高级驾驶辅助系统与自动驾驶控制策略

需积分: 9 138 浏览量更新于2024-07-17 4 收藏 12.11MB PDF 举报

"Control Strategies for Advanced Driver Assistance Systems and Autonomous Driving" 是一本专注于高级驾驶辅助系统（ADAS）和自动驾驶功能（ADF）控制策略的书籍，由国际专家撰写，旨在介绍和测试这些系统的算法开发、测试和验证方法。书中强调了在不断变化的操作条件下，控制算法必须确保安全、可靠和最优响应，并符合安全和排放法规。面对车辆与环境及交通参与者复杂互动产生的无数可能场景，该书探讨了新的处理复杂性的方法，包括人类交互建模、现实世界场景的理论定义，以及实践案例，以促进ADAS和ADF的高效开发和测试。在自动驾驶领域，控制策略是至关重要的，因为它们决定了车辆如何理解周围环境并做出决策。这本书涵盖了以下几个关键知识点： 1. **控制算法的开发**：书中详细介绍了如何设计和实施适用于真实驾驶环境的控制算法。这涉及到了对车辆动态模型的理解、控制理论的应用，以及如何在算法中考虑不确定性、实时性和安全性。 2. **测试与验证**：对于ADAS和ADF，测试和验证是确保其性能的关键步骤。书中讨论了如何模拟各种驾驶条件，创建代表性的场景，以及如何利用仿真和实地测试来评估算法的性能。 3. **人类交互建模**：由于自动驾驶系统需要理解和预测人类驾驶员的行为，人类交互建模是重要一环。这部分可能包括了心理学、行为科学和统计学的方法，以创建准确的人类驾驶员行为模型。 4. **理论场景定义**：书中的理论方法可以帮助开发者定义和分析可能的驾驶场景，这些场景可能包括复杂的交通情况、天气条件、道路特征等，从而确保算法能够适应各种复杂环境。 5. **实践案例**：书中包含的实际应用例子为读者提供了直观理解，展示如何将理论知识应用于实际的ADAS和ADF开发中，这有助于研究人员和工程师更好地理解如何解决实际问题。 6. **安全法规遵从性**：自动驾驶技术必须符合国家和地区的安全法规，书中的内容可能涵盖了如何在设计和验证过程中考虑这些法规要求。这本书对学术研究人员、工业研究者、汽车公司和供应商来说是一份宝贵的资源，它提供了一个全面的视角，涵盖了从理论到实践的各个层面，对于推动自动驾驶技术的进步具有重要意义。通过深入学习，读者可以更深入地了解如何设计和实现安全、智能的自动驾驶系统。

As a consequence, classical approaches, such as certiﬁcation and ﬂeet testing,

are not capable of handling the complexity, and so new methods to allow efﬁcient

development and testing are required. Although many academic and industrial

research projects focus on this topic and aim to provide guidelines and solutions, up

to now no consensus on the best practice is established. Moreover, with self-driving

vehicles, new challenges arise which are not only related to technical aspects but

require insights into legal, psychological, and social issu es including human

interaction.

It appears that simulations using models and hardware-in-the-loop facilities will

be a key enabler for testing, validation, veriﬁcation, and certiﬁcation of the auto-

mated driving functions.

This testing in “virtual world” or mixed “real–virtual world” allows to cover the

large range of potential scenarios and complexity levels involved in automated

driving. To manage complexity, a modular environment which allows to simulate

the system at different levels and with different real or virtual components seems to

be necessary. Still, it is the question which detail levels are sufﬁcient and can be

handled. A very important component is the scenarios, i.e., the driving and envi-

ronment situations, which are used in the simulation studies. These scenarios should

cover all (realistic but also worst case) situations which the vehicle may encounter

during the driving mission including a realistic interaction with other trafﬁc par-

ticipants and the environment. Further, it is also necessary to understand (and

model) the interactions between the driver and automated driving functions, in

particular, as the near-term systems will likely be supervised by the drivers and/or

handover the control of the vehicle to the driver under some conditions.

During September 29–30, 2016, a workshop on the subjects of development,

testing, and veriﬁcation of ADAS and ADF has taken place at the Johannes Kepler

University Linz to assess the status and discuss ways to improve the strategies for

development and testing of such advanced driving functions. The workshop con-

sisted of 14 talks by experts from industry and academia, covering topics in human

interaction, the impact on the powertrain and emissions, testing frameworks, sce-

nario design, and function development.

