In this thesis, the problem of interest is to provide measurement data about common algorithms in motion planning. These are industrial algorithms that control multi-joint robotic arms and drive wheeled vehicles. The whole framework and the implementation of algorithms have been developed on Web platform for educational purposes.
A brief description of the chosen platform and the used technologies are summarized in the first chapter. After that, terminology, definitions (like Configuration space and differential constraints) and mathematical formulation of the problem statement follows.
In the main part, there are six planning tasks described. Some of them are classical problems in motion planning, there are planar and spatial ones. The descriptions include implementation notes and details. Three well-known algorithms have been chosen as subjects of measurement: Probabilistic RoadMap and two variations of Rapidly exploring Random Tree. These algorithms are described in detail, alongside with their implementation.
Finally, relevant metrics are chosen in the last part to characterize the selected algorithms. The algorithms are run on the planning problems, and statistics are collected about these metrics. These statistics are then used to compare the algorithms’ behaviour in different situations.