This full day workshop to bring top researchers from China and the rest of world together to study the uncertainty issues in automation. Due to its booming manufacturing and other industry sectors, China can be viewed as the largest test bed for researchers in automation science and engineering. Automation has potential to improve quality (consistency), efficiency, safety, and cost for manufacturing, and has many other applications, such
as healthcare and security. However, a major issue for automation systems is how to effectively cope with uncertainty and dynamics arising in the physical and human environment. This workshop will provide a gathering ground for researchers with common interests in the uncertainty issues to share and develop recent progresses such as new models and methods for effectively modeling and resolving uncertainty in automation.
We would like to invite more researchers in this area to give a talk and/or
participate this exciting workshop.
The full call for proposal can
be downloaded here. The workshop can be registered through ICRA
conference registration website. If you are interested in presenting you
recent work please send the following items to us (email: email@example.com)
on or before March 1st,
1. Full contact info including affiliation
2. Proposed title (if there is any change)
3. A short (150 words) abstract addressing how the talk relates to
4. A URL to the home page describing related research or a related
(does not matter whether it is published elsewhere.)
In this talk I will provide a brief introduction to ensemble control theory, which is a new and useful way to deal with bounded uncertainty in dynamical systems. To steer one system with an uncertain parameter, we pretend to steer a continuous ensemble of systems, each with a particular value of that parameter. Despite the corresponding jump from a finite-dimensional to an infinite-dimensional configuration space, I will show how to establish controllability results and to derive practical motion planning algorithms based on these results. Hardware experiments will demonstrate the utility of this approach when applied to real robots.
Dr. Timothy Bretl
Assistant Professor of Aerospace Engineering Affiliate,
Coordinated Science Laboratory,
University of Illinois at Urbana-Champaign
Office: CSL 148
In this talk we explore the relationship between grasps and cages of a rigid body, in particular into the use of cages as waypoints to grasp an object. We introduce the concept of pregrasping cage, a caging configuration from which the object can be reached while maintaining the cage on it, and therefore guaranteeing that the grasp will be reached with absolute certainty. In the case of two-fingered manipulators, all cages are indeed pregrasping cages and, as a consequence, are useful waypoints to grasp an object. We will show in this talk that the same does not hold with more than two fingers. There are caging configurations from which a grasp of the object cannot be reached without breaking the cage on it. We explore the natural generalization of the squeezing/stretching characterization to the case of
n fingers and exploit it to give sufficient conditions for a cage to be a pregrasping cage
Mr. Alberto Rodriguez and Professor Matt Mason
The Robotics Institute,
Carnegie Mellon University,
5000 Forbes Avenue - Newell Simon Hall A519
15213 Pittsburgh, PA, USA
Manufacturing systems flexibility is critical to meet the challenges arising from the uncertain demands. In this talk, we propose a framework to evaluate the effect of utilizing flexibility in manufacturing systems which have the capability of producing multiple products at multiple plants, and to optimize to achieve good designs of capacity planning for the plants. A nonlinear polynomial stochastic programming model was proposed to describe the production planning problem. Analysis shows that a simplified version of the problem could also be NP-hard. Due to the uncertainty, the complexity and the large search space involved, we try to apply Ordinal Optimization (OO) method to our problem to search for good enough designs. To estimate the performance of the designs sampled following OO method, a Reformulation-Linearization/convexification Technique (RLT) could be applied, or the problem could be transformed to a MIP problem and solved with a MIP solving technique. The problem is based on practical background and could provide important insights for manipulating uncertainty in manufacturing systems.
Prof. Qianchuan Zhao,
Department of Automation,
Beijing 100084, China
In the field of Micro/Nano manipulation, the key feature is its performance in terms of accuracy, dexterity and dynamic character. The performance will be affected by the uncertain errors of the system caused by mechanical architecture parameters, manufacturing tolerances and clearances, installation, performance of driving actuators, dynamic model, the noises from the sensors, and the task uncertainty. To achieve high accurate characteristics, the architectural parameters of the manipulator should be designed and optimized carefully. The optimization of architectural parameters for parallel kinematic machine (PKM) can be performed with the particle swarm optimization (PSO) to find the global optimal solutions. Most micromanipulators are driven by either PZT or SMA. The actuators are always with high hysteresis and nonlinear characters. The successful compensation for hysteresis relies on the design of suitable control strategies. Typically, the hysteresis is compensated by feed-forward control resorting to an inverse of the hysteresis model such as the Preisach model, Duhem model, Maxwell model, and Bouc-Wen model, etc. Sliding mode control (SMC) strategy can be adopted since SMC is an effective and simple way to deal with model imperfection and uncertainties for nonlinear systems. Besides, sliding mode control with perturbation estimation (SMCPE) can be adopted as an enhanced version of the conventional SMC. Hence a suitable selection of control methods will be also very helpful to improve the performance of the whole system.
