Content uploaded by Shuran Li
Author content
All content in this area was uploaded by Shuran Li on Dec 28, 2023
Content may be subject to copyright.
Towards Designing Smart Public Spaces: A Framework
for Designers to Leverage AI and IoT Technologies
Shuran Li1[0000-0002-0823-6208] Chengwei Wang2[0009-0006-5901-8039] Liying Rong3Yuwei
Wu1 [0000-0002-7875-8343] Zhiqiang Wu1
1Tongji University, Shanghai 200000, CHINA
2Tongji Architectural Design and Research Institute (Group) Co., Ltd., Shanghai 200000,
CHINA
3Future City (Shanghai) Design Consulting Co., Ltd., Shanghai 200000, CHINA
sl2935@cornell.edu
Abstract. Artificial Intelligent (AI) and Internet of Things (IoT) will provide
novel solutions in the area of public spaces design if the designers could
understand how these technologies can be best utilized. This study aims to
address the question, "How can practitioners be supported in applying AI and
IoT technologies in the early design process of smart public spaces?" In order to
answer the question, the author developed a framework includes three
categories and 48 technologies that can be utilized in smart public spaces
design. A focus group was run to evaluate the feasibility. The evaluation
suggests that the framework can be used as design stimuli in the concept design
phase. At the end, the paper discusses the usage and iteration direction for the
framework.
Keywords: Artificial Intelligent, Public Spaces, Design Tools, Internet of
Things.
2
1 Introduction
Public space refers to the unfenced areas situated between architectural entities in
urban settings, serving as a forum for public interaction and diverse activities among
urban residents. The public nature of urban public space renders it a crucial enabler of
civic social life, while simultaneously offering a pivotal platform for showcasing a
city's appearance and spirit. Along with the technological advancements, people’s
demand for public spaces have been shaped as well. With the daily activities and life
style were reshaped by information and communication technologies, the usage of
public spaces is undergoing transformation, which presents new challenges for public
space designers.
In recent years, with the proliferation of sensing networks and the development of
smart cities, public space has acquired interactive and responsive attributes [1].
Artificial intelligence (AI) and the Internet of Things (IoT) present numerous
possibilities for the design of public spaces [8]. Designers can utilize these
technologies to perceive the usage patterns of users and enable the environment to
respond promptly. By meeting the needs of different user groups, designers can offer
various application scenarios and personalized experiences for users. Numerous
practical examples of smart public space implementation have been successfully
completed. Smart public spaces comprise interactive facades, augment reality, smart
urban furniture[2,3], smart transportation[4] and data management platform[5],
transforming the public space from purely physical spaces into smart public space
with the ability to sensing, decision make, and respond.
There is no doubt that AI is a fundamental technology that can enhance
competitiveness, increase productivity, and bring new perspectives to address
traditional challenges in public space design [9]. However, several researchers have
suggested that practitioners face challenges in comprehending AI capabilities and
applying AI as a design material in their design process [7]. For example, some
researchers point out that user experience designers struggle to understand the
capabilities and limitations of machine learning [6]. Other scientists indicates that
designers often unable to identify obvious questions can be optimized by machine
learning[10].
3
In recent years, leverage AI into the design process became a popular research
topic. To support designer practitioners to apply AI in their design, various tools,
methods and aids [11,12] have been developed. Saleema Amershi et al have
developed a list contain 18 guidelines for Human-AI interaction [13]. Qian yang et al
demonstrate how to work with Machine Learning by series of interview [14]. Zhibin
Zhou et al developed a ML-Process Canvas, which is a design tool to support the UX
design of machine learning-empowered products[15]. The majority of design tools are
primarily focused on user experience and product design, with limited exploration in
the field of urban public space design. Most design tools lacking a comprehensive
consideration of potential technologies for designers. This lack of overview puts
designers at a disadvantage because smart public space design often requires a
combination of multiple technologies to address specific problems. Therefore, it
becomes crucial for designers to have a comprehensive understanding of potential
technologies. With a holistic perspective, designers can selectively combine and
integrate technologies based on specific design problems and requirements, thus
formulating design solutions for intelligent public spaces.
