Cited time in webofscience Cited time in scopus

Robust and Resilient Control Design for Vehicle Platooning

Title
Robust and Resilient Control Design for Vehicle Platooning
Alternative Title
강인하고 복원력있는 군집차량 제어설계
Author(s)
Gyujin Na
DGIST Authors
Gyujin NaYongsoon EunKyoung-Dae Kim
Advisor
은용순
Co-Advisor(s)
Kyoung-Dae Kim
Issued Date
2022
Awarded Date
2022/02
Type
Thesis
Subject
Robustness, resiliency, platooning system, autonomous vehicle, fault detection, probing signal, disturbance observer, unknown input observer
Description
Robustness, resiliency, platooning system, autonomous vehicle, fault detection, probing signal, disturbance observer, unknown input observer
Abstract
자율군집차량기술이란 차량간의 거리와 속도를 일정하게 유지하여 후행차량이 받는 공기저항을 줄여 연료효율을 높일 수 있는 기술을 말한다. 차량군집주행시스템을 완벽하게 구성하기 위해서는 많은 중요한 문제를 해결해야하며, 이 논문은 몇가지의 중요한 문제를 다룬다. 첫번째 문제는 차량과 도로의 불확실성이 존재하는 상황에서 차량간의 간격을 유지할 수 있는 기술이며, 두 번째 문제는 센서 또는 액추에이터 오작동에 대한 군집주행의 복원력에 대한 기술이다. 첫번째 문제에 대해 본 논문은 군집주행 제어에 외란관측기를 적용하여 차량모델의 불확실성을 보상하고, 환경의 주요교란으로 작용하는 도로경사의 영향을 보상하는 방법에 대한 기술을 제시한다. 군집차량은 외란관측기에 의한 교란보상으로 향상된 추적성능을 보이며, 나아가 더욱 향상된 군집주행 성능을 보여준다. 두번째 문제는 주변차량의 센서를 활용하여 오작동 및 센서공격을 견디는방법과 활성신호를 활용하여 오동작을 감지하는 기법을 제시한다. 결함차량의 주변차량이 상대거리 센서를 사용하여 결함차량의 움직임을 관찰할 수 있을경우, 주변차량이 결함차량에 보다 신뢰할 수 있는 센서 데이터를 보낼 수 있다. 본 논문은 이러한 원리를 활용하여 복원성을 보장하고, 동시에 외란강인성을 보장할 수 있는 군집차량 기술을 제시한다. 본 논문은 센서 공격을 능동적으로 감지하기위해 프로빙 신호를 액추에이터에 주입하는 기법도 소개한다. 기존의 프로빙 기반 고장검지 기술들은 제어입력에 액티브한 신호를 인가할 경우, 차량의 제어성능을 악화시키는 문제를 가진다. 이러한 단점을 해결하고자, 외란관측기를 활용하여 제어성능을 보장하고 동시에 고장을 탐지할 수 있는 기술을 개발한다. 본 기술의 효력을 입증하기위해, 군집차량, 자율주행차량, 실제 무인차량로봇에 탐지기술을 적용하여 그 성능을 입증한다. 마지막으로 액추에이터 고장검지를 위해 미지입력관측기을 사용하여 각 구동기의 개별고장을 탐지하는기법을 제시한다. 자율주행 전기자동차에서 모터 오작동에 대한 다양한 시나리오를 해결하기위해 차량모델을 4륜구동선형 매개변수 가변시스템 모델로 고려한다. 차량역학은 선형매개변수 가변시스템으로 표현되기에 액추에이터결함 감지방법은 선형시변 시스템에대해 개발된 미지입력관측기를 활용하여 구현된다.|A platoon is a group of vehicles that can travel very closely together, safely at high speed. Enabled by advances in control, communication and computer technologies, it has already been implemented and demonstrated with heavy duty trucks, and is a part of many features that self-driving cars may embrace in the future. Benefits from vehicle platooning include the increase of the road capacity by achieving more steady-state traffic flow, the reduction of fuel consumption by short inter-vehicle distance that reduces the aerodynamic drag of the following vehicles, and gained efficiency due to all the cars travelling close together and moving at the same time. Many important problems must be addressed to perfect the vehicle platooning methods and systems. This dissertation addresses several of the important problems. First, the performance of maintaining gap between the vehicles is important especially in the presence of uncertainties from both vehicles and roads. Second, resiliency against sensor or actuator malfunctions is of high importance to deploy the technology in our daily lives. This is especially so with many prior occurrences of malicious attacks on control systems. We address the mentioned problems and provide solutions. For the first problem, we apply disturbance observer (DOB) in platooning control. The DOB not only compensates the uncertainties in the vehicle model but also provides a means to compensate the effect of road slope, which acts as a major disturbance from the environment. When platooning vehicles follow the desired velocity reference with elevated tracking performance thanks to the disturbance compensation by the DOB, higher fuel efficiency follows for the vehicles in the platoon. For the second problem, we provide a method of actively detecting attacks or malfunctions on the sensor, and also a method of tolerating the attack by utilizing sensors of neighboring vehicles. For active detection of sensor attack, a probing signal is injected in the actuator. Employing probing signals in actuators is a well-known and effective strategy to detect replay attacks. The usage of probing signals may