Embedded Files
Massachusetts Institute of Technology, USA
Talk title: A Bayesian approach to breaking things: efficiently predicting and repairing failure modes via sampling
Talk abstract: Before robots can be deployed in safety-critical applications, we must be able to understand and verify the safety of these systems. For cases where the risk or cost of real-world testing is prohibitive, we propose a simulation-based framework for a) predicting ways in which a robotic system is likely to fail and b) automatically adjusting the system’s design to mitigate those failures preemptively. We frame this problem through the lens of approximate Bayesian inference and use differentiable simulation for efficient failure case prediction and repair. I will introduce several applications of this approach on a range of robotics and control problems, including optimizing search patterns for robot swarms and reducing the severity of outages in power transmission networks.
Bio: Chuchu Fan is an Assistant Professor in the Department of Aeronautics and Astronautics (AeroAstro) and Laboratory for Information and Decision Systems (LIDS) at MIT. Before that, she was a postdoc researcher at Caltech and got her Ph.D. from ECE at the University of Illinois at Urbana-Champaign. She earned her bachelor’s degree from Tsinghua University, Department of Automation. Her research group Realm at MIT works on using rigorous mathematics, including formal methods, machine learning, and control theory, for the design, analysis, and verification of safe autonomous systems. Chuchu is the recipient of an NSF CAREER Award, an AFOSR Young Investigator Program (YIP) Award, and the 2020 ACM Doctoral Dissertation Award.
Talk title: A Correct-by-Construction Paradigm for Designing Autonomous Systems
Talk abstract: Correct-by-construction synthesis is at the forefront of merging formal methods with control theory, particularly in the design of safety-critical systems. This methodology transcends the traditional, exhaustive cycle of redesign, verification, and validation, advocating for a streamlined process that iterates on formal requirements through rigorous proof chains, embedding system correctness directly into the design stage. Over the past two decades, significant strides have been made in expanding the scope of correct-by-construction synthesis, especially for cyber-physical systems that integrate discrete-event control with continuous dynamics. The field has advanced by combining symbolic techniques with state-space reduction strategies, enhancing the feasibility of applying correct-by-construction principles to complex autonomous systems. Additionally, recent research has explored new avenues, such as barrier certificates and data-driven methods, which are particularly beneficial in scenarios where system models are unknown. This talk will highlight our recent breakthroughs in establishing a solid foundation for the correct-by-construction synthesis of cyber-physical systems.
Bio: Bio: Majid Zamani is an associate professor in the Computer Science Department at University of Colorado Boulder and leading Hybrid Control Systems Lab. Between May 2014 and January 2019, he was an assistant professor (W2 grade) in the Department of Electrical Engineering at Technical University of Munich. He received a Ph.D. degree in Electrical Engineering and an MA degree in Mathematics both from University of California, Los Angeles in 2012, an M.Sc. degree in Electrical Engineering from Sharif University of Technology in 2007, and a B.Sc. degree in Electrical Engineering from Isfahan University of Technology in 2005. He received the NSF Career award in 2022 and ERC starting grant award from the European Research Council in 2018. His research interests include verification and control of cyber-physical systems, hybrid systems, embedded control software synthesis, networked control systems, and incremental properties of nonlinear control systems.
University of Colorado Boulder. USA
Marie Farrell
Marie Farrell
The University of Manchester, UK
Talk title: Formal Specification and Verification for Autonomous Robots
Talk abstract: This talk will provide an overview of the state-of-the-art formal methods used in the specification and verification of autonomous robots. It will also provide some examples and specific use cases of these tools and techniques applied to robotic systems.
