Hindawi

Wireless Communications and Mobile Computing

Volume 2020, Article ID 2138643, 12 pages
https://doi.org/10.1155/2020/2138643

Research Article
Simultaneous Localization and Mapping Based on Kalman Filter
and Extended Kalman Filter

Inam Ullah ,1 Xin Su ,1 Xuewu Zhang ,1 and Dongmin Choi 2

1
College of Internet of Things (IoT) Engineering, Hohai University (HHU), Changzhou Campus, 213022, China
2
Division of Undeclared Majors, Chosun University, Gwangju 61452, Republic of Korea

Correspondence should be addressed to Dongmin Choi; [email protected]

Received 28 January 2020; Revised 14 March 2020; Accepted 22 May 2020; Published 8 June 2020

Academic Editor: Yin Zhang

Copyright © 2020 Inam Ullah et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

For more than two decades, the issue of simultaneous localization and mapping (SLAM) has gained more attention from
researchers and remains an influential topic in robotics. Currently, various algorithms of the mobile robot SLAM have been
investigated. However, the probability-based mobile robot SLAM algorithm is often used in the unknown environment. In this
paper, the authors proposed two main algorithms of localization. First is the linear Kalman Filter (KF) SLAM, which consists of
five phases, such as (a) motionless robot with absolute measurement, (b) moving vehicle with absolute measurement, (c)
motionless robot with relative measurement, (d) moving vehicle with relative measurement, and (e) moving vehicle with relative
measurement while the robot location is not detected. The second localization algorithm is the SLAM with the Extended
Kalman Filter (EKF). Finally, the proposed SLAM algorithms are tested by simulations to be efficient and viable. The simulation
results show that the presented SLAM approaches can accurately locate the landmark and mobile robot.

1. Introduction can work cooperatively. The capability to collaborate is

dependent on the robot’s capability to connect and commu-
Wireless sensor networks (WSNs) grasp the potential of var- nicate with each other’s. In this work, the authors consider
ious new applications in the area of management and control. the procedure of simultaneous localization and mapping
Examples of such applications include detection, target track- (SLAM). For this purpose, a linear Kalman Filter (KF) with
ing, habitation monitoring, catastrophe management, and SLAM and Extended Kalman Filter (EKF) with SLAM are
climate management such as temperature and humidity. applied [3, 4].
The key technology that drives the development of sensor SLAM plays a key role in the field of robotics and espe-
applications is the quick growth of digital circuit mixing. In cially in a mobile robot system. The key objective of SLAM
the recent future, these applications will provide a small, is to jointly measure the position of the robot as well as the
cheap, and efficient sensor node. Localization is also crucial model of the surrounding map [5–7]. For the safe interaction
for various applications in WSNs. In both universal comput- of robots within the operation area, this information is
ing and WSNs, there has been considerable consideration of important. A variety of the SLAM algorithm has been pre-
localization [1, 2]. Characteristically, the WSN system offers sented over the last decade. Most of the early algorithms for
the range and/or bearing angle measurements between each SLAM used a laser rangefinder [8] which works as the core
landmark and vehicle. sensor node, and visual sensor nodes are the most used
Mobile robot localization is also one of the attractive option currently, whichever is active or passive [9, 10]. In
researches that support a truly self-governing mobile robot contrast to a laser rangefinder, currently, small, light, and
performance. Various independently working robots can affordable cameras can offer higher determination data and
accomplish tasks more rapidly in many situations. However, virtually unrestricted estimation series. These cameras work
there is a possibility of even better productivity gains if robots as passive sensor nodes and, therefore, do not affect one
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 Wireless Communications and Mobile Computing

another while deploying in similar operation areas. Distinct is usually applicable for the nonlinear functions by approxi-
in the designed light range sensor nodes, cameras are also mating the mobile robot motion model by means of linear
able to apply for both interior and exterior situations. There- functions. A variety of the SLAM algorithms use the EKF
fore, such features can make the camera the best choice for and IF applied by propagating the state error covariance
mobile robotic platforms and SLAM. inverse [17–19]. IF is more advantageous as compared to
In SLAM, the need for using the environment map is the KF. Initially, the information is filtered out by summing
twofold or double [11, 12]. The first one is the map often the vector and matrices of information which resultantly give
essential to support or back up other responsibilities; for a more precise estimate. Next, the IF is steadier than the KF.
example, a map can notify a track arrangement or offer an Lastly, the EKF is comparatively slow while estimating the
initiative imagining for a worker. Secondly, the map or plot maps of having dimensions, because the measurement of
follows in restraining the fault performed in measuring the every vehicle normally affects the Gaussian parameters. Con-
state of the robot. While without a map, the dead reckoning sequently, the updates need prohibitive times when faced
would rapidly point energetically. On the other hand, by with a situation having several landmarks.
using a map, for example, a set of distinct landmarks, the In recent years, the SLAM and autonomous mobile robot
robot can reorganize its localization error by reentering the combinations play an important role in the controlling disas-
known areas. Therefore, SLAM applications are more useful ter field. Particularly, the autonomous robots are widely used
in such situations in which a preceding plan is not existing for the maintenance and rescue operations in the disaster
and require to be constructed. In some aspects of the robots, controlling such as radioactivity leaks. Since the area is
a set of landmark location is known prior. For example, a unreachable, simultaneous mapping of the environment
robot is operational on the floor of a workshop that can be and the robot localization is crucial to determine the exact
supplied with a physically assembled chart of artificial source spot [20–23]. Therefore, SLAM has been an important
guidelines in the operation area. Alternatively, in another issue as the localization degree hangs on active mapping.
case, in which the robot has admittance to the global posi- However, the SLAM implementation by using the EKF is
tioning system (GPS), the GPS satellite can be chosen as a pretty exciting because of the approximation of the sensor
moving beacon at a prior known position. In this case, the noises and real-time stochastic system as Gaussian. There-
SLAM may not be needed if the localization is done consis- fore, inappropriate alteration of the noise covariance may
tently concerning the prior known landmark of the robot. result in filter divergence over time, resulting in the complete
Through the development of indoor localization uses of system becoming unstable. The researchers presented some
mobile robots, the popularity of SLAM is increased. Most alternate methods that are moderately straightforward but
of the indoor procedures rule out the practice of GPS to severe computationally which have the benefit to accommo-
assure the error of localization. The SLAM algorithm also date the noise model other than the Gaussian such as UKF,
provides an interesting substitute to the maps which is built FastSLAM, and Monte Carlo localization [24–26].
by the user, which represents that the process of the robot is
also conceivable in the nonappearance of ad hoc networks 1.1. Contributions. In the above paragraphs, the authors
for localization [13]. investigated the SLAM with KF and EKF. The performance
KF derivatives are concerned with the first branch of of such models under localization is not yet well-thought-
those methods which apply a filter [14, 15]. The KFs assume out. Therefore, in this work, the authors analyzed SLAM by
that Gaussian noises affect data, which is not inevitably accu- suing linear KF and EKF. The basic contribution of this work
rate in our case. KFs are planned to solve the problems of lin- included one dimensional (1D) SLAM using a linear KF (a)
ear systems in their basic form and are rarely used for SLAM, motionless robot with absolute measurement, (b) moving
although they have great convergence properties. On the vehicle with absolute measurement, (c) motionless robot
other hand, in the nonlinear filtering systems such as in with relative measurement, (d) moving vehicle with relative
SLAM, the EKF is a common tool. EKF introduces a step of measurement, and (e) moving vehicle with relative measure-
linearization for the nonlinear systems, and a first-order Tay- ment while the robot location is not detected. Furthermore,
lor expansion performs linearization around the current esti- the authors analyzed the localization performance of SLAM
mate. The optimality of EKF’s is shown as long as with EKF. The proposed SLAM-based algorithms are evalu-
linearization is performed around the state vector’s exact ated and compared with each other and also with other algo-
value. It is the value to estimate in practice and is therefore rithms regarding SLAM. More precisely, the proposed SLAM
not usable, and this can lead to problems of accuracy. None- algorithms present good accuracy while maintaining a sensi-
theless, estimates are close enough to the reality, for the most ble computational complication.
part, to allow the EKF to be used.
KF is Bayes filters which signify posteriors by using the 1.2. Organization of the Paper. The structure of this paper is
Gaussians [16], for example, the distributions of unimodal as follows: Section 2 demonstrates the work related to SLAM
multivariate that can be denoted efficiently by a minor sum and Section 3 demonstrates the proposed SLAM algorithms.
of parameters. The KF SLAM is based on the hypothesis that The subsections of Section 3 are SLAM with KF and SLAM
the transformation and estimation functions are linear with with EKF, respectively. Section 4 demonstrates the compari-
the introduction of Gaussian noise. In state-of-the-art SLAM, son of the proposed and other algorithms. Finally, Section 5
KF has two main variations. The first one is the EKF, and the demonstrates the conclusion and future direction of the pro-
second one is the information filtering (IF) or EIF. The EKF posed algorithms.
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Wireless Communications and Mobile Computing 3