During the talks and presentation at the workshop, several interesting observa-

tions have been made related to the current status and future directions of this ﬁeld.

Examples of these observations are summarized below:

• Currently, in both automated driving and advanced driver assistance function

development, the main focus is on improving safety and comfort of the passengers.

Additional performance aspects, such as related to fuel/energy consumption, CO

and other tailpipe emissions, are not considered explicitly. The general assumption

is that these will be dealt with later while the cur rent focus is on vehicle trajectory

and velocity planning. It is expected that once the fundamental challenges of

automated dr iving are addressed sat isfactorily, fuel econ omy and emission-related

aspects will be fully considered. This somewhat incremental development

approach may lead to automated vehicles with suboptimal overall system perfor-

mance which are difﬁcult to redesign at a later stage. Thus, pursuing holistic

xvi Introduction

opportunities involved in automated driving that involves fuel/energy efﬁciency

and emissions appears to be worthwhile.

• There is an increasing emphasis on trafﬁc scenario development in simulation

environments. These environments seem to be a suitable tool to test and develop

control functions in safe and scalable manner. It is important for the models used

to be realistic including presenting responses of other vehicl es, the environment

and the physical properties and limitations of sensors and actuators. Depending

on the intended system, test, and available time, different models and detail

levels are necessary.

• The role of mixed (virtual–real) testing is also increasing, but is mainly limited

to laboratory environment. There is a large potential to perform tests in the ﬁeld

of HIL setups with engine, powertrain, or vehicle in the loop. Modular setups

which allow to exchange virtual and real components and execute the same

scenario need to be developed, and one of the main challenges in this area is on

a clean interface design. Her e, functional mock-up units and interfaces

(FMU/FMI) seem to be a valuable approach.

• Currently, a large focus is on simulations with models that are not validated

from experiments, because such validations may require billions of real-world-

driven kilometers. In this case, much can be gained, when the loop from the road

to virtual development is closed by feeding back the ﬁeld results to the virtual

environments which can lead to improved uncertainty models. There seems to

be a large potential for public data sharing, between OEMs, academia, and

governmental agencies.

• Up to now, a standardized method for ADAS an d ADF testing and certiﬁcation is

missing. Here, it may be important to deal with public perception and addr ess the

upcoming issues in a clear and still understandable manner. Furthermore, there are

large regional differen ces which make the establishment of a standardized method

even more difﬁcult. Additionally, in actual testing and certiﬁcation of veh icles, the

impact of ADAS and ADF on CO

and pollutant emissions is presently not

covered by legislation. Consequently, it is expected that this impact will become

important for the development of future representative test cycles.

• The workshop focused mainly on passenger car applications; however, many

presented metho ds may be applied in commerc ial an d he avy-duty applications too .

This book combines and extends selected results presented at the workshop. It

covers different aspects of development , testing, and veriﬁcation of control systems

for ADAS and ADF and provides insight into the underlying issues from both,

theoretical and application, points of view.

Introduction xvii

Chapter 1

Cooperation and the Role of Autonomy

in Automated Driving

Gina Wessel, Eugen Altendorf, Constanze Schreck,

Yigiterkut Canpolat and Frank Flemisch

Abstract While automation in human–machine systems can increase safety and

comfort, the 2016 lethal crash of an automated vehicle demonstrates that automation

is not without its risk. Indeed, accidents from the air trafﬁc domain also demonstrate

that cooperation between human and machine is crucial and interaction design must

be devoted great care. Therefore, the present paper aims at developing design recom-

mendations to reduce the risks of lethal crashes with automated vehicles. To this end,

the concepts of cooperation and autonomy are closely investigated. These two terms

are central to research on human machine cooperation; however, the present deﬁni-

tion of cooperationand the role of autonomy might be further speciﬁed for the domain

of automated driving. Therefore, selected perspectives from different scientiﬁc ﬁelds

(e.g., sociology and psychology) will be presented in order to develop a differentially

inspired working deﬁnition of cooperation, which is tailored to the automated driving

domain. Another goal of this approach is to investigate different views on the concept

of autonomy, which is often entailed in work on cooperation. This can help clarify

the role of autonomy in automated driving in particular. Moreover, insights from the

presented theories and ﬁndings on cooperation can be transferred to the interaction

design of automated vehicles. Accordingly, recommendations for interaction design

will be presented. Finally, an example for the implementation of the working deﬁ-

nition and the design recommendations will be presented by describing a prototype

for automated driving—the H-mode prototype.

