Conference PaperPDF Available

Efficacy of Localization Through Magnets Embedded in Infrastructure

1,2New Jersey Institute of Technology, United States of America
Abstract. This paper investigates localization and guidance systems as
important future considerations for autonomous mobility within the built
environment. Specifically, it looks at embedding magnets within build-
ing construction assemblies, using magnetic sensors for autonomous
navigation, and understanding the impact construction materials may
have on magnetic-field-based localization and guidance systems of au-
tonomous agents.
Keywords. Autonomous Car; Localization; Infrastructure;
1. Introduction
The developments in personal and autonomous mobility unavoidably raise the
question of the interface between the built environment (buildings, cities, and
infrastructure) and autonomous agents (cars, drones, house robots, or assisted
wheelchairs). This interface needs to consider and accommodate new players with-
out an adverse impact on current users. This inclusive, yet well-tuned, integration
will be critical in a successful transition from the current inert habitation to future
autonomously empowered habitation. While the focus of these transformations
may seem to be machine based, accommodating various forms of autonomous
mobility, ultimately it will lead into more inclusive environments that will em-
power human users, particularly the elderly, children, and people with mobility
impairments (Cowan et al. 2012; Madarasz et al. 1986).
The need for the seamless integration of new and old users may suggest a partic-
ular approach to development of autonomous-mobility-friendly environments that
focus on redefinition of materiality and placement of strategic micro-gestures (em-
bedded capabilities) without an overall typological rethinking of what is expected
from buildings today. This attitude could form the first step in making buildings
and cities ready for an effective integration of autonomous agents. It would also
help in defining a strategy for a transition toward autonomous mobility.
This paper considers localization and guidance systems as important future
considerations for autonomous mobility within built environment. Specifically, it
P. Janssen, P. Loh, A. Raonic, M. A. Schnabel (eds.), Protocols, Flows and Glitches, Proceedings of the
22nd International Conference of the Association for Computer-Aided Architectural Design Research in Asia
(CAADRIA) 2017, 735-745. © 2017, The Association for Computer-Aided Architectural Design Research
in Asia (CAADRIA), Hong Kong.
looks at embedding magnets within building construction assemblies, using mag-
netic sensors for autonomous navigation, and understanding the impact construc-
tion materials may have on magnetic-field-based localization and guidance of au-
tonomous agents.
Numerous methods for localizing autonomous robots (including cars) have
been studied for decades (Borenstein et al. 1996). In many cases, technology such
as GPS can guide both human and robot drivers, with small adaptations either to
be calculated by the robot or visualized for the human. Built-in technology such as
road infrastructure provides an implicit layer for humans to interpret. Complica-
tions arise in decision-making, particularly in relation to other vehicles, but also in
unknown circumstances. While a painted line across a stop sign can be interpreted
by a person as graffiti, the difference from an untouched sign makes it difficult
for image recognition. These variations create difficulty in localization for robots,
and as such require multiple strategies and approaches to effectively navigate the
environment. If any technology could be provided at a higher rate of reliability
and consistency, the reliance on other technologies could be reduced. In the case
of GPS, the satellite must be visible, eliminating tunnels; RGB cameras require
lighting to detect elements; and laser scanning can have interference with reflec-
tive and transparent materials. If the infrastructure allowed for the technology to
offset these challenges, the rate of adoption would greatly increase.
Indoor environments also provide ample examples of hallways with glass cor-
ridors or open doors; of glazed ceramic or stone wall coverings; and of free-form
designed and undulated surfaces with posters, flyers, and unattended objects that
could significantly affect the autonomous agent’s ability to use computer vision
and to navigate. Just as with the changing car scene on roads, anything from peo-
ple to wheelchairs to dogs can populate an indoor environment. Between stands
for posters, conversations around a coffee shop, and people passing by, the record-
ing and comparison of image data can be nearly useless. The ability for a “ground
truth” localization method would allow for the guaranteed path center, while sup-
plementary data from Light Detection and Ranging (LiDAR) devices or RGB
videos and could offer collision avoidance.
2. Background
Induction through a metal coil embedded in the road has been used for a long
time to study traffic patterns. The induction loop is able to detect when a large
ferrous metal object is near. When multiple loops are implemented in the road,
timings for cars to pass over can be calculated. These induction loops can be used
for traffic signals to detect when vehicles are present and trigger various traffic
controls. Furthermore, these induction sensors can be fine-tuned to distinguish
between various vehicle categories, such as cars, vans, trucks, and buses (Sun et
al. 2003). This removes the need for image sensing of vehicles, which can be less
reliable depending on light and atmospheric conditions, while improving on the
basic timed system.