2. Related Work schemes comprise the use of high-priced laser sensor nodes
and comparatively innovative and inexpensive RGB-D cam-
Before presenting the proposed SLAM algorithms, it would eras. These sensors are too costly for some applications, and
be better to present some background knowledge and related RGB-D cameras consume much power, CPU, or communi-
work on SLAM algorithms. In this section, the authors pres- cation specifications for on-board or PC processing of data.
ent a detailed description of the SLAM that forms the basis of Thus, the authors tried to model an uncertain setting using
the proposed SLAM algorithms. Compared to the current a low-cost device, EKF, and dimensional features such as
solutions, many people still do not have highly accurate walls and furniture.
instruments; they still have challenging piloting capabilities Usually, the typical filter uses the scheme model and for-
and can solve the SLAM problem. Their mapping, therefore, mer stochastic info to approximate the subsequent robot
depends on the toughness policy of acting as a replacement state. Though in the real-time condition, the sound statistics
for the accurate world definition. With linear KF, this possessions are comparatively unidentified, and the system is
approach is a new research concept for SLAM. imprecisely demonstrated. Therefore, the filter deviation
Regarding the SLAM, readers may not be familiar with might arise in the incorporation scheme. Furthermore, the
the origin and its derivation may refer to the standard and predictable precision might be stimulating to be grasped
current work on SLAM [27, 28]. For the SLAM problem, due to the nonappearance of the receptive time-varying of
the first method was introduced between 1986 and 1991. mutually the process and measurement noise statistic. So,
Smith and Chesseman [29] published a paper in 1986 for the outdated approach desires to be upgraded pointing to
the solution of SLAM problems. They present the EKF to deliver an aptitude to guesstimate those belongings. To solve
solve this problem. In that paper, they established a numeri- this problem, the new adaptive filter is proposed in [38]
cal basis for explaining the relation between landmarks and named as an adaptive smooth variable structure filter
operating the geometric uncertainty. (ASVSF). The upgraded SVSF is consequential and executed;
Several other researchers have worked on various SLAM the process and measurement noise statistics are appraised
issues. For example, in [30–32], the authors presented a new by using the maximum a posteriori creation and the weighted
architecture that applies one monocular SLAM system for exponent concept. The authors applied ASVSF to overwhelm
the tracking of unconstraint motion of the mobile robot. the SLAM problem of a self-directed mobile robot; hereafter,
The improved oriented FAST and rotated BRIEF (ORB) it is shortened as an ASVSF-SLAM algorithm.
characteristics show the landmarks to design a network fea- In [39], the authors presented a 3D cooperative SLAM for
ture procedure of detection. An enhanced matching feature a joint air grounded robotic system which is intended to
system has enhanced function matching strength. The succeed an indoor quadrotor flying done composed with a
updated EKF measures the free-moving visual sensor’s mul- Mecanum-wheeled omnidirectional robot (MWOR) in
tiple dimensional states rather than the standard EKF. Fur- indoor unidentified and no GPS environments. Moreover,
thermore, in [33, 34], the authors address the issue of the an Oriented Fast and Rotated BRIEF- (ORB-) SLAM 2.0
applications of SLAM for navigation problems. For the solu- method is applied to yield a 3D chart and determine concur-
tion of high-accuracy problems, an EKF or particle filter (PF) rently the location of the indoor quadrotor, and a particle-
algorithm [35] is frequently applied to the processing of data. filter SLAM (FastSLAM 2.0) method is applied to shape the
The PF algorithm, which is often applied for the G-mapping 2D chart of the universal atmosphere for the MWOR. An
SLAM technique, is well-matched for the nonlinear system’s additional accurate 3D quadrotor location estimation tech-
investigation. Though, PF computational dimensions are nique for the quadrotor is planned with the help of the
larger than those of EKF. Therefore, EKF and PF also have MWOR. A cooperative SLAM applying fuzzy Kalman filter-
some disadvantages in the process of navigation. The Gauss- ing is presented to fuse the productions of the ORB-SLAM
ian smoothing filter and its modification are used which is 2.0, FastSLAM 2.0, and quadrotor location estimation
based on the distributed computing scheme. This algorithm methods, in order to localize the quadrotor further precisely.
is meaningfully better to the EKF and PF algorithms regard- Mutually, SLAM methods, quadrotor position estimation
ing the computational speed. method, and cooperative SLAM have been executed in the
For the reduction of the linearization error of KF algo- robotic operation system atmosphere.
rithms, the authors presented three techniques and their via- Recent work on SLAM [40] attempted to address the
bility and efficiency are assessed by SLAM [36]. In the issue of SLAM landmarks [41]. The authors presented an
derivative-based approaches of the KF system, the lineariza- AUV vision-based SLAM, in which the submerged nonnat-
tion error is undetectable owing to the practice of the Taylor ural landmarks are utilized for visual sensing of onward and
expansion for the linearization of the nonlinear motion pro- down cameras. The camera can also estimate the AUV loca-
cess. The presented three techniques reduce the error of lin- tion data, along with several navigation sensor nodes such as
earization by substituting the Jacobian observation matrix depth sensor node, Doppler velocity log (DVL), and an iner-
with new formulations. Similarly, in [37], a SLAM with lim- tial measurement unit (IMU). The landmark detection algo-
ited sensing by applying EKF is proposed. The robot’s prob- rithm is organized in a framework of conventional EKF
lem with creating a map of an unidentified atmosphere while SLAM to measure the landmark and robot status. Further-
adjusting its particular location which is the basis on a similar more, partial observability of mobile robot based on EKF
map and sensor information is called SLAM. Because sensor is explored in [42, 43] to find the answer that can avoid
accuracy plays a major part in this issue, most of the planned erroneous measurements. When considering only certain
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 Wireless Communications and Mobile Computing

