TYPE Original Research
PUBLISHED 30 May 2025
DOI 10.3389/frobt.2025.1580289
Human interactions with delivery
OPEN ACCESS drones in public spaces: design
EDITED BY
Ginevra Castellano, recommendations from recipient
Uppsala University, Sweden
REVIEWED BY and bystander perspectives
Sofia Thunberg,
Chalmers University of Technology, Sweden
Ana Tanevska,
Shiva Nischal Lingam1,2*, Rutger Verstegen2
Uppsala University, Sweden ,
3 2,4
*CORRESPONDENCE Sebastiaan M. Petermeijer and Marieke Martens
Shiva Nischal Lingam,
1
s.n.lingam@tue.nl Aerospace Operations Safety and Human Performance, Royal Netherlands Aerospace Center,
Amsterdam, Netherlands, 2Industrial Design, Eindhoven University of Technology, Eindhoven,
RECEIVED 20 February 2025 Netherlands, 3Aerospace Operations Training and Simulation, Royal Netherlands Aerospace Center,
ACCEPTED 29 April 2025 Amsterdam, Netherlands, 4Integrated Vehicle Safety, TNO, Helmond, Netherlands
PUBLISHED 30 May 2025
CITATION
Lingam SN, Verstegen R, Petermeijer SM and
Martens M (2025) Human interactions with Drones will likely deliver packages in public spaces, where humans interact
delivery drones in public spaces: design
recommendations from recipient and as recipients of the package and as bystanders passing by. Understanding
bystander perspectives. the human needs and uncertainties that may arise during these interactions
Front. Robot. AI 12:1580289. is crucial to ensure safety. This user-centered design study employed twelve
doi: 10.3389/frobt.2025.1580289
interviews and four focus groups to identify key requirements for recipients
COPYRIGHT
and bystanders interacting with delivery drones in public spaces. Findings
© 2025 Lingam, Verstegen, Petermeijer and
Martens. This is an open-access article demonstrate different information needs and preferred interface modalities
distributed under the terms of the Creative between recipients and bystanders across various interaction stages, from
Commons Attribution License (CC BY). The
ordering a package to the drone’s retraction after delivery. This paper highlights
use, distribution or reproduction in other
forums is permitted, provided the original essential design features and offers concrete design recommendations based
author(s) and the copyright owner(s) are on the interaction requirements. These recommendations can inform the
credited and that the original publication in
standardization and customization of design features for each interaction
this journal is cited, in accordance with
accepted academic practice. No use, stage, enhancing safety and facilitating natural human-drone interaction. Future
distribution or reproduction is permitted research should build on these recommendations and validate the design
which does not comply with these terms.
concepts through experimental user studies involving human interactions with
delivery drones in public spaces.
KEYWORDS
human-robot interaction, human-drone interaction, robot design, human-machine
interface, delivery application, interview, focus group, public space
1 Introduction
Drones are increasingly becoming part of daily life, with the expanding global drone
market creating new opportunities for consumer interactions and integration into public
spaces. Consumers now receive package deliveries from drones, fostering interactions
between drones and humans (Shankland, 2023). Delivery is highlighted as a major
application in the Human-Drone Interaction (HDI; a sister domain to Human-Robot
Interaction) literature, as reviewed by Herdel et al. (2022). An expert interview study
on HDI by Lingam et al. (2024) also identifies delivery as a key application in public
spaces for the coming decade. These findings suggest that delivery will be a primary use
case for drones in public spaces, involving interaction with the individuals. However,
designing robots (e.g., drones) to safely operate and adapt to the uncertainties of public
spaces is challenging due to complexities, arising from diverse situational factors as well
Frontiers in Robotics and AI 01 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
as the involvement of various stakeholders, including recipient, to address uncertainty in a public space (e.g., park) delivery context,
bystanders, and vulnerable populations (Yu et al., 2024). focusing on the roles of recipients and bystanders. By investigating
Consideration of public needs and acceptance is crucial for the user needs and preferences regarding drone interactions, the study
successful integration of drone delivery technology into society. seeks to identify and discuss essential design features for each role.
Zenz and Powles (2024) examine public resistance to drone delivery These insights contribute to developing design guidelines for future
services, with a particular focus on Google Wing’s operations researchers and drone designers and improving the natural HDI
in Canberra, Australia. The authors highlight how inadequate experience.
community engagement can lead to resistance, disrupting corporate The contributions of this study for the HDI and HRI
plans and emphasizing the importance of aligning technological community are:
development with public interests. In public spaces, humans
primarily interact with delivery drones in two roles: as recipients • Exploration of the roles of humans, as recipients and bystanders,
who actively interact by receiving packages or as bystanders in interactions with delivery robots in public spaces.
who may be nearby but do not participate in the interaction • Identification of uncertainty factors and user requirements for
(Lingam et al., 2024). Expectations and informational needs differ HDI with delivery robots in public spaces.
between these roles; for example, while recipients anticipate their • Highlighting essential design features for standardization and
interaction with the drone, bystanders might be unaware of the customisation, along with user reflections on current delivery
drone’s purpose and require contextual information. Prior research drone designs.
(Lingam et al., 2025; Obaid et al., 2015; Tan et al., 2018) has • Provision of design recommendations for each interaction stage
primarily focused on the perspective of recipients in public spaces, to reduce uncertainty and improve safety.
however, bystander perspectives have been less frequently studied.
Previous Human-Robot Interaction (HRI) research (Nielsen et al.,
2023; Pelikan et al., 2024; Yu et al., 2024) underscore the need 2 Background
to consider bystander roles, as delivery robots are likely to
interact with bystandersmore frequently than recipients (Rosenthal- Our research is informed by four key areas in the literature.
von der Pütten et al., 2020) and can contribute to interaction First, we present the background on the human roles in public
breakdowns (Nielsen et al., 2023). The lack of consideration for spaces within HDI and HRI. Next, we explore the significance of
both (i.e., recipient and bystander) perspectives in the HDI design addressing human perceived uncertainty in HDI. Then, we consider
space leads to feelings of uncertainty (henceforth referred to as the challenges and strategies for managing uncertainty through
uncertainty), which can affect their trust in automated systems, such design features in drones. Finally, we discuss the background and
as drones (Lee and See, 2004). relevance of the user-centered approach in the fields of HDI and
A possible approach to reducing uncertainty involves carefully HRI, which forms the foundation of our research methods.
designing the appearance of drones and implementing Human-
Machine Interfaces (HMIs). For instance, drones could adopt visual
cues similar to delivery vehicles on the road, making their purpose 2.1 Human roles in public space
more evident to the public (Lingam et al., 2024). Additionally, interactions
HMIs such as speakers and ground projections can communicate
specific intentions of drones to the public (Obaid et al., 2015). Individuals from the public interact with drones in two primary
However, research exploring user requirements for the delivery ways: actively, as recipients engaging with the drone service,
drones in public spaces remains limited. As drones increasingly or passively, as bystanders situated in the vicinity of the drone
enter public spaces for deliveries, variations in current delivery (Lingam et al., 2024). Previous HDI studies have primarily focused
drone designs (c.f., Matternet, 2024; Shankland, 2023; Vuleta, on the role of the recipient in public spaces. For instance, Obaid et al.
2021) prompt questions about the impact of these designs on (2015) evaluated the use of HMIs for drones to assist road users in
human perception. A lack of discussion on standardization and discarding garbage, and Tan et al. (2018) examined the delivery of a
customisation of design elements in appearance and interfaces can wedding ring by a drone in a public event. However, the emerging
lead to varied interpretations, potentially causing confusion and passive role of the bystander has been rarely investigatedwithinHDI.
increasing uncertainty among the public. The exploration of bystander interactions has been gaining
The literature lacks clarity on which specific design elements attention inHRI, particularly for service robots in public spaces (e.g.,
need standardization to effectively communicate delivery intentions, Nielsen et al., 2023; Pelikan et al., 2024; Yu et al., 2024). Bystanders,
which aspects should be customized based on the type of human though not the primary users, are likely to encounter delivery robots
role (recipient and bystander), and how these elements interrelate more often than recipients (Rosenthal-von der Pütten et al., 2020).
in an interaction space. Understanding the balance between A design probe study (Yu et al., 2024) and a video-based study
standardization and customisation in drone design is crucial for (Pelikan et al., 2024) found that ground robots disrupt bystander
developing user-centered design principles that address the needs of activities when not designed to adapt to context, particularly
public users. Our study attempts to fill these gaps and contribute to when the robots’ objectives differ from bystander interests. The
the future development of natural interaction between public users authors (Pelikan et al., 2024; Yu et al., 2024) proposed contextually
(i.e., recipients and bystanders) and delivery drones. adaptive robot designs that account for diverse bystander needs
This study aims to explore user requirements and offer to facilitate natural HRI in public environments. In another video
recommendations for designing drones and their interaction spaces study (Nielsen et al., 2023), findings indicated that over 30% of
Frontiers in Robotics and AI 02 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
disruptions in interactions occur between a public service robot gap in the literature regarding the investigation of user needs for
and bystanders in an airport, in addition to challenges related to handling uncertainty, particularly in the context of delivery drones
environmental disruptions and control features. Interactions with within HDI.
bystanders frequently broke down due to interruptions caused by
children and adults, who are curious. Collectively, these studies
(Nielsen et al., 2023; Pelikan et al., 2024; Yu et al., 2024) highlight 2.3 Managing uncertainty through drone
the importance of considering bystander interests to ensure safe and design
natural HRI in public spaces.
Recipients and bystanders may share the same space during Design elements such as flying patterns, propeller noise, drone
drone deliveries, but their roles differ. Recipients expect to appearance, and HMIs play a crucial role in conveying drone
interact with the drone to receive a package, while bystanders intentions, as highlighted in recent expert interviews and user
may be unaware of the drone’s purpose and have limited studies (Bevins and Duncan, 2021; Lingam et al., 2024; 2025;
involvement (Lingam et al., 2024). These differences necessitate Obaid et al., 2015; Fink et al., 2023; Tan et al., 2018), and managing
distinct requirements and interaction protocols for safer and more user uncertainty inHDI. For instance, experts in Lingamet al. (2024)
natural HDI. Addressing these differences presents challenges in suggested that propeller noise, similar to an ambulance siren, could
drone design and interaction that are rarely covered in existing HDI signal a drone’s presence and purpose, thereby reducing uncertainty
literature. in HDI. Bevins and Duncan (2021) examined how flying patterns
affect user perception of drone intentions in a video-based study,
observing that an undulating flight pattern and a straight descent
2.2 Feelings of uncertainty in HDI were both interpreted as signals to avoid approaching the drone. A
virtual reality experiment by Lingam et al. (2025) investigated drone
Due to the complexity of public spaces and the novelty of flight paths and delivery methods, finding that drones approaching
drone technology, uncertainty may arise during HDI. In accordance recipients in curved paths and delivering packages through a
with the explanation provided by the research experts in the field cable while hovering above eye level were associated with lower
of HDI across academia and industry (Lingam et al., 2024), we uncertainty and higher trust compared to drones following a straight
define uncertainty as “a state of doubt experienced by humans path and landing on ground to deliver packages. Flying patterns
when interactions with drones deviate from the expected, leading and propeller noise serve as implicit cues for drone intentions;
to a loss of understanding of the drone’s intentions or its next however, humans may interpret the same cue in different ways.
actions.” Handling uncertainty is a critical challenge in integrating For example, Bevins and Duncan (2021) found that while some
drones into public spaces (Lingam et al., 2024). Uncertainty may participants perceived a U-shaped flight pattern as a signal to avoid
rise during the human interactions with robots, such as drones, approaching the drone, others interpreted it as an invitation to look
that negatively affects the decision making of the human (Lindley, at the drone.
2013) and defines the boundaries of trust in automated systems Another approach is to explore explicit forms of communication
(Lee and See, 2004). Uncertainties about drone identification and such as drone appearance and HMIs. Such cues improve clarity
purpose can cause confusion and discomfort, making it essential and interpretability of the drone intentions by conveying explicit
to address these issues to prevent miscommunication, enhance information (Fink et al., 2023). In Lingam et al. (2024), experts
trust and safety, and ensure natural HDI. Limited familiarity with underscored the importance of purpose-reflective drone design,
drones contributes to the uncertainty. Vuleta (2021) found that drawing inspiration from vehicles like delivery trucks and
only 15% of U.S. residents have experience operating drones. ambulances. On similar lines, Tan et al. (2018) explored user
Due to current safety regulations (Federal Aviation Administration, preferences for delivery drone design, recommending propeller
2024), the proportion of individuals with experience interacting guards and explicit communication of intentions to enhance safety
with delivery drones to receive packages is likely even lower. and comfort. In contrast, Wojciechowska et al. (2019) conducted
The lack of knowledge and information can lead to uncertainty an online study where participants rated 63 static images of drones
and higher perception of risks towards robots, such as drones using Likert scale statements, regardless of the application area,
(Clothier et al., 2015; Meissner et al., 2020). and recommended excluding propeller guards on delivery drones
A possible direction to address uncertainty regarding to improve trustworthiness, interaction likability, and friendliness.
interactions with delivery drones involves comprehending and Post-experiment interviews from Lingam et al. (2025) showed that
designing as per the user’s expectations and needs regarding recipients desired design features reflecting the drone’s purpose,
both the drone and the interaction. Fink et al. (2023) discussed suggesting ambulance-like features for medical drones. Recipients
the importance of understanding user requirements by involving also expressed the need for delivery intentions to be communicated
humans in the development process of medical drone services, via sound and lights, particularly during package drop-offs. Experts
concerns can be addressed, and the security, usability, and (Lingam et al., 2024) have suggested using ground projections to
acceptance of the technology can be improved. Shapira and indicate landing spots and implementing audio signals to signal
Cauchard (2022) highlighted the significance of crafting public safety warnings to nearby recipients. Additionally, Obaid et al.
service drones and the elements of interaction spaces, such as (2015) found that combining audio and ground projections in
drone appearance, with user experience in mind to enhance HMIs was more effective than audio alone in persuading recipients
public acceptance. While it is crucial to reduce uncertainty to to clean the garbage in the public space, demonstrating greater
promote a natural HDI experience (Jane et al., 2017), there is a influence. While these studies highlight the potential for managing
Frontiers in Robotics and AI 03 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
recipient uncertainty through design elements, they offer limited spaces in delivery contexts, draws inspiration from the user-centered
insights into how these design elements compare with the needs of design approach.
bystanders.
Current delivery drone models vary in terms of appearance
and the availability of HMIs to communicate with users. For 3 Methods
instance, TU Delft’s ambulance drone (Momon, 2014) includes an
onboard audio interface to guide recipients during emergencies by The study implemented a user-centered design process
delivering Automated External Defibrillators. In contrast, drones consisting of two stages: online interviews and focus groups with
from Matternet (2024), Wing (2024), and Zipline (Shankland, ideation sessions (see Figure 1). Online interviews were conducted
2023) used for medical and commercial deliveries lack both such to explore individual experiences, needs, and uncertainties in depth.
interfaces and ambulance-like designs. The drone models also vary Follow-up focus group sessions were utilized to stimulate group
significantly in appearance, including colors and forms, as well as discussions to let participants raise issues that might not have been
in their delivery methods. This diversity raises questions regarding identified in the interviews (Lazar et al., 2017). The approach was
the impact of appearance and HMIs on user perceptions and inspired by iterative design processes in HDI literature, including
expectations. The appearance of drones, including their design interviews, surveys, and focus groups with ideation sessions (e.g.,
and forms, may influence how users perceive the technology and Herdel et al., 2021; Tan et al., 2018; Yeh et al., 2017).The study design
the services provided (Wojciechowska et al., 2019). A possible was approved by the Eindhoven University Ethical Review Board.
direction is to identify the design elements of drone models
and HMIs that require standardization according to the use
case (i.e., delivery) to manage the uncertainty experienced by 3.1 Interviews
public users (Lingam et al., 2025).
The interviews were conducted and recorded via Microsoft
Teams by the first author. Before the interviews, participants were
2.4 User-centered design approach provided with a document detailing two (fictional) scenarios that
illustrated the roles of recipient and bystander, to set context and
The user-centered design approach has been widely applied expectations (Buskermolen and Terken, 2012). A semi-structured
in the fields of HRI and HDI to investigate and incorporate interview approach was used, with questions iteratively developed
human needs throughout the design process of robots, ensuring through research group discussions, insights from prior studies on
that human needs are appropriately considered (Alon et al., 2021; HDI for delivery drones (Lingam et al., 2024; 2025), and results from
Amiche et al., 2024; Herdel et al., 2021; Karjalainen et al., 2017; the pilot study.
Tan et al., 2018; Yeh et al., 2017). The design process often involves
iterative and multi-stage activities, such as user needs exploration
and design workshops, in order to provide recommendations for 3.2 Focus groups
future developments of robots (Amiche et al., 2024).
Previous HDI studies have conducted interviews to The first and second authors moderated the group discussions
investigate expert user requirements (Khan and Neustaeder, and ideation sessions, which were audio-recorded to capture
2019; Ljungblad et al., 2021). For instance, Ljungblad et al. humans’ perceptions and visions of interacting with delivery drones,
(2021) interviewed drone pilots to identify their needs for drone as well as to understand the rationale behind their preferred features.
applications and proposed corresponding design guidelines. Khan Participants were initially encouraged to discuss their requirements
and Neustaeder (2021) interviewed firefighters to understand for each role separately within their groups. They storyboarded
their needs and provided design recommendations for drones interactions with delivery drones, a method commonly used in
assisting in firefighting operations. Other studies have conducted HCI literature (e.g., Kantola and Jokela, 2007; Truong et al., 2006),
focus groups and design workshops to explore and derive design and sketched the drone (e.g., Herdel et al., 2021; Yeh et al., 2017)
implications for drones interacting with recipients (Herdel et al., based on their discussions and preferences. During the pilot sessions
2021; Karjalainen et al., 2017; Tan et al., 2018; Yeh et al., 2017). and interviews, participants desired information on the phone.
For example, Herdel et al. (2021) conducted focus groups with Consequently, interface cards shaped like smartphone cutouts were
participants, asking them to sketch potential capabilities for police provided to represent mobile information requirements during
drones in public spaces, considering varying levels of context storyboarding.
severity. Karjalainen et al. (2017) organized design workshops
to understand and elaborate on appearance requirements for a
companion drone. Tan et al. (2018) explored recipient preferences 3.3 Procedure
for delivery drone design using focus groups, while Yeh et al. (2017)
conducted sketching sessions to capture how recipients envision a Participants were asked to complete a questionnaire on their
social drone and to elaborate on preferred design features. These demographics (gender, age, ethnicity, educational background),
HDI studies provided recommendations for drone design and their attitudes towards technology interaction (Franke et al., 2019),
integration into human environments. Our research, which aims experience with drones, and provide consent.They read the scenario
to explore human (i.e., recipient and bystander) requirements and document and were then interviewed (see Supplementary Material
to offer design recommendations for drones and their interaction for the scenario description and questions) to understand the
Frontiers in Robotics and AI 04 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
FIGURE 1
Study methods implemented to explore requirements for the recipient and bystander roles.
their preferences and perceptions. Adjustments to their sketches
were permitted based on the observed features. Although the
sample of drone models in the videos was not exhaustive, it served
as inspiration, given participants’ limited real-world experience
with delivery drones. The interviews and focus groups lasted
approximately 1 hour and about 2 hours, respectively. Participants
were thanked and compensated with a €35 voucher at the end
of the study.
