MobileHCI '25 Adjunct: Adjunct Proceedings of the 27th International Conference on Mobile Human-Computer Interaction

To enable automated testing of gesture-based applications, synthetic touch gestures must closely mimic real touch gestures. Although various generation methods exist, comparing them remains difficult due to the lack of standardized evaluation procedures. This work builds upon an existing initial evaluation framework that organizes relevant metrics into categories based on a taxonomy. We specify open aspects of the framework to support its practical and consistent application. We also propose a hierarchical structure for weighting and comparing metrics across categories. Finally, we outline how the framework can be extended to evaluate multi-stroke and multi-touch gestures.

Locomotion is a key factor in virtual reality (VR) navigation but remains underexplored in asymmetric setups combining VR and mixed reality (MR). Our study investigates whether visualized teleportation can serve as a middle ground, balancing the social connection afforded by real walking with the efficiency of teleportation within asymmetric VR-MR setups. We conducted a mixed-design study (N = 24, 12 pairs) to examine how locomotion methods (teleport, real walking, visualized teleport), platform (VR vs. MR), and task type (social vs. navigation) affect user experience and performance. Results show that locomotion technique significantly influences both user experience and navigation effectiveness. Our findings highlight visualized teleport as a promising compromise, offering improved presence and user experience without imposing physical demand, providing practical guidance for designing locomotion in asymmetric mixed reality environments.

The integration of Large Language Models (LLM) into systems that combine voice and graphical user interfaces causes a paradigm shift from designing for humans only to designing for collaborative interaction. However, the non-deterministic, natural language-based nature of LLMs also causes conflicts with the deterministic, structured GUI interaction. We argue that to seize collaboration and mitigate conflicts, UX designers must design the LLM as a second user, creating one interface for human users and one for the LLM. We designed an exemplary prototype as a case study, identifying through this hands-on process three unique challenges for practical UX design: (a) Since humans express objectives rather than commands, the information architecture should be intention-based; (b) the system should translate interactions between human and AI to foster mutual understanding; and (c) the natural language dialog with the LLM should be integrated with the structured interaction typical for GUIs through targeted use of modalities.

With the widespread use of online conferencing tools like Zoom on both PCs and mobile devices, there is a growing risk of unintentional exposure of private information through screen sharing. To address this issue, automated systems have been developed to detect and obscure private information on screens. However, the effectiveness of visual obfuscation methods—such as masking and blurring—has not been thoroughly evaluated. This study aims to identify an effective obfuscation method that achieves a balance between privacy protection and viewing experience for on-screen information. We conducted an online experiment with 55 participants, in which six types of filters were evaluated, including variations in blurring, pixelation, and masking strength and color. The results indicate that users preferred filters with high privacy protection, particularly those that appeared more natural. Among the tested methods, background-color masking received the highest ratings for both privacy protection and viewing experience, suggesting that it is a well-balanced obfuscation method.

Connecting personal devices to the in-vehicle infotainment system has become mainstream in modern vehicles, contouring a distinctive context of use characterized by the user interface being distributed across multiple interactive systems, including those in the vehicle and the users’ personal digital devices, likely involving different input and output modalities. However, distributed user interfaces (DUIs), despite extensively studied in other application domains, have not been addressed to the same extent for in-vehicle interactions. In this context, we examine in-vehicle DUIs and report the results of an exploratory study conducted with twenty-four drivers, who shared their preferences regarding digital device use inside the vehicle. To complement our findings, we also present a demonstrative application featuring a user interface with interaction modalities distributed across the in-vehicle infotainment system and the driver’s smartwatch.