environmental landmarks, the computational costs of Table 1: Index of notation.

mobile robots can be minimized, but with an increase in
device uncertainties. The fuzzy logic methodology is pre- Notation Description
sented to guarantee that the calculation has attained the Xk Current state
desired output even though some of the landmarks have X k−1 Previous state
been omitted for reference purposes. For the measurement
Qk Covariance matrix of prediction
invention of KF, fuzzy logic is used to exact the location of
the mobile robots and any sensed landmarks all throughout Rk Covariance matrix of observation
the process of observations. X k+1 Estimated state vector
Researchers have proposed several algorithms for SLAM; Pk+1/k Covariance matrix for prediction
some of which are already discussed in the above pages. Most
Q Process noise matrix
of them focused on the landmark’s estimation, performance,
accuracy, and effectiveness of the SLAM algorithm. However, R Measurement noise matrix
there are still some important and fundamental issues that wk ∼ N ð0, Qk Þ Process noise
need to be addressed, such as an optimal solution for SLAM,
vk ∼ N ð0, Rk Þ Observation noise
active SLAM for SLAM development, SLAM failure detec-
tion, SLAM front end robust algorithm, and SLAM algorithm K +1 Time instant
that considers various aspects at once. Therefore, in this Z k+1 Estimated measurement vector
paper, the authors attempted to propose a modified SLAM ∇F x , ∇F u Jacobian matrices of the function f
algorithm by applying KF and EKF. The authors presented
B, F State transition matrix
SLAM algorithms that consider several aspects of the SLAM
such as velocity, distance, coverage area, maximum range, U Input of the process noise
and localization time. The notations used in this work are W, V Vectors of state measurement noise
listed in Table 1. v Mobile robot velocity
3. Proposed Simultaneous Localization and p Landmark position
Mapping Algorithms JF Jacobian of state equation
K k+1 Kalman gain
This section presents the proposed SLAM algorithms based X k+1/k+1 Updated estimate
on KF and EKF. The proposed algorithms are analyzed and
evaluated in the next subsections. Pk+1/k+1 Updated covariance
H Measurement Jacobian or linearization matrix
3.1. Simultaneous Localization and Mapping with Kalman
dt Global time
Filter. In the following section, the authors presented the
theory of SLAM which results in efficient localization and T Time
mapping in WSNs. Specifically, the author presents the t Initial time
analysis of the operating environment and finally discussed LM Landmark
the proposed algorithm and compared it with other SLAM Pk New state covariance matrix
algorithms. In the existence of Gaussian white noise, the KF
provides a well-designed and statically optimum explana-
tion for the linear systems. It is a technique that uses linear tions that govern the dynamics and observation from the
estimation associated with the states and error covariance method.
matrixes for the purpose to produce gain stated to as the
Kalman gain. Such benefit is added to the estimation of a X k+1/k = F k × X k/k , ð3Þ
preceding condition, thereby generating an estimate of a
posteriori [44]. The below equations define the dynamic
model of the system and the measuring model used for Pk+1/k = F k × Pk/k × F Tk + Ak × Qk × ATk , ð4Þ
the linear state approximation in general which consists of  Ã−1
two f and h functions. K k+1 = Pk+1/k × H Tk H k+1 × Pk+1/k × H Tk+1 + Rk , ð5Þ

Xk + 1 = f ðXk, Uk, WkÞ, ð1Þ X k+1/k+1 = X k+1/k + K k+1 ½Z k+1 − H k+1 × X k+1/k Š, ð6Þ

Pk+1/k+1 = ½I − K k+1 × H k+1 Š × Pk+1/k : ð7Þ

Zk + 1 = hðXk, VkÞ, ð2Þ
Such equations from the KF-based method are used iter-
which administrate state proliferation and state measure- atively in conjunction with Equations (1) and (2). Equation
ments, where U is the input of the process, W and V are (3) generalizes the prior state estimate, and Equation (4)
the vectors of state and measurement noise, while k repre- represents the equivalent state covariance error. The gain
sents the discrete-time. In the above equations, f and h of Kalman can be estimated by Equation (5) which is
are typically based on a set of discretized difference equa- applied to update the state approximation and covariance
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Wireless Communications and Mobile Computing 5

error, defined by Equations (6) and (7), correspondingly. SLAM with kalman filter
80
EKF is practically comparable to the iterative KF method,
and sometimes, it is used for the nonlinear systems. By 70
applying the Jacobian, which is a first-order partial deriva-
tive, the measurement H k+1 and nonlinear system F k matri- 60
ces are linearized.
A one-dimensional SLAM with KF is applied for a 50
motionless robot, and the measurement is considered an

y (m)
40
absolute measurement. A 1-DoF mobile robot is used which
is motionless in a fixed position of a straight line. The mobile 30
robot is used for detecting the motionless/stationary land-
marks. The mobile robot velocity v and position p of the 20
landmarks are calculated by applying SLAM with linear KF.
10
Ten numbers of landmark positions are considered. For the
real trajectory, the robot is motionless at a given position
which is v = 0 m/s. The landmark distance is relative to the 100 200 300 400 500 600 700 800 900 1000 1100 1200
mobile robot’s location/position which had a moderate mea- x (m)
surement noise as shown in Figure 1. The state vector is the
Figure 1: SLAM with motionless robot and absolute measurement
diagonal of those that correspond to the robot’s present state
while having a moderate measurement noise.
by projecting the next one. The vector used for the control is
null; it shows that there are no exterior inputs to vary the
SLAM with kalman filter
mobile robot’s state; i.e, the velocity and position of the robot 80
are constant. The initial matrix of covariance is not prevalent;
it is characterized by a broad diagonal ambiguity in both the 70
robot’s landmark location and state and equal ambiguity/un-
60
certainty. The result of mobile robot localization with abso-
lute measurement is shown in Figure 2. 50
Next, a one-dimensional SLAM with KF is applied for a
y (m)

moving vehicle and the measurement is considered an abso- 40

lute measurement. A 1-DoF mobile robot is traveling on a
straight path. The mobile robot is sensing the motionless/sta- 30
tionary landmarks. The robot velocity v and the landmark 20
position/velocity p are calculated by applying SLAM using a
linear KF, and in this case, all the measurements are absolute, 10
see Figure 3. The landmark positions are the same as the pre-
vious one.
100 200 300 400 500 600 700 800 900 1000 1100 1200
Furthermore, a one-dimensional SLAM with KF is
x (m)
applied for a motionless robot, and the measurement is con-
sidered a relative measurement. Here, a 1-DoF mobile robot Figure 2: SLAM with motionless robot and absolute measurement.
is used in a motionless and fixed position of a straight lane
that detects the motionless/stationary landmarks. The robot
SLAM with kalman filter
velocity and the position/location landmarks are calculated 80
by using the SLAM with a KF, see Figure 4. The fourth one
is a one-dimensional SLAM with linear KF. In this case, a 70
moving vehicle is considered with a relative measurement
60
and a 1-DoF robot is traveling on a straight line that detects
the motionless/stationary landmarks. The robot position/lo- 50
cation, velocity, and landmark position are calculated
y (m)

through SLAM with linear KF. Here, all the measures are 40
comparative to the position/location of the mobile robot,
30
see Figure 5.
The last one is almost different from the previous four 20
SLAM algorithms. In this case, a one-dimensional SLAM
with linear KF is considered and the vehicle is moving with 10
a relative/comparative motion. The position/location of the
mobile robot is not observed in this case. A mobile robot is 100 200 300 400 500 600 700 800 900 1000 1100 1200
traveling on a straight line that detects the landmarks which x (m)
are motionless as shown in Figure 6. The velocity of the robot
and its landmark are calculated by applying SLAM with Figure 3: SLAM with moving vehicle and absolute measurement.
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 Wireless Communications and Mobile Computing

SLAM with kalman filter linear KF. As mentioned before, the position is not observed
80
and all the measurements are relative/comparative to the
70 mobile robot position/location.

60 4. Simulation Results and Discussion

50 The proposed SLAM algorithm is evaluated by simulation. In
this simulation, the author evaluates the SLAM algorithm by
y (m)

40
conducting a different experiment with different landmarks.
30 The simulation is divided into five steps, such as a motionless
robot with absolute measurement, a moving vehicle with
20 absolute measurement, a motionless robot with relative mea-
surement, a moving vehicle with relative measurement, and a
10
moving vehicle with relative measurement while the robot
location is not detected. The number of time-stamps is
100 200 300 400 500 600 700 800 900 1000 1100 1200 1200 with the map of dimension [180]. The landmark posi-
x (m) tions are set to be LM = 10 which are denoted by p. For the
real trajectory, the velocity and position are v = 1:0 m/s and
Figure 4: SLAM with motionless robot and relative measurement.
p = 40, respectively, at state xðkÞ = xðk − 1Þ and vðkÞ = vðk −
1Þ, i.e., motionless at a given position having a moderate
SLAM with kalman filter
80 measurement noise as shown in Figure 1. The process pn
and measurement noise r is added, and the landmark dis-
70 tance is relative to the robot position, see Figure 2. The state
equation is a diagonal of those, which ensures that the next
60
state’s estimate or prediction is equal to the present state.
50 The control vector is null; it shows that there are no exterior
inputs that vary the state of the robot because, as we stated
y (m)

40 earlier, the velocity v and position p are constant. Also, the

primary covariance matrix is well-defined by a higher diago-
30 nal uncertainty mutually in the position of the landmark and
20
the robot state and by a comparable uncertainty, which
means that none prevails over the other.
10 The process noise matrix represented by Q and the mea-
surement noise matrix represented by R are computed in
100 200 300 400 500 600 700 800 900 1000 1100 1200
which the landmarks are motionless. For the next state pre-
diction, the measurement is done at the prediction position,
x (m)
and for observation, it is measured at the right position/loca-
Figure 5: SLAM with moving vehicle and relative measurement. tion xðk + 1Þ, vðk + 1Þ, and p. The landmark positions are
similar for all five methods. However, for this case, a vehicle
is considered with constant velocity v = 0:02 m/s and the
SLAM with kalman filter
80 position are p = 30. Also, in this case, the landmark distance
is absolute. In the third case, the robot is motionless and
70 the measurement is relative at a given velocity and position
v = 1:0 m/s and p = 40, respectively. The fourth one is the
60
SLAM with linear KF in which the vehicle is moving and
50 the measurement is relative. The landmark distance is rela-
tive to robot position and a vehicle with a constant velocity
y (m)

40 of v = 0:02 m/s and at the position, see Figure 5, the red line
denotes the position. The last one is the SLAM with linear
30
KF and a vehicle is moving, and the measurement is relative.
20 Note, in this case, the position is not observed as the previous.
The constant velocity of the vehicle is set to be v = −0:2 m/s
10 and the position is 20, as can be seen in Figure 6. The above-
mentioned algorithms for SLAM with KF are evaluated in
100 200 300 400 500 600 700 800 900 1000 1100 1200
deep detail. The authors considered a variety of aspects
x (m)
regarding the SLAM localization. The offered SLAM algo-
rithms present a high level of accuracy in various conditions
Figure 6: SLAM with moving vehicle and relative measurement and perform well in terms of velocity, distance, coverage
while the position of the robot is not observed. area, etc.
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Wireless Communications and Mobile Computing 7

2 3
4.1. Simultaneous Localization and Mapping with Extended 0 0 −dt × uð1Þ × sin ðxð3ÞÞ
Kalman Filter. EKF is well-known as a widespread resolution 6 7
to the SLAM problem for mobile robot localization. In this JF = 6
40 0 dt × uð1Þ × cos ðxð3ÞÞ 7
5, ð14Þ
section, the authors realized the EKF SLAM-based algorithm 0 0 0
for a mobile robot that follows a specific trajectory. EKF
SLAM relies on present elements of the navigation system
and the global initialization Jacobian G can be written as
known as landmarks to change the location of the robot.
follows:
EKF SLAM for a mobile robot is executed in a defined field
with a specific feature. The authors considered two basic " pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi #
mathematical models such as the EKF state and observation − δð1Þ − δð2Þ 0 δð1Þ δð2Þ
model that are represented below. G=
δð2Þ −δð1Þ −1 −δð2Þ δð1Þ

X k+1 = f ðX k , U k , wk Þ, ð8Þ ð15Þ

In the observation and update phase, the observation

Z k+1 = hðX k+1 , υk+1 Þ, ð9Þ model Zk+1 at X k+1/k can be represented as

where wk ∼ Nð0, Ok Þ and vk ∼ Nð0, RkÞ which characterize Zk+1 = hðXk+1/k Þ + Hk+1 × ðXk+1/k − Xk+1 Þ: ð16Þ
the process and observation noise. Here, X k+1 denotes the
estimated state vector at time k + 1. The time is the discrete
time for a known input U k assuming all noise to be wk . In To apply the KF update cycle, i.e., Xk+1/k and Pk+1/k , the
Equation (9), Z k+1 represents the estimated measuring vector KF gain can be computed.
at the time instant k + 1, where vk is the observation noise. Qk
 Ã−1
and Rk denote the covariance matrix of prediction and obser- K k+1 = Pk+1/k × H Tk+1 × H k+1 × Pk+1/k × H Tk+1 + Rk , ð17Þ
vation, respectively. EKF offers an approximation of the opti-
mal state estimate. The EKF-SLAM objectives are to estimate
where K k+1 is the Kalman gain. With measurement of Z k+1 ,
recursively the landmark state X k as stated by the Z k+1 mea-
the updated estimate can be
surement. EKF is basically divided into several steps which
are represented as at the initial state, the state vector X k+1 will
become K k+1/k+1 = X k+1/k + Kk+1 × ½Z k+1 − Hk+1 × Xk+1/k Š: ð18Þ

  If the Z k of measurement is available, EKF calculates the

X k+1 = f ðX k , U k+1 Þ+∇F x × X k − X k : ð10Þ matrix of Kalman gain and integrates the invention of mea-
k

surement to obtain the approximate state X k , accompanied

In the prediction stage, the covariance matrix for predic- by the update of the state error matrix. Therefore, the mea-
tion Pk+1/k can be represented as surement updated step from the above equation will become

Pk+1/k = ∇F x × Pk/k × ∇F Tx +∇F u × Qk × ∇F Tu : ð11Þ X k = X k + K k × ½Z k − hðX k ފ, ð19Þ

In the above equation, the ∇F x and ∇F u denote the P k = ½I − K k × H k Š × P k : ð20Þ

Jacobian matrices of the function f concerning the state
vector which is X k+1 . The state transition matrix is denoted Therefore, the update covariance Pk+1/k+ 1 can be repre-
by B, and F is the state equation which can be represented sented as
as follows:
Pk+1/k+1 = ½I − K k+1 × H k+1 Š × Pk+1/k , ð21Þ
2 3
dt × cos ðxð3ÞÞ 0
6 7 where K k is the Kalman gain and Pk is the new state covari-
B=6
4 dt × cos ðxð3ÞÞ 0 5
7 ð12Þ ance matrix. H is the measurement Jacobian or linearization
0 dt matrix and X k denotes the state vector X k estimate.
After evaluating EKF in deep detail, the authors conclude
2 3 that the EKF also has some disadvantages that is if the pro-
1 0 0 cess and measurement noise are not accurately displayed,
6 7 the robot will diverge from its route which resultantly give
F=6
40 1 07
5 ð13Þ
a contradiction. Particularly, in the case of the robot velocity,
0 0 1 the robot is sensitive to the velocity as by varying the velocity
the robot is diverging from its route as shown in Figure 7 .
Therefore, the Jacobian of the state equation will However, in the first case, the velocity is v = 1:0 m/s as shown
become in Figure 8.
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 Wireless Communications and Mobile Computing

EKF-SLAM EKF-SLAM
20
20

15
15
y (m)

y (m)
10 10

5
5

0
0
–10 –5 0 5 10 15 –5 0 5 10 15
x (m) x (m)
Dead reckoning True LM Dead reckoning True LM
Ground truth Predicted LM Ground truth Predicted LM
EKF-SLAM EKF-SLAM
(a) (b)
EKF-SLAM EKF-SLAM
20
25

20 15

15
y (m)

y (m)

10

10
5
5

0
0
–10 –5 0 5 10 15 20 –5 0 5 10 15
x (m) x (m)
Dead reckoning True LM Dead reckoning True LM
Ground truth Predicted LM Ground truth Predicted LM
EKF-SLAM EKF-SLAM
(c) (d)

Figure 7: The SLAM EKF performance for various v values: (a) v = 0:9 m/s, (b) ‘ , (c) v = 0:7 m/s, and (d) v = 0:6 m/s, where the green line
represents the dead reckoning calculation of the existing location, which is the outcome of the displacement function applied for the
previous location. The black mark is the ground truth information that is necessary for real-time localization. The red mark is the EKF-
SLAM. The blue asterisks represent the true landmarks, and the black circle is the estimated landmarks.

5. Simulation Results and Discussion mark coordinates are [xy], i.e., LM = ½0 15 ; 10 0 ; 15 20Š:
The maximum range is set to be 20 at the initial stage and
The proposed SLAM EKF algorithm is evaluated through parameter α = 1. For the input parameters, the time is set
simulation. In this simulation, the author evaluates the to be T = 10 sec, the velocity is v = 1:0 m/s, and yaw rate =
SLAM EKF algorithm by performing simulation with various 5 deg/sec. At the initial stage, the velocity is limited to v =
factors. Firstly, the time is t = 0, end time is t = 60 sec, while 1:0 m/s as can be seen in Figure 8; however, in the next
the global time is dt = 0:1 sec: In this simulation, the state stage, the velocity is varying.
vector is considered xEst = ½0 0 0ŠT in which the xTrue = x In the case of varying the velocities as can be seen in
Est, while at the dead reckoning state xd = xTrue. The land- Figure 7, the velocities are set to be v = 0:9 m/s, v = 0:8 m/s,
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Wireless Communications and Mobile Computing 9

EKF-SLAM

20
20

15
15
10
y (m)

y (m)
10 5

0
5
–5

0 –10
–10 –5 0 5 10 15
x (m) –20 –20 0 10 20
Dead reckoning True LM x (m)
Ground truth Predicted LM
Figure 9: Performance of SLAM with Extended Kalman Filter in
EKF-SLAM case of higher range.
Figure 8: Performance of SLAM with Extended Kalman Filter.
The technique is applied that the adaptive neurofuzzy EKF
provides development in performance effectiveness.
v = 0:7 m/s, and v = 0:6 m/s. By varying the velocity of the An EKF-based SLAM system for a mobile robot with sen-
robot, the robot is diverging from its route and, therefore, sor bias estimation is presented in [46]. The authors pro-
reduces the coverage area as can be seen in Figure 7(a)- posed an improved method for EKF which is practical to
7(d). Also, the error between the true landmark and pre- the issue of mobile robot SLAM which has taken into consid-
dicted landmark is increasing. On the other hand, for higher eration the sensor bias issue. Mobile robot Pioneer 3-AT is
velocities (more than v = 1:0 m/s), the proposed algorithm is taken as the model for studying the theoretical derivation
not applicable, because in the SLAM, the robot is following and the authentication of the investigation in this work. At
the prior defined map and the robot keeps communication first, the kinematic model of Pioneer 3-AT mobile robot is
with the surrounding. However, in our previous study, we introduced; then, the improved EKF method, taking into
mentioned the higher velocities for the robot, in the case of account the issue of bias estimation and compensation, is
EKF, UKF, and PF, the coverage area, and localization were anticipated to increase the precision of the location estima-
increasing by increasing the velocity. Furthermore, the max- tion. In addition, a study explores the autonomous location
imum range was set to be 20 as shown in Figure 6, but by and atmosphere mapping of stirring substances under the
modifying the maximum range to 30 or above, in this case dust and low lighting situations in underground underpasses.
also, the robot diverges from its route of localization as The typical EKF algorithm has a problem that machine noise
shown in Figure 9. Resultantly, the authors conclude that and the prior statistical characteristics of the observed noise
the proposed algorithm is more suitable for constant velocity cannot be predicted accurately. Thus, the authors presented
which presents a high level of accuracy. an enhanced EKF algorithm to accomplish a fuzzy adaptive
SLAM [45, 47, 48]. Therefore, to predict the position, a laser
6. Comparison of the Proposed and matching is applied to the EKF prediction process, and the
Other Algorithms weighted average location is used as the final location of the
predicted component. The machine noise and the weighted
In the above sections, the authors investigated and evaluated value of experiential noise become fuzzily recognizable by
well about the proposed SLAM algorithms. However, to observing the variation of mean value and covariance. The
demonstrate the effectiveness and better performance of the improved filtering algorithm is applied to a SLAM simulation
planned algorithms, the authors present a brief comparison study and measure the impact on position estimation of four
of the proposed algorithms with other algorithms in this dissimilar landmark measurements.
section. A recent approach strong tracking second-order (STSO)
In [45], the authors presented a neurofuzzy-based adap- central difference SLAM is presented in [49] which it is based
tive EKF method. The purpose of this method is to estimate on the tracking second-order central difference KF. The pro-
the right value of matrix R at every stage. They plan an adap- posed procedure gathers the second-order central differential
tive neurofuzzy EKF to lessen the variance among the theo- filter (SOCDF), strong tracking filter (STF), and PF. Using
retical and actual covariance matrices. The parameters for Cholesky decomposition, the algorithm uses the Sterling
this technique are then skilled offline by using a particle Interpolation second-order method to solve a nonlinear sys-
swarm optimization method. A mobile robot steering with tem problem. This methodology transmits directly in the
a number of landmarks under two situations is assessed. probabilistic estimation of SLAM by adding the covariance
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 Wireless Communications and Mobile Computing

square root factor. Furthermore, in [50], a visual-inertial the findings of this study are currently under embargo while
SLAM feedback mechanism is presented for the real-time the research findings are commercialized. Requests for data,
motion assessment of the SLAM map. The main aspect of based on the approval of patents after project closure, will
this mechanism is that the front-end and the back-end can be considered by the corresponding author.
support each other in the VISLAM. The output from the
back-end is fed to the KF-based front-end to decrease the
Conflicts of Interest
motion estimation error produced by the linearization of
the KF estimator. On the other hand, this more accurate The authors declare that they have no conflicts of interest.
front-end motion estimation will improve back-end optimi-
zation as it provides the back-end with an exact primary
state. Similarly, the EKF-based SLAM approaches are pre- Acknowledgments
sented in [33, 51, 52] which focus on the performance and
This research is supported by the National Key Research and
effectiveness of the SLAM. Each algorithm presents well in
Development Program under Grant 2018YFC0407101 and
its domain, but the proposed SLAM algorithms perform well
in part by the National Natural Science Foundation of China
compared to the other SLAM algorithms.
under Grant 61801166. It was also supported by the Funda-
mental Research Funds for the Central Universities under
7. Conclusion and Future Directions Grant 2019B22214 and in part by the Basic Science Research
Program through the National Research Foundation of
In this work, the SLAM algorithm is proposed in two differ-
Korea (NRF) funded by the Ministry of Education under
ent methods such as SLAM with linear KF and SLAM with
Grant NRF-2018R1D1A1B07043331.
EKF. Firstly, SLAM with linear KF is implemented in five dif-
ferent methods such as the motionless robot with absolute
measurement, moving vehicle with absolute measurement, References
a motionless robot with relative measurement, moving vehi-
cle with relative measurement, and moving vehicle with rela- [1] I. Ullah, Y. Liu, X. Su, and P. Kim, “Efficient and accurate tar-
get localization in underwater environment,” IEEE Access,
tive measurement while the robot location is not detected.
vol. 7, pp. 101415–101426, 2019.
The landmark position was set to be 10 for all five cases.
[2] K. Sha, T. A. Yang, W. Wei, and S. Davari, A survey of edge
The mobile robot position or velocity and landmark position
computing-based designs for iot security, Digital Communica-
are calculated by applying SLAM using a linear KF. Secondly, tions and Networks, 2019.
the SLAM with EKF is implemented and an analytical
[3] G. Zand, M. Taherkhani, and R. Safabakhsh, “A novel frame-
expression for the EKF-based SLAM algorithm is derived work for simultaneous localization and mapping,” in 2015
and their presentation is evaluated. The SLAM algorithm Signal Processing and Intelligent Systems Conference (SPIS),
with EKF is evaluated in various scenarios, and several itera- pp. 109–113, Tehran, Iran, December 2015.
tions are applied to explain the performance of EKF-based [4] X. Ma, R. Wang, Y. Zhang, C. Jiang, and H. Abbas, “A name
SLAM well. The proposed algorithm is simulated for varying disambiguation module for intelligent robotic consultant in
velocities, and their performance is presented in Figure 8. industrial internet of things,” Mechanical Systems and Signal
Each process of localization is effective in its domain. In this Processing, vol. 136, article 106413, 2020.
analysis, many localization factors such as velocity, coverage [5] T. Pire, T. Fischer, J. Civera, P. De Cristóforis, and J. J. Berlles,
area, localization time, and cross section area are taken into “Stereo parallel tracking and mapping for robot localization,”
consideration. The proposed SLAM-based algorithm perfor- in 2015 IEEE/RSJ International Conference on Intelligent
mance is intensively assessed by executing numerous itera- Robots and Systems (IROS), pp. 1373–1378, Hamburg,
tions as can be seen in the figures above. The planned Germany, October 2015.
SLAM-based algorithms present a high precision while pre- [6] T. A. Johansen and E. Brekke, “Globally exponentially stable
serving realistic computational complexity. The simulation Kalman filtering for slam with ahrs,” in 2016 19th Interna-
outcomes indicate that the planned SLAM algorithms can tional Conference on Information Fusion (FUSION), pp. 909–
916, Heidelberg, Germany, July 2016.
accurately locate the landmark and mobile robot.
Future research will use more simulation and tests to [7] S. Safavat, N. N. Sapavath, and D. B. Rawat, “Recent advances
in mobile edge computing and content caching,” Digital Com-
show the robustness of the SLAM in different scenarios and
munications and Networks, 2019.
landmarks. We will try to make a robot pilot more originally
[8] S. Fu, H.-y. Liu, L.-f. Gao, and Y.-x. Gai, “Slam for mobile
and also apply SLAM with UKF and PF algorithms. In addi-
robots using laser range finder and monocular vision,” in
tion, improving the development of some standards for esti- 2007 14th International Conference on Mechatronics and
mating SLAM approaches, particularly for large-scale SLAM Machine Vision in Practice, pp. 91–96, Xiamen, China,
in dynamic situations, is also important to make the results of December 2007.
SLAM algorithms more valuable. [9] C. Kerl, J. Sturm, and D. Cremers, “Dense visual slam for rgb-d
cameras,” in 2013 IEEE/RSJ International Conference on Intel-
Data Availability ligent Robots and Systems, pp. 2100–2106, Tokyo, Japan,
November 2013.
Since the funding project is not closed and related patents [10] Y. Li, S. Xia, M. Zheng, B. Cao, and Q. Liu, “Lyapunov optimi-
have been evaluated, the simulation data used to support zation based trade-off policy for mobile cloud offloading in
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Wireless Communications and Mobile Computing 11

