Design of a controller for wheeled mobile robots based on automatic movement sequencing

TELKOMNIKA Telecommunication, Computing, Electronics and Control
Vol. 18, No. 6, December 2020, pp. 3088~3095
ISSN: 1693-6930, accredited First Grade by Kemenristekdikti, Decree No: 21/E/KPT/2018
DOI: 10.12928/TELKOMNIKA.v18i6.16431  3088
Journal homepage: http://journal.uad.ac.id/index.php/TELKOMNIKA
Design of a controller for wheeled mobile robots based on
automatic movement sequencing
Holman Montiel Ariza, Fredy H. Martínez S., Fernando Martínez S.
Facultad Tecnológica, Universidad Distrital Francisco José de Caldas, Colombia
Article Info ABSTRACT
Article history:
Received Apr 19, 2020
Revised Jul 06,2020
Accepted Jul 26, 2020
There are many kinds of robots and among them the wheeled mobile robots
(WMR) stand out, because they are relatively cheap and easy to build. These
features make WMRs the test prototypes for control strategies or motion
generation. In general, the controllers developed are based on sensory schemes
that give an WMR the ability to travel through flat or obstructed environments.
However, these strategies are highly reactive, i.e. they are based on the control-
action scheme and are not adaptive; or, they are motion schemes built from
simulations that assume the environmental conditions to determine the robot's
path. In both cases, WMRs do not adapt perfectly to the change of
environment, since the controller does not find appropriate movements for the
robot to move from one point to another. Therefore, this article proposesa
partial solution to this problem, with a controller that generates sets of adaptive
movements for an WMR to travel around its environment from the sensory
perception information.
Keywords:
Controller
Genetic algorithm
Optimization algorithm
Sensor
Wheeled mobile robots
This is an open access article under the CC BY-SA license.
Corresponding Author:
Holman Montiel Ariza
Facultad Tecnológica,
Universidad Distrital Francisco José de Caldas,
Cll 68 D Bis A Sur No. 49F–70, Bogotá D.C, Colombia.
Email: hmontiela@udistrital.edu.co
1. INTRODUCTION
A robot is basically a system that performs one or more tasks automatically, which increases the level
of precision and accuracy that a human being achieves when performing high-risk, heavy or repetitive
activities. For example, virtual assistants that help online users or robotic manipulators that perform welding
tasks in automotive manufacturing lines [1-4]. As can be seen, the robot definition is not only applied to designate
prototypes with mechanical and electrical parts, but also to designate virtual prototypes that have a digital
representation or are available in computer systems. Therefore, there is many robots and ways to classify them, such
as mobile robots that are transport machines with a certain degree of autonomy and locomotion to travel around their
environment [3, 5, 6].
Wheeled mobile robots (WMR) are a case of a mobile robot and have advantages such as; the ease of
physical implementation or the ability to travel long distances in a short time. The ease of implementation has
increased the mass of this type of robot and the frequency in industrial or research applications (e.g. the “Mars
Exploration Rover” used in space missions) [6-8]. Another feature of WMRs is that they incorporate groups of
instruments, sensors or accessories to interact with their environment, which attempts to reduce or eliminate
human intervention and increase the level of autonomy during the execution of certain tasks. Although
the autonomy level of an WMR is high enough to solve a problem, there are electrical (energy supply),