Bio: Dr. Marie Farrell is a distinguished Royal Academy of Engineering Fellow, currently leading a pioneering project entitled "Strong Software Reliability for Autonomous Space Robotics." This fellowship focuses on devising innovative methodologies to describe, analyze, and ensure the autonomous behavior of robotic systems designed for space exploration. Prior to this prestigious fellowship, Dr. Farrell served as a Senior Post-Doctoral Researcher in the Department of Computer Science at Maynooth University, contributing to the VALU3S project. This role involved eliciting and verifying requirements for an aircraft engine controller, showcasing her expertise in rigorous system validation. Dr. Farrell's earlier work includes significant contributions to the EPSRC-funded FAIR-SPACE Hub, as well as participation in the RAIN and ORCA Hubs. Her research in these projects concentrated on the application and integration of formal methods to provide certification evidence for robotic systems operating in hazardous environments, underscoring her commitment to advancing safety and reliability in robotics. In 2017, Dr. Farrell earned her PhD from Maynooth University, where she developed a semantics, modularization constructs, and interoperability framework for the Event-B formal specification language using the theory of institutions. This work laid the foundation for her ongoing contributions to the field of autonomous and space robotics.
Talk title: Certifying Robustness of Learning-Based Keypoint Detection and Pose Estimation Methods
Talk abstract: This talk addresses the certification of the local robustness of vision-based two-stage 6D object pose estimation. The two-stage method for object pose estimation achieves superior accuracy over the single-stage approach by first employing deep neural network-driven keypoint regression and then applying a Perspective- n-Point (PnP) technique. Despite advancements, the certification of these methods’ robustness, especially in safety-critical scenarios, remains scarce. This research aims to fill this gap with a focus on their local robustness on the system level—the capacity to maintain robust estimations amidst semantic input perturbations. The core idea is to transform the certification of local robustness into a process of neural network verification for classification tasks. The challenge is to develop model, input, and output specifications that align with off-the-shelf verification tools. Under certain conditions, we demonstrate that the main components of our certification framework are both sound and complete. We validate its effectiveness through extensive evaluations on realistic perturbations, shedding light on the certification of complex, learning-based perception systems.
Bio: Dr. Changliu Liu is an assistant professor in the Robotics Institute, School of Computer Science, Carnegie Mellon University (CMU), where she leads the Intelligent Control Lab. Prior to joining CMU in Jan 2019, Dr. Liu was a postdoc at Stanford Intelligent Systems Laboratory in 2018. She received her Ph.D. in Engineering together with Master degrees in Engineering and Mathematics from University of California at Berkeley (in 2017, 2014, 2015 respectively) and her bachelor degrees in Engineering and Economics from Tsinghua University (in 2012). Her research interests lie in the design and verification of human-centered intelligent systems with applications to manufacturing and transportation and on various robot embodiments, including robot arms, mobile robots, legged robots, and humanoid robots. She published the book “Designing robot behavior in human-robot interactions” with CRC Press in 2019. She is the co-founder of the International Neural Network Verification Competition launched in 2020. Her work has been recognized by NSF Career Award, Amazon Research Award, Ford URP Award, Advanced Robotics for Manufacturing Champion Award, Young Investigator Award at International Symposium of Flexible Automation, and many best/outstanding paper awards.
Carnegie Mellon University, USA
King Abdullah University of Science and Technology, Saudi Arabia
Talk title: Passivity, Stability, and Learning in Multi-Robot Games
Talk abstract: In this talk, we discuss a multi-robot game framework, where we design and analyze learning models for multiple robots to adopt effective strategies for carrying out a team mission. In the multi-robot games, given a set of strategies, each robot selects a strategy to engage in repeated strategic interactions with others. Instead of computing and adopting the best strategy based on a known cost function, robots learn this strategy selection from the instantaneous payoffs they receive at each stage of the repeated interactions. Distinct from existing formulations relevant to multi-agent games, our framework introduces dynamic payoff mechanisms with an underlying mechanism that has its own dynamics. We discuss how stability analysis and passivity methods from feedback control theory can be used as formal methods to analyze the performance of the framework. We also examine the practical application of this framework to multi-robot task allocation scenarios, where a decentralized decision-making model is necessary for a team of mobile robots to select and carry out a set of tasks in dynamically changing environments.
Bio: Shinkyu Park received the B.S. degree from Kyungpook National University (경북대학교), Korea, in 2006 (summa cum laude), the M.S. degree from Seoul National University (서울대학교), Korea, in 2008, and the Ph.D. degree from the University of Maryland, College Park, MD, USA, in 2015, all in electrical engineering. From January to June in 2016, he was a Postdoctoral Researcher at National Geographic Society, where he investigated distributed decision problems in multi-agent systems with applications to analysis of animal group behavior. From June 2016 to May 2019, he was a Postdoctoral Associate at Massachusetts Institute of Technology, where his research focused on conceiving a fleet of reconfigurable robotic vessels in Roboat project and developing a distributed sensing/sampling platform for the urban epidemiology project, Underworlds. Prior to joining KAUST, he was appointed as an Associate Research Scholar at Princeton University working in cross-departmental robotics projects. His current research interests are in robotics, multi-robot control/coordination, feedback control theory, and game theory.
Talk title: The Role of Control in Timed Temporal Logic-based Planning of Autonomous Systems
Talk abstract: Temporal logics provide an efficient and user-friendly way to express complex task specifications for autonomous systems as well as provably correct methodologies for their execution. The incorporation of deadlines (a.k.a., time constraints) in such specifications can offer greater efficiency and the ability to perform a large variety of tasks, imposing however challenging constraints on the underlying systems. In particular, how can we guarantee that an autonomous system can execute trajectories within specific time limits? This is further complicated by potential uncertainties and exogenous disturbances in the system’s dynamics, obstacle-cluttered workspaces, or a priori unknown environments. In this talk, I will discuss the concept of funnel control and how it can be used to facilitate safe planning subject to timed temporal logic goals. I will elaborate on how funnel controllers can adapt in real time to uncertainties in the dynamics and the environment and on their connection to high-level planning.
Bio: Christos K. Verginis is an assistant professor at the School of Electrical Engineering, Uppsala University. He received his Ph.D. in automatic control from KTH Royal Institute of Technology in 2020. Before joining Uppsala University in 2022, he was a postdoctoral researcher at the University of Texas at Austin. His research interests include planning and control of multi-robot systems, safety-critical and adaptive control of uncertain nonlinear systems, temporal-logic-based planning, and learning. His Ph.D. thesis received the EECI award for the best thesis in control of complex and heterogeneous systems and was a finalist for the George Giralt Ph.D. award in Robotics.
Uppsala University, Sweden
Indian Institute of Science Bangaluru
Talk title: Real-time constrained data-driven tracking control of redundant manipulators with formal guarantees
Talk abstract: In this talk, we consider data-driven identification and control of redundant serial manipulators characterized by a high-dimensional nonlinear input-output map. We present an efficient data-driven model synthesis framework that relies on the Koopman operator for realizing constrained real-time control with formal convergence guarantees. To this end, an autoencoder-based neural architecture is employed to discover the bilinear Koopman model for manipulator dynamics in joint space using input-output data, which is subsequently integrated with a feed-forward neural network that maps the joint coordinates to end-effector Cartesian coordinates, thus allowing for the learning of highly accurate models with a significantly lower number of observable states compared to the previous studies. The learning architecture allows for coupling with a zeroing dynamics neural controller, which offers a computationally inexpensive alternative to state-of-the-art Nonlinear Model Predictive Control controllers whose computational burden might render real-time control infeasible. The low-dimensional learned model, when combined with the computationally inexpensive control policy, facilitates real-time control applications with improved tracking accuracy. Simulation and experimental studies of trajectory tracking on a KUKA IIWA 7-DOF robot demonstrate the efficacy of the proposed schemes.
Bio: Dr. Jishnu Keshavan is an Assistant Professor in Mechanical Engineering at the Indian Institute of Science Bangalore, India. Prior to this appointment, he was an Assitant Professor at Mississippi State University, MS USA. His research interests are broadly in the areas of unmanned systems, nonlinear dynamics and control, and autonomous vision. He obtained a PhD (2012) and a MS (2007) in Aerospace Engineering from University of Maryland, College Park, and a BTech in Aerospace Engineering (2004) from the Indian Institute of Technology Bombay.
Page updated
Google Sites
Report abuse