The Hierarchical Agent Control Architecture (HAC) is a general toolkit for specifying an agent's behavior. By organizing the hierarchy around tasks to be accomplished, not the agents themselves, it is easy to incorporate multi-agent actions and planning into the architecture. In addition, HAC supports action abstraction, resource management, sensor integration, and is well suited to controlling large numbers of agents in dynamic environments. Unlike other agent architectures, HAC does not conceptually distinguish reactive from deliberative, or single-agent from multi-agent behaviors. There is no pre-determined number of cognitive "levels" in the hierarchy - all actions share the same form and are implemented with the same functions. GRASP is a multi-goal partial hierarchical planner that has been implemented using the HAC framework. GRASP illustrates two points: Firstly, that the same HAC mechanisms used to write reactive actions can be used to implement a cognitive activity such as planning; and secondly, that the problem of integrating reactive and deliberative behavior itself can be viewed as having to simultaneously achieve multiple goals. Throughout the paper, we show how HAC and GRASP were applied to an adversarial, real-time domain based on the game of "Capture the Flag".
|Original language||English (US)|
|Number of pages||16|
|Journal||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|State||Published - Jan 1 2001|
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)