The reverse use of this technology would allow vehicles to detect their location
within a road or a relationship to various obstacles and infrastructure elements. By
decreasing the size and increasing the spacing, accurate location information for
the road can be detected. More specifically, the knowledge that these devices are
centered in the road can provide a reliable and trustworthy localizing information
source to the robot (or car). The same induction technologies are used, for example,
as an “invisible fence” to control boundary conditions for house animals and as
household robotic devices such as autonomous lawn mowers.
Induction loops could be designed to provide unique characteristics, as is the
case with Radio-Frequency IDentification (RFID) technology that allows for short-
to mid-range distance detection. However, RFIDs can be used to track general
inventories (presence) with long-range implementations or specific locations with
short-range implementations.
While there is ample (1) research and implementation of using magnetic field
and sensors for autonomous guidance systems and (2) research into shielding mag-
netic fields and how various materials contribute to it (Shi et al. 1995; Sumner et
al. 1987; Graham and Cullity 2009), little research has been done on materials
specific to construction and architecture with an integration of magnets into con-
struction assemblies as a base for autonomous localization and guidance.
The research on material shielding in physics and material sciences may not
always be fully applicable to architecture and construction. Construction, espe-
cially executed on-site, introduces a high level of imprecision and usually requires
a high tolerance for errors as compared to other manufacturing technologies and
sciences. Materials used are often produced locally and may vary in their ferrous
metal contents from similar materials elsewhere. This means that uniformity and
measured outcomes may have a broad range of variations.
Construction assemblies such as floors or walls are hardly ever composed of a
single material. For example, concrete slabs are made of a single or double pour
with a top coating providing a finished surface. In the case of a single pour, the top
of the concrete slab is often covered with carpet, wood, or tiles, usually set in with
a thin set or an adhesive. Within the slab, depending on the desired structural per-
formance, properties, and construction technology, there may be additional metal
deck panning, reinforcement steel bars (rods), or metal wire mesh to reinforce
concrete top surfaces against point loads. In addition to structural steel, floor slabs
may also integrate a number of other elements, such as cable conduits or piping
for radiant heating.
For the reason of its complexity, it is desirable to develop mock-ups for vari-
ous types of floor slab assemblies and test them as composites. The case studies
below show magnetic shielding tests of individual materials such as floor tiles,
wood composites, and carpets as well as thin concrete slabs with various types of
metal mesh reinforcements. These mock-ups were used to set up parameters and
constraints necessary for the magnetic-field-based localization and guidance case
3. Methodology
In order to correlate the magnetic field to the material interference, base measure-
ments of the magnet were performed. Following this, numerous measurements
with different materials and locations of the testing unit were done. Milliteslas
were the units used in the experiments, while the values received from the sensors
mounted on the robot were analog values between 48 and 4048.
To test the magnetic field, a Kanetec TM-601 tesla meter was used, with a resolu-
tion of 0.1 milliteslas. For repeatability across magnets and materials, an X-carve
CNC machine was fitted with a custom holder in the ER-11 collet to maintain the
position of the TM-601 sensor (figure 1). The machine was controlled through the
software to specify exact height distances from the testing materials.
Figure 1. Testing device mounted to the ER-11 Collet of a CNC machine. The bar magnet is
raised and secured via POM plastic.
Implementation of the localization was done using magnets, with a KUKA
youBot and custom magnetic sensor array. Importantly, the robot had mecanum
wheels, allowing for full directionality of movement. An aluminium box contain-
ing 30 AD22151 hall sensors was mounted to the rear of the robot (figure 2). The
sensors were connected to an Arduino board, which connected to the computer
in the robot base, transmitting serial information through USB. Data values repre-
senting the magnetic field were between 48 and 4048, corresponding to south and
north poles, respectively. The base was controlled through a Python 2.7 interface
with ROS, and the sensor was read using pySerial. The program on the robot was
started and stopped over an Ethernet connection. While in practice this would be
a wireless connection, reliability of stopping and ease of programming required a
tether connection in these experiments.
Multiple distances between the sensor and magnet were tested to understand the
falloff of the magnetic field, and whether any materials interfered with this falloff.
Base measurements of the magnet only were done at 0, 5, 10, 20, 30, 40, and
50 mm distances (vertical). Measurements with materials separating the testing
device and the material were done at the same distances, provided the material
height allowed. For example, a tile 8 mm thick would allow for samples 10 mm
and further from the magnet.
Figure 2. The KUKA Youbot used for the experiment. The four wheels allow for omni
directional movement. The magnetic field sensors are mounted in the rear of the robot and
connect to the USB port of the robot.
While many materials do not contain ferro-metallic particles that could affect mag-
netic field/flux, various construction materials were used to validate the insignifi-
cant effects such materials have. Materials used for testing included (1) ceramic
tiles, 100 x 100 mm, 8 mm thick, with cobalt and white glazing; (2) pine wood
planks, 20 mm thick; (3) floor carpet with non-cut loop, 6 mm thick; (4) mar-
ble stone tile, 10 mm thick; (5) vinyl flooring tiles, 3 mm thick; and (6) concrete
mix. The concrete used had a manufacturer statement that the mixture contained
approximately 20% sand, with a recommended ratio of 0.14 ml/g.
As concrete is often poured with rebar or metal mesh, additional material com-
binations were created. A steel reinforcement mesh with a 3 mm diameter in a
100 mm square was used to test top surface reinforcement used in many concrete
slabs, while a 3.5 mm x 6.5 mm diamond mesh of 0.5 mm thickness was used to
test concrete mesh pouring. Last, a concrete slab 18 mm thick (seen in figure 2)
was poured containing magnets at the bottom for use in the centering experiments.
Two tests were performed to demonstrate the ability of the robot to localize based
on magnets. The first test was a sideways movement using the omnidirectional
wheels to locate the robot center to the magnet. Second, magnets were placed at
various distances apart to allow the robot to move forward and stop when a new
magnet was found, centering upon that magnet and then continuing.
4. Results
Two main results are presented. First, the relationship between magnetic fields
and location of the sensor demonstrates the feasibility as well as the difficulty of
integrating magnets in the construction process. Second, the implementation of a
centering robot based on embedded magnets is demonstrated.
As suspected, materials showed no significant variation in interference of the mag-
netic field. While it is possible these materials can interfere with electromagnetic
fields, as seen in some cases of concrete (Shi and Chung 1995), the concrete used
in the present experiments did not show any interference. However, there was sig-
nificant interference of the magnetic field when metal was introduced. The values
measured from the magnet varied depending on the location. Figure 3 shows that
the difference between the edge and center of the bar magnet had significant dif-
ferences when tested close to the surface, but became similar as the measurement
device moved further away.
Figure 3. Base measurements of the bar magnet tested at the edge and center.
When using the wire mesh with a cast of concrete, the measurements showed
a significant decrease depending on the location of the mesh. However, the effect
of measuring 10 mm higher from the magnet caused a larger difference than the
introduction of wire mesh. The wire mesh closest to the magnet showed the most
significant difference in the magnetic field, especially at the center. When a wire
mesh was introduced closest to the surface, a 26% reduction in the measured mag-
netic field occurred, whereby the difference between 20 mm and 30 mm reduced
the magnetic field strength by 52% (figure 4).
Figure 4. Left: Measurements from the center of the bar magnet. Right: Measurements from
the edge of the magnet. Base is the measurement with no concrete or mesh. Bottom represents
the mesh facing closest to the magnet with the concrete above. Middle is the mesh embedded
in the middle of the concrete. Top is the wire mesh above the concrete. All data is based on the
average of 3 separate measurements. .
Further illustrating the results showing that the distance of metal mesh from the
magnet has a significant impact, additional measurements demonstrated the vari-
ous distances and their effects on magnetic strength. In the case of a 5.5 mm gap
between the magnet and mesh at 10 mm, the measured magnetic field decreased to
60% of the original base measurement. Comparatively, placing the mesh directly
on the magnet resulted in a measured field decreasing to 96% of the original base
measurement (figure 5).
Figure 5. Various distances of the mesh from the magnet and the corresponding measured
magnetic field.
Figure 6. Results of the 3mm bar square being placed with the magnet in the center and
magnet under the bar. .
The final testing was done with the 3 mm metal bar. Notably, the spacing used
between bars was wide enough to have no impact when the magnet was centered
(figure 6). However, placement of the bar directly over the magnet did have an
impact on the measured field. As metals often used in construction are magnetic,
special care must be taken to avoid the magnet attaching to the bar.
