Modeling takeover time based on non-driving-related task attributes in highly automated driving

Area of Science:

• Human-Computer Interaction
• Automotive Engineering
• Cognitive Psychology

Background:

• Highly automated driving (HAD) systems require drivers to retake control in certain situations.
• Non-driving-related tasks (NDRTs) performed by drivers can interfere with this critical takeover process.
• Understanding the impact of NDRT characteristics on takeover performance is crucial for safety.

Purpose of the Study:

• To investigate how physical, visual, and cognitive attributes of NDRTs affect the transition of control in HAD.
• To develop a conceptual model of the takeover process, distinguishing between motor and mental reactions.
• To create and validate a predictive model for takeover time based on NDRT attributes.

Main Methods:

• A laboratory experiment was designed to assess the influence of individual NDRT attributes on specific stages of the takeover process.
• Multiple linear regression analysis was employed to develop a prediction model for takeover time.
• A validation experiment involving nine different NDRTs and a baseline condition was conducted to test the model's explanatory power.

Main Results:

• The study found that the motor reaction component of control transition in HAD is significantly affected by the physical attributes of NDRTs.
• The developed prediction model demonstrated a notable ability to explain variations in takeover time across different NDRT conditions.
• Physical characteristics of NDRTs were identified as a primary factor influencing the timing of control transitions.

Conclusions:

• The physical attributes of non-driving-related tasks play a dominant role in the motor aspects of control transition during highly automated driving.
• The developed model provides valuable insights into predicting takeover time, aiding in the design of safer HAD systems.
• Future research should consider the interplay of physical, visual, and cognitive NDRT factors to further refine understanding of driver behavior in HAD.