All-inclusive TORs: Cross-Cultural and Age-Sensitive Design for Take-Over Requests in Level 3 Cars

Research Background and Issues

• Problems and Challenges: The core issue of this study is how to design Take-Over Requests (TORs) for Level 3 autonomous vehicles to optimize the transition between autonomous and manual driving. Existing designs fail to consider the combined effects of age, cultural differences, and various types of Non-Driving Related Tasks (NDRTs) on driver behavior, leading to designs that lack adaptability and compromise safety.

• Significance: Level 3 autonomous vehicles allow drivers to focus on non-driving tasks but require them to quickly regain control of the vehicle when necessary. As these vehicles are already in use globally, such as in the United States and Germany, there is an urgent need for safe TOR designs. Optimized designs not only enhance safety but also promote the widespread adoption of autonomous driving technology.

• Research Motivation and Related Work: This study identifies that age, cultural background, and NDRT types significantly influence drivers' responses to TORs. Specifically:

  • Older drivers may be affected by declines in cognitive and physical abilities when switching control tasks.
  • Different cultures exhibit unique driving habits and legal compliance standards, which influence TOR design requirements.
  • The varying visual, cognitive, and physical demands of different NDRTs further increase the complexity of design. Existing research often overlooks the interactive effects of these factors, adopting a "one-size-fits-all" design approach.

Solution

• Methods and Solutions: Through cross-cultural participatory design, this study explores for the first time how to customize TORs based on drivers' age, cultural background, and NDRT types. Participants from two cultural backgrounds (the UK and Israel) and two age groups (22-32 years and 60+ years) designed TORs for four types of NDRTs (reading, watching videos, chatting, texting).

• Innovations:

  • Proposed a dynamic, culturally sensitive, and age-adaptive TOR design method, integrating NDRT devices with in-vehicle information systems.
  • Developed a taxonomy of TOR interface components, including information content, presentation methods, location, and technological integration.
  • Created five new TOR conceptual designs to directly address the needs of cross-cultural and age-diverse users.
  • Introduced the concept of a "risk scale" for real-time risk assessment, addressing the sensitivity to alerts in Israeli driving culture.

• Implementation Steps and Techniques:

  1. Participatory Design Experiment:
     • Using actual vehicles, participants designed TORs tailored to their needs after completing NDRTs.
     • Data included NASA-TLX questionnaire measurements of task load, audio recordings, and design sketches.
  2. Thematic Analysis:
     • Extracted common themes and issues underlying discussions.
  3. Taxonomy Development:
     • Established a hierarchical taxonomy of TOR features, including task type, content, mode, location, and technology.
  4. Conceptual Design Synthesis:
     • Proposed five design concepts based on the taxonomy, corresponding to the needs of different age groups and cultural backgrounds.

Research Outcomes

• Specific Findings:

  1. Theoretical Insights:
     • Age significantly affects TOR preferences: older drivers require more direct, multimodal (auditory and tactile) TORs, while younger drivers prefer titrated, gradual notifications.
     • Cultural differences are also significant: Israeli participants, due to regional conflicts, emphasized the need for dynamic risk assessment, while UK participants focused more on normative and legal compliance.
  2. Practical Designs:
     • Proposed five TOR conceptual designs, including dynamic risk-aware TORs, three-phase progressive TORs, and dual TORs synchronized between devices and in-vehicle systems.

• Advantages:

  • Culturally and age-sensitive designs allow for personalized adjustments to meet diverse driver needs while maintaining commonalities to enhance universality.
  • Integration of NDRT devices, such as smartphones, improves TOR responsiveness and acceptance.
  • Data-driven designs are based on actual driver preferences and characteristics, supporting design rationality.

• Experimental and Evaluation Results:

  • NASA-TLX analysis revealed significant differences in task load across NDRTs, with video-watching tasks being significantly less demanding than chatting or texting.
  • Older drivers overwhelmingly preferred tri-modal TORs, while younger drivers opted for bi-modal TORs, demonstrating greater flexibility in handling task interruptions.

• Limitations and Future Directions:

  1. Experimental Limitations:
     • The study was conducted in stationary vehicles and did not test response performance in real driving environments.
     • Contextual efficiency of task boundary switching was not considered.
     • TOR performance in high-risk scenarios was not tested.
  2. Future Plans:
     • Test the design's validity and acceptance in dynamic driving scenarios.
     • Explore the impact of other cultural backgrounds and regional laws on design.
     • Enhance design adaptability, such as integrating augmented reality (AR) technology into NDRT devices.

Conclusion

This study provides an in-depth exploration of TOR design for Level 3 autonomous vehicles from cultural and age perspectives, proposing a comprehensive innovation pathway from theory to practice. The findings not only offer concrete guidance for designers but also contribute valuable insights into the standardization and inclusivity of global autonomous driving technology applications.