Potential field based approach for multi-threat containment with cooperative robots

Show full item record

Redirect: RIT Scholars content from RIT Digital Media Library has moved from http://ritdml.rit.edu/handle/1850/2626 to RIT Scholar Works http://scholarworks.rit.edu/theses/5491, please update your feeds & links!
Title: Potential field based approach for multi-threat containment with cooperative robots
Author: Mehendale, Bhushan
Abstract: Target or threat containment is a process where a team of robots forms a particular formation around a target so that it gets engulfed in it. The robots under consideration run decentralized Swarm-Algorithms that make them cooperate with each other using their local sensing capabilities. Previous work in this area has dealt with robot formations and single threat containment where the target is either fixed or dynamic. In either case the same targets exist throughout the containment process. The challenge is to use such collaborative robots to contain multiple targets that appear at random times, and expire after some time. In order to accommodate multiple threats, an algorithm named MUltiple Threat Containment Algorithm (MUTCA), was proposed and analyzed in this work. Upon sensing threat occurrences, the robots split up and attempt to surround all threats with a circular formation. MUTCA is based on Quadratic Artificial Potential Functions (QAPF). Multiple QAPFs are carefully designed to allow robots to surround the threats evenly and to avoid collisions. Using MUTCA, each robot periodically and independently calculates the sum of forces acting at their location in order to decide its movements. This work investigates the capabilities and limitations of autonomous robots utilizing MUTCA via simulation using MAHESHDAS - a simulator developed to simulate autonomous robots. Application specific success rates were examined for robot sensor capabilities and parameters used by QAPF. Though working well for the tested scenarios, MUTCA has been found to be sensitive to application specifics and sensor errors. As no prior work has been found to solve such a multi-threat containment problem, this work reveals the issues involved with this challenging problem and analyzes the proposed novel and robust solution approach.
Record URI: http://hdl.handle.net/1850/2626
Date: 2006-09-28

Files in this item

Files Size Format View
BMehendaleThesis09-2006.pdf 2.164Mb PDF View/Open

The following license files are associated with this item:

This item appears in the following Collection(s)

Show full item record

Search RIT DML

Advanced Search