Coordinated execution and goal management in a reactive cognitive architecture
Abstract/Contents
- Abstract
- Coordinated execution and goal management are important human-level capabilities studied in cognitive science and psychology. Evidence suggests there are several di erent types of constraints that govern the coordinated execution of multiple, concurrent tasks. We took three of these types - shared objects, logical relations, and resource requirements - and developed an extension to a cognitive architecture, Icarus, that supports coordinated execution under such constraints. The extended architecture assumes that it should consider subgoals in its procedures concurrently unless the constraints prevent such execution. The existing framework supports constraints on shared objects and logical relations, but it requires a new facility to handle resource constraints. We specify resource requirements for each action that Icarus can use in the world, and the system assigns resources to executable skill paths during the skill retrieval process. The architecture prevents the execution of any skill path that requires already assigned resources and ensures that execution is coordinated according to the resource constraints. We are also interested in the ability to manage goals that arise through changes in the environment. Such changes can lead people to pursue a new goal, abandon an old goal, and revise priorities among existing goals. This ability is especially important when executing multiple tasks at the same time, because concurrently pursued goals can interact with each other in both positive and negative ways. An agent should be able to give priority to more important ones if other goals are known to interfere. In response, we made a second extension to the Icarus architecture that lets the system manage its top-level goals in a reactive manner. The revised Icarus distinguishes long-term and short-term goals, and it nominates instantiated versions of long-term goals when the state matches their corresponding relevance conditions. During the nomination process, the architecture computes the degree of match for the relevance conditions and uses it to modulate priority values of nominated goals. Icarus revises the priority of nominated goals based on the modulated values. To evaluate these extensions, we chose a variety of scenarios in an urban driving domain to demonstrate the new abilities. Due to the high-level nature of our research on cognitive architectures, the demonstrations focus on ability and parsimony of the extended Icarus framework. We have also examined connections to earlier work on these topics and explored future directions for research.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2010 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Choi, Dongkyu |
|---|---|
| Associated with | Stanford University, Department of Aeronautics and Astronautics |
| Primary advisor | Rock, Stephen M |
| Thesis advisor | Rock, Stephen M |
| Thesis advisor | Lall, Sanjay |
| Thesis advisor | Langley, Pat |
| Advisor | Lall, Sanjay |
| Advisor | Langley, Pat |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Dong Kyu Choi. |
|---|---|
| Note | Submitted to the Department of Aeronautics and Astronautics. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2010. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2010 by Dong Kyu Choi
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...