Access to master’s theses is restricted to current ERAU students, faculty, and staff.

Date of Award

4-26-2011

Document Type

Thesis - Open Access

Degree Name

Master of Science in Mechanical Engineering

Department

Mechanical Engineering

Committee Chair

Dr. Charles Reinholtz

Committee Member

Dr. Brian Butka

Committee Member

Dr. Marc Compere

Abstract

There are many different proposed methods for Supervisory Control of semi-autonomous robots. There have also been numerous software simulations to determine how many robots can be successfully supervised by a single operator, a problem known as fan-out, but only a few studies have been conducted using actual robots. As evidenced by the MAGIC 2010 competition, there is increasing interest in amplifying human capacity by allowing one or a few operators to supervise a team of robotic agents. This interest provides motivation to perform a more in-depth evaluation of many autonomous/semiautonomous robots an operator can successfully supervise. The MAGIC competition allowed two human operators to supervise a team of robots in a complex search-and mapping operation. The MAGIC competition provided the best opportunity to date to study through practice the actual fan-out with multiple semi-autonomous robots. The current research provides a step forward in determining fan-out by offering an initial framework for testing multi-robot teams under supervisory control. One conclusion of this research is that the proposed framework is not complex or complete enough to provide conclusive data for determining fan-out. Initial testing using operators with limited training suggests that there is no obvious pattern to the operator interaction time with robots based on the number of robots and the complexity of the tasks. The initial hypothesis that, for a given task and robot there exists an optimal robot-to-operator efficiency ratio, could not be confirmed. Rather, the data suggests that the ability of the operator is a dominant factor in studies involving operators with limited training supervising small teams of robots. It is possible that, with more extensive training, operator times would become more closely related to the number of agents and the complexity of the tasks. The work described in this thesis proves an experimental framework and a preliminary data set for other researchers to critique and build upon. As the demand increases for agent-to-operator ratios greater than one, the need to expand upon research in this area will continue to grow.

Share

COinS