Coordinated path following control and formation control of mobile robots [Elektronische Ressource] / vorgelegt von Kiattisin Kanjanawanishkul

Kiattisin Kanjanawanishkul

Informations

• Publié par: eberhard_karls_universitat_tubingen
• Publié le: 01 janvier 2010
• Nombre de lectures: 18
• Langue: English
• Poids de l'ouvrage: 3 Mo

Extrait

CoordinatedPathFollowingControl
andFormationControlof
MobileRobots
Dissertation
der Fakulta¨t fu¨r Informations- und Kognitionswissenschaften
der Eberhard-Karls-Universita¨t Tu¨bingen
zur Erlangung des Grades eines
Doktors der Naturwissenschaften
(Dr. rer. nat.)
vorgelegt von
M.Sc. Kiattisin Kanjanawanishkul
aus Trang, Thailand
Tu¨bingen
2010Tag der mu¨ndlichen Qualifikation: 23.07.2010
Dekan: Prof. Dr.-Ing. Oliver Kohlbacher
1. Berichterstatter: Prof. Dr. Andreas Zell
2. Berichterstatter: Prof. Dr.-Ing. Frank Allgo¨werAbstract
Rapidadvancesinsensing,computingandcommunicationtechnologieshaveledtocon-
siderably increased research activities in multi-robot systems over the last decade. Top-
ics include multi-robot motion planning, cooperative manipulation, aerial applications
involvingcooperativeexplorationoftheunknownenvironment,automatedhighwaysys-
tems,softwarearchitecturesformulti-robotsystems,andformationcontrol. Multi-robot
systemshavebeenproventoofferadditionaladvantagesintermsofflexibilityinoperat-
ingagroupofrobotsandfailuretoleranceduetoredundancyinavailablemobilerobots.
However,thebenefitsofusingmulti-robotteamsdonotcomewithoutcost. Coordinating
teamsofautonomousrobotsismuchmorechallengingthanmaneuveringasinglerobot.
Thisdissertationaddressesformationcontrolproblems,whichareamongthemostac-
tive research topics in multi-robot systems. Over the last two decades, there have been
a large number of publications on this field, and it is still growing. Recently, this re-
search has been extended to some related research areas, e.g., consensus problems and
distributed control systems, imposing new challenges on formation control problems.
In general, formation control subproblems addressed in the literature can be classified
as formation shape generation, formation reconfiguration/selection, formation tracking,
and role assignment in formation. The main purpose of this dissertation is to address
two important and correlated subproblems in formation control: formation tracking and
role assignment in formation. The goal of the former is that a team of mobile robots is
required to maintain a geometric formation while tracking a reference or a set of refer-
ences. The latter arises when a mobile robot in the team must decide what role to take
on in a desired formation configuration.
In particular, we study coordinated path following control of omnidirectional mobile
robots and unicycle mobile robots. This problem can be seen as a subtask of formation
tracking. Path following is one of the three basic motion control tasks in mobile robot
research. The others are trajectory tracking and point stabilization. Even though less
attention is drawn to this problem in the literature, it offers some advantages over tra-
jectory tracking in some cases. The objective of path following control is to be on the
path rather than at a certain point at a particular time. To solve this problem, we employ
a model predictive control (MPC) technique to generate a sequence of optimal veloci-
ties of a so-called virtual vehicle which is followed by a real robot. This approach can
eliminate stringent initial condition constraints because the velocity of a virtual vehicle
iscontrolledexplicitly. Usingthistechnique,wecangainsomebenefitsoverotheravail-
ablecontrolschemes,e.g.,theabilitytoincorporategenericmodels,linearandnonlinear,
andconstraintsintheoptimalcontrolproblemandtheabilitytousefuturevaluesofref-
iiierences when they are available, allowing to improve system performance. However,
the main drawback is significant computational burden associated with solving a set of
nonlinear differential equations and a nonlinear dynamic optimization problem online.
Then, we extend path following control to coordinated path following control. A
group of mobile robots not only follow a reference path but also maintain a geometric
formation shape. The main challenge is to design a decentralized control law using only
local information to achieve a formation tracking objective. In this study, we propose
two solutions. In the first solution, the MPC framework for path following control is
extended to the coordinated path following control problem. In spite of great theoretical
propertiesofsuchMPCcontrollers,thestabilityandfeasibilityofdecentralizedschemes
are rather conservative. The second solution is computationally simple so that it may be
suitableforlow-computationalsystemswhentheadvantagesofMPCschemesincluding
constraint handling are not a dominating factor. Its controller design is based on a Lya-
punov technique and a second-order consensus protocol with a reference velocity. It is
worth noting that the path variable has been used as a coupling variable synchronizing
each member in formation in both solutions.
In the second formation control subproblem, we study role assignment in formation.
This problem becomes more challenging when robots in the team do not have complete
information and they do not know the number of robots participating in the formation
tasks. With the assumption that the formation graph is connected and bidirectional, we
proposeanonlineanddistributed roleassignment. Thisapproachisprovenbyextensive
simulation and experimental results.
ivAcknowledgments
I would first like to acknowledge my advisor, Prof. Dr. Andreas Zell, for giving me an
opportunity to do my research work on a team of mobile robots and for his continuing
support and guidance throughout my years at the University of Tu¨bingen. I would also
like to thank the dissertation committee member, Prof. Dr.-Ing. Frank Allgo¨wer, for his
constructive advice and efforts in reading and commenting on my dissertation.
I am thankful to the Attempto Tu¨bingen Robot Soccer team members for developing
the RobocCup robots which I have used for validating my control algorithms. I owe
specialthankstoDr. XiangLiforhelpingmeusetheRoboCupsoftwareframeworkand
controltheRoboCuprobotsduringmyfirstyearandforintroducingmetopathfollowing
control problems and model predictive control.
IwouldliketothankallcolleaguesattheDepartmentofComputerArchitectureforan
excellent working atmosphere. I appreciate the help from Matthias Mu¨ller (from the In-
stituteforSystemsTheoryandAutomaticControloftheUniversityofStuttgart),Philipp
Vorst, Marius Hofmeister, Philippe Komma, and Karl E. Wenzel for their proofreading
of this dissertation and many valuable suggestions. I am especially indebted to my of-
fice mate, Philipp Vorst, for always willing to give me suggestions and assistance, the
interesting discussions, and for answering all my programming and Linux questions. I
gratefully acknowledge his help and support all the time. Furthermore, I am very thank-
fultoMariusHofmeisterforhelpingmeduringmyfirstyearlectures,andforhissupport
in experiments. Last but not least I would like to thank Dr. Christian Weiss and Karsten
Bohlmann for their help.
Additionally,IwouldliketothankVitaSerbakovafordoingagreatjobasourdepart-
ment’s secretary. Many thanks go to Klaus Beyreuther who is always helpful with all
computer problems.
Most importantly, I profoundly thank my parents and my sisters for their love, their
encouragement and constant support throughout the years.
This dissertation was made possible by a Thai government scholarship. I am grateful
to the Thai government for the scholarship which enabled me to undertake a PhD study
at the University of Tu¨bingen.
vviContents
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Dissertation Organization . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 BackgroundControlTheory 5
2.1 Nonlinear System Theory . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Lipschitz Functions . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Lyapunov Stability . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3 The Invariance Principle . . . . . . . . . . . . . . . . . . . . . 8
2.1.4 Nonautonomous Systems . . . . . . . . . . . . . . . . . . . . . 8
2.1.5 Barbalat’s Lemma and Stability of Time-varying Systems . . . 10
2.1.6 Boundedness . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Model Predictive Control (MPC) . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Principles and Formulation . . . . . . . . . . . . . . . . . . . . 12
2.2.2 Issues on Nonlinear MPC . . . . . . . . . . . . . . . . . . . . 15
2.2.3 Optimization Solvers . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.4 Centralized MPC vs. Decentralized MPC . . . . . . . . . . . . 21
2.3 Consensus Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 RobotSystems 25
3.1 System Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Heterogeneity vs. Homogeneity . . . . . . . . . . . . . . . . . 25
3.1.2 Communication Structures . . . . . . . . . . . . . . . . . . . . 26
3.1.3 Centralization vs. Decentralization . . . . . . . . . . . . . . . . 27
3.2 Robot Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Omnidirectional Mobile Robots . . . . . . . . . . . . . . . . . 28
3.2.2 Unicycle Mobile Robots . . . . . . . . . . . . . . . . . . . . . 32
3.3 Software Frameworks . . . . . . . . . . . . . . . . . . .