Giving Santos™ Intelligence: Artificial Intelligence (AI)
The objective is for Santos™ to develop and implement necessary intelligence in terms of cognitive behavior for specific tasks performed by soldiers in the field. A digital human will then be able to function realistically in a virtual environment, and as a result evaluate designs and tasks virtual prototypes.
We use research in cognitive science, particularly that of cognitive task analysis, schema theory, and the student modeling of intelligent tutoring systems, to design a cognitive engine for our digital human models. The targeted population is based on typical (modal) levels of assistance actually provided by soldiers in the fi eld while performing a specifi c task (e.g., performance maintenance of an FCS (Future Combat System) vehicle).
Another goal is to develop a model of this cognitive engine as an intelligent agent where cognition is cognition is conceptualized as a property of the activities of the system. In this perspective, cognition is seen as distributed, or spread over, the person (digital human), and situation in which the activity is taking place. This is in contrast to more traditional informational processing models in which cognition is conceived of as a property of an individual, who is impacted by properties of the world.
Virtual humans will require the ability to learn, reason, and coordinate task plans with increasing autonomy in order to simulate complex multi-agent tasks. Without being endowed with these artifi cial brains, virtual humans will continue to require mainly manual manipulation within their artifi cial immersive environments. Research in multi-agent planning and coordination has usually taken place in the context of software agents or physical robotic agents. Our work with virtual humans will include investigations into how virtual human agents can learn tasks and coordinate these tasks by communicating and merging local plans distributed among them.
