ROS深度解析：机器人操作系统全面指南第二卷

需积分: 9 132 浏览量更新于2024-07-18 1 收藏 27.53MB PDF 举报

《机器人操作系统（ROS）：完整参考指南第二卷》是Studies in Computational Intelligence系列中的一部重要著作，由Anis Koubaa编辑，该系列由Janusz Kacprzyk担任系列编辑，隶属于波兰科学院，华沙。此书作为Robot Operating System (ROS) 的详尽参考手册，深入探讨了ROS在机器人技术领域的核心作用和广泛应用。 ROS是一个开源的、强大的机器人操作系统平台，旨在简化软件开发过程，使得创建和部署机器人系统变得更加高效和模块化。它提供了一种标准的框架，支持各种传感器数据处理、任务规划、控制和通信，适用于各种类型的机器人，包括工业、服务、个人和科研领域。本第二卷涵盖了ROS的核心概念、设计理念、API的详细解析以及实际应用案例分析。读者可以在这里了解到ROS架构的各个组件，如消息传递系统、节点管理、包管理、以及其丰富的第三方库和工具。此外，书中还讨论了ROS与各种机器学习算法、计算机视觉和自主导航技术的集成，以及如何利用ROS进行多机器人协作和网络部署。 "Studies in Computational Intelligence"系列的定位是快速发布计算智能领域的最新研究成果和进步，包括神经网络、连接主义系统、遗传算法、进化计算、人工智能、细胞自动机、自组织系统、软计算、模糊系统和混合智能系统等多个分支。该系列书籍不仅为研究者提供了深入探讨的平台，也使广大工程师和爱好者能够迅速获取到最新的技术和实践经验。通过阅读《Robot Operating System (ROS) The Complete Reference Volume 2》，读者不仅能掌握ROS的基础理论，还能学习如何将其应用于实际项目中，解决复杂的机器人问题。对于想要在机器人技术领域深入发展的人来说，这是一本不可或缺的参考资料。无论是从事机器人设计、研发，还是希望了解当前ROS的发展趋势和技术动态，这本书都提供了丰富的信息资源。

Weijia Yao College of Mechatronics and Automation, National University of

Defense Technology, Changsha, China

Qinghua Yu College of Mechatronics and Automation, National University of

Defense Technology, Changsha, China

Zhiqiang Zheng College of Mechat ronics and Automation, National Univ ersity of

Defense Technology, Changsha, China

Editor and Contributors xvii

4 M.Kameletal.

Precise trajectory tracking is a demanding feature for aerial robots in general in

order to successfully perform required tasks, especially when operating in realistic

environments where external disturbances may heavily affect the ﬂight performance

and when ﬂying in the vicinity of structure. In this chapter, several model predictive

control strategies for trajectory tracking are presented for multi-rotor systems as well

as for ﬁxed-wing aircraft. The general control structure followed by this chapter is a

cascade control approach, where a reliable and system-speciﬁc low-level controller

is present as inner loop, and a model-based trajectory tracking controller is running

as an outer loop. This approach is motivated by the fact that many critical ﬂight

software is running on a separate navigation hardware which is typically based on

micro-controllers, such as Pixhawk PX4 and Navio [1, 2] while high level tasks are

running on more powerful -but less reliable- on-board computers. This introduces a

separation layer to keep critical tasks running despite any failure in the more complex

high-level computer.

By the end of this chapter, the reader should be able to implement and test various

Model Predictive Control strategies for aerial robots trajectory tracking, and integrate

these controllers into the Robot Operating System (ROS) [3]. Various implementation

hints and practical suggestions are provided in this chapter and we show several

experimental results to evaluate the proposed control algorithms on real systems.

In Sect.2 the general theory behind MPC is presented, with focus on linear MPC

and robust linear MPC. In Sect. 3 we present the multi-rotor system model and

give hints on model identiﬁcation approaches. Moreover, linear and robust MPC

are presented and we show how to integrate these controller into ROS and present

experimental validation results. In Sect. 4 we present a Nonlinear MPC approach for

lateral-directional position control of ﬁxed-wing aircraft with implementation hints

and validation experiments.

