|
|
Expert Systems and Advanced Algorithms in Mobile Robots Path Planning
Abbadi, Ahmad ; Šeda, Miloš (oponent) ; Volná,, Eva (oponent) ; Matoušek, Radomil (vedoucí práce)
Motion planning is an active field in robotics domain, it is responsible for translating high-level specifications of a motion task into low-level sequences of motion commands, which respect the robot and the environments constraints. In this work many path-planning approaches have been reviewed, mainly, the rapidly exploring random tree algorithm (RRT), the cell decomposition approaches (CD), and the application of fuzzy expert system (FES) in motion planning. These approaches have been adapted to solve some of mobile robots motion-planning problems efficiently, i.e. motion planning in small and narrow areas, the global path planning in dynamic workspace, and the improvement of planning efficiency using available information about the working environments. New planning approaches have been introduced based on exploiting and combining the advantages of cell-decomposition, and RRT, in addition to use other tools i.e. fuzzy expert system, to increase the efficiency and completeness of finding a solution. This thesis also proposed solutions for other motion-planning problems, for example the identification of narrow area and the important regions when using sampling-based algorithms, the path shortening for RRT, and the problem of planning a safe path. All proposed methods were implemented and simulated in Matlab to compare them with other methods, in different workspaces and under different conditions. Moreover, the results are evaluated by statistical methods using Matlab and Minitab environments.
|
host ::
RSS.