Model Predictive Controller Design for an Autonomous Corner Module

Model Predictive Controller Design for an Autonomous Corner Module

De Geus, V.

Verhaegen, M. (mentor)
Gerard, M.P. (mentor)

Mechanical, Maritime and Materials Engineering

Delft Center for Systems and Control

2010-04-12

With the growing interest for vehicles with electric propulsion, research has been done on more modular car designs where the driveline and steering system is integrated in the wheel corners. Such a corner module containing an in-wheel motor and active brake and steering system is denoted as an Autonomous Corner Module (ACM). Using this module tyre forces can be freely assigned within the chassis solution, which could lead to improved stability. The goal of the thesis is to design a controller for an ACM which needs to generate two desired output forces (longitudinal and lateral) using the inputs (wheel spin velocity and steering angle). The controller design is restricted to the field of Model Predictive Control, since this method can deal with nonlinearities, multivariable systems and constraints. Throughout the thesis the potential of MPC for this particular application is investigated. Two types of controllers are proposed, the first type is a Non-linear MPC where the non-linear model is directly used in the online optimization. The design method allows for easy adjustment of the model, but the non-linear optimization is time consuming and therefore only applicable at velocities up to 20 km/h. To speed up the computation time, a second controller is proposed, which is a Linear Time Varying MPC. This type of controller is faster than the NMPC, since the optimization is performed on a local linearization of the non-linear model. Simulation results show that this controller could be a serious option for online implementation for velocities up to 50 km/h.

http://resolver.tudelft.nl/uuid:bbc47cd4-1d55-445a-828b-7261e08b4621

Student theses

master thesis

(c) 2010 De Geus, V.