Seminar

Professor Robin Murphy

Professor, Department of  Computer Science & Engineering, Texas A&M University

Time: 12:30 pm – 1:30 pm

Date: Friday, March 7

Location: BC 220

Abstract: This talk will review 30 years of disaster robotics, tracing its start from the 1995 Kobe Earthquake and the Oklahoma City bombing to the current state of the art.  Since 2001, disaster robotics have made a positive contribution to both research and society. Mini ground robots were first used for the immediate response phase of the 9/11 World Trade Center collapse, small unmanned aerial systems for assessing damage and needs of civilians have become common since Hurricane Harvey, and marine vehicles that routinely assist lifeguards on beaches have helped with mass casualty events such as the Syrian Refugee Crisis in Greece. Perhaps more surprisingly, robots of all types were readily adopted by civilians, not just governments, during the COVID pandemic. Most recently, exciting breakthroughs in computer vision and machine learning are enabling responders to rapidly make more informed decisions to better save lives and accelerate recovery. Based on personal involvement with over 30 disasters and analysis of the use of robotics in dozens more, three summative observations about AI and robotics have emerged. First, and foremost, disaster robotics remains a formative domain better suited for qualitative and field methodologies than traditional hypothesis-driven laboratory studies. Second, understanding the unique socio-technical attributes of the work domain is vital in identifying high-impact fundamental research topics. Third, disasters pose ethical challenges for responsible research and innovation. The talk concludes with personal suggestions for those interested in careers in research. 

Bio: Dr. Robin R. Murphy, Ph.D. (’92) and M.S. (‘89) in computer science and B.M.E. (‘80) from the Georgia Institute of Technology,  is the Raytheon Professor of Computer Science and Engineering at Texas A&M University and a director of the Center for Robot-Assisted Search and Rescue. Her research focuses on artificial intelligence, robotics, and human-robot interaction for emergency management. She is an AAAS, ACM, and IEEE Fellow, a TED speaker, and the author of over 400 papers and four books including the award-winning Disaster Robotics which captures her research deploying ground, aerial, and marine robots to over 30 disasters in five countries including the 9/11 World Trade Center, Fukushima, Hurricane Ian, and the Surfside collapse. Her contributions to robotics have been recognized with the ACM Eugene L. Lawler Award for Humanitarian Contributions and a US Air Force Exemplary Civilian Service Award medal.  Dr. Murphy has served on numerous professional and government boards, including the Defense Science Board and National Science Foundation, as well as the AI for the Benefit of Humanity prize committee.

Professor Calin Belta

Professor, Departments of Electrical Engineering and Computer Engineering, University of Maryland

Time: 12:30 pm – 1:30 pm

Date: Friday, February 21th

Location: BC 220

Abstract: In control theory, complicated dynamics such as systems of (nonlinear) differential equations are mostly controlled to achieve stability and to optimize a cost. In formal synthesis, simple systems such as finite state transition graphs modeling computer programs or digital circuits are controlled from specifications such as safety, liveness, or richer requirements expressed as formulas of temporal logic. With the development and integration of cyber-physical and safety-critical systems, there is an increasing need for computational tools for controlling complex systems from rich, temporal logic specifications, while ensuring safety. Recent works proposed computational efficient approaches for safety-critical control using Control Barrier Functions (CBF) and Control Lyapunov Functions (CLF). In this talk, I will show how these approaches can be extended to accommodate systems with high relative degrees, (partially) unknown dynamics, and temporal logic specifications, and to improve the feasibility of the associated optimization problems.

Bio: Calin Belta is the Brendan Iribe Endowed  Professor of  Electrical and Computer Engineering and Computer Science at the University of Maryland, College Park, which is also part of the Maryland Robotics Center (MRC) and the Institute for Systems Research (ISR).  His research focuses on making control and machine learning systems safe and Interpretable,  with particular emphasis on robotics and systems biology. Notable awards include the 2008 AFOSR YIP and the 2005 NSF CAREER. He is a Fellow of the IEEE.

Dr. Abhinav Verma

Assistant Professor, Departments of Electrical Engineering and Computer Science, Penn State University

Time: 12:30 pm – 1:30 pm

Date: Friday, January 31st

Location: BC 220

Abstract: Specifications in linear temporal logic (LTL) offer a simplified way of specifying tasks for policy optimization that may otherwise be difficult to describe with scalar reward functions. However, the standard Reinforcement Learning (RL) frameworks can be too myopic to find maximally satisfying policies. In this talk we will discuss eventual discounting, a value-function based proxy under which one can find policies that satisfy a specification with the highest achievable probability. To improve the efficiency of learning from specifications we combine eventual discounting with LTL-guided Counterfactual Experience Replay, a method for generating off-policy data from on-policy rollouts via counterfactual reasoning. Finally, we will discuss a mechanism for exploiting the compositionality of a LTL specification to provide formal guarantees on the behavior of learnt policies for reach-avoid tasks.

Bio: Dr. Verma Is an Assistant Professor in the Department of Computer Science and Engineering at The Pennsylvania State University. Previously, He was a postdoc at the Institute of Science and Technology (IST) Austria in the Henzinger Group. Before joining IST, He completed his PhD from the University of Texas at Austin advised by Prof. Swarat Chaudhuri. His research lies at the intersection of machine learning and formal methods, with a focus on building intelligent systems that are reliable, transparent, and secure. This work builds connections between the symbolic reasoning and inductive learning paradigms of artificial intelligence.

Dr. S. Farokh Atashzar

 Assistant Professor, Departments of Mechanical and Aerospace Engineering and Electrical and Computer Engineering, New York University

Time: 12:30 pm – 1:30 pm

Date: Friday, October 4th

Location: BC 220

Abstract: In this talk, Atashzar will present his team’s recent work towards achieving human-robot symbiosis. In the first part, he will discuss a new family of data-driven nonlinear control systems based on passivity theory to enable resilient and transparent physical human-robot interaction by accounting for the passivity-based energetic signature of human biomechanics when interacting with robot dynamics. In the second part, he will present the team’s recent efforts in AI for neural engineering to decode complex neural signals for (a) predicting human intention for the control of neurorobots and (b) assessment of human neuromusculoskeletal disability for closed-loop interfacing. This ranges from novel computational models and deep networks to wearable myographic sensors. He will discuss the practical limitations of non-invasive interfaces like high-density electromyography and how AI can help to address them. Finally, he will briefly introduce two new directions of his team: (a) how AI can support the design and fabrication of soft robots and (b) how flexible probabilistic models can enable generalizable human-to-robot skill transfer, overcoming limitations of existing probabilistic learning from demonstration approaches. The overall vision for the talk will be integrated human-robot systems that combine the best of human and machine capabilities for application in medical robotics and neurorobotics.

Bio: S. Farokh Atashzar is currently an Assistant Professor at the School of Engineering, New York University (NYU). He holds joint appointments between the Departments of Mechanical and Aerospace Engineering and Electrical and Computer Engineering, which he joined in August 2019. Previously, he was a postdoctoral scientist in the Department of Bioengineering at Imperial College London and was sponsored by the National Sciences and Engineering Research Council (NSERC) of Canada postdoctoral award. At NYU, he leads the Medical Robotics and Interactive Intelligent Technologies (MERIIT) lab, focusing on the intersection of robotics, control, and AI for broad applications in human-robot interaction and neural engineering. His research aims to achieve human-robot symbiosis at Physical, Cognitive, and Metacognitive levels by developing technical and technological means to bridge human intelligence and physics with machine cognition and kinodynamics. The outcome will be next-generation machines that augment and learn from rather than replace human skills, merging the best of human and machine capabilities. He has published ~90 journal papers, ~60 conference papers, and two book chapters. He received several awards, including the MathWorks Research Award, NSERC PDF award, and IEEE RAS RAL Distinguished AE award. His research is funded by NIH R01, and five NSF grants in addition to multiple industrial grants besides a $2M equipment grant. He currently serves as Associate Editor for IEEE Transactions on Robotics, IEEE Transactions on Haptics, and IEEE Robotics and Automation Letters. Atashzar is also the chair of the IEEE RAS Cluster for human-centered robotics. He is also the General Chair for the IEEE SPS PROGRESS diversity initiative.

Dr. Guoquan (Paul) Huang

Associate Professor, Mechanical Engineering (ME) and Computer and Information Sciences (CIS), University of Delaware

Time: 12:30 pm – 1:30 pm

Date: Friday, September 20

Location: BC 220

Abstract: As cameras and IMUs are becoming ubiquitous, visual-inertial systems for spatial perception, in analogy to biological visual-vestibular neural systems, hold great potential in a wide range of applications from extended reality (XR) to autonomous robots. While visual-inertial navigation systems (VINS), alongside with SLAM, have witnessed tremendous progress in the past decade, yet certain critical aspects in the design of visual-inertial systems remain poorly explored, hindering the widespread deployment of these systems in practice. In this talk, I will present some recent research efforts of my group on advancing the state of the art of visual-inertial sensing, navigation and perception, including consistent and efficient 3D motion tracking and scene understanding. Many of the codebases have been open sourced to promote visual-inertial systems, broadly benefiting the community.

Bio: Guoquan (Paul) Huang is an Associate Professor of Mechanical Engineering (ME) and Computer and Information Sciences (CIS) at the University of Delaware (UD), where he is leading the Robot Perception and Navigation Group (RPNG). He is also a Principal Scientist at Meituan. He was a Postdoctoral Associate (2012-2014) at MIT CSAIL (Marine Robotics) after receiving his PhD in Computer Science from the University of Minnesota. From 2003 to 2005, and was a Research Assistant (Electrical Engineering) at the Hong Kong Polytechnic University. His research interests focus on state estimation and spatial computing for robotics and XR, including optimal sensing, calibration, localization, mapping, tracking, perception, navigation and locomotion of autonomous robots and mobile devices. He has served as an Associate Editor for the IEEE Transactions on Robotics (T-RO), IEEE Robotics and Automation Letters (RA-L), and IET Cyber-Systems and Robotics (CSR), as well as the robotics flagship conferences (ICRA and IROS). Dr. Huang has received many honors and awards, including the 2015 UD Research Award (UDRF), 2015/2023 NASA DE Space Research Seed Award, 2016 NSF Research Initiation Award, 2018 Google Daydream Faculty Research Award, 2019 Google AR/VR Faculty Research Award, 2020 ARL SARA Award, 2022 Google AI Faculty Research Award, 2023 Meta Reality Labs Faculty Research Award. He was recognized among the AI 2000 Scholars (Top 100 in robotics) and World Top 2% Scientists. He was also the recipient of the ICRA 2022 Best Paper Award (Navigation), 2022 Best Paper Award of GNC Journal, and the Finalists of the ICRA 2024 Best Paper Award (Robot Vision), RSS 2023 Best Student Paper Award, ICRA 2021 Best Paper Award (Robot Vision), and RSS 2009 Best Paper Award.

Dr. Qiyu Sun

Professor, Department of Mathematics, University of Central Florida

Time: 12:30 pm – 1:30 pm

Date: Friday, September 13

Location: BC 220

Abstract: Taylor expansion and Fourier expansion have been widely used to represent functions.  The question to be discussed in this talk is whether there is some analog for nonlinear dynamic systems.  In particular, we consider Carlemen linearization and Carleman-Fourier linearization of nonlinear dynamic systems and show that the primary block of the finite-section approach has exponential convergence to the solution of the original dynamic system.

Bio: Qiyu Sun received the Ph.D. degree in mathematics from Hangzhou University, Hangzhou, China, in 1990. He is currently a Professor of mathematics with the University of Central Florida, Orlando, FL, USA.  His research interests include applied and computational harmonic analysis, optimal control theory, mathematical signal processing and sampling theory. Together with Nader Motee, he received the 2019 SIAG/CST Best SICON Paper Prize for making a fundamental contribution to spatially distributed systems theory.  He is on the editorial board of several journals, including Journal of Fourier Analysis and Applications, Frontiers in Signal Processing, and  Sampling Theory, Signal Processing, and Data Analysis.

Dr. Sarah Dean

 Assistant Professor, Computer Science Department, Cornell University

Time: 12:00 pm – 1:00 pm

Date: Friday, September 6

Location: BC 220

Abstract: Modern robotic systems benefit from machine learning and human interactions. In this talk, I will discuss recent and ongoing work on developing algorithmic foundations for learning with and from human interactions. I will start with motivation: a collaboration on building a robot for assistive feeding that adaptively asks for help. The first key algorithmic question is how to decide when to query a human expert. I will describe a recently developed interactive bandit algorithm with favorable regret guarantees. The second key question is how to learn dynamical models of human mental state, like cognitive load or boredom, from partial observations. I will describe a learning algorithm based on ideas from system identification that comes with sample complexity guarantees. This is based on joint work with Rohan Banerjee, Tapomayukh Bhattacharjee, Jinyan Su, Wen Sun, Yahya Sattar, and Yassir Jedra.

Bio: Sarah is an Assistant Professor in the Computer Science Department at Cornell. She is interested in the interplay between optimization, machine learning, and dynamics, and her research focuses on understanding the fundamentals of data-driven control and decision-making. This work is grounded in and inspired by applications ranging from robotics to recommendation systems. Sarah has a PhD in EECS from UC Berkeley and did a postdoc at the University of Washington.

Dr. Gregory J. Stein

 Assistant Professor, Department of Computer Science, George Mason University

Time: 12:30 pm – 1:30 pm

Date: Friday, October 25

Location: BC 220

Abstract: The next generation of service and assistive robots will need to operate under uncertainty, expected to complete tasks and perform well despite missing information about the state of the world or the future needs of itself and other agents. Many existing approaches turn to learning to overcome the challenges of planning under uncertainty, yet are often brittle and myopic, limiting their effectiveness. Our work introduces a family of model-based approaches to long-horizon planning under uncertainty that augments (rather than replaces) planning with estimates from learning, allowing for both high-performance and reliability-by-design.

In this talk, I will present a number of recent and ongoing projects that improve long-horizon navigation and task planning in uncertain home-like environments. First, I will discuss our recent developments that improve performance and reliability in unfamiliar environments—environments potentially dissimilar from any seen during training—with a technique we call “offline alt-policy replay,” which enables fast and reliable deployment-time policy selection despite uncertainty. Second, I will discuss “anticipatory planning,” by which our robot anticipates and avoids side effects of its actions on undetermined future tasks it may later be assigned; our approach guides the robot towards behaviors that encourage preparation and organization, improving its performance over lengthy deployments.

Bio: Greg is an Assistant Professor of Computer Science at George Mason University, where he runs the Robotic Anticipatory Intelligence & Learning (RAIL) Group and is the director of the GMU Autonomous Robotics Lab. His research, at the intersection of robotics, planning, and machine learning, is centered around developing representations for planning and learning that allow robots to better understand the impact of their actions, so that they may plan quickly, intelligently, and reliably in a dynamic and uncertain world. Before joining Mason, he received his PhD in 2020 from the MIT Department of Electrical Engineering and Computer Science and previously graduated summa cum laude from Cornell University with a B.S. in Applied and Engineering Physics. His work was a finalist for Best Paper at the 2018 Conference on Robot Learning, at which he was additionally awarded Best Oral Presentation.

Website: https://cs.gmu.edu/~gjstein/

Dr. Matei Ciocarlie

Associate Professor, Department of Mechanical Engineering, Columbia University

Time: 11:00 am – 12:00 pm

Date: Friday, April 26

Location: PA 466

Abstract: At long last, robot hands are becoming truly dexterous. It took advances in sensor design, mechanisms, and computational motor learning all working together, but we’re finally starting to see true dexterity, in our lab as well as others. This talk will focus on the path our lab took to get here, and questions for the future. From a mechanism design perspective, I will present our work on optimizing an underactuated hand transmission mechanism jointly the grasping policy that uses it, an approach we refer to as “Hardware as Policy”. From a sensing perspective, I will present our optics-based tactile finger, providing accurate touch information over a multi-curved three-dimensional surface with no blind spots. From a motor learning perspective, I will talk about learning tactile-based policies for dexterous in-hand manipulation and object recognition. Finally, we can discuss implications for the future: how do we consolidate these gains by making dexterity more robust, versatile, and general, and what new applications can it enable?

Bio: Matei Ciocarlie is an Associate Professor in the Mechanical Engineering Department at Columbia University, with affiliated appointments in Computer Science and the Data Science Institute. His work focuses on robot motor control, mechanism and sensor design, planning and learning, all aiming to demonstrate complex motor skills such as dexterous manipulation. Matei completed his Ph.D. at Columbia University in New York; before joining the faculty at Columbia, Matei was a Research Scientist and then Group Manager at Willow Garage, Inc., and then a Senior Research Scientist at Google, Inc. In these positions, Matei contributed to the development of the open-source Robot Operating System (ROS), and led research projects in areas such as hand design, manipulation under uncertainty, and assistive robotics. In recognition of his work, Matei was awarded the Early Career Award by the IEEE Robotics and Automation Society, a Young Investigator Award by the Office of Naval Research, a CAREER Award by the National Science Foundation, and a Sloan Research Fellowship by the Alfred P. Sloan Foundation.