Designing for Passengers' Information Needs on Fellow Travelers: A Comparison of Day and Night Rides in Shared Automated Vehicles

The paper examines how time of day (day vs. night) and provision of information about fellow passengers influence perceived security, user experience (UX), trust, acceptance, and emotions in shared automated mobility-on-demand (SAMoD) rides. As automated vehicles (AVs) enable driverless ride-sharing, the absence of a human authority may heighten concerns when riding with strangers, particularly at night and especially among women. Prior work suggests that providing information about co-passengers could increase acceptance and perceived security, but the roles of gender and time of day remain underexplored. Research question: How are SAMoD passengers’ perceptions of security and corresponding UX, trust, acceptance, and emotions influenced by the time of day of a shared ride and their knowledge of co-passengers? To address limitations of survey-only research, the authors conducted an empirical, within-subjects simulator study to provide participants with realistic experiences of shared rides and systematically varied co-passenger information. The study aims to inform resilient, privacy-aware UI designs that support security needs and inclusive adoption of SAMoD.

Background and related work identify major acceptance barriers for shared automated mobility, including demographics, system reliability, trust, perceived security, and privacy concerns. Perceived security and system resilience are interlinked: resilient systems (and cultures promoting reporting and responsiveness) can foster trust and perceived security, which in turn influence willingness to use SAMoD. UI design is central because communication occurs via in-vehicle and companion interfaces without a driver present. Prior studies show preferences for familiar modalities (visual/auditory) and front-area displays, and highlight multimodal, context-aware design. Because real AV testbeds are limited, context-based prototyping methods (VR, WoOz, simulators, and immersive video) are used to evaluate HMIs. Immersive video-based simulators offer controllable yet dynamic environments appropriate for passenger-focused studies without driver input. Prior findings suggest information about co-passengers can reduce reluctance to share rides, but privacy tensions and risks of discrimination exist. Security needs differ by time of day and gender, with heightened concerns at night and among women; some work also indicates cultural variability in ride-sharing preferences. These insights motivated testing HMI variants that disclose co-passenger information to assess impacts on perceived security, UX, acceptance, and emotions across day vs. night contexts.

Design: Exploratory within-subjects mixed-methods study with N = 24 (12 women, 12 men; age 18–81, M = 40.5, SD = 21.3). Each participant experienced four simulated SAMoD rides: 2 day and 2 night, each once with personal information about co-passengers and once without (order randomized and counterbalanced). In the “with information” condition, the UI displayed co-passenger name, age (young: 20–30; older: 50–60), destination, and an AI-generated profile photo (neutral expression). Age and gender of co-passengers were balanced across rides; one co-passenger (woman) always boarded first to limit condition proliferation; only one co-passenger was present at a time. Setting and apparatus: An immersive video-based simulator with three 55.1-inch LCD screens (front, left, right views) showing synchronized urban day/night ride footage recorded via action cameras; enhanced with door sounds and voice prompts. A tent-based vehicle mock-up with 2x2 seating increased immersion. The in-vehicle passenger information display prototype was shown on a 24.1-inch screen. Audio via a 2.1 sound system. The UI featured a split view with a route map, current position, upcoming stops, and boarding/alighting announcements. Two variants were used: without co-passenger personal information (only generic counts) and with co-passenger personal information (name, age, destination, photo), with corresponding voice prompts. Scenarios: Four leisure-trip scenarios (two daytime bakery-to-park round trips; two nighttime home-to-restaurant round trips) designed via storytelling; participants received paper tickets with name and itinerary to enhance realism. In each ride, one man and one woman boarded sequentially at successive stops (only one on board at a time); participants alighted at the third stop (destination). Procedure and measures: Sessions (60–90 minutes) included briefing and consent, demographics (including BFI-10), and STAI trait anxiety. During rides, participants repeatedly marked emotions using an adapted emoji-based Russell Affect Grid (pleasure, arousal) at key boarding/alighting announcements. After each ride, participants drew an emotion curve and completed questionnaires: UEQ-s (pragmatic and hedonic UX), UEQ+ Usefulness and Attractiveness subscales, Trust in Automation (Körber), Technology Acceptance Model measures (Perceived Usefulness, Intention to Use; Chen), Security Concerns (Dekker), and Perceived Risk. After the final ride, presence/immersion was assessed via IPQ. Post-session semi-structured interviews explored preferences for UI variants, security perceptions, and desired information types. Ethics: Conducted per Declaration of Helsinki guidelines; informed consent; option to withdraw any time. One participant (P21) terminated one ride due to simulator sickness; missing data were imputed via maximum likelihood. Analysis: Quantitative analyses via RM-ANOVA with within-subject factors time of day (day vs. night) and passenger information (without vs. with), and between-subject factor gender (women vs. men). Alpha = .05; Holm-adjusted pairwise comparisons; effect sizes per Cohen. Qualitative interviews transcribed verbatim and analyzed via inductive qualitative content analysis (MAXQDA) with iterative coding.

User experience: The SAMoD UI received excellent ratings for pragmatic and hedonic UX across conditions. Attractiveness was higher during daytime than nighttime (F(1,22) = 6.820, p = .016, η2G = 0.026; mean difference = 0.3, SE = 0.1; Cohen’s d = 0.533). An interaction of information x gender appeared in RM-ANOVA (F(1,22) = 5.059, p = .035, η2G = 0.021) but was not confirmed in pairwise comparisons. Acceptance: Women reported higher Perceived Usefulness (F(1,22) = 7.586, p = .012, η2G = 0.194) and higher Intention to Use (F(1,22) = 6.490, p = .018, η2G = 0.159) than men. Pairwise comparison example: Intention to Use difference t = 2.547, p_holm = .018, d = 0.520. No within-subject effects of time of day or passenger information on acceptance scales were found. Security, trust, and risk: Trust in automation was medium-high, and security concerns were moderate, without significant differences by time or information. Perceived risk was significantly lower with co-passenger information than without (F(1,22) = 7.321, p = .013, η2G = 0.013; mean difference = 0.2, SE = 0.1; t = 2.706, p_holm = .013; d = 0.552). Emotions: Daytime rides were rated more pleasant than nighttime (Affect Grid pleasure: F(1,43) = 12.386, p = .001, η2G = 0.032; Day M = 7.6, SD = 2.2 vs. Night M = 6.8, SD = 2.2; mean diff = 0.8, SE = 0.2; t = 3.519, p_holm = .001; d = 0.525). Arousal showed no meaningful effects. Visual inspection suggested rides without information felt more pleasant overall during the day, while information increased dispersion in affect ratings. Presence: IPQ indicated medium-to-high immersion (Realism M = 4.0, SD = 1.1; Involvement M = 3.3, SD = 1.1; Spatial Presence M = 4.1, SD = 0.9; General M = 4.9, SD = 0.8). Qualitative insights: Most participants preferred having co-passenger information (15; 9 women, 6 men), primarily for security (22 mentions; 12 women, 9 men) and a more pleasant, connected experience (9). The main reason to prefer no information was privacy (17). Time-of-day mattered strongly: participants emphasized higher security needs at night and preferred receiving co-passenger information then; during the day, such information was often deemed unnecessary or even detrimental. Both women and men expressed a preference for sharing rides with women at night. The most valued information element was the profile photo (23 mentions), followed by gender (13), age (10), name (8), and destination (7). Participants noted the tension between privacy and security and suggested context-dependent disclosure (e.g., more at night).

Findings indicate that daytime shared rides are perceived more positively, while nighttime rides heighten security concerns. Co-passenger information reduces perceived risk, particularly at night, suggesting that contextually increasing transparency can improve perceived security and resilience. However, providing personal information introduces privacy tensions and potential harms (e.g., targeting based on name/destination), leading to heterogeneous preferences and increased emotional variability. Gender differences emerged: women rated usefulness and intention to use higher and, along with men, preferred sharing rides with women at night—echoing broader public transport findings about safety. From a human factors and resilience perspective, one-size-fits-all UIs are unlikely to suffice. Systems should provide flexible, customizable disclosure options and consider timing and placement of information: information may be more beneficial in the booking phase and minimized in-vehicle. Emphasizing elements that afford a feeling of control (e.g., photos) can increase perceived security, but designs must avoid discrimination and protect privacy. Alternative concepts (e.g., buddy systems or social passengering) might provide security through supportive social presence rather than identity disclosure.

The study shows that time of day and co-passenger information shape UX, perceived security, acceptance, and emotions in shared automated rides. Daytime rides are generally evaluated more positively; at night, providing co-passenger information lowers perceived risk and can enhance feelings of security. Most participants valued seeing a photo of fellow travelers as the most impactful information element. Yet, privacy concerns create a clear trade-off, underscoring the need for context-dependent and customizable disclosure strategies. The authors recommend exploring provision of co-passenger information during booking rather than in-vehicle and complementing information with safety features (e.g., emergency/support options). Future research should cover the full journey (booking, ride, on/off-boarding), include richer social context (e.g., multiple/physical co-passengers), examine cultural differences, and investigate modalities and features that improve perceived control and security without compromising privacy.

External validity is limited by the lab-based, immersive video simulation and virtual (not physical) co-passengers. One participant experienced simulator sickness; otherwise, presence ratings suggest good immersion. The small, German-based, gender-balanced sample with relatively low trait anxiety limits generalizability; COVID-19 conditions may have influenced recruitment and responses. The study evaluated a narrow set of co-passenger attributes (photo, gender, age, name, destination), which may have reinforced stereotypes and does not capture broader social dimensions (e.g., race, culture). Order of co-passenger boarding was fixed to control complexity, and only single co-passenger situations were tested. Cultural context and security baselines in Germany may not generalize to regions with different public transport safety profiles.