Abstract
The presentation will have two parts. First, recognition techniques in computer vision are beginning to work, making the next question, “What should we recognize?” I will present some work on increasing the label space for recognition toward large numbers of semantic labels embedded in a hierarchy, toward multiple attribute labels, and toward detailed spatial parsing. Predictions of these labels are improving results on problems from face recognition to large-scale similar image retrieval and building stronger connections between computer vision and natural language processing. The second part of the presentation will attempt to sketch out places were vision is failing to address the needs of application areas, including robotics.
Bio
Alex Berg is an assistant professor in the Department of Computer Science at the University of North Carolina at Chapel Hill. His research concerns computational visual recognition. Berg has worked on general object recognition in images, action recognition in video, human pose identification in images, image parsing, face recognition, image search, and machine learning for computer vision. He co-organizes the ImageNet Large Scale Visual Recognition Challenge, and has co-organized a series of workshops on large scale recognition in computer vision.
Before joining the Department of Computer Science in 2013, Berg was on the faculty at Stony Brook University, served as a research scientist at Columbia University, and was a research scientist at Yahoo! Research. While earning his Ph.D. at the University of California, Berkeley, he developed a novel approach to deformable template matching. Berg earned a B.A. and M.A. in mathematics from Johns Hopkins University and learned to race sailboats at the Severn Sailing Association in Annapolis.
]]>Abstract
Despite considerable progress in all aspects of machine perception, using machine vision in autonomous systems remains a formidable challenge. This is especially true in applications such as robotics, in which even a small error rate in the perception system can have catastrophic consequences for the overall system.
This talk will review a few ideas that could be used to start formalizing the issues revolving around integrating vision systems. They include a systematic approach to the problem of self-assessment of vision algorithms and predicting quality metrics on the inputs to the vision algorithms, ideas on how to manage multiple hypotheses generated from a vision algorithm rather than relying on a single “hard” decision, and methods for using external (non-visual) domain- and task-dependent information. These ideas will be illustrated with examples of recent vision for scene understanding, depth estimation, and object recognition.
Bio
Martial Hebert’s work is in the areas of computer vision and perception for autonomous systems. His interests are in the interpretation of perception data (both 2-D and 3-D), including building models of environments. Current research directions include:
Abstract
End users expect appropriate robot actions, interventions, and requests for human assistance. As with most technologies, robots that behave in unexpected and inappropriate ways face misuse, abandonment, and sabotage. Complicating this challenge are human misperceptions of robot capability, intelligence, and performance. This talk will summarize work from several projects focused on this human-robot interaction challenge. Findings and examples will be shown from work on human trust in robots, deceptive robot behavior, robot motion, robot characteristics, and interaction with humans who are blind.
Bio
Dr. Aaron Steinfeld is an associate research professor in the Robotics Institute (RI) at Carnegie Mellon University. He received his B.S.E., M.S.E., and Ph.D. degrees in Industrial and Operations Engineering from the University of Michigan and completed a postdoctoral appointment at U.C. Berkeley. He is the co-director of the Rehabilitation Engineering Research Center on Accessible Public Transportation (RERC-APT), director of the DRRP on Inclusive Cloud and Web Computing, and the area lead for transportation related projects in the Quality of Life Technology Center (QoLT). Steinfeld’s research focuses on operator assistance under constraints; i.e., how to enable timely and appropriate interaction when technology use is restricted through design, tasks, the environment, time pressures, and/or user abilities. His work includes intelligent transportation systems, crowdsourcing, human-robot interaction, rehabilitation, and universal design.
]]>Abstract
Many industrial robotic tasks involve moving a manipulator along a defined trajectory where both the path and the velocity are constrained to ensure process quality. Examples are processes like water-jet cutting, composite fiber placement, and robotic sealing. For high dynamic trajectories, a redundant manipulator consisting of a low bandwidth macro-manipulator and high bandwidth mini-manipulator can preserve the high dynamic trajectory requirements while maintaining the large work envelope of the macro manipulator. My talk presents an efficient global solution to the redundancy resolution problem, with the goal of minimizing the integral jerk of the actuators. Minimum jerk trajectories are advantageous for minimizing structural oscillations and improving the smoothness of control. The method presented achieves up to 75% reduction in integral jerk and 98% reduction in peak jerk vs. a greedy approach, while being embarrassingly parallel making it suitable for a high performance computing implementation. Additionally, a simple real-time controller is presented that maintains the planned low jerk trajectories given task-space feedback on the trajectory following error.