Therefore, this study aims to address the question, "How can practitioners be
supported in applying AI and IoT technologies in the early design process of smart
public spaces?" This paper focus on developing a framework with potential
technologies that can be used in designing smart public spaces. The framework aims
to facilitate designers in building a comprehensive understanding of the potential
technologies available and generated more creative ideas during the design process.
The first part of the paper reports the development process of the framework.
Second, the framework is described by sections. Third, a focus group session was run
to test the feasibility of the framework. Last, the implications and limitations of the
framework are discussed and future research steps are proposed.
2 Development the Framework of Potential AI and IoT
Technologies
This part describes the framework development procedure. The first stage was
collecting a long list of AI and IoT technologies from existing literature and practice.
For the second stage, all technologies were clustered into three categories.
2.1 Development the Framework of potential AI and IoT Technologies
The first step in our study was to create an overview of related technologies reported
in the literature. To conduct the literature review, we utilized Google Scholar due
to its comprehensive coverage. In the initial exploration, searching keywords “smart
public space”, “AI technology” and “Interaction design” were used. After a forward
and backward search, keyword “sensing” and “smart cities”was added. According to
the relevance of the content, a list included 131 papers was created. Besides literature,
this work also tried to learn from existing design practice. This study collected 119
completed cases of smart public space practices, in which the IoT and AI technologies
4
employed were categorized.
2.2 Cluster the technologies
In the second stage, one researcher from IoT background and two experienced AI
engineer were invited. All collected technologies was discussed during this stage. The
technologies that have the similar purpose with different terms were merged.
Technologies have that have similar object were clustered together. A list includes 48
potential technologies were created. The 48 technologies were subsequently clustered
into three categories: sensing, decision, and reaction.
3 Three categories of potential technologies
To help designers understand these technologies as design materials, we further
clustered the technologies in three categories based on their purpose. Figure 1
demonstrates the framework of the clustered technologies, which include sensing,
decision and reaction. Designers can combine technologies from different categories
in their design process.
Fig. 1. Framework of clustered technologies.
3.1 Sensing
Sensing technology has revolutionized public space design by providing designers
with a wealth of individual and group information as well as environmental data.
With IoT-enabled sensors, designers can gather valuable insights about individuals in
public spaces, such as their voice [18], facial features [19,20], and even emotions
[21,22]. This allows for personalized and adaptive interactions, tailoring experiences
to meet the specific needs and preferences of individuals. By analyzing group
information such as number of people [23], heat maps [24,25] and group activities
N
5
[26,27], designers can gain a deeper understanding of collective behaviors and
dynamics within public spaces. This knowledge enables the creation of inclusive and
engaging environments that promote social interactions and foster a sense of
community. Furthermore, sensing technology allows designers to monitor
environmental conditions such as temperature [28], humidity [29], and light [30],
enabling them to optimize the design of public spaces for comfort and safety. Overall,
the sensing category provides technologies that can assist designers in gathering in-
time information about the user and the public space environment.
3.2 Decision
The decision category suggests the potential algorithms that can be embedded during
the public spaces design. Based on the information collected through sensing stage,
decision technologies empower public spaces to make informed and adaptive
decisions. Numerous applications of decision-making enabled public spaces have
already been implemented. In order to further enhance urban traffic efficiency,
Copenhagen has invested 60 million kroner in the creation of 380 "smart traffic
signals." These new signals are equipped to gather real-time traffic information and
adjust signal timings accordingly to improve traffic flow. Additionally, "green wave"
signals have been installed on bicycle lanes, suggesting an ideal speed of 20 km/h.
Cyclists can adjust their riding speed based on the brightness of the green light,
thereby improving the overall throughput [16]. In Beijing, China, the AI-powered
park offers a personalized running track. Embedded facial recognition cameras within
the track can assess visitors' physical fitness and exercise completion, enabling them
to be matched with suitable exercise routes[18].
