Technical Report

Sacagawea: a testbed for experimenting with real-time problem solving techniques

Public Deposited

Downloadable Content

Download PDF


Attribute NameValues
  • The Sacagawea testbed has been designed to support testing and devel­opment of decision making models, and associated uncertainty manage­ment techniques, that are appropriate under real-time problem solving con­straints. Conceptually, it provides a simulated temporal and spatial environment, the latter consisting of a terrain grid of elevation and tree cover data, plus associated costs of travel, in which mobile autonomous agents operate. Intelligent real-time problem solving is a poorly defined and poorly un­derstood field of artificial intelligence which addresses the control of system behavior in dynamic and resource-limited situations. Its basic character­istics, applicability and limitations are open to conjecture and pioneering investigation. Most proposed methodologies are theoretical, and yet to be implemented. The Sacagawea testbed is intended to facilitate the testing of these methodologies by creating environments that will challenge the capa­bilities of decision making architectures designed to contend with relatively unstructured problems. Initial experiments with the testbed will focus on engineering reasoning systems that will allow agents to move from a starting location to a preset destination as rapidly as possible. Through movement and sensing actions, agents will accumulate a "mental" image of the terrain that will influence their choice of travel routes. An agent's initial knowledge of the terrain will be limited to a line-of-sight view from the starting position. Unpredicted encounters with other types of agents of a hostile or friendly nature may influence agent behavior and alter an agent's plan for reaching its objective.
Resource Type
Date Issued
Academic Affiliation
Rights Statement
Peer Reviewed



This work has no parents.