3.4 Participants
FIGURE 2
Participants in a focus group. They were given A3 papers, colored
markers, and pens to create sketches. The study prioritized the depth and quality of qualitative data
over larger sample size (Lingam et al., 2024; Ljungblad et al.,
2021). Previous user-centered design studies in HDI have reached
acceptable results and saturation with fewer than 10 participants
requirements for the two roles in the delivery context within a (e.g., Alon et al., 2021; Herdel et al., 2022; Tan et al., 2018). Twelve
public park. Participants used the Miro board (https://miro.com/) participants were recruited for this study to ensure balanced focus
to categorize requirements according to theMoSCoW prioritization groups and rich data collection, with sessions lasting approximately
method into four categories: 1) “Must Have,” 2) “Should Have,” 3 hours for each participant and involvingmultiplemethods for both
3) “Could Have,” and 4) “Won’t Have” (Clegg and Barker, 1994). recipient and bystander roles. All the participants had seen drones
The MoSCoW principle has been used previously to prioritize user in the media or from a distance in reality, but none had experience
requirements in the design of human-technology interactions, such piloting, owning, or seeing a delivery drone. Participants were
as those involving autonomous vehicles (Hallewell et al., 2022; selected based on their limited experience with drones and their
Rodak et al., 2020) and smartphones (Vos et al., 2016). background in design. Their limited prior interaction with drones
Participants were invited to focus groups where they could provide insights from a novice public user perspective, as
storyboarded interaction scenarios and sketched the drone in most residents in the Netherlands (where the study was conducted)
separate sessions for the roles of recipient and bystander. They were currently lack experience with delivery drones. This is due to the
provided with A3 papers, colored markers, and pens (see Figure 2) state of technology and existing regulations, which prohibit delivery
for these tasks and could choose to incorporate attributes identified drones (>500 g) from flying near humans, requiring a minimum
in the interviews or not. Additionally, participants were provided horizontal distance of 50 m (Netherlands Enterprise Agency,
with a drone silhouette (i.e., a hybrid VTOL model of the Zipline 2024). During the pilot study, participants without a design
drone) on anA3 paper to initiate their sketches, inspired by previous background faced difficulty to provide design solutions
studies (Herdel et al., 2021; Yeh et al., 2017). They were informed during the focus group discussions. To facilitate a clearer
that this silhouette was for inspiration and not mandatory to adopt articulation of complex ideas through sketches and storyboards,
in their designs. They could sketch HMIs directly on the drone or participants with a background in visual communication design
on interface cards, and were asked to explain their design choices were recruited.
while remaining open to alternative solutions. Participants (five male, seven female) were aged 25–34 years
After completing the storyboard and sketching (M = 29.2, SD = 3.2) and were from diverse backgrounds: five
phases, participants watched videos of existing drone Chinese, two Dutch, two Indian, and one each from Brazil, Greece,
models (see Supplementary Material for videos), including Amazon and Indonesia. Overall, they indicated a positive attitude towards
Prime Air, Mana, Wing, and Zipline, to prompt reflections on technology interaction (M = 3.8, SD = 0.6). Twelve interviews
Frontiers in Robotics and AI 05 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
and four focus groups, each consisting of three participants, were 4.1.1.1 Differences in human roles
conducted in June and July 2024. The levels and types of uncertainties differed between the two
human roles: recipient and bystander. As the recipient has “(...)
much more of an expected engagement” (P6) with the drone than
3.5 Analysis methods a bystander, the possible interactions and the uncertainties are
different and higher for a bystander:
A thematic analysis was conducted to qualitatively identify
and report patterns in participant responses (Braun and Clarke, “(...) you can probably have like lists of possible use cases of
2006), consistent with prior HDI research (Lingam et al., 2024; possible archetypes [for recipients]. But for bystanders, you
2025; Ljungblad et al., 2021). The analysis was performed on the could have like, so many unpredictable variations, reactions
transcriptions, sketches and storyboards. Interview and focus group and dynamics in the park” (P1).
recordings were automatically transcribed and then reviewed and “I will definitely feel more uncertain as [a bystander]
corrected for accuracy by the first and second authors (referred to compared to being a recipient” (P9).
as analysts). Personal information, such as names, was removed.
First, the analysts familiarized themselves with the transcriptions to Majority of the participants, as recipients, expressed feeling
extract insights from the interview data. Second, they familiarized less uncertain during the interaction. Uncertainties regarding safety
themselves with the focus group transcriptions, sketches and involve interactions, flying behavior, drone identification, drone and
storyboards to identify recurring visual motifs, narrative structures, package size, and recipient positioning for package delivery. On
and thematic elements. Codes were created to categorize elements, the other hand, as a bystander, most of the participants reported
which were then organized into sub-themes and themes. For feeling uncertain when a delivery drone approaches the vicinity.
example, interview codes related to the role of external agents, The bystanders expressed mixed emotions and uncertainty towards
the presence of multiple recipients and drones, and environmental the propeller noise, the purpose of the drone, the context and the
factors contributing to user uncertainty were organized under the delivery location.
sub-theme, public space dynamics. The codes, sub-themes and
themes were compared across the two human roles to identify 4.1.1.2 Public space dynamics
similarities and differences. The analysis was conducted in a data- The dynamics of a public space, such as the existence of
driven and emergentmanner.The results from the analysis indicated environmental factors, the role of external agents, and the presence
a degree of saturation within the utilized sample, as observed by the of multiple recipients and drones, could influence uncertainty as,
consistency of sub-themes and themes across participants. “(...) there is a lot going to be happening in that [delivery] situation”
Information requirements and attributes mentioned by (P6). These uncertainties, related to public space dynamics, might
participants were categorized using the MoSCoW method and prompt recipients to alter the delivery location shortly before the
quantified. drone arrives.
Environmental factors in and around the park, such as weather
conditions, geographical features (e.g., park layout, rivers), cars,
4 Results and trees, could contribute to feelings of uncertainty during the
delivery interaction, especiallywhen compared to drone deliveries to
4.1 Interview results homes. If multiple groups are ordering deliveries in a public space,
recipients might be uncertain about identifying the correct drones
The thematic analysis identified three main themes: factors and could receive misplaced packages. Additionally, both recipients
contributing to user uncertainty about HDI, user requirements to and bystanders could be annoyed by the presence and noise of
feel certain during HDI, and drone design solutions to address several drones. Uncertainties caused by external agents, such as
uncertainty in HDI, along with a total of 12 sub-themes. These recipients and bystanders engaging in activities with friends, family
themes (see Table 1) are presented in the following sections, (including elders and children), pets, and birds, were recognised as
accompanied by selected participant quotes. Participant quotes from hindrances to the delivery process and raise safety concerns: “(...)
the interviews are labeled with “P,” followed by the corresponding have people running up to the drone (...) because that’s going to
interview number. create chaos” (P8).
4.1.1 Factors contributing to user uncertainty 4.1.1.3 Familiarity with delivery drone technology and
about HDI processes
Participants expressed uncertainty when interacting with Uncertainty was expressed about “(...) how the drone will
delivery drones in public spaces, primarily due to the behave” (P11) due to participants’ lack of familiarity with drone
unpredictability associatedwith the novelty of the technology and its technology and delivery processes. Instead, they drew comparisons
autonomous functioning. They identified handling uncertainty as a with existing services, such as DHL and UberEats for package
significant challenge, citing factors such as differences in human roles, deliveries (e.g., “imagine the drone is the food delivery guy”
public space dynamics, familiarity with delivery drone technology and (P3)), and ambulances for medical deliveries. Participants expressed
processes, criticality of the situation, and privacy concerns. These uncertainty about the delivery methods, for instance, “I don't
elements (see below sub-sections) collectively shaped their feelings know if the drone would land or how are the package[s] being
of uncertainty toward delivery drones in public spaces like parks. delivered” (P12).
Frontiers in Robotics and AI 06 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
TABLE 1 Themes and corresponding sub-themes identified through thematic analysis.
Factors contributing to user User requirements to feel certain Drone design solutions to address
uncertainty about HDI during HDI uncertainty in HDI
Differences in human roles Tracking information for the recipient Drone appearance
Public space dynamics Recognition of the drone and recipient Human-machine interfaces to communicate drone
intentions
Familiarity with delivery drone technology and Landing/take-off intention of the drone Use case dependency
processes
Criticality of the situation Limited user intervention in drone control
Privacy concerns
4.1.1.4 Criticality of the situation User requirements included information on tracking
Uncertainties and reactions varied depending on the criticality information for the recipient, recognition of the drone and recipient,
of the situation in which the drone delivered packages. As recipients, landing/take-off intentions of the drone, and limited user intervention
participants noted that critical situations increased their uncertainty in drone control. Recognition of the drone and recipient, landing/take-
regarding waiting time, delivery location, and process, stating that off intentions of the drone, and limited user intervention in drone
“when we change the scenario, there is also a change in terms of control werementioned for both roles, while tracking information for
priority” (P1). As bystanders, some participants expressed that they the recipient was specifically required for the role of the recipient. As
would not intervene, others indicated they would step aside for the shown in Table 2, for recipients, recognition of the drone and recipient
drone, and remaining mentioned they would follow the drone to was identified with the highest number of “inclusion requirements”
assist the recipient in an emergency. (22), followed by tracking information for the recipient (17). Limited
A majority of participants noted a potential shift in the role of user intervention in drone controlwas regarded as the least significant
bystander to recipient in critical situations. It was anticipated that a “inclusion requirement” (1) and had the “exclusion requirements”
bystander might need to receive a medical package from the drone (1). For the role of bystander, Table 3 shows that recognition of the
and assist the unwell patient. This role change was expected to lead drone and recipient was given the highest number of “inclusion
to uncertainties regarding the delivery process, interactions with the requirements” (15), followed by landing/take-off intentions of the
drone to collect the package, and the use of medical contents to aid drone (6). Tracking information for the recipient was deemed as not a
the patient. requirement, while limited user intervention in drone control received
“exclusion requirements” (2).
4.1.1.5 Privacy concerns
Both recipients and bystanders expressed privacy concerns.
Recipients were primarily worried about their personal information 4.1.2.1 Tracking information for the recipient
Participants emphasized the need for recipients to have access
being disclosed publicly and preferred discreet, non-intrusive to timely delivery tracking information, including delivery process
communication methods (e.g., using codes instead of announcing and live location, as, “those [pieces of information] are for the
names). Bystanders, on the other hand, were concerned about expectation and [they] will feel more transparent (...)” (P3) and
drones equipped with cameras and the possibility of being recorded. allowing them to plan their time and actions in advance.
They emphasized the need for “reassurance of privacy” (P8) through Information on the delivery process was found to provide
data protection and ethical practices. recipients with “context on the delivery status” (P9) and to help
them understand “the drone is arriving or what is happening”
4.1.2 User requirements to feel certain during (P8). Participants noted the potential for inaccuracies in the
HDI estimated delivery time, drawing examples from current food
Participants suggested that managing uncertainty could involve delivery services. They added that receiving live updates on the
understanding user expectations and requirements, and proposing drone’s location while en route would help them not only track
design solutions accordingly. Participants noted that designs aligned the package and compensate for delivery time inaccuracies but
with user requirements could address uncertainty, promote safety, also distinguish their drone from others and determine its arrival
and help manage anxiety during interactions: direction.This would reduce uncertainty and provide a greater sense
of control.
“As a person I have a lot of anxiety. I would like to know some
stuff beforehand” (P8).
“I think drones are in general still quite dangerous that you 4.1.2.2 Recognition of the drone and recipient
want to make it as safe as possible. So I probably put some A majority of participants highlighted the need for information
requirements” (P10). to recognise both the drone and the recipient. Participants, as
recipients, required details to identify and verify their drone, as
Frontiers in Robotics and AI 07 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
TABLE 2 Number of recipient requirements from the MoSCOW prioritization notes during the interviews.
Sub-theme User requirements
Must have Should have Could have Won’t have
Tracking information for the recipient 8 7 2 0
Recognition of the drone and recipient 12 6 4 0
Landing/take-off intentions of the drone 10 2 0 0
Limited user intervention in drone control 0 0 1 1
The Must Have, Should Have and Could Have requirements (green cells) are referred to as “inclusion requirements” and the Won’t Have requirements (red cells) as “exclusion requirements.”
TABLE 3 Number of bystander requirements from the MoSCOW prioritization notes during the interviews.
Sub-theme User requirements
Must have Should have Could have Won’t have
Tracking information for the recipient 0 0 0 0
Recognition of the drone and recipient 11 2 2 0
Landing/take-off intentions of the drone 5 1 0 0
Limited user intervention in drone control 0 0 0 2
The Must Have, Should Have and Could Have requirements (green cells) are referred to as “inclusion requirements” and the Won’t Have requirements (red cells) as “exclusion requirements.”
well as guidance on interaction. Bystanders, on the other hand, (P2). Awareness of the drone’s presence was considered to reduce
preferred information about the recipient, the drone’s presence, and uncertainty and mitigate surprise, especially for those preoccupied
its purpose in their vicinity. in the park who might not otherwise visually notice the drone:
Participants highlighted the need for recipients to identify and “It’s for the person that is within the space to know what is
verify their drone, in case of a scenario withmultiple delivery drones around them. I think that is a basic—It’s a safety thing. It’s a social
and recipients in a public space were to happen. Identifying the thing” (P5).
drone allows recipients to: “(..) immediately know like, oh, this
[drone] is mine, or this [drone] is not mine and you can act quickly” 4.1.2.3 Landing/take-off intention of the drone
(P4). The verification process was found to enhance safety and Participants, as recipients, expected information on the drone’s
reduce uncertainty, ensuring that the package was not lost and that delivery methods, and both recipients and bystanders were found
the delivery process was complete: to require information on the delivery location through a signal.
Recipients imagined various delivery scenarios, such as the drone
“What if I didn't finish the receiving process? If it starts to fly, landing on the ground, dropping the package, hovering above and
then I will lose my package” (P3). lowering the packagewith a cable, or even “giving it tomy [recipient]
“I put if I have a key to open the drone to receive my package. hand” (P9). Recipients preferred to “(...) have more instructions”
(...) I think it is very important to have it to avoid that (P12) on the delivery methods. In addition, they wanted to be
somebody else will not take your package” (P11). informed about whether they needed “to wait for the drone to
land” (P8) and “how close I [recipient] can approach [the drone]
Especially during the early adoption phases, a tutorial on the or not” (P8).
delivery interaction, the drone and the user tasks was deemed In addition to the delivery methods, information on the delivery
necessary before the drone arrival to facilitate safe interaction, set location, including a landing/take-off signal, was required for both
expectations, save time, and reduce uncertainty. recipient and bystander roles. This information was seen as essential
Most participants, as bystanders, highlighted the need to for allowing recipients to prepare to collect the package and request
understand the drone’s purpose and presence in their vicinity “so a location change if deemed unsafe. Bystanders and external agents
that they are informed about the [basic] intentions” (P5) and to need to be notified that drones will be present near the delivery
address uncertainties, safety, and privacy concerns. They were less location, allowing them to decide whether to move to “less crowded
concerned with specific flying behavior and focusing instead on space or empty space” (P1). Without this information, it is likely to
the drone’s overall activity, such as “whether it is delivering a be “hard to get people to feel happy and give them a safe feeling about
package or emergency or involving in some of the other activities” the process” (P4).
Frontiers in Robotics and AI 08 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
4.1.2.4 Limited user intervention in drone control on both drones and packages to enhance identification and reduce
Overall, recipients and bystanders did not prefer to take over the uncertainty. Participants criticized “classical drones” as “very ugly”
control of the drone. Bystanders showed no interest in interacting (P11) and “mechanical ormasculine” (P12), recommending features
with or controlling the drone and would be annoyed if asked to like propeller guards, rounded shapes, and the avoidance of sharp
do so, except in safety critical situations. In contrast, recipients edges to create a sense of safety. Some proposed that a package
were interested in passive interactions for basic functions, such as attached beneath the drone could signal its delivery function.
verifying delivery with a QR code: “I’ll see my snacks show up, scan
my QR code, and I’m done with this” (P6). They were also willing 4.1.3.2 Human-machine interfaces to communicate
to provide directional suggestions for safe landing if needed: “You drone intentions
just have to tell the drone to go more left, more right, and wait for Participants mentioned the use of HMIs, including audio and
something to drop” (P8). visual elements on the drone, for communicating the drone’s
intentions and establishing a connection with recipients and
4.1.3 Drone design solutions to address bystanders. A participant suggested that the use of visual and audio
uncertainty in HDI interfaces should depend on the distance between the drone and the
Design solutions aligned with user expectations were proposed user: “Maybe when it [drone] is far, it [drone] can have lights and
to address uncertainties in interactions with delivery drones in when it [drone] is close by, it can be sound” (P12). A few participants
public parks. Table 4 shows that solutions for drone appearance recommended using multiple HMIs to provide redundancy and
and HMIs (excluding phones) were mentioned almost equally inclusivity for physically challenged users. However, one participant
for both roles. However, the phone interface was required by all advised against overloading the user with too many interfaces.
participants for the recipient role but mentioned by only one for the The suggested visual interfaces include lights, projections, and
bystander role. displays attached to the drone. Lights were intended to communicate
landing/take-off intentions, maintain safety distance, identify the
4.1.3.1 Drone appearance recipient’s drone, and indicate video recording. Projections on the
Participants expressed the need for the drone’s design to ground were primarily meant to signal the landing location, while
be interaction-friendly and aligned with its delivery purpose, displays were designed to help identify the “correct” drone by
as its appearance impacts their emotions and uncertainties. The showing the recipient’s information. Visual interface advantages
majority suggested using colors and brand stickers on the drone’s include: “If it is just LED lights, it is simple and it can be seen from
body to indicate its delivery purpose and adopting a friendly far away” (P12), andwith a display, “it is easy to change on the drone,
design to enhance safety and approachability. Colors and stickers different recipients, different names” (P11). A few participants raised
were deemed as effective for attracting attention, aiding in the practical concerns about the visibility of visual interfaces: “if it is
identification of the drone’s purpose from a distance, and reducing really bright Sun and in a park, I don’t know how easy is it to see
uncertainty. Participants highlighted the need for intuitive color on the drone what is happening” (P8).
use and visibility considerations based on the delivery location’s The audio interfaces were identified as including both vocal
geography: “If it is about to land in a park, it is gonna be preferable and non-vocal messages. Vocal messages were recommended to
to not be green, for example, (...) [If] I see something red, maybe it convey the drone’s purpose, landing/take-off intentions for safety,
is something to not approach” (P8). and to communicate with bystanders: “By language, tell everybody:
Drawing from familiar food delivery services like DHL and I’m landing, I’m landing (...) If it [drone] needs help then I
UberEats, it was suggested that colors and brand stickers be used would expect voice of sentences rather than just [non-vocal]
sounds” (P10). Some participants found vocal messages to be
intrusive, intimate, anddifficult to hear in “open environments” (P2).
TABLE 4 Number of participants mentioned design solutions during the They recommended using “minimal” (P12) non-vocal messages
interviews for the roles of recipient and bystander. to communicate basic intentions, such as the drone’s presence,
Drone design solutions to Recipient Bystander landing/take-off intentions, and post-delivery acknowledgment.The
address uncertainty in HDI library of non-vocal messages included: “(...) beeping with a low
frequency” (P5), “noise that washing machine does sometimes,
Colors and stickers 7 7 when it is finished” (P11), and “turning off laptop sound” (P12).
Drone appearance While audio interfaces were found to add value by grabbing the
Friendly Looks 6 6
user’s attention, especially when preoccupied, some participants
Lights 4 5 “imagine [that], in public space, sound [audio] would be too
intrusive” (P12).
Projection 3 4 An application on the recipient’s mobile device was a popular
Human-machine suggestion for receiving timely and specific information about
Display 2 2
interfaces to the delivery process and drone’s intentions. Phone application
communicate
Non-vocal message 6 6 (henceforth referred as phone) was considered functional and
drone intentions
simple to use for everyone. The use of the phone was mentioned
Vocal message 4 2 to reduce the necessity for HMIs on the drone, such as displays
and vocal messages, while maintaining privacy. As recipients,
Phone 12 1
participants wanted to receive information on their phones
Frontiers in Robotics and AI 09 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
regarding tracking, identification and verification with the majority of participants suggested incorporating colors, stickers,
drone, precise landing/take-off locations, tutorials on interaction lights, and sirens similar to those on ambulances to convey the
procedures and delivery methods: urgency of the situation. Audio interfaces were considered less
intrusive given the criticality of the situation and were deemed
“I can track onmyphone or the device, how far it is andwhere necessary to “notify that something [emergency] is happening”
it flies, the current location of the delivery” (P3). (P8) and to request bystanders to move away or assist with the
“The app will tell you that the drone is like number 27. There emergency: “It can be like an audio announcement saying, Hey, this
is a number on the drones body, also showing number 27, for person [recipient name] is stuck in this [specific] place, this person
you to identify, and then see if it is the right one” (P10). [recipient name] needs help” (P7).
As bystanders, a majority of participants were not comfortable
using the phone “(...) to know about the drones” (P10) and preferred
to “shift most of the interaction to the drone and move away from 4.2 Focus group results
the app” (P1). A participant, however, mentioned the need for an
application to locate the drones in public space: “I would like to The thematic analysis of drone sketches and storyboards
know, if I go somewhere and see that there aremany deliveries about produced design solutions and interaction procedures aligned with
to happen where I am and I don’t want to have the sound, I would user requirements. Figures 3, 4 illustrate example sketches of a drone
like to go somewhere else” (P8). and an interaction storyboard, respectively. The drone sketches
indicated the types of design solutions, while the storyboards
detailed the stages of interaction and the application of the design
4.1.3.3 Use-case dependent
Participants expressed the need for design solutions to solutions. Reflections on existing drone models were included, with
reflect the situation criticality as the uncertainties and the focus group results labeled as “G” followed by the corresponding
expectations differed: “We should have [the emergency drone] group number.
more distinguishable, so people around also know that this is for
a medical reason and it is not like a bag of chips or my Amazon 4.2.1 Drone sketches
[package]” (P8). Design solutions included protected propellers/wings, encased
The relevance of HMIs and appearance was found to vary for packages, colors/stickers, lights, displays, and projections for safety,
emergency deliveries, in contrast to grocery package deliveries. A recognition and landing/take-off intentions (see Figures 5, 6).
FIGURE 3
Example sketch of a drone, created by Group 4, for the bystander role.
Frontiers in Robotics and AI 10 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
FIGURE 4
Example sketch of a storyboard, created by Group 4, for the bystander role.
Participants suggested using protection around the propellers 4.2.2 Storyboards
and wings (G1, G2, G3) to enhance perceived safety for those in the Based on the thematic analysis of storyboards, the stages of
vicinity of the drone. Encasing the package inside the drone (G1, G2, interaction were identified as Order and Ship, Arrival, Delivery,
G3, G4) was recommended for package safety, while placing lights and Retract for recipients, and Arrival, Delivery, and Retract for
around the propellers (G2) and on the motor (G4) was advised to bystanders. Table 5 indicates that appearance related solutions were
help recipients and bystanders identify the propellers and maintain mentioned only for theArrival stage. Lightswere noted for recipients
a safe distance, reducing safety concerns when the drone enters a across the Arrival, Delivery, and Retract stages, but only during the
public park space for delivery. Retract stage for bystanders. Projection was identified for recipients
Colors and (brand) stickers were suggested on the body and in the Arrival and Delivery stages, while for bystanders, it was
wings of the drone to indicate the delivery purpose, with two relevant only during the Delivery stage. All groups utilized phone
groups (G1, G3) recommending this for recipients and four groups interface cards for the recipient role, but none did so for the
(G1, G2, G3, G4) for bystanders. Lights on the propellers (G3), bystander role. The uncertainties were reported below for every
a display attached under the nose of the drone (G2), and ground stage. The following text (reported often in past tense form) outlines
projections (G4) were recommended to help recipients identify their the interaction stages based on the storyboarding results, reflecting
respective drones. The visual interfaces were intended to work in participants’ expectations rather than actual experiences.
conjunction with the phone; for example, the recipient receives a
delivery-specific code on their phone that is reflected on the visual During the Order and Ship stage, the recipient places the
interfaces. For the package verification process, a QR code was delivery order and receives tracking information.
recommended as either a sticker on the drone (G2) or a projection
on the ground (G4), which recipients could scan with their phones. Participants were uncertain about the delivery process and
Ground projections (G1, G2, G4) were recommended to indicate methods of interaction. The order was placed on the phone by
landing/take-off intentions to both recipients and bystanders, the recipient with the required details similar to the current food
while lights on the motor mount (G3) were suggested specifically delivery applications (e.g., UberEats; G1, G2, G3, G4). Once the
for recipients. order was accepted, the recipient received the delivery tracking
Frontiers in Robotics and AI 11 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
FIGURE 5
Sample display of design solutions proposed by participants using drone sketches for the recipient role. Numbers in brackets represent the number of
group suggestions for each solution. For lights, (A–D) denote the light interfaces around the protected propellers, on the motor, on the motor mount,
and on the propellers, respectively.
FIGURE 6
Sample display of design solutions proposed by participants using drone sketches for the bystander role. Numbers in brackets represent the number of
group suggestions for each solution. For lights, (A, B) denote the light interfaces around the protected propellers and on the motor, respectively.
Frontiers in Robotics and AI 12 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
TABLE 5 Number of groups mentioned design solutions for each human role and each stage of interaction. R and B represent recipient and bystander
roles, respectively.
Design solutions Stages of interaction and human roles
Order and Arrival Delivery Retract
ship
R B R B R B R B
Colors and stickers 0 0 2 4 0 0 0 0
Appearance
Friendly Looks 0 0 3 3 0 0 0 0
Lights 0 0 1 0 1 0 1 1
Projection 0 0 1 0 3 3 0 0
HMIs Display 0 0 1 0 0 0 0 0
Audio message 0 0 0 0 2 1 1 1
Phone 4 0 2 0 4 0 1 0
information, such as order details (G1, G2, G3, G4), instructions on Participants expressed uncertainty about the delivery location
the interaction procedure (G3, G4), estimated time of arrival (G1, and concerns regarding bystanders and external agents potentially
G2), flying route (G2, G3), and drone identification details (G2, G3) interrupting the delivery process by entering the interaction
on the phone. While the drone flew to the delivery address (G1, G2, space (G2, G4).
G3, G4), the recipient waited in anticipation. The delivery process was initiated by the recipient after phone
confirmation (G1, G3, G4). The delivery location was highlighted
In the arrival stage, the recipient identifies the drone, whereas by the drone through lights or a projection on the ground while
the bystander identifies the recipient, drone’s presence it hovered above and observed by the recipient (G1, G2, G4). The
and purpose. package was then delivered either by landing (G1, G2, G4) or by
using a cable to drop it (G3), with lights on the motor mount and
Participants (G3) were uncertain about being unable to identify audio used as a landing signal (G4). The recipient was provided the
the drone and were concerned about the drone attracting attention ability to terminate the delivery process via phone (G1), if deemed
and eliciting reactions frombystanders and external agents (G2,G4). unsafe. Verification with the drone was performed by the recipient
Notifications were received on the phone when the drone was either before the delivery by scanning aQR code sticker on the drone
close by, which helped the recipient prepare for the interaction (G1). (G2) or as a projection (G4), or after the delivery by confirming on
As the drone arrived in the park, it was identified by the recipient their phone (G1, G3).
using a code through lights on the propellers, colors/stickers, display, The delivery location was highlighted by the drone with a
or projection (G2, G3, G4). If not identified, the service was projection on the ground (G1, G2, G4), and the bystanders and
contacted via phone (G3). Propeller guards/wings and lights on the external agents were expected to stay away or not to avoid hindering
propeller guards and motor were used to reflect a “safe” interaction the delivery process. The bystander showed little concern about the
(G1, G2, G4). Once identified, the drone either located a precise delivery method (G1, G2, G3). The drone delivered the package with
delivery location (G2, G3, G4) or the recipient selected a location projection and audio used as a landing signal (G4). In scenarios where
from those suggested on the phone (G1) andmoved to that location. hindrance was possible, one group (G4) indicated that the recipient
The responsibility for identifying the delivery location was handed wasresponsible forclearing thehindranceandpreventing interruption
to the drone by three groups and to the recipient by one group. of the delivery process, while three groups (G1,G2,G3) suggested that
Bystander identified the recipient in the vicinity and the the responsibility lay with the drone rather than the recipient.
presence of the approaching drone with propeller noise and
its purpose through colors/stickers (G1, G2, G3, G4). Different In the retract stage, the recipient and bystander observe the
reactions were observed from bystanders (and external agents): drone retracting and flying away from their vicinity.
some were curious and approached the drone, some were annoyed
and sought drone-free areas, and some were tempted to place an Participants were uncertain about potential interruptions to
order later (G2, G4). the retracting process by bystanders and external agents (G2, G4)
and about the drone retracting before the recipient completed the
During theDelivery stage, the recipient identifies the delivery package collection (G4).
location, verifies with the drone and handles the delivery After the delivery, the recipient scanned the area for a safe
procedure, while the bystander observes from afar. take-off and confirmed readiness via phone (G2), or the drone
Frontiers in Robotics and AI 13 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
autonomously scanned the area for a safe take-off without recipient and bystanders, drone flying behavior, identifying their service
involvement and waited for a fixed duration for package pickup drone, and determining their positioning for a safe pickup.
(G4). While not explicitly mentioned by two groups (G1, G3), the While previous studies have individually studied the roles of
responsibility for ensuring a safe take-off was assigned to the drone bystanders (Pelikan et al., 2024; Yu et al., 2024) and recipients
by one group (G4) and to the recipient by another group (G2). (Tan et al., 2018; Yeh et al., 2017) in HRI, our study offers a
Before the drone retracted, the recipient and bystander observed novel contribution by discussing the interplay between these roles,
the drone using lights on the motor or an audio message to request the differing levels of uncertainty experienced with both the roles
clearing the landing location (G2, G4). The location was vacated and presenting their underlying causes. Future research in HRI
by recipients, bystanders, and any external agents present, and the should focus more on codesign that incorporates both recipient
drone retracted. and bystander perspectives to improve public acceptance of service
robots in public spaces.
4.2.3 Reflection on the existing drone models Participants explained that current commercial drone designs
Overall, participants found the current designs of Amazon, (e.g., Amazon Prime, Mana, Wing, Zipline) are better suited for
Mana, Wing, and Zipline drones to be more suitable for home home deliveries than public spaces, where uncertainty is expected to
deliveries with open spaces, where recipients are familiar with the be greater.The evolving dynamics of public spaces—such asmultiple
exact delivery location and purpose, rather than for public park recipient groups, pets, and environmental factors—contribute
deliveries. The perceived differences in drone sizes led participants to this uncertainty, complicating interactions and raising safety
to associate them with different use cases; for example, the Amazon concerns. While Nielsen et al. (2023) found that over 60% of
drone was seen as large and suitable for delivering bigger packages, interaction breakdowns with public service robots are due to
while the smaller Mana drone was deemed more appropriate environmental disturbances, our study extends this by identifying
for snack deliveries (G1). The dead-drop delivery method onto the specific causes of uncertainty stemming from public space
designated landing pads raised safety concerns (G1, G2, G3, G4), dynamics. Future research should conduct naturalistic studies
and the necessity of carrying landing pads in parks was criticized involving public interactions with drones, investigating how factors
due to the efforts required to transport the pads (G1, G2). The like multiple recipient groups, pets, and environmental elements
visible package on the Wing drone and the cable delivery methods influence uncertainties for both recipients and bystanders. This will
of Wing and Mana led participants to believe that these drones help guide the design of safer drones for public environments.
would perform well in pleasant weather but could struggle in harsh Uncertainties varied across different stages of interaction,
wind conditions (G1, G2, G3). Participants recommended encased includingOrder and ship, Arrival, Delivery, andRetract, aligningwith
packages for safety, as seen in the designs of the Mana and Zipline thefindingsofLingametal. (2025)fortheArrival,Delivery,andRetract
drones (G1, G2, G3). stages. Consequently, user requirements differed, prompting tailored
Participants criticized the Mana and Wing drone designs for design recommendations to manage uncertainties in each stage.
lacking propeller and wing protection (G1, G3) and found the
Amazon drone’s protective design to be sturdy and safe (G1,
G2, G3, G4). Wing drone was not preferred for its unprotected 5.1 Order and ship stage
propellers, intrusive, sharp-edged, and mechanical appearance
(G1, G2, G3, G4). The intimidating size of the Amazon drone The interaction with the recipient begins passively even
discouraged hovering during the delivery phase (G1, G2, G3, G4). before the drone arrives, building expectations that help manage
The Zipline drone design was viewed as safe and appealing due to uncertainties. Recipients expressed uncertainty about the delivery
its futuristic design and mini-droid feature (G1, G2, G3, G4), which methods, drone size, and interaction procedures, highlighting the
allowed precise package delivery and was deemed stable for windy need for tutorials that outline the process and align their mental
conditions (G1, G3). Participants recommended the design focus to models prior to the drone’s arrival. It is recommended to share
shift from themain drone to themini-droid, suggesting zoomorphic standardized tutorials alongside package tracking information on
features for improved user appeal (G1, G3). the phone application where the order is placed. This underscores
the need to provide information to build recipient expectations
before the onset of HDI, a novel observation that adds to the
5 Discussion current HDI studies (Bevins and Duncan, 2021; Obaid et al., 2015;
Tan et al., 2018; Yeh et al., 2017) that have primarily focused on the
Our user-centered design study, using individual interviews and interaction element.
focus groups, identified key uncertainty factors, user requirements,
and design solutions for delivery drones interacting with humans Recommendation 1: Recipients should be informed on the
in public spaces like parks. Among the uncertainty factors, interaction protocols via phone before the drone arrives.
participants highlighted how uncertainty levels varied depending on
human roles and the dynamic nature of public spaces. Bystanders,
who rarely expect engagement with drones, experience higher
uncertainty compared to recipients, who anticipate deliveries 5.2 Arrival stage
and face fewer uncertainties. Bystanders expressed uncertainties
regarding the drone’s purpose, noise, and the delivery location. In According to Lingam et al. (2025), recipients experience the
contrast, recipient uncertainties focused on safety of themselves most uncertainty during the Arrival stage. Our study reveals
Frontiers in Robotics and AI 14 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
that recipients may struggle with identifying the “correct” drone, and recommending the use of projection to communicate drone
particularly in the presence of multiple recipient groups, while intentions about occupying ground space (e.g., package drop-off
bystanders may feel uncertain about the drone’s presence and spot), thereby reducing uncertainty and guiding humans tomaintain
purpose. These uncertainties cause interruptions and safety issues, a safe distance.
particularly with unpredictable behavior of bystanders or external The phone application should include a customised feature that
agents. In order to address these concerns, drones should use allows recipients to verify and, if necessary, interrupt deliveries for
visual aids like stickers and colors to indicate purpose and safety.This aligns with the current safety standards of global delivery
non-vocal audio cues, such as propeller noise, for presence. companies, such asUPS (2025) andDHL (2025), which use recipient
This recommendation based on our user-centered study aligns verification by delivery truck drivers to preventmisplaced deliveries.
with the expert recommendations of Lingam et al. (2024). Extending this approach to the context of automated drone delivery
It is further recommended to draw design inspiration from using a mobile device for recipient verification is a novel finding.
delivery and taxi services, incorporating familiar elements such Lights and audio warnings should also be integrated into the drone’s
as bike or taxi designs and application interfaces, to reduce design to inform humans to maintain a safe distance during the
uncertainty. drone’s retraction. Standardized visual and audio interfaces for
Designs should prioritise safety and convey “friendly” landing and take-off intentions would allow recipients, bystanders,
intentions, incorporating protective features around propellers and and other external agents to quickly and accurately understand
wings, contrary to the recommendation by Wojciechowska et al. the drone’s intentions, reducing uncertainty, and enhancing safety.
(2019) to exclude propeller guards. This difference Future research should evaluate which interfaces (e.g., lights versus
arises because Wojciechowska et al. (2019) used static images of audio messages) are most effective for communicating specific
similarly sized drone models without providing context on their landing or take-off information, including notifying recipients
delivery use case, whereas our study presented information and about package collection and alerting bystanders to keep a
videos that conveyed both the delivery context and an estimate safe distance.
of drone size. This highlights the importance of informing the
public about the delivery context and drone characteristics to better Recommendation 3: Humans should be informed of the
understand their needs on design elements and reduce uncertainty delivery location through ground projections, and the
in public spaces. drone should only proceed with the delivery after receiving
Standardized visual and audio cues can help humans identify the confirmation from the recipient via phone. Additionally,
drone’s purpose and presence, while customized interfaces can assist lights and audio warnings should be employed to keep
recipients in distinguishing the “correct” drone. While previous bystanders and external agents at a safe distance.
studies (Tan et al., 2018; Wojciechowska et al., 2019; Yeh et al.,
2017) provided design recommendations for one-to-one HDI, our Participants are divided on defining control responsibilities
study uniquely contributes by recommending customizable cues during the Arrival, Delivery, and Retract stages. Some felt that
that facilitate drone identification as the scale of drones and of recipients should be responsible for identifying the delivery location
recipient orders increase. The above design solutions should be and ensuring safe landing and take-off, while others expressed
tested in user studies to assess how quickly the solutions instill a that the drone should handle these tasks. This division may
sense of certainty among users. stem from a lack of familiarity with drone capabilities in public
spaces. Unclear control responsibilities can increase uncertainty and
miscommunication for humans, leading to interaction breakdowns
Recommendation 2: Humans should be informed about during critical moments. Future studies should explore how
the drone’s purpose through the appearance, presence by humans respond to different control assignments, whether to
propeller noise, and identification through appearance and the drone, the recipient, or shared between them, examine the
visual interfaces when the drone arrives. resulting uncertainties, and work towards standardizing control
responsibilities to enhance safety.
5.3 Delivery and retract stage
5.4 Limitations and future research
During this stage, uncertainty primarily stems from difficulties
in identifying the delivery location and concerns about bystanders Although the authors in our study made efforts to minimize
and external agents interrupting the delivery process, which can lead confirmation bias and enhance diversity in the interpretation by
to safety issues and interaction breakdowns. Lingam et al. (2025) involving multiple coders, thematic analysis inherently involves
found that recipients experience high levels of uncertainty before the subjective interpretation of the data (Braun and Clarke, 2006).
drone attempts to deliver and retract. It is recommended that drones The qualitative nature of the research introduces potential biases,
use ground projections to indicate delivery locations and provide which could influence the results and limit the generalizability
audio warnings for safety, inline with Lingam et al. (2024). Previous of the findings. Future research should examine the extent to
user studies in public spaces have used projection to receive recipient which the viewpoints presented in our study can be generalized
input for task execution (Cauchard et al., 2019) or to guide recipients to the broader public by conducting large-scale studies on HDI in
in discarding garbage by highlighting it (Obaid et al., 2015). Our simulated environments. Additionally, employing mixed methods,
study adds value by incorporating the perspective of bystanders involving Likert scale statements or behavioral observations around
Frontiers in Robotics and AI 15 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
drones, could provide objective measures to complement the for conveying specific types of information and facilitating
qualitative insights. communication.
Interviews and focus groups required participants to
envision drone interactions, and their limited experience with
drones may have influenced their ability to propose effective Data availability statement
design solutions. Additionally, the proposed design solutions
need empirical validation. Future research should refine these The data analysed in the study are included in the
solutions into prototypes or conceptual models and evaluate Supplementary Material. Further inquiries can be directed to the
them through user testing in virtual environments or real-world corresponding author.
experiments, for delivery scenarios involving both recipients and
bystanders.
While our study reflects the limited exposure to drone deliveries, Ethics statement
representing the perspective of novice users in public spaces, it is
limited in considering the perspective of experienced users, such as The study involving humans was approved by Ethical
drone pilots. Such users could deepen the observations by reflecting Review Board, Eindhoven University of Technology, Eindhoven,
on their experiences with interaction scenarios, challenges, HMI Netherlands. The study was conducted in accordance with the local
requirements, and their implications. Future research should legislation and institutional requirements.The participants provided
extend our work by including participants with experience their written informed consent to participate in this study.
flying drones.
The participants’ design background enabled them to formulate
concrete design solutions with rationale; however, it might have
biased their suggestions toward existing design principles. To Author contributions
minimize this bias and improve the validity, future research
should develop concepts based on our design recommendations SNL: Conceptualization, Data curation, Formal Analysis,
and evaluate them with participants from non-design Investigation, Methodology, Project administration, Software,
backgrounds. Visualization, Writing – original draft, Writing – review and
Participants in our interviews stressed designing systems editing. RV: Conceptualization, Data curation, Formal Analysis,
based on situation criticality and transitions from bystander to Investigation,Methodology, Software, Validation,Writing – original
recipient roles, as user needs vary with urgency. While our draft, Writing – review and editing. SMP: Funding acquisition,
study offered preliminary design directions, the focus groups did Methodology, Project administration, Resources, Supervision,
not comprehensively address this aspect due to time constraints. Writing – original draft,Writing – review and editing.MM: Funding
It is recommended to explore how varying levels of situation acquisition, Methodology, Project administration, Resources,
criticality impact user requirements for both recipients and Supervision, Writing – review and editing.
bystanders, and develop design recommendations for HDI in
public spaces.
Funding
6 Conclusion The author(s) declare that financial support was received
for the research and/or publication of this article. This research
Our user-centered design study, conducted through interviews work was supported by the Dutch Research Agenda (NWA)
and focus groups, identified key uncertainty factors and user under Dutch Research Council (NWO, the Netherlands)
requirements, providing design recommendations for interactions through the project Safety Solutions for Autonomous Vehicle
between delivery drones and humans (recipients and bystanders) Integration in Urban Mobility: Efficient and Reliable Acting
in public parks. It is crucial to address these aspects for the in an Uncertain and Unreliable World under case number
effective integration of delivery drones into public environments. NWA.1292.19.298.
The study identified that information needs and preferred
interface modalities vary between recipients and bystanders
and across different interaction stages. The study highlighted Acknowledgments
the necessary design features that require standardization and
customisation to support the development of effective design We thank Tram Thi Minh Tran for providing feedback on the
guidelines and improve natural HDI in public spaces. Drone study background and methods.
designers may face the challenge of implementing these features
and addressing human requirements, particularly related to
uncertainty and safety concerns. Prior to the implementation, Conflict of interest
future research should validate the proposed recommendations
through experimental studies involving interactions between The authors declare that the research was conducted in the
different human roles and delivery drones. Furthermore, research absence of any commercial or financial relationships that could be
is necessary to identify which interfaces are most effective construed as a potential conflict of interest.
Frontiers in Robotics and AI 16 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
Generative AI statement organizations, or those of the publisher, the editors and the
reviewers. Any product thatmay be evaluated in this article, or claim
Theauthor(s) declare thatGenerativeAIwas used in the creation thatmay bemade by itsmanufacturer, is not guaranteed or endorsed
of this manuscript. ChatGPT was used to enhance the language by the publisher.
of the text.
Supplementary material
Publisher’s note
The Supplementary Material for this article can be found
All claims expressed in this article are solely those of the online at: https://www.frontiersin.org/articles/10.3389/frobt.2025.
authors and do not necessarily represent those of their affiliated 1580289/full#supplementary-material
References
Alon, O., Rabinovich, S., Fyodorov, C., and Cauchard, J. R. (2021). “Drones the 2017 CHI Conference on Human Factors in Computing Systems, 6794–6799.
in firefighting: a user-centered design perspective,” in Proceedings of the doi:10.1145/3025453.3025755
23rd International Conference on Mobile Human-Computer Interaction, 1–11.
doi:10.1145/3447526.3472030 Kantola, N., and Jokela, T. (2007). “Svsb: simple and visual storyboards: developing
a visualisation method for depicting user scenarios,” in Proceedings of the 19th
Amiche, F., Juillot, M., Vigné, A., Brock, A. M., and Clodic, A. (2024). “First steps of Australasian Conference on Computer-Human Interaction: Entertaining User
designing adaptable user interfaces to mediate human-robot interaction,” in 2024 IEEE Interfaces, 49–56. doi:10.1145/1324892.1324901
Conference on Telepresence, 192–196. doi:10.1109/Telepresence63209.2024.10841532 Karjalainen, K. D., Romell, A. E. S., Ratsamee, P., Yantac, A. E., Fjeld, M., and
Bevins, A., and Duncan, B. A. (2021). “Aerial flight paths for communication: how Obaid, M. (2017). “Social drone companion for the home environment: a user-centric
participants perceive and intend to respond to drone movements,” in Proceedings of exploration,” in Proceedings of the 5th International Conference on Human Agent
the 2021 ACM/IEEE International Conference on Human-Robot Interaction, 16–23. Interaction, 89–96. doi:10.1145/3125739.3125774
doi:10.1145/3434073.3444645 Khan, M. N. H., and Neustaedter, C. (2019). “An exploratory study of the use
Braun, V., and Clarke, V. (2006). Using thematic analysis in psychology. Qual. Res. of drones for assisting firefighters during emergency situations,” in Proceedings
Psychol. 3 (2), 77–101. doi:10.1191/1478088706qp063oa of the 2019 CHI Conference on Human Factors in Computing Systems, 1–14.
Buskermolen, D. O., and Terken, J. (2012). Co-constructing stories: a participatory doi:10.1145/3290605.3300502
design technique to elicit in-depth user feedback and suggestions about design Lazar, J., Feng, J. H., andHochheiser,H. (2017).Researchmethods in human-computer
concepts. Proceedings of the 12th ParticipatoryDesignConference: Exploratory Papers, interaction 2nd Edn. Cambridge, MA: Elsevier: Morgan Kaufmann.
Workshop Descriptions, Industry Cases 2, 33–36. doi:10.1145/2348144.2348156 Lee, J. D., and See, K. A. (2004). Trust in automation: designing for appropriate
Cauchard, J. R., Tamkin, A., Wang, C. Y., Vink, L., Park, M., Fang, T., et al. (2019). reliance. Hum. factors 46 (1), 50–80. doi:10.1518/hfes.46.1.50.30392
“Drone. io: a gestural and visual interface for human-drone interaction,” in 14th
ACM/IEEE International Conference on Human-Robot Interaction (HRI), 153–162. Lindley, D. V. (2013). Understanding uncertainty. John Wiley & Sons.
doi:10.1109/HRI.2019.8673011 Lingam, S. N., Franssen, M., Petermeijer, S. M., and Martens, M. (2024). Challenges
Clegg, D., and Barker, R. (1994). Case method fast-track: a RAD approach. Addison- and future directions for human-drone interaction research: an expert perspective. Int.
Wesley Longman Publishing Co., Inc. J. Human Computer Interact., 1–17. doi:10.1080/10447318.2024.2400756
Clothier, R. A., Greer, D. A., Greer, D. G., and Mehta, A. M. (2015). Risk Lingam, S. N., Petermeijer, S.M., Torre, I., Bazilinskyy, P., Ljungblad, S., andMartens,
perception and the public acceptance of drones. Risk Anal. 35 (6), 1167–1183. M. (2025). Behavioral effects of a delivery drone on feelings of uncertainty: a virtual
doi:10.1111/risa.12330 reality experiment. ACM Trans. Human Robot Interact. doi:10.1145/3729538
DHL (2025). Deliver parcels safely with Ident-Check. Available online at: https:// Ljungblad, S., Man, Y., Baytaş, M. A., Gamboa, M., Obaid, M., and Fjeld, M. (2021).
www.dhl.de/en/geschaeftskunden/paket/leistungen-und-services/services/service- “What matters in professional drone pilots’ practice? An interview study to understand
loesungen/ident-check (Accessed March 28, 2025). the complexity of their work and inform human-drone interaction research,” in
Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems,
Federal Aviation Administration (2024). Is package delivery allowed under the 1–16. doi:10.1145/3411764.3445737
Small UAS Rule. Part (107)? Available online at: https://www.faa.gov/faq/package-
delivery-allowed-under-small-uas-rule-part-107#:∼:text=Getting%20Started-,Is%20 Matternet (2024). Matternet: about product. Available online at: https://www.mttr.
package%20delivery%20allowed%20under%20the%20Small%20UAS%20Rule%20 net/ (Accessed February 07, 2025).
(Part,drone%20within%20his%2Fher%20sight (Accessed March 26, 2025). Meissner, A., Trübswetter, A., Conti-Kufner, A. S., and Schmidtler, J. (2020). Friend or
Fink, F., Paulicke, D., Grünthal, M., and Jahn, P. (2023). “Of course, drones delivering foe? understanding assembly workers’ acceptance of human-robot collaboration. ACM
urgent medicines are necessary. But I would not use them until” Insights from a Trans. Human Robot Interact. 10 (1), 1–30. doi:10.1145/3399433
qualitative study on users’ needs and requirements regarding the use of medical drones. Momon, A. (2014). Drones for good. Master’s thesis. Delft, Netherlands: Delft
PLos One 18 (5), e0285393. doi:10.1371/journal.pone.0285393 University of Technology. Available online at: http://resolver.tudelft.nl/uuid:36ce77ad-
Franke, T., Attig, C., and Wessel, D. (2019). A personal resource for 1b06-4149-8da4-a231dcdfec69 (Accessed February 07, 2025).
technology interaction: development and validation of the affinity for technology Netherlands Enterprise Agency (2024). Rules for flying drones. Available
interaction (ATI) scale. Int. J. Human Computer Interact. 35 (6), 456–467. online at: https://business.gov.nl/regulation/drones/?gad_source=1&gclid=CjwKCA
doi:10.1080/10447318.2018.1456150 jw68K4BhAuEiwAylp3ktRHcDphMcDlHhX2-b-s5Dv3rW7qmfGIBJdtJiKgoz28CZ
Hallewell, M. J., Hughes, N., Large, D. R., Harvey, C., Springthorpe, J., and hDIhrDQBoCs44QAvD_BwE (Accessed March 25, 2025).
Burnett, G. (2022). Deriving personas to inform HMI design for future autonomous Nielsen, S., Skov, M. B., Hansen, K. D., and Kaszowska, A. (2023). Using user-
taxis: a case study on user requirement elicitation. J. Usability Stud. 17 (2), 41–64. generated YouTube videos to understand unguided interactions with robots in public
doi:10.5555/3542830.3542832 places. ACM Trans. Human Robot Interact. 12 (1), 1–40. doi:10.1145/3550280
Herdel, V., Yamin, L. J., and Cauchard, J. R. (2022). “Above and beyond: a scoping Obaid, M., Mubin, O., Basedow, C. A., Ünlüer, A. A., Bergström, M. J.,
review of domains and applications for human-drone interaction,” in Proceedings and Fjeld, M. (2015). “A drone agent to support a clean environment,” in
of the 2022 CHI Conference on Human Factors in Computing Systems, 1–22. Proceedings of the 3rd International Conference on Human-Agent Interaction, 55–61.
doi:10.1145/3491102.3501881 doi:10.1145/2814940.2814947
Herdel, V., Yamin, L. J., Ginosar, E., and Cauchard, J. R. (2021). “Public Pelikan, H. R., Reeves, S., and Cantarutti, M. N. (2024). “Encountering autonomous
drone: attitude towards drone capabilities in various contexts,” in Proceedings of robots on public streets,” in Proceedings of the 2024 ACM/IEEE International
the 23rd International Conference on Mobile Human-Computer Interaction, 1–16. Conference on Human-Robot Interaction, 561–571. doi:10.1145/3610977.3634936
doi:10.1145/3447526.3472053 Rodak, A., Jamson, S., Kruszewski, M., and Pędzierska, M. (2020). “User
Jane, L. E., Ilene, L. E., Landay, J. A., and Cauchard, J. R. (2017). “Drone & requirements for autonomous vehicles–A comparative analysis of expert and non-
wo: cultural influences on human-drone interaction techniques,” in Proceedings of expert-based approach,” in 2020 AEIT International Conference of Electrical
Frontiers in Robotics and AI 17 frontiersin.org
Lingam et al. 10.3389/frobt.2025.1580289
and Electronic Technologies for Automotive (AEIT AUTOMOTIVE), 1–6. Vos, S., Janssen, M., Goudsmit, J., Lauwerijssen, C., and Brombacher, A. (2016).
doi:10.23919/AEITAUTOMOTIVE50086.2020.9307415 From problem to solution: developing a personalized smartphone application for
Rosenthal-von Der Pütten, A., Sirkin, D., Abrams, A., and Platte, L. (2020). “The recreational runners following a three-step design approach. Procedia Eng. 147,
forgotten in HRI: incidental encounters with robots in public spaces,” in Companion of 799–805. doi:10.1016/j.proeng.2016.06.311
the 2020 ACM/IEEE International Conference on Human-Robot Interaction, 656–657. Vuleta, B. (2021). Drone statistics: commercial and governmental use. Available
doi:10.1145/3371382.3374852 online at: https://seedscientific.com/drone-statistics/ (Accessed September 04, 2024).
Shankland, S. (2023). Zipline’s new drone brings aerial delivery closer for millions Wing (2024). Learn about how Wing delivery works. Available online at: https://
more of us. Available online at: https://www.cnet.com/tech/computing/new-zipline- wing.com/about-delivery/ (Accessed February 06, 2025).
drone-brings-aerial-delivery-closer-for-millions-more-of-us/ (Accessed February 06,
2025). Wojciechowska, A., Frey, J., Mandelblum, E., Amichai-Hamburger, Y., and Cauchard,
J. R. (2019). Designing drones: factors and characteristics influencing the perception of
Shapira, S., andCauchard, J. R. (2022). Integrating drones in response to public health flying robots. Proceedings of the ACMon Interactive,Mobile,Wearable andUbiquitous
emergencies: a combined framework to explore technology acceptance. Front. Public Technologies 3 (3), 1–19. doi:10.1145/3351269
Health 10, 1019626. doi:10.3389/fpubh.2022.1019626 Yeh, A., Ratsamee, P., Kiyokawa, K., Uranishi, Y., Mashita, T., Takemura, H., et al.
Tan, H., Lee, J., and Gao, G. (2018). “Human-drone interaction: drone delivery and (2017). “Exploring proxemics for human-drone interaction,” in Proceedings
services for social events,” in Proceedings of the 2018 ACM Conference Companion of the 5th International Conference on Human Agent Interaction, 81–88.
Publication on Designing Interactive Systems, 183–187. doi:10.1145/3197391 doi:10.1145/3125739.3125773
.3205433 Yu, X., Tran, T. T. M., Wang, Y., Mah, K., Cao, Y., Johansen, S. S., et al. (2024). “Out
Truong, K. N., Hayes, G. R., and Abowd, G. D. (2006). “Storyboarding: an empirical of place robot in the wild: envisioning urban robot contextual adaptability challenges
determination of best practices and effective guidelines,” in Proceedings of the 6th through a design probe,” in Extended Abstracts of the CHI Conference on Human
Conference on Designing Interactive Systems, 12–21. doi:10.1145/1142405.1142410 Factors in Computing Systems, 1–7. doi:10.1145/3613905.3651002
UPS (2025). Identification requirements. Available online at: https://www.ups. Zenz, A., and Powles, J. (2024). Resisting technological inevitability: Google
com/nl/en/support/shipping-support/id-requirements-for-tendering-shipment Wing’s delivery drones and the fight for our skies. Philos. Trans. A 382, 20240107.
(Accessed March 28, 2025). doi:10.1098/rsta.2024.0107
Frontiers in Robotics and AI 18 frontiersin.org