IROS 2022 Workshop

Neuromechanics meet Deep Learning: new opportunities and avenues for robotics, human-robot interaction and in-silico robotics mediated rehabilitation

Join us at IROS 2022 in Kyoto, Japan

Where: Kyoto International Conference center, Kyoto, Japan

When: October 23-27

Complexity of motor behavior an organism exhibits is a direct expression of intelligence it possesses. An important question to consider is -- what enables such complex decision making, and motor control to execute them?

Synthesis and expression of such diverse behaviors requires effective coordination between the central nervous system – where intelligent controllers are created by networks of billions of neurons – and the peripheral musculoskeletal system which translates the intentions into actions. Akin to the central nervous system, the field of Artificial Intelligence has been pursuing the emulation of intelligent behaviors via neural structures (Neural Networks). At the same time, and mostly independently, the biomechanics community has been developing musculoskeletal systems to understand peripheral actuation. Since the uniqueness of vertebrate motor control evolved at the juncture of neural decision making and musculoskeletal control, via this workshop, we seek a platform for the experts of two fields to come together to share respective progress and deliberate joint opportunities.

After incredible progress over the last decade in both fields, we stand today at crossroads where (a) musculoskeletal models are reasonably accurate and computationally efficient, and (b) deep learning techniques are capable of scaling to complex problems. The time is ripe when both fields can catalyze and challenge each other. The development of large scale neuromusculoskeletal models could provide the next frontier for deep learning to synthesize controllers matching human performance. On the other hand, such controllers embedded in the same perception-action loop as biological counterparts can provide a deeper understanding on how physiological effectors are controlled. Potential impacts can be efficient bio-inspired robotics systems and controllers, effective rehabilitation techniques, refinement of human-wearables, cobot interaction, etc.

Via this workshop, we hope to (a) to provide an overview of existing approaches used at the intersection of neuromusculoskeletal modeling and machine learning and (b) to bring the two fields together for a dialogue and exchange of ideas. To achieve these goals, we have assembled a panel of experts in musculoskeletal modeling, human motor control, and deep learning, to provide an overview of the field and opportunities for cross pollination. We also have arranged for a balanced set of invited talks, tutorials and facilitated discussions which will encourage individuals across disciplines to interact, share experience and explore ideas to further develop this emergent field.

Poster Presentations:

Phd and Master students are encouraged to participate via a poster presentation. Please send us an email (g[dot]v[dot]durandau[at]utwente[dot]nl) with a short explanation about your project and results. The best project will be able to present their research during the poster session.

PROGRAM:

08:30 - 08:35

Opening remarks

08:35 - 08:55

Invited talk 1

08:55 - 09:15

Invited talk 2

09:15 - 09:35

Invited talk 3

09:35 - 09:55

Invited talk 4

09:55 - 10:25

Coffee break and poster session

10:25 - 10:45

Invited talk 5

10:45 - 11:05

Panel discussion

11:05 - 11:45

Tutorial 1

11:45 - 12:25

Tutorial 2

12:25 - 12:30

Conclusion

Speakers:

  • Fabio Pardo

  • Josh Merel

  • Andrea D’Avella

    After being trained in Physics, Andrea d'Avella obtained a Ph.D. in System and Computational Neuroscience at Massachusetts Institute of Technology, Cambridge, USA, in 2000 working on the modular organization of the frog motor system in the laboratory of Emilio Bizzi. In 2003 he joined the Laboratory of Neuromotor Physiology at Santa Lucia Foundation, Rome, Italy. Since 2015 he is Professor of Physiology in the Department of Biomedical Sciences and Morphological and Functional Images at the University of Messina, Italy. His research is focused on investigating sensorimotor control of reaching, throwing and catching movements, muscle synergies in healthy subjects and after neurological lesions, and motor adaptation. He has developed a novel computational technique to identify time-varying muscle synergies from multi-muscle EMG recordings and a new approach using myoelectric control and virtual reality(adaptation to "virtual surgeries") to directly probe the synergistic organization of the motor system. 

  • Seugmoon Song

Organizations:

  • Vikash Kumar
  • Vittorio Caggiano
  • Huawei Wang

    Huawei Wang, a postdoc researcher in the Neuromechanical Modelling and Engineering Lab. His interests are neuromusculoskeletal modeling; wearable measurement devices; real-time state estimation; and wearable robotics.

  • Guillaume Durandau

    Guillaume Durandau is a postdoctoral researcher at the Department of Biomechanical Engineering, University of Twente, Enschede, the Netherlands. He obtained his engineering diploma at ISEN, Toulon, France and his master degrees at the University of Sherbrooke, Sherbrooke, Canada. During his master, he spend four months at the CINESTAV, Mexico. He also obtained during his master, the Mitacs Globalink Research Award. At the beginning of 2015, he started as a research assistant at the Institute of Neurorehabilitation Systems, UMG, Gottingen, Germany. In 2017, he started is PhD at the University of Twente, Enschede, the Netherlands. During his PhD, He won the Best Demo Award in 2018 at the 7th IEEE BIOROB, Best paper Published by IEEE-EMBS in 2017-18, 3rd place at the 7th Dutch Bio-medical Engineering Conference and the best thesis Award 2021 from the European Society of Biomechanics.

  • Massimo Sartori

    Massimo Sartori is a Professor and Chair of Neuromechanical Engineering at the University of Twente where he also direct the Neuromechanical Modelling & Engineering Lab. It's research focuses on interfacing wearable robotic technologies with the neuromuscular system for enhancing human movement. On these topics he is directing personal grants (e.g. ERC Starting Grant) and is PI in consortium-based projects (e.g. H2020-ITN Project SimBionics). He conducted his PhD (2009-2011) across the Universities of Padova (Italy), Western Australia (Australia) and Stanford (USA). He continued with a post-doc at the University of Göttingen (Germany, 2011) where he became Junior Research Group Leader in 2015. Since In April 2017 he joined the University of Twente as a tenure-track scientist where he is leading an expanding independent research group. Throughout his career he received awards (e.g. OpenSim Outstanding Research), was guest editor in academic journals (e.g. IEEE TBME, Front Comput Neurosci), and was Workshop Chair at leading congresses (e.g. IEEE BioRob 2018) in the field. He is currently Associate Editor at the IEEE Transactions on Neural Systems and Rehabilitation Engineering: https://www.embs.org/tnsre/associate-editors/. Moreover, he is a member of leading scientific societies spanning across the fields of robotics and biomechanics including: IEEE Robotics and Automation Society, IEEE Engineering in Medicine and Biology Society, IEEE International Consortium on Rehabilitation Robotics, and European Society of Biomechanics.

    More about him via the UT Featured Scientist page.