Fly Away: Evaluating the Impact of Motion Fidelity on Optimized User Interface Design via Bayesian Optimization in Automated Urban Air Mobility Simulations

Research Background and Issues

Problems or Challenges Identified by the Authors

1. The widespread adoption of Urban Air Mobility (UAM) requires enhancing passenger trust and sense of security, yet research addressing this aspect remains limited.
2. In simulations of air taxis, motion fidelity may influence passengers' evaluation of the user interface (UI), but the impact of high-fidelity motion simulation in the UAM domain has not been thoroughly studied.
3. Current research primarily focuses on the design of individual visualization elements (e.g., flight trajectories) without optimizing the complete UI, and the interactions between components have not been deeply explored.
4. Simple standardized UI designs may fail to meet the diverse needs of different passengers.

Why This Problem is Important

• Automated air taxis are expected to become a critical future transportation mode for alleviating urban traffic congestion and reducing emissions. Their adoption rate heavily depends on passengers' trust and understanding of this new mode of transportation.
• High-fidelity motion simulation may better approximate real-world conditions, but its impact on UI design optimization and user experience remains unexplored, which is crucial for future UAM system design.

Research Motivation and Related Work

• Existing research has demonstrated that visualized information significantly enhances passenger trust and sense of security. However, most studies use low/medium-fidelity VR environments, with fewer studies employing high-fidelity motion simulation.
• Multi-objective Bayesian optimization (MOBO) methods have been successfully applied in other interaction design domains but have not been fully explored in UAM UI design research.
• The authors aim to investigate the impact of motion fidelity on UI design and provide more scientific and comprehensive recommendations for optimizing UAM UI design based on passenger feedback.

Solution

Proposed Methods or Solutions

1. Experimental Design:
   • The authors designed a controlled study involving 40 participants, with one group experiencing simulations in a high-fidelity VR environment with motion cues (using a 3-DoF motion chair), and the other group in a standard VR environment without motion cues.
2. UI Optimization Method:
   • Using multi-objective Bayesian optimization (MOBO), six objectives (trust, perceived safety, understanding, cognitive load, acceptance, aesthetics) were optimized across 12 UI design parameters.
   • Subjective measurements and user feedback were utilized to iteratively adjust the UI design, identifying the optimal parameter combinations and converging on Pareto-optimal solutions.

Innovations of the Solution

1. Integrating motion simulation with MOBO, this study is the first to explore how motion simulation affects user experience and design decisions in air taxi UI design.
2. Optimizing personalized UIs for different participants, providing a more detailed design framework to address diverse passenger needs.
3. Proposing a systematic testing and optimization framework to balance trade-offs among multiple objectives (e.g., trust, safety, and aesthetics).

Implementation Steps and Key Technologies

1. Experimental Environment:
   • Developed a Unity-based VR air taxi flight simulation, incorporating environmental visualizations (e.g., flight trajectories and map information).
   • Synchronized motion cues with physical feedback from the 3-DoF motion chair.
2. Bayesian Optimization Process:
   • Initial sampling phase using Sobol sampling (5 samples).
   • Optimization phase (25 iterations), dynamically adjusting UI design parameters based on user feedback.
   • Optimization results were generated using the Python package BoTorch, dynamically producing Pareto front designs.
3. Parameter Settings:
   • The 12 optimized parameters included trajectory length, transparency, information display, etc. (e.g., whether to display maps or bounding boxes).
   • Optimization objectives were measured using normalized subjective scales to ensure consistent quantification.

Research Findings

Specific Results

1. Compared to VR simulations without motion cues, high-fidelity simulations with motion cues significantly reduced passengers' trust, understanding, and acceptance.
2. While motion simulation had limited impact on UI design parameters, certain specific elements (e.g., bounding box display and size of other aircraft) showed significant differences.
3. A single UI design may not satisfy passenger needs; diversified and personalized UI designs have greater potential to enhance user experience.

Advantages Over Existing Solutions

• Systematic Methodology: Compared to traditional experimental methods, MOBO enables more efficient UI design, dynamically adapting to passengers' subjective preferences.
• Future Adaptability: Emphasizing personalized design, offering customized experiences for different passengers, surpassing the static "one-size-fits-all" UI design.
• Closer to Reality: Incorporating motion cues into the experiment provides a more realistic air taxi riding scenario.

Experimental or Evaluation Results

1. User Effects:
   • Tests revealed significant differences between the motion cue group and the non-motion cue group in subjective trust (Bayes factor BF = 3970.14), understanding, and acceptance.
2. UI Design Optimization:
   • MOBO successfully improved UI design performance, but the optimized UI could not eliminate the trust loss caused by motion cues.
   • Pareto front designs concentrated on specific features, such as reducing interference from large icons under motion conditions.
3. Objective Correlations:
   • Trust, understanding, acceptance, and aesthetics were strongly correlated, suggesting future research should simplify the dimensions of optimization objectives.

Limitations and Future Directions

1. Limitations of Motion Simulation Equipment:
   • This study used only one type of motion simulation equipment (3-DoF motion chair), excluding higher-fidelity 6-DoF equipment or real flight experiences.
2. Participant Background Limitations:
   • Most participants lacked real-world air taxi riding experience, which may affect the external validity of the results.
3. Future Research Suggestions:
   • Explore the effects of more high-fidelity simulation equipment and incorporate real user groups to simulate actual flight experiences.
   • Combine physiological signals (e.g., heart rate or skin conductance response) in UI optimization to obtain more comprehensive user experience evaluations.
   • Investigate shared public transportation UI design standards to accommodate shared modes like "air metros."