Speakers
Luca Carlone
coming soon
Abstract: coming soon
Bio: Luca Carlone is the Charles Stark Draper Assistant Professor in the Department of Aeronautics and Astronautics at the Massachusetts Institute of Technology, and Principal Investigator in the MIT Laboratory for Information & Decision Systems (LIDS). He received his PhD from the Polytechnic University of Turin in 2012. He joined LIDS as a postdoctoral associate (2015) and later as a Research Scientist (2016), after spending two years as a postdoctoral fellow at the Georgia Institute of Technology (2013-2015).
His goal is to enable human-level perception and world understanding on mobile robotics platforms (micro aerial vehicles, self-driving cars, ground robots) operating in the real world. Towards this goal, his work involves a combination of rigorous theory and practical implementations. In particular, his research interests include nonlinear estimation and probabilistic inference, numerical and distributed optimization, and geometric vision applied to sensing, perception, and decision-making in single and multi-robot systems.
His work includes seminal results on certifiably-correct algorithms for localization and mapping, as well as approaches for visual-inertial navigation and distributed mapping. He is a recipient of the 2017 Transactions on Robotics King-Sun Fu Memorial Best Paper Award, and the best paper award at WAFR 2016.
Noemie Jaquier
Geometry-aware Bayesian optimization for efficient robot skill learning
Abstract: Fast and data-efficient skill learning is a key challenge in robotics, where robots often need to generalize previously-learned policies to unforeseen settings. In this context, Bayesian optimization has emerged as a powerful tool to efficiently optimize parametric policies in challenging robotics scenarios. However, improving the performance, scalability, and safety of Bayesian optimization techniques remains a central goal for robotics applications.
This talk introduces the idea of incorporating inductive bias into Bayesian optimization algorithms by leveraging task-specific information on the geometry of the search space. Many robotic parameters belong to non-Euclidean spaces with inherent geometric properties. For instance, orientations can be viewed as elements of the unit sphere or the special orthogonal group; control gains, inertia, and manipulability ellipsoids lie in the manifold of symmetric-positive-definite matrices; and robot joint configurations belong to the torus. First, I will discuss how geometry-awareness can be brought into Bayesian optimization by leveraging Riemannian manifold theory, both in the surrogate model and in the optimization of the acquisition function. Second, I will discuss how this framework can be extended to optimize high-dimensional non-Euclidean objective functions. Finally, I will show that leveraging geometry as inductive bias for Bayesian optimization generally leads to significant performance improvements in a variety of benchmarks and robotics applications.
Bio: Noemie Jaquier is a visiting postdoctoral scholar working with Prof. Oussama Khatib at the Stanford Robotics Lab and a postdoctoral researcher working with Prof. Tamim Asfour in the High Performance Humanoid Technologies Lab (H²T) at the Karlsruhe Institute of Technology (KIT).
From August 2016 to July 2020, she was a PhD student affiliated to EPFL and working at the Idiap Research Institute. She did a PhD sabbatical in the Bosch Center for Artificial Intelligence (BCAI), Germany from April to September, 2019. She obtained a Bachelor in Microengineering (2014), a Master in Robotics and Autonomous Systems (2016) and a Minor in Computational Neurosciences (2016) from EPFL.
Her research brings a novel Riemannian perspective to robot learning, optimization, and control by leveraging Riemannian geometry as inductive bias and as a theory to provide sound theoretical guarantees. She investigates data-efficient methods that build on geometric spaces and exploit the geometric information naturally arising in robotic data. Noemie's work focuses on skills learning via human demonstrations and adaptation techniques with geometry as a cornerstone. It spans various applications in the field of robot manipulation.
Web: https://njaquier.ch/
David M. Rosen
coming soon
Abstract: coming soon
Bio: David M. Rosen is an Assistant Professor in the Departments of Electrical & Computer Engineering and Mathematics, and the Khoury College of Computer Sciences (by courtesy). Prior to joining Northeastern, he was a Research Scientist at Oculus Research (now Meta Reality Labs) from 2016 to 2018, and a Postdoctoral Associate at MIT’s Laboratory for Information and Decision Systems (LIDS) from 2018 to 2021.
He is broadly interested in the mathematical and algorithmic foundations of trustworthy autonomy. His research applies tools from nonlinear optimization, differential geometry and topology, abstract algebra, and probability and statistics to devise principled, computationally efficient, and provably robust algorithms for machine perception and control. Much of his recent work has explored the use of principled approximation schemes (such as convex relaxation) to efficiently compute provably-good solutions of challenging machine perception problems (such as SLAM).
His work has been recognized with multiple awards at flagship international venues, including the inaugural Best Paper Award at the International Workshop on the Algorithmic Foundations of Robotics (2016), selection as an RSS Pioneer (2019), a Best Student Paper Award at Robotics: Science and Systems (2020), and an Honorable Mention for the IEEE Transactions on Robotics King-Sun Fu Memorial Best Paper Award (2021).
Paul Goulart
coming soon
Abstract: coming soon
Bio: Paul Goulart joined the University of Oxford in 2014 as an Associate Professor in Engineering Science. He received his SB and MSc degrees in Aeronautics and Astronautics from the Massachusetts Institute of Technology (MIT). Following his undergraduate studies he was a software developer in the flight operations centre for the Chandra X-Ray Observatory at the Harvard-Smithsonian Centre for Astrophysics, and later an engineer in the Autonomous Systems group at the Charles Stark Draper Laboratory.
In 2003, Paul was selected as a Gates Scholar at the University of Cambridge, where he received a PhD in Control Engineering in 2007. From 2007 to 2011 he was a Lecturer in control systems in the Department of Aeronautics at Imperial College London, and from 2011 to 2014 a Senior Researcher in the Automatic Control Laboratory at ETH Zurich. He is a member of the Control Group in the department of Engineering Science.
Paul's research interests are in high speed optimization and control for embedded systems, data-driven and robust optimization, and machine learning. He works on embedded and large-scale optimization applications ranging from energy systems to fluid mechanics. His research group are the developers of the open-source optimization packages OSQP, Clarabel and COSMO.
Francesco Biral
InDi: An Indirect/Direct method for Optimal Control Problems
Abstract: Optimal control problems are widely present in engineering and with an increasing number of successful applications. Direct methods emerge as the most popular approach to solve these problems because they are simpler to implement and require less symbolic computation. Instead, the indirect approach requires more symbolic manipulation, but through the Pontryagin maximum principle, it allows for a better understanding of the solution's structure, and the control part is captured more naturally.
This work introduces a tailored derived numerical method for solving optimal control problems which can be seen interchangeably as derived from a direct approach or from an indirect approach. The search for a connection between the solution of the discretization of direct methods with that obtained with the discretization of indirect methods is not a new idea. The principle of covector mapping (CMP) has demonstrated a mapping that makes the solution of the direct method equivalent to that of the indirect method when using pseudospectral collocation techniques. Similarly to CMP, this study introduces a mapping approach for finite difference scheme that connects the discretized equations of indirect methods with those of direct transcription applied to implicit dynamics. Revisiting the discretization of the indirect method as that obtained from a direct transcription allows us to hybridize the solution techniques and algorithms of the two families of methods by exploiting their respective strengths.
This innovative formulation holds great promise for the indirect method, facilitating the adoption of regularization techniques commonly associated with direct approaches, all while retaining the explicit solution of the control as a function of the states and co-states within a bi-level optimization framework that significantly improves the numerical performance. This flexible formulation allows the introduction of numerical techniques derived from direct methods, as proximal methods, with the aim of obtaining a robust solver while maintaining the efficient structure of Newton's damped method used for the solution of the indirect method. Moreover, the proposed mapping extends its utility to encompass implicit dynamics, thereby providing a comprehensive solution strategy for a broader array of optimal control problems and applications.
Bio: Francesco Biral received the Master Degree in Mechanical Engineering at the University of Padova, Italy, and the Ph.D. in Mechanism and Machine Theory from University of Brescia, Italy, in 2000. He is currently Associate Professor at the Department of Industrial Engineering at University of Trento.
His research interests include symbolic and numerical multibody dynamics and optimisation, constrained optimal control, mainly in the field of sport engineering and vehicle dynamics with special focus on minimum lap time problems and Intelligent Transportation Systems.
Russ Tedrake
Planning through contact with Graphs of Convex Sets
Abstract: We've recently introduced Graphs of Convex Sets (GCS) as a general framework for optimizing mixed continuous + combinatorial optimization problems that are naturally specified on a graph, and have started to apply this technique to motion planning. In this talk, I'll describe how we transcribe planning through contact into GCS with additional work on making tight SDP relaxations of the contact dynamics. Amazingly, this approach allows us to solve multi-modal planning through contact problems with convex optimization plus a little rounding. So far, we have studied this transcription carefully on relatively simple problems; I will also discuss the potential and limitations.
Bio: Russ Tedrake is the Toyota Professor of Electrical Engineering and Computer Science, Aeronautics and Astronautics, and Mechanical Engineering at MIT, the Director of the Center for Robotics at the Computer Science and Artificial Intelligence Lab, and the leader of Team MIT's entry in the DARPA Robotics Challenge. Russ is also the Vice President of Robotics Research at the Toyota Research Institute. He is a recipient of the 2021 Jamieson Teaching Award, the NSF CAREER Award, the MIT Jerome Saltzer Award for undergraduate teaching, the DARPA Young Faculty Award in Mathematics, the 2012 Ruth and Joel Spira Teaching Award, and was named a Microsoft Research New Faculty Fellow.
Russ received his B.S.E. in Computer Engineering from the University of Michigan, Ann Arbor, in 1999, and his Ph.D. in Electrical Engineering and Computer Science from MIT in 2004, working with Sebastian Seung. After graduation, he joined the MIT Brain and Cognitive Sciences Department as a Postdoctoral Associate. During his education, he has also spent time at Microsoft, Microsoft Research, and the Santa Fe Institute.
Armin Nurkanovic
Numerical methods for solving nonsmooth optimal control problems
Abstract: In control applications contact and friction phenomena are often modeled as nonsmooth dynamical systems. In order to optimally control such systems or to implement MPC controllers, nonsmooth optimal control problems have to be solved. This involves three steps: (re)formulating the problem, discretizing it, and solving it. In contrast to standard smooth trajectory optimization, these steps are highly nontrivial. In this talk, we review recent developments in all three steps and show with counterexamples why naive approaches must fail. For the first step, we investigate the time-freezing reformulation - which simplifies systems with state jumps; for the second, finite elements with switch detection (FESD) - which ensures high accuracy and correct sensitivities; and we discuss how to efficiently solve mathematical programs with complementarity constraints - which frequently arise in the third step. Examples show how special care is needed in all three steps to obtain efficient and meaningful solutions.
Bio: Armin Nurkanovic received a Bachelor's degree in Electrical Engineering from the University of Tuzla in 2015 and a Master's degree in Electrical Engineering and Information Technology from the Technical University of Munich in 2018. He received a DAAD scholarship for his master's studies. From 2018 he was a PhD student at the University of Freiburg (first external-industrial, then internal) under the supervision of Prof. Moritz Diehl. Until October 2021, he was an industrial PhD student at Siemens Technology in the research group Autonomous Systems and Control. Together with his co-authors, he received the IEEE Control Systems Letters Outstanding Paper Award for 2022. He successfully defended his PhD thesis (summa cum laude) in November 2023.
Armin is currently interested in developing numerical methods for optimal control of hybrid and nonsmooth dynamical systems, including ODEs with discontinuous vector fields (Filippov systems, switched systems, dynamic complementarity systems, and similar) and systems with state jumps (mainly systems from nonsmooth mechanics).
Robin Deits
Taskable Model-Predictive Control on the Electric Atlas
Abstract: The Atlas humanoid robot can walk, jump, run, and manipulate, and it can combine those behaviors seamlessly to create flashy dance moves, parkour sequences, or dynamic throws. In this presentation, I'll discuss how we've built the generations of model-predictive control that power Atlas's behaviors, as well as the challenges of deploying MPC in reality. Finally, I'll show how we're developing new behaviors on top of that controller to allow the new electric Atlas to start doing useful work in the world.
