Development of a Curling Stone Tracking System Using Infrared LEDs,

and an Accompanying Application

Yoshinari Takegawa

1 a

, Noa Sasaki

, Shimpei Aihara

2 b

and Fumito Masui

2 c

School of Systems Information Science, Future University Hakodate, Hokkaido, Japan

Department of Sport Science and Research, Japan Institute of Sport Sciences, Tokyo, Japan

Keywords:

Curling, Position Measurement, Infrared LED, Image Processing, Measurement Software.

Abstract:

The purpose of this study is the design and implementation of a real-time position measurement system for

use on curling stones. Often called ‘chess on ice’, curling is a sport that requires a high level of strategy.

Accordingly, how the stones move around the vast 40m curling rink constitutes important data. However,

in the unique environment of the icy and vast rink, it is difﬁcult to monitor the position of the stones by a

simple method without hindering the players. Therefore, in our research, we proposed a system using infrared

LEDs and an infrared camera. Infrared LED modules are installed on the stones and the rink, and infrared

cameras installed around the edge of the rink ﬁlm the LED modules and perform calibration. Then, using

four coordinates of the LED modules on the rink, the system employs perspective transformation technology,

which is a type of image processing. In so doing, it is possible both to measure the position of the stones, and

solve problems. Through experiments, performance evaluation was conducted to asses what degree of error

occurs in position measurement when the proposed system is used. Experiments were conducted on a curling

rink. The average error was 0.189m in the experiment at the curling rink.

1 BACKGROUND

Curling is a sport in which players slide stones on

ice, aiming to achieve a higher score than their op-

ponents. Its high level of strategy and skill has led

curling to be called ‘chess on ice’. However, to form

a strategy, it is necessary to consider various factors,

such as the current positions of the stones, the con-

dition of the ice, and the state of play, while skill is

dependent upon a player’s experience and intuition.

Furthermore, there is, even now, no established the-

ory regarding what causes the stones to curve as they

travel (Murata, 2022). Due to the complex and ad-

vanced element of strategy and the fact that technique

is dependent on players, there are few scientiﬁc ap-

proaches to curling, in comparison with other sports.

Nevertheless, Masui et al. have started research ti-

tled ‘Curling Science’ which is an initiative to cre-

ate new strategy support that integrates information

https://orcid.org/0000-0003-1947-0021

https://orcid.org/0000-0002-8513-0204

https://orcid.org/0000-0001-9979-8734

technology’

. This project involves research on dig-

ital curling (Ito and Kitasei, 2015), tactical analysis,

measurement of stone behavior, and sweeping (Gwon

et al., 2020; Won et al., 2018).

Digital curling refers to the proposal of a virtual

curling space, created using a computer’s physical

simulator, that acts as a space to enable discussion of

curling strategy. This concept has been developed by

a large number of people (Yamamoto et al., 2015).

In addition, systems, such as the Portable Tactical

Support DB System, have been developed to record

shots, stone layout, match scores, players taking part,

and the condition of the ice, on a tablet device (Masui

et al., 2015; Masui et al., 2016; Otani et al., 2016).

These works are examples of tactical support car-

ried out on a computer, but we consider it necessary

to apply the technologies to actual play and provide

support in real time. It is thought that grasping the po-

sitions of the stones is important as one factor towards

achieving this. The points that must be considered in

actual measurement of stone position are as follows:

• Measurement on ice

https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT

-15H02797/

136

Takegawa, Y., Sasaki, N., Aihara, S. and Masui, F.

Development of a Curling Stone Tracking System Using Infrared LEDs, and an Accompanying Application.

DOI: 10.5220/0012182600003587

In Proceedings of the 11th International Conference on Sport Sciences Research and Technology Support (icSPORTS 2023), pages 136-143

ISBN: 978-989-758-673-6; ISSN: 2184-3201

As curling is played on ice, room temperature is be-

low 10 degrees, and the curling rink maintains a tem-

perature of -5 degrees Celsius. Accordingly, the sys-

tem must be able to operate in this environment. Also,

it must be durable enough to run not only a few times,

but throughout the entire curling season.

• Not hindering players

As curling is played on a 40m-long rink, it is possi-

ble that installing a system on the rink will hinder the

players.

• Highly accurate position measurement

In curling, the stone closest to the centre of the house

obtains points, which is why stone position informa-

tion is important. The occurrence of a large calcula-

tion error would affect the outcome of a game. There-

fore, we require a system that can measure positions

across the entire 40m rink, without any large calcula-

tion errors.