TELKOMNIKA Telecommun Comput El Control 
Design of a controller for wheeled mobile robots based on automatic movement (Holman Montiel Ariza)
3089
mechanical (repair or replacement of parts), computer (generation of control instructions) or electromechanical
(integration of components) limitations that do not allow it to achieve this capacity in its entirety [9, 10].
Some proposals that achieve a certain level of autonomy of an WMR are related to the capacities to
perceive, shape, plan and act, which have been worked on from different points of view. One of these proposals
is the reactive scheme based on perception and action, i.e. the WMR implements a sensor that, on receiving
a stimulus from the environment, modifies the state of the actuators. Although this scheme reduces
the processing speed of the robot controller (it requires few resources to operate), it limits the amount of
movements of the actuators because they are reduced to the indications given at the time of generating
the controller [11-13].
Increasing the number of motions available on an WMR controller is not a simple task, since increasing
the robustness of the controller reduces the robot's operating speed when performing a task. Therefore, motion
and control schemes have been widely studied and there are remote or local implementation algorithms for
operating an WMR. Some of the most used algorithms currently are the route mapping algorithms, machine vision
algorithms and local operation algorithms. First, there are the routing algorithms that from a defined environment
create a sequence of instructions that are sent to the robot, to go from one place to another. In second place, there
are the machine vision algorithms that by means of an image processing algorithm determine the location of
obstacles so that the robot avoids them with the advantage that the environment can be partially known. Finally,
the local operation algorithms are operating routines that are implemented directly on the robot controller and
depend on the response of the sensors to the environment [12-14].
In summary, the range and speed of response of an WMR depends on the robustness of the controller.
However, the efficiency of a controller is a subject under study, since, current controllers are subject to
variables that limit the amount of movement of the actuators and the ability to explore their environment.
Therefore, this article proposes a control strategy for WMR that generates instruction libraries without a defined
depth using an optimization algorithm so that the robot performs movements in an adaptive way. The algorithm
developed and the results obtained are reported in the following sections, which are organized as follows:
section 2 describes the concepts required to understand the functioning of the algorithm, section 3 presents the
proposed algorithm and describes its operation. Finally, section 4 presents the results obtained.
2. MATERIALS AND METHODS
The WMR controller proposed in this article codes the movements of the actuators of a unicycle robot
using the ROS (robot operating system) language, from the motion sets built using an optimization algorithm
in a simulator set up for that purpose. These concepts are explained in detail in this section.
2.1. Unicycle robot
The combination of different types of wheels or their location within the structure of an WMR changes
the kinematics of the robot, so they have been grouped in different ways. Although the functionality of
the WMR allows it to be classified according to its specialty (e.g., search or rescue), the arrangement of its
wheels is the most common form of classification, since it differentiates robots with respect to their
maneuverability, and among them are: omnidirectional, unicycles, tricycles and quadricycles [9-13].
Omnidirectional robots have a greater number of degrees of freedom (number of axes in which
a movement is made) than other types of robot, since they move in almost any direction without the need to
re-orient themselves when they are going to rotate. Unicycles have a simple kinematics; their structure is
composed of two fixed wheels aligned on the same axis and a support wheel that allows them to turn on
the turning axis of the platform as shown in Figure 1. Then there are the tricycles and quadricycles that incorporate
three or four wheels (respectively) in their structure trying to keep the center of gravity in the center of the structure
when increasing the robot speed. In this case, the TurtleBot unicycle robot was used, which consists of a modular
chassis to place electronic components, such as: control cards, cameras, sensors or accessories. In addition,
the controller of this robot interprets ROS locally, which allows the user to use a standard language form to control
the robot movements [15].
On the one hand, the robotic operating system or ROS is a standardized work environment for
developing applications for robots and implements some basic services that contemplate: the abstraction of
the hardware, the control systems, the way of communication and the transmission of messages between
the robot and the controller. This environment is based on networks where the processor receives, sends and
multiplexes the messages from the sensors, actuators and states of the platform. Furthermore, it is entirely free
and can be implemented in a UNIX operating system (e.g. Ubuntu) [16, 17].
On the other hand, the TurtleBot generates motion schemes from an absolute reference system, which
indicates that finding a robot position is equivalent to finding a coordinate with translation and rotation components
[ x, y, θ ]. This means that the robot has three degrees of freedom, shown in Figure 2 (a). Also, the position and
orientation of the robot is determined using a technique called odo-metry, which achieves it through a partial time

