Revolutionizing Firefighting Training with Digital Twins: Real-Time Fire Spread Simulation and Escape Route Optimization

International Journal of Computational Science and Information Technology (IJCSITY) Vol. 12, No.
1/2/3/4, November 2024
DOI : 10.5121/ijcsity.2024.12401 1
REVOLUTIONIZING FIREFIGHTING TRAINING
WITH DIGITAL TWINS: REAL-TIME FIRE
SPREAD SIMULATION AND
ESCAPE ROUTE OPTIMIZATION
Ying Zhang1
and Boon Giin Lee2
1
Faculty of Humanities and Social Science, University of Nottingham Ningbo China,
Ningbo, China
2
School of Computer Science, University of Nottingham Ningbo China, Ningbo, China
ABSTRACT
The advancement of digital twin (DT) systems had revolutionized various industrial and safety applications,
offering virtual replicas of physical processes for improved monitoring and training. Fire scenarios are
highly dynamic, with conditions changing rapidly due to factors like wind direction, material flammability,
and structural integrity. This study explored the application of a smart fire scene DT system (SFSDTS) for
firefighting training and safety management. The novelty and contributions of this study was the proposed
SFSDTS integrated machine learning models to predict fire spreading paths and estimate escape routes in
real-time, providing an immersive and interactive training environment for firefighters. Various ML
models, including random forest (RF), XGBoost, decision tree (DR), logistic regression (LR), and K-
nearest neighbors (KNN), were utilized for predicting fire spreading. The performance of these models was
evaluated using metrics such as accuracy, precision, recall, and F1 score, with XGBoost and RF models
demonstrating superior performance. The proposed SFSDTS also employed the A* algorithm for optimized
escape route estimation based on dynamic fire conditions. User experience was assessed through a
standardized questionnaire, user experience questionnaire (UEQ), revealing positive ratings for the
proposed SFSDTS’s efficiency, stimulation, and perspicuity. Compared to commercial products, the
proposed SFSDTS offered improved accessibility and real-time simulation functionalities. The study
highlighted the potential of proposed SFSDTS in transforming firefighting training and safety management.
KEYWORDS
Digital Twin, Firefighting, Machine Learning, Training Simulation
1. INTRODUCTION
With the advancement of computational simulations integrated with real-world scenarios, the
concept of digital twin (DT) systems became a significant research topic, providing virtual
replicas for physical processes. A DT served as a virtual duplicate of a physical object,
continually simulating real-world conditions. It was frequently employed for monitoring, design,
optimization, maintenance, and remote access in industrial production [1]. While initial research
on the integration of Industry 4.0 and DT technologies concentrated on the connection of
manufacturing machines and systems [2], DTs were also applied to enhance the safety
management of workers in various fields [3]. This was particularly crucial in hazardous
environments such as chemical plants, manufacturing industries, and fire scenes. In firefighting,
which involved different environmental factors and harsh conditions, the DT concept was further