Fig. 2. The Green Wave at Copenhagen. Source: https://www.swarco.com/stories/bike-
friendly-cities
6
Fig. 3. The Intelligent track at Haidian AI park
Source: https://language.chinadaily.com.cn/a/201811/06/WS5be12540a310eff303286db1.html
http://travel.people.com.cn/n1/2018/1104/c41570-30380593.html
By leveraging AI algorithms in the decision model, public spaces can become
intelligent entities capable of enhancing user satisfaction, promoting engagement, and
fostering a sense of belonging within the community.
3.3 Reaction
The reaction category encompasses a collection of actionable responses that can be
undertaken in accordance with the decisions made. It enables public spaces to
promptly respond and adapt based on the judgments derived from the sensing and
decision components. These reactions encompass a range of possibilities, from
physical to virtual. For example, the reaction can by the adjustment of environmental
lighting [31], the playback of carefully selected audio and music [32], and the
provision of relevant visual information [33].
The timeliness of responses in smart public space design can vary. It can involve
immediate reactions based on specific circumstances, such as adjusting lighting based
on the proximity of individuals. Additionally, it can entail long-term optimization
achieved through continuous learning from the collected information.
By harnessing the insights garnered through sensing and the subsequent decision-
making process, the reaction category facilitates instantaneous and contextually
tailored responses, thereby elevating the overall user experience within public spaces.
4 Case Study
To evaluate the feasibility of the framework, the first author conducted a focus group
with three experienced designers in the fields of urban planning and landscape
architecture. During the workshop, the author provided a 10-minute presentation to
introduce the framework and the existing design practices that applied the
technologies in the framework to the participants. A design site was demonstrated
afterward (a riverside park in Shanghai, China). All participants were invited to
generated as many concepts as possible in 20 minutes by utilizing the framework.
Each designer was given five minutes to present their outcomes. Afterward, all
participants were asked to discuss their feedback and experiences of designing using
the framework. The entire focus group was audio recorded. All data were translated
7
and iteratively reviewed to cluster similar responses using thematic analysis [34] and
meaning make techniques such as affinity diagramming [35].
During the focus group session, the three participants employed a variety of design
elements provided by the framework, encompassing sensing, decision-making, and
reaction. Table 1 provides a comprehensive overview of the design outputs and
applied design strategies for each participant.
#
Design Description
Participant
Number
Sensing
Decision
Reaction
1
A range of devices responded to the ambient
temperature and humidity to create a
comfortable visiting environment for tourists,
such as adjustable pavilions and cooling mist
sprays.
#1
Temperature,
Humidity,
Light,
Air,Sound
N/A
Motion,
Water
2
Optimizing tourists the touring route based on
crowd density. Utilizing lighting and sound
cues along pathways to guide the flow of
people.
#1
Heat map
Operations
and
Optimization
Light,
Information,
Sound,
AR/VR
3
Tailoring to the preferences and physical
abilities of different individuals, provides
personalized recommendations of fitness
dance music and exercise guidance that suit
their needs.
#1
Pose,
Emotion,Vital
Sign
Element
Matching,
Recommenda
tion System
Sound, Image
4
Provide visitors with different tour
atmospheres and interactions based on their
relationships
#1
Face, Age,
Gender, Group
Activity
Element
Matching
Sound,
Water,Ligh
t,AR/VR
5
Collect the long-term usage data of the park
and provide suggestions for its planning,
design, and operation
#1
Heat Map,
Trajectory
Pattern
Recognition
Information
6
Offer personalized tour routes based on the
preferences of the visitors
#2
Heat map,
Trajectory,
Feature
Pattern
Recognition
Information
7
Illuminate special warning lights when a large
dog passes by in the vicinity
#2
Pressure,objec
t
Element
Matching
Light、
Sound
8
Provide energy-efficient and secure lighting
during nighttime solo walks in the park for
everyone's comfort and safety
#2
Distance
N/A
Light
9
Predict the crowd flow in the park to assist
visitors in avoiding peak visiting times
#3
Number of
People
Pattern
Recognition
Information
10
A photo taking installation that allow visitor
take a photo with virtual reality environment
#3
Gender,
Temperature,
Age、Feature
Element
Matching
Image
8
Fig. 4. Participants designed smart public spaces with the framework during the focus group
5 Discussion
10 concept design ideas were generated during the 20 minutes session. The high
usage rate of the technologies from the framework suggests its feasibility. In this part,
the key insights from the focus group were discussed.
