Past Events

Title: Human-robot collaborative manipulation

Abstract: Research at the Center for Robotics and Biosystems at Northwestern University includes bio-inspiration, neuromechanics, human-machine systems, and swarm robotics, among other topics.  In this talk, I will give an overview of our work on manipulation, including autonomous robot manipulation, neuroprosthetics to restore human manipulation capability, and human-robot collaborative manipulation.

Biography: Kevin Lynch is a professor of mechanical engineering and director of the Center for Robotics and Biosystems at Northwestern University. His research is on robotic manipulation, locomotion, human-robot systems, and robot swarms. He is the Editor-in-Chief of the IEEE Transactions on Robotics, former Editor-in-Chief of the IEEE Conference on Robotics and Automation Conference Editorial Board, and coauthor of three textbooks on robotics and mechatronics the instructor of six Coursera online courses forming the Modern Robotics specialization. He received a B.S.E. degree in electrical engineering from Princeton University and a Ph.D. in robotics from Carnegie Mellon University.

Image-guided Histotripsy, a novel non-invasive platform for future procedure automation

Abstract:

Histotripsy uses very high amplitude and short pulses (microseconds) of focused ultrasound to induce and control acoustic cavitation in the form of a histotripsy “bubble cloud.” The negative pressure, which can exceed -25MPa in the focal zone, allows the rapid formation and collapse of nano and microbubbles (within the bubble cloud) derived from endogenous gases naturally present in the targeted tissue. As the bubbles within the cloud form and collapse in microseconds, creating mechanical forces strong enough to destroy tissue at cellular and sub-cellular levels without the need for ionizing or thermal energy. An image-guided robotic platform is used to deliver the therapy. HistoSonics is currently focused on developing the broad platform capability to non-invasively treat across the body, initially focused on significant unmet needs in the abdominal and liver. The company is also researching and developing novel image-guided techniques to allow further procedure simulation, planning, and automation over setup and therapy delivery.  This talk will focus on recent research in these areas.

Bio:

Dr. Josh Stopek has 20+ years of R&D leadership experience, including a background in developing minimally invasive image-guided technologies, new cancer diagnostic and therapeutic platforms, biomaterial/biosurgery, and combination devices to market. He is currently the Vice President of R&D for HistoSonics, focused on new breakthrough oncology solutions. He formerly led R&D in various business areas at Medtronic, Covidien, and US Surgical. Before that, he was the co-founder and VP of a startup medical device company, VMSI, working on new minimally invasive and tissue regeneration therapies. Dr. Stopek has over 200 issued and pending patents. He received his B.S., M.S., and Ph.D. in Materials Science and Engineering from the University of Florida, where he also completed a fellowship in Neurosurgery and Neuroscience.

Title: The Robotics Part of Micro and Nano Robots

Abstract: Micro and nanorobots have made great strides since becoming a focused research topic over two decades ago. Much of the progress has been in material selection, processing, and fabrication, and paths forward in developing clinically relevant biocompatible and biodegradable micro and nanorobots are becoming clear. Our group, as well as others, maintain that using biocompatible magnetic composites with externally generated magnetic fields and field gradients is perhaps closest to clinical application. One of the most challenging aspects of the field is the development of the magnetic navigation system (MNS) that generates the fields and field gradients needed for microrobot locomotion. In this talk, I will present an overview of MNSs and show how these systems are fundamentally robotic in the way they must be designed and controlled. Decades of work in robotic manipulation can be brought to bear on this problem as we bring MNS technology to the clinic. I will also look at recent efforts in creating more intelligent micro and nanorobots that exhibit increasingly complex behaviors, some of which can even be programmed in situ. The field appears to be on the cusp of realizing the fantastic voyage.

Bio: Brad Nelson is the Professor of Robotics and Intelligent Systems at ETH Zürich and has recently become the Chief Scientific Advisor of Science Robotics. He has over thirty years of experience in the field and has received several awards in robotics, nanotechnology, and biomedicine. He serves on the advisory boards of several academic departments and research institutes across North America, Europe, and Asia. Prof. Nelson has been the Department Head of Mechanical and Process Engineering at ETH twice, the Chairman of the ETH Electron Microscopy Center, and a member of the Research Council of the Swiss National Science Foundation. He also serves on the boards of three Swiss companies and is a member of the Swiss Academy of Engineering (SATW). Before moving to Europe, Nelson worked as an engineer at Honeywell and Motorola and served as a United States Peace Corps Volunteer in Botswana, Africa. He has also been a professor at the University of Minnesota and the University of Illinois at Chicago.

Title: Decentralized allocation of tasks to agents and robots

Speaker: Maria Gini, CSE Distinguished Professor, Distinguished University Teaching Professor, Department of Computer Science and Engineering

Abstract: Task allocation and distributed decision-making are common in computer science, robotics, and many related fields.  Most algorithms solve the problem in a centralized way as an optimization problem.  I am interested in solving task allocation problems with no central authority or only a minimalistic central authority, such as an auctioneer. The methods I have developed work for physical agents that move in 2D space, such as robots or delivery vehicles, and methods for virtual agents. I am especially interested in methods that work with temporal and precedence constraints or for tasks that have costs that increase over time. In this talk, I will present examples of task allocation problems in multiple contexts, some requiring planning in advance and some requiring decisions in real time.

Bio: Maria Gini is a College of Science & Engineering Distinguished Professor in the Department of Computer Science and Engineering at the University of Minnesota.  She works on decision-making for autonomous agents in many application domains, ranging from swarm robotics to distributed methods for the allocation of tasks, methods for robots to explore an unknown environment, and navigation in dense crowds.  She is a Fellow of the AAAI, ACM, and IEEE.  She has published more than 60 journal articles and more than 300 conference papers, and book chapters.  She is Editor in Chief of Robotics and Autonomous Systems and is on the editorial board of numerous journals, including Autonomous Agents and Multi-Agent Systems, Current Robotics Reports, and Integrated Computer-Aided Engineering.

Developing, Designing, and Deploying Model Predictive Control for Satellite Rendezvous and Proximity Operations

Asst. Prof Chris Petersen, Department of Mechanical & Aerospace Engineering, University of Florida

2:30 PM on 2022-11-04

Bio: Dr. Christopher "Chrispy" Petersen is an Assistant Professor at the University of Florida in the Mechanical & Aerospace Engineering Department as of Fall 2023. He leads the Spacecraft Technology And Research (STAR) Laboratory, which is focused on 4 pillars of research; 1) Exploring & exploiting spacecraft dynamics, 2) Advanced guidance, navigation, control, and autonomy (GNCA), 3) Real-time, computationally aware optimization for spacecraft and 4) Immersive human-satellite interfaces. While he enjoys everything in space, his group's research focuses primarily on rendezvous, proximity operations, and docking (RPOD) and eXtra GEOstationary (XGEO, which is above Geostationary orbit to the Moon, and beyond). Before that, he was at the Space Vehicles Directorate of the U.S. Air Force Research Laboratory (AFRL/RV) at Kirtland Air Force Base in New Mexico. He received his B.S. from Syracuse University in Aerospace Engineering in 2012 and his M.S. and Ph.D. from the University of Michigan in 2016 in Spacecraft Dynamics & Control. While at AFRL he worked on 10+ satellite experiments, developing, deploying, and executing GNCA algorithms for ground and on-orbit use. As a highlight, Dr. Petersen was the PI for advanced autonomous guidance algorithms used by the Mycroft flight experiment, which has been recognized as "...the AF's biggest game changer" for space warfighters. He also served as Deputy Program Manager of the Autonomous Demonstrations and Orbital eXperiments (ADOX) Portfolio, a series of satellite demonstrations focused on autonomy technologies to enable satellite inspection, XGEO space domain awareness, and logistics in GEO, including advanced propulsion and refueling. For his accomplishments, in 2021, he was awarded the AFRL Early Career Award.

Abstract: The process of developing, designing, and deploying algorithms in relevant environments is one that couples theory and application tightly. This is especially prevalent in satellite navigation and control. Deploying these methods requires rigorous theory to ensure confidence that the algorithm will not cause damage to a multi-million dollar asset. However, such methods may require repeated execution in seconds to minutes with minimum memory impact. Such considerations are important to the United States Space Force (USSF) as it strives to remain at the forefront of technology development for future satellite architectures.

This talk will discuss the process of developing, designing, and deploying algorithms, in particular Model Predictive Control (MPC), for satellite rendezvous and proximity operations (RPO). These RPO missions must facilitate numerous requirements. Firstly, the ability to operate in unknown, communication limited environments, such as in geostationary orbit and beyond. Secondly, these algorithms must enable precise, time-critical maneuvering and replanning for missions like on-orbit assembly and manufacturing. Some key aspects that will be highlighted throughout the talk are (1) understanding and exploiting underlying spacecraft dynamics, (2) how to design algorithms to meet mission needs, and (3) steps to take to ensure that algorithms can be maximized for satellite and operator use. In the context of MPC, this will entail developing the optimization problem to ensure stability, robustness, and recursive feasibility while tailoring the methods for real-time feedback control and planning.

Associate Professor, Department of Electrical & Computer Eng.  

Title: New Quantitative Bounds on Recursive Stochastic Algorithms with Applications to Reinforcement Learning and System Identification.

Abstract: Recursive stochastic algorithms are pervasive in machine learning applications, system identification, and control. Examples include stochastic gradient descent, Q-learning, recursive least-squares, and Langevin algorithms. In reinforcement learning, control, and system identification applications, measurements have dependencies across time that make the quantitative analysis of the associated algorithms more challenging. In this talk, we describe how a wide variety of these algorithms can be analyzed in a unified framework known as L-mixing. The theory of L-mixing processes quantifies how statistical dependencies in a time series decay over time. Our first result will be to show how in scenarios, if the system generating the data is stable, then the data will be L-mixing. From here, we will show several algorithms from machine learning, and system identification can be cast into this L-mixing framework. Finally, we will show the L-mixing properties lead to clean, quantitative bounds on the convergence of these algorithms. 

Infinite Dimensional Direct Adaptive Control and Quantum Information Systems

Prof Mark Balas, Mechanical Engineering, Texas A & M

2:30 PM on 2022-10-21

Bio: Mark Balas is the Leland Jordan Professor in the Mechanical Engineering Department at Texas A&M University. He was formerly the Guthrie Nicholson Professor of Electrical Engineering and former Head of the Electrical and Computer Engineering Department at the University of Wyoming. He has the following technical degrees: Ph.D. in Mathematics, MS in Electrical Engineering, MA in Mathematics, and BS in Electrical Engineering. He has held various positions in industry, academia, and government. Among his careers, he has been a university professor for over 40 years with RPI, MIT, University of Colorado-Boulder, University of Wyoming, and Embry-Riddle Aeronautical University and has mentored 45 doctoral students. He has over 350 publications in archive journals, refereed conference proceedings, and technical book chapters. He has been visiting faculty with the Institute for Quantum Information and the Control and Dynamics Division at the California Institute of Technology, the US Air Force Research Laboratory-Kirtland AFB, the NASA-Jet Propulsion Laboratory, the NASA Ames Research Center, and was the Associate Director of the University of Wyoming Wind Energy Research Center and adjunct faculty with the School of Energy Resources. He is a life fellow of the AIAA, a life fellow of the IEEE, and a fellow of ASME. But if he ever becomes famous, it will be because he is the father of the Denver drum and bass DJ known as Despise, who is his daughter Maggie.

Abstract: Many control systems are inherently infinite dimensional when they are described by partial differential equations. There is renewed interest in controlling these systems, especially in the quantum information field. Since the dynamics of these systems will not be perfectly known, it is especially of interest to control these systems adaptively and autonomously via low-order finite-dimensional controllers. In our work, we have developed a direct model reference adaptive control and disturbance rejection with very low-order adaptive gain laws for infinite-dimensional systems on Hilbert spaces.

Quantum Information Systems are fundamentally infinite-dimensional. And the basic operations that can be performed on quantum systems to manipulate information are unitary quantum gates. Because of the nature of entanglement at the quantum level, these gates suffer from decoherence and cannot operate in a fully unitary way. It is also quite difficult to perform experiments identifying all the parametric data needed to create precise models of a particular quantum system. Instead, direct adaptive control that is suited to infinite dimensional systems could provide a reduction in the decoherence and allow the quantum gates to function in a more idealized unitary way.

This talk will focus on the effect of infinite dimensionality and some of the issues in controlling quantum information systems. The topics here may sound esoteric, but I hope to give you a version of them that will be reasonably accessible and will still remain as exciting and attractive to you as they are to me.

Karthik Desingh is an Assistant Professor in the Department of Computer Science and Engineering at the University of Minnesota. Karthik is associated with the Minnesota Robotics Institute (MnRI). He recently completed his Postdoctoral position at the University of Washington (UW). Before joining UW, he received his Ph.D. in Computer Science and Engineering from the University of Michigan. During his Ph.D., he was closely associated with the Robotics Institute and Michigan AI. He researches the intersection of robotics, computer vision, and machine learning, primarily focusing on providing perceptual capabilities to robots using deep learning and probabilistic techniques to perform manipulation tasks in unstructured environments. His work has been awarded the best workshop paper award at RSS 2019 and nominated as a finalist for the best systems paper award at CoRL 2021. He is serving as an Associate Editor for IROS 2022.

Title: Explicit and Implicit Object Representations for Robust and Generalized Perception in Robotics

Abstract: My long-term goal is to build general-purpose robots that can care for and assist the aging and disabled population by autonomously performing various real-world tasks. To robustly execute various tasks, a general-purpose robot should be capable of seamlessly perceiving and manipulating various objects in our environment. To achieve a given task, a robot should continually perceive the state of its environment, reason with the task at hand, and plan and execute appropriate actions. In this pipeline, perception is largely unsolved and one of the more challenging problems. Common indoor environments typically pose two main problems: 1) inherent occlusions leading to unreliable observations of objects and 2) the presence and involvement of a wide range of objects with varying physical and visual attributes (i.e., rigid, articulated, deformable, granular, transparent, etc.). Thus, we need algorithms that can accommodate perceptual uncertainty in the state estimation and generalize to a wide range of objects.

In my research, I develop 1) probabilistic inference methods to estimate the world-state with the notion of uncertainty and 2) data-driven methods to learn object representations that can generalize the state estimation to a wide range of objects. This talk will highlight some of my research efforts in these two research thrusts. In the first part of the talk, I will briefly describe an efficient belief propagation algorithm - Pull Message Passing for Nonparametric Belief Propagation (PMPNBP) - for estimating the state of articulated objects using a factored approach. Here, objects are modeled explicitly with representations that aid in robust state estimation in cluttered settings. In the second part of the talk, I will describe - Spatial Object-centric Representation Network (SORNet) - for learning object-centric representation grounded for sequential manipulation tasks. Here, objects are modeled implicitly as learned embeddings to aid generalization in perception. I will also discuss the open research problems on these thrusts toward realizing general-purpose domestic robots.

Assistant Professor, Department of Computer Science & Engineering

Title: Deep Image Prior (and Its Cousin) for Inverse Problems: the Untold Stories.

Abstract: Deep image prior (DIP) parametrizes visual objects as outputs of deep neural networks (DNNs); its cousin neural implicit representation (NIR) directly parametrizes visual objects as DNNs. These stunningly simple ideas, when integrated into natural optimization formulations for visual inverse problems, have matched or even beaten the state-of-the-art methods on numerous visual reconstruction tasks, although not driven by massive amounts of training data. Despite the remarkable successes, the over parametrized DNNs used are
typically powerful enough to also fit the noise besides the desired visual contents (i.e., overfitting), and the fitting process can take up to tens of minutes on sophisticated GPU cards to converge to a reasonable solution. In this talk, I’ll describe our recent efforts to combat these practicality issues around DIP and NIR, and how careful calibration of DIP models (or variants) can help to solve challenging visual reconstruction problems, such as blind image deblurring and phase retrieval, in unprecedented regimes.

Joint work with Taihui Li, Hengkang Wang, Zhong Zhuang, Hengyue Liang, Le Peng, and Tiancong Chen.

Related papers:
Early Stopping for Deep Image Prior  https://arxiv.org/abs/2112.06074
Self-Validation: Early Stopping for Single-Instance Deep Generative Priors https://arxiv.org/abs/2110.12271
Blind Image Deblurring with Unknown Kernel Size and Substantial Noise https://arxiv.org/abs/2208.09483

Professor, Department of Computer Science and Engineering

Title: From Surveying Farms to Tidying our Homes with Robots

Abstract: For decades, the robotics community has been working on developing intelligent autonomous machines that can perform complex tasks in unstructured environments. We are now closer than ever to delivering on this promise. Robotic systems are being developed, tested, and deployed for various applications. In this talk, I will present our work on building robots for agriculture and home automation which are two application domains with distinct sets of associated challenges. In agriculture, robots must be capable of operating on very large farms under rough conditions while maintaining precision to efficiently perform tasks such as yield mapping, fruit picking, and weeding. In these applications, state-of-the-art perception algorithms can generate intermediate geometric representations of the environment. However, the resulting planning problems are often hard. I will present some of our work on tracking and mapping and give examples of field deployments. In-home automation, the robots must be able to handle a large variety of objects and clutter. In such settings, generating precise geometric models as intermediate representations is not always possible. To address this challenge, I will present our recent and ongoing work on developing state representations for coupled perception and action planning for representative home automation applications such as decluttering.