To clearly demonstrate the efficacy of using magnets for centering the robot, figure
7 shows a magnet being moved, with a real-time calculation of the new magnet
location and the movement required to align to it. The process was repeatable and
reliable, demonstrating the efficacy of the process.
Figure 7. Centering algorithm of the robot being demonstrated. Top Left: Robot is centered
on the magnet embedded in the concrete. Top Right, an additional magnet with higher force
than the embedded one forces the robot to move sideways. Bottom Left and Right: Robot
autonomously centers again in respect to the embedded magnet.
The second test performed was to have the robot autonomously center on a
magnet, after which it was intended to move straight forward until another magnet
was found, and repeat the process. Figure 8 shows the experimental setup in which
magnets were placed at random distances apart. The robot was then placed with
the sensor near the first magnet to begin localization.
Figure 8. Three magnets were placed on the wooden platform various distances apart and
with slight offsets.
The robot was able to successfully center and track until the following magnets
were found. Figure 9 shows an overlay for the video. Notably, the double images
of the robot at similar positions show the ability to center on the unevenly placed
magnets along the path. The largest obstacle in this implementation is the spacing
between magnets. As the distance increases, the error for the center of the robot
increases. When this error is too high, the robot can pass the magnet without
the sensor passing over it. To address this problem, additional attention to the
magnet spacing is required. Another success of this experiment is seen in the free-
standing magnets on the platform. While ideally the magnets are embedded into
the infrastructure, the robot was able to pass over the magnets without the magnet
attaching to the steel base. This demonstrates that the strength of the magnet can
be low enough to have little force on metal objects while still being readable by a
dedicated sensor.
Figure 9. Overlays of the video in which the robot centers on a magnet, continues forward
until another magnet is found, centers again, and continues forward until the third magnet is
found, centering and ending the program.
5. Conclusion
The results presented here provide key guidelines for magnetic line following in-
tegrated into a construction site. Furthermore, the efficacy of a robot relying on
magnets for basic localization, such as centering on a line, has been shown to be
simple and easy to implement. When designing to integrate magnets for this pur-
pose, the most important consideration is the distance between the magnet and the
aboveground sensor. A second consideration is that the peak of the magnetic field
varies across a typical bar magnet when measured at various heights. This is com-
pounded by the introduction of magnetic shielding material, such as metal mesh
or metal bars. The elimination of these materials is the best option, but if they
must be used between the magnet and the floor surface, effort should be taken to
reduce the distance between the magnet and the shielding material, as the amount
of shielding increases when the gap increases.
Furthermore, the strength of the natural magnets greatly outweighs the strength
of the Earth’s electromagnetic fields by several magnitudes. The strength of
the Earth’s electromagnetic field is in the range of 25 to 65 microteslas, while
the smallest measurement we considered in navigation was in tens of milliteslas.
Demonstrating the ability to strongly detect these magnets, the sensor used in this
study was able to detect the magnets lying free under the robot, while the chassis of
the robot was high enough not to attract the magnet onto itself. As other research
has pointed out, various materials and aggregates in concrete play an important
role in electromagnetic shielding. The ability for the system demonstrated in this
paper to work likely hinges on the strength of the natural magnets and the lack of
high-powered electromagnets or high sensitivity to electromagnetic fields. When
dealing with all magnetic fields at a high resolution, the Earth itself will create
interference. However, as shown here, this level of noise is insignificant for this
type of robotic navigation. Finally, the method and algorithm implemented is also
sensitive enough to track a magnet under concrete while simultaneously able to be
overcome by a magnet placed by hand on top of the concrete.
6. Future
The main goal of this paper has been to show that through infrastructure, the pace
at which society can adopt autonomous robotics that can navigate through space
can significantly increase. While magnets themselves are not an end-all solution
to navigation, with little work they have been shown to aid in the centering and
localization of a wheeled robot. This concept could be expanded to electromagnets
as well. In this type of situation, data could be communicated to the robot as well,
through various methods. For example, simple on-off commands similar to Morse
code could be deciphered by the robot reading the magnet. This could provide
navigation information to various robots, or act as an embedded traffic system that
only robots can detect, not interfering with humans within the space. However,
the benefits of using natural earth magnets may be lost, and further research is
The authors would like to thank Jaehwan Kim from Autonomous Vehicle Lab at
the Advanced Institutes of Convergence Technology, Seoul National University
for the development of the magnetic sensor used in this research.
