Trajectory Tracking Control of Autonomous Ground Vehicles Using Adaptive Learning MPC.

Area of Science:

• Robotics and Control Systems
• Autonomous Vehicle Navigation
• Adaptive Control Theory

Background:

• Autonomous ground vehicles (AGVs) require precise trajectory tracking for safe and efficient operation.
• Perturbations, input constraints, and unknown system parameters pose significant challenges to AGV control.
• Existing robust model predictive control (MPC) methods can be conservative, leading to suboptimal performance.

Purpose of the Study:

• To propose an adaptive learning model predictive control (ALMPC) scheme for perturbed AGVs.
• To enhance trajectory tracking accuracy and robustness under input constraints and system uncertainties.
• To reduce the conservatism associated with traditional robust control methods.

Main Methods:

• Developed a set-membership-based parameter estimator using recursive least-squares (RLS) to estimate unknown system parameters.
• Integrated the estimated parameters into an MPC framework to improve prediction accuracy.
• Introduced an adaptive robustness constraint within the MPC optimization, with an off-line determined shape and an online updated shrinkage rate based on estimation error bounds.

Main Results:

• The proposed ALMPC scheme effectively handles parametric and additive uncertainties through the adaptive robustness constraint.
• The method demonstrated reduced conservatism compared to standard robust MPC techniques.
• Theoretical analysis confirmed the recursive feasibility of the ALMPC algorithm and the input-to-state stability (ISS) of the closed-loop system.

Conclusions:

• The developed ALMPC scheme provides a robust and adaptive solution for trajectory tracking of AGVs.
• The combination of parameter estimation and adaptive robustness constraints enhances control performance.
• Numerical simulations validated the efficacy and superiority of the proposed method over existing approaches.