2 Background

2.1 Concepts of Receding Horizon Control

Receding Horizon Control (RHC) corresponds to the historic evolution in control

theory that aimed to attack the known challenges of ﬁxed horizon control. Fixed

horizon optimization computes a sequence of control actions {u

, u

,...,u

N −1

}

over a horizon N and is characterized by two main drawbacks, namely: (a) when

an unexpected (unknown during the control design phase) disturbance takes place

or when the model employed for control synthesis behaves different than the actual

system, then the controller has no way to account for that over the computed control

sequence, and (b) as one approaches the ﬁnal control steps (over the computer ﬁxed

horizon) the control law “gives up trying” since there is too little time left in the ﬁxed

horizon to go to achieve a signiﬁcant objective function reduction. To address these

limitations, RHC proposed the alternative strategy of computing the full control

Model Predictive Control for Trajectory Tracking … 5

sequence, applying only the ﬁrst step of it and then repeating the whole process

iteratively (receding horizon fashion). RHC strategies are in general applicable to

nonlinear dynamics of the form (considering that state update is available):

x = f(x, u) (1)

where the vector ﬁeld f : R

× R

, x ∈ R

n×1

represents the state vector, and u ∈

m×1

the input vector. The general state feedback-based RHC optimization problem

takes the following form:

min

F(x

t+N

) +

N −1



k=0

||x

t+k

− x

ref

t+k



+||u

t+k



s.t. x

t+k+1

= f (x

t+k

, u

t+k

)

t+k

∈ U

t+k

∈ X

= x(t)

(2)

where z ={u

, u

t+1

,...,u

t+N −1

} is the optimization variables,  denotes some

(penalized) metric used for per-stage weighting, F (x

t+N

) represents the terminal

state weighting, x

ref

t+k

is the reference signal, the subscript t + k is used to denote the

sample (using a ﬁxed sampling time T

) of a signal at k steps ahead of the current

time t, while t + k + 1 indicates the next evolution of that, U

represents the set of

input constraints, X

the state constraints and x(t) is the value of the state vector at

the beginning of the current RHC iteration. The solution of this optimization problem

leads again to an optimal control sequence {u



, u



t+1

,...,u



t+N −1

} but only the ﬁrst

step of that u



is applied while the whole process is then repeated iteratively.

Within this formulation, the term F(x

t+N

) has a critical role for the closed-loop

stability. In particular, it forces the system state to take values within a particular set

at the end of the prediction horizon. It is relatively easy to prove stability per local

iteration using Lyapunov analysis. In its simplest case, this essentially means that

considering the regulation problem (x

ref

t+k

= 0fork = 0,...,N − 1), and a “decres-

cent” metric , then the solution of the above optimization problem makes the system

stable at x

= 0, u

= 0 – that is that a terminal constraint x

t+k

= 0 is introduced (a

simpliﬁed illustration is provided in Fig. 1). However, the question of global stability

Fig. 1 Illustration of the terminal constraint set (Ω)

剩余651页未读，继续阅读

独孤尚亮dugushangliang

粉丝: 556
资源: 5

ROS深度解析：机器人操作系统全面指南第二卷

ROS与Python入门教程.pdf

Robot Operating System (ROS) The Complete Reference (Volume 4).pdf

Robot Operating System (ROS) The Complete Reference (Volume 2) 2017

RobotOperatingSystem(ROS)_TheCompleteReference(Volume3)byAnisKoubaa

Robot Operating System (ROS) The Complete Reference (Volume 1) 2016

No package 'orocos-bfl' found

Robot Operating System (ROS) 2.pdf

ROS机器人操作系统Robot Operating System

Robot Operating System (ROS) The Complete Reference (Volume 1)

ROS（Robot Operating System）安装配置.zip

最新资源