[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

NTRL: Encounter Generation via Reinforcement Learning for Dynamic Difficulty Adjustment in Dungeons and Dragons

Created by
  • Haebom

Author

Carlo Romeo, Andrew D. Bagdanov

Outline

In this paper, we propose a novel approach to automate the manual balancing task of Dungeon Masters (DMs) in Dungeons & Dragons (D&D), called 'Encounter Generation via Reinforcement Learning (NTRL).' NTRL generates encounters based on real-time party member attributes by framing the situational bandit problem. Compared to existing DM heuristics, it increases encounter intensity by increasing the duration of the fight (+200%), increasing the damage inflicted to party members, decreasing post-fight health loss (-16.67%), and increasing the number of player deaths (while keeping the total party wipe low). It maintains a high win rate (70%) while enhancing strategic depth and increasing difficulty to maintain game fairness, and outperforms encounters designed by human DMs.

Takeaways, Limitations

•
Takeaways:
◦
Provides an automated difficulty adjustment system to reduce the burden on the DM in D&D.
◦
Provides a more immersive gaming experience through increased battle duration and enhanced strategic depth.
◦
Validation of the utility of an automated difficulty adjustment system based on reinforcement learning.
◦
Demonstrating the practicality of automated systems through superior performance compared to human DM.
•
Limitations:
◦
NTRL's performance evaluation may be limited to specific D&D systems or game sessions.
◦
Further research is needed to account for different player types and combat styles.
◦
There may be limitations in perfectly reflecting complex game situations and rules.
◦
Additional validation of NTRL's performance and stability for long-term gameplay is needed.
👍
Made with Slashpage