Model Predictive Controller Approach for Automated Vehicle's Path Tracking

Area of Science:

• Automotive Engineering
• Control Systems
• Robotics

Background:

• Automated vehicle steering requires precise path tracking for safety and efficiency.
• Existing control methods often simplify vehicle dynamics, potentially limiting performance.
• Model Predictive Control (MPC) offers a promising framework for complex trajectory following.

Purpose of the Study:

• To develop and evaluate a Model Predictive Control (MPC) strategy for automated vehicle steering path tracking.
• To investigate the impact of different steering dynamics models (first-order vs. second-order lag) on control performance.
• To introduce a novel method for reference trajectory generation in the vehicle's ego coordinate system.

Main Methods:

• A Linear Parameter-Varying (LPV) vehicle plant model incorporating steering dynamics was developed.
• A cascade MPC structure was implemented, separating steering dynamics into a second MPC.
• The proposed MPC approach and reference trajectory generation were validated through simulations and on a test vehicle.

Main Results:

• The first-order steering model demonstrated slightly more accurate path-following than the second-order model.
• Including steering dynamics in the prediction model significantly enhanced controller performance.
• Both simulation and experimental results confirmed the effectiveness of the novel reference definition method.

Conclusions:

• The proposed MPC approach effectively controls automated vehicle steering for accurate path tracking.
• The inclusion of detailed steering dynamics models is crucial for achieving high controller performance.
• The novel reference trajectory generation method proved effective in practical applications.