도심자율주행을 위한 도로정보모델기반 통합운전계획 시스템에 관한 연구

Abstract

최근 자동차 연구분야에 있어서 차량 안전도 및 탑승자 편의향상을 위해 자율주행 자동차관련 기술실현이 핵심 이슈로 자리잡고 있다. 이와 같은 목적을 달성하기 위한 자율주행 자동차는 주행환경을 인지하고 운전자의 개입 없이 스스로 주행할 수 있어야 한다. 이러한 자율주행 자동차의 개발에 있어서 위치추정, 환경인지, 운전계획, 차량제어 및 시스템 운영관련 등의 최신 기술들은 필수적인 요소이다. 이 최신 기술 중에서 운전계획은 자율주행 자동차의 안전한 주행을 위한 전반적인 거동 및 주행전략을 결정짓는 핵심이라 말할 수 있다. 이 논문은 도심주행환경에서 자율주행자동차의 안정적인 주행전략을 제공하기 위한 운전계획 알고리즘에 대해 제안한다. 제안한 알고리즘은 복잡한 도심도로환경을 분석하고 상황에 맞는 운전경로를 제공하기 위한 도로정보모델기반 통합운전계획 시스템으로써 총 3가지 서브 시스템(전역경로 계획시스템, 행동 계획시스템, 지역경로 계획시스템)으로 구성된다. 첫째로, 전역경로 계획시스템은 출발지에서 목적지까지의 경로를 검색할 뿐만 아니라 Open Street Map (OSM) 데이터 모델을 사용하여 구축한 지도 데이터베이스로부터 도로정보모델을 생성한다. 생성된 도로정보모델을 바탕으로 행동 계획시스템은 주행상황판단 및 규칙기반 결정 프로세스를 통해 자율주행차량의 전반적인 거동을 결정한다. 최종적으로 지역경로 계획시스템은 안정적인 자율주행을 위한 장애물 회피 경로를 탐색한다. 제안한 통합운전계획 시스템을 구현 및 검증하기 위해, A1 및 H-CAB의 두 자율주행 자동차를 개발하였다. 이 두 자율주행차량의 구현을 위해, 네트워크기반 분산시스템 플랫폼 및 AUTOSAR 방법론 기반 개발 프로세스가 적용되었다. 이를 통해 연구개발한 자율주행자동차 및 운전계획 알고리즘에 대한 평가를 다양한 실험을 통해 수행할 수 있었다. 결과적으로 제안한 통합운전계획 시스템이 자율주행차량의 최적경로를 검색할 뿐만 아니라 연산효율측면에 있어서도 우수한 성능을 보임을 확인할 수 있었다. |Realization of autonomous vehicle is one of key issues in automotive research areas for improving vehicle safety and passenger comfort. For achieving these objectives, autonomous vehicles should perceive road environments and drive by itself without human intervention. The development of autonomous vehicle requires the state-of-the-art technologies in localization, perception, planning, control, and system management. Among these core technologies, the planning is the key element which determines the behavior and motion of the autonomous vehicle for safe vehicle navigation. This dissertation presents the planning algorithms that provide reliable navigation capability to the autonomous vehicle under urban environments. In order to analyze the complex urban road situations and decide appropriate vehicle motion, a road information model-based integrated motion planning system is proposed in this dissertation. The proposed integrated motion planning system is composed of three subcomponents: global path planning system, behavior planning system, and local trajectory planning system. The global path planning system finds a global route from a start point to a destination, and it derives a road information model from a developed road map database using Open Street Map (OSM) data model. Based on the road information model, the behavior planning system determines overall maneuvers of the autonomous vehicle through situation assessment and rule-based decision process. Finally, the local trajectory planning system finds a collision-free and smooth trajectory for safe and reliable autonomous driving. In order to implement and validate the proposed integrated motion planning system, the two autonomous vehicles were developed: A1 and H-CAB. For the implementation of both autonomous vehicles, in-vehicle network (IVN)-based distributed system platform and AUTOSAR methodology-based development process were applied. Using the system platform and development process, intensive experimental studies were performed for evaluation of the developed autonomous vehicles and planning algorithms. From the experimental results, the proposed integrated motion planning system was validated to find an optimal motion of vehicle as well as to show excellent computational efficiency for urban autonomous driving.; Realization of autonomous vehicle is one of key issues in automotive research areas for improving vehicle safety and passenger comfort. For achieving these objectives, autonomous vehicles should perceive road environments and drive by itself without human intervention. The development of autonomous vehicle requires the state-of-the-art technologies in localization, perception, planning, control, and system management. Among these core technologies, the planning is the key element which determines the behavior and motion of the autonomous vehicle for safe vehicle navigation. This dissertation presents the planning algorithms that provide reliable navigation capability to the autonomous vehicle under urban environments. In order to analyze the complex urban road situations and decide appropriate vehicle motion, a road information model-based integrated motion planning system is proposed in this dissertation. The proposed integrated motion planning system is composed of three subcomponents: global path planning system, behavior planning system, and local trajectory planning system. The global path planning system finds a global route from a start point to a destination, and it derives a road information model from a developed road map database using Open Street Map (OSM) data model. Based on the road information model, the behavior planning system determines overall maneuvers of the autonomous vehicle through situation assessment and rule-based decision process. Finally, the local trajectory planning system finds a collision-free and smooth trajectory for safe and reliable autonomous driving. In order to implement and validate the proposed integrated motion planning system, the two autonomous vehicles were developed: A1 and H-CAB. For the implementation of both autonomous vehicles, in-vehicle network (IVN)-based distributed system platform and AUTOSAR methodology-based development process were applied. Using the system platform and development process, intensive experimental studies were performed for evaluation of the developed autonomous vehicles and planning algorithms. From the experimental results, the proposed integrated motion planning system was validated to find an optimal motion of vehicle as well as to show excellent computational efficiency for urban autonomous driving.