A Study on Dynamic Motion Planning for Autonomous Vehicles Based on Nonlinear Vehicle Model.

Area of Science:

• Robotics
• Artificial Intelligence
• Control Systems

Background:

• Autonomous driving relies heavily on motion planning and trajectory tracking, areas requiring urgent improvement.
• Current path planning methods often simplify vehicle dynamics using linear models to achieve real-time performance, sacrificing accuracy.
• Accurate simulation of vehicle dynamics and kinematics alongside real-time computation is crucial for enhanced autonomous driving performance.

Purpose of the Study:

• To develop an advanced motion planner that integrates nonlinear vehicle characteristics for superior performance.
• To enhance dynamic obstacle avoidance capabilities in intelligent interconnected vehicles.
• To ensure real-time path planning that aligns with the vehicle's dynamic behavior.

Main Methods:

• Implementation of a Takagi-Sugeno fuzzy-model-based closed-loop rapidly exploring random tree (CL-RRT) algorithm for motion planning.
• Direct integration of a nonlinear vehicle model into the motion planner design.
• Application of Takagi-Sugeno fuzzy modeling for real-time utilization of vehicle state in path planning.

Main Results:

• The proposed motion planner effectively improves dynamic obstacle avoidance.
• Local obstacle avoidance paths are generated in accordance with the vehicle's dynamic characteristics.
• Numerical simulations demonstrate the planner's ability to generate efficient reference trajectories with a lower replanning rate.

Conclusions:

• The Takagi-Sugeno fuzzy-model-based CL-RRT approach successfully addresses the limitations of linear models in motion planning.
• This method enables real-time path planning that accurately reflects vehicle dynamics, enhancing safety and efficiency.
• The developed motion planner offers a significant advancement for autonomous driving systems, particularly in complex dynamic environments.