Adaptive autonomous emergency braking model based on weather conditions

Area of Science:

• Automotive Safety Engineering
• Artificial Intelligence in Transportation
• Machine Learning for Vehicle Dynamics

Background:

• Vehicle active safety systems, such as autonomous emergency braking (AEB), are crucial for collision avoidance.
• Traditional AEB systems rely on safety distance calculations optimized for normal weather conditions.
• Adverse weather significantly degrades the performance and early warning capabilities of conventional AEB systems.

Purpose of the Study:

• To develop an adaptive AEB system algorithm capable of functioning effectively under adverse weather conditions.
• To enhance the safety and reliability of AEB systems in challenging environmental scenarios.
• To improve collision avoidance performance during inclement weather like rain and haze.

Main Methods:

• A multilayer perceptron (MLP) model was employed to analyze accident and weather data.
• The MLP model was trained to predict accident severity, which informed the adaptive AEB algorithm.
• The adaptive AEB model was validated using Prescan simulations and a driver-in-the-loop system.

Main Results:

• The adaptive AEB system demonstrated superior performance compared to traditional AEB systems in adverse weather simulations.
• Testing confirmed that the adaptive AEB algorithm enhances safety and reliability under conditions such as rain and haze.
• The system successfully increased safety distances in rainy conditions and prevented collisions in hazy environments.

Conclusions:

• The developed adaptive AEB system effectively mitigates the performance degradation of AEB in adverse weather.
• The adaptive AEB algorithm offers a significant improvement in vehicle safety and collision avoidance during inclement weather.
• This research highlights the potential of AI-driven adaptive systems for enhancing automotive safety in real-world driving conditions.