Hands-off detection in electric power assist systems using neural network

OData support
Dr. Harmati István
Department of Control Engineering and Information Technology

In many fields of science and technology, so also in car industry, especially in the development of autonomous functions, the artificial intelligence-based applications are spreading continuously and reaching promising results.

On the way of fully automated vehicles increasing number of autonomous functions are introduced (such as lane keeping assistant, or parking assistant systems). Some of them are based on artificial intelligence algorithms. Currently the lack of legal regulations and the prevention of possible failures imply that the operation of autonomous functions is bonded to the attention of the driver. They are allowed to operate only if the system is sure, that the driver pays attention.

There is a need for recognizing the attention of the driver in electric power assisted steering systems (EPAS) also. To detect it, the Hands-off Detection (HFD) function is developed. The point of the function is to be able to detect, whether the driver pays attention to the driving or not, without additional sensors, using only the originally measured signals. In EPAS the function is developed by using neural network.

This thesis gives an overview of the architecture and operation of the currently used thyssenkrupp Presta EPAS systems. After that, it goes through the general operation and architecture of the neural networks, describing its mathematical background, and gives also some application examples. Then, it describes the development steps of the HFD function, the implementation of the neural network, the selection and processing of the used signals, and the optimization of the network’s parameters.

At the end of the thesis, the developed function is presented through simulation tests in Matlab.


Please sign in to download the files of this thesis.