The design that emerges from the focus group is characterized by its diversity. It
encompasses various aspects such as enhancing visitors' sense of security, improving
the aesthetics of the park, fostering social interactions among individuals, providing
enriched functional services, and assisting in park management. Designers can
seamlessly integrate their designs with ongoing projects, as participant 1 eloquently
stated, "I have recently been assisting a park in optimizing its operations and
management. I believe that analyzing and utilizing crowd relationships can be highly
beneficial."
The framework provides the designers with an overview of the potential
technologies. All three participants expressed their previous understanding and
attempts at employing similar intelligent techniques in public space design. However,
due to a lack of comprehensive awareness about intelligent technologies, their designs
often remained limited to replicating existing cases or applying isolated technological
solutions. This framework, on the other hand, enables them to swiftly grasp the
potential technological means and facilitates the generation of original design
proposals.
The participants acknowledged that understanding technologies in the sensing and
reaction category is relatively straightforward. However, they expressed difficulties in
distinguishing between the various technologies in the decision category. Throughout
9
the design process, the participants sought clarification from the organizers regarding
the meaning of certain aspects within the decision section. The organizers attempted
to provide a simple explanation of the algorithms in the decision section, covering
both fundamental principles and application examples. During the discuss session, the
designers' outcomes demonstrated a correct understanding of the decision section and
applied decision strategies in their design. Their feedback once again confirmed the
challenges of treating AI as a design material. The highly specialized terminology of
algorithms hindered the designers' comprehension of what the technology can
accomplish. In the course of the discussions, the author observed that, compared to
the principles of algorithms, designers were more interested in the application cases
and practical capabilities of algorithms. Participant 2 remarked, "The mathematics
makes me feel confused as well. However, when you explain the algorithms with
examples, I can immediately visualize them in my mind." The decision section not
only features the main decision algorithms but also includes subtitles for each
category. However, the participants expressed confusion about the purpose of the
subtitles, finding it challenging to differentiate between the sub-techniques and
determine when to apply each technology. One participant mentioned that they were
not concerned about the specific branches within the technologies since addressing
those issues falls under the purview of algorithm engineers. As designers, their focus
is on providing design solutions rather than delving into the algorithmic
implementation process.
The participants have raised questions regarding the next steps for implementing
the design. In the realm of traditional public space construction, designers readily
acquaint themselves with an array of prospective collaborators, spanning the realms
of architectural, construction, and electrical and plumbing teams. However, in the
design of smart public spaces, they are uncertain about which teams they will be
cooperating with. They vaguely mention the involvement of teams such as software
development and hardware design, but they lack specific understanding of their
precise engagement timeline, responsibilities, and collaboration models. The
participants express their desire for more guidance to help them comprehend how to
collaborate with multiple teams during the design process and ensure the feasibility of
the design.
6 Conclusion
6.1 General Discussion and Conclusion
This research significantly contributes to the realm of public space design and
human-computer interaction through the development of a comprehensive framework
comprising potential AI and IoT technologies for the design of intelligent public
spaces. By conducting an extensive analysis of relevant literature and real-world
practice examples, a total of 48 technologies were meticulously extracted and
categorized into three categories: sensing, decision, and reaction. To assess the
efficacy of the framework, a evaluation was conducted utilizing a focus group
approach. In a specific case study involving the design of a riverside park, three
10
designers generated ten distinct concept designs. Notably, the implementation of the
framework yielded a remarkable utilization rate, with 20 out of 22 sensing strategies,
4 out of 6 decision-making strategies, and all 7 reaction strategies being effectively
employed. The pronounced utilization of the framework underscores its inherent
comprehensibility and applicability. Additionally, the ensuing discussion session
underscored the immense potential of employing the framework as a potent design
stimulus tool during the nascent conceptual design phase, effectively harnessing the
capabilities of AI and IoT technologies in the design process.
