Avatars, Agency and Performance: The Fusion of Science and Technology within the Arts
Richard Andrew Salmon 2014
1.1 The Project
The aim of this thesis was to develop a deeper understanding of the nature and context within which interactions between a robotic exhibit (dynamic avatar and conversational agent) called the Articulated Head (AH) and a human interacting with it took place. The main focus of the research was directed at understanding the nature of the experience of avatar, conversational agent and audience interaction within the constraints and context of the Articulated Head’s exhibited status: that of an enclosed situated robotic exhibit in a public exhibition space. The Articulated Head is predominantly referred to in the past tense because the project is essentially over. The hardware components of the Articulated Head were decommissioned from exhibition at the Powerhouse Museum in Ultimo, Sydney, Australia in November 2012.
The purpose of this research
The purpose of this research was to consider how interaction between humans and the Articulated Head could be improved. Therefore, a primary goal of this investigation was to find ways in which human engagement was enhanced during these interactions.
The following research questions underpinned the investigation:
a. In which ways ‘has and can’ the avatar be given agency? b. In which ways ‘is and can’ the avatar be multi-sensory and multi- present? c. In which ways ‘can and do’ the characteristics of a `virtual’ performer (e.g., personality, nuance of response, context) condition affect and engagement?
Auditory-visual creative new media additions (see section 6) were added to the Articulated Head’s exhibit space with the express intention that they would help contribute to enhancing the interactive experience for the active audience visiting the Articulated Head. These auditory-visual additions were tested to determine whether they would enhance active audience engagement and thereby improve the exhibits interaction design.
An introduction the Articulated Head and Stelarc (the concept artist) are given in Section 2 followed by a historical and contextual view of the exhibit in relation to other robotic artworks. The general history of human-machine interaction and established concepts are then presented, followed by an exposition of the operational architecture of the Articulated Head, including other practical and theoretical considerations pertinent to the investigation.
The methodological framework within which this investigation was situated had at its centre Grounded Theory and Phenomenology. The reasons for choosing this methodological framework are explained in Section 3.
The methods utilised in data collection were:
1. Electronic storage of textual input/output conversational data that was exchanged between the human and the machine in these interactions. 2. Video Cued Recall Interviews and questioning of research participants in relation to their interactions. 3. Researcher observational data.
The methods used in analysis of data were:
1. A word frequency/phrase analysis of textural input/output data. 2. An Interpretive Phenomenological analysis of the Video Cued Recall Interviews and their associated interaction data using a computer assisted qualitative data analysis software (CAQDAS) program called NVIVO (“NVivo 10 research software for analysis and insight,” 2012) (see Section 7).
The findings from this research identified and examined various phenomena that occurred during both the recorded Video Cued Recall Interviews, and the observed human machine interactions that took place. A significant finding of this investigation was the identification of “The Freakish Peak Phenomenon” (see Figure 2-1 section 2.4.3 & Figure 9-2 section 9.2.12). This Phenomena and other findings as examined in Sections 7, 8 & 9 formed the basis of “The Blueprint of Emergent Recommendations”, a group of suggested design refinements presented in a sequence of diagrams (Figures 9-3, 9-4 & 9-5) and their accompanying text in Section 9. The suggested design refinements take direct account of the findings from this investigation and target improvement of the interaction design of this and similar types of interactive exhibits accordingly.
Findings from this research are of relevance and consequence to other similar interactive exhibits and situated robotic designs that incorporate Chatbots and involve human machine interaction. The recommendations for realignment of the design and layout are also of consequence for the conclusions of this work in that they directly address the core question of this research project: How can interaction between humans and machines be improved? The design refinements link back to the research questions a, b, and c on page 1 of this thesis in that they offer a range of suggestions as to how the avatar can be given agency, can be multisensory and multi present, and how the characteristics of a ‘virtual’ performer can enhance the affect and engagement of its audience. The suggested refinements directly address several unresolved issues raised by the artist Stelarc, and the wider performance team comprising Stelarc, Associate Professor Garth Paine and myself during installation of this exhibit in the Powerhouse Museum, as well as those issues identified by analysis of the research participant Video Cued Recall Interviews and associated data.