 ISSN: 1693-6930
TELKOMNIKA Telecommun Comput El Control, Vol. 18, No. 6, December 2020: 3088 - 3095
3090
integration of movements [15-17]. One of the advantages of odo-metry is that it is simple, easily implemented and
requires relatively few computational resources, which allows for an increase in the sampling time to determine the
orientation of the mobile. However, the margin of error increases as the wheels wear or slide, so periodic
calibrations are required. This means that the relative orientation of the robot depends on the speed of the wheels,
which when rotating at the same speed do not generate a variation in position and when increasing the speed in one
of them also does it the tendency towards a direction, as shown in Figure 2 (b) [18].
(a) (b)
Figure 1. TurtleBot mobile robot; (a) font view, and (b) bottom view
(a) (b)
Figure 2. TurtleBot reference frames; (a) reference system, and (b) variablea involved in a turn
More specifically, to determine the robot position, mathematical transformations based on the robot
reference frames are used, where the path of wheel 1 and 2 are represented by (1), the form of the average path
by (2) and the orientation by is constructed by subtracting two coordinates and taking into account the wheel
spacing in (3) [19]. It is worth noting that there are other ways to determine the orientation of a robot, such as;
measurement with a GPS, digital goniometers or gyroscopes. However, the scheme implemented by ROS for
the handling of the TurtleBot is based on the odo-metric technique and thus determines an orientation and
relative position [20]. This information is relevant for the generation of the controller using an optimization
algorithm, which is described in the following section.
∆𝑥𝑖 = 𝑟𝑖∆𝜃 (1)
∆𝑥 =
∑ ∆𝑥 𝑖
2
1
2
(2)
∆𝜃 =
∑ ∆𝑥 𝑖
2
1
𝑎
(3)
2.2. Optimization algorithm
An optimization algorithm is a technique that, when implemented in a software, finds approximate
values to an optimal value. Usually, this value can be the largest or the smallest of a set of values and to find it
the algorithm performs a series of iterations that alter the input values according to an aptitude value. Where,
the proficiency value represents the performance of the variable when modifying the input variables
(individual) of the function [21, 22].
a) Reference system b) Variables involved in a turn

3091
A case of optimization algorithms is genetic algorithms (GA), which iteratively updates a population
of individuals to select the best among them. In a similar way to natural selection, GA implement selection,
crossing and mutation methods, which from an aptitude value assigned to everyone determine whether it
survives when updating the population (generational change) [23]. GAs are very versatile tools and can be
adapted to almost any need, so there are many varieties and classes. In its simplest form is the conventional
GA that has no restrictions on the application of selection, crossing and mutation operations, and a variant of
its computational implementation converts individuals into binary numbers to facilitate crossing and mutation
operations (Algorithm 1) [24-27].
Algorithm 1. Conventional genetic algorithm
i=0
P0=0
While i < condition of termination do
𝑓 ← 𝐸𝑣𝑎𝑙𝑢𝑎𝑡𝑖𝑛𝑔 𝑡ℎ𝑒 𝑎𝑝𝑡𝑖𝑡𝑢𝑑𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛(𝑃0)
𝑃1 ← 𝑆𝑒𝑙𝑒𝑐𝑡𝑖𝑜𝑛(𝑃0, 𝑓)
𝑃1 ← 𝑐𝑟𝑜𝑠𝑠𝑖𝑛𝑔(𝑃1, 𝑓)
𝑃1 ← 𝑚𝑢𝑡𝑎𝑡𝑖𝑜𝑛(𝑃1)
𝑖 + +
Selection operators rank individuals based on their skill value, including roulette and tournament
methods. First, the roulette method assigns everyone a probability value based on his or her skill value, i.e.
a probability distribution is constructed by adding up all the skill values of the individuals and dividing them
by that result. Then the method generates and searches for the individual with the closest probability, to that of
a randomly generated number. Secondly, the tournament method selects four individuals at random and
compares their skill value, among them it selects two and from the two resultants it selects one.
The binary crossing operator takes two individuals from the population, codes them into a binary
number and cuts their resulting binary strings at a random point and recombines them to generate two new
individuals, see Figure 3 (a). Similarly, the mutation operator selects an individual, converts it into a binary
number and changes the value of a bit, see Figure 3 (b). The proposed controller builds motion sets using
a modified GA, in which the number of crossing operators is increased and together with the mutation operator
they are selected with a predefined threshold value. These values and the implementation of the modified GA
are described in detail in the following section.
(a) (b)
Figure 3. Genetic operators; (a) crossing operator, and (b) mutation operator
3. DEVELOPMENT AND IMPLEMENTATION
Initially, the controller generates a set of 20 individuals, which are sets of parameters coded in ROS
language that allow the robot to execute movement routines, where, the suitability value is the distance the
robot travels with the indications given by the individual during 10 seconds, see Figure 4 (a). Each set of
parameters is composed by groups of up to 5 components (generated at random), that can be forward
movements (duration between 0 and 5 s), backward movements (duration between 1 and 5 s) and turning
(between − π and π), see Figure 4 (b).
Like the conventional GA, the selection method incorporated in this version of the GA is tournament
and the selection in each phase is made following a Bernulli distribution, shown in Figure 5. That is, everyone
has a fifty percent (50%) chance of advancing to the next phase. In addition, this method is complemented by
the selection and crossing methods, since thresholds were set for the algorithm to select among the genetic
operators of crossing and mutation.
On the one hand, there are three options for selecting a genetic cross operator. The first option shown
in Figure 6 (a), the combination of characteristics between two randomly selected individuals. The second
shown in Figure 6 (b), the combination of similar parameters (if any) between individuals, e.g. if both
individuals have the turn attribute their turn angles are exchanged. The third shown in Figure 6 (c), operator
simply reverses the components of a selected individual. On the other hand, the genetic mutation operator is
simpler, see Figure 6 (d), since it changes an individual in the population for a new one.

