The DARPA Robotics Challenge Finals: Humanoid Robots To The Rescue

692 Pages·2018·37.087 MB·English

by  Matthew Spenko, Stephen Buerger, Karl Iagnemma

Springer Tracts in Advanced Robotics 121
Matthew Spenko
Stephen Buerger
Karl Iagnemma Editors
The DARPA
Robotics
Challenge Finals:
Humanoid Robots
To The Rescue
121
Springer Tracts in Advanced Robotics
Editors
Prof. Bruno Siciliano Prof. Oussama Khatib
Dipartimento di Ingegneria Elettrica Artificial Intelligence Laboratory
e Tecnologie dell’Informazione Department of Computer Science
Università degli Studi di Napoli Stanford University
Federico II Stanford, CA 94305-9010
Via Claudio 21, 80125 Napoli USA
Italy E-mail: [email protected]
E-mail: [email protected]
Editorial Advisory Board
Nancy Amato, Texas A & M, USA
Oliver Brock, TU Berlin, Germany
Herman Bruyninckx, KU Leuven, Belgium
Wolfram Burgard, Univ. Freiburg, Germany
Raja Chatila, ISIR - UPMC & CNRS, France
Francois Chaumette, INRIA Rennes - Bretagne Atlantique, France
Wan Kyun Chung, POSTECH, Korea
Peter Corke, Queensland Univ. Technology, Australia
Paolo Dario, Scuola S. Anna Pisa, Italy
Alessandro De Luca, Sapienza Univ. Rome, Italy
Rüdiger Dillmann, Univ. Karlsruhe, Germany
Ken Goldberg, UC Berkeley, USA
John Hollerbach, Univ. Utah, USA
Lydia Kavraki, Rice Univ., USA
Vijay Kumar, Univ. Pennsylvania, USA
Bradley Nelson, ETH Zürich, Switzerland
Frank Park, Seoul National Univ., Korea
Tim Salcudean, Univ. British Columbia, Canada
Roland Siegwart, ETH Zurich, Switzerland
Gaurav Sukhatme, Univ. Southern California, USA
More information about this series at http://www.springer.com/series/5208
Matthew Spenko Stephen Buerger
(cid:129)
Karl Iagnemma
Editors
The DARPA Robotics
Challenge Finals: Humanoid
Robots To The Rescue
123
Editors
MatthewSpenko KarlIagnemma
TheRobotics Lab Robotic Mobility Group
Illinois Institute of Technology (IIT) Massachusetts Institute of Technology
Chicago, IL Cambridge, MS
USA USA
StephenBuerger
HighConsequence Automation and
Robotics
SandiaNational Laboratories
Albuquerque, NM
USA
ISSN 1610-7438 ISSN 1610-742X (electronic)
SpringerTracts inAdvanced Robotics
ISBN978-3-319-74665-4 ISBN978-3-319-74666-1 (eBook)
https://doi.org/10.1007/978-3-319-74666-1
LibraryofCongressControlNumber:2017964438
©SpringerInternationalPublishingAG,partofSpringerNature2018
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpart
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
orinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar
methodologynowknownorhereafterdeveloped.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfrom
therelevantprotectivelawsandregulationsandthereforefreeforgeneraluse.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authorsortheeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinor
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictionalclaimsinpublishedmapsandinstitutionalaffiliations.
Printedonacid-freepaper
ThisSpringerimprintispublishedbytheregisteredcompanySpringerInternationalPublishingAG
partofSpringerNature
Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland
Series Foreword
Robotics is undergoing a major transformation in scope and dimension. From a
largely dominant industrial focus, robotics is rapidly expanding into human envi-
ronments and is vigorously engaged in its new challenges. Interacting with,
assisting, serving, and exploring with humans, the emerging robots will increas-
ingly touch people and their lives.
Beyond its impact on physical robots, the body of knowledge robotics has
produced is revealing a much wider range of applications reaching across diverse
research areas and scientific disciplines, such as biomechanics, haptics, neuro-
sciences,virtualsimulation,animation,surgery,andsensornetworksamongothers.
In return, the challenges of the newly emerging areas are proving an abundant
source of stimulation and insights for the field of robotics. It is indeed at the
intersection of disciplines that the most striking advances happen.
TheSpringerTractsinAdvancedRobotics(STAR)isdevotedtobringingtothe
research community the latest advances in the robotics field on the basis of their
significance and quality. Through a wide and timely dissemination of critical
researchdevelopmentsinrobotics,ourobjectivewiththisseriesistopromotemore
exchanges and collaborations among the researchers in the community and con-
tribute to further advancements in this rapidly growing field.
