Resource-Bounded KARO

Abstract

Usually, a logic is not concerned with computing power. This is a problem for logic-based agents that need to react fast to their environment. Especially in complex situations and/or when the agent has little computing power. The agent must also take into account the time it takes to complete an action. If the agent needs to change its environment fast it is not useful to react with an action that takes a long time to complete. Thus there is a constraint on the time a reaction takes to complete and the time it takes to find a suitable reaction.

In the KARO framework agents can plan a sequence of actions that will fulfil the goals of the agent. The KARO framework is extended to take into account the two types of time constraints. A goal of the agent must be fulfilled before a certain point in time. The resource-bounded KARO framework doesn't allow action sequences that finish after the end time of a goal. It is also not allowed to reason about the right action sequence for so long that the found action sequence cannot be finished in time. The KARO framework is also extended with parallel actions. A goal can be reached faster if the actions to reach the goal can be performed in parallel.