Trust and Visual Focus in Automated Vehicles: A Comparative Study of Beginner and Experienced Drivers

Research Background and Issues

• What problems or challenges did the authors identify?
  Automated Vehicles (AVs) are developing rapidly, but there remains a knowledge gap regarding the interaction behaviors between human drivers and automated systems, particularly in the areas of trust and visual attention allocation. Overtrust may lead to drivers losing effective control of the vehicle during critical moments, while distrust may result in low adoption rates of automated driving technology. Furthermore, the impact of different levels of driving experience on trust in automated driving systems and interaction patterns has not been fully clarified.

• Why is this issue important?
  Trust is a key factor in user acceptance of automated driving technology, and failure to effectively calibrate trust may lead to accidents and a decline in system performance. A deeper understanding of how driving experience influences trust and driving behavior can help improve human-computer interaction interface design, enhancing the safety and user experience of automated driving.

• Research Motivation and Related Work
  Research indicates that driving experience affects drivers' adaptability to automation technology and decision-making behavior. Skilled drivers perform better in trust calibration and system control transitions, but their advantages under complex conditions may be limited. On the other hand, the impact of trust on monitoring behavior and Non-Driving-Related Tasks (NDRT) requires further quantitative analysis.

Solutions

• What methods or solutions did the authors propose?
  This study used simulated driving experiments to analyze trust dynamics, driving behavior, and gaze allocation when novice and experienced drivers interacted with automated driving systems. The experiment included three driving conditions (manual driving, non-critical automated driving, and critical automated driving) and combined trust questionnaires (TiA), eye-tracking data, and Standard Deviation of Lateral Position (SDLP) to evaluate driving performance.

• What are the innovative aspects of this solution?

  1. The study categorized drivers into novice and experienced groups to explore how driving experience influences trust dynamics and responses to critical events.
  2. It integrated driving stability (SDLP), eye-tracking behavior (MGD), and trust questionnaires, providing a comprehensive analytical framework for the interaction mechanisms between driving behavior and trust.
  3. Non-critical and critical scenarios were designed to evaluate driver performance under varying levels of automated driving complexity.

• What are the implementation steps and key technologies used?

  1. Experimental Design
     Twenty participants (divided into inexperienced and experienced groups) completed a simulated driving task under multiple driving conditions, during which they also performed a distracting non-driving task (e.g., using Spotify).
  2. Simulation Equipment
     The CARLA driving simulator and eye-tracking devices were used to record driving behavior data, with voice prompts guiding specific scenarios.
  3. Data Collection
     • Standard Deviation of Lateral Position (SDLP): Evaluated the variability of the vehicle's lateral position.
     • Trust Questionnaire (TiA): Measured drivers' trust in the automated driving system and its changes.
     • Eye-tracking Data (MGD): Analyzed drivers' attention distribution across different visual areas during driving or non-driving tasks.
  4. Data Analysis
     Mixed Analysis of Variance (ANOVA) was used to deconstruct the performance differences between groups under the three driving conditions and to explore correlations with trust scores.

Research Findings

• What specific findings were obtained?

  1. Driving Behavior

     • Novice drivers exhibited greater lateral position variability (SDLP) under critical conditions, indicating overtrust in the automated system and delayed responses.
     • Experienced drivers demonstrated more stable driving performance and better adaptability to lane changes in critical conditions.

  2. Trust Dynamics

     • All participants showed a significant increase in overall trust in the automated driving system after completing the experiment.
     • Novice drivers exhibited a faster increase in trust but often lacked an understanding of the system's limitations, while experienced drivers demonstrated a higher level of understanding of system predictability and reliability.

  3. Visual Behavior

     • Under critical conditions, drivers showed a significant increase in gaze time on the right-side rearview mirror, indicating heightened awareness of environmental risks.
     • Novice drivers displayed higher levels of monitoring behavior in automated driving states, especially under non-critical conditions.

• What are the advantages compared to existing solutions?

  • The study provides a more detailed quantification of the differential impacts of driving experience on trust and monitoring behavior.
  • It offers a cross-analysis of lateral driving stability, trust questionnaires, and eye-tracking data, which is innovative in the context of automated driving research.

• What were the experimental or evaluation results?

  • The correlation analysis between trust and eye-tracking behavior showed that novice drivers increased their monitoring behavior as trust grew, while experienced drivers reduced their monitoring frequency as trust increased.

• Limitations and Future Directions
  Limitations:

  • The sample size was relatively small, and cultural and age differences were not thoroughly explored.
  • The experiment was based on simulated driving, which may not fully reflect risk perception and behavior choices in real-world driving.

  Future Directions:

  • Further research is needed on the dynamic process of long-term trust development and its profound impact on driving behavior.
  • Investigate the role of different cultural and social backgrounds in building trust in automated driving.

Based on the findings of this study, the authors emphasize the importance of trust calibration in future automated driving design and propose practical recommendations for improving human-computer interaction systems, such as enhancing interface transparency and the predictability of system feedback.