Learning environment with mobile robot using only contact sensor

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

Due to the fast advencements of control theory and electronics in the 20th. century, people have been able to contstruct devices which need no human supervision during their operation and relieve humans of burden at physical work, tasks with loads of complex calculations or at medical interventions. Such an operations can be done with them, which was mostly impossble with merely human power and skills. Robots belong to these kind of devices, like robots on the industrial assembly lines, surgical robots or a Mars rover.

Mobile robots were also get used by sciencific researches. They let us to reach places that can hardly or even never be accessible without risking a human life, because of limits of the human body. However, the minimum condition to control one of the mobile units staying in the uncertainty is to have some information about its environment and be able to localize the robot in it. The question arises, that what degree of details we need to know about the unknown environment to be able to localize the robot and navigate it to an arbitrary point. This is directly proportional to the complexity of the robot and therefore to its costs.

This thesis deals with a strategy that can be used with a simple mobile robot unit to map a polygonal environment. Then, with possession of the information about the environment another algorithm is able to construct a path in it provided the path exists. Traveling along the path result in catching any evaders moving unpredictably in the environment. First of all a graphical user interface (GUI) was developed in MATLAB framework. The algorithms mentioned above can be simulated with it. On the other hand a simple robot was produced which can be placed to a real environment. Then in connection to the GUI and it can execute both the actions given by the mapping and pursuit-evasion algorithms running at the GUI in order to do mapping and pursuit evaders in a real environment.


Please sign in to download the files of this thesis.