In the peculiar environment of the icy and wide

rink, it is difﬁcult to monitor the positions of the

stones by a simple method without hindering the play-

ers. To the extent of our knowledge, there is no ex-

isting stone-tracking system that meets the aforemen-

tioned requirements.

Therefore, in this research, we aim to develop a

curling-stone tracking system. In the proposed sys-

tem, where a hindrance to players is resolved by

mounting infrared LEDs on the stones and having an

infrared camera installed at the edge of the rink per-

forming image processing. In addition, we develop

applications that utilize the stone position information

measured by the stone tracking system. This applica-

tion provides a graphic presentation of the trajectories

of the stones, by projection mapping on the surface of

the rink.

2 RELATED RESEARCH

Many studies and methods have been implemented to

measure the position of people or things. There is

a method that uses Lidar, created by Velodyne Lidar

Inc., to measure position information in real time

Lidar is capable of measurement with an error of ap-

proximately 3cm. However, as Lidar measures on a

horizontal, measurement cannot be performed if there

is any obstacle between a Lidar sensor and the item to

be measured. For this reason, we consider it difﬁ-

cult to use Lidar for position measurement in curling,

which uses multiple stones. Lidar can recognize the

https://en.wikipedia.org/wiki/Lidar

Figure 1: Measurement Conditions.

shape of a stone if it is nearby, but fails when a stone

is further away.

There has been research on measuring the position

and angular velocity of curling stones, using video

and camera images (Hattori et al., 2023). With an

error of 2mm, the system is highly accurate. How-

ever, because the system is focused on accuracy and

aims to measure curl ratio, i.e., the way the stones curl

over a short distance, the photographic range is lim-

ited to approximately 4.5m. This differs from our re-

search, which aims for simple and wide-ranging mea-

surement. There is further research involving mea-

surement by cameras, though not targeted at curling

stones. In the research by Aoshima, measurement is

accurate to within 1.0mm, but measurement range is

limited to 1-4m (Tao et al., 2017). Furthermore, it is

necessary to prepare several cameras and install them

all facing the same direction. In contrast, our research

aims for a simple measurement method that covers the

wide area of the curling rink. There are also methods

that use Bluetooth or wireless LAN (Sawada et al.,

2016)(Tsuda et al., 2013). These methods are capable

of measuring across a range of approximately 100m,

however, measurement is only accurate to within sev-

eral meters. This makes it difﬁcult to apply such

methods to curling, as an error of a few meters is too

large for a sport in which a few millimeters can deﬁne

a difference between winning and losing.

3 MEASUREMENT METHOD

3.1 System Installation

Figure 1 shows the conditions at the time of curling

stone position measurement. In this system, an LED

module is installed on top of each curling stone, and

LED modules are also installed under the ice of the

curling rink, with a module installed every 1m along

the length of the rink, up to 10m, and a module in-

stalled every 1m across the width of the rink, up to

3m, making a total of 30 modules. Also, an infrared

camera is installed at the edge of the rink.

Development of a Curling Stone Tracking System Using Infrared LEDs, and an Accompanying Application

137

Figure 2: Modules Installed on the Stone.

Figure 3: Flow up to Lighting Up/Turning off.

3.1.1 Stones

Figure 2 presents the LED module and esp8266 mod-

ule that are installed on the top of a curling stone.

To light up the LED module on the curling stone, an

esp8266 module, capable of Wi-Fi connection, was

used. This enables remote adjustment of the bright-

ness of the LED module. In addition, the esp8266

module is powered by a lithium ion battery, meaning

that it can be controlled wirelessly and so will not stop

working during play. Next, the ﬂow up to the lighting

of the LED module is presented in Figure 3. First,

esp8266 module 1 is wired to the computer and a

value between 0 and 255, representing the light inten-

sity of the LED module, is uploaded by serial commu-

nication. Then, the value received by esp8266 module

1 is sent to the data store of the Milkcocoa server. The

value in the data store is received by esp8266 module

2, then sent to the LED module, causing it to light up

or turn off. The LED module turns off when the value

is 0, and lights up when the value is anything other

than 0.

3.2 Flow of Position Measurement

• Calibration

The coordinates of the LED module, which are nec-

essary for position measurement, are recorded. Then

calibration is performed, by the following process.

1. Designate the range as the four corners of the rink

at which LED modules are installed.

2. Light up the LED modules within the range, one

by one.

3. Adjust the light intensity until the surface area of

LED module light detected by the infrared camera

is 10 pixels (+-3).

4. Extract the outline of the LED module light and

record the coordinates at the centre of the outline.

As long as the infrared camera is not moved, calibra-

tion needs only be performed once. Also, LED mod-

ule coordinate measurement at the time of calibration

is carried out automatically once implemented. Thus,

LED modules are lit up in consecutive order from

no.1, and the program automatically measures the po-

sition, in meters, at which the modules are present on

the rink.

• Stone Sliding

Position measurement is made possible by using the

values obtained in calibration along with perspective

transformation technology, which is a type of image

processing. The ﬂow of the stone position measure-

ment is as follows:

1. Slide a stone onto which a LED module is

mounted

2. Calculate four coordinate points (hereafter re-

ferred to as reference points) froom the nearest

LEDs surrounding the stone’s LED module

3. Using the four reference points, carry out perspec-

tive transformation

3.2.1 Perspective Transformation

Perspective transformation is an image processing

technique for depicting a three-dimensional object

position on a two-dimensional plane. In the layout

of the proposed system, the infrared camera is tilted

slightly downward to ﬁlm the LED modules on the

rink. However, as the camera image reﬂects a three-

dimensional space, distances of the same length ap-

pear different in the foreground to the background.

Therefore, perspective transformation is used to cor-

rect the image to a two-dimensional plane. During

this process, ﬁrst the image position of LED mod-

ule on top of the stone is extracted in pixels. Next,

a perspective transformation series is calculated from

the surrounding four reference points obtained in cal-

ibration. By this process, the image is transformed

to a two-dimensional plane, and stone position can be

measured.

icSPORTS 2023 - 11th International Conference on Sport Sciences Research and Technology Support

138

Figure 4: Program Execution Screen of Position Measure-

ment Application.

3.2.2 Position Measurement Application

Figure 4 shows the program execution screen of the

position measurement application. The application

has three modes, and stone position measurement is

made possible by implementing the three modes in

order.

1. Rink mode: In this mode, the range of calibration

is decided. This is done by setting red circles at

the positions corresponding to the corners of the

rink. The reason it is necessary to set the range

is that, because curling is an indoor sport, some-

times indoor lighting or light reﬂection off walls

can prevent calibration from being carried out suc-

cessfully.

2. Reference mode: In this mode, the infrared LED

modules within the range set in Rink mode are

calibrated. Calibration comprises lighting up the

LED modules one by one and recording the x and

y coordinates and light intensity of each module.

3. Position mode: This mode records the position of

a stone that has been thrown.

Next, we explain the program execution screen.

‘A’ in Figure 4 presents and records the x coordinate,

y coordinate and light quantity of each LED module,

obtained in calibration. The numbers from 0 to 5 on

the vertical axis and 1 to 20 on the horizontal axis rep-

resent the positions of the LED modules installed on

the rink, and correspond to 1m, 2m, ... 20m. The posi-

tions of LED modules for which current coordinates

were obtained during calibration are represented by

purple squares, the positions of which can be altered

manually.

‘B’ in Figure 4 represents the position information

of a stone thrown when using Position mode. Here,

the distance in the x and y directions, and the pixel

position, are expressed.

‘C’ in Figure 4 presents the keys used to oper-

ate the program, as well as the mode status and LED

module light intensity. From left to right, the ﬁgure

presents the keys used, a simple explanation of key

operation, and the current mode and LED light inten-

sity in red letters. The functions assigned to each key

are described in detail below.

• ‘Space’ key This key is used to switch between

’Start’ and ’stop’ in the program. The program is

in the ’Stop’ state at the time of start-up, at which

time pressing the key once switches to ’Start’ sta-

tus and the infrared camera starts up. Pressing

the key once more switches back to ’Stop’ status,

causing the camera to stop recording.

• ‘Enter’ key This key is used to switch between

Rink mode, Reference mode, and Position mode.

The program is in Rink mode at the time of start-

up. In this state, pressing the enter key once tran-

sitions to Reference mode, and pressing the key

once more transitions to Position mode. Pressing

the key once more transitions to Rink mode again.

• ‘g’ key This key can only be used in Reference

mode. Pressing the key initiates calibration, caus-

ing the LED modules to light up in order from the

module closest to the connection point, and the

x coordinate, y coordinate, and light quantity of

each module to be recorded automatically.

