adaptive_mpc

This repository models vehicles using vehicle driving data. It implements MPC using that model for lateral control of vehicle track following. Comparing the track following error with MPC using a two-wheel model, it can be seen that MPC using driving data is able to follow the track with a smaller RMSE.

How to use

1. Open data_driven_mpc.mlx.
2. If you do not provide your own driving data or reference trajectory for testing, execute all code.
3. Finally, a graph is displayed showing the tracking error between the MPC using the two-wheel model and the MPC using driving data.

If you need to make driving data or reference trajectory, you can use Manual_Trajectory_Generator.slx to make them.

Required toolbox

The following toolbox is used in this model.

• MATLAB®
• Simulink®
• Model Predictive Control Toolbox™
• Vehicle Dynamics Blockset™
• System Identification Toolbox™

License

Since this is a modified version of The MathWorks, Inc. sample code, the license is in accordance with the one created by The MathWorks, Inc. See LICENSE for details.

Reference websites

1. Simulation of longitudinal and lateral vehicle motion with Vehicle Dynamics Blockset

2. Design adaptive MPC in a black box

3. MPC Implementation Example

4. About MPC implemented in Autoware