Smartphone-based pedestrian dead-reckoning (PDR) has become promising in indoor localization since it locates users with a smartphone only. However, existing PDR approaches are still facing the problem of accumulated localization errors due to low-cost noisy sensors and complicated human movements.ThispaperpresentsanovelPDRindoorlocalizationalgorithmcombinedwithonlinesequential extreme learning machine (OS-ELM). By analyzing the process of PDR localization, this paper ο¬rst formulatestheprocessofPDRlocalizationasanapproximationfunction,andthen,asliding-window-based scheme is designed to preprocess the obtained inertial sensor data and thus to generate the feature dataset. At last, the OS-ELM-based PDR algorithm is proposed to address the localization problem of pedestrians. Due to the fact of universal approximation capability and extreme learning speed within OS-ELM, our algorithmcanadapttolocalizationenvironmentdynamicallyandreducethelocalizationerrorstoalowscale. Inaddition,bytakingthemovementhabitsofpedestrianintotheprocessofextremelearning,ouralgorithm can predict the position of pedestrian regardless of holding postures. To evaluate the performance of the proposed algorithm, this paper implements OS-ELM-based PDR on a real android-based smartphone and comparesitwiththestate-of-the-artapproaches.Extensiveexperimentresultsdemonstratetheeffectiveness of the proposed algorithm in various different postures and the practicability in indoor localization.
Pedestrian dead reckoning indoor localization based on os-elm
1. 1
Pedestrian Dead-Reckoning Indoor Localization
Based on OS-ELM
MINGYANG ZHANG, YINGYOU WEN, JIAN CHEN, XIAOTAO YANG,
RUI GAO AND HONG ZHAO
January 10, 2018
IEEE Access
2. 2
Motivation
The objective of this paper is to reduce the problems related to
accumulated localization error and complicated human
movements for indoor localization. A novel PDR indoor
localization algorithm combined with online sequential
extreme learning machine (OS-ELM) is used for localization.
3. 3
Contributions
ο¬ Proposed the first OS-ELM based PDR algorithm.
ο¬ Zero-crossing detection with a threshold-based peak detection for step detection.
ο¬ The proposed system will not affect the different postures of holding the phone.
ο¬ Designed a framework of OS-ELM based PDR for localizing pedestrians.
4. 4
Introduction to PDR
ο¬ The pedestrian position can be
computed as
π₯ π+1
π¦ π+1
=
π₯ π
π¦ π
+ππΏ π+1
sin(π»π· π+1)
cos(π»π· π+1)
(1)
ο¬ The three procedures used in PDR can
be extracted as the following functions
ππ· = ππ π(π)
π»π· = πβπ(π, π)
πL = ππ π(π)
Example of pedestrian dead-reckoning.
Where a, m and g are the values obtained from accelerometer, magnetometer and gyroscope.
πβπ , ππ π and ππ π are the rules for estimating heading angles, detecting steps and estimating
stride length.
SD,HD and SL are the values of step detection, heading angles and stride length.
5. 5
STEP DETECTION
ο¬ To overcome the tilting effect, the proposed algorithm transforms the raw acceleration
from smartphone coordinate system (SCS) to earth coordinate system (ECS).
ο¬ To compute the acceleration in ECS, the proposed algorithm computes the rotation
matrix from SCS to ECS.
π π§ π π‘ =
πππ π π‘ π ππ π π‘ 0
β sin π π‘ cos π π‘ 0
0 0 1
π π₯ ππ‘ =
1 0 0
0 cos ππ‘ sin ππ‘
0 βπ ππππ‘ cos ππ‘
π π¦ π π‘ =
cos π π‘ 0 sin π π‘
0 1 0
β sin π π‘ 0 cos π π‘
Where π π‘, ππ‘ and π π‘ are the a azimuth angle, pitch angle and roll angle at the t-th sampling
moment.
ο¬ The total rotation matrix of the z-x-y axes can be written as
π π‘
π§π₯π¦
= π π§ π π‘ π π₯ ππ‘ π π§ π π‘ -------- (8)
ο¬ Transformation of acceleration from SCS to ECS can be written as
π π‘
πΈπΆπ
= π π‘
π§π₯π¦
π π‘
ππΆπ
------ ----- (9)
ο¬ The z-axis component of the acceleration contains gravity, and then the proposed
algorithm eliminate the effect of gravity as
π π‘
πΏπππππ
= π π‘
πΈπΆπ
β π[0,0,1] π
6. 6
ο¬ To reduce the effect of noise, the proposed algorithm performs a moving average filter
operation as
π π‘ =
1
π π π
π=π‘βπ π π+1
π‘
π π§,π
πΏπππππ
Where the π π πis the order of moving window. The filter linear acceleration π π‘ is used for
detecting steps. Example of step detection.