As one of robotics pioneering symposia, the International Symposium on
Robotics Research (ISRR) has established over the past two decades some of the
field’s most fundamental and lasting contributions. Since the launching of STAR,
ISRR and several other thematic symposia in robotics find an important platform
for closer links and extended reach within the robotics community.
This 16th edition of “Robotics Research,” edited by Masayuki Inaba and Peter
Corke, brings a collection of a broad range of topics in robotics including control,
design, intelligence and learning, manipulation, perception, and planning. The
content of these contributions provides a wide coverage of the current state of
robotics research: the advances and challenges in its theoretical foundation and
technology basis, and the developments in its traditional and novel areas of
applications.
v
vi SeriesForeword
The novelty and span of the work presented in this volume reveal the field’s
increasedmaturityandexpandedscope.This16theditionofISRRculminateswith
thisimportantreferenceonthecurrentdevelopmentsandnewdirectionsinthefield
of robotics—a true tribute to its contributors and organizers!
Stanford, California Oussama Khatib
November 2015 STAR Editor
Preface
Since the early 2000s, the United States’ Defense Advanced Research Projects
Agency (DARPA) has used competitions to spur progress in robotics and auton-
omous systems research. The model was first employed in 2004 in a series of
autonomous driving challenges. The 2004 DARPA Grand Challenge required
competitors to field vehicles capable of autonomously traversing a 150-mile-long
desert course, but ended in disappointment when no team traveled further than
7.3miles.Undeterred,DARPArepeatedthechallengethefollowingyear,andsaw
four teams finish the course. This success was followed in 2007 by the DARPA
Urban Challenge, which required autonomous vehicles to operate in a dynamic
environment that included pedestrians, traffic signals, and other vehicles. These
challengesinducedresearcherstocreatepracticalautonomousgroundvehicles,and
acted as a key springboard to the formation of the burgeoning self-driving car
industry. A decade removed from these challenges, this capability is now widely
expected to disrupt the global transportation sector, with the promise of reduced
vehicular fatalities, reduced transportation cost, and improved fuel economy.
Riding on the success of those prior events, DARPA initiated the DARPA
Robotics Challenge (DRC) in 2012 to advance research in autonomy and legged
locomotion. The DRC was motivated by the Fukushima nuclear disaster in Japan,
which was caused by an earthquake-induced tsunami in March 2011. The disaster
resulted in meltdowns in three nuclear reactors. Because of the high radiation
exposure risk, responders had a challenging time assessing the damage. This
dangerous situation was an ideal application for robots,and they were deployedin
multipleattemptstosurveythedamage.Unfortunately,duetotechnicallimitations,
the robotic disaster response was largely ineffectual. This experience exposed the
shortcomings of robots in confronting challenging, real-world disaster scenarios.
In response, DARPA initiated the DRC with the goal of creating robots that
could operate effectively in environments similar to the Fukushima disaster. The
emergency response task requires complex functions from a robot, including
operatinginanenvironmentdesignedforhumans(e.g.,oneincludingdoors,stairs,
andladders)thatmaybephysicallydegraded(e.g.,strewnwithrubblecausedbyan
earthquake, or suffering from water damage). Emergency response tasks may also
vii
viii Preface
requirerobotstooperatecomplextools,machines,mechanisms,orvehiclesinorder
to execute a disaster recovery task.
The DRC required robots to complete a scenario that included eight specific
tasksdesignedtomimicanenvironmenttheymightencounterinaFukushima-like
disaster site. These tasks included mounting and driving a vehicle, locomoting
across a rubble pile, clearing debris, opening a door and entering the doorway,
climbinganindustrial ladderandtraversing acatwalk, usingapowertooltobreak
through a concrete panel, locating a leaking pipe among numerous pipes and
closing a nearby valve, and replacing a cooling pump installed by fasteners
(DARPA, “DARPA Robotics Challenge Broad Agency Announcement,” DARPA
Tactical Technology Office (TTO), 2012). Original concept artwork depicted
multiple humanoid robots engaged in heavy work in a contaminated and partially
ruinedindustrial environment.To accomplishthese tasks,therobotswererequired
to be proficient in perception, semiautonomous and/or autonomous decision-
making, mounted mobility (e.g., driving a vehicle), dismounted mobility
(e.g., walking and climbing), and manipulation. Each of these would require
dramatic improvement over the state of the art. To ensure proficiency in autono-
mous tasks, the communication channel between the robot and teleoperators was
intentionally degraded.