• ‘j’ key, ’k’ key These keys can only be used in

Reference mode. Pressing the ’j’ key increases

the light intensity of the illuminated LED mod-

ule. Pressing the ’k’ key decreases the light inten-

sity. Light intensity can be adjusted within a range

from 0 to 255.

• ‘z’ key This key can only be used in Reference

mode. Pressing the key deletes the coordinates

and light intensity data of the currently selected

LED module.

• ‘d’ key This key can only be used in Reference

mode. After calibration has ended, pressing this

key saves the coordinates and light intensities ob-

tained in calibration.

3.3 Implementation

The program used when calibrating and measuring

stones employs the following devices. The PC used

was a Lenovo ThinkPad (CPU Intel(R) Core(TM) i7-

4650U 1.70GHz); software development on the PC

was carried out using Visual C++ and OpenCV library

and openFrameworks library on Windows 10. To ﬁlm

the curling stones on the wide rink, from diagonally

above, an infrared camera (DMK23UX236) was in-

stalled on top of a tripod with a maximum height

of 3.6m (ManFrott SKU1004BAC), and an IR ﬁlter

(NEEWER IR950) was attached to the infrared cam-

era to block visible light.

Development of a Curling Stone Tracking System Using Infrared LEDs, and an Accompanying Application

139

Figure 5: Infrared LED Module Circuit Diagram.

Figure 6: Infrared LED Module Circuit Board.

3.4 LED Module

3.4.1 Speciﬁcation

Figure 5 shows the circuit diagram of the propsed

LED module. The parts used in the circuit board con-

sist of an infrared LED, resistor, capacitor, transistor

and micro-processor. Next, Figure 6 shows the actual

circuit board, with infrared LED attached, that was

installed on the rink and the stones. The V, GND,

In, and Out parts of circuit boards are connected to

control multiple substrates. A circuit board itself, in-

cluding its component parts, has a length of 2.4mm,

width of 3.7mm, and height of 0.9mm. The red circle

in the centre of the ﬁgure is the infrared LED.

3.4.2 Operation in Curling

In this research, we embedded infrared LED modules

in the ice of a curling rink. The reason for this is that

installing the modules on the surface of the rink would

prove a hindrance to players, and modules would have

to be re-installed every time the rink was used. How-

ever, there is still the problem that the LED modules

are exposed, besides which it is uncertain whether

they can operate for a long time within the ice. To

resolve these issues, the LED modules were coated

with crystal resin. First, the circuit board is inserted

into an 8mm plastic case and afﬁxed with a glue gun.

Then, resin is poured into the case and left to harden

for 24 hours.

Figure 7: Distribution of LEDs Embedded in the Rink.

Figure 8: Arrangement of the System in the Curling Hall.

4 EXPERIMENTS

The aim of this experiment was to verify the accuracy

of stone position measurement in the case of having

altered certain conditions (distance between camera

and LED lines, camera angle).

Experiment Environment. The experiment took

place at the Kawanishi Construction Curling Hall, in

Kitami, Hokkaido.

In addition, Figure 7 presents the LED lines laid

out on the rink before it was covered with ice. 10m

worth of LED modules were installed at 1m intervals

along the length of the rink, and 3m worth at 1m in-

tervals across the width of the rink.

The infrared camera was installed 5.5m in front of

the edge of the rink. This is because the construction

design of the rink makes it impossible to position the

camera any further away, and the camera cannot de-

tect all the LED modules if it is positioned any closer

to the rink. The arrangement of the system in the curl-

ing hall is shown in Figure 8.

Experiment Method. Regarding the procedure, af-

ter ﬁlming with the infrared camera and performing

calibration, an LED module representing a curling

stone is placed at the point 1m on the length and 1m

on the width of the rink, and position measurement is

begun. Then, the measured value of the position on

the screen is obtained. Once this value is obtained,

the LED module is shifted 0.25m lengthwise, while

icSPORTS 2023 - 11th International Conference on Sport Sciences Research and Technology Support

140

Figure 9: Measurement Points on the Curling Rink.

maintaining a position of 1m horizontally. This is re-

peated until the module reaches the 10m point. After

reaching a length of 10m, the module is shifted 0.25 at

a time along the width of the rink, until it reaches the

2m point. Then, the module is returned to the point

1m along the length of the rink and the whole process

is repeated until the module reaches the 10m point.

As shown in Figure 9, we measure 37 points on the

length of the rink and 9 points on the width of the

rink, making a total of 333 points measured.