Borenstein, J., Everett, H. R. and Feng, L.: 1996, Where am I? Sensors and methods for mobile
robot positioning, University of Michigan, 27.
Cowan, R.E., Fregly, B.J., Boninger, M.L., Chan, L., Rodgers, M.M. and Reinkensmeyer, D.J.:
2012, Recent trends in assistive technology for mobility, Journal of neuroengineering and
rehabilitation,9-1, 1.
Graham, C. D. and Cullity, B. D.: 2009, Introduction to magnetic materials, John Wiley &
Madarasz, R., Heiny, L., Cromp, R. and Mazur, N.: 1986, The design of an autonomous vehicle
for the disabled., IEEE Journal on Robotics and Automation,2-3, 117-126.
Sumner, T. J., Pendlebury, J. M. and Smith, K.: 1987, Convectional magnetic shielding, Journal
of Physics D: Applied Physics,20-9, 1095.
Sun, C., Ritchie, S. G. and Oh, S.: 2003, Inductive Classifying Artificial Network for Vehicle
Type Categorization, Computer-Aided Civil and Infrastructure Engineering,18, 161–172..
Zheng-Qiang, S. and Chung, D. D. L.: 1995, Concrete for magnetic shielding, Cement and
concrete Research,25-5 , 939-944.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Loss of physical mobility makes maximal participation in desired activities more difficult and in the worst case fully prevents participation. This paper surveys recent work in assistive technology to improve mobility for persons with a disability, drawing on examples observed during a tour of academic and industrial research sites in Europe. The underlying theme of this recent work is a more seamless integration of the capabilities of the user and the assistive technology. This improved integration spans diverse technologies, including powered wheelchairs, prosthetic limbs, functional electrical stimulation, and wearable exoskeletons. Improved integration is being accomplished in three ways: 1) improving the assistive technology mechanics; 2) improving the user-technology physical interface; and 3) sharing of control between the user and the technology. We provide an overview of these improvements in user-technology integration and discuss whether such improvements have the potential to be transformative for people with mobility impairments.
The use of steel paper clips in the amount of 5 vol.% in concrete was found to result in concrete with magnetic shielding (60 Hz) ability comparable to steel mesh. Furthermore, the paper clips did not affect the compressive strength of the concrete. The paper clips were discontinuous enough for mixing with the concrete mix, while their intertwining tendency allowed sufficient continuity for magnetic shielding.
A general review of conventional magnetic shielding is presented. Several possible definitions of the term 'shielding factor' are given, taking into account the ferromagnetic nature of shielding materials. An alternative treatment of single-layer axial shielding for a cylindrical shield is presented and compared to a previous approximation. A new algorithm for calculating multi-layer axial shielding factors is derived and shown to be an improvement on existing methods when applied to some assemblies, including one reported for the first time.
As transportation surveillance technology continues to advance, the measurement of more complete traffic information is becoming increasingly feasible. ICAN stands for Inductive Classifying Artificial Network and is used to conveniently describe a self–organizing feature map (SOFM) for vehicle type categorization using inductive signatures as input. Vehicle type categorization is the separation of vehicles into predefined classes and can be useful for improving transportation efficiency, cost, environmental sustainability, enforcement, safety, and education. ICAN mainly focuses on the challenging task of differentiating between two–axle vehicles such as passenger car, sports utility vehicle (SUV), van, truck, and bus. This is in contrast to systems that classify according to the number of axles. One characteristic of ICAN is the simplicity of the 13–neuron one–dimensional neural network, and the employment of a small training set of 13 signatures. The overall classification results of 87% (data set 1) and 82% (data set 2) for seven categories coupled with consistent performance across all vehicle categories was significant and encouraging. Field freeway data was used for testing ICAN; representative signatures and video images of different vehicle types are presented.
The first steps in the design and construction of an autonomous vehicle for the physically and sensory disabled are described. The vehicle is basically a self-navigating wheelchair, which is designed to transport a person to a desired room within an office building given only the destination room number. The wheelchair has been equipped with an on-board microcomputer, a digital camera, and a scanning ultrasonic rangefinder. The goal of the project is to develop a vehicle that can operate without human intervention within a crowded environment with little or no impact on the building or its inhabitants. An overview of the project objectives and a detailed description of the components that are currently working are presented.