Sonar-based Mobile Robot Navigation with Probabilistic Reasoning
Thesis submitted for the M.SC. Degree

Departement of Mathematics and Computer Science
Ben-Gurion University, Beer-sheva, 84-105, Israel.
e-mail : ami@cs.bgu.ac.il
February 1997

Abstract

Mobile robot navigation in an unknown world is the principal subject of this research. In this world we have to supply the mobile robot with the appropriate sensing equipment for receiving information from the world and decoding it.
The thesis represents knowledge as a probabilistic model of sensing system actions. Our research combines theoretical knowledge representation models, of uncertainty with implementation problems.
The central items in the thesisi are sensor-fusion problems, orientation in uncertain enviroment and mapping the obstacles in the world.
Robot navigation in an unknown world with sonars is formalized as an uncertainty problem. The mapping model allows for solving the problem in reasonable computation time.
We reduce the number of states that are necessary for the world representation, by dividing the world into zones, and implement this with a grid, and build a probabilistic network based on the zones.
Our model emphasizes the dependence between the zones. The model's architecture is implemented by a computer program.
Initial testing, in different configurations, demonstrates the advantage of our model in contrast with another models with independence between the zones



Back to the publications page
Back to the main page