Jennie Gallimore (Advisor)
Doctor of Philosophy (PhD)
The dynamic nature of complex systems and the overwhelming amount of data that must be handled by human operators requires the development of smart systems that augment human capabilities. For these systems human-agent collaboration is part of the system dynamic. Unfortunately the computer agent is not typically perceived by the human operator as a collaborative team member. This research addressed the key question of how to develop and evaluate computer agents with personality to enhance human-machine collaboration. To move towards this goal three research objectives were established: (1) develop agents with personality, (2) validate that humans perceive these agents as having a personality, and (3) evaluate if agents with personality enhance human-machine collaboration. Experimentation and development was carried out in three phases: Phase I: Identify actions, language, and/or behaviors that signify personality traits within a well known personality model (Big Five Factor). Participants rated personality traits and identified actions, language and behaviors that gave them their impressions. Phase II: Model agents in a multimodal environment and validate the personalities. Actions, language and behaviors were modeled into computer agents via visual, auditory, and tactile output supporting a uninhabited combat aerial vehicle (UCAV) military mission. Phase III: Empirically evaluate human-machine collaboration performance when using agents in a complex semi-autonomous UCAV mission. Empirical results indicated that humans were able to identify and perceive significant actions, language, and/or behaviors associated with the computer agent personalities modeled. In addition, human performance (overall simulation score) was significantly greater when interacting with agents having a personality versus an agent with no personality. This research has made a significant contribution to the field of human-machine collaboration by (a) outlining a systematic approach based on modern personality theory for developing computer agents with personality and incorporating personality theory into model development, (b) developing a foundation for an agent personality model by providing methods to capture actions, language, and behaviors that can be used in agent models, (c) identifying significant personality traits that can be modeled in the context of collaboration/teamwork, (d) evaluating whether computer agents that communicate personality via multimodal inputs provide users with perception of personality.
Department or Program
Department of Biomedical, Industrial & Human Factors Engineering
Year Degree Awarded
Copyright 2007, all rights reserved. This open access ETD is published by Wright State University and OhioLINK.