ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Algorithm D*

Matjaž Verbole (2012) Algorithm D*. EngD thesis.

Download (863Kb)


    Path planning is an important part of the navigation of autonomous mobile robots. Mobile robots often operate in a partially or entirely unknown world. In fact, it would be impractical for a robot to fully know the world before driving through it, as the robot would have to memorise a large quantity of data. It is also true that cases when information about the environment is entirely unknown prior to planning the path are very rare. In 1994 Dr Anthony Stentz discovered algorithm D*, which combines global and local planning, and enables effcient use of information obtained from the prepared map of the environment and from the robot's sensors as it moves around the area. The algorithm is effective, because when it needs to find a new optimal path (replan) it makes use of the already calculated information instead of recalculating everything. The diploma thesis examined several algorithms and techniques for finding optimal paths in graphs. Algorithms have been described which use uninformed techniques (BFS, DFS, and Iterative Deepening) for exploring, as well as informed techniques which use heuristic approaches for exploring. Since algorithm D* has several versions, these were also given a closer look. Thus in addition to algorithm D*, algorithms Focused D* and D* Lite were described as well. The algorithm D* Lite has been derived from algorithm LPA*, hence the latter was described too. Algorithms D*, Focused D*, D* Lite and LPA* were implemented in the Python programming language, their speeds measured, and the results of the measurements presented at the conclusion of the paper.

    Item Type: Thesis (EngD thesis)
    Keywords: path search, heuristic, algorithm, D*, unknown environment, replanning
    Number of Pages: 63
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    prof. dr. Borut Robič28Mentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=50070&select=(ID=00009508180)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 1915
    Date Deposited: 26 Oct 2012 15:28
    Last Modified: 15 Nov 2012 10:20
    URI: http://eprints.fri.uni-lj.si/id/eprint/1915

    Actions (login required)

    View Item