AbstractsComputer Science

Theoretical thesis. Bibliography: pages 203-230. Chapter 1. Introduction  – Chapter 2. Background  – Chapter 3. Related work  – Chapter 4. Approach  – Chapter 5. Evaluating the plausibility of the proposed agent architecture and the communication model  – Chapter 6. Evaluating the impact of agent multimodal communication on human-IVA teams  – Chapter 7. The impact of virtual agent personality on human-IVA team performance  – Chapter 8. Discussion  – Chapter 9. Conclusion. There is a growing interest in the use of heterogeneous teams comprised of humans and Intelligent Virtual Agents (IVAs). Human teamwork studies have provided cumulative knowledge about team features and performance. How well this knowledge transfers to human-IVA teams requires further investigation. The development of a Shared Mental Model (SMM) between team members and effective communication of the shared knowledge have been found to improve human teamwork performance. In human-IVA heterogeneous teams, the communication required to develop a SMM is further complicated as each party belongs to different worlds (i.e. real and virtual). Communication is a vital factor in the collaboration between team members. Creating IVAs that are able to communicate with humans in Virtual Environments (VEs) is a challenging research area. When both the IVA and the human user should communicate together while performing a collaborative activity, communication becomes more critical and the challenge becomes more difficult. Moreover, humans may differ in how they produce and perceive communication acts according to their personality traits. The main aim of my PhD is to study the factors that tend to improve team performance and foster collaboration between humans and IVAs in VEs. To understand the requirements of human-IVA collaboration in VEs, we present the design of a framework based on Activity Theory called Multi-Agent Collaborative VirtuaL Learning Environment (MACVILLE), which is a framework to understand the nature of collaboration in human teams. The MACVILLE framework indicated the importance of communication for collaboration in the VE. In addition, the proposed framework demonstrates the need to extend the design of an IVA to include collaborative and social abilities. To address this essential extension we propose an agent architecture that handles two-way human-agent collaboration. To support human-IVA communication in VEs, we present Human-Agent Teamwork Communication Model (HAT-CoM). HAT-CoM was designed based on Speech Act Theory (SAT), which is a methodology to understand the structure of human speech. HAT-CoM was implemented and integrated into our agent architecture. VII To evaluate the impact of HAT-CoM on developing/breaking a SMM between a human and an agent, a study was conducted with 66 undergraduate students. The evaluation was conducted by analytical and inductive means. The analytical evaluation aims at investigating the impact of HAT-CoM components, i.e. verbal and non-verbal, on the development of SMM features, i.e.… Advisors/Committee Members: Macquarie University. Department of Computing.