 ISSN: 1693-6930
3092
(a) (b)
Figure 4. General diagram of an individual; (a) composition of an individual and the population, and
(b) view of the parameters to be evaluated and calculation of the suitability value
(a) (b)
Figure 5. Functional diagrams of an individual's assessment method and selection method;
(a) individual assessment scheme, and (b) representation of the selection method
(a) (b)
(c) (d)
Figure 6. Genetic cross and mutation operators; (a) genetic crossing operator 1,
(b) genetic crossing operator 2, (c) genetic crossing operator 3, and (d) genetic crossing operator 4
Genetic crossing and mutation operators are combined with a conventional GA, which incorporates
a fixed threshold for the selection of each individual operator. That is, in each iteration of the modified GA
the execution of an operator depends on a random number that follows a uniform distribution. It should be
noted that when an individual is evaluated, the execution of the modified GA is stopped until the simulator
responds with a value for the distance travelled (Algorithm 2).
Algorithm 2. Modified genetic algorithm
Function Genetic algorithm ()
P0[20][5] =initial population;
Generations=0;
P0←Generate 20 individuals at random;
While Generations<Termination condition do
d←Evaluate (P0)
P1←Tournament (P0, d)

3093
S~U [0,1]
If S< 0.25 then
P1←Crossing operator 1(P1)
Else If S>=0.25 and S<0. 5 then
P1←Crossing operator 2 (P1)
Else If S>=0.5 and S<0. 75 then
P1←Crossing operator 3 (P1)
Else
P1←Mutation Operator (P1)
P0←P1
Generations++
Function Evaluation (P)
T←0
Establishing a connection with the virtual machine
If T<20 then
Encode Individual with ROS tags (P0[T] [1:5])
p_1=read current position ()
Send motion parameters to the simulator (P0[T] [1:5])
Wait 10 seconds
p_2=read current position ()
d[T]=|p_1-p_2 |
T++
Return d
In order to evaluate the performance of the modified GA, an implementation of the conventional GA
was carried out. This implementation was done according to Algorithm 1, where the selection method is
roulette and the cross and mutation operator is presented in Figure 6 (a) and Figure 6 (d) respectively. In both
cases, the test environment is the working space of the gazebo simulator and the main objective of the TurtleBot
is to travel the greatest distance possible, in a flat environment with and without obstacles, see Figure 7. This
workspace is installed in a virtual machine with LINUX operating system, which ran on a computer with
WINDOWS 10 operating system, an Intel inside TM core i3 processor, 8Gbs of RAM memory and a 240Gbs
hard disk.
(a) (b)
Figure 7. Test scenarios; (a) unobstructed environment, and (b) obstacle environment
4. RESULTS AND ANALYSIS
The AGs mentioned in section 2 and section 3 carried out 1000 aptitudevalue evaluations over 50 algorithm
runs. That is, each algorithm had a generational change over 50 generations in both scenarios. In the unobstructed
environment, the information collected allowed the construction of the graphs shown in Figure 8 that show
the standard deviation, the average, the best and the worst with respect to the distance traveled. In the environment
with obstacles, the information collected allowed the construction of the graphs in Figure 9, where the behavior of
the robot when trying to avoid obstacles and travel through its environment during a conventional and modified GA
execution respectively is presented. Considering that 50 GA runs were performed in the obstacle environment,
the best suitability value found by each algorithm is presented in Table 1. Finally, the interpretation of the results
obtained is presented in the following section.
Table 1. Best fit value found in the obstacle environment by both GA
Conventional GA Modified GA
Distance travelled [cm] 43.12 ± 5.76 35.15 ± 6.88

