Towards Intelligent Automated Forces for Simnet
Semiannual rept. 1 Apr-30 Sep 91,
UNIVERSITY OF SOUTHERN CALIFORNIA MARINA DEL REY INFORMATION SCIENCES INST
Pagination or Media Count:
one of our long-term research goals is to construct Soar-based autonomous intelligent agents that can act as independent players in multi-agent simulation environments. The Soar architecture has been designed with the goal of being able to support general intelligence, and various systems built within Soar have demonstrated a range of planning, learning, language, and robotics behaviors. However, there is not yet a single agent that combines the architecture with sufficient knowledge to allow it to survive and behave appropriately in complex simulated environments. The focus of this contract is on performing seed research towards this end. The work is being pursued in the context of a simplified battlefield environment - called GridWorld and developed at the Hughes AI Center - which provides a two-dimensional, multiagent, realtime world in which both space and time are represented as continuous quantities.
- Computer Programming and Software