Yangmin Li, Ph.D., Professor
Department of Electromechanical Engineering,
Faculty of Science and Technology,
University of Macau,
Room: Choi Kai Yau Building N412
fax: (853) 28838314
Official Webpage: http://www.fst.umac.mo/en/staff/fstyml.html
Personal Webepage: http://www.sftw.umac.mo/~yangmin
There is significant potential for using aerial robots in the assembly and construction of structures, especially in aerial lifting or transport to hard-to-access
sites and in tasks requiring assembly of tall towers. In this talk we will explore the use of rotorcrafts in the assembly of three-dimensional structures, similar to
those involved in construction of scaffolds, tower cranes, skyscrapers and high-voltage towers. We will study the effects of positioning accuracy, part tolerance, and
process variation on the effectiveness of assembly and show how aerial robots can use force/moment sensing to adapt their behaviors to improve performance.
Prof. Vijay Kumar,
Departments of Mechanical Engineering & Applied Mechanics,
Computer & Information Science (Secondary Appointment),
School of Engineering & Applied Sciences,
University of Pennsylvania Tsinghua University,
Although mobile robots are increasingly being used in real-world
applications, the ability to robustly sense and interact with the
environment is still missing. A major challenge to the widespread
deployment of mobile robots is the ability to operate autonomously
despite the partial observability and uncertainty of real-world
domains. The proposed approach exploits the rich information encoded
in visual inputs to address these challenges, and makes the following
significant contributions: (a) a bootstrap learning scheme that uses
temporal, local and global visual cues to autonomously learn
hierarchical models of novel objects---the learned models are
incrementally refined and used to track these objects over time; and
(b) a novel hierarchical sequential decision-making scheme that
incorporates constrained convolutional policies and belief propagation
to automatically tailor visual sensing and information processing to
the task at hand. All algorithms are implemented and evaluated on
simulated and physical robots operating in dynamic environments.
Dr. Mohan Sridharan
Department of Computer Science,
Texas Tech University,
8th and Boston,
Lubbock, TX 79409
Our group works on autonomous systems that are designed to assist nature observation. We would like to bring automation into this dangerous, tedious and stressful task. Recently, we develop a new filter to assist the search for rare bird species. Since a rare bird only appears in front of a camera with very low occurrence (e.g., less than ten times per year) for very short duration (e.g., less than a fraction of a second), our algorithm must have very low false negative rate. We verify the bird body axis information with the known bird flying dynamics from the short video segment. Since a regular extended Kalman filter (EKF) cannot converge due to high measurement error and limited data, we develop a novel Probable Observation Data Set (PODS)-based EKF method. The new PODS-EKF searches the measurement error range for all probable observation data that ensures the convergence of the corresponding EKF in short time frame. The algorithm has been extensively tested using both simulated inputs and real video data of four representative bird species. In the physical experiments, our algorithm has been tested on rock pigeons and red-tailed hawks with 119 motion sequences. The area under the ROC curve is 95.0%. During the one-year search of ivory-billed woodpeckers, the system reduces the raw video data of 29.41TB to only 146.7MB (reduction rate 99.9995%).
Dezhen Song, Ph.D., Associate Professor
Department of Computer Science and Engineering,
Texas A&M University,
College Statin, TX 77843,
Official Webpage: http://faculty.cse.tamu.edu/dzsong
Imperfect sensor and imperfect controller cause both world uncertainty
and manipulator uncertainty in robot grasping control. Advanced grasp
planning algorithm will benefit greatly if we're able to achieve
accurate tracking of the grasp object as well as obtain an accurate
all-around knowledge of the system when the robot attempts a grasp.
This motivates our study of the G-SL(AM)^2 problem, in which two goals
are simultaneously pursued: object tracking relative to the robot hand
and estimation of parameters of the dynamic model. We view G-SL(AM)^2
problem as a filtering problem. Because of stick-slip friction and
collisions between the object and hand, suitable dynamic models
exhibit strong nonlinearities and jump discontinuities. This fact
makes Kalman filters (which assume linearity) and extended Kalman
filters (which assume differentiability) inapplicable, and leads us
develop a particle filter which was verified in a planar grasp test
rig. Besides visual sensory data, we also incorporated synthetic
tactile sensor data in the particle filter to combat the common visual
occlusion problem during grasping.
Ms. Li (Emma) Zhang, Ph.D. student
Department of Computer Science,
Rensselaer Polytechnic Institute
110 8th Street,
Troy, NY, 12180
Discrete manufacturing systems commonly produce streams of events -- parts arrive, machines start and finish, etc. With dozens of concurrent processes, these events do not have a well-defined order. Faults and other anomalies may manifest themselves in these data streams, but not be visible to even an expert observer.
