IoT 기반 실시간 교통사고 감지 시스템의 설계 및 구현

국문 초록 (Abstract)

최근 5년간 도로에서 발생한 교통사고 중 차량 단독사고의 치사율은 전체 사고 대비 4.7배 높게 집계되며, 차량 단독으로 발생하는 사고를 실시간으로 감지하고 대처할 수 있는 시스템 구축이 필요하다. 본 연구는 도로의 가드레일에 차량 충돌로 발생하는 충격과 차량의 진입을 감지하는 IoT(Internet of Thing) 센서를 부착해 가드레일의 충격 감지 시 해당 지점의 영상을 인공지능 기반으로 분석하여 구조작업 등 빠른 후속조치를 통해 교통사고로 발생하는 피해를 최소화 시킬 수 있는 시스템을 제안한다.
교통사고 감지에 있어 IoT 센서나 CCTV가 단독으로 교통사고 감지 시 발생할 수 있는 오인식의 문제를 개선하고 더욱 신뢰도 높은 상황 감지를 위해 IoT와 인공지능 기술을 융합하여 활용하였다.
시스템이 설치된 해당 가드레일 구간 내 차량의 진입과 차량 충돌로 발생하는 충격을 감지하는 IoT 센서와 다양한 차량의 데이터 학습을 기반으로 차량과 사람을 참지하는 객체 탐지 모듈을 구현한다. 센서를 통해 수집되는 정보와 객체인식 정보를 통합적으로 관리하는 모니터링 시스템을 구현한다.
구현된 시스템의 성능과 유효성을 검증하기 위해 실제 도로에 현장적용을 한 후 차량진입과 충격감지 센서 상황을 실시간으로 확인할 수 있는 모니터링 시스템 상의 상황기록 확인과 사람과 자동차 등 객체탐지 모듈의 객체탐지 정확도를 검증하였다. 또한 센서의 충격감지 시 LED 전광판의 정보표출이 변화되는 기능과 영상 데이터의 수집 및 삭제 등에 관련된 기능을 검증하였으며, 모두 만족하는 결과를 도출하였다. 향후에는 해당 기능들을 더욱 고도화할 수 있는 연구를 지속하여 상용화를 추진할 계획이며, 본 시스템은 본 시스템은 도로 안전 향상에 최소한의 역할을 하는 방안으로 활용성이 높을 것으로 기대된다.

다국어 초록 (Multilingual Abstract)

Among the traffic accidents that occurred on the road in the last five years, the fatality rate of single-vehicle accidents is 4.7 times higher than all accidents, and it is necessary to establish a system that can detect and respond to accidents that occur alone in real-time. In this study, an IoT sensor detects the impact caused by a vehicle collision, and the vehicle's entry is attached to a guardrail on the road.
In detecting traffic accidents, IoT sensors or CCTVs are used to improve the problem of misrecognition that can occur when traffic accidents are detected independently. IoT and artificial intelligence technologies are converged and utilized for more reliable situation detection.
When the impact of a guardrail is detected, we propose a system that can minimize the damage caused by a traffic accident through quick follow-up measures such as rescue work by analyzing the video of the corresponding point based on artificial intelligence.
Implement an object detection module that detects vehicles and people based on data learning from various vehicles and IoT sensors that detect impacts caused by vehicle entry and vehicle collision within the guardrail section where the system is installed. Implement a monitoring system that manages information collected through sensors and object recognition information in an integrated manner.
The implemented system was applied on-site to actual roads to confirm its effectiveness. Vehicle entry and shock sensor conditions could be checked in real-time, and objects such as people and cars were accurately detected. In addition, the function of changing the information display of the LED signboard when the sensor detects an impact and the function related to the collection and deletion of image data were verified. This system is expected to contribute to improving road safety.

목차 (Table of Contents)

• 제 1 장 서 론 ················································································1
• 제 2 장 관련연구 ·········································································3
• 제 1 절 국내 도로교통 관련 연구 동향 및 문제 인식 ······· 3
• 1. 차세대 지능형 교통 시스템(C-ITS) 연구 동향 ·········· 3
• 2. 자율주행차 관련 연구 동향 ··············································5
• 3. 국내 연구 동향과 문제 인식 ············································7
• 제 2 절 IoT(사물인터넷) 관련 국내외 선행 연구 ················8
• 1. IoT(사물인터넷)의 개념 및 정의 ····································8
• 2. IoT 기술을 활용한 국내 선행 연구 ·····························10
• 3. IoT 기술을 활용한 해외 선행 연구 ·····························11
• 제 3 절 인공지능 관련 선행 연구 ·········································13
• 1. 인공지능 정의 및 활용 ····················································13
• 2. 인공지능 기술을 활용한 국내 선행연구 ······················15
• 3. 인공지능 기술을 활용한 해외 선행연구 ······················16
• 제 3 장 IoT 기반 실시간 교통사고 감지 시스템 설계 및 구현···· 18
• 제 1 절 시스템 구성 ····························································· 18
• 1. 시스템 기능 구성 ······························································18
• 2. 시스템 운영 프로세스 ····················································19
• 제 2 절 IoT 기반 센싱 모듈 ·················································20
• 1. IoT 기반 센싱 모듈 설계 및 구현 ···································20
• 2. 센서 데이터 통신 설계 및 구현 ····································22
• 제 3 절 인공지능 기반 객체 탐지 알고리즘 ·····················24
• 1. 인공지능 학습 데이터 구성 ············································25
• 2. Yolo v5 기반 객체탐지 알고리즘 구현 ·····················27
• 제 4 절 모니터링 시스템 ·······················································31
• 1. 모니터링 시스템 설계 및 구현 ······································31
• 제 5 절 운용지원 모듈 ·····························································34
• 1. 통합제어 장치 및 태양광 운영 모듈 설계 및 구현 ·· 34
• 2. 후방차량 사고 안내 LED 전광판 모듈 설계 및 구현 ····34
• 제 4 장 현장 실증 및 검증 ····················································35
• 제 1 절 현장 실증 ·································································35
• 1. 현장 실증 구역 선정 근거 ··············································35
• 2. 현장 설치 ··········································································36
• 제 2 절 검증 ·············································································36
• 1. 기능검증 ··············································································36
• 제 5 장 결론 ···············································································40
• 참고문헌 ·························································································42
• Abstract ·························································································45