| In this thesis, I present a Multi-Input Multi-Output (MIMO) data-acquired controller using a system of Hierarchy-Structured Decision Tables for a simulated vehicle driving control problem. The simulator incorporates the dynamic mathematical model of a vehicle driving on a track. Sensor readings and expert driver control actions are accumulated to derive the vehicle control model. Sensor readings include random error to reflect realistic data acquisition conditions. The methodology of Rough Sets is being used to process the data and to automatically derive the control algorithm.; With the help of a novel progressive binary-split discretization method, a novel algorithm HDTL is adapted to efficiently derive Hierarchy-Structured Decision Tables from training sample data. Compared to the traditional quantization methods, (such as fixed-level, best-discernibility binary-split, etc.), this new discretization method can maintain a better balance between high-precision and wide-coverage for the resultant rule-base, thus achieving better performance in dealing with continuous condition variables encountered in the project.; In the experiments, with the help of a novel stability enhancement training protocol, the automated vehicle control algorithms derived from different driving patterns and with substantial sensor errors consistently demonstrate astonishing robustness in their ability to properly drive the vehicle. |
