Behavior of RALPH-MEA

Behavior of RALPH-MEA

The various execution architectures produce differing behaviors. The Decision-Theoretic EA, if given enough response time, will produce completely rational behavior by maximizing expected utility. However, under stricter time constraints, the Condition-Action EA can produce highly reactive behavior. The use of utility functions in the influence diagrams of the execution architectures insures that tasks are salient to the current environmental situation by the encoding of event priorities.

The designers discuss an Autonomous Underwater Vehicle implementation, aimed at surveying and obstacle avoidance. The former usually involves planning tasks, while the latter leads to invocation of the replanning architecture. Whether proper behavior can be maintained when more complex tasks are required remains an open issue.

The agent's behavior depends greatly on the choice of utility functions. Since the execution architectures make local decisions, it is possible that in the long run, goals will remain unsatisfied (this myopia is mentioned by the authors). For instance, in the current state, a certain action is preferable since it advances toward goal A, although it moves away from goal B. In the resulting state, perhaps goal B actions have higher utility, and reverse the previous results. In addition, there is also the problem of local minima. Of course, correct choice of utility function can correct both problems, but sometimes the correct choice is unknown.


Return to the top of this architecture.

Go to a discussion of this property for multiple architectures.


Current Location: RALPH-MEA - Properties - Behavior

Go to NEXT page.