Multi-Paths Quest Generation For Structural Analysis

Abstract

Existing quest generation systems that used quest structures had an important limitation. They were not compatible with dynamic environment games, where player actions may have unforeseen repercussion from chain-reaction, such as locking the quest or preventing it from being completed. In addition, quests generated by such systems were not guaranteed to have multiple completion paths.

In this thesis, action-based-objective quests were replaced with game state-based-objective quests, while retaining the previous grammar rule. Each quest’s contents and objectives were filled using Token system and Full Condition State system to ensure the connectivity within the quest’s contents and to prevent players from locking the quest. The Quest Query system simulated a dynamic environment game. It analysed every combination of player’s possible action to discover how many paths the player could complete the quest.

The state-based objectives were compatible with dynamic environment simulation. The query system discovered paths that used multiple steps of planning, including using the dynamic environment to complete the quest. The proposed systems functioned as intended and could be used as a framework for further studies. However, the simulation only used a small number of objects because of performance issue. Further study is advised to focus on improving the performance of the systems and applying the systems and its simulator to commercial products.