Research on AI-based Aim Accuracy Improvement Support for FPS Game Players
In this study, we aim to establish a mechanism whereby AI generates appropriate feedback by comprehensively analyzing FPS game play footage and controller input data. We automatically extract information such as damage, number of bullets, and effects from game screen analysis, and combine it with device operation data to analyze the relationship between player actions and results. Additionally, we will utilize LLM to present personalized feedback in text form tailored to individual challenges based on the analysis data. We will verify changes in feedback quality due to different data combinations and the optimal presentation format, and identify features perceived as “beneficial” through evaluations by multiple users. Based on these research findings, we plan to release a Thumaster alpha version available for general users and collect empirical usage data.
