I am 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, I was a PhD student affiliated to EPFL and working at the Idiap Research Institute. I did a PhD sabbatical in the Bosch Center for Artificial Intelligence (BCAI), Germany from April to September, 2019. I obtained a Bachelor in Microengineering (2014), a Master in Robotics and Autonomous Systems (2016) and a Minor in Computational Neurosciences (2016) from EPFL.

My 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. I investigates data-efficient methods that build on geometric spaces and exploit the geometric information naturally arising in robotic data. My 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.

News!

[Jan. 2024] We got 4 papers accepted in ICRA! Great works by Tilman on incremental learning of movement primitives, Jienfeng on visual imitation learning, Andre on human-like motion retargeting, and a very nice collaboration with Leonel on the "single tangent space fallacy"!
[Jan. 2024] The second edition of our tutorial Riemann and Gauss meet Asimov: A Tutorial on Geometric Methods in Robot Learning, Optimization and Control will take place at ICRA'24! More infos coming soon!
[Dec. 2023] New preprint on the challenges and promises of transfer learning in robotics resulting from a cool collaboration with Michael Welle and many others! Check it here.
[Oct. 2023] New preprint unraveling the single tangent space fallacy when applying Riemannian geometry in robot learning, explaining why and how to avoid it! Check it here.
[Oct. 2023] Great news! Our IROS paper "On the Design of Region-Avoiding Metrics for Collision-Safe Motion Generation on Riemannian Manifolds" was selected as a finalist for the IROS Best Paper Award on Mobile Manipulation!
[June 2023] Two papers accepted to IROS! Great works by Holger on collision-avoidance metrics and by Andre on bimanual segmentation! Preprints available here.
[June 2023] Jianfeng's paper on keypoints-based visual imitation learning was accepted to T-RO! Read it here.
[April 2023] The German Federal Ministry of Education and Research and German Informatics Society awarded me with the title of AI newcomer 2023 of Technical and Engineering Science! I am truly honored that the committee saw value in my work on geometric methods for robot learning. Thank you all for your supporting votes! You can check the KIT press release here.
[Feb. 2023] We released the recordings of our IROS'22 tutorial Riemann and Gauss meet Asimov: A Tutorial on Geometric Methods in Robot Learning, Optimization and Control! You can watch them here!
[Jan. 2023] I was shortlisted as AI newcomer by the German Informatics Society! You can support me with your vote!
[Oct. 2022] We have a new preprint that brings the hyperbolic manifold to robotics applications! This was a nice collaboration with Leonel, Miguel, Slava, and Tamim! Check our paper here!
[July 2022] Our blue sky paper on Riemannian geometry as a unifying theory for robot motion learning and control was accepted to ISRR! Check it here.
[July 2022] One paper accepted to IROS! This is a nice work by Holger on human motion analysis and transfer using Riemannian methods! Check it here.
[June 2022] I am co-organizing the tutorial Riemann and Gauss meet Asimov: A Tutorial on Geometric Methods in Robot Learning, Optimization and Control at IROS 2022! More information here!
[June 2022] Our paper on sequencing and blending robot skills via differentiable optimization was accepted to RA-L! Check it here.
[Oct. 2021] I am giving a lecture on Riemannian methods for robot learning at KIT! It covers topics at the intersection of robotics and machine learning with geometry, and discusses Riemannian-based methods in the context of robot learning. More information (for KIT students) here.
[Oct. 2021] One paper accepted to CoRL! Check it here.
[April. 2021] I am co-organizing the workshop on Geometry and Topology in Robotics: Learning, Optimization, Planning, and Control at R:SS 2021! More information here.
[Oct. 2020] We got one paper accepted in NeurIPS'20! Check it here!
[July 2020] I succesfully defended my PhD thesis entitled Robot skills learning with Riemannian manifolds: Leveraging geometry-awareness in robot learning, optimization and control ! My thesis is available here.
[July 2020] I am co-organizing the workshop on Bringing geometric methods to robot learning, optimization and control at IROS 2020! More information here.
[July 2020] We got one paper accepted in IJRR and two papers accepted in IROS'20! The preprints are available here.
[Dec. 2019] I received the Idiap PhD student award 2019!
[Nov. 2019] Our paper Bayesian Optimization meets Riemannian Manifolds in Robot Learning received the best presentation award in CoRL 2019!

Research

Geometry-aware learning and control in robotics

Many relevant robotic parameters and data have particular geometric properties. For example, manipulability ellipsoids, stiffness and damping matrices, as well as inertia matrices and some control parameters are instances of symmetric positive definite (SPD) matrices and lie in a particular manifold. As an other common example, orientations are usually represented with quaternions, represented as points in the surface of a high-dimensional sphere. Moreover, in sensing applications, data may often be represented in the form of multidimensional arrays. I believe that considering the underlying structure and geometry of these parameters can be beneficial in many robotics applications in terms of accuracy, stability, data-efficiency and scalability.

Relevant robotic parameters belong to Riemannian manifolds. For example, the quaternions belong the sphere manifold and stiffness matrices to the manifold of symmetric positive definite matrices.


Geometry-aware Bayesian Optimization (GaBO) to minimize the Ackley function on the sphere. The Gaussian process (GP), the surrogate model of BO, is represented on the surface of the sphere (top) and in 2D-projections of the sphere (bottom).

Geometry-aware Bayesian optimization

Control policies often need to be refined and adapted for the robot to cope with uncertainties in the task or to generalize to unseen situations. Bayesian optimization (BO) has gained increasing interest in robotics due to its data-efficiency in the learning process. In our work, we propose to include geometry-awareness in the BO framework to optimize non-Euclidean parameters, such as orientations and stiffness matrices, in robotic tasks. This approach aims at improving the performance of BO in terms of convergence rate, accuracy and solution variance.




Manipulability learning, tracking and transfer

Body posture greatly influences human performance when carrying out manipulation tasks. Adopting an appropriate pose helps us regulate our motion and strengthen our capability to achieve a given task. Similarly, in robotic manipulation, the robot joint configuration affects both the ability to move freely in all directions in the workspace and the capability to generate forces along different axes.

Examples of manipulability for different postures in pushing and pulling tasks.

Manipulability transfer. A manipulability trajectory is kinesthetically teached to a Baxter robot. A Franka robot is then able to reproduce the task.

In this context, manipulability ellipsoids are posture-dependent measures indicating the preferred directions to perform velocity, force or dynamic control commands. We propose to transfer manipulability-based posture variation between robots or from human to robot: a teacher (robot or human) demonstrates how to perform a task with a desired manipulability profile and a learner robot reproduces the task while matching the learned manipulability profile. This approach allows to transfer posture-dependent task requirements such as preferred directions for motion and force exertion. We encode, retrieve and track robot manipulability ellipsoids using geometry-aware learning and tracking frameworks based on Riemannian geometry and tensor formulation.


Publications

2024

Preprints

M. Braun, N.Jaquier, L. Rozo, and T. Asfour. Riemannian Flow Matching Policy for Robot Motion Learning, arXiv preprint, 2024.
[pdf] [bibtex]

N.Jaquier*, M. Welle*, Andrej Gams, Kunpeng Yao, Bernardo Fichera, Aude Billard, Ales Ude, Tamim Asfour, and Danica Kragic. Transfer Learning in Robotics: An Upcoming Breakthrough? A Review of Promises and Challenges, arXiv preprint, 2023.
[pdf] [bibtex]

N.Jaquier, L. Rozo, M. Gonzalez-Duque, V. Borovitskiy, and T. Asfour. Bringing Motion Taxonomies to Continuous Domains via GPLVM on Hyperbolic Manifolds, arXiv preprint, 2022.
[pdf] [bibtex]


Conference papers

N.Jaquier*, L. Rozo*, and T. Asfour. Unraveling the Single Tangent Space Fallacy: An Analysis and Clarification for Applying Riemannian Geometry in Robot Learning, in IEEE Intl. Conf. on Robotics and Automation (ICRA), 2024.
[pdf] [bibtex]

T. Daab, N. Jaquier, C. Dreher, A. Meixner, F. Krebs, and T. Asfour. Incremental Learning of Full-Pose Via-Point Movement Primitives on Riemannian Manifolds, in IEEE Intl. Conf. on Robotics and Automation (ICRA), 2024.
[pdf] [bibtex]

J. Gao, X. Jin, F. Krebs, N. Jaquier, and T. Asfour. Bi-KVIL: Keypoints-based Visual Imitation Learning of Bimanual Manipulation Tasks, in IEEE Intl. Conf. on Robotics and Automation (ICRA), 2024.

A. Meixner, M. Carl, F. Krebs, N. Jaquier, and T. Asfour. Towards Unifying Human-Likeness: Evaluating Metrics for Human-Like Motion Retargeting on Bimanual Manipulation Tasks, in IEEE Intl. Conf. on Robotics and Automation (ICRA), 2024.


Invited talks

26.01 Robotics Institute Seminar Serie, University of Toronto, Canada.

23.02 IEEE RAS TC on Model-Based Optimization for Robotics Seminar Serie, online.


2023

Journal papers

J. Gao, Z. Tao, N.Jaquier, and T. Asfour. K-VIL: Keypoints-based Visual Imitation Learning, IEEE Transactions on Robotics, 39(5), pp.3888-3908, 2023.
[pdf] [bibtex] [webpage and video] [code]


Conference papers

H. Klein, N.Jaquier, A. Meixner, and T. Asfour. On the Design of Region-Avoiding Metrics for Collision-Safe Motion Generation on Riemannian Manifolds, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2023. Finalist for the IROS Best Paper Award on Mobile Manipulation sponsored by OMRON Sinic X Corp.
[pdf] [bibtex] [video]

A. Meixner, F. Krebs, N.Jaquier, and T. Asfour. An Evaluation of Action Segmentation Algorithms on Bimanual Manipulation Datasets, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2023.
[pdf] [bibtex] [video]


Invited talks

17.03 Robotics, Perception and Learning Lab, KTH, Sweden.

10.05 Ninja Spring School , Bielefeld, Germany.

02.06 Workshop on Geometric Representations: The Roles of Screw Theory, Lie Algebra, and Geometric Algebra, ICRA 2023.

02.06 Workshop on Transferability in Robotics, ICRA 2023.

02.06 2nd Workshop on Compliant Robot Manipulation: Challenges and New Opportunities, ICRA 2023.

14.06 Research Group Differential Geometry, KIT, Germany.

29.08 Gepetto Team, LAAS Toulouse, France.

27.11 AUTOLab, UC Berkeley, USA.

30.11 IPRL Lab, Stanford University, USA.


2022

Journal papers

N.Jaquier, Y. Zhou, J. Starke, and T. Asfour. Learning to Sequence and Blend Robot Skills via Differentiable Optimization, IEEE Robotics and Automation Letters, 7(3), pp.8431-8438, 2022.
[pdf] [bibtex] [video] [code]


Conference papers

N.Jaquier and T. Asfour. Riemannian geometry as a unifying theory for robot motion learning and control, in International Symposium on Robotics Research (ISRR) - Blue sky track, 2022.
[pdf] [bibtex] [video]

H. Klein, N.Jaquier, A. Meixner, and T. Asfour. A Riemannian Take on Human Motion Analysis and Retargeting, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), pp.5210-5217, 2022.
[pdf] [bibtex] [webpage and video]


Invited talks

05.10 AI Hub @ Karlsruhe, Germany.

16.11 Learning and Adaptive Systems group , Department of Computer Science, ETH Zurich, Switzerland.

15.12 Geometry, Physics, and Human Knowledge as Inductive Bias in Robot Learning Workshop, Conference on Robot Learning, 2022.


2021

Journal papers

N. Jaquier, L. Rozo, D. G. Caldwell and S. Calinon. Geometry-aware Manipulability Learning, Tracking and Transfer, International Journal of Robotics Research (IJRR), 20:2-3, pp.624-650, 2021.
[pdf] [bibtex] [webpage] [code]

N. Jaquier, R. Haschke and S. Calinon. Tensor-variate Mixture of Experts for Proportional Myographic Control of a Robotic Hand, Robotics and Autonomous Systems, 142, 2021.
[pdf] [bibtex] [webpage] [code]


Conference papers

N. Jaquier*, V. Borovitskiy*, A. Smolensky, A. Terenin, T. Asfour, and L. Rozo. Geometry-aware Bayesian Optimization in Robotics using Riemannian Matérn Kernels, in Conference on Robot Learning (CoRL), 2021.
[pdf] [bibtex] [code] [video]


Invited talks

25.11 Secondmind, Cambridge, United Kingdom. Talk title: Bayesian optimization on Riemannian manifolds for robot learning.


2020

PhD thesis

N. Jaquier. Robot skills learning with Riemannian manifolds: Leveraging geometry-awareness in robot learning, optimization and control, PhD thesis, Ecole Polytechnique Fédérale de Lausanne (EPFL), 2020.
Nominated for the Asea Brown Boveri Ltd. Award.
[pdf]


Conference papers

N. Jaquier and L. Rozo. High-dimensional Bayesian Optimization via Nested Riemannian Manifolds, in Conference on Neural Information Processing Systems (NeurIPS), 2020.
[pdf] [bibtex] [code]

N. Jaquier, L. Rozo and S. Calinon. Analysis and Transfer of Human Movement Manipulability in Industry-like Activities, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), pp.11131-11138, 2020.
[pdf] [bibtex] [webpage]

H. Girgin, E. Pignat, N. Jaquier and S. Calinon. Active Improvement of Control Policies with Bayesian Gaussian Mixture Model, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), pp. 5395-5401, 2020.
[pdf] [bibtex]


2019

Conference papers

N. Jaquier, L. Rozo, S. Calinon and M. Bürger. Bayesian Optimization Meets Riemannian Manifolds in Robot Learning, In Conference on Robot Learning (CoRL), 2019, (Oral presentation - Best presentation award ).
[pdf] [bibtex] [webpage] [code] [ presentation ]

N. Jaquier, D. Ginsbourger and S. Calinon. Learning from demonstration with model-based Gaussian process, In Conference on Robot Learning (CoRL), 2019.
[pdf] [bibtex] [webpage] [code]


Invited talks

13.07 Festival Scientastic, EPFL campus Valais-Wallis, Switzerland

19.05 Minisymposium on Algebraic Geometry for Kinematics and Dynamics in Robotics, SIAM conference on Applied Algebraic Geometry, Bern, Switzerland

2018

Journal papers

N. Jaquier and S. Calinon. Improving the control of prosthetic hands with tactile sensing, Micro & Nano Magazine, Micronarc.
[pdf] [bibtex]


Conference papers

N. Jaquier*, L. Rozo*, D. G. Caldwell and S. Calinon. Geometry-aware Tracking of Manipulability Ellipsoids, in Robotics: Science and Systems (R:SS), 2018.
[pdf] [bibtex] [webpage] [video] [code]


Workshop papers

N. Jaquier, L. Rozo and S. Calinon. Geometry-aware Robot Manipulability Transfer, in Learning and Inference in Robotics: Integrating Structure, Priors and Models (LAIR) Workshop at Robotics: Science and Systems Conference (R:SS), 2018.
[pdf] [bibtex]

N. Jaquier and S. Calinon. Geometry-aware Control and Learning in Robotics, in Robotics: Science and Systems (R:SS) Pioneers Workshop, 2018.
[pdf] [bibtex]


