ICRA 2017 Full-Day Workshop:
Recent Advances in Dynamics for Industrial Applications
Date: May 29th (Mon), 2017
Time: 08:30 - 17:10
Venue: Room 4811/4812, Sands Expo and Convention Center, Singapore
The benefits of using dynamic models in the trajectory generation, planning, and control of industrial robots is a widely accepted notion within the robotics research community. Even if the models are not completely accurate, when available they can still be used to advantage, and offer superior performance over non-model-based methods. Yet with only a few exceptions, such dynamic model-based methods are not widely used in today's industrial robots. Why? Are there computational or other fundamental limits that prevent their effectiveness, or are purely kinematics-based methods sufficient for today's industrial robot applications?
This workshop aims to explore these and other related questions, and to shed light on the gap between the state-of-the-art on dynamics-based methods for robot planning and control and the needs of the practitioner.
Our intended audience would include researchers and developers who are working in dynamics and are interested in solving real challenges in the industry using the state-of-the-art robotics techniques developed in the academic community. This workshop will demonstrate the progress in the robot-related industry (e.g., service robots, manufacturing, logistics), introduce main problems and challenges in real industry, and finally discuss the gap and its possible solutions while applying research outputs to the real industry.
Call For Contributions
Participants are invited to submit abstracts related to key challenges while applying dynamics techniques in real world industrial applications. Topics include but are not limited to: dynamics computation, motion planning, task planning, manipulation, and mechanism design. We invite submissions in the form of extended abstracts (up to 4 pages) following ICRA formatting guidelines. The abstracts will be reviewed by the organizers. Accepted contributions will be featured in a poster session and will be included in the workshop proceedings, which will be available at the workshop webpage. We encourage work-in-progress to be submitted and will take this into account in the review process.
Poster submission opening: Feb 15th, 2017. Please submit your abstracts to firstname.lastname@example.org with keyword "poster". Submission deadline (extended): Apr 1st, 2017. Submission Acceptance Notification: Apr 10th, 2017.
Please take one minute to fill in this questionnaire.
- Alin Albu-Schäffer (DLR-German Aerospace Center)
- David J. Braun (Singapore University of Technology and Design)
- Sébastien Briot (Centre National de la Recherche Scientifique)
- Wenjie Chen (FANUC)
- Torsten Kroeger (Stanford University / Google)
- Sergey Levine (UC Berkeley)
- Andreas Müller (Johannes Kepler University)
- Todd Murphey (Northwestern University)
- Quang-Cuong Pham (Nanyang Technological University)
- Paolo Rocco (Politecnico di Milano)
- Luis Sentis (University of Texas at Austin)
- Katsu Yamane (Disney Research)
Alin Albu-Schäffer graduated in electrical engineering at the Technical University of Timisoara, in 1993 and received the PhD in automatic control from the Technical University of Munich in 2002. Since 2012 he is the head of the Institute of Robotics and Mechatronics at the German Aerospace Center, which he joint in 1995 as a PhD candidate. Since 2013 he is a professor at the Technical University of Munich, holding the Chair for “Sensorbased Robotic Systems and Intelligent Assistance Systems” at the Computer Science Department.
He received several awards, including the IEEE King-Sun Fu Best Paper Award of the Transactions on Robotics in 2012 and 2014, several ICRA and IROS Best Paper Awards, the Eurobotics TechTransfer Award 2011 (first prize) and 2016 (third prize) as well as the DLR Science Award in 2007. He is an IEEE Fellow since 2016.
David J. Braun (Singapore University of Technology and Design)
- Title: Constrained Optimization for Robot Control Application
- Abstract: The benefit of using dynamic optimization for control of industrial robots has been recognized by the robotics community. However, many of the methods for dynamic optimization use heuristic treatment of physical constraints, limitations on the control inputs and constraints on the temporal aspect of the motion, common in practical application. In this talk, I will present our recent effort to develop a numerically efficient dynamic optimization method which takes physical constraints rigorously into account. The method complements a typical model-based planning approach with an on-line optimal constrained feedback controller. This makes it suitable for real-world application under imperfect model information.