1/2/3/4, November 2024
2
expanded, which included interconnected networks of complex environmental situations and
human conditions [5].
The application of DT technology held significance in the digital transformation of firefighting.
DT technology provided firefighters with an immersive and interactive training environment,
eliminating the risks associated with traditional real fire training exercises [5]. By simulating
realistic building and fire scenarios, DTs were used to train firefighters, helping them familiarize
themselves with various fire scenarios and learn optimal response strategies and operational
techniques [6]. Diverse fire scenarios could be simulated, and conditions of the fire scenes could
be remodeled in different ways to challenge trainees and enhance their problem-solving skills.
The advantages of DTs in fire training included three key aspects [7]. Firstly, they provided a
safe platform for trainees to experience and respond to high-risk scenarios without the dangers of
burns, smoke inhalation, and structural collapse. Additionally, the training could be personalized,
allowing instructors to tailor scenarios to address specific learning objectives and trainees’
weaknesses. Furthermore, the integration of real-time scene data and analytics allowed for
immediate feedback, enabling trainees to review their actions and learn the best strategies to
respond to similar situations in actual fire scenarios.
In terms of building safety and security, DTs played a crucial role in advanced obstacles
detection, precise localization, fire condition analysis, and real-time fire spreading modeling [8].
Additionally, DT facilitated safety management through building information modeling and
machine learning algorithms. These technologies enabled real-time data collection and analysis
of dynamic safety information and building data, thereby enhancing firefighter safety during
emergencies [9]. Historical fire scene data could also be integrated into the DT, allowing it to
utilize artificial intelligence (AI) models to analyze this data in real-time which enabled instant
predictions and improved firefighter training [10].
The existing fire scenario simulation and prediction models required significant computational
time, ranging from days to weeks, to simulate fire scenarios [11]. Various methods had been
proposed by existing studies to overcome this limitation through the use of AI models [12]. For
instance, Verlekar et al. [13] utilized a Convolutional Long Short-Term Memory neural network
(Conv-LSTM) to link spatio-temporal temperature distributions with the number, size, and
location of fire sources. Similarly, Wang et al. [14] demonstrated the effectiveness of using
Conv-LSTM in a DT during a full-scale fire test chamber experiment. Additionally, Lei et al. [15]
improved the fire prediction effectiveness at the China Palace Museum by utilizing the XGBoost
model to categorize buildings into different risk levels based on the information of building
materials and environmental factors (temperature, humidity, etc.). Zhang et al. [17] proposed a
Transformer network to predict fires in tunnels, which also provided a 3D visual representation of
the fire scene, aiding in firefighting operations, evacuation, and training exercises.
On the other hand, numerical fire modeling was implemented through a computational fluid
dynamics approach [11]. Crucial steps included mesh generation, capturing combustion physics,
turbulence modeling, and heat transmission between solid obstacles. The model addressed
ignition, fuel combustion, and conservation of mass, focusing on accurate fire dynamics
simulation. To ensure accuracy, the model was validated against experimental data and
continuously modified to improve predictive capability. However, numerical modeling required a
long duration to simulate sample scenes and lacked real-time fire predictions in firefighting
practice. SmokeView [17] served as the visualization companion for the fire dynamics simulator
(FDS), displaying simulation results where the virtual smoke, fire, and heat spread were
presented in a 3D visual representation, enabling researchers and firefighters to visually analyze
and interpret the data generated by FDS.

1/2/3/4, November 2024
3
Meanwhile, fire prediction models utilizing datasets from sensors have gained increasing
significance since the 2010s. For instance, Han et al. [18] proposed the FireGrid system, which
quickly and accurately forecasted fire spread using real-time sensor data. Optimization methods
were employed to enhance FireGrid’s computational efficiency, enabling better adaptation to
changing fire scenarios. Additionally, various optimization strategies, such as inverse
computational fluid dynamics predictions and regional modeling approaches, were utilized to
decrease the computational response time to sudden changes in fire scenes and simplify fire
models [19].
One of the research gaps in the field of firefighting DT technology was the lack of creation of
interior layout information [20]. This project aimed to design a customizable scenario that
considered various building layouts, fire intensity, obstacles in different locations, and different
building floors. Designing and implementing such scenario allowed firefighters to gain training
and experience in different fire scenarios. The contributions of this study included designing and
implementing a machine learning (ML)-based fire and smoke prediction model that were
customizable in real-time to simulate dynamic fire changes in a DT model. Additionally, the
study proposed an AI-based fire escape route recommended model, predicting escape routes
based on current fire and smoke conditions.
2. METHODS
2.1. System Overview
Figure 1. Overview design of the proposed SFSDTS
Figure 1 depicted the overview design of the proposed a smart fire scene DT system (SFSDTS)
which consisted of several elements. For demonstration purpose, the SFSDTS started in a
building without fire. The fire was started after 5 minutes. Then, the spreading of fire was
initially simulated through a machine learning (ML) model, analyzed based on the environmental
condition. Then, the evacuation guidance module is triggered, utilizing different AI model to
provide an estimated escape route. The estimated escape route was constantly updated, depending
on the fire spreading condition. Here, three data values, including temperate, wind level, and fire
spread index (see Figure 2), can be adjusted manually in real-time by the users to simulate
unexpected events, e.g., rising of temperature in short period due to explosion. Lastly, the
realistic 3D fire scene was visualized with the predicted escape route, along with virtual flames
and smokes. Similarly, the scene was updated dynamically.

