research

We developed a new cognitive architecture based on predictive coding and free-energy minimization for learning and re-creating actively memory sequences robustly, even with intrinsic noise or external perturbation.

The neural architecture is called INFERNO for Iterative Free-Energy Optimization for Recurrent Neural Networks. By learning the cause-effect relationship in a neural network, the architecture regenerates the specific dynamics learned in its memory by active inference. It is used on vocal learning and handwriting to learn the sensorimotor primitives.

current Phd student : Louis Annabi

Article Source: Iterative free-energy optimization for recurrent neural networks (INFERNO)

Alexandre Pitti, Mathias Quoy, Sofiane Boucenna, Catherine Lavandier. Brain-inspired model for early vocal learning and correspondence matching using free-energy optimization. PLOS Computational Biology 17(2): e1008566

Louis Annabi, Alexandre Pitti, Mathias Quoy. Autonomous learning and chaining of motor primitives using the Free Energy Principle. IJCNN 2020. preprint ⟨hal-02567225⟩

Alexandre Pitti, Mathias Quoy, Catherine Lavandier, Sofiane Boucenna. Gated spiking neural network using Iterative Free-Energy Optimization and rank-order coding for structure learning in memory sequences (INFERNO GATE). Neural