A model of dyadic merging interactions explains human drivers' behavior from control inputs to decisions.

Area of Science:

• Robotics and Artificial Intelligence
• Human-Computer Interaction
• Transportation Engineering

Background:

• Safe and socially acceptable interactions between human-driven vehicles and automated vehicles (AVs) are critical for widespread AV adoption.
• Current driver models often focus on isolated aspects of driving behavior, limiting a holistic understanding of traffic interactions.
• Understanding the principles of human traffic interactions is essential for developing effective AVs.

Purpose of the Study:

• To develop a novel driver model that captures the complexities of human-driven vehicle interactions.
• To explicitly incorporate risk perception and inter-driver communication into a unified model.
• To improve the ability of automated vehicles to understand and predict human driving behaviors.

Main Methods:

• Development of a Communication-Enabled Interaction model.
• The model is based on risk perception and does not assume rational driver behavior.
• Explicitly accounts for communication between drivers in traffic scenarios.

Main Results:

• The model successfully explains and reproduces observed human interactions in a simplified merging scenario.
• The model captures joint behaviors at high-level decisions, safety margins, and low-level control inputs.
• Demonstrates the importance of communication and risk perception in traffic interactions.

Conclusions:

• The developed model enhances the understanding of underlying mechanisms in human traffic interactions.
• This research represents a significant step towards interaction-aware automated driving systems.
• Future AVs can leverage this model for safer and more socially acceptable navigation.