1/2/3/4, November 2024
4
The fire scene had been modeled on a three-floor building in an urban area, as illustrated in
Figure 3a. The interior layouts of each floor were depicted in Figures 3b and 3c. Note that the
modeled building had no elevators, but with staircases connected each floor, situated on both the
left and right sides, as shown in Figure 3d. To simulate realistic fire scenes, fire and smoke
particle effects were employed. Three distinct fire effects, representing large, medium, and small
fire intensities, were designed, as depicted in Figure 4. These variations in fire effects enabled
users to identify and assess the current risks associated with the fire situations.
Figure 2. Illustration of a user interface that allows users to manually adjust the data for stimulating
changes in environment, including temperature, wind level and fire spread index
(a) (b)
(c) (d)
Figure 3. Views of the modeled building from the (a) outside, (b) first floor, (c) second floor, and (d)
staircase

1/2/3/4, November 2024
5
Figure 4. Three different types of fire effects, ranging from large to small fire intensities
2.2. Fire Spreading Prediction And Escape Route Estimation Models
This study adopted ML models to predict fire spreading paths and estimate escape routes.
Initially, the paths were modeled in grid-based regions, represented as rectangles, as illustrated in
Figure 5a. Each region displayed three different fire risk levels: yellow for low or no risk, orange
for medium risk, and red for high risk. These risks were simulated using the associated fire effect
types and smoke particles (see Figure 4). Various ML models, including random forest (RF),
XGBoost, decision tree (DR), logistic regression (LR), and K-nearest neighbors (KNN),
predicted the occurrence of fire in a specific region, r(x, y). This prediction was based on data
from neighboring regions, if present, such as top-left (r(x-1, y-1)), top-middle (r(x, y-1)), top-right
(r(x+1, y-1)), center-left (r(x-1, y)), center-right (r(x+1, y)), bottom-left (r(x-1, y+1)), bottom-
middle (r(x, y+1)) and bottom-right (r(x+1, y+1)) regions.
This study utilized the Algerian forest fires dataset [21] to train the ML models for fire spreading
prediction. This dataset contained 11 attributes and 244 instances, with 138 instances labeled as
fire and 106 as non-fire. Key attributes used for model training included temperature (°C),
relative humidity (%), wind speed (km/h), rain (mm), Fine Fuel Moisture Code (FFMC), and
Duff Moisture Code (DMC).
For escape route estimation, the study employed the A* algorithm [22] to compute optimized
escape routes (pathfinding) based on the predicted fire spreading index. The escape routes were
modeled as continuous green 3D circles, as shown in Figure 5b. Referring to Figure 5b, it is
important to note that the user was in a room with only one exit route (upper left), which the
green circles were displayed despite two regions along the route having high fire risk levels.
To assess the feasibility and usability of the proposed SFSDTS, participants were recruited to
interact with the system. Prior to the experiment, participants were introduced to the SFSDTS and
briefed on their role and the study's purpose. Consent forms were signed by those who agreed to
participate. After the experiment, participants completed a questionnaire to evaluate their
experiences with the SFSDTS. The study received approval from the university's research ethics
committee, following the ethical review process outlined in the university's code of research
conduct and ethics.

1/2/3/4, November 2024
6
(a) (b)
Figure 5. Illustration of the modeling for fire spreading, including the (a) paths divided into
grid-based regions and (b) color indicators showing different risk levels for each region
The SFSDTS’s fire scene 3D modeling was implemented using the Unity3D tool (version
2021.2.7f1). Flask (version 2.0.1) was utilized as the server to connect the front-end (viewer) and
back-end (fire spreading prediction and escape route estimation models), implemented in Python
(version 3.8) programming language.
2.3. Questionnaire
A user experience questionnaire (UEQ) [23] was utilized as a standardized framework to assess
the UX of participants interacting with the proposed SFSDTS. The UEQ comprised three main
aspects: valence (VL), pragmatic quality (PQ), and hedonic quality (HQ) [24]. Specifically, VL
included attractiveness (AT); PQ included efficiency (EF), perspicuity (PP), and dependability
(DP); while HQ included stimulation (ST) and novelty (NV). Thus, the UEQ evaluated a total of
six scales. These scales measured overall impressions, ease of use, efficiency of the models,
control over interaction, innovation and creativity. Table 1 provided descriptions of each scale
associated with the respective aspect.
Table 1. Descriptions of each scale associated with the respective aspect in the UEQ.
Aspect Scale Description
Total
Items
VT AT
Overall impression of the game. Do participant like or dislike
the proposed SFSDTS?
6
PQ
PP
Is it easy to get familiar with the game? Is it easy to learn how
to use the proposed SFSDTS?
4
EF
Can participants understand the shown escape route without
unnecessary effort?
4
DP Does the participant feel in control of the interaction? 4
HQ
ST Is it exciting and motivating to use the proposed SFSDTS? 4
NV
Is the proposed SFSDTS innovative and creative? Does the
proposed SFSDTS capture the interest of participants?
4
3. RESULTS AND DISCUSSION
The performance of the fire spreading prediction models was assessed using five different
evaluation metrics: accuracy (AC, see Equation (1)), precision (PR, see Equation (2)), recall (RC,