David Braun is Assistant Professor at Singapore University of Technology and Design (SUTD). He received the Ph.D. degree in Mechanical Engineering from Vanderbilt University. Following that he was a postdoctoral research fellow at the University of Edinburgh, and a visiting researcher at the German Aerospace Center. Currently he leads the Dynamics & Control Laboratory at SUTD. His work spans fundamental aspects of dynamics, control and design of compliant robots. Application areas of interest include: (1) legged locomotion, (2) optimal control of variable impedance robots, and (3) analytical design of compliant actuators. He has received the 2013 IEEE Transactions on Robotics King-Sun Fu Memorial Best Paper Award and was a Scientific Program Co-Chair of the 2015 IEEE/RAS-EMBS International Conference on Rehabilitation Robotics.
Sébastien Briot (Centre National de la Recherche Scientifique)
- Title: Exploiting Dynamics Singularities for Increasing the Robot Capabilities
- Abstract: Dynamics singularities appear when the system of wrenches applied on the mechanism under control loses its rank (locally) and is no more able to sustain the efforts applied on the mechanism. Such types of singularities appear in several systems such as UAVs, ROVs, but above all, parallel robots.
During long years, it was thought that these singularities were impossible to cross for parallel robots and that they limited their workspace size, which was already much smaller than the workspace of serial robots. We showed that they are possible to cross if we define a proper trajectory respecting a criterion based on the analysis of the dynamic model degeneracy and if we design a proper controller. Our most recent results in the field will be introduced for fixed-base parallel robots. We will also briefly present some preliminary results obtained for the design and control of an aerial parallel robot whose dynamic reconfiguration is possible via singularity crossing.
Sébastien Briot received the B.S. and M.S. degrees in Mechanical Engineering in 2004 from the National Institute of Applied Sciences (INSA) of Rennes (France). Then, he began a PhD thesis, under the supervision of Prof. Vigen Arakelian, at the INSA of Rennes and received the PhD degree in 2007.
He worked at the Ecole de Technologie Supérieure of Montreal (Canada) with Prof. Ilian Bonev as a postdoctorate fellow in 2008. Since 2009, he is a full-time CNRS researcher at the LS2N (ex-IRCCyN Lab.) in Nantes (France). Since 2017, he is the head of the ARMEN research team at LS2N. His research fields concern the design optimization of robots and the analysis of their dynamic performance. He also studies the impact of sensor-based controllers on the robot performance. He is the author of more than 30 referred journal papers and 3 inventions.
Dr. Briot received the Award of the Best Ph.D. Thesis in Robotics from the French CNRS for year 2007. In 2011, he received two other awards: the Award for the Best Young Researcher from the French Region Bretagne and the Award for the Best Young Researcher from the French Section of the American Society of Mechanical Engineering (SF-ASME).
Wenjie Chen (FANUC)
- Title: Robotic Learning in Industrial Applications
- Abstract: The incapability of flexible automation has prohibited today’s robots from wide use in various industries. Also aggressive operation schedules and production quotas push most robots to work near their hardware design limits. Therefore, significant efforts are sought to make robots more intelligent and capable through the application of advanced control, learning, and optimization technologies. In this talk, we will present several examples where learning technologies at three different levels (motion control, motion planning, and decision making) have advanced to meet industrial objectives, such as vibration suppression, easy teaching from demonstration, bin picking, and failure prediction. We will also show some practices that we are doing at FANUC to bridge the gap between academia and industry.
Wenjie Chen received a B.Eng from Zhejiang University, China, in 2007, and a M.S. and a Ph.D. in Mechanical Engineering from UC Berkeley, USA, in 2009 and 2012, respectively. After his Ph.D., he stayed as a Post-Doctoral Scholar at UC Berkeley, working on intelligent control of robot manipulators, as well as exoskeleton design and control for brain-machine interfaces. In 2013, he joined Robot Laboratory, FANUC Corporation, Japan, where he currently leads the research development of concept controller for next generation robots, and the frontier robotic research collaboration with university. His current research interests include advanced control, sensing, and learning algorithms with applications to robots for advanced manufacturing.
Torsten Kroeger (Stanford University / Google)
- Title: Robot Manipulation: Real-time Motion Planning for Industrial Manipulators
- Abstract: Online and instantaneous robot motion generation is an important feature for robot motion control systems to let robots respond instantaneously to unforeseen events. An algorithmic concept that enables instantaneous changes from sensor-guided robot motion control (e.g., force/torque or visual servo control, distance control) to trajectory-following motion control, and vice versa, will be presented. The resulting class of on-line trajectory generation algorithms serves as an intermediate layer between low-level motion control and high-level sensor-based motion planning. Online motion generation from arbitrary states is an essential feature for autonomous hybrid switched motion control systems. It enables robotic systems to perform coordinated and deterministic motions in response to sensor signals. Samples and use-cases - including manipulation and human-robot interaction tasks - will accompany the talk in order to provide a comprehensible insight into this interesting and relevant field of robotics. [PDF]
Torsten Kroeger is Head of Robotics Software at Google X and a visiting researcher at Stanford University. He received his Master's degree in Electrical Engineering from TU Braunschweig, Germany, in 2002. From 2003 to 2009, he was a research assistant at Robotics Research Institute at TU Braunschweig, from which he received his Doctorate degree in Computer Science in 2009 (summa cum laude). In 2010, he joined the Stanford AI Laboratory, where he worked on instantaneous trajectory generation, autonomous hybrid switched-control of robots, and distributed real-time hard- and software systems. He is the founder of Reflexxes GmbH, a spin-off of TU Braunschweig working on the development of deterministic real-time motion generation algorithms. In 2014, Reflexxes has joined Google. Torsten is an editor or an associate editor of multiple IEEE conference proceedings, books, and book series. He received the IEEE RAS Early Career Award, the Heinrich Büssing Award, the GFFT Award, two fellowships of the German Research Association, and he was a finalist of the IEEE/IFR IERA Award and the euRobotics TechTransfer Award.
Sergey Levine (UC Berkeley)
- Title: Deep Robotic Learning
- Abstract: Deep learning methods have provided us with remarkably powerful, flexible, and robust solutions in a wide range of passive perception areas: computer vision, speech recognition, and natural language processing. However, active decision making domains such as robotic control present a number of additional challenges, standard supervised learning methods do not extend readily to robotic decision making, where supervision is difficult to obtain. In this talk, I will discuss experimental results that hint at the potential of deep learning to transform robotic decision making and control, present a number of algorithms and models that can allow us to combine expressive, high-capacity deep models with reinforcement learning and optimal control, and describe some of our recent work on scaling up robotic learning through collective learning with multiple robots. [PDF]
Sergey Levine received a BS and MS in Computer Science from Stanford University in 2009, and a Ph.D. in Computer Science from Stanford University in 2014. He joined the faculty of the Department of Electrical Engineering and Computer Sciences at UC Berkeley in fall 2016. His work focuses on machine learning for decision making and control, with an emphasis on deep learning and reinforcement learning algorithms. Applications of his work include autonomous robots and vehicles, as well as computer vision and graphics. His research includes developing algorithms for end-to-end training of deep neural network policies that combine perception and control, scalable algorithms for inverse reinforcement learning, deep reinforcement learning algorithms, and more.
Andreas Müller (Johannes Kepler University)
- Title: Optimal Control, Modelling, Identification and Calibration of Robotic Manipulators
- Abstract: Model-based control is becoming an integral part of the operation of robotic manipulators not only for specially tailored systems but for standard industrial robots. The aim to operate with maximal possible speed and at the same time with high accuracy calls for advanced optimal control schemes. As any model-based control scheme relies on an accurate kinematical and dynamical model, they must be accompanied with (desirably online and non-intrusive) geometric calibration and dynamic identification methods.
Dynamics modeling, despite being indispensable for motion planning and control, is still regarded as difficult topic. It will be shown that this is not the case and that the concept of screws and transformation groups provides a conceptually simple and computationally efficient setting for modeling industrial manipulators.
Model-based optimal control amounts to solving highly non-linear boundary value problems with non-linear equality and inequality constraints. To this end, tailored approaches to solve the optimal control problem are presented. In particular, point-point and trajectory following control problems are considered. Here again the geometric modeling yields an efficient formulation for the higher order constraints that are necessary to limit joint accelerations, jerk, jounce, etc. according to the technical limitations of the drives as well as for flatness-based control.
Andreas Müller is full professor and head of the Institute of Robotic at the Johannes Kepler University Linz, Austria. He obtained a diploma degree in mathematics at University Mittweida, Germany, a master degree in electrical engineering at University of Northumbria at Newcastle, UK (1998), and a diploma in mechanical engineering at the TU Chemnitz, Germany, where 2004 he also earned a PhD in mechanics. In 2008 he completed his habilitation at the University Duisburg-Essen, Germany. His work focuses on computational methods for efficient modeling and control of robotic systems, mechanism kinematics and singularities, mobile platforms, redundant serial and parallel kinematics manipulators and flexible lightweight robots.
Todd Murphey (Northwestern University)
- Title: Principles of Statistical Mechanics for Sharing Autonomy with Robotic Systems
- Abstract: Shared autonomy--the execution of tasks involving both people and robots--is increasingly important in applications ranging from manufacturing to rehabilitation. Static and quasistatic tasks are a challenging class of shared autonomy problems, and dynamic tasks are often even more challenging because they require tight synchronization in time. This talk will focus on software-enabled techniques that integrate autonomy and an operator during dynamic task execution. The software approaches combine numerical methods for dynamics and numerical methods for optimal control, subject to real-time constraints. Moreover, the software uses novel metrics for measuring and evaluating operator signals, with measures originating in statistical mechanics often providing unexpected insight into control approaches used by operators. I will also discuss case studies using these methods, including some arising from rehabilitation. [PDF]
Dr. Todd D. Murphey is an Associate Professor of Mechanical Engineering at Northwestern University. He received his B.S. degree in mathematics from the University of Arizona and the Ph.D. degree in Control and Dynamical Systems from the California Institute of Technology. His laboratory is part of the Neuroscience and Robotics Laboratory, and his research interests include computational methods for mechanics and real-time optimal control, physical networks, and information theory in physical systems. Honors include the National Science Foundation CAREER award in 2006, membership in the 2014-2015 DARPA/IDA Defense Science Study Group, and Northwestern's Charles Deering McCormick Professorship of Teaching Excellence. He is a Senior Editor of the IEEE Transactions on Robotics.
Quang-Cuong Pham (Nanyang Technological University)
- Title: Robotic manipulation with contact and dynamics
- Abstract: In this talk, I will survey a number of recent developments in our group (www.ntu.edu.sg/home/cuong) regarding robotic manipulation, in particular: kinodynamic motion planning, (initial steps in the) autonomous assembly of an IKEA chair, automatic precision drilling, manipulation of a rotating chain, etc. The common denominator of these tasks is the prominence of contact and dynamics. I will discuss the challenges we faced and the ideas we developed to tackle some of them.
Cuong was born in Hanoi, Vietnam and grew up in Vietnam and then in France. He graduated from École Normale Supérieure rue d'Ulm, France, in 2007. He obtained a PhD in Neuroscience from Université Paris VI and Collège de France in 2009. In 2010, he was a Visiting Researcher at the University of São Paulo, Brazil. From 2011 to 2013, he was as a Fellow of the Japan Society for the Promotion of Science (JSPS), doing research in Robotics at the University of Tokyo. He joined NTU as an Assistant Professor in 2013. He was the recipient of the Best Paper Award at the conference Robotics: Science and Systems, 2012. His team won the second prize at the Airbus Shopfloor Challenge at ICRA 2016.
Paolo Rocco (Politecnico di Milano)
- Title: Constraint-based reactive motion planning for industrial robot manipulators
- Abstract: Industrial robot manipulators are used today in a variety of applications that require interaction with an unstructured environment, including humans in the case of collaborative robots. Correspondingly, motion programming has become increasingly difficult, as new capabilities are required to the robots, like reactivity and adaptability. Purely pipelined planning and control architectures might turn out to be inadequate, as they require a large effort in defining what-if handling strategies to cope with each possible non-nominal behaviour at execution time. Next generation robot controllers should then transform application constraints into motion commands only at a lower and real-time level, where updated sensor information or other kind of events can be handled consistently with the higher level specifications.
In this talk, we will describe a methodology for motion specification and robust reactive execution. Traditional trajectory generation techniques and optimization-based control strategies are merged into a unified framework for simultaneous motion planning and control, accounting also for the dynamic properties of the robot. Applications in image-based tasks and in safe human-robot interaction will be discussed. [PDF]
Paolo Rocco is a full professor in automatic control and robotics at Politecnico di Milano, Italy, where he serves as Chair of the BSc and MSc Programs on Automation and Control Engineering. He is also a co-founder of Smart Robots, a spin-off company of Politecnico di Milano.
A Senior Member of IEEE, he has served in various positions in the Editorial Boards of journals and conferences. At present he serves as a Senior Editor for the IEEE Robotics and Automation Letters and as an Associate Editor for the IFAC journal Mechatronics.
He has been in charge of several research projects with industrial partners and public bodies. Currently his research interests concern a few aspects related to industrial robotics, with particular focus on safe and productive human-robot interaction. He is the author of about 150 papers in the areas of robotics, motion control, and mechatronics.
Luis Sentis (University of Texas at Austin)
- Title: Uncertainty in Human-Centered Robots
- Abstract: Uncertainty permeates in all control approaches and significantly complicates controller design. This is specially true for human-centered robots which rely on over-simplifications such as ignoring high-frequency behaviors or real-time delays to central computers. In this talk I join forces with two of my students to present detailed mathematical work on choosing structure for measuring uncertainty in a meaningful statistical sense, motivate the nature of uncertainty in hardware systems involving high performance series elastic actuators, and devise a planning and control framework that embraces uncertainty to external disturbances via reinforcement learning of locomotion responses.
Luis Sentis is an Associate Professor in Aerospace Engineering at the University of Texas at Austin and co-founder of Apptronik Systems. He received a Ph.D. in Electrical Engineering from Stanford University and was a La Caixa Foundation Fellow. He worked in Silicon Valley in the high-tech sector leading R&D projects in clean room automation. In Austin, he leads the Human Centered Robotics Laboratory, an experimental facility focusing on control and embodiment of humanoid robots. He writes extensively in areas related to real-time control of human-centered robots, design of high performance humanoid robots, and safety protocols in robotics. He was awarded the NASA Elite Team Award for his contributions to NASA’s Johnson Space Center Software Robotics and Simulation Division.
Katsu Yamane (Disney Research)
- Title: Teaching Animators to Work with Robots through Simulation
- Abstract: In this talk, I will discuss the lessons we learned from working with an animator to program an interactive quadruped robot using hand-crafted animations. In the entertainment industry, robots are often required to perform highly stylized motions to given the sense of life and personality. The robots are therefore usually programmed based on professionally created animations. Unfortunately, 3D animation software packages, which animators prefer to work with, are not capable of indicating whether an animation is physically reasonable. This can be especially problematic with legged robots because the ability to maintain balance has utmost importance in addition to motion style. To solve this problem, we developed a tool for animators to check the feasibility of an animation without affecting their normal pipeline. The tool also allows the animator to easily run physics simulation and visualize the result so that the animation can be adjusted to realize the creative intent on the robot.
Dr. Katsu Yamane is a Senior Research Scientist at Disney Research, Pittsburgh and an Adjunct Associate Professor at the Robotics Institute, Carnegie Mellon University. He received his B.S., M.S., and Ph.D. degrees in Mechanical Engineering in 1997, 1999, and 2002 respectively from the University of Tokyo, Japan. Prior to joining Disney in 2008, he was an Associate Professor at the University of Tokyo and a postdoctoral fellow at Carnegie Mellon University. Dr. Yamane is a recipient of King-Sun Fu Best Transactions Paper Award and Early Academic Career Award from IEEE Robotics and Automation Society, and Young Scientist Award from Ministry of Education, Japan. His research interests include humanoid robot control and motion synthesis, physical human-robot interaction, character animation, and human motion simulation.
- 08:30-08:45 Opening ceremony
- 08:45-09:10 Alin Albu-Schäffer, "Modelling, dynamics identification and model bases control for torque controlled robots in industrial co-worker scenarios"
- 09:10-09:35 David J. Braun, "Constrained Optimization for Robot Control Application"
- 09:35-10:00 Sébastien Briot, "Exploiting Dynamics Singularities for Increasing the Robot Capabilities"
- 10:00-10:30 Coffee break
- 10:30-10:55 Wenjie Chen, "Learning & Optimization for Robot Dynamics in Industrial Applications"
- 10:55-11:20 Luis Sentis, "TBD"
- 11:20-11:45 Sergey Levine, "Deep Robotic Learning"
- 11:45-12:30 Poster session
- Changsu Ha, Hackchan Kim, and Dongjun Lee, "Experiments on the Tracking and Vibration Suppression Control of Stage-Manipulator System on Flexible Structure"
- Claudio Gaz and Alessandro De Luca, "Collision Detection and Reaction when Holding an Unknown Payload" [slides][poster]
- Emanuele Magrini and Alessandro De Luca, "A Model-Based Approach for Contact Handling in Physical Human-Robot Collaborative Tasks"
- Janne Koivumäki and Jouni Mattila, "Recent Advances Towards High-Performance and Energy-Efficient Control of Hydraulic Robotic Manipulators" [slides][poster]
- Cristian Militaru, Ady-Daniel Mezei and Levente Tamas, "Lessons Learned from a Cobot Integration into MES"
- Rekha Raja and Swagat Kumar, "Advances in Motion Planning for Industrial Robot Manipulators" [slides]
- Taeyoon Lee and Frank C. Park, "The Use of Natural Distance Measure on Inertial Parameters for Ill-posed Multibody Dynamic Identification Problem"
- Youngsuk Hong, Jinkyu Kim, and Frank C. Park, "Comparative Analysis of Energy-Based Criteria for Dynamics-Based Robot Motion Optimization"
- Joonyoung Kim, Hyun-Kyu Lim, and Elizabeth A. Croft, "Online Time-Optimal Trajectory Planning and Control for Industrial Robots"
- 12:30-13:45 Lunch break
- 13:45-14:10 Andreas Müller, "Optimal Control, Modelling, Identification and Calibration of Robotic Manipulators"
- 14:10-14:35 Todd Murphey, "Principles of Statistical Mechanics for Sharing Autonomy with Robotic Systems"
- 14:35-15:00 Quang-Cuong Pham, "Robotic manipulation with contact and dynamics"
- 15:00-15:30 Coffee break
- 15:30-15:55 Paolo Rocco, "Constraint-based reactive motion planning for industrial robot manipulators"
- 15:55-16:20 Torsten Kroeger, "Robot Manipulation: Real-time Motion Planning for Industrial Manipulators"
- 16:20-16:45 Katsu Yamane, "Teaching Animators to Work with Robots through Simulation"
- 16:45 Closing remarks
Frank C. Park, Seoul National University
Frank Chongwoo Park received his B.S. in electrical engineering from MIT in 1985, and Ph.D. in applied mathematics from Harvard University in 1991. From 1991 to 1994 he was on the faculty at UC Irvine, and since 1995 he has been professor of mechanical and aerospace engineering at Seoul National University. His research interests are in robot mechanics, planning and control, vision and image processing, and related areas of applied mathematics. He has been an IEEE Robotics and Automation Society Distinguished Lecturer, and has held adjunct faculty positions at the NYU Courant Institute, the Interactive Computing Department at Georgia Tech, and the HKUST Robotics Institute. He is a Fellow of the IEEE, editor-in-chief of the IEEE Transactions on Robotics, developer of the EDX course Robot Mechanics and Control I, II, and co-author (with Kevin Lynch) of the book Modern Robotics: Mechanics, Planning, and Control (Cambridge University Press, 2017).