Parallel and real-time simulation of vehicle dynamics

OData support
Supervisor:
Dr. Kiss Bálint
Department of Control Engineering and Information Technology

The purpose of my thesis is to develop a nonlinear controller for autonomous vehicles based on GPGPU programming, which becomes more and more popular these days.

%The core of the algorithm is an NMPC (Nonlinear Model Predictive Control), although the optimal control signals are not calculated with an regular control mechanism (such as LQR, PID) but It is calculated by an optimizer method on the GPU.

The core of the algorithm is an NMPC (Nonlinear Model Predictive Control), although the optimal control signals are calculated by an optimizer method on the GPU instead of a regular control mechanism (such as LQR, PID).

The main purpose of the algorithm is to exploit the resources provided by the GPU, which enables the real-time use of parallel optimization procedures. During my thesis I have implemented several modern optimization algorithms, of which the PSO proved to be the most effective in terms of the task, so the thesis also deals with a separate chapter. In order to use the algorithms in real time, the thesis used some heuristic ideas. The implemented algorithm has been validated during simulations and on real vehicles.

Downloads

Please sign in to download the files of this thesis.