Evaluation Indices. Once position measurement

was complete, we compared the original values and

the obtained values. We calculated the position mea-

surement error in the case of having reduced the ref-

erence points as shown in Figure 10.

Results. The average error was 0.189m. The max-

imum error was 0.63m, at the point at length 9.75m

and width 2.75m. There was no problem regarding

the operation of the LED modules embedded in the

ice.

Next, we describe the the results when reference

points are reduced. The transition of the average error

in the case of reduced reference points is presented in

Figure11.

Figure 10: Measurement Range when Reference Points are

Reduced (On Curling Rink).

Figure 11: Transition of Average Error When Reference

Points are Reduced (Curling Rink).

Consideration. Regarding position measurement

accuracy, in the same manner as the preliminary ex-

periment, the further away the measurement point

was, the larger the measurement error became. This

is because, at a distant point, measurement error be-

comes large if detection of the LED support is out by

even one pixel.

Overall, the error was larger than in the prelim-

inary experiment. This was because refraction was

caused by the ice of the curling rink, causing the op-

tical axis of the LED module to shift forward. At ev-

ery point on the rink, it was conﬁrmed that even if

the stone LED was actually placed in the correct po-

sition, the measurement result was further away. As a

revision method to compensate for for refraction, we

reduced all the length measurement results in the esti-

mated values by 0.100. In this case, the average error

became 0.129m.

Even inside the ice, the LED modules within the

LED line operated successfully. This is because the

modules were protected from moisture and cold by a

Development of a Curling Stone Tracking System Using Infrared LEDs, and an Accompanying Application

141

Figure 12: System Structure.

resin coating.

When the reference points are reduced, the four

reference points surrounding the stone LED, which

are used in perspective transformation, become fur-

ther away. As a result, the inﬂuence of LED sup-

port detection error at these reference points becomes

greater.

5 APPLICATION

As mentioned in Section 1, this research is targeted at

curling beginners, children, spectators, etc., and aims

to enable people to enjoy curling, and learn all its

component techniques. The target users are beginners

with no special knowledge or skill related to curling,

and curling spectators. The games developed in the

Curling Projection Mapping research include a game

played on the rink and a game that uses a smartphone

application.

5.1 Curling Projection Mapping

The system structure of Curling Projection Mapping

is presented in Figure 12. Infrared LEDs are attached

to the curling stones, and the trajectory of those stones

is recognized by multiple infrared cameras installed

on the ceiling. The corresponding position informa-

tion is estimated by a computer, which generates an

image. Then, the generated image is projected on the

rink in real time by multiple projectors.

In Curling Projection Mapping, there are a total of

three games that we have developed: ‘MiracleFlower’

and ‘CoinCollector’, which are games projected onto

the rink, and ‘StoneSpeedChecker’, which is a tablet

application. Each type of game is described below.

Here we explain about ‘MiracleFlower’ and ‘Co-

inCollector’, which are the games involving projec-

tion on the rink.

MiracleFlower. As shown in Figure 13, ‘Miracle-

Flower’ is a simple game in which ﬂowers bloom

following the trajectory of a curling stone. Players

are free to slide the stones however they wish, with-

out considering direction, velocity, and so on, which

Figure 13: Actuar usage of ‘MiracleFlower’.

Figure 14: Actuar usage of ‘CoinCollector’.

means that even beginners or people with no curling

knowledge can enjoy the game. In addition, specta-

tors can also enjoy watching the graphics of Miracle-

Flower.

CoinCollector. As shown in Figure 14, ‘CoinCol-

lector’ is a game in which players aim to slide stones

onto coins scattered across the rink, with the winner

being the player who collects the most points. With

its concept of aiming for as many targets as possible

in a single turn, this is a novel game that does not

conform to existing curling rules, although it enables

players to experience the thrill of competing.

5.2 Tablet Application

Here we introduce our tablet application, ‘Stone-

SpeedChecker’. In StoneSpeedChecker, as shown in

Figure 15, a tablet is installed on a stone and the cur-

rent velocity of the stone is presented in real time,

based on the position information generated by the

tracking system. The aim of this application is for the

sweepers to get an accurate grasp of the speed of the

stone. The sweepers announce to their team mem-

bers the ten-zone value that expresses where on the

house the stone will stop. Accordingly, it is essen-

tial for sweepers to grasp the speed of the currently

sliding stone in real time. The speed display meter

is equipped with a function that rotates the meter in

response to the rotation of the stone, to enable the

sweepers to read the speed smoothly even when the

stone rotates. The angle of rotation is estimated based

on the tablet’s internal geomagnetic sensor. In addi-

icSPORTS 2023 - 11th International Conference on Sport Sciences Research and Technology Support

142

Figure 15: Actuar usage of ‘StoneSpeedChecker’.

tion to aiding practice, this application can be used to

conﬁrm the state of the stones and rink in night prac-

tice.