Bio: Robin Deits is a roboticist at Boston Dynamics.
Marc Toussaint
Comparing Concepts from Constrained Optimization and Sampling
Abstract: I will discuss relations between typical MCMC (e.g. Langevin, manifold & polytope sampling) and constrained optimization techniques, and argue for a unified approach. My original motivation is robustness in task-and-motion planning.
Bio: Marc Toussaint is professor in the area of AI & Robotics at TU Berlin, lead of the Learning & Intelligent Systems Lab at the EECS Faculty, and member of the Science Of Intelligence cluster of excellence.
His research interests are in the intersection of AI and robotics, namely in using machine learning, optimization, and AI reasoning to tackle fundamental problems in robotics. The integration of learning and reasoning, of data-based and model-based decision making is of particular interest to him. Concrete research topics he works on are models and algorithms for physical reasoning, task-and-motion planning (logic-geometric programming), learning heuristics, the planning-as-inference paradigm, algorithms and methods for robotic building construction, and learning to transfer model-based strategies to reactive and adaptive real-world behavior. To this end, Marc builds on methodologies from optimization, reinforcement learning, machine learning, search, planning, and probabilistic inference. Some of his earlier work was on evolutionary algorithms (esp. evolving genetic representations and compression), relational reinforcement learning, and active learning. His physics diploma research was on gravity theory as a gauge theory.
Brandom Amos
Differentiable optimization for robotics
Abstract: Optimization is a crucial technology for robotics and provides functionality such as optimal control, motion planning, state estimation, alignment, manipulation, tactile sensing, pose tracking, and safety mechanisms. These solvers are often integrated with learned models that estimate and predict non-trivial parts of the world. *Differentiable optimization* enables the learned model to receive a learning signal from these downstream optimization problems. This signal encourages the model to improve on regions that are important for the optimization problem to work well, rather than making accurate predictions under a supervised loss. This talk will overview the foundations, applications, and recent advancements on these topics.
Bio: Brandon Amos is a Research Scientist in Meta AI’s Fundamental AI Research group in NYC. He holds a PhD in Computer Science from Carnegie Mellon University and was supported by the USA National Science Foundation Graduate Research Fellowship (NSF GRFP). Prior to joining Meta, he has worked at Adobe Research, Google DeepMind, and Intel Labs. His research interests are in machine learning and optimization with a recent focus on reinforcement learning, control, optimal transport, and geometry.
Rika Antonova
Differentiable simulation
Abstract: Differentiable simulation is a promising toolkit for fast gradient-based policy optimization and for adapting simulators to reality. However, existing approaches to differentiable simulation have focused on scenarios with well-behaved gradients, such as systems with mostly smooth dynamics. This talk will outline the challenges arising with contact-rich interactions, deformables and fluids, where simulators yield rugged optimization landscapes. The talk will discuss an approach that combines global search and local optimization to overcome the above challenges, and also touch upon other interesting directions that benefit from differentiable simulation, such as differentiable rendering and optimizing tool morphologies.
Bio: coming soon
Lin Zhao
coming soon
Abstract: coming soon
Bio: Lin Zhao is currently an assistant professor in the Department of Electrical and Computer Engineering at the National University of Singapore. He received a PhD degree in ECE and MSc in Mathematics from The Ohio State University, USA, in 2017, supervised by Prof. Wei Zhang and Prof. Ghaith A. Hiary, respectively. Before that, he received a B.S. and M.S. in Control Science and Engineering from Harbin Institute of Technology, China, in 2010 and 2012, respectively, supervised by Prof. Huijun Gao. Before joining NUS, he was a research scientist at Aptiv Pittsburgh Technology Center (now Motional) working on autonomous driving.
Lin's research intersects control theory, optimization, and machine learning, with applications to unmanned aerial vehicles (UAV). One recent focus is on design and analysis of reinforcement learning algorithms, to boost training efficiencies and control performance with theoretical guarantees. On the UAV side, his group develop AI-assisted planning and control algorithms to enable more intelligent and robust autonomous operations and multi-agent collaborations.