1/2/3/4, November 2024
7
see Equation (3)), and F1 score (see Equation (4)). The respective equations for the evaluation
metrics were as follows:
AC = (TP + TN) / (TP + TN + FP + FN) (1)
PR = TP / (TP + FP) (2)
RC = TP / (TP + FN) (3)
F1 = (2  PR  RC) / (PR + RC) (4)
where TP, TN, FP, FN referred to true positive, true negative, false positive and false negative,
respectively.
Table 2 summarized the performance of each ML model in predicting fire spreading. The results
showed that the XGBoost model (µ = 0.972) performed the best across all metrics. The RF model
(µ = 0.97) demonstrated similar performance to XGBoost. In contrast, the LR model (µ = 0.92)
had the lowest performance, likely due to its limited capability in predicting fire spreading
involving both discrete and continuous data types. The KNN model (µ = 0.93) indicated that
unsupervised learning was less effective in distinguishing between fire and non-fire classes based
on both data types.
Table 2. Performances metrics of ML models for fire spreading prediction.
Model AC PR RC F1 Mean,
LR 0.92 0.91 0.93 0.92 0.920
DR 0.95 0.95 0.95 0.95 0.950
RF 0.97 0.97 0.98 0.96 0.970
KNN 0.93 0.93 0.93 0.93 0.930
XGBoost 0.97 0.97 0.98 0.97 0.972
Figure 6 displayed the UEQ results with mean scores for six scales. The rating scale ranged from
-3 to +3 and was normalized to a range from 0 to 6. A mean value below 3 indicated a negative
attitude, above 3 indicated a positive attitude, and 3 indicated neutrality. Overall, the mean values
for all scales were above 3, suggesting that the proposed SFSDTS received positive ratings from
participants. The top three scales, with the highest mean values, were perspicuity, stimulation,
and efficiency, all rated above 5.3. Attractiveness and dependability also performed well, with
mean ratings above 5. Novelty received the lowest rating, below 5 but still above 3. Some
participants noted that the virtual scenes were less enjoyable, less realistic, and had a lower level
of immersion, but generally were satisfied with the SFSDTS.
The proposed SFSDTS differed from commercial products in several aspects. For example, the
FDS [25] was a computational fluid dynamics model of fire-driven fluid flow used to address fire
protection engineering problems and study fundamental fire dynamics and combustion. However,
the FDS was complex and required a significant amount of time to simulate a scenario due to
numerous parameters, limiting its effectiveness in real-time fire rescue missions. Similarly,
Autodesk Revit [26] developed a design analysis tool for fire and smoke simulation in buildings,
requiring users to specify building construction materials and initial fire types for accurate
predictions. These building information modeling (BIM) software required substantial user
training, and their complexity restricted their use to experts, making them less suitable for
ordinary users. In contrast, the proposed SFSDTS offered advantages in accessibility and real-
time simulation functionalities.

1/2/3/4, November 2024
8
Figure 6. UEQ results displaying the mean scores for six scales, normalized on a 0 to 6 range. The x-axis
represents the six GEQ scales, while the y-axis indicates the mean scores derived from participants’ GEQ
ratings
4. LIMITATION AND FUTURE WORK
The current study concentrated on developing a SFSDTS designed for a single building, along
with an initial examination of indoor weather conditions. Future efforts will focus on expanding
the system to simulate more complex fire scenarios, including the consideration of escape routes
for multiple individuals. This will involve integrating computer vision-based human detection
within indoor environments to track individuals and estimate escape routes for those still inside
the building.
5. CONCLUSIONS
This study had proposed the design and implementation of SFSDTS which had demonstrated
substantial benefits in the field of firefighting training and safety management. By utilizing
different ML models, the SFSDTS effectively predicted fire spreading paths and estimated escape
routes in real-time. The significant performance of the models highlighted the SFSDTS
robustness in handling dynamic fire scenarios. Additionally, the use of the model for escape route
optimization further improved the SFSDTS practicality and effectiveness. Compared to existing
fire simulation tools, the SFSDTS offers improved accessibility and real-time functionalities,
making it an important asset for firefighter training and emergency response planning. Future
work could explore further integration of advanced AI techniques and expanded real-world
applications to continue improving the proposed SFSDTS capabilities.
ACKNOWLEDGEMENTS
This work was supported by the UNNC Education Foundation through Li Dak Sum Innovation
Fellowship under Grant LDS202307.

