Human-Like Lane Change Decision Model for Autonomous Vehicles that Considers the Risk Perception of Drivers in Mixed...

Area of Science:

• Autonomous Vehicle Safety
• Human-Machine Interaction
• Road Traffic Psychology

Background:

• Lane change decisions in autonomous vehicles (AVs) are critical for safety.
• Existing models often neglect the risk perception of surrounding drivers.
• Human-like lane change behavior requires understanding driver psychology.

Purpose of the Study:

• To develop a human-like lane change decision model for AVs.
• To establish two-level thresholds for safe lane changes, considering rear and subject vehicle driver risk perception.
• To enhance AV safety and politeness during lane change maneuvers.

Main Methods:

• Utilized signal detection theory and extreme moment trials on a real highway.
• Defined the Minimum Safe Deceleration (MSD) of the rear vehicle as the safety indicator.
• Calculated MSD using a proposed human-like lane-change decision model.

Main Results:

• Drivers in rear-approaching scenarios are more conservative than those in front-approaching scenarios.
• Determined a primary safe threshold of 0.85 m/s² and a secondary threshold of 1.76 m/s².
• The developed model successfully calculated the MSD for lane change safety.

Conclusions:

• The lane change decision model enhances AV safety and politeness.
• The identified thresholds meet the safety expectations of both subject and rear vehicle drivers.
• Improved AV lane change behavior increases acceptance and ensures road safety.