heterogeneous wireless networks,” IEEE Transactions on Mediterranean Conference on Control and Automation,
Cloud Computing, 2019. pp. 19–26, Chania, Greece, June 2013.
[11] C. Cadena, L. Carlone, H. Carrillo et al., “Past, present, and [27] S. Huang and G. Dissanayake, “A critique of current develop-
future of simultaneous localization and mapping: toward the ments in simultaneous localization and mapping,” Interna-
robust-perception age,” IEEE Transactions on Robotics, tional Journal of Advanced Robotic Systems, vol. 13, no. 5,
vol. 32, no. 6, pp. 1309–1332, 2016. article 1729881416669482, 2016.
[12] M. Raja, “Application of cognitive radio and interference can- [28] G. Dissanayake, S. Huang, Z. Wang, and R. Ranasinghe, “A
cellation in the l-band based on future air-to-ground commu- review of recent developments in simultaneous localization
nication systems,” Digital Communications and Networks, and mapping,” in 2011 6th International Conference on Indus-
vol. 5, no. 2, pp. 111–120, 2019. trial and Information Systems, pp. 477–482, Kandy, Sri Lanka,
[13] I. Ullah, J. Chen, X. Su, C. Esposito, and C. Choi, “Localization August 2011.
and detection of targets in underwater wireless sensor using [29] R. C. Smith and P. Cheeseman, “On the representation and
distance and angle based algorithms,” IEEE Access, vol. 7, estimation of spatial uncertainty,” The International Journal
pp. 45693–45704, 2019. of Robotics Research, vol. 5, no. 4, pp. 56–68, 2016.
[14] G. Bresson, Z. Alsayed, L. Yu, and S. Glaser, “Simultaneous [30] K.-K. Tseng, J. Li, Y. Chang, K. L. Yung, C. Y. Chan, and C.-
localization and mapping: a survey of current trends in auton- Y. Hsu, “A new architecture for simultaneous localization
omous driving,” IEEE Transactions on Intelligent Vehicles, and mapping: an application of a planetary rover,” Enterprise
vol. 2, no. 3, pp. 194–220, 2017. Information Systems, pp. 1–17, 2019.
[15] Y. Zhang, H. Wen, F. Qiu, Z. Wang, and H. Abbas, “Ibike: [31] Z. Miljković, N. Vuković, and M. Mitić, “Neural extended
intelligent public bicycle services assisted by data analytics,” Kalman filter for monocular slam in indoor environment,”
Future Generation Computer Systems, vol. 95, pp. 187–197, Proceedings of the Institution of Mechanical Engineers, Part
2019. C: Journal of Mechanical Engineering Science, vol. 230, no. 5,
[16] J. Aulinas, Y. R. Petillot, J. Salvi, and X. Lladó, “The slam prob- pp. 856–866, 2015.
lem: a survey,” CCIA, vol. 184, no. 1, pp. 363–371, 2008. [32] S. Prakash and G. Gu, Simultaneous localization and mapping
[17] P. Jensfelt, D. Kragic, J. Folkesson, and M. Bjorkman, “A with depth prediction using capsule networks for uavs, 2018,
framework for vision based bearing only 3d slam,” in Proceed- http://arxiv.org/abs/1808.05336.
ings 2006 IEEE International Conference on Robotics and Auto- [33] N. Ayadi, N. Derbel, N. Morette, C. Novales, and G. Poisson,
mation 2006. ICRA 2006, pp. 1944–1950, Orlando, FL, USA, “Simulation and experimental evaluation of the ekf simulta-
May 2006. neous localization and mapping algorithm on the wifibot
[18] A. J. Davison and D. W. Murray, “Simultaneous localization mobile robot,” Journal of Artificial Intelligence and Soft Com-
and map-building using active vision,” IEEE Transactions on puting Research, vol. 8, no. 2, pp. 91–101, 2018.
Pattern Analysis and Machine Intelligence, vol. 24, no. 7, [34] G. Wang and A. Fomichev, “Simultaneous localization and
pp. 865–880, 2002. mapping method for a planet rover based on a gaussian filter,”
[19] S. Se, D. Lowe, and J. Little, “Mobile robot localization and in InAIP Conference Proceedings, vol. 2171, no. 1, article
mapping with uncertainty using scale-invariant visual land- 160003, 2019AIP Publishing, 2019.
marks,” The international Journal of robotics Research, [35] I. Ullah, Y. Shen, X. Su, C. Esposito, and C. Choi, “A localiza-
vol. 21, no. 8, pp. 735–758, 2016. tion based on unscented kalman filter and particle filter local-
[20] M. N. Santhanakrishnan, J. B. B. Rayappan, and R. Kannan, ization algorithms,” IEEE Access, vol. 8, no. 1, pp. 2233–2246,
“Implementation of extended kalman filter-based simulta- 2020.
neous localization and mapping: a point feature approach,” [36] F. F. Yadkuri and M. J. Khosrowjerdi, “Methods for improving
Sādhanā, vol. 42, no. 9, pp. 1495–1504, 2017. the linearization problem of extended kalman filter,” Journal of
[21] T. Rahman, X. Yao, and G. Tao, “Consistent data collection Intelligent & Robotic Systems, vol. 78, no. 3-4, pp. 485–497,
and assortment in the progression of continuous objects in 2015.
iot,” IEEE Access, vol. 6, pp. 51875–51885, 2018. [37] O. Ozisik and S. Yavuz, “Simultaneous localization and map-
[22] Y. Li, J. Liu, B. Cao, and C. Wang, “Joint optimization of radio ping with limited sensing using extended kalman filter and
and virtual machine resources with uncertain user demands in hough transform,” Tehnicki vjesnik - Technical Gazette,
mobile cloud computing,” IEEE Transactions on Multimedia, vol. 23, no. 6, pp. 1731–1738, 2016.
vol. 20, no. 9, pp. 2427–2438, 2018. [38] Y. Tian, H. Suwoyo, W. Wang, and L. Li, “An asvsf-slam algo-
[23] X. Su, I. Ullah, X. Liu, and D. Choi, “A review of underwater rithm with time-varying noise statistics based on map creation
localization techniques, algorithms, and challenges,” Journal and weighted exponent,” Mathematical Problems in Engineer-
of Sensors, vol. 2020, 24 pages, 2020. ing, vol. 2019, 17 pages, 2019.
[24] P. Yang and W. Wu, “Efficient particle filter localization algo- [39] C.-C. Tsai, C.-F. Hsu, X.-C. Lin, and F.-C. Tai, “Cooperative
rithm in dense passive rfid tag environment,” IEEE Transactions slam using fuzzy kalman filtering for a collaborative air-
on Industrial Electronics, vol. 61, no. 10, pp. 5641–5651, 2014. ground robotic system,” Journal of the Chinese Institute of
[25] P. Yang, “Efficient particle filter algorithm for ultrasonic Engineers, vol. 43, no. 1, pp. 67–79, 2020.
sensor-based 2d range-only simultaneous localisation and [40] J. Jung, Y. Lee, D. Kim, D. Lee, H. Myung, and H.-T. Choi,
mapping application,” IET Wireless Sensor Systems, vol. 2, “Auv slam using forward/downward looking cameras and arti-
no. 4, pp. 394–401, 2012. ficial landmarks,” in 2017 IEEE Underwater Technology (UT),
[26] G. Cotugno, L. D’Alfonso, W. Lucia, P. Muraca, and pp. 1–3, Busan, South Korea, February 2017.
P. Pugliese, “Extended and unscented kalman filters for mobile [41] H. Abdelnasser, R. Mohamed, A. Elgohary et al., “Semantic-
robot localization and environment reconstruction,” in 21st slam: using environment landmarks for unsupervised indoor
 6302, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2020/2138643, Wiley Online Library on [18/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 Wireless Communications and Mobile Computing

localization,” IEEE Transactions on Mobile Computing, vol. 15,

no. 7, pp. 1770–1782, 2016.
[42] H. Ahmad, N. A. Othman, and M. S. Ramli, “A solution to
partial observability in extended kalman filter mobile robot
navigation,” Telkomnika, vol. 16, no. 1, pp. 134–141, 2018.
[43] P. Thulasiraman and K. A. White, “Topology control of tacti-
cal wireless sensor networks using energy efficient zone rout-
ing,” Digital Communications and Networks, vol. 2, no. 1,
pp. 1–14, 2016.
[44] F. Demim, A. Nemra, K. Louadj, Z. Mehal, M. Hamerlain, and
A. Bazoula, “Simultaneous localization and mapping algo-
rithm for unmanned ground vehicle with svsf filter,” in 2016
8th International Conference on Modelling, Identification and
Control (ICMIC), pp. 155–162, Algiers, Algeria, November
2016.
[45] C. H. Do and H.-Y. Lin, “Incorporating neuro-fuzzy with
extended kalman filter for simultaneous localization and map-
ping,” International Journal of Advanced Robotic Systems,
vol. 16, no. 5, article 1729881419874645, 2019.
[46] X. Xie, Y. Yu, X. Lin, and C. Sun, “An ekf slam algorithm for
mobile robot with sensor bias estimation,” in 2017 32nd Youth
Academic Annual Conference of Chinese Association of Auto-
mation (YAC), pp. 281–285, Hefei, China, May 2017.
[47] Z.-L. Ren, L.-G. Wang, and L. Bi, “Improved extended kalman
filter based on fuzzy adaptation for slam in underground tun-
nels,” International Journal of Precision Engineering and
Manufacturing, vol. 20, no. 12, pp. 2119–2127, 2019.
[48] C. H. Do, H.-Y. Lin, and Y.-C. Huang, “Simultaneous localiza-
tion and mapping with neuro-fuzzy assisted extended kalman
filtering,” in 2017 IEEE/SICE International Symposium on Sys-
tem Integration (SII), pp. 393–398, Taipei, Taiwan, December
2017.
[49] J. Dai, X. Li, K. Wang, and Y. Liang, “A novel stsoslam algo-
rithm based on strong tracking second order central difference
kalman filter,” Robotics and Autonomous Systems, vol. 116,
pp. 114–125, 2019.
[50] J. Bai, J. Gao, Y. Lin, Z. Liu, S. Lian, and D. Liu, “A novel feed-
back mechanism-based stereo visual-inertial slam,” IEEE
Access, vol. 7, pp. 147721–147731, 2019.
[51] A. Giannitrapani, N. Ceccarelli, F. Scortecci, and A. Garulli,
“Comparison of ekf and ukf for spacecraft localization via
angle measurements,” IEEE Transactions on Aerospace and
Electronic Systems, vol. 47, no. 1, pp. 75–84, 2011.
[52] D. Fethi, A. Nemra, K. Louadj, and M. Hamerlain, “Simulta-
neous localization, mapping, and path planning for unmanned
vehicle using optimal control,” Advances in Mechanical Engi-
neering, vol. 10, no. 1, Article ID 168781401773665, 2018.