1/2/3/4, November 2024
9
REFERENCES
[1] M. Singh, E. Fuenmayor, E. P. Hinchy, Y. Qiao, N. Murray and D. Devine, "Digital Twin: Origin to
Future," Applied System Innovation, vol. 4, no. 2, pp. 36, 2021.
[2] B. R. Barricelli and D. Fogli, "Digital Twins in Human-Computer Interaction: A Systematic
Review," International Journal of Human-computer Interaction, vol. 40, no. 2, pp. 79–97, 2022.
[3] G. P. Agnusdei, V. Elia and M. G. Gnoni, "Is Digital Twin Technology Supporting Safety
Management? A Bibliometric and Systematic Review," Applied Sciences, vol. 11, no. 6, pp. 2767,
2021.
[4] X. Wu, X. Zhang, Y. Jiang, X. Huang, G. G. Huang and A. Usmani, "An Intelligent Tunnel
Firefighting System and Small-Scale Demonstration," Tunnelling and Underground Space
Technology, vol. 120, no. 1, pp. 104301, 2022.
[5] T. Kaarlela, S. Pieskä and T. Pitkäaho, "Digital twin and Virtual Reality for Safety Training," 2020
11th IEEE International Conference on Cognitive Infocommunications (CogInfoCom), Mariehamn,
Finland, 2020, pp. 115–120.
[6] X. Fang, H. Wang, G. Liu, X. Tian, G. Ding and H. Zhang, "Industry Application of Digital Twin:
From Concept to Implementation," International Journal of Advanced Manufacturing Technology,
vol. 121, no. 7, pp. 4289–4312, 2022.
[7] A. Harichandran, K. W. Johansen, E. L. Jacobsen and J. Teizer, "A Conceptual Framework for
Construction Safety Training Using Dynamic Virtual Reality Games and Digital Twins,"
Proceedings of the 38th
ISARC, Dubai, UAE, 2021, pp. 621–628.
[8] T. Zohdi, "A Digital Twin Framework for Machine Learning Optimization of Aerial Fire Fighting
and Pilot Safety," Computer Methods in Applied Mechanics and Engineering, vol. 373, no. 1, pp.
113446, 2021.
[9] L. Jiang, J. Shi, C. Wang and Z. Pan, "Intelligent Control of Building Fire Protection System Using
Digital Twins and Semantic Web Technologies," Automation in Construction, vol. 147, no. 1, pp.
104728, 2023.
[10] A. Sharma, E. Kosasih, J. Zhang, A. Brintrup and A. Calinescu, "Digital Twins: State of the Art
Theory and Practice, Challenges, and Open Research Questions," Journal of Industrial Information
Integration, vol. 30, no. 1, pp. 100383, 2022.
[11] R. K. Janardhan and S. Hostikka, "Predictive Computational Fluid Dynamics Simulation of Fire
Spread on Wood Cribs," Fire Technology, vol. 55, no. 1, pp. 2245–2268, 2019.
[12] T. Zhang, Z. Wang, Y. Zeng, X. Wu, X. Huang and F. Xiao, "Building Artificial-Intelligence
Digital Fire (AID-Fire) System: A Real-Scale Demonstration," Journal of Building Engineering, vol.
62, no. 1, pp. 105363, 2022.
[13] T. T. Verlekar and A. Bernardino, "Video Based Fire Detection Using Xception and Conv-LSTM,"
In: Bebis, G., et al. Advances in Visual Computing. ISVC 2020. Lecture Notes in Computer Science,
Springer, Cham, vol. 12510, no. 1, 2022.
[14] Z. Wang, T. Zhang, X. Wu and X. Huang, "Predicting Transient Building Fire based on External
Smoke Images and Deep Learning," Journal of Building Engineering, vol. 47, no. 1, pp. 103823,
2022.
[15] Y. Lei, Z. Shen, F. Tian, X. Yang, F. Wang, R. Pan, H. Wang, S. Jiao and W. Kou, "Fire Risk Level
Prediction of Timber Heritage Buildings based on Entropy and XGBoost," Journal of Cultural
Heritage, vol. 63, no. 1, pp. 11–22, 2023.
[16] X. Zhang, Y. Jiang, X. Wu, Z. Nan, Y. Jiang, J. Shi, Y. Zhang, X. Huang and G. G. Huang, "AIoT-
Enabled Digital Twin System for Smart Tunnel Fire Safety Management," Developments in the
Built Environment, vol. 18, no. 1, pp. 100381, 2024.
[17] G. P. Forney, SmokeView (Version 6) A Tool for Visualizing Fire Dynamics Simulation, Data
Volume I: User’s Guide, 2013.
[18] L. Han, S. Potter, G. Beckett, G. Pringle, S. Welch, S. Koo et al., "FireGrid: An E-infrastructure for
Next-Generation Emergency Response Support," Journal of Parallel and Distributed Computing,
vol. 70, no. 11, pp. 1128–1141, 2010.
[19] W. Jahn, G. Rein and J. Torero, "Forecasting Fire Dynamics Using Inverse Computational Fluid
Dynamics and Tangent Linearisation," Advances in Engineering Software, vol. 47, no. 1, pp. 114–
126, 2012.

