Robot obstacle avoidance by geometric modeling techniques

Authors

Sabyasachi Ghoshray, Florida International University

Document Type

Thesis

Degree

Master of Science (MS)

Major/Program

Electrical Engineering

First Advisor's Name

Kang K. Yen

First Advisor's Committee Title

Committee Chair

Second Advisor's Name

Jean Andrian

Third Advisor's Name

Gustavo Roig

Date of Defense

4-6-1995

Abstract

Avoiding collisions is the primary objective for a robot moving in a workspace with obstacles. The conventional obstacle avoidance algorithms suffer from the drawbacks due to considerable approximations and lack of flexibility. The proposed obstacle avoidance algorithms are developed based upon the geometric modeling of robot workspace using quadtree data structure. Quadtree is a hierarchical decomposition method in which the region is recursively divided into four contiguous squares. A quadtree is generated from the polygonal representation of the obstacles in the workspace, called the obstacle-quadtree. By examining the obstacle-quadtree, nodes that can allow safe passage for the mobile robot are identified. By connecting all such nodes, a set of collision-free paths e generated and subsequently the optimal path is chosen. The advantages of the proposed algorithms include: (a) a quick computation of the shortest path for the mobile robot, (b) flexibility to merge different parts of the robot environment to investigate different navigation strategies, (c) modeling of multiple robots in the same quadtree, and (d) ability to vary the required accuracy of collision detection by changing the resolution of hierarchical decomposition of workspace.

Identifier

FI15101572

Recommended Citation

Ghoshray, Sabyasachi, "Robot obstacle avoidance by geometric modeling techniques" (1995). FIU Electronic Theses and Dissertations. 3988.
https://digitalcommons.fiu.edu/etd/3988