Invited talks

22.11 Valais/Wallis AI Workshop - AI for Rehabilitation and Prosthetics, HES-SO Valais, Switzerland


2017

Journal papers

N. Jaquier, M. Connan, C. Castellini and S. Calinon. Combining electromyography and tactile myography to improve hand and wrist activity detection in prostheses, Technologies, 5:4, Special issue on assistive robotics.
[pdf] [bibtex] [code]

Conference papers

N. Jaquier and S. Calinon. Gaussian mixture regression on symmetric positive definite matrices manifolds: Application to wrist motion estimation with sEMG, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), pp.59-64, 2017.
[pdf] [bibtex]

L. Rozo, N. Jaquier, S. Calinon and D. G. Caldwell. Learning manipulability ellipsoids for task compatibility in robot manipulation, in IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), pp.3183-3189, 2017.
[pdf] [bibtex] [webpage]



N. Jaquier, C. Castellini and S. Calinon. Improving hand and wrist activity detection using tactile sensors and tensor regression methods on Riemannian manifolds, in Myoelectric control (MEC) Symposium, 2017.
[pdf] [bibtex]


Invited talks

28.11 Swiss Machine Learning Day, EPFL, Switzerland

14.11 Applied Mathematics Seminar, Applied Mathematic Dept., UCL, Belgium

24.09 Workshop on Micro-data: the next frontier in robot learning?, IROS 2017, Vancouver, Canada

23.05 Numerical Analysis Seminar. Mathematic Dept., University of Geneva

31.01 Rapid-fire talk. Applied Machine Learning Days, EPFL, Switzerland

20.01 Operating room technicians class, Upper School of the Health (ES Santé), Lausanne, Switzerland

Awards

Finalist for the IROS Best Paper Award on Mobile Manipulation sponsored by OMRON Sinic X Corp

For our paper On the Design of Region-Avoiding Metrics for Collision-Safe Motion Generation on Riemannian Manifolds.

AI newcomer 2023 of Technical and Engineering Science

The Federal Ministry of Education and Research (BMBF) and German Informatics Society (GI) honored ten outstanding researchers (doctoral or postdoc phase) who are promoting the development of artificial intelligence in Germany with their innovative ideas. © BMBF / Christina Czybik.


PhD thesis nominated for the EPFL Asea Brown Boveri Ltd. Award

The prize is awarded for particularly excellent master’s and/or doctoral work in the fields of energy and information technology and automation technology.

Idiap PhD Student award 2019

At the end of each year, the Idiap Research Institute awards a research award to a PhD student of the institute.


Best Presentation award at CoRL 2019

for our paper Bayesian Optimization Meets Riemannian Manifolds in Robot Learning


Selected as a Robotics: Science and Systems (R:SS) 2018 Pioneer

R:SS Pioneers is a day-long invitation-only workshop for senior graduate students and postdocs, that seeks to bring together a cohort of the world's top early career researchers in all areas of robotics.

Past projects

Master project: Improving the drawing skills of a humanoid robot with visual feedback
(Idiap Research Institute)

I developed controllers for the humanoid robot Baxter to reproduce a simple image with pen strokes while exploiting the compliance of the arms of the robot and the visual feedback provided by its cameras.

Travail de maturité: Le pendule double, un système chaotique (Gymnase Intercantonal de la Broye)

Ce travail présente une analyse détaillée d'un cas particulier du système du pendule double, visant à démontrer la présence de chaos dans le système. Les équations caractérisant le pendule double sont établies, puis appliquées. Les méthodes de résolution numériques des équations différentielles sont présentées en détails. On observe ensuite la présence de chaos grâce à différents diagrammes illustrant la forte sensibilité aux conditions initiales, et finalement le chaos est quantifié pour des énergies différentes par le calcul de l'exposant de Lyapunov.

Contact

Noémie Jaquier
Karlsruhe Institute of Technology (KIT)
Institute for Anthropomatics and Robotics (IAR)
High Performance Humanoid Technologies (H²T)
Adenauerring 2
76131 Karlsruhe
Germany

Email: noemie.jaquier@kit.edu