IEEE Consumer Electronics Magazine - May 2018 - 13

provide inferior accuracy caused by an unstable signal
because the signal strength can be influenced by objects or
nearby base station numbers.
In recent years, several indoor positioning systems have
been proposed. In [6]-[9], the studies outlined a positioning system that uses Bluetooth beacons to locate users
with host devices. These systems compute the user's
position by measuring the signal strength between
the host device and the Bluetooth beacons. However, the Bluetooth positioning system may be interfered with by objects, e.g., walls or doors. To be
effective, this system requires hundreds of beacons to be installed to get the precise location,
which is inefficient.
Several studies in [10]-[13] have developed positioning systems based on the
matched features on corresponding frames.
These systems develop a database that contains several types of information, including
the location information and the corresponding scene features. When a frame of the pictures is taken, the system can locate the user
through matching features between the picture
and scenes in the database. Nevertheless, scene
features could easily be interfered with by changes in the environment. Furthermore, powerful computation is needed for matching. In [14] and [15], a
method that uses an augmented reality (AR) marker is
developed to obtain the user's location. The system
locates the user by matching the marker rather than the
scene. Since this system requires the markers to be installed
in every location, it would be arduous to implement it globally.
People with visual impairment need not only the position
but also information about the surrounding objects. Generally, canes are used to detect objects and avoid obstacles, but
canes are impractical to detect objects above the waistline.
To overcome this problem, Bharathi et al. [16] presented a
positioning system using a cane and a pair of glasses installed
with ultrasonic sensors to detect obstacles above the ground
and under the human body. Kim et al. [17] proposed a jacket
and a bag combined array of ultrasonic sensors, camera, and
other sensors to implement the guidance system. Unfortunately, these devices are impractical for daily use.
In this article, an indoor positioning system and obstacle
detection are implemented on a pair glasses and on a mobile
phone. The device was installed with a camera and three
ultrasonic transducers. The system was developed based on
image markers. Owing to the limitation of visual markers,
this article presents a custom color marker embedded with
the location information. This marker can be detected at a
greater distance than the previous markers and processed in
real time on a mobile phone. Compared to other markers that
require information about locations from the database, this
marker can embed information independently. The obstacle
detection uses three ultrasonic transducers to widely detect
obstacles above the waistline and to notify its users through

audio and haptic feedback. The positioning and obstacle
detection can work independently or in parallel. To ensure the
portability of the proposed wearable device, microsensors
and a small-sized microprocessor module are utilized.

rELaTEd WorkS
In this section, the previous marker designs and the existing
systems are summarized before elaborating on the proposed
marker. The visual markers include many types, including
road signs and signboards. With the advancement of computer
vision, computers could recognize markers through machine
learning or other methods [18], [19]. Although proposing an
algorithm to robustly recognize different markers is technically plausible, there are vast numbers of markers with distinct
patterns, making its actual implementation arduous. To properly achieve a desirable accuracy, a significant number of
training sets are required for generating the detection model,
which requires tedious labeling efforts. The barcode markers
can offer better performance for a positioning system. Different types of barcodes have their own protocols and include
varying data.

AN AR mARkeR
Most of the existing markers for the positioning systems utilize
AR positioning markers. The AprilTag [20] is an AR marker
that is similar to the matrix barcode. The structure of the
AprilTag comprises a black square and various-sized white
blocks on the base. Based on this structure, the AR-based system can be used to track angles and relative distances between
the camera and the marker. Each AprilTag has a unique identification (ID), but it cannot be used to store specific text information, thus requiring a database. It is a challenge to make
universal rules for database integration and implementing the
limited possibilities of a marker.

A Quick RespoNse code
A quick response (QR) code is one of the matrix barcodes that
includes three positioning points, one or more alignment
points, and the main data patterns. The number of alignment
points and data patterns varies according to the standard of the
corresponding QR code. The current version of the QR code
has larger data capacity than the previous, and it consumes
more decoding and searching time. Additionally, the marker
possibility will be limited depending on the encoding scheme.
A QR code is a more suitable type of marker for positioning systems because of its ability to store text information.
The text information can help the system get brief position
information without a database, thus establishing the position
of the user. Nevertheless, the detection range of QR-based
systems is extremely limited. With regard to visually
impaired individuals, it is a major drawback to use a QR code
as a marker.

THe TARgeT mARkeR
Christen et al. [19] proposed a marker that comprises a black
circle on the outside for the detection step and a concentric
MAy 2018

IEEE Consumer Electronics Magazine

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - May 2018