Design and implement MPC path tracking controller on embedded platform

Model predictive control is an advanced control method that is used to control a process while satisfying a set of constraints. In the field of autonomous driving it is widely used as a path tracking control technique. Different implementation of MPC exist for the path tracking problem, such as linear MPC, non-linear MPC, lateral and longitudinal MPC as two separate problems, lateral and longitudinal combined MPC etc.

At Capgemini Engineering we are testing an Automated Valet Parking system in a Hardware-in-the-loop environment. The autonomous driving software we are using is Autoware.Auto. Currently, there is an MPC controller running as part of Autoware.Auto in a ROS2/C++ environment on a Linux based machine. Our whish is to have a similar MPC implementation, but implemented on an automotive grade ECU platform, such as the MicroAutobox II. Therefore, the implementation of the MPC controller will have to be done in MATLAB/Simulink.      

1. 1. Assignment

The goal of the assignment is to build an MPC path tracking controller in MATLAB/Simulink that is to be implemented on an automotive grade ECU platform. The assignment includes:

• Find a reference implementation in another programming language, such as python or C++, which can serve as the starting point for implementation in MATLAB/Simulink. On platforms like Github several opensource initiatives are availabe.
• Develop the algorithm in MATLAB/Simulink without using any paid toolboxes
• Compile the model and implement it on the automotive ECU platform
• Test the path tracking controller with the complete setup, where it will replace the Autoware.Auto MPC implementation running in ROS2
• Benchmark performance to Stanley Path tracking controller
  1. Technical aspects

•  Python, C++, MATLAB/Simulink dSPACE Scalexio, “Hardware In the Loop” (HIL) setup, dSPACE MicroAutobox/other platform.
•  Development is done using MATLAB/Simulink, knowledge of python, C++, ROS or Autoware is expected as well as experience with Control Systems design.
• Strong mathematical background

1. 1. Other:

Level of the assignment: This assignment can be done by a HBO+/WO (master) student, where the assignment can be adjusted according to the educational direction/level.

Educational direction: Automotive, Mechanical, Electrical, Computer science.

Contact persons: Jeroen Schutte - jeroen.schutte@capgemini.com or Lucas Raemakers - lucas.raemakers@capgemini.com

Duration of assignment: 4-6 months, standard HBO/WO graduation period, including research, implementation, thesis writing.