Bio
Phil Freeman is a technical fellow with Boeing Research and Technology. He works in the newly opened BR&T Technology Center in Charleston, South Carolina as part of the Advanced Production Systems group. His work is focused on developing advanced industrial robotics and automation for aerospace manufacturing. The Advanced Production Systems team includes Assembly Automation, Measurement Technology, Human Factors, and Lean Production Systems.
]]>Abstract
In this talk, I will describe our progress in developing distributed control policies to achieve rendezvous by a set of robots, even when some robots in the system do not follow the prescribed policy. These nonconforming robots correspond to faults in the multi-robot system, and our control policies are thus fault-tolerant policies. We consider the case in which each robot is an autonomous decision maker that is anonymous (i.e., robots are indistinguishable to one another), memoryless (i.e., each robot makes decisions based upon only its current information), and dimensionless (i.e., collision checking is not considered). Each robot has a limited sensing range, and is able to directly estimate the state of only those robots within that sensing range, which induces a network topology for the multi-robot system. We assume that it is not possible for the fault-free robots to identify the faulty robots (e.g., due to the anonymous property of the robots). Our approach provides an efficient computational framework and analysis of algorithms that achieve approximate convergence in the face of faulty robots under a few assumptions on the network topology and sensing sensing abilities.
Bio
Seth Hutchinson received his Ph.D. from Purdue University in 1988. In 1990, he joined the faculty at the University of Illinois in Urbana-Champaign, where he is currently a professor in the Department of Electrical and Computer Engineering, the Coordinated Science Laboratory, and the Beckman Institute for Advanced Science and Technology. He served as associate department head of ECE from 2001 to 2007. Hutchinson currently serves on the editorial boards of the International Journal of Robotics Research and the Journal of Intelligent Service Robotics, and chairs the steering committee of the IEEE Robotics and Automation Letters. He was Founding Editor-in-Chief of the IEEE Robotics and Automation Society’s Conference Editorial Board (2006-2008), and Editor-in-Chief of the IEEE Transaction on Robotics (2008-2013). He has published more than 200 papers on the topics of robotics and computer vision, and is coauthor of the books Principles of Robot Motion: Theory, Algorithms, and Implementations, published by MIT Press, and Robot Modeling and Control, published by Wiley. Hutchinson is a Fellow of the IEEE.
]]>Abstract
As computer vision and natural language processing techniques are maturing, there is heightened activity in exploring the connection between images and language. In this talk, I will present several recent and ongoing projects in my lab that take a new perspective on problems like automatic image captioning, which are receiving a lot of attention lately. In particular, I will start by describing a new methodology for evaluating image-captioning approaches. I will then discuss image specificity — a concept capturing the phenomenon that some images are specific and elicit consistent descriptions from people, while other images are ambiguous and elicit a wider variety of descriptions from different people. Rather than think of this variance as noise, we model this as a signal. We demonstrate that modeling image specificity results in improved performance in applications such as text-based image retrieval. I will then talk about our work on leveraging visual common sense for seemingly non-visual tasks such as textual fill-in-the-blanks or paraphrasing. We propose imagining the scene behind the text to solve these problems. The imagination need not be photorealistic; so we imagine the scene as a visual abstraction using clipart. We show that jointly reasoning about the imagined scene and the text results in improved performance of these textual tasks than reasoning about the text alone. Finally, I will introduce a new task that pushes the understanding of language and vision beyond automatic image captioning — visual question answering (VQA). Not only does it involve computer vision and natural language processing, doing well at this task will require the machine to reason about visual and non-visual common sense, as well as factual knowledge bases. More importantly, it will require the machine to know when to tap which source of information. I will describe our ongoing efforts at collecting a first-of-its-kind, large VQA dataset that will enable the community to explore this rich, challenging, and fascinating task, which pushes the frontier towards truly AI-complete problems.
Bio
Devi Parikh is an assistant professor in the Bradley Department of Electrical and Computer Engineering at Virginia Tech (VT) and an Allen Distinguished Investigator of Artificial Intelligence. She leads the Computer Vision Lab at VT and is also a member of the Virginia Center for Autonomous Systems (VaCAS) and the VT Discovery Analytics Center (DAC).
Prior to her current appointments, Parikh was a research assistant professor at Toyota Technological Institute at Chicago (TTIC), an academic computer science institute affiliated with University of Chicago. She has held visiting positions at Cornell University, University of Texas at Austin, Microsoft Research, MIT, and Carnegie Mellon University. She received her M.S. and Ph.D. degrees from the Electrical and Computer Engineering Department at Carnegie Mellon University in 2007 and 2009, respectively. She received her B.S. in electrical and computer engineering from Rowan University in 2005.
Parikh’s research interests include computer vision, pattern recognition, and AI, particularly visual recognition problems. Her recent work involves leveraging human-machine collaboration for building smarter machines and exploring problems at the intersection of vision and language. She has also worked on other topics such as ensemble of classifiers, data fusion, inference in probabilistic models, 3D reassembly, barcode segmentation, computational photography, interactive computer vision, contextual reasoning, hierarchical representations of images, and human debugging.
Parikh has received numerous awards and honors, including a National Science Foundation Graduate Research Fellowship, the 2011 Marr Best Paper Prize awarded at the International Conference on Computer Vision (ICCV), and the 2014 Army Research Office Young Investigator Program (YIP) award. The recipient of three Google Faculty Research Awards, Parikh was named an Outstanding New Assistant Professor by the College of Engineering at Virginia Tech in 2015.
]]>Abstract
Using a group of robots in place of a single complex robot to accomplish a task has many benefits, including simplified system repair, less down time, and lower cost. Combining heterogeneous groups of these multi-robot systems allows addressing multiple subtasks in parallel, reducing the time it takes to address many problems, such as search and rescue, reconnaissance, and mine detection. These missions demand different roles for robots, necessitating a strategy for coordinated autonomy while respecting any constraints the environment may impose. Synthesis of control policies for heterogeneous multirobot systems is particularly challenging because of inter-robot constraints such as communication maintenance and collision avoidance, the need to coordinate robots within groups, and the dynamics of individual robots.
I will present approaches to synthesizing feedback policies for navigating groups of robots in constrained environments. These approaches automatically and concurrently solve both the path planning and control synthesis problems, and are specified at a high level, for example, using an iPad interface to navigate a complex environment with a team of UAVs. I will also present some preliminary work on novel approaches to developing controllers for many types of multirobot tasks, by using crowdsourced multi-player game data.
Bio
Nora Ayanian is a Gabilan Assistant Professor of Computer Science at University of Southern California. Her research focuses on creating end-to-end solutions for coordinating teams of robots that start from truly high-level specifications and deliver code for individual robots in the team, such as using simple multitouch inputs to control a team of UAVs. Ayanian brings a unique approach to multirobot systems, creating unified solutions that address task assignment, path planning, and control that are broadly applicable across all aspects of multirobot systems and mobile sensor networks. Her solutions provide guarantees of convergence and safety on real robotic systems. She won the Best Student Paper Award at ICRA 2008 for her work on constrained multirobot control. In 2013 she was named one of IEEE Intelligent Systems “AI's 10 to watch” and in 2014 was named to NerdScholar’s “40 Under 40: Professors who Inspire.” In 2015, she was honored to be selected for the inaugural Mic 50, Mic.com’s list of 50 influential young people. Ayanian is a co-founder and current co-chair of the IEEE Robotics and Automation Society Technical Committee on Multi-Robot Systems, and co-produced the first ICRA trailer in 2015.
]]>