This presentation describes a method for detecting anomalies in streams of event data, for systems which do not have a pre-defined formal model. Commonly-available information about the system is required as input to the method (e.g., which events are associated with which processes and resources). Since it is not known whether a formal model exists that can completely characterize the manufacturing system, the method builds a set of models, thereby allowing uncertainty about the "true" behavior of the system to persist through the process.
The method has been applied to a Ford machining line to find an anomaly associated with a gantry incorrectly waiting for a machine to become available.
Prof. Dawn Tilbury
Mechanical Engineering Department
Unversity of Michigan
2250 GGBrown, 2350 Hayward St.
Ann Arbor, MI 48109-2125 USA
Measuring the performance of intelligent systems situated across
different domains, for example in robotics and automation, requires
scientifically sound and practically meaningful metrics, and evaluation
methodologies. A key element in such performance measurement is the notion
of uncertainty. A reasonably good understanding of what uncertainties affect
system performance is an essential first step. Once the sources of
uncertainty are identified, it is possible to characterize uncertainty in
performance measurement via model-based techniques, statistically
significant results, and quantifiable performance data.
This presentation will draw from our experiences from evaluation exercises
and field testing on how to account for uncertainty of intelligent systems. In
conjunction, we will discuss relevant scenarios from our work in performance
evaluation, benchmarking, and standardization ranging from urban search and
rescue to manufacturing automation, and including field tests and robot
Dr. Raj Madhavan
Institute for Systems Research,
University of Maryland, College Park & National Institute of Standards
firstname.lastname@example.org or email@example.com
Ms. Elena Messina
National Institute of Standards and Technology
 E. Messina, "Stimulating Innovation: NIST Developing Performance
Standards for Response Robots," Unmanned Systems, May 2009.
 R. Madhavan, A.P. del Pobil, and E. Messina, "Performance Evaluation
and Benchmarking of Robotic and Automation Systems [TC Spotlight]", IEEE
Robotics and Automation Magazine, March 2010.
 R. Madhavan, R. Lakaemper, and T. Kalmar-Nagy, "Benchmarking and
Standardization of Intelligent Robotic Systems", Proceedings of the
International Conference on Advanced Robotics, June 2009.
(This is a joint work of Benedikt Hupfauf, Heiko Hahn, Leon Bodenhagen, Dirk Kraft, Norbert Kruger, and Justus Piater.)
Automatic picking of randomly distributed objects from a bin has been
denoted the "Holy Grail" in the world of robot automation. A particular
property of the bin-picking scenario (in contrast to most other
industrial robot applications) is that grasp errors are allowed to
occur: Usually bin-pickers fill feeding stations such that occasional
errors do not disturb the main process, since the feeding station has a
buffer of a number of objects which it provides to the assembly
process. Grasping errors are usually detected by haptic feedback. We
intend to utilize the huge amount of grasp data generated in industrial
bin-picking for grasp learning. This aim is achieved by using the novel
concept of grasp densities (Detry et al., 2010). Grasp densities can
describe the full variety of grasps that apply to specific objects using
specific grippers. They represent the likelihood of grasp success in
terms of object-relative gripper pose, can be learned from empirical
experience, and allow the automatic choice of optimal grasps in a given
scene context (object pose, workspace constraints, etc.). We will show
grasp densities extracted from empirical data in a real industrial bin
picking context, constituting (to our knowledge) one of the first
examples of grasp learning in industrial robotics.
Dr. Heiko Hahn
University of Innsbruck,
Institute of Computer Science,
6020 Innsbruck, Austria
(This is a joint work of Dr. Tamás Haidegger, Dr. Balazs Benyó and Prof. Zoltán Benyó.)
Surgical robots have been introduced to the operating rooms originally to provide the same
advantages as their industrial ancestors: faster and more accurate task execution. However,
even today, the most successful surgical robot system (the da Vinci) relies on the direct control
of the surgeon through master-slave teleoperation due to the uncertainties in the system and
the environment. On one hand, the human-in-the-loop control strategy allows for more flexible
hardware solutions, where the surgeon is entirely responsible for patient safety and adaptation
to unforeseen events. On the other hand, research and development is towards improved
machine intelligence and automation. The inherent precision of future image-guided surgical
systems has to be enforced for the benefit of the patients.
Due to the various sources of uncertainty-inherent inaccuracies, calibration, registration
procedures, system latencies, patient motion-validation and assessment of image-guided
robotic systems can be cumbersome, yet essential. Our group has been focusing on various
approaches to improve these systems and methods, to handle uncertainties, supporting the
development towards advanced, automated medicine.
