06.03.2013 Views

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

24.5 Offline Planning by Generalized<br />

Voronoi Diagram (GVD)<br />

The construction process of GVD [28] by Cellular Automata consists of two<br />

main steps. In step 1, the boundaries of each obstacle <strong>and</strong> the inner space<br />

boundary of the environment are filled in with a numeral say 1(One). As the<br />

distance from the obstacles <strong>and</strong> the inner boundary increases, the coordinates<br />

of the space will be filled in with gradually increasing numbers. The process<br />

of filling in the workspace, by numerals, is thus continued until the entire<br />

space is filled in. In step 2, the highest numerals are labeled <strong>and</strong> its<br />

neighboring coordinates containing the highest numerals or next to the highest<br />

are labeled. The process of labeling is continued until each obstacle is<br />

surrounded by a closed chain of labeled numerals. Such closed curves are<br />

retained <strong>and</strong> the rest of the numerals in the space are deleted. The graph, thus<br />

constructed, is called the GVD. An example shown above (vide fig. 24.4 (a) &<br />

(b)) demonstrates the construction process of GVD.<br />

Once the construction of a Voronoi diagram is complete, the shortest<br />

path from the starting position <strong>and</strong> the goal position of the robot to the graph,<br />

depicting the Voronoi diagram, are evaluated. A heuristic search may now be<br />

employed to find the shortest path on this graph.<br />

24.6 Path Traversal Optimization Problem<br />

There exist many approaches to h<strong>and</strong>le this type of problem. We will however<br />

limit our discussion to i) Quadtree based heuristic search <strong>and</strong> ii) Genetic<br />

algorithm based scheme.<br />

24.6.1 The Quadtree Approach<br />

The quadtree is a tree, in which each node in the tree will have four children<br />

nodes. We can represent the given 2- dimensional image map in the form of<br />

quadtree by a recursive decomposition. Each node in the tree represents a<br />

square block of the given image map. The size of the square block may be<br />

different from node to node. The nodes in the quadtree can be classified into<br />

three groups: i) free nodes, ii) obstacle nodes <strong>and</strong> iii) mixed nodes.<br />

A free node is a node that has no obstacle in the square region<br />

represented by it. An obstacle node is a node whose square region is totally<br />

filled with obstacles. A mixed node's square region is partially filled with<br />

obstacles. For example, consider the image map given in fig. 24.5.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!