Drones will deliver packages in public spaces and interact with humans as recipients of the packages and as bystanders passing by. Clear communication of drone intentions is critical for reducing uncertainty and ensuring public safety. Limited research exists on interface designs addressing communication with diverse human roles. This user-centered study introduces six interface concepts utilizing lights, displays, and projection technologies to communicate delivery intentions with both recipients and bystanders. The concepts also offer cues for vertical movements, proximity-based warnings, and spatial guidance. An online survey demonstrated that all six concepts were perceived as more intuitive, clearer, and socially acceptable than having no interface. Projections were favored for requiring less visual scanning, displays for conveying direction and progress, and lights for offering familiar cues. Specifically, helipad-like projection was appreciated for providing information that could guide recipients toward the package and signal bystanders to maintain distance. Future research should address practical implementation challenges and potential interpretation issues, particularly for bystanders positioned farther from the drone.

This paper explores the use of triboelectricity to enable gesture-based interactions with interior textile surfaces. We present low-cost, self-powered electrostatic (e-static) sensors built using off-the-shelf textiles and simple construction tools, making triboelectric nanogenerator (TENG)-based sensing more accessible for e-textile-focused HCI research. The sensor design is robust and customizable, supporting both single- and multi-channel configurations that can recognize multiple gestures and capture qualitative aspects such as interaction speed. We show that gesture recognition remains effective even when the sensor’s physical properties are modified, and that performance can be quickly restored through lightweight fine-tuning. To evaluate classification performance, we compare shallow machine learning models with a compact CNN-LSTM deep learning model, demonstrating the latter’s strong cross-user accuracy and suitability for real-time applications. This work lowers the barrier to prototyping and integrating TENG-based sensors into textiles, supporting future research into gesture-based, self-powered e-textile interfaces for everyday environments.

Empirical findings in gesture-based interaction often stem from highly controlled experimental settings, which raises concerns about their generalizability. To explore how variations in such settings influence discoveries on user-defined gestures, we selected an end-user elicitation study involving smart rings that had been replicated at least once. By reusing the same stimuli, equipment, and data collection method, we conducted four new replications of the original study, involving a total of 120 participants across four different research teams. Our results show that smart ring gestures elicited in these replications overlap only partially, with differences in agreement rate, thinking time, and goodness of fit with corresponding system functions. We argue that systematic replication of gesture elicitation studies is essential for generalizable gesture sets.

We present an exploratory study comparing LEGO building experiences across three environments: the physical world, a Virtual Reality (VR) counterpart, and an enhanced VR setting with "superpowers" (e.g., infinite brick supply and first-person miniature perspective). Our aim is to investigate how creative physical activities can be translated into digital experiences and reimagined for ubiquitous computing platforms. 22 participants engaged in both structured assembly and creative free-building tasks across the three environments. We investigated differences in user performance, engagement, and creativity, with a focus on how the additional functionalities influenced the building experience. The findings reveal that while the physical environment offers a familiar tactile experience, VR, particularly with added superpowers, was clearly favored by participants in the creative free-building scenario. We discuss implications for future ubiquitous XR systems that enable intuitive creative exploration across diverse physical and virtual contexts, supporting learning and play.

With the rapidly growing number of articles published in the field of Human-Computer Interaction (HCI), it has become increasingly difficult to keep track of a given research area. Consequently, secondary research methods such as evidence synthesis are of growing importance. In this work, we transfer an evidence synthesis method from epidemiology to the field of HCI, develop a novel visualization technique specifically aimed at evidence synthesis, and make this technique available to the HCI community via the open source web application LayEv. To demonstrate how our approach can reveal new relational and causal paths between variables, we apply it to a set of publications identified in a prior literature review and visualize the results with LayEv. Future researchers can leverage our method to obtain comprehensive overviews of research areas, identify trends, and visualize existing evidence synthesis results.

The rapid evolution of lightweight consumer augmented reality (AR) smart glasses (a.k.a. optical see-through head-mounted displays) offers novel opportunities for learning, particularly through their unique capability to deliver multimodal information in just-in-time, micro-learning scenarios. This research investigates how such devices can support mobile second-language acquisition by presenting progressive sentence structures in multimodal formats. In contrast to the commonly used vocabulary (i.e., word) learning approach for novice learners, we present a “progressive presentation” method that combines both word and sentence learning by sequentially displaying sentence components (subject, verb, object) while retaining prior context. Pilot and formal studies revealed that progressive presentation enhances recall, particularly in mobile scenarios such as walking. Additionally, incorporating timed gaps between word presentations further improved learning effectiveness under multitasking conditions. Our findings demonstrate the utility of progressive presentation and provide usage guidelines for educational applications—even during brief, on-the-go learning moments.

Music creation has shifted from a collaborative, social experience to an individual and technical process through the use of digital audio workstations (DAWs), which often have steep learning curves that deter novice users. To address this, many new music creation applications prioritize social play through collaboration and exploration. However, tools that combine the structured concepts of traditional music production with social play remain relatively unexplored. This paper introduces SnapTunes, a tablet-based application where users individually create music tracks and connect their devices to form a combined composition. SnapTunes allows users to spatially arrange their devices, changing how the individual tracks are layered and sequenced. We playtested a Wizard-of-Oz-style prototype of SnapTunes in a workshop with 15 university students to gather insight on usability and social engagement. Findings demonstrate that SnapTunes was intuitive, enjoyable and created opportunities for playful collaboration. Usability feedback and directions for future development are discussed.

As individuals accumulate large numbers of images, many of which are redundant or low-quality, identifying and deleting unnecessary photos becomes a time-consuming task. This study explores a gallery cleaning app that aims to use machine learning to help users manage their digital photo collections more efficiently. The app adapts to individual preferences by learning from user behavior, offering personalized suggestions for deletions. We evaluate the effectiveness of the app in cleaning participants’ galleries and examine whether its personalization features impact the overall cleaning process. While personalization did not significantly improve the results in the current study, the findings contribute to the growing field of Human-Centered AI, demonstrating how adaptive systems may improve user experiences by aligning with individual needs in AI-driven applications.

Asynchronous communication tools, such as text and voice messaging, are designed for a turn-taking style where participants exchange complete messages. However, this structure lacks the dynamic elements of real-time communication, such as overlaps and backchanneling, which are central to synlogue, a turn-coupling conversation style fostering co-creative and sympathetic interactions. To address this limitation, we propose integrating synlogic interactions, including overlaps and aizuchi (short verbal backchannels), into asynchronous communication. To demonstrate this, we developed SynVoice, a voice messaging app that records the receiver’s reactions while playing the sender’s message and layers these reactions onto the sender’s original message. SynVoice enables receivers to utter spontaneous overlaps and aizuchi, allowing senders to perceive the receiver’s immediate reactions to their messages. Our preliminary user study with 12 participants showed promising improvements in empathy, understanding, and interpersonal closeness, demonstrating SynVoice’s potential to create pseudo-synchronous experiences while maintaining asynchronous flexibility.

The tests in ISO 9241 to evaluate pointing, selecting, and dragging tasks are well-known and widely used. However, although tracing is crucial in many mobile applications, like drawing or artwork, the tracing tests (e.g., path-following, line tracing) are mostly overlooked. In this paper, we detail the calculation of throughput for tracing tasks, noting ambiguities and inconsistencies in the definitions and equations in the ISO standard. A user study with 16 participants was conducted to explore the equations for the effective index of difficulty (\({\it ID}_\text{e}\)) for the two ISO tests for tracing. Results show the importance of the logarithmic term to preserve the theoretical foundation of Fitts’ law and comparability between studies. We also recommend the use of throughput, calculated as describe herein, as a performance metric when using the tracing tests.

Carpal tunnel syndrome (CTS) and cervical myelopathy (CM) are neurological disorders that impair hand function but are often underdiagnosed due to limitations in conventional screening methods. We propose a scalable, drawing-based screening approach that classifies stylus-drawn figures using image-based machine learning. Participants traced geometric shapes on a tablet, and the resulting images were used to train binary classifiers. The models achieved sensitivity and specificity comparable to or exceeding those of traditional physical tests, while minimizing examiner-dependent variability. To enhance accessibility, we integrated the models into a smartphone application that classifies paper-based drawings captured via the device camera. Unlike prior work focused on a single condition, our method addresses both peripheral (CTS) and central (CM) neuropathies within a unified framework. These results demonstrate the feasibility of using hand-drawn figures as digital biomarkers for screening neurological disorders affecting manual dexterity in clinical and home settings.

Researchers in human-computer interaction have often approached interactive system design and engineering involving atypical users, novel technological advancements, and unconventional environments that deviate from the average, moderate, or standardized. This paper examines the extent to which researchers characterize their work as extreme, through a targeted literature review and an analysis of the linguistic connotations of the term. We emphasize extreme users, who display exaggerated behaviors, acquire extraordinary abilities, and engage in high-stakes, high-performing collaborations. We also highlight extreme platforms that facilitate input and output modalities surpassing conventional limits, foster intense synergy between users and machines, and employ boundary-pushing design approaches. Lastly, we discuss extreme environments, whether physical, digital, virtual, or mixed, which pose very serious, severe, and potentially risky conditions. We conclude with the benefits for the HCI field of engaging in more scientific and technical explorations driven by “extreme” interaction elements.

Artificial intelligence (AI), particularly Large Language Models (LLMs), has created opportunities to improve user experiences by enabling the development of more interactive applications in various implementation scenarios. This paper proposes a mobile application as a virtual self-guided tour, enabling landmark recognition and enhanced user interaction with LLMs. A landmark classifier is employed for Cloud-based image classification, with accuracy further improved by incorporating GPS-based matching of classification results. These preliminary tests proved that the use of GPS to match the location improved the results and that the London Eye improved from 82 to 88 percent. Subsequently, users are provided with audio information about the identified landmark and access to extended landmark details generated by the used LLM. Users can also engage in text or voice-based interactions with the system. The system architecture integrates real-time image processing, location optimisation, and generative AI, creating interactive and engaging user interfaces.

Texting while walking, often referred to as a “smartphone zombie”, has become a growing social issue that increases the risk of traffic accidents and pedestrian collisions. In this study, we propose a novel guidance display system that selectively presents visual guidance or warnings only to pedestrians using their smartphones while walking. The system employs high-speed projection of stripe-decomposed patterns, which are perceptible only when the user is walking with a fixed gaze while focusing on their smartphone. The display remains imperceptible to pedestrians walking normally and appears as a white image. This allows guidance to be provided only to those with a fixed gaze on their smartphone, while preserving the visual field and surrounding landscape for others. This paper introduces the proposed technique and describes a demonstration setup, along with results from a preliminary user study conducted to evaluate its effectiveness.

The automobile is a complex mobile multimedia system with an operating system that is mostly hidden and closed. For research, this creates two issues: collecting vehicle data for AI is arduous and rapidly building dashboard prototypes requires specialized expertise. We introduce CANViz to overcome these barriers and allow for data capture, AI integration, and rapid visualizations. CANViz logs real time CANBUS/OBD-II vehicle data combined with a GPS logger to collect real world position data. It additionally uses computer vision (Berkeley Deep Drive) to identify road signs, vehicles, and pedestrians. Beyond logging, CANViz pipes these signals into Node-RED where an end user programmer can combine the inputs into a desired web socket that outputs to a web page instrument. The system is built on ROS and Raspberry Pi OS and operates on a local network of low cost CPUs, enabling simplified data collection and rapid prototyping research for the automotive vehicles.

For blind or low vision individuals, tactile graphics (TGs) provide essential spatial information but are restricted in the amount of data they can convey – especially semantic information, which is represented using the small amount of braille abbreviations that fit on the TG. Using computer vision, TGs can be enhanced by adding audio labels, in which audio descriptions are triggered when the user touches elements on the TG; these audio descriptions can contain unlimited amounts of semantic information and are accessible even to those who don’t read braille. Unfortunately, existing audio label systems are closed systems with severe limitations, some of which are costly and/or tied to specific hardware platforms. We address this problem by creating CamIO-Web, an open-source version of our CamIO (short for “Camera Input-Output”) audio label system, which runs in the browser of virtually any computer or mobile device. The system includes facilities that allow users to create their own TGs with any audio labels (audio recordings or Text-to-Speech), and the open-source code base makes it extensible.

Traditional hands-on laboratory education, while fundamental for skill development in the sciences, presents significant hurdles—including prohibitive costs, safety risks, and issues of accessibility. This paper presents ChemVR-AI, an immersive Virtual Reality (VR) laboratory designed to overcome these barriers in the context of chemistry. Featuring a real-time, multilingual AI assistant named Sahayak (a Hindi name for assistant), the system provides a safe, interactive environment for conducting experiments. Sahayak offers intelligent guidance, responds to conceptual queries, and clarifies theoretical principles in the user’s preferred language through natural conversation. Additionally, it is both context-aware and scientifically safety-aware—capable of detecting user actions in real time, interpreting experimental progress, and offering step-by-step guidance or warnings based on procedural logic. Interactive features include guided experiments, in-situ quizzes, systematic data recording, and adaptive support aligned with the learner’s actions. We demonstrate ChemVR-AI’s core functionalities, showcasing its potential to transform STEM education into a more engaging, inclusive, and effective experience for learners everywhere.

We present a design system for dynamic light performances utilizing portable swarm robots. The system is lightweight and highly portable, enabling users to easily deploy it in various environments without spatial constraints. By assembling modular A3-sized metal plates, users can construct various performance stages, including those that use vertical surfaces. A graphical user interface (GUI) on a PC enables intuitive design of robot positioning, movement trajectories, and lighting patterns, while also supporting interactive real-time control. Each robot is equipped with magnets to ensure stable operation on vertical metal surfaces, thereby expanding the staging of performance space beyond conventional horizontal layouts. This paper introduces the overall system architecture and demonstrates its capabilities for authoring and executing portable, swarm-robot-based light performances.

Timely diagnosis of joint bleeds is crucial for preventing long-term damage in patients with hemophilia. However, access to specialized care and the operator-dependent nature of ultrasound imaging pose significant challenges for remote monitoring. We present GAJA (Guided self-Acquisition of Joint ultrAsound images), a mobile system that interactively guides patients during the acquisition of joint ultrasound images without requiring real-time supervision. The version of GAJA presented in this paper extends support beyond the knee to include elbow and ankle joints and integrates with the CADET platform, enabling clinicians to remotely assess the acquired images. In this demo, we showcase GAJA’s real-time guidance interaction and its integration with remote clinical workflows.

Patients with cervical pain are often referred for rehabilitation treatment with a physiotherapist. This in-clinic treatment is usually combined with prescribed therapeutic exercises to perform at home. The dedication to the prescribed Home Exercise Program (HEP) is crucial to the effectiveness of the treatment and its beneficial long-term effects. However, involvement in HEP is usually low mainly due to lack of motivation and the feeling of external supervision. To overcome this, we propose RehbeCa, a complete functional prototype. It is a platform consisting of a mobile application (for patients) and a web application (for physical therapists). This allows the customization of HEP according to the needs of the patients. These new interaction techniques to monitor the HEP also ensure its proper performance. This maintains closer oversight and adjusts treatment as necessary. The mobile application consists mainly of a serious game designed to encourage patients to perform neck therapeutic exercises. This introduces a more appealing experience for patients. The exergame innovates the HEP integrating a camera-based head-tracker. Therefore, the interaction with the mobile application is based on the movement of the head (no additional sensors needed). This allows patients to perform their prescribed HEP on their own mobile devices, transforming rehabilitation into an innovative mobile experience: HEP anywhere and anytime.