G. Wessel (

) · E. Altendorf · C. Schreck · Y. Canpolat

Institute of Industrial Engineering and Ergonomics,

RWTH Aachen University, Aachen, Germany

e-mail: g.wessel@iaw.rwth-aachen.de

F. Flemisch

Fraunhofer Institute for Communication, Information Processing and Ergonomics FKIE,

Wachtberg, Germany

H. Waschl et al. (eds.), Control Strategies for Advanced Driver Assistance

Systems and Autonomous Driving Functions, Lecture Notes in Control

and Information Sciences 476, https://doi.org/10.1007/978-3-319-91569-2_1

2 G. Wessel et al.

1.1 Introduction

“Alone we can do so little; together we can do so much”

Helen Keller

This quote of a deaf-blind author seems strikingly simple and true, yet the application

of its meaning is not that easy. Cooperation between humans can be a complex issue,

and cooperation between humans and machines might be even more intricate. Which

roles do human and machine play? Who is responsible for joint actions? How much

autonomy should a machine have? Or how much autonomy of a machine is needed

before cooperation can take place? And is cooperation with a machine even possible

at all?

A special ﬁeld of application of these questions is the domain of automated driv-

ing. In this domain, the concept of highly automated and cooperative automation

prevails and emphasis is put on the advantages of combining the assets of driver and

automation for creating a high-performing system, e.g., [36, 38]. The term cooper-

ative guidance and control is used to denote a focus on these two aspects of driving.

In the framework of these concepts, driver and automation need to act jointly, which

makes questions about the automation’s autonomy inevitable [6, 21, 22, 25].

Even though the concepts of cooperation and autonomy have been deﬁned in

a general way by dictionaries, their range of application is very wide. Therefore,

several domains have speciﬁed and deﬁned both concepts in a tailored way in order

to organize research, for example, in teamwork [31] or biology research [32]. For

the automated driving domain, the following deﬁnition of cooperative guidance and

control has been presented:

the action or process of working together of at least one human and at least one computer

on the guidance and control of at least one vehicle [25].

Furthermore, Flemisch et al. [21] claim that at higher levels of automated driving the

capability of the machine to function autonomously is a necessary prerequisite for

cooperative driving. Additionally, they point out that at higher levels of autonomy,

there might be a tradeoff with cooperativeness. These two statements demonstrate

the interdependence of the two concepts and point out the need to provide further

explanation.

While this shows that these concepts already received attention and tailoring to

the domain of automated driving, this domain is still a relatively new scientiﬁc ﬁeld.

Therefore, deﬁnitions and research ﬁndings concerning cooperation and the role of

autonomy from different domains could inspire formulating an elaborated working

deﬁnition of cooperation and clarify the role of machine autonomy in driving. Fur-

thermore, a solid foundation of the concepts of cooperation and autonomy could

help develop recommendations for interaction design. This is of major importance,

because some crashes with automated planes and the 2016 fatal Tesla crash can

(partly) be ascribed to bad interaction design.

Accordingly, in the following, general concepts and frameworks from the domain

of automated driving will be provided and an airplane crash and the 2016 Tesla

剩余234页未读，继续阅读

wang1062807258

粉丝: 13
资源: 272

高级驾驶辅助系统与自动驾驶控制策略

ADASIS v2 接口协议 (Advanced Driver Assistance Systems Interface）

ADASIS v2 接口协议 (Advanced Driver Assistance Systems Interface Specifications )

ADASIS_v3协议.zip

Dynamic event-triggered control for discrete Markov jump systems

给我找一些关于eps控制的文献

an introduction to stochastic differential equations version微盘

poly-encoders: architectures and pre-training strategies for fast and accura

Give me a sample IELTS essay with highly Advanced vocabulary and complex phrases on medical insurance

a practical guide to algorithmic strategies and trading systems pdf

最新资源