 ISSN: 1693-6930
3094
(a) (b)
Figure 8. Trend of conventional and modified GA in an unhindered environment;
(a) conventional GA (X50), and (b) modified GA (X50)
(a) (b)
Figure 9. Trend of conventional and modified GA in a hindered environment;
(a) conventional GA, and (b) modified GA
5. CONCLUSIONS
The conventional GA is versatile enough to adapt to the needs of each situation, which indicates that
this type of optimization algorithms is not exclusively used for the solution of complex mathematical problems.
As shown in Figure 4 (b) the function to be optimized has only one arithmetic operator and depends on two
values to fix the aptitude value of an individual, which indicates that determining an aptitude value can be
achieved from simple mathematical transformations. In addition, this type of transformation requires a low
computational cost, which gives the user the flexibility to implement it in a conventional desktop computer.
The graphs in Figure 8 show that both GAs find individuals that allow the robot to travel up to 50 cm
in a straight line, which indicates that during the evolution the turning and backward movements disappear.
Although, both GA find relatively good configurations of individuals, in conventional GA it has a lower speed
of convergence, since, the mutation operator makes very strong changes to the population increasing in
convergence time. However, by limiting the effect of the mutation operator on the modified GA, the effect of
the mutation operator is reduced and so is the convergence time. In addition, increasing the number of crossing
operators takes advantage of the characteristics of good individuals in the population.
The graphs in Figure 9 were not made in a similar way to the unobstructed environment, as the robot
makes random movements when it encounters an obstacle, which does not allow a convergence of the
proficiency value. As shown in the lower part of the graph, sometimes the proficiency value is less than
10 cm which means that the robot is trapped or cannot avoid the obstacle for a certain time. However, when
dodging the obstacle, the aptitude value increases rapidly, which leads both GAs to follow that path and thus
the robot travels through much of the environment. Although, the same characteristics of Figure 8 were not
measured, Table 1 was constructed showing the best individual found by both GA, of which the modified GA
found some partially better ones.

