An experiment in automatic content generation for platform games

摘要

Computer platform games are an innovative measure to improve the strategic abilities of the player. However due to the repetitive nature of game levels, the players lack interest. Procedural Content Generation (PCG) is a technique to overcome such issues. PCG generates endless and adaptive levels but most of the PCG techniques have catastrophic failures that make the game unplayable. The focus of our work is to produce reliable levels for platform games. For this purpose, a well known game Mario was selected. In our previous effort, we identified some catastrophic failures in offline level generation for Mario and now we present a novel technique that is catastrophic failure free and generates adaptive and reliable levels. This content generation process will insure dynamic difficulty adjustment (DDA) in games as players start with different skill levels and games become unexcitingly easy for some players and disturbingly difficult for others.