6.2 Limitation and future work
The participants' frequent utilization of strategies provided by the framework serves
as a validation of its utility while also indicating the requirements for its further
development. Building upon the designers' unique cognitive approach to technology,
the author aims to expound on the technologies within the decision category using
designerly language in the upcoming phase. Concrete design cases and visual aids
will be employed to facilitate designers' comprehension of these technologies. Equally
important are the application guidelines for the framework.
Considering diverse design objectives, such as environmental beautification, social
facilitation, and enhancing security, the author will furnish designers with practical
guidance on applying the framework. This will prove invaluable in aiding designers
unfamiliar with AI and IoT in swiftly acquainting themselves with these technologies.
References
1. Suurenbroek, F., Nio, I., & de Waal, M. Responsive public spaces: exploring the use of
interactive technology in the design of public spaces. (2019).
2. Premier, A., GhaffarianHoseini, A., & GhaffarianHoseini, A. Solar-powered smart urban
furniture: preliminary investigation on limits and potentials of current designs. Smart and
Sustainable Built Environment. (2022).
3. Ciaramella, A., Bellintani, S., Savio, L., Carbonaro, C., Pagani, R., Pennacchio, R., ... &
Thiebat, F. Smart furniture and smart city. In IOP Conference Series: Materials Science
and Engineering (Vol. 365, No. 2, p. 022012). IOP Publishing. (2018).
4. Oberheitmann, A. The Development of Smart Cities in China. Smart Urban Regeneration,
201-213. (2017).
5. Burange, A. W., & Misalkar, H. D. Review of Internet of Things in development of smart
cities with data management & privacy. In 2015 International Conference on Advances in
Computer Engineering and Applications (pp. 189-195). IEEE. (2015).
6. Dove, G., Halskov, K., Forlizzi, J., & Zimmerman, J.UX design innovation: Challenges
for working with machine learning as a design material. In Proceedings of the 2017 chi
conference on human factors in computing systems (pp. 278-288). (2017).
7. Yang, Q., Steinfeld, A., Rosé, C., & Zimmerman, J. Re-examining whether, why, and how
human-AI interaction is uniquely difficult to design. In Proceedings of the 2020 chi
conference on human factors in computing systems (pp. 1-13). (2020).
11
8. Townsend, A.M.: Smart Cities: Big Data, Civic Hackers, And The Quest For A New
Utopia. W.W. Norton & Company, New York (2013)
9. Yang, Q. Machine learning as a UX design material: how can we imagine beyond
automation, recommenders, and reminders?. In AAAI Spring Symposia (Vol. 1, No. 2.1,
pp. 2-6). (2018).
10. Yang, Q., Zimmerman, J., Steinfeld, A., & Tomasic, A. Planning adaptive mobile
experiences when wireframing. In Proceedings of the 2016 ACM Conference on
Designing Interactive Systems (pp. 565-576). (2016,).
11. Bond, R. R., Mulvenna, M. D., Wan, H., Finlay, D. D., Wong, A., Koene, A., ... & Adel,
T. Human Centered Artificial Intelligence: Weaving UX into Algorithmic Decision
Making. In RoCHI (pp. 2-9). (2019).
12. Xu, W. Toward human-centered AI: a perspective from human-computer interaction.
interactions, 26(4), 42-46. (2019).
13. Amershi, S., Weld, D., Vorvoreanu, M., Fourney, A., Nushi, B., Collisson, P., ... &
Horvitz, E. Guidelines for human-AI interaction. In Proceedings of the 2019 chi
conference on human factors in computing systems (pp. 1-13). (2019).
14. Yang, Q., Scuito, A., Zimmerman, J., Forlizzi, J., & Steinfeld, A. Investigating how
experienced UX designers effectively work with machine learning. In Proceedings of the
2018 designing interactive systems conference (pp. 585-596). (2018).
15. Zhou, Z., Sun, L., Zhang, Y., Liu, X., & Gong, Q. ML lifecycle canvas: designing machine
learning-empowered UX with material lifecycle thinking. Human–Computer Interaction,
35(5-6), 362-386. (2020).
16. Cardwell, D. Copenhagen lighting the way to greener, more efficient cities. New York
Times, 8. (2014).