3095
REFERENCES
[1] T. Tokunaga, K. Oka and A. Harada, "1segment continuum manipulator for automatic harvesting robot - prototype and
modeling," 2017 IEEE International Conference on Mechatronics and Automation (ICMA), pp. 1655-1659, 2017.
[2] Y. Wan and H. Xu, "On dynamics simulation of 3 DOF manipulator," 2016 35th Chinese Control Conference (CCC),
pp. 6290-6294, 2016.
[3] S. Kucuk and B. D. Gungor, "Inverse kinematics solution of a new hybrid robot manipulator proposed for medical
purposes," 2016 Medical Technologies National Congress (TIPTEKNO), pp. 1-4, 2016.
[4] J. Meng, A. Liu, Y. Yang, Z. Wu and Q. Xu, "Two-Wheeled Robot Platform Based on PID Control," 2018 5th
International Conference on Information Science and Control Engineering (ICISCE), pp. 1011-1014, 2018.
[5] H. Watanabe, "Development of Wafer Transfer Simulator Based on Cellular Automata," IEEE Transactions on
Semiconductor Manufacturing, vol. 28, no. 3, pp. 283-288, Aug. 2015.
[6] Y. Ninomiya, Y. Arita, R. Tanaka, T. Nishida and N. I. Giannoccaro, "Automatic Calibration of Industrial Robot and
3D Sensors using Real-Time Simulator," 2018 International Conference on Information and Communication
Technology Robotics (ICT-ROBOT), pp. 1-4, 2018.
[7] B. Mu, J. Chen, Y. Shi and Y. Chang, "Design and Implementation of Nonuniform Sampling Cooperative Control
on A Group of Two-Wheeled Mobile Robots," IEEE Transactions on Industrial Electronics, vol. 64, no. 6,
pp. 5035-5044, 2017.
[8] O. Calla, S. Mathur and K. L. Gadri, "Possible Landing site for Chandrayaan-2 Rover," 2016 International
Conference on Recent Advances and Innovations in Engineering (ICRAIE), pp. 1-5, 2016.
[9] J. Meng, A. Liu, Y. Yang, Z. Wu and Q. Xu, "Two-Wheeled Robot Platform Based on PID Control," 2018 5th
[10] K. Piemngam, I. Nilkhamhang and P. Bunnun, "Development of Autonomous Mobile Robot Platform with Mecanum
Wheels," 2019 First International Symposium on Instrumentation, Control, Artificial Intelligence, and Robotics
(ICA-SYMP), pp. 90-93, 2019.
[11] D. Cui, X. Gao, W. Guo and H. Dong, "Design and Stability Analysis of a Wheel-Track Robot," 2016 3rd
[12] H. Wang, B. Li, J. Liu, et al., "Dynamic modeling and analysis of Wheel Skid steered Mobile Robots with the different
angular velocities of four wheels," Proceedings of the 30th Chinese Control Conference, pp. 3919-3924, 2011.
[13] Z. Fan, Q. Qiu and Z. Meng, "Implementation of a four-wheel drive agricultural mobile robot for crop/soil
information collection on the open field," 2017 32nd Youth Academic Annual Conference of Chinese Association of
Automation (YAC), pp. 408-412, 2017.
[14] K. Priandana, et al., "Design of A Task-Oriented Autonomous Wheeled- Robot for Search and Rescue," 2018
International Conference on Advanced Computer Science and Information Systems (ICACSIS), pp. 259-263, 2018.
[15] H. Aagela, et al., "An Asus_xtion_probased indoor MAPPING using a Raspberry Pi with Turtlebot robot Turtlebot
robot," 2017 23rd International Conference on Automation and Computing (ICAC), pp. 1-5, 2017.
[16] R. Mishra and A. Javed, "ROS based service robot platform," 2018 4th International Conference on Control,
Automation and Robotics (ICCAR), pp. 55-59, 2018.
[17] A. Koubâa, et al., "Turtlebot at Office: A Service-Oriented Software Architecture for Personal Assistant Robots Using
ROS," 2016 International Conference on Autonomous Robot Systems and Competitions (ICARSC), pp. 270-276, 2016.
[18] M. M. Kassir and M. Palhang, "Novel qualitative visual odometry for a ground: Vehicle based on funnel lane
concept," 2017 10th Iranian Conference on Machine Vision and Image Processing (MVIP), pp. 182-187, 2017.
[19] Q. Lin, X. Liu and Z. Zhang, "Mobile Robot Self-LocalizationUsing Visual Odometry Based on Ceiling Vision,"
2019 IEEE Symposium Series on Computational Intelligence (SSCI), pp. 1435-1439, 2019.
[20] M. Galli, R. Barber, S. Garrido and L. Moreno, "Path planning using Matlab-ROS integration applied to mobile
robots," 2017 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC),
pp. 98-103, 2017.
[21] M. Patrascu and A. Ion, "Self-Adaptation in Genetic Algorithms for Control Engineering: The Case of Time Delay
Systems," 2017 21st International Conference on Control Systems and Computer Science (CSCS), pp. 19-25, 2017.
[22] A. V. Mokshin, V. V. Mokshin and L. Sharnin, “Adaptive genetic algorithms used to analyze behavior of complex
system,” Communications in Nonlinear Science and Numerical Simulation, vol. 71, pp. 174-186, 2019.
[23] A. Hussein, “Improve the Performance of K-means by using Genetic Algorithm for Classification Heart
Attack,” International Journal of Electrical and Computer Engineering, vol. 8, no. 2, pp. 1256-1261, 2018.
[24] E. B. Asmae, B. Benhala, and I. Zorkani, “A genetic algorithm for the optimal design of a multistage
amplifier,” International Journal of Electrical and Computer Engineering, vol. 10, no. 1, pp. 129-138, 2020.
[25] S. Harun and M. F. Ibrahim, “A genetic algorithm-based task scheduling system for logistics service robots,” Bulletin
of Electrical Engineering and Informatics, vol. 8 no. 1, pp. 206-213, 2019.
[26] S. Shan, et al., "An adaptive genetic algorithm for demand-driven and resource-constrained project scheduling in
aircraft assembly," Information Technology and Management, vol. 18, no.1, pp. 41-53, 2017.
[27] M. J. Varnamkhasti, “A genetic algorithm rooted in integer encoding and fuzzy controller,” International Journal of
Robotics and Automation, vol. 8 no. 2, pp. 113-124, 2019.

Design of a controller for wheeled mobile robots based on automatic movement sequencing

Recommended

More Related Content

What's hot (20)

Similar to Design of a controller for wheeled mobile robots based on automatic movement sequencing (20)

More from TELKOMNIKA JOURNAL (20)

Recently uploaded (20)

Design of a controller for wheeled mobile robots based on automatic movement sequencing