MSc Thesis #92158
|Bee-Haviour in a Mobile Robot the Construction of a Self-Organized Cognitive Map and Its Use in Robot Navigation Within a Complex Natural Environment
|This thesis describes a biologically inspired mobile robot map building and navigation system which is modeled on the strategy thought to be employed by honey bees. These insects navigate over long distances via a dead reckoning mechanism and over short distances, where target recognition is important, by a visual image comparison process called global localization. A cognitive map can be formed by combining remembered images, with the compass-bearing from which they were taken, into a topographically ordered map which preserves the important spatial relationships between elements depicted therein. This Department's Ben Hope robot, as used in this project, employs a self-organizing neural network to construct an ordered map from (i) ultra-sound range images and (ii) metric positional data (from the robot's high resolution encoders). Navigation via this cognitive map involves using sensor readings to associate the robot's world position with its corresponding map position, and then planning a route (between this map location and a goal map location) which corresponds to a safe and efficient path through the robot's environment. This system has been developed, for the most part, in simulation and is now ready for full deployment on the robot.
|NO ONLINE COPY