detect the replay attacks. For the attack tolerating strategy, the use of sensors from the neighboring vehicles are proposed. When neighboring vehicles of a faulty vehicle can observe faulty vehicle’s movements using relative distance sensors, surrounding vehicles may send more reliable sensor data to the faulty vehicle. Finally, for the attack on actuators, we provide a method of detection using unknown input observers. In this particular case, we have extended the model of the vehicle to a four wheel driven one (in other part of the dissertation, the vehicle model is a simple point mass) to address various scenarios of motor malfunction or attack in self-driving electric vehicles. The vehicle dynamics is represented into linear parameter varying system and an actuator fault detection method is derived using unknown input observers for linear time varying systems.
Table Of Contents
I. Introduction 1
1.1 Necessity of Vehicle Platooning 1
1.2 Contributions and Outline of Dissertation 2
II. Disturbance Observer Approach for Fuel-efficient Heavy-duty Vehicle Platooning 11
2.1 Platoon Architecture 11
2.2 Modelling for Vehicle Dynamics 14
2.3 Vehicle Controller and Reference Generator 15
2.3.1 Vehicle Controller with Disturbance Observer 15
2.3.2 Reference Generator 18
2.4 Main Results 19
2.4.1 Generation of v using Estimated Road Slope 19
2.4.2 Closed-loop Platoon Operation 24
2.4.3 Comparison with Alternative Solution 28
III. Robust and Resilient Vehicle Platooning Method 33
3.1 Platoon Architecture with Sensor Resiliency 33
3.2 Vehicular Control Systems with Robustness and Resiliency 35
3.3 Fault Detection Method 37
3.4 Platoon Operation Results under Sensor Faults 40
3.5 Error Bounds for Avoiding Vehicle Collision 44
IV. A Probing Signal-based Sensor Attack Detection Method Avoiding Control Performance Degradation 51
4.1 Motivation 51
4.1.1 Systems with Synchronous Attack Detection 51
4.1.2 Motivating Example 53
4.2 Disturbance Observer-based Replay Attack Detector 54
4.3 Detector Optimization 59
4.4 Application to Vehicle Platooning 63
4.4.1 Platooning Architecture with a Proposed Detector 63
4.4.2 Detector Parameter Selection 65
4.5 Experiment Results Using Unmanned Ground Vehicle 67
V. Application of Probing Detection Method into Autonomous Vehicular Systems 73
5.1 An Autonomous Vehicular System with Attack Detection 73
5.1.1 Architecture 73
5.1.2 Attack Detection 75
5.1.3 Nonlinear Vehicle Dynamics 76
5.1.4 Velocity Controller and Steering Controller 78
5.1.5 Design of DOBy and DOBδ 79
5.2 Simulation Results 81
VI. Actuator Fault Detection Method for Autonomous Unmanned Ground Vehicles 87
6.1 Understanding of Unknown Input Observer 87
6.2 Four Wheel Unmanned Ground Vehicle Dynamics 88
6.3 Actuator Fault Detection Method 90
6.4 Simulation Results 94
6.4.1 Actuator Fault Detection of individual UGV 94
6.4.2 Discussion on Friction Coefficient Estimation 96
6.4.3 Performance Evaluation under Swarm Scenario 96
6.5 A Data-Driven Motor Fault Detection Method 98
6.6 A Testbed Architecture for Motor Fault Detection 102
6.7 Motor Fault Detection Experiment 104
6.7.1 DC Motors Considered 104
6.7.2 Experiment Results 104
VII. Conclusions of Dissertation 115
Bibliography 117
국문초록 129
URI
http://dgist.dcollection.net/common/orgView/200000592058

http://hdl.handle.net/20.500.11750/16318
DOI
10.22677/thesis.200000592058
Degree
Doctor
Department
Information and Communication Engineering
Publisher
DGIST
Related Researcher
  • 은용순 Eun, Yongsoon
  • Research Interests Resilient control systems; Control systems with nonlinear sensors and actuators; Quasi-linear control systems; Intelligent transportation systems; Networked control systems
Files in This Item:

There are no files associated with this item.

Appears in Collections:
Department of Electrical Engineering and Computer Science Theses Ph.D.

qrcode

  • twitter
  • facebook
  • mendeley

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE