A STOCHASTIC POSITION SENSING SYSTEM BASED ON MACHINE VISION AND WHEEL ENCODERS
Date
1992-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
Type
Degree Level
Masters
Abstract
This thesis presents a system to determine the position of a free-roving autonomous robot in an unknown environment. Data from multiple sensors and multiple readings are fused together to increase the sensing accuracy over that available from any one of the
sensors. The analysis of the data also provides a means to evaluate the statistical level of uncertainty in the position of the robot. The fused sensor data are used to build a World Model or map; since the fused sensor data are stochastic then the features (parameters,
objects) in the World Model are also stochastic. The position is then determined by the relative position of features with respect to the World Model. The sensors used for the analysis are cameras at two locations on the robot, and wheel encoders. Positional accuracy of approximately 9 mm was achieved while the robot moved in an unstructured office environment
Description
Keywords
Citation
Degree
Master of Science (M.Sc.)
Department
Electrical Engineering