Introducing ROADS: A Systematic Comparison of Remote Control Interaction Concepts for Automated Vehicles at Road Works

Research Background and Issues

• What problems or challenges did the authors identify?

  • Fully autonomous driving technology still faces numerous challenges, particularly in uncontrolled complex environments (e.g., urban construction sites), where autonomous vehicles (AVs) may struggle to independently make decisions or handle specific situations. Additionally, current designs for remote operation (RO) interfaces lack systematic comparisons, especially when remote operators need to manage multiple vehicle requests simultaneously.

• Why is this issue important?

  • The widespread adoption of autonomous driving technology depends on its reliability and safety in constrained environments. Remote operation systems can fill the capability gaps of autonomous driving technology, providing supplementary support for AVs, especially when operating outside their Operational Design Domain (ODD). However, the lack of comprehensive studies on operational interfaces may limit the efficiency of these systems.

• Research Motivation and Related Work

  • Early studies analyzed remote driving scenarios, interaction requirements, and corresponding technical and human factors challenges. However, the authors noted that no existing research compares the performance of different user interfaces or their feasibility in multi-vehicle concurrent control scenarios. This study experimentally validates different remote operation interaction methods, providing foundational insights for interface design to improve future system efficiency and user experience.

Solutions

• What methods or solutions did the authors propose?

  • The authors designed, implemented, and evaluated three distinct remote operation interaction concepts: Trajectory Guidance, Path Planning, and Waypoint Guidance.
  • They tested scenarios with up to four simultaneous requests in a closed simulation environment to understand user behavior and system performance under different interaction modes.

• What is innovative about this solution?

  • It is the first systematic comparison of the performance of three interaction modes in multi-task operations.
  • A new simulation platform, Remote Operation Automated Driving Suite (ROADS), was introduced to replicate remote operation requirements for AVs in complex scenarios.
  • Multiple evaluation metrics, such as task load, System Usability Scale (SUS) scores, and path deviation, were considered and quantified to comprehensively assess the effectiveness of the interaction modes.

• What are the implementation steps and key technologies used?

  1. User Interface:
     • Specific user interfaces were designed for each interaction mode, such as:
       • Path Planning: Listing multiple recommended paths for the vehicle.
       • Trajectory Guidance: Allowing users to draw the vehicle's trajectory using a mouse.
       • Waypoint Guidance: Enabling users to manually set several waypoints.
  2. Experiment Design:
     • A simulated road construction scenario was developed using the Unity platform.
     • Experiments tested the feasibility of controlling multiple vehicles (1 to 4 concurrent tasks), with task load combinations randomly presented using a Latin square design.
  3. Evaluation of Key Performance Metrics:
     • Objective metrics (e.g., task completion rate, path deviation, mouse movement distance).
     • Subjective metrics (e.g., NASA-TLX task load index, acceptance, and satisfaction scores).

Research Findings

• What specific results were achieved?

  1. Among the three modes, Path Planning performed the best, achieving the highest System Usability Scale (SUS) score (82.47) and the lowest task load.
  2. Waypoint Guidance ranked second, while Trajectory Guidance scored the lowest due to its higher operational complexity.
  3. Users reported the highest satisfaction when handling multiple vehicles (especially two).

• What advantages does it have compared to existing solutions?

  • Through higher automation design (e.g., the "Path Planning" mode), the task workload of remote operators was significantly reduced.
  • This study addresses gaps in existing literature by comparing different interaction schemes and is the first to validate the feasibility of a single remote operator managing multiple vehicles.

• What were the experimental or evaluation results?

  • The experiments demonstrated:
    • In single-vehicle scenarios, all three interaction modes were capable of completing tasks, but "Path Planning" performed best in multi-task scenarios, effectively distributing attention.
    • When the number of tasks increased to three or four, the completion rate for "Trajectory Guidance" significantly decreased.
    • Participants generally preferred the "Path Planning" mode but suggested incorporating switchable interaction options to handle special situations.
  • Limitations and Future Directions:
    1. The simulation experiments did not account for potential communication delays in real-world environments, which could significantly impact actual operations.
    2. User evaluations primarily involved younger participants, necessitating further validation with a more diverse sample.
    3. The study was limited to specific road construction scenarios; expanding to other complex conditions (e.g., urban environments) would be valuable.
    4. Users suggested further improvements to the interface (e.g., prompts for optimal task focus and options for undo operations).

By comparing the three interaction modes, the study underscores the importance of well-designed user interfaces and highlights the need for appropriate path recommendation systems to reduce user workload. These findings provide critical references for the design of future remote control systems.