17. PK, J., Sen, A., Chen, X., & Murali, A. Inside Haidian Park: A stroll through the world’s
first “ai-park.” Retrieved from https://archive.factordaily.com/inside-haidian-park-the-
worlds-first-artificial-intelligence-park/ (2018).
18. Arriany, A. A., & Musbah, M. S. Applying voice recognition technology for smart home
networks. In 2016 International Conference on Engineering & MIS (ICEMIS) (pp. 1-6).
IEEE. (2016)
19. Li, L., Mu, X., Li, S., & Peng, H. A review of face recognition technology. IEEE access, 8,
139110-139120. (2020).
20. Praveen, G. B., & Dakala, J. Face recognition: Challenges and issues in smart
city/environments. In 2020 International Conference on COMmunication Systems &
NETworkS (COMSNETS) (pp. 791-793). IEEE. (2020).
21. Pujol, F. A., Mora, H., & Martínez, A. Emotion recognition to improve e-healthcare
systems in smart cities. In Research & Innovation Forum 2019: Technology, Innovation,
Education, and their Social Impact 1 (pp. 245-254). Springer International Publishing.
(2019).
22. Hossain, M. S., & Muhammad, G. Emotion recognition using deep learning approach from
audio–visual emotional big data. Information Fusion, 49, 69-78. (2019).
23. Marsden, M., McGuinness, K., Little, S., & O'Connor, N. E. Resnetcrowd: A residual deep
learning architecture for crowd counting, violent behaviour detection and crowd density
level classification. In 2017 14th IEEE international conference on advanced video and
signal based surveillance (AVSS) (pp. 1-7). IEEE. (2017).
24. Jarvis, N., Hata, J., Wayne, N., Raychoudhury, V., & Gani, M. O. Miamimapper: Crowd
analysis using active and passive indoor localization through wi-fi probe monitoring. In
Proceedings of the 15th ACM International Symposium on QoS and Security for Wireless
and Mobile Networks (pp. 1-10). (2019).
12
25. Atta, S., Sadiq, B., Ahmad, A., Saeed, S. N., & Felemban, E. Spatial-crowd: A big data
framework for efficient data visualization. In 2016 IEEE International Conference on Big
Data (Big Data) (pp. 2130-2138). IEEE. (2016).
26. Yan, R., Tang, J., Shu, X., Li, Z., & Tian, Q. Participation-contributed temporal dynamic
model for group activity recognition. In Proceedings of the 26th ACM international
conference on Multimedia (pp. 1292-1300). (2018).
27. Deng, Z., Vahdat, A., Hu, H., & Mori, G. Structure inference machines: Recurrent neural
networks for analyzing relations in group activity recognition. In Proceedings of the IEEE
conference on computer vision and pattern recognition (pp. 4772-4781).(2016).
28. Channi, H. K., & Kumar, R. The role of smart sensors in smart city. In Smart Sensor
Networks: Analytics, Sharing and Control (pp. 27-48). Cham: Springer International
Publishing. (2021).
29. Siregar, B., Nasution, A. B. A., & Fahmi, F. Integrated pollution monitoring system for
smart city. In 2016 International conference on ICT For smart society (ICISS) (pp. 49-52).
IEEE. (2016).
30. Hancke, G. P., de Carvalho e Silva, B., & Hancke Jr, G. P. The role of advanced sensing in
smart cities. Sensors, 13(1), 393-425. (2012).
31. Dheena, P. F., Raj, G. S., Dutt, G., & Jinny, S. V. IOT based smart street light
management system. In 2017 IEEE international conference on circuits and systems
(ICCS) (pp. 368-371). IEEE. (2017).
32. Sarmento, P., Holmqvist, O., & Barthet, M. Musical Smart City: Perspectives on
Ubiquitous Sonification. arXiv preprint arXiv:2006.12305. (2020).
33. Murshed, S. M., Al-Hyari, A. M., Wendel, J., & Ansart, L. Design and implementation of
a 4D web application for analytical visualization of smart city applications. ISPRS
International Journal of Geo-Information, 7(7), 276. (2018).
34. Clarke, V., Braun, V., & Hayfield, N. Thematic analysis. Qualitative psychology: A
practical guide to research methods, 3, 222-248. (2015).