- Title
- Evaluating a virtual human storyteller for improved decision support
- Creator
- Blackmore, K. L.; Smith, S. P.; Nesbitt, K. V.; North, L.; Wark, S.; Nowina-Krowicki, M.
- Relation
- 23rd International Congress on Modelling and Simulation (MODSIM2019). Proceedings of 23rd International Congress on Modelling and Simulation (MODSIM2019) (Canberra, Australia 01-06 December, 2019) p. 207-213
- Publisher Link
- http://dx.doi.org/10.36334/modsim.2019.B2.blackmore
- Publisher
- Modelling and Simulation Society of Australia and New Zealand Inc
- Resource Type
- conference paper
- Date
- 2019
- Description
- Defence and security organisations face increasing challenges to maximise the real-time integration of data and information for improved decision making. This is particularly true in applications of human-machine teaming. These teams create complex interaction environments for human operators that integrate autonomous systems, multi-modal displays, and information streams at differing temporal scales. Examples of these complex environments can be seen in simulation pilots operating in large scale synthetic training scenarios, and operators of multi-unmanned aerial vehicle (multi-UAV) systems. In both these cases, an individual must perform a primary task (directing simulated or real entity behaviour), whilst simultaneously following a pre-defined scenario plan. On top of this, emergent information, with varying levels of uncertainty, must be acted on or ignored as appropriate. Additional contextual information may also be required such that these actions can occur with an understanding of broader contextual factors. The efficiency with which this primary, planned, emergent, and contextual information is presented underpins decision support and superiority in increasingly complex military contexts. In this research, we evaluate the use of storytelling via a virtual human to deliver complex geo-spatial conflict briefing information. The role of virtual humans in information delivery is an area of active research. At the intersection of education, psychology, and computing, the development of effective virtual humans is a difficult task. Many factors contribute to their effectiveness, including cognitive loading associated with different forms of multimedia, as well as the design and features of the virtual humans themselves. Further complicating this is the inherent learning abilities and inclinations of individuals, and their underlying preferences for different learning and information delivery approaches. In this research we conducted an initial evaluation of the delivery of narrative content using the Defence Science and Technology (DST) Virtual Human Storytelling system (Figure 1). An experimental methodology was adopted, using a between-subjects design. A total of 87 participants viewed a 3-minute multimedia video on a hypothetical conflict scenario in a fictious location that included details of actors in the conflict, actions occurring between these actors, and the geographic location. Three experimental conditions were used: a control group with the scenario delivered using text and images; a virtual human condition, with the scenario delivered using an animated avatar, synthesised speech, and images; and an audio condition, with the scenario delivered using synthesised speech and images. Preliminary analysis of the results suggests differences in performance on the information recall task, with participants in the virtual human group recording a greater range of, and higher maximum, scores. While simple between group comparisons of mean scores were not statistically significant, the higher maximum recall scores for participants in the virtual human group are promising. Directions for future work are also outlined, including incorporation of a measure of real-time cognitive load to isolate learning content for optimum delivery of information at a more granular level.
- Subject
- decision support; virtual humans; simulation; cognitive load; storytelling
- Identifier
- http://hdl.handle.net/1959.13/1460078
- Identifier
- uon:45853
- Identifier
- ISBN:9780975840092
- Language
- eng
- Reviewed
- Hits: 1392
- Visitors: 1387
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|