Preparations for the final competition lasted 33 months with two intermediate
sub-challenges.Therobotsweregiven1htodriveanopen-airUtilityTaskVehicle
(UTV),egressfromthevehicle,openandmovethroughadoor,turnavalve,usea
powertooltocutaholeindrywall,completea“surprise”task,moveacrosseithera
rubble field or a debris-strewn passage, and then climb a set offour stairs. Details
of these tasks are provided in the first chapter of this book. By the day of the
competition, the DRC attracted considerable international interest. Twenty-five
teamshailingfromtheUSA,Germany,HongKong,Japan,China,andSouthKorea
competedinthechallenge.Ofthose,19wereabletosuccessfullycompleteatleast
one task.
Though the resultsof the DRC were promising, they highlighted the significant
gap between current state-of-the-art humanoid performance and the level of
performance required for real-world system operation. The competing robots
required nearly an hour to complete tasks that a healthy adult could complete in
under10min.Theyalsodisplayedkeyshortcomingsthatwouldpreventnear-term
deployment of humanoid robots for disaster response, most notably a common
inabilitytoperformtasksthatrequireexertionofsubstantialforce,anddifficultyin
performing tasks requiring whole-body coordination. Legged locomotion also
continues to be a challenge: while the systems were generally able to travel con-
fidently over flat ground, locomotion over uneven or deformable terrain proved
difficultorimpossible.Theprevalenceoffallsstoodinstarkcontrasttothescarcity
offall recovery techniques, as teams of humans were required to physically assist
robotsafterallbutonefall.Evidenceofsystemfrailtycanbefoundthroughoutthe
team’s technical reports, suggesting that operation in a real-world disaster envi-
ronment would pose a major challenge.
Preface ix
From a research standpoint, a stated goal of the DRC was to advance the state
of the art in supervised autonomy for mobile robots—specifically, those that must
interact with multiple, varied objects in complex environments (e.g., a valve, a
power tool, and a vehicle). The final competition featured only relatively modest
levels of autonomy, as teams opted to adopt more conservative teleoperation
strategies. Generally, only the lowest level, single-feature behaviors (such as
reachingtograspavalve)wereconductedwithoutdirectoperatorintervention.This
likely stemmed from the persistent technical difficulty of autonomy in complex
environments as well as the competition’s structure. For example, although com-
municationswiththehumanoidweredegraded,thedegradationwasstructuredand
predictable;therewasnolimitplacedonthenumberofoperators;andtherelatively
small number of possible tasks enabled direct training and practice for specific
activities. (The extent to which some teams tailored their development to specific
tasks is highlighted by their several reports of failures induced by late, minor
changes to some tasks).
LiketheresultsofthefirstDARPAGrandChallenge,thelimitationshighlighted
at the DRC suggest that this landmark event may represent only the first, halting
stepsonalongjourneytowardacoexistencewithroboticsandautonomoussystems
that previously resided only in the realm of science fiction.
Despite these limitations, the fact that 25 teams were able to field operational,
fully untethered humanoid robots capable of performing even a subset of the
requiredtasksinahuman-scalecompetitionenvironmentrepresentsmajorresearch
progress. Advances were made in diverse technical areas that will ensure a strong
legacy for the DRC. More specifically, the development of robotic hardware cap-
ableofsustainingrepeatedfallswhilebeingsufficientlynimbletodriveaUTVand
climbstairsisanimpressiveaccomplishment.Inaddition,numerousteamsreported
significant progresson advanced operatorcontrolmethods; othersintegrated novel
robots from predominately Commercial Off-The-Shelf components (COTS); while
others still developed highly innovative, fully original hardware designs.
Furthermore,theteams’widespreaduseandcreationofopen-sourcecodeopeneda
wealth of new software tools for the robotics community.
Thesocietalimpactofthecompetitionmustalsobenoted.Thefinalcompetition
eventheldonJune5–6,2015attheFairplexinPomona,California,USAwasopen
to the general public and covered extensively in the global media. Although some
of the most popular reports focused primarily on the robots’ amusing follies, the
eventcapturedthepublic’simaginationandfocusedattentiononthestateoftheart
in emerging humanoid robotics. Several reports noted that the robots won sub-
stantial empathy and concern from the crowd in attendance, perhaps helping to
counterthefearandnegativitythathasbeenhistoricallyprevalentinsomecultural
views of robots.
The purpose of this volume is to collect reports describing the technical
approaches of many of the competing teams in the DRC. The chapters outline the
technical approaches that enabled successes and the shortcomings that led to fail-
ures.Additionalchaptersprovideanoverviewofthecompetitionanditsresults,an
evaluation of the operator–robot interfaces employed by the teams, and some