The Glaive Narrative Planner
Glaive is a fast planning algorithm which leverages computational models of narrative to solve multi-agent coordination problems often encountered in stories. Each character has its own goals and should only act on those goals, but the group as a whole needs to be shepherded toward an author goal as if by an invisible puppetmaster. Each character should appear to act only on its own goals, cooperating and conflicting with other agents along the way to the author's desired outcome.
The Glaive planner:
- Is based on Hoffmann and Nebel's Fast Forward algorithm.
- Uses Riedl and Young's model of intentionality to reason about individual agent goals and cooperation.
- Uses Ware and Young's model of conflict to reason about how agents plans can interfere with one another and fail. By treating the problem's search space as a set of possible worlds, Glaive can find an explanation for the actions of a failed plan by reasoning about other ways the story might have played out.
- Pure Java 1.7 (no additional libraries required)
- Input and output based on PDDL 3
- Full ADL support (negative preconditions, typed objects, equality, conditional effects, first-order quantifiers)
- Domain axioms
- Can read in a partial plan which must appear as a subsequence in the solution
The current version of Glaive can be downloaded here:
- Executable JAR file
- Full Java 1.7 source as an Eclipse project archive
- README file
- 8 test problems and a batch file to run them
- Glaive: a state-space narrative planner supporting intentionality and conflict. In Proceedings of the 10th AAAI international conference on Artificial Intelligence and Interactive Digital Entertainment, pp. 80-86, 2014. (awarded Best Student Paper).
- The Intentional Fast-Forward narrative planner. In Proceedings of the 5th Intelligent Narrative Technologies Workshop at the 8th AAAI international conference on Artificial Intelligence and Interactive Digital Entertainment, pp. 57-62, 2012..