β’ This paper proposes an accurate step detection approach that combines the zero crossing
detection with peak detection.
7. 7
Stride Length and Heading Direction Estimation
ο¬ The pedestrian stride length can be computed as
ππΏπ = π
4
π π‘π
π
β π π‘π
π
Where π π‘π
π
(π π‘π
π
) is the peak (valley) of filtered linear acceleration at the i-th time step and K
is the coefficient.
π =
πππΏπ
4
π π‘π
π
β π π‘π
π
πππΏπ
2
π π‘π
π
β π π‘π
π
ο¬ The heading angle at time t can be written as
π»π·π‘ = πβπ π π‘, ππ‘ = π π‘
ο¬ The proposed algorithm replaces the aforementioned heading direction and stride length
estimation with an OS-ELM based localization approach.
9. 9
FRAMEWORK OF PROPOSED PDR LOCALIZATION
ο¬ The framework contains two phases
1. The model training phase (dashed arrows)
ο¬ Sensor data are processed into features and labels ο used for training OS-ELM models.
ο¬ The proposed algorithm constructs two OS-ELM models ο The stride length estimation and heading
direction estimation
2. The PDR localization phase (solid-line arrows)
ο¬ Estimates the stride length and heading direction by substituting the localization request data into trained
OS-ELM models.
12. 12
Specification
ο¬ The threshold πΏ π
+
and πΏ π
β
to be 0.5
ο¬ The size of sliding window W to be 20
ο¬ Coefficient K to be 0.47
ο¬ Moving average π π πto be 3,
πβπto be 15, π π π to be 4
ο¬ Expanding times of heading direction
epochβπ
to be 5
ο¬ Expanding times of stride length
epoch π πto be 20
13. 13
SELECTION OF PARAMETERS FOR
OS-ELM MODELS
ο¬ This paper evaluates the performance of three different activation functions:
1. Radial basis function
2. Sigmoid function
3. Sine function
β’ The number of hidden nodes for heading direction model : 300
β’ The number of hidden nodes for stride length model: 300
β’ The sine activation function is chosen as the activation function for stride length model
and heading direction model.
14. 14
EXPERIMENT RESULTS
ο¬ The data in path 1 is chosen to compare the proposed step detection approach with some
popular step detection approaches. The relative error is employed to evaluate the
performance, which is defined as
π =
ππ β ππ
ππ
Γ 100%
where ππ is the number of detected steps, and ππ is the ground truth.
15. 15
The performance of stride length and
heading direction estimation
ο¬ To evaluate the performance of stride
length, the proposed approach is
compared with the typical linear
approach and nonlinear approach.
ο¬ The path 2 is chosen to evaluate the
performance of heading direction estimation
approaches.
16. 16
Evaluation of the training time of the
proposed algorithm in real smartphone
ο¬ In the experiment of training stride length model:
The total number of samples =1020
The total training time = 0.945
The training time of initialization phase =0.112s
The average training time of sequential phase = 0.0203s
ο¬ In the experiment of training heading direction model:
The total number of samples =7105
The total training time = 40.35s
The training time of initialization phase =4.033s
The average training time of sequential phase = 0.1117s
β’ The training time of sequential learning phase can satisfy the requirement of online
learning.
β’ Therefore, it is practicable to deploy the propose localization algorithm in a real
smartphone.
17. 17
Conclusions
ο¬ Proposed an OS-ELM based PDR indoor localization algorithm for android-based
smartphone.
ο¬ The proposed localization algorithm does not force the smartphone to be held in
fixed posture.
ο¬ Zero-crossing detection with a threshold based peak detection method is used
for step detection.
ο¬ OS-ELM localization frame work is used for stride length and heading direction
estimation.
ο¬ Sliding-window based scheme is used for preprocessing feature data.
ο¬ The proposed PDR algorithm can continuously train OS-ELM online and generate
OS-ELM models for pedestrians movements.
ο¬ The experiment results demonstrate the effectiveness of the proposed algorithm in
various different postures.