Page 84 - Kaleidoscope Academic Conference Proceedings 2022

2022 ITU Kaleidoscope Academic Conference

The absence of GPS signals in public places such as 2. PROPOSED SYSTEM MODEL
underground shopping malls, subway stations, and
basements of multistorey buildings constitutes a requirement The architecture of the proposed MAR–PNS consists of
of a new user localization mechanism in digital online maps three modules namely the AR Navigation (ARN) module,
to support walking directions. Counting the footsteps of the Step count-based Distance Estimation (SDE) module, and
user is useful in estimating the distance covered by them the Landmark Recognition (LR) module (Figure 2). The AR
between successive GPS signals. Real-time step counting Navigation module involves getting the optimum route
algorithms are useful for indoor pedestrian navigation by between source and destination on digital map, details about
creating a probabilistic map which stores geographic data the pedestrian user (such as estimated length of distance
primitives [7]. Accurate step counts can be estimated by between legs), turning point coordinates, and generating /
using a high pass filter on the accelerometer readings. rendering AR turning point indicator objects according to the
Another key parameter while estimating distance from step proposed algorithm. The SDE module helps in localizing the
counts is to take into account the step length of the user. user’s position with respect to the turning point in the
However, step length changes dynamically while the user absence of GPS. There are two major inputs from the user
walks and this step length needs to be detected at every point i.e., the destination address and the live feed captured by
[8]. One of the shortcomings of existing step counters for their mobile camera. The destination point-based route
Android devices is the latency in generating the step count. information provided by the digital map service is integrated
According to the official Android documentation [9], the with the ARN module along with the LR to enhance the
built-in step counter requires at least 10 seconds to display pedestrian navigation in real time. The LR module estimates
accurate step count. In our proposed work, we aim to reduce the location of the landmark which the user’s camera is
this step count latency by leveraging a temporal filtering pointed to and displays relevant information about the
technique coupled with a threshold-based magnitude filter. landmark in the MAR device. The AR objects generated by
the ARN and LR modules are anchored to their geo-positions,
Recommendation ITU-T Y.5462 [10] discusses spatio- localized with respect to the user’s camera view and rendered
temporal information services for smart cities requiring on the user’s screen.
navigational and positioning systems for citizens to
determine a person’s location as well as provide a digital 2.1 AR Navigation (ARN) module
mapping system displaying instructions for moving from one
place to another. The MAR-based solution for pedestrian The ARN module is designed to display walking direction in
navigation needs to be considered for the enhancement of an AR environment based on turning point markers. It uses
user experience. a digital mapping service to obtain an optimal route from
source to destination. The last known location of the user is
In this paper, we present a marker-less Mobile Augmented obtained from GPS and the user’s position is localized by
Reality-based Pedestrian Navigation System (MAR–PNS), polling the user’s position either by using GPS or the step
which provides a visual solution for the enhancement of user count-based distance estimation module. Whenever the user
experience by incorporating landmark detection in pathways, is near a turning point, the turning point direction (AR object)
and improves the accuracy of a navigational system in the is generated (right or left direction depending on the turning
absence of GPS signals. The system uses an AR framework point) and rendered on the user’s screen.
to render the turning points and landmarks on demand by the
mobile user. It allows landmark recognition by relying on the 2.2 Step count-based Distance Estimation (SDE)
user’s current location and direction in which they’re module
pointing their phone. The system mitigates the absence of
GPS in scenarios such as underground subways by This module mitigates the unreliability or unavailability of
integrating a step count-based user localization method. The GPS during navigation. It uses the step count of the user and
MAR–PNS model complies with Recommendation ITU-T performs arbitrary calculations to translate the number of
Q.4066 “Testing procedures of augmented reality steps to the distance covered by the user using parameters
applications” [11] and can be a solution for smart city such as the user’s height and waking duration. The trade-off
navigational systems, while meeting the requirements of decision in minimizing latency was taken based on the fact
Recommendation ITU-T Y.4562 “Functions and metadata of that in pedestrian navigation a few missteps do not affect the
spatiotemporal information service for smart cities” [10]. overall working of the application and an error threshold can
be maintained. This design consideration is also taken
The remainder of the paper is organized as follows. Section keeping in mind that even intermittent GPS signals can help
2 provides the architectural details of the proposed system in localization of the user with the distance estimation
and Section 3 describes the algorithms developed during the module. The other design requirement includes the overall
course of our work. Section 4 describes the implementation physique of the user which can affect the foot stride.
environment and evaluates our system in terms of locational
accuracy and other metrics. Section 5 provides concluding
remarks and offers scope for future work.

– 38 –