<![CDATA[Background: Non-Playable Characters (NPCs) play an important role in the quality of interaction in Virtual Reality (VR) games. As NPC systems become more complex and multi-role, their behavior must be evaluated not only from a design perspective but also in terms of technical stability and interaction consistency. However, systematic methods for validating NPC behavior in VR games remain limited. Objective: This study proposes and implements a scenario-driven black-box testing framework for evaluating NPC behavior in VR games based on observable interaction outcomes. Methods: NPC behavior was evaluated using normal, boundary, and stress-based interaction scenarios. Four behavioral quality criteria were assessed: determinism, state integrity, role consistency, and recovery stability. Selected dimensions of the Game Experience Questionnaire (GEQ), namely Social Behaviour Activity and Post-game Module, were also evaluated. Results: Six test scenarios were evaluated. Four scenarios (66.7%) met all criteria, while two (33.3%) were categorized as partial due to minor delays in state transitions and temporary idle-state resets. No crashes, system failures, or fatal interaction breakdowns were observed. GEQ results indicated moderate-to-good social behavior involvement (mean = 3.3) and a comfortable post-game experience (mean = 3.5). Conclusion: The proposed framework provides a structured and reproducible approach for assessing NPC behavior in VR games without requiring access to internal implementation details. The findings highlight the importance of technical behavioral testing, as positive player experience does not necessarily indicate fully stable and consistent NPC behavior.]]>
Copyrights © 2026
