JSAI2024

Presentation information

Organized Session

Organized Session » OS-31

[1I4-OS-31a] OS-31

Tue. May 28, 2024 3:00 PM - 4:40 PM Room I (Room 41)

オーガナイザ:三宅 陽一郎(株式会社スクウェア・エニックス)、森川 幸人(モリカトロン株式会社)

4:00 PM - 4:20 PM

[1I4-OS-31a-03] System Development of a Looping 2D Adventure Game Combining Automatic Generation Technologies Utilizing AI

○Hiroto Oyanagi1, Ryoma Tanaka1, Masaki Mori1, Keiryu Nakamura1, Hiromi Muraoka1, Konan Matsushita1, Yuta Yamasaki1, Takaki Sekiya1, Ren Sato1, Kuu Tokinaga1, Yutaro Nishiguchi1, Haruhiko Ozaki1, Hajime Suzuki1, Haruta Nagase1, Yamamoto Takuma1, Kazushi Mukaiyama1, Emiko Tayanagi1, Keiji Hirata 1, Hironori Yoshida 1,Hajime Murai1 (1. Future University Hakodate)

Keywords:AI, game

Automatic generation by AI has the great advantage of obtaining a large amount of similar output at low cost. In addition, works with looping structures, which have become popular in recent years, are characterized by the repetition of similar developments. In this study, we attempted to develop a looping 2D adventure game system by combining several techniques of automatic AI generation in order to develop a game system that makes the best use of AI that can generate a large number of similar outputs. Specifically, we developed a system that integrates automatic generation of map design, scenarios, sound, and character animation with reinforcement learning to change the behavior patterns of enemies. By combining these elements, we constructed a system for a looping 2D adventure game. As a result, for example, in the automatic generation of scenarios, efficiency was achieved by producing a large number of plots for each loop by generating outputs that have similar developments but differ in details. We have shown that it is possible to produce games by integrating the results of generation using the above multiple AI automatic generation technologies.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.

Password