Unreal Engine을 활용한 피난 영향인자별 인간행동 데이터 수집 및 저장 자동화 플랫폼 개발 연구

국문 초록 (Abstract)

급격한 도시화로 인해 초고층 건축물 및 대형 건축물이 증가하면서 건축물의 대형화 및 구조적 복잡성이 심화되고 있다. 이러한 건축 환경의 변화는 화재 발생 시 피난자들의 신속하고 효율적인 대피를 어렵게 하여 생존 가능성을 저해하는 주요 요인으로 발생하고 있다. 특히 복잡한 공간 구조, 다수의 피난 경로, 시야 제한 등의 요인은 피난자의 의사결정과 이동 행동에 영향을 미치며, 이는 전체 피난 효율성을 크게 저하되는 요인이 된다. 이러한 문제를 해결하기 위해서는 다양한 피난 상황에서 피난자의 행동 특성과 의사결정 과정을 정밀하게 분석하여 복합적으로 발생하는 피난 상황에 대응할 수 있는 효율적인 대피 전략을 수립하는 것이 필수적이다. 기존 연구에서는 피난자의 행동을 분석하기 위해 실험적 접근을 기반으로 피난자의 인간행동 데이터를 수집하였다. 기존 실험 연구 대부분은 제한적인 실험 환경과 변수 통제로 인한 단순화된 행동 및의사결정 데이터를 제한적으로 수집하고 개별적으로 수집하여 실제 재난 상황의 복잡성을 반영하지 못하는 한계가 존재한다. 이러한 한계를 극복 하기 위해 본 연구에서는 가상현실(Virtual Reality, VR)을 활용하여 대피 과정에서 나타나는 인간행동을 체계적으로 분석하고 피난 행동에 영향을 미치는 행동 패턴을 세분화하여 정량적인 데이터를 수집할 필요가 있다. 본 연구는 피난 행동에 영향을 주는 요인별 인간행동 데이터를 수집하고 저장할 수 있는 플랫폼을 개발하였다. 개발한 플랫폼은 첫째, 현실적이면서 구축 및 수정에 용이한 가상환경을 구현하기 위해 Lidar Scanner를 통해 수집한 점군(Point Cloud) 데이터를 활용하여 현실적인 가상환경을 Unreal Engine에 구현하였다. 둘째, 대피 행동에 영향 미치는 인간행동을 정량적으로 수집 및 저장하기 위해 인간행동을 체계적으로 분류하여 인간행동 요인 분류체계를 개발하였다. 셋째, 인간행동 데이 터를 자동으로 분류하여 인간행동 요인별 수집 및 저장하는 기능을 개발 하였다. 마지막으로 본 연구에서 제안된 플랫폼의 유효성을 검증하기 위해 가상환경 실험을 구축하고 이를 기반으로 검증 실험을 수행하였다.

다국어 초록 (Multilingual Abstract)

Rapid urbanization has led to a substantial increase in super high-rise and large-scale buildings, intensifying the scale and structural complexity of modern built environments. These architectural transformations significantly hinder rapid and efficient evacuation of occupants during fire emergencies, thereby reducing evacuees’ likelihood of survival. In particular, complex spatial configurations, multiple, intersecting evacuation routes and visibility constraints significantly influence evacuees’ decision-making processes and movement behavior, ultimately diminishing overall evacuation efficiency. Addressing these challenges requires a precise analysis of evacuees’ behavioral responses and decision-making mechanisms under diverse evacuation scenarios, enabling the establishment of effective evacuation strategies capable of responding to the multifaceted conditions of real fire situations.
Previous research has attempted to analyze evacuee behavior through experimental approaches that collect human behavioral data. Most previous experimental studies collected simplified behavioral and decision-making data in a limited manner due to constrained experimental environments and strict variable controls, and because these data were gathered individually rather than in an integrated manner, they were unable to reflect the complexity of actual disaster situations. As a result, these studies have struggled to capture the dynamic complexity of actual fire evacuation scenarios, as data were often simplified or examined in isolation. To overcome these limitations, the present study employs Virtual Reality (VR) to analyze evacuation behavior in a systematic manner and to collect detailed quantitative data by segmenting evacuation-related behavioral patterns based on influencing factors.
This study developed an integrated data-collection platform capable of acquiring and storing human behavior data based on factors influencing evacuation dynamics. First, a realistic yet easily modifiable virtual environment was created by acquiring point-cloud data using a Lidar scanner, enabling the construction of an authentic VR-based evacuation simulation environment.
Second, a hierarchical behavioral factor classification framework was developed to systematically categorize evacuation-related behaviors and support the structured and quantitative collection of human behavior data. Third, an automated system was implemented to classify, collect, and store human behavior data according to the defined behavioral factors. Finally, a VR-based experimental environment was established, and validation experiments were conducted to evaluate the effectiveness and applicability of the proposed platform.

목차 (Table of Contents)

• 제1장 서 론 ············································································································ 1
• 제1절 연구의 배경 및 목적 ············································································ 1
• 1. 연구의 배경 ······························································································ 1
• 2. 연구의 목적 ······························································································ 4
• 제2절 연구의 범위 및 방법 ············································································ 4
• 1. 연구의 범위 ······························································································ 5
• 2. 연구의 방법 ······························································································ 5
• 제2장 연구에 관한 이론적 배경 ········································································· 8
• 제1절 기존 피난 실험 연구 방법 ··································································· 8
• 1. 현장실험 ···································································································· 8
• 2. 세트장 실험 ···························································································· 10
• 3. 가상환경 실험 ························································································ 11
• 제2절 인간행동 데이터 수집 연구 ······························································ 13
• 1. 개인적 행동 요인 ·················································································· 13
• 2. 환경적 정보 요인 ·················································································· 14
• 3. 심리적 상태 요인 ·················································································· 15
• 제3절 플랫폼 개발에 관한 선행 연구 ························································· 17
• 1. 가상환경 기반 인간행동 데이터 수집 플랫폼 연구 ························ 17
• 2. 사물 인터넷 기반 실시간 행동 데이터 수집 플랫폼 연구 ············ 18
• 제3장 Unreal Engine을 활용한 피난 영향인자별 인간행동 데이터 수집및 저장 자동화 플랫폼 개발 ············································································ 19
• 제1절 연구 방법 ······························································································ 19
• 제2절 가상환경 구축 ······················································································ 20
• 1. 가상환경 구축 방법 ··············································································· 21
• 제3절 인간행동 요인 분류체계 개발 ··························································· 24
• 1. 개인적 행동 요인 ·················································································· 24
• 2. 환경적 정보 요인 ·················································································· 27
• 3. 심리적 상태 요인 ·················································································· 28
• 제4절 인간행동 데이터 수집 및 저장 기능 개발 ····································· 31
• 제4장 플랫폼 검증 ······························································································ 34
• 제1절 장비 및 소프트웨어 ············································································ 34
• 1. 장비 ·········································································································· 34
• 2. 소프트웨어 ······························································································ 35
• 제2절 가상환경실험 구축 및 실험 ······························································ 35
• 1. 가상환경실험 구축 ················································································· 35
• 2. 피험자 값 설정 ······················································································ 41
• 3. 실험 데이터 수집 및 저장 ··································································· 41
• 제3절 실험 ······································································································· 42
• 제4절 플랫폼 검증 결과 ················································································ 42
• 제5장 결 론 ········································································································· 48
• 참 고 문 헌 ···································································································· 51
• ABSTRACT ·········································································································· 65