Virtual parking path planning in narrow roads based on fuzzy pure pursuit algorithm.

Area of Science:

• Robotics and Control Systems
• Automotive Engineering
• Artificial Intelligence

Background:

• Traditional Pure Pursuit (PP) algorithms struggle with adaptability and accuracy in parking due to fixed look-ahead distances.
• Parking maneuvers require precise path tracking and heading angle control, which are challenging for standard algorithms.

Purpose of the Study:

• To develop an adaptive automatic parking path tracking control algorithm.
• To enhance the adaptability and tracking accuracy of Pure Pursuit (PP) algorithms in parking scenarios.

Main Methods:

• Proposed a Fuzzy Pure Pursuit (FPP) algorithm integrating a fuzzy controller.
• The fuzzy controller adaptively adjusts the look-ahead distance using speed and curvature proportionality coefficients.
• Validated the algorithm using Prescan/CarSim/Simulink simulations and the Apollo advanced platform.

Main Results:

• FPP demonstrated superior tracking performance compared to traditional PP algorithms in vertical parking simulations.
• Achieved a 4.8% reduction in tracking error and a 7.3% reduction in heading angle error.
• Successful path tracking and collision-free parking were confirmed on the Apollo platform under various initial conditions.

Conclusions:

• The Fuzzy Pure Pursuit (FPP) algorithm significantly improves path tracking accuracy and adaptability in automatic parking.
• FPP exhibits excellent environmental adaptability, making it suitable for real-world parking applications.
• The developed algorithm offers a robust solution for precise vehicle control during parking maneuvers.