept and
architecture for programming industrial robots
using augmented reality with mobile devices
like microsoft HoloLens," 2017 22nd IEEE
International Conference on Emerging
Technologies and Factory Automation
(ETFA), Limassol, Cyprus, 2017, pp. 1-4, doi:
10.1109/ETFA.2017.8247749.
[2] Oleinikov, S. Kusdavletov, A. Shintemirov
and M. Rubagotti, "Safety-Aware Nonlinear
Model Predictive Control for Physical
Human-Robot Interaction," in IEEE Robotics
and Automation Letters, vol. 6, no. 3, pp.
5665-5672, July 2021, doi:
10.1109/LRA.2021.3083581.
[3] M. Rubagotti, T. Taunyazov, B. Omarali and
A. Shintemirov, "Semi-Autonomous Robot
Teleoperation With Obstacle Avoidance via
Model Predictive Control," in IEEE Robotics
and Automation Letters, vol. 4, no. 3, pp.
2746-2753, July 2019, doi:
10.1109/LRA.2019.2917707.
[4] N. Zhang, Y. Zhang, C. Ma, and B. Wang,
"Path planning of six-DOF serial robots based
on improved artificial potential field method,"
2017 IEEE International Conference on
Robotics and Biomimetics (ROBIO), Macau,
Macao, 2017, pp. 617-621, doi:
10.1109/ROBIO.2017.8324485.
[5] K. -H. Lee, V. Pruks and J. -H. Ryu,
"Development of shared autonomy and virtual
guidance generation system for human
interactive teleoperation," 2017 14th
International Conference on Ubiquitous
Robots and Ambient Intelligence (URAI), Jeju,
Korea (South), 2017, pp. 457-461, doi:
10.1109/URAI.2017.7992775.
[6] A. Vick, J. Guhl, and J. Krüger, "Model
predictive control as a service — Concept and
architecture for use in cloud-based robot
control," 2016 21st International Conference
on Methods and Models in Automation and
Robotics (MMAR), Miedzyzdroje, Poland,
2016, pp. 607-612, doi:
10.1109/MMAR.2016.7575205.
[7] S. Kleff, A. Meduri, R. Budhiraja, N. Mansard
and L. Righetti, "High-Frequency Nonlinear
Model Predictive Control of a Manipulator,"
2021 IEEE International Conference on
Robotics and Automation (ICRA), Xi'an,
China, 2021, pp. 7330-7336, doi:
10.1109/ICRA48506.2021.9560990.
[8] S. V. Ghoushkhanehee and A. Alfi, "Model
Predictive Control of transparent bilateral
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
467
Volume 18, 2023
teleoperation systems under uncertain
communication time-delay," 2013 9th Asian
Control Conference (ASCC), Istanbul,
Turkey, 2013, pp. 1-6, doi:
10.1109/ASCC.2013.6606130.
[9] Nicola Piccinelli, Riccardo Muradore,”A
bilateral teleoperation with interaction force
constraint in unknown environment using
nonlinear model predictive control,”
European Journal of Control, Vol. 62, 2021,
pp.185-191, ISSN: 0947-3580.
[10] Bukeikhan Omarali, Tasbolat Taunyazov,
Askhat Bukeyev, and Almas Shintemirov.
2017. Real-Time Predictive Control of an
UR5 Robotic Arm Through Human Upper
Limb Motion Tracking. In Proceedings of the
Companion of the 2017 ACM/IEEE
International Conference on Human-Robot
Interaction (HRI '17). Association for
Computing Machinery, New York, NY, USA,
237–238.
[11] H. Lim, Y. Kang, C. Kim, J. Kim and B. -J.
You, "Nonlinear Model Predictive Controller
Design with Obstacle Avoidance for a Mobile
Robot," 2008 IEEE/ASME International
Conference on Mechtronic and Embedded
Systems and Applications, Beijing, China,
2008, pp. 494-499, doi:
10.1109/MESA.2008.4735699.
[12] M. Noroozi and L. Kies, "Performance
Analysis of Universal Robot Control System
Using Networked Predictive Control," 2022
7th International Conference on Robotics and
Automation Engineering (ICRAE), Singapore,
2022, pp. 227-232, doi:
10.1109/ICRAE56463.2022.10056165.
[13] Barnali Das, Gordon Dobie,” Delay
compensated state estimation for Telepresence
robot navigation,”Robotics and Autonomous
Systems, Vol. 146, 2021, 103890, ISSN 0921-
8890.
[14] F. Cursi, V. Modugno and P. Kormushev,
"Model Predictive Control for a Tendon-
Driven Surgical Robot with Safety
Constraints in Kinematics and Dynamics,"
2020 IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS), Las
Vegas, NV, USA, 2020, pp. 7653-7660, doi:
10.1109/IROS45743.2020.9341334.
[15] T. Erez, K. Lowrey, Y. Tassa, V. Kumar, S.
Kolev and E. Todorov, "An integrated system
for real-time model predictive control of
humanoid robots," 2013 13th IEEE-RAS
International Conference on Humanoid
Robots (Humanoids), Atlanta, GA, USA,
2013, pp. 292-299, doi:
10.1109/HUMANOIDS.2013.7029990.
[16] J. Lee, M. Seo, A. Bylard, R. Sun and L.
Sentis, "Real-Time Model Predictive Control
for Industrial Manipulators with Singularity-
Tolerant Hierarchical Task Control," 2023
IEEE International Conference on Robotics
and Automation (ICRA), London, United
Kingdom, 2023, pp. 12282-12288, doi:
10.1109/ICRA48891.2023.10161138.
[17] Cho J, Park JH. Model Predictive Control of
Running Biped Robot. Applied Sciences.
2022; 12(21):11183.
https://doi.org/10.3390/app122111183.
[18] Künhe, F., Gomes, J., & Fetter, W. (2005,
September). Mobile robot trajectory tracking
using model predictive control. In II IEEE
Latin-Americat robotics symposium, Vol. 51.
[19] Taïx Michel, Flavigné David, Ferré Etienne.
Human interaction with motion planning
algorithm. Journal of Intelligent & Robotic
Systems, 2012, 67: 285-306.
[20] M. S. Ahn, H. Chae, D. Noh, H. Nam, and D.
Hong, "Analysis and Noise Modeling of the
Intel RealSense D435 for Mobile Robots,"
2019 16th International Conference on
Ubiquitous Robots (UR), Jeju, Korea (South),
2019, pp. 707-711, doi:
10.1109/URAI.2019.8768489.
Contribution of Individual Authors to the
Creation of a Scientific Article
- Chih-Jer Lin, Yu-Sheng Chang carried out the
simulation and the optimization.
- Yu-Sheng Chang has implemented the Algorithms
in Visual Studio C# (Unity).
- Ting-Yi Sie and Yu-Sheng Chang have organized
and executed the experiments of Section 4.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
The authors would like to thank the National
Science Council of the Republic of China, Taiwan,
for financial support of this research under Contract
No. NSTC 112-2218-E-027-007.
Conflict of Interest
The authors have no conflicts of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2023.18.49
Chih-Jer Lin, Ting-Yi Sie, Yu-Sheng Chang
E-ISSN: 2224-2856
468
Volume 18, 2023
