RALPH: Utility Function Available
Description
- Decision theory requires that the utility of a state be calculable, which then gives the value of the action which achieved that state.
- The utility of a state is a measure of the desirability of that state based on the values of the state variables, which include the condition of the environment, current goals, and current knowledge of the agent.
- The use of utilities provides RALPH with utility-driven behavior.
- Several of RALPH's execution architectures make explicit use of the utility value of a state and the others make indirect use of such information. For example, the decision-theoretic execution architecture determines which action to take based on the utility of all possible next states, which is a direct use of the utility function. The condition-action architecture, on the other hand, compiles away any explicit references to the utility function in order to increase speed of operation, but the rule is compiled from decisions made based on the utility function.
Detail from RALPH