BROWSE
Detail
Conference Paper
인간 행동 경향을 반영한 식사 후 테이블 장면 생성 방법
Cited - time in 
Share 
Share
- Authors
- 정희원, 유영재
- Issue Date
- 2026-02
- Citation
- 한국로봇학회 종합 학술 대회 2026, pp.1056-1059
- Publisher
- 한국로봇학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- Recent advancements in service robots have accelerated research on various household tasks. Among these, post-meal bussing is a complex task involving diverse objects and actions such as manipulation, navigation, and procedural reasoning. However, existing tabletop generation methods in simulation environments often rely on random object placement, which fails to capture the physical consistency and behavioral patterns observed in real post-meal settings. To address this, we propose PostMealGen (Generating Human-Like Post-Meal Table Scenes through Behavioral Awareness), a human-like method that reflects behavioral tendencies by integrating social regularity and individual variability to generate post-meal table scenes. The proposed method was quantitatively evaluated in terms of placement accuracy and behavioral tendency, showing significant improvement over random generation. Specifically, PostMealGen achieved a 60 %p higher location accuracy and a 28 %p improvement in behavioral tendency, demonstrating its ability to generate human-like, physically consistent table layouts.
- Keyword
- Human-Like Scene Generation, Embodied Simulation, Behavioral Awareness
- KSP Keywords
- Behavioral Patterns, Embodied simulation, Human-like, Random object, Scene Generation, Service robots, complex task, location accuracy