Conceptualization and Implementation of a Reinforcement Learning Approach Using a Case-Based Reasoning Agent in a FPS Scenario

Marcel Kolbe, Pascal Reuß, Jakob Michael Schönborn, Klaus-Dieter Althoff

In: Robert Jäschke, Matthias Weidlich (Hrsg.). Lernen, Wissen, Daten, Analysen. GI-Workshop-Tage "Lernen, Wissen, Daten, Analysen" (LWDA-2019) September 30-October 2 Berlin Germany Seiten 280-291 CEUR 2019.


This paper describes an approach that combines case-based reasoning (CBR) and reinforcement learning (RL) in the context of a first person shooter (FPS) game in the game mode deathmatch. Based on an engine written in C#, Unity and a simple rule-based agent, we propose a FPS agent who is using a combination of case-based reasoning and reinforcement learning to improve the overall performance. The reward function is based on learned sequences of performed small plans and considers the current win chance in a given situation. We describe the implementation of the reinforcement algorithm and the performed evaluation using different starting case bases.

Weitere Links

paper_48.pdf (pdf, 1 MB)

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence