Date of Award
2006
Degree Type
Thesis
Degree Name
Master of Engineering Science
Program
Electrical and Computer Engineering
Supervisor
Dr. Kenneth McIsaac
Abstract
This thesis describes an algorithm to successfully maneuver a team of two mobile robots in an unknown environment without formation and the ability to recover from temporary loss of communication. The team searches out the environment completely while maintaining a map of the environment. The robots avoid obstacles and collisions with each other while also staying within a communications range. The algorithm is designed to minimize the search time by having the robots search in parallel as much as possible to map the greatest amount of area. Currently there are many diverse algorithms for map building, all having their own advantages and disadvantages. Soft computing is easily updatable for minor changes but require a training set for an environment that is unknown. Algorithmic control may be harder to manipulate to compensate new behavior but is consistent in the results given by the algorithm used. The algorithm implemented in this thesis makes use of potential fields in order to minimize complexity and make best use of computational power. Since potential fields require a single target to strive towards, an algorithm to find the next best target is implemented. The next step was to expand the potential field to achieve a hierarchy of multiple goals and constraints. These constraints are to avoid obstacles (and each other) while maintaining communication and to take advantage of parallelism and have the robots search in parallel whenever possible.
Recommended Citation
Webber, Erik, "A Potential-field Based Method for Resolving Multiple Navigation Goals" (2006). Digitized Theses. 4684.
https://ir.lib.uwo.ca/digitizedtheses/4684