The sight of a self-driving system in an autonomous ground vehicle (AGV), or the sensor system of an autonomous vehicle, is limited due to its low viewpoint. An autonomous unmanned aerial vehicle (UAV) flying within cooperative control with the AGV can provide much longer and wider bird-eye-view map on the diverse road conditions. By pairing an AGV and a UAV, the newly proposed system gains better access to blind spots rather than AGV itself. However, verification of such system in real world causes injuries from unavoidable accidents. Using modularized system architectures, we propose the novel CarSim and Unreal Engine simulator-based evaluation framework for evaluating path planning and tracking performance of an AGV-UAV cooperative system in a virtual environment. Furthermore, this system is under verification process by an indoor lab experiment. The performance evaluation of the AGV-UAV cooperative system in safety scenarios with blind spots are presented for demonstration. With the aid of UAV’s early detection of obstacles, the proposed AGV-UAV system convinces that not only the possibility of the collision is reduced, but also the autonomous vehicle maneuvering driving performance is substantially improved.|본 연구는 차량 단독 주행의 경우에 발생하는 사각지대에 대한 정보를 차량-드론 연계 시스템을 통해 제공한다. 지상에서 운용되는 자율주행 차량의 센서는 차량의 표면 위에 배치되어 인지할 수 있는 정보가 제한된다. 한편 자율주행 차량과 함께 비행하는 드론을 활용하면 더 넓은 시야를 이 용해 개선된 Top-view 맵을 제작할 수 있다. 시뮬레이션 및 실내 실험실 환경에서 차량-드론 연계 시스템을 구성하여 평가를 진행하며 다양한 경로 계획 및 추종 제어기를 비교한 결과 드론의 도로 물체 사전 인지를 통해 자율주행 차량의 회피 성능과 기동 성능 평가 프레임워크를 개발하여 이를 활용하여 본 연구에서 제안된 시스템이 기존의 자율주행 차량 단독으로 운행되는 것보다 그 성능 이 확연히 개선됨을 확인할 수 있다.
Table Of Contents
1. Introduction 1 2. Theoretical Background 3 2.1 System Dynamics 3 2.1.1 Autonomous Ground Vehicle (AGV) Dynamics 3 2.1.2 Autonomous Unmanned Aerial Vehicle (UAV) Dynamics 5 2.2 Path Planning Algorithms 6 2.2.1 A* Algorithm 6 2.2.2 Artificial Potential Field (APF) Occupancy Function 7 2.3 Path Tracking Controllers 8 2.3.1 Pure Pursuit 8 2.3.2 Stanley Control 9 2.3.3 Model Predictive Control (MPC) 10 2.3.4 PID Cascade Control Loop for UAV 11 2.4 Localization and Mapping 12 2.4.1 Ultrasonic Indoor GPS 12 2.4.2 IR Motion Capture 13 2.5 Object Detection 14 3. Simulation and Experimental Setup 15 3.1 Simulation Setup 15 3.1.1 Simulation Framework 15 3.1.2 System Validation with Unreal Engine 16 3.2 Experimental Setup 16 3.2.1 Hardware Configuration 16 3.2.2 Experimental Framework 18 3.3 Scenarios of Road Environments 19 4. Simulation and Experimental Results 21 4.1 Simulation Results 21 4.1.1 Simulation Case 1: Emergency Braking Performance 21 4.1.2 Simulation Case 2: Vehicle Maneuvering Performance 24 4.2 Experiment Results 30 4.2.1 Localization and Mapping Evaluation 30 4.2.2 AGV Evasive Maneuvering Performance 31 5. Discussion and Remarks 33 6. Conclusion 34 Bibliography 35
Research Interests
Autonomous Vehicle and Aerial Robotic Systems and Control; Theory and Applications for Mechatronic Systems and Control; 자율 주행 및 비행 시스템 제어; 로봇공학 및 지능제어