You are here
The industrial objective of this research project, called PRESTO (Plausible Representation of Emergency Scenarios for Training Operations), is the creation of a system for the study of contexts
operating based on virtual reality.
The advantage of this system compared to its competitors will reside in the richness and the ease of defining the behavior of artificial characters in simulated scenarios.
It will be possible to specify procedures, psychological profiles, and other factors that influence the behavior of individuals and/or groups in any role (emergency teams, victims, observers, terrorists, criminals, etc.) and to build scenarios, for instance a car accident, in which part or all of the people involved are simulated by artificial characters.
To this end, the system will include an environment for building the training scenarios by the domain experts, tools for the specification of cognitive and perceptual models used for augmenting profiles psychological of non-player characters, and execution engines cognitive behaviors, actions and perceptions within a virtual reality environment.
The system can be used, for example, for training safety personnel, for the verification and the optimization of operational procedures, and for the analysis of work environments.
The system will be tested in a pilot project selected in a specific application domain of large interest in both commercial and research fields: training for emergency management within close environments (such as fires, evacuations, overload of users due to external factors such great disasters scale, etc.).
The pilot will be conducted in collaboration with the Health Services of the Trentino local government (APSS).
The open problems addressed by this project may be summarized as follows:
• the perception of the virtual environment by an artificial character and the execution of its models and procedures must be able to adapt to the context, to its history and status (fatigue, emotions, intake of stimulants such as caffeine or depressants such as alcohol) and must maintain a level of variability (i.e. in the accuracy of the vision, the rate of reaction, in the choices among alternatives) such that the behavior plausible but not trivially predictable;
• the representation of procedures and patterns of behavior must be independent of one specific usage scenario and accessible to domain experts (i.e. industrial safety or civil protection) rather than just a computer, in an environment that facilitates the definition and configuration of training scenarios by such specialists.