RAIL: A Domain-Specific Language for Generating NPC Behaviors in Action/Adventure Game

Abstract

Domain-Specific Modeling (DSM) has shown its effectiveness of improving software productivity in many software domains [1], where Domain Specific Language (DSL) plays a key role. Also in the domain of video games, researchers have proposed various DSLs for developing different aspects of several game genres. This paper presents a DSL named RAIL for generating Non-Playable Character (NPC) behaviors in Action/Adventure Games. Our DSL borrows concepts from State Machines and adds some features to better suit the target domain. Further, we have implemented a tool-chain for RAIL using the Eclipse language workbench, and the tool-chain has been integrated with the level editor of the Torque2D game engine. To evaluate the DSL, we developed a prototype game and collected data regarding the development time and code lines. The results showed that RAIL significantly improves the productivity of developing NPC behaviors in the target game with a reasonable associated cost. In addition, the integration of the RAIL and the Torque 2D tool-chains provides a smooth development workflow.