In addition to industrial applications, mobile robotics are increasingly present in everyday life. The most common topic of speech is the self-driving car, which, besides the great interest, raises a number of problems to solve. One such fundamental question is the motion planning in the presence of obstacles. There are many algorithms for mobile robot motion planning, but the kinematic constraints of car-like robots are a further complication. In the course of my thesis I deal with one of the subtasks of motion planning, pathplanning, especially for car-like robots.
In the first part of my thesis I present two pathplanning algorithms and their implementations. During pathplanning, I pay special attention to the kinematic constraints of car-like robots and the associated difficulties. I compare the algorithms according to their effectiveness and usability.
In the second part of the thesis I deal with the application of motion planning in real environment and the problems that arise during this task. I present a framework that allows components of a software running on a real mobile robot to work together. An important part of motion planning is mapping the environment and taking advantage of it to detect obstacles. The framework provides several tools for this, which I use to integrate the system with the algorithms I have previously presented. Using the potentials of the framework, the path planning algorithms will be able to function as part of a real system and provide an important subtask of self-driving in real time. The system I have created is tested in simulation and then in real environment.