Dr. Tamás Haidegger,
Assistant research fellow at the Laboratory of Biomedical Engineering,
Dept. of Control Engineering and IT,
Budapest University of Technology and Economics (BME IIT), Hungary
Visiting researcher at the Austrian Center for Medical Innovation and Technology (ACMIT)
We will consider uncertainty in the context of grasp planning and feeder
design. We study the computation of grasps that are insensitive to
inaccuracies of the grasping hand in the sense that immobilization is still
assured despite misplacements of the fingers.
We also review ideas for augmenting the design space of a feeder device to
cope with discrepancies between observed and modeled part behavior. Based on
these results we identify two fundamental questions related to uncertainty,
which will be the focus of a new project.
Dr. Frank van der Stappen, Assiocate Professor Center for Geometry, Imaging and Virtual Environments
Department of Information and Computing Sciences
PO Box 80089
3508 TB Utrecht
Considering that energy use in buildings represents more than 40% of global energy consumption and that humans spend 90% of the time indoors, technologies enabling smarter buildings can lead to significant reductions in greenhouse gas emissions, and produce a comfortable, efficient, safe, and secure environment. This is to be achieved through intelligent sensing, advanced automation, modern computing and communication technologies to efficiently operate, monitor, and maintain buildings.
A seemingly unrelated, but complementary and synergistic, research theme is the Smart Grid. It brings together information technology with the current electrical infrastructure to enable the integration and optimization of renewable generation (such as wind and solar) and plug-in electric vehicles; lead to significant increases in the efficiency and reliability of power network; and empower consumers to manage their energy usage and reduce costs in an environmentally friendly manner without compromising their lifestyle. Smart buildings will benefit from the Smart Grid, while Smart Grid needs smart buildings to fully realize its potential. We have a unique opportunity to develop technologies that integrate smart buildings and the Smart Grid.
In this talk, the above vision will be elaborated, supplemented by some specific topics worthy to be addressed. Recent activities at University of Connecticut will also be highlighted.
Prof. Peter B. Luh,
SNET Professor, Electrical & Computer Engineering,
University of Connecticut,
Storrs, CT 06269-2157,
Although automated systems based on CAD/CAM technologies are increasingly
used in modern manufacturing, the workpiece holding and localization process
is relatively untouched by modern technology. The lag in fixture automation
can be attributed largely to the complexity involved in the design of the
automation system. A primary task of the design is to automatically generate
the most reliable fixture configuration for a specific 3D workpiece, usually
among a vast set of feasible configurations.
This presentation describes research results on design and planning of
workpiece fixturing and localization systems. In the class of problem
considered, fixture locators and clamps are to be selected from a large set
of discrete candidate locations on the workpiece. The primary focus of the
research is the reliability of workpiece restraint and localization with
respect to the fixturing configuration and associated errors in fixturing.
In addition, other criteria are considered for fulfilling the complete
fixturing requirements, including force closure of passive forces. In the
presentation, analysis of the localization accuracy and the conditions of
passive force closure is described. An approach based on optimal design of
experiments to automated fixture design will be presented, with 2D and 3D
Prof. Michael Y. Wang, PhD, FASME, FHKIE, FIEEE
Professor, Dept. of Mechanical & Automation Engr.,
The Chinese University of Hong Kong,
Shatin, NT, Hong Kong
Tel: +852-26098487; Fax: +852-26036002
(This is a joint work of Ken Goldberg and Melissa Goldstein.)
Grasping of an object by a robot (or human) is complicated by fundamental uncertainty about the pose of the object, the pose of the gripper, the shape of the object, and the mechanics of friction, mass distribution, and surface motion. Researchers have developed a series of methods to bound grasp behavior given bounds on these properties.
We are exploring alternatives based on advances in parallel processing that allow parallel sampling of distributions of these properties to identify grasps that approximately optimize the likelihood of succeeding. In this initial work we consider push-grasp planning for the parallel jaws of the Willow Garage PR2 gripper. We sample distributions on relative pose, object shape, and center of mass, using conservative friction bounds to identify desirable push-grasps.
We present initial results and open questions.
Prof. Ken Goldberg
University of California, Berkeley
IEOR, EECS, and Center for New Media
Berkely, CA 94720
Ken Goldberg, Professor, IEOR and EECS College of Engineering and School of Information,
Univ. of California, Berkeley, CA 94720-1758, United States +1(510) 643-9565,
Homepage: http://goldberg.berkeley.edu email:
Dezhen Song, Associate Professor, Department of Computer Science and
Engineering, Texas A&M University, College Station, TX 77843, United
States +1(979)845-5464, Homepage: http://faculty.cs.tamu.edu/dzsong email: