To request a Distinguished Lecturer (DL) for your next event, complete the DL Application Form.
Katja Mombaur (AM’04-M’05) Prof. Dr. Katja Mombaur joined the University Waterloo in March 2020 as Full Professor and Canada Excellence Research Chair (CERC) for Human-Centred Robotics & Machine Intelligence. Her research focuses on understanding human movement by a combined approach of model-based optimization and experiments and using this knowledge to improve motions of humanoid robots and the interactions of humans with exoskeletons, prostheses and external physical devices. Her goal is to endow humanoid and wearable robots with motion intelligence that allow them to operate safely in a complex human world. Prior to coming to Canada, she has been a Full Professor at the Institute of Computer Engineering of Heidelberg University and head of the Optimization, Robotics & Biomechanics group, as well as coordinator of the Heidelberg Center for Motion Research. She holds a diploma degree in Aerospace Engineering from the University of Stuttgart and a PhD In Applied Mathematics from Heidelberg. She has coordinated the European project KoroiBot, has been part of several other European projects such as Spexor, MOBOT and ECHORD, and still is a partner in the ongoing European projects Eurobench and Agilis, and one of the directors of the HeiAge project in Heidelberg.
Model-based optimization for improving the motion intelligence of human-centred robots
Human-centred robots have the potential to support and facilitate people’s lives, ranging from improved well-being and increase independence to reduced risk or harm and a removal of boring jobs. They can take the form of humanoid robots, wearable robots or other types of mobility assistance robots and have to enter in in close physical interactions with humans or support them physically. For this, human-centred robots require motion intelligence or embodied intelligence that makes the robot aware of how it moves in and interacts with its dynamic environment and with humans. In addition to biomechanical studies of human behavior, model-based optimization or optimal control is a widely used approach for generating and controlling motions of human-centered robots. Optimization can tackle the challenges of such systems which include a high complexity, redundancy, underactuation, and a high risk of instability and falls. In this talk, I will present different examples from my research group on using model-based optimal control to control different humanoid robot platform and to improve the design and control of wearable robots for the lower limbs and the lower back and other assistive devices. I will discuss different levels of modeling robots – and in some cases also the interacting humans - to address specific research questions. In addition, I will discuss possible combinations of optimal control methods with reinforcement learning and movement primitive approaches to reduce computation times and improve robot control.
What do we optimize? Inverse Optimal Control as a Tool to Understand Human
Gaining a fundamental understanding of the movements of the human body has long been an important research topic in biomechanics, sports science, physiology, neuroscience, computer animation and many areas of robotics and human-robot interaction. How do humans choose their motions out of the infinite number of ways to perform a given task? And how do motions change based on the situation, or based on the person’s age, training level or medical condition? It is a common assumption that motions of humans and animals – similar to many other processes in nature - are performed in an optimal way due to evolution, learning and training. Optimality principles can be found in the mechanical properties of the executed movements, but also in the closed loop sensory motor system. However, the particular criterion optimized is highly dependent on the specific case and situation and not easy to determine. In this talk, I discuss inverse optimal control as a very promising systematic approach to identify the underlying optimality principles of human movement. Starting from (partial) motion capture data of a specific
human movement or a set of movements of multiple subjects and subject-specific mathematical models, inverse optimal control tells us which objective function – or typically combination of multiple objective functions - gives the closest approximation of the recorded data. I will present algorithmic approaches for solving inverse optimal control as well as a number of examples, including walking on different terrains, running motions of amputees and non-amputees, painting, lifting motions with and without back pain, and interactions with robotic manipulandum. Inverse optimal control has lead to very promising results in these biomechanical studies, and certainly has a huge potential for a much wider
Cường was born in Hanoi, Vietnam. He is an alumnus of École Normale Supérieure, rue d’Ulm (France) and holds a Ph.D. in Neuroscience from Université Pierre et Marie Curie (France). He was a visiting researcher at the University of São Paulo (Brazil) in 2010, and a JSPS Fellow at the University of Tokyo (Japan) in 2011-2013. He joined NTU (Singapore) in 2013 and is currently an Associate Professor in the School of Mechanical and Aerospace Engineering. He was a recipient of the Best Paper Award at the conference Robotics: Science and Systems, 2012. His research has featured in major international media, including The New York Times, The Guardian, The Economist, CNN, Science, Nature, etc. He is also a Founder and Director of Eureka Robotics
(https://eurekarobotics.com/), a deep-tech startup devoted to solving the toughest automation challenges in manufacturing. Eureka Robotics is a recipient of the 2019 IEEE N3XT Star award.
Model-based robotic manipulation with contact and dynamics
Planning and executing motions in the presence of contacts and significant dynamics effects still constitute a major challenge in robotics. Of particular interest, for example, is the computation of dynamically-feasible motions in fractions of a second for real-world industrial applications, such as robotic pick-and-place. We have developed new methods for planning dynamically-feasible motions within milliseconds, based on completely new approaches to solving the Time-Optimal Path Parameterization (TOPP) problem, and to utilizing it as a subroutine in kinodynamic motion planners (Admissible Velocity Propagation). I will discuss our theoretical framework, as well as applications ranging from the "waiter motion" to critically-fast pick-and-place with suction cups. Along the way, I will also present a number of complex tasks involving contacts and dynamics we have tackled in recent years: automatic precision drilling, autonomous assembly of an IKEA chair, large-scale 3Dprinting by a team of mobile robots, etc. These complex tasks illustrate the need for building robust and scalable robotic systems that address multiple challenges, from precise localization, to motion planning, to control of contact forces. Videos of the demos can be found on our channel: https://www.youtube.com/c/CRIGroupRobotics
Pierre-Brice Wieber is a full-time researcher at INRIA Grenoble and has been a visiting researcher at AIST/CNRS Joint Research Lab in Tsukuba. He has advised 14 PhD students and 6 Post-Docs on topics covering modeling, optimization and control of autonomous vehicles, humanoid and legged robots, industrial and collaborative robots. His specific focus of interest is on model-based safety guarantees. He has been serving as Associate Editor for IEEE Transactions on Robotics, Robotics and Automation Letters and conferences such as ICRA and Humanoids.
Talk # 1
A mathematical approach to Isaac Asimov's Three Laws of Robotics
1/ A robot may not injure a human being. 2/ A robot must obey orders except where they conflict with the First Law. 3/ A robot must protect its own existence as long as this does not conflict with the First or Second Law. I propose to discuss how these broad, eight decades old statements can be approached and implemented in today's autonomous vehicles, humanoids, and collaborative robots, introducing a general mathematical approach to provide the corresponding safety guarantees. This will naturally raise the question of models in decision making, and a few ethical issues.