Access to healthcare is widely recognized as a fundamental human right. However, language and cultural barriers often prevent equitable medical access for millions globally. In this Demo, we designed Medylan, a Medical Interpretation Agent, which bridges the language and cultural gaps by providing real-time, AI-driven interpretation services with a hand sketch input option tailored for healthcare settings. Medylan Supports the world’s major languages, including spoken and sign languages, which ensures accurate and context-aware translations of medical terminologies. Beyond language support, Medylan incorporates cultural sensitivity guidelines and AI-driven adaptations to facilitate effective cross-cultural communication. Medylan addresses disparities in healthcare by empowering patients and reducing communication-driven medical errors, particularly for marginalized populations, refugees, and immigrants, fostering inclusiveness and equity.

We present Surrogate Avatar, an adaptive telepresence method that enhances user mobility and situated co-presence in symmetric avatar-mediated communication. The system enables a remote user’s avatar to autonomously position itself in socially and environmentally appropriate locations within the local user’s space—based on spatial affordances, interactional norms, and environmental constraints—supporting fluid interaction without requiring a shared environmental context. Through a formative study, we derived key adaptation objectives and implemented them using a distributed optimization framework based on the AUIT system. The framework distributes adaptation tasks across server and client to balance responsiveness and computational efficiency. A user study involving both stationary and nomadic scenarios demonstrated consistently high usability and presence, with some limitations observed under walking conditions. An additional exploratory field study in a semi-structured public setting demonstrated the system’s viability beyond controlled lab conditions. These findings motivate future designs of mobile telepresence systems that dynamically adapt to spatial and conversational context while mitigating misunderstandings that can arise from asymmetric environmental awareness and supporting privacy-sensitive interaction.

The quest for enhanced cognition is still a driving force behind human advancement. Following the success of the first installment of the mobiCHAI workshop in 2024, mobiCHAI2 will continue to explore the intersection of Mobile Cognition-Altering Technologies (CAT) and Human-Centered AI (HCAI), focusing on their potential to augment and modify human cognition in real-world applications. Building upon previous discussions, this workshop aims to bridge insights from cognitive science, AI, and human-computer interaction (HCI) to address key challenges and opportunities in developing trustworthy, effective, and ethical AI-driven cognitive augmentation tools. Core themes of the workshop include ubiquitous sensing for cognitive tracking, AI-driven cognitive modeling, interactive augmentation methods, and ethical implications of deploying CAT in education, healthcare, and productivity. Special attention is given to the societal and ethical impact of cognition-altering AI, ensuring that augmentation technologies enhance human agency rather than compromise autonomy. Through interdisciplinary collaboration, the workshop fosters discussions on how AI can complement, rather than replace, human cognitive abilities, setting a foundation for responsible and socially acceptable digitization.

We propose this workshop to discuss and brainstorm new interactions across multiple mobile applications. Today’s users navigate a complex app ecology for leisure, e-commerce and communications, each designed to serve different purposes within their daily activities. In response, emerging technologies like iOS Shortcuts, Tasker, and uLink aim to enable fluid, multi-app interactions —yet they often fall short due to trade-offs between expressiveness and usability. We invite participants to explore how to create adaptable tools that integrate multiple apps and leverage recent advances in automation. The workshop centers around three perspectives: Researchers, End-Users, and Practitioners. For researchers, we will explore how they can prototype studies more efficiently by reusing existing app features through Shortcuts*, reducing development overhead and enabling rapid iteration. For end-users, we will examine how recent automation technologies empower individuals to build and adapt personal routines. For practitioners, we will discuss how developers and designers can move beyond siloed app experiences and think about modular, cross-application solutions and open interfaces.

As interactive technology increasingly shifts towards more seamless and natural user experiences, there is a growing need for adaptive and responsive interactions that can intuitively react to user behavior and environmental changes. Embedded AI, which enables real-time, on-device processing, plays a crucial role in achieving this by reducing latency, enhancing privacy, and ensuring continuous functionality even in offline scenarios. Meanwhile, smart textiles offer a novel and versatile input modality, allowing for richer interaction through touch, motion, and physiological sensing. The “ODISI: On-Device Intelligence for Smart Interactions” workshop explores the integration of these technologies, focusing on deploying AI models on resource-constrained devices such as microcontrollers, smartphones, and smartwatches. By combining embedded AI with smart textiles, this workshop aims to advance research in real-time adaptive interfaces, covering key topics such as AI deployment strategies, innovative sensor integration, design considerations, efficient processing pipelines, and ethical concerns. Featuring interactive presentations, hands-on tutorials, and collaborative activities, this full-day event will foster new insights and developments in AI-driven wearable and mobile interactions.

Social media platforms constitute an essential part of many people’s mobile device usage. Their contents have adapted to the mobile form factor, e.g., through an increase of short-form video and content recommendation instead of navigation and active selection. Social media systems thereby have a strong influence on individuals and society, for example, concerning public discourse and opinion-making. The rise of AI-generated content and LLM-backed autonomous agents even pushes such developments. This workshop discusses social media’s recent developments and yielding positive and negative effects on our society. Participants will share their perspectives of HCI research on social media systems and the research aims they are pursuing. In this workshop, we outline opportunities in joining insights from social sciences with the potential of recent developments in Human-Computer Interaction approaches. Interface design ideas will be explored and discussed with the research community, synthesizing collective challenges, promising future directions, and strategies for research that mitigate the negative effects of AI in social media systems on our society.

In mobile, AI-enhanced work environments, many professionals find themselves both empowered and overwhelmed. While automation promises to save time, conversations with designers, analysts, and managers reveal a different reality: expectations accelerate, interruptions multiply, and the space for focused, meaningful work continues to shrink. This article draws on real-world experiences and workplace reflections to explore the paradox of digital productivity. We examine how fragmented attention, notification overload, and performance pressure are reshaping not only workflows but also workers’ sense of clarity and fulfillment. Through three illustrative examples from industry, we underscore the need to redesign work with intention. This perspective encourages readers to reconsider how presence, rhythm, and small rituals can help restore attention and satisfaction in an age of constant connectivity.

Blind and low-vision (BLV) individuals often face difficulties when moving through complex indoor spaces like hospitals. This work investigates how commercially available technologies can be integrated to assist BLV visitors navigate outpatient clinics in a hospital in Greece. The project involved creating a functional prototype of an indoor navigation system that uses off-the-shelf Bluetooth-based sensors for accurate indoor positioning, a third-party indoor navigation service, and a custom mobile application that integrates those components. BLV users were included early in the process, and a demonstration was conducted to assess how well the system worked and how users responded to it. The project revealed both technical and infrastructure-related challenges, along with areas for improving accessibility. This work offers practical insights and lessons learned from adapting existing technologies to make indoor spaces of public interest more accessible.

Femtech applications developed to support menstruation tracking, fertility, pregnancy, and broader reproductive health are widely used but often fall short in addressing user privacy, accessibility, and design transparency. This dissertation explores the intersection of usability and privacy in femtech by integrating three interrelated studies: large-scale analysis of app store reviews, persona-based walkthroughs of mobile apps, and a user survey on privacy expectations and reproductive needs. The research surfaces design gaps, legal concerns, and usability breakdowns that affect diverse user groups. The goal is to develop a mobile application that integrates user-informed privacy-aware features. Feedback is sought on how to best align technical design with social needs in reproductive health technology.

I plan to design earthquake preparedness, evacuation and rescue systems using extended reality. The targeted disaster, which is an earthquake, causes humans to be trapped inside collapsed buildings, where time is critical for finding and saving them. The public is made aware of how they would feel inside a collapsed building by using the virtual reality environment. Then an evacuation post the earthquake is planned using the augmented reality system where the dynamic QR codes placed at the critical locations direct the public to the nearest exit. For the rescuers post the earthquake, a VR interfaced robotic system is designed in such a way that it can traverse through the debris and the feedback is sent to the VR head mounted display. Thus this work proposes an extended reality-based system integrating VR for public awareness and rescue operations, and AR-guided evacuation to enhance earthquake preparedness and response. I believe that presenting this work in the Doctoral Consortium will help in receiving constructive feedback from the experts in this field.

Social Mixed Reality (MR) enables new forms of interaction, but its broader adoption hinges on achieving strong Social Presence. Supporting Social Presence is challenging due to the diverse forms and configurations of MR systems. This research pursues two goals: (1) understanding how Social Presence can be supported in MR, and (2) proposing design strategies for effective communication and collaboration. A systematic literature review and thematic analysis uncovered common social use cases, key design requirements, and future directions for MR development. Drawing from these insights, we propose three contributions: (1) a decoupled representation model that separates how users see themselves from how they appear to others in immersive MR; (2) proxemic interaction techniques for monocular smart glasses; and (3) gesture-based navigation and tracking for non-immersed users in asymmetric VR setups. Our findings reveal that representation, proxemics, and intent communication are central to the quality and realism of social MR experiences.

FemTech, or female technology, is a growing sector of technological innovation focused on addressing historically overlooked areas of women’s health, including menstruation, contraception, fertility, and menopause. In this space, mobile health (mHealth) apps have become especially popular due to their accessibility and convenience. While existing research often evaluates FemTech systems based on clinical efficacy or structural critique, this dissertation takes a user-centered approach to explore their ethical, emotional, and experiential dimensions. It follows a two-phase research plan: first, analyzing user reviews of popular commercial apps to uncover unmet needs, design flaws, and user dissatisfaction; second, developing future-design experiments that explore ways to build trust, support meaning-making, and encourage interpersonal care. The project aims to produce a design framework and actionable guidelines for more human-centered, experience-driven FemTech design and offer a perspective to reconsider how FemTech systems operate in commercial contexts where health, data, and commerce intersect in complex and ethically significant ways.

Noise sensitivity affects both neurodivergent and neurotypical individuals, posing challenges for self-regulation and quality of life. While existing research suggests various methods to support this condition, there is a notable gap in systems that can identify and monitor characteristics indicating the onset of noise sensitivity. Thus, allowing people with noise sensitivity (PWNS) and those around them to take preventive measures before becoming overstimulated. My research investigates how PWNS and their care networks manage and regulate reactions to noise. It aims to inform the design of a novel application that incorporates sensing and tracking technology to enhance awareness, promote information sharing, and offer strategies for effectively managing noise sensitivity experiences. This research seeks to deepen our understanding of noise sensitivity and contribute solutions for those affected.

The proliferation of mobile technologies has significantly increased the speed at which users create and manage personal data, yet users often remain unaware of how their data is accessed. While smartphone privacy management relies on permission systems, existing user interfaces frequently fall short of aligning with user privacy expectations. Prior studies have mainly explored runtime permissions, leaving the permissions manager interface, key for privacy control post-installation, under-examined. My dissertation aims to redesign the permissions managers interface to enhance usability and support more efficient and privacy-aware decision-making. It addresses three core challenges: (1) high cognitive load from numerous permissions; (2) absence of prompts for regular review; and (3) limited user understanding of the data accessed by permissions. To improve permission discoverability and comprehension, the proposed redesign will leverage personalization and customization techniques to support more informed user privacy decisions.

Advancements in virtual reality (VR) and portable head-mounted displays are transforming how surgical training is delivered, particularly in minimally invasive procedures like endoscopic neurosurgery. Current VR-based surgical training systems often rely on expensive proprietary hardware and lack features such as patient specificity, adaptive difficulty, or standardized evaluation metrics. This work addresses these challenges by utilizing commercially available VR headsets and open-source components to build a unified platform that integrates anatomy visualization, procedural simulation, performance evaluation, and collaborative learning. The platform includes features such as AI-based patient-specific modeling, gamified progression, "ghost surgery" instructional modules, and a privacy-preserving training environment. The research contributes new methods for performance benchmarking in VR surgery, cost-effective system design, and pedagogically informed simulation design. The project is validated through user studies and technical evaluations in collaboration with neurosurgical experts.