6 CONCLUSION

In this research, we developed an image-processing

based real-time curling stone tracking system. AS a

result of analyzing the peculiarities of the curling rink

and the game itself, we adopted an image-processing

based measurement method by infrared LED and in-

frared camera. In addition, we actually implemented

the system and, after conducting a preliminary exper-

iment, involving altering the infrared camera position,

inside the university, we conducted an experiment in

the real environment of the Kawanishi Construction

Curling Hall in Kitami. The average measurement

error was 0.189m in the curling venue experiment,

demonstrating that it was possible to measure with a

high degree of accuracy while using a single camera.

Furthermore, as applications that utilize our tracking

system, we developed two kinds of game incorporat-

ing projection mapping on the rink.

ACKNOWLEDGEMENTS

This work was supported by the “Functional Develop-

ment Project for Resilient Athlete Support” of Japan

Sports Agency.

REFERENCES

Gwon, J., Kim, H., Bae, H., and Lee, S. (2020). Path plan-

ning of a sweeping robot based on path estimation of

a curling stone using sensor fusion. Electronics, 9(3).

Hattori, K., Tokumoto, M., Kashiwazaki, K., and Maeno,

N. (2023). High-precision measurements of curling

stone dynamics: Curl distance by digital image analy-

sis. Proceedings of the Institution of Mechanical En-

gineers, Part P: Journal of Sports Engineering and

Technology, 237(2):102–108.

Ito, T. and Kitasei, Y. (2015). Proposal and implementa-

tion of ”digital curling”. In 2015 IEEE Conference on

Computational Intelligence and Games (CIG), pages

469–473.

Masui, F., Hirata, K., Otani, H., Yanagi, H., and Ptaszynski,

M. (2016). Informatics to support tactics and strate-

gies in curling. International Journal of Automation

Technology, 10(2):244–252.

Masui, F., Ueno, H., Yanagi, H., and Ptaszynski, M. (2015).

Toward curling informatics — digital scorebook de-

velopment and game information analysis. In 2015

IEEE Conference on Computational Intelligence and

Games (CIG), pages 481–488.

Murata, J. (2022). Study of curling mechanism by preci-

sion kinematic measurements of curling stone’s mo-

tion. Scientiﬁc Reports, 12(1).

Otani, H., Masui, F., Hirata, K., Yanagi, H., and Ptaszyn-

ski, M. (2016). Analysis of curling team strategy

and tactics using curling informatics. In Correia,

P. P. and Cabri, J., editors, icSPORTS, pages 182–187.

SciTePress.

Sawada, K., Hanada, Y., and Mori, S. (2016). User-

installable indoor positioning system using a wi-ﬁ

beacon and pdr module. Journal of Information Pro-

cessing, 24(6):843–852.

Tao, T., Chen, Q., Feng, S., Hu, Y., Da, J., and Zuo, C.

(2017). High-precision real-time 3d shape measure-

ment using a bi-frequency scheme and multi-view sys-

tem. Appl. Opt., 56(13):3646–3653.

Tsuda, Y., Kong, Q., and Maekawa, T. (2013). Detecting

and correcting wiﬁ positioning errors. In Proceedings

of the 2013 ACM International Joint Conference on

Pervasive and Ubiquitous Computing, UbiComp ’13,

page 777–786, New York, NY, USA. Association for

Computing Machinery.

Won, D.-O., Kim, B.-D., Kim, H.-J., Eom, T.-S., Muller,

K.-R., and Lee, S.-W. (2018). Curly: An ai-based

curling robot successfully competing in the olympic

discipline of curling. In Proceedings of the Twenty-

Seventh International Joint Conference on Artiﬁcial

Intelligence, IJCAI-18, pages 5883–5885. Interna-

tional Joint Conferences on Artiﬁcial Intelligence Or-

ganization.

Yamamoto, M., Kato, S., and Iizuka, H. (2015). Digital

curling strategy based on game tree search. In 2015

IEEE Conference on Computational Intelligence and

Games (CIG), pages 474–480.

Development of a Curling Stone Tracking System Using Infrared LEDs, and an Accompanying Application

143