Innovating driving solution for numerical multibody systems Oriented towards bikes

Details

Supervisors

Docquier, Nicolas ; Fisette, Paul

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil mécanicien, à finalité spécialisée

Abstract

enIn the study of multibody system, it is sometime required to control the system in a certain aspect in order to see how it will behave. And this is the case for bikes, where we need to keep it in balance throughout the trajectory we desire to follow. We will thus create one of those controller, in our case it is a predictive control solver capable of handling all type of tree-like topologies multibody system. In a predictive control solver, we predict the movement of the system for a certain amount of time and optimise that prediction via a certain cost function. This solver requires knowing the equation of motion of the system and the control variable it can change in order to control the behaviour of the system. Those kinds of controller are very capable, robust and versatile controller with the main drawbacks of needing some computation time. We have created this controller by using two numerical tools, the first one is Casadi, a nonlinear solver develop at the KUL, and the second is Robotran, a multibody environment develops at the UCL. We will explain how the controller work and present the result and experience to show the capabilities of such controller. We will also talk about its advantages and drawbacks. Furthermore, we will go from a simple analytical model to a complete multibody system of a bike, and each time we will improve the controller. This work is oriented towards bike, but this solver could be used for every tree-like topologies multibody system.