19-05-2014, 02:21 PM
Self-Recognition of Vehicle Position Using UHF Passive RFID Tags
Self-Recognition of Vehicle.pdf (Size: 654.22 KB / Downloads: 27)
Abstract
This paper proposes a method that enables self-
recognition of a mobile vehicle’s current position by utilizing
ultrahigh frequency (UHF) passive radio-frequency identification
(RFID) tags. The proposed method can be used in real industry
environments such as complex storage warehouses where many
different goods are dispersed throughout a wide area. In partic-
ular, the proposed method makes use of two UHF RFID readers
with identical emission configuration attached to a vehicle to iden-
tify a reference RFID tag. By utilizing the received signal strength
indicator obtained by the readers from the reference RFID tag,
the precise position of the moving vehicle can be obtained. The
experiments prove the effectiveness of the proposed method in
accurately estimating the vehicle position.
I NTRODUCTION
DURING THE past five years, the ultrahigh frequency
(UHF) passive radio-frequency identification (RFID)
technology has been widely adopted as a direct response to
the needs of the supply chain management. When products
affixed with UHF passive RFID tags (“Tag(s)”) are released,
they travel from manufacturing plants to warehouses to retail
shops. For supply chain management operators, it would be
of great interest to be able to detect the current location of
such products in real time. In real-life applications, since most
products are shipped on “Global Positioning System” tracked
vehicles, their locations can be readily identified while they are
en route. However, in order to identify the current locations of
such products in an indoor environment, one needs to either
manually record their exact locations or locate the indoor vehi-
cles carrying such products. Generally, to identify momentary
locations of such vehicles, odometry is widely used. Odometry
enables a vehicle to estimate the total distance traveled from
a starting point. However, odometry is often inaccurate since
estimation errors accumulate over time without corrections by
external reference signals. Thus, there has been a growing
interest in supplementing odometry to improve the localization
of mobile vehicles, particularly by using Tags [1]–[8].
C ONVENTIONAL PARTICLE F ILTER M ETHOD
FOR P OSITION E STIMATION
In UHF passive RFID technology, a reader can receive and
obtain RSSI information from Tags. However, the distance
between the Tags and the reader cannot be readily obtained
[5], [10], [12]. Thus, the calculation of the distance between
the Tags and a reader by using RSSI requires the first step
of determining the spatial variant RSSI model by classifying
L = {l0 , l1 , l2 , . . .} ({0, 1, 2, . . .} ∈ integer) as the preselected
positions for the likely moving path of the vehicle and obtaining
the signal strength observation o from each marked position L.
Next, it would be necessary to find the position equating to the
observed RSSI from the spatial variant RSSI model to estimate
the distance between the Tags and the reader. Numerous studies
have been devoted to accurately identifying the position by
using a particle filter, a well-known statistical deduction method
[10], [18], [22].
Influence of the Number of Particles
In the first experiment, the influence of the number of par-
ticles on the position estimation accuracy for the conventional
particle filter method and the proposed method was tested. As
shown in former studies related to localization, such as [3],
[13]–[15], [18], and [24], with a greater number of samples
applied to the algorithm, high localization accuracy can be
obtained. However, in real operation, it is difficult to secure
many particles in one location of a moving vehicle. Thus, it is
necessary to investigate the performance of the position estima-
tion method when reduced numbers of particles are available
to the method. Fig. 7 shows the error distance between the
conventional particle filter method and the proposed method
with 20, 50, and 100 particles.
C ONCLUSION
In this paper, a novel method for achieving a highly accurate
self-localization and industrially useful application has been
presented. In order to identify the location of a vehicle, first,
while the vehicle is in motion, the odometry of the vehicle
is used to predict the vehicle’s position hypothesis. Second,
the RSSI values for the position estimations of the vehicle are
obtained. Third, by finding the RSSI equating to the location of
the position estimation from the spatial variant RSSI model, the
measured RSSI is replaced. Fourth, by recomparing the updated
observations of the two readers and using the Tag position
of the unique observation with identical RSSI or the smallest
difference of RSSI, the actual Tag location is corrected. Finally,
using the Tag location, the vehicle’s location is corrected.