1/2/3/4, November 2024
10
[20] M. Almatared, H. Liu, O. Abudayyeh, O. Hakim and M. Sulaiman, "Digital-Twin-Based Fire Safety
Management Framework for Smart Buildings," Buildings, vol. 14, no. 1, pp. 4, 2024.
[21] F. Abid and N. Izeboudjen, "Predicting Forest Fire in Algeria Using Data Mining Techniques: Case
Study of the Decision Tree Algorithm," Ezziyyani M. (eds) Advanced Intelligent Systems for
Sustainable Development (AI2SD’2019), pp. 363–370, 2020.
[22] P. Cheng, S. Li, N. Wu and F. Meng, "Research on Fire Escape Path Planning based on Improved
A* Algorithm," 2020 IEEE 9th Joint International Information Technology and Artificial
Intelligence Conference (ITAIC), 2020.
[23] L. Sun, B. G. Lee and W. Y. Chung, "Enhancing Fire Safety Education Through Immersive Virtual
Reality Training with Serious Gaming and Haptic Feedback," International Journal of Human–
Computer Interaction, pp. 1–16, 2024.
[24] B. G. Lee, H. Tang and F. Fang, "Enhancing Critical Thinking and Engagement through Puzzle Box
Integration in Virtual Reality-based Digital Game-Based Learning," 2023 IEEE International
Conference on Teaching, Assessment and Learning for Engineering (TALE), pp. 1-8, 2023.
[25] K. McGrattan, R. McDermott, C. Weinschenk and G. Forney, "Fire Dynamics Simulator Users
Guide," Sixth Edition, Special Publication (NIST SP), National Institute of Standards and
Technology, Gaithersburg, MD, 2013.
[26] B. Wang, G. Ren, H. Li, J. Zhang and J. Qin, "Developing a Framework Leveraging Building
Information Modelling to Validate Fire Emergency Evacuation," Buildings, vol. 14, no. 1, pp. 156,
2024.
AUTHORS
Zhang received the B.A. degree in commerce from Swinburne University of Technology, the M.S. degree
in interpreting and translation from University of Nottingham Ningbo China, China. From 201x, she was
employed with the Faculty of Humanities and Social Sciences with the University of Nottingham Ningbo
China. Her research interests include education, linguistic study and extended reality.
Lee received the B.I.T. degree in information technology from Multimedia University, Malacca, Malaysia,
in 2007, the M.S. degree in electronic engineering from Dongseo University, Busan, South Korea, and the
Ph.D. degree in electronic engineering from Pukyong National University, Busan. From 2014 to 2019, he
was employed as an Assistant Professor with the Department of Electronic Engineering, Keimyung
University, South Korea. Since 2019, he was employed as an Associate Professor with the School of
Computer Science, University of Nottingham Ningbo China. His research interests include human-
computer interaction, extended reality, human-centric multi-sensing, and wireless sensors network.

Revolutionizing Firefighting Training with Digital Twins: Real-Time Fire Spread Simulation and Escape Route Optimization

More Related Content

Similar to Revolutionizing Firefighting Training with Digital Twins: Real-Time Fire Spread Simulation and Escape Route Optimization (20)

More from ijcsity (20)

Recently uploaded (20)

Revolutionizing Firefighting Training with Digital Twins: Real-Time Fire Spread Simulation and Escape Route Optimization