Integrated RBF-EKF observer and MPC for simultaneous trajectory tracking and stability control of 4WID-EVs

Area of Science:

• Vehicle Dynamics and Control
• Robotics and Autonomous Systems
• Artificial Intelligence in Engineering

Background:

• Achieving accurate trajectory tracking and stable motion control for 4WID-EVs under extreme conditions is challenging.
• Direct measurement of critical states like sideslip angle is difficult, hindering precise vehicle control.
• Existing control methods may not adequately balance tracking performance with vehicle stability.

Purpose of the Study:

• To develop novel algorithms for enhanced state estimation and trajectory tracking control of 4WID-EVs.
• To improve the accuracy of estimating key vehicle states, particularly the sideslip angle.
• To design a robust controller that ensures both high tracking accuracy and vehicle stability.

Main Methods:

• A validated 7-DOF vehicle dynamics model was used as a simulation platform.
• A hybrid Radial Basis Function Neural Network (RBFNN)-Extended Kalman Filter (EKF) observer was proposed for accurate sideslip angle estimation.
• A Linear Time-Varying Model Predictive Control (LTV-MPC) strategy was implemented, incorporating stability constraints.

Main Results:

• The RBF-EKF observer improved sideslip angle estimation accuracy, reducing Root-Mean-Square Error (RMSE) by up to 30.43% compared to standard EKF.
• The LTV-MPC controller significantly reduced lateral tracking error (0.30 m) and yaw angle error (3.11°) in double-lane-change maneuvers.
• The proposed methods ensured vehicle stability within operational boundaries during dynamic maneuvers.

Conclusions:

• The developed hybrid observer and LTV-MPC controller effectively address the challenges in 4WID-EV state estimation and trajectory tracking.
• The proposed approach enhances vehicle safety and performance, particularly under demanding driving conditions.
• Simulation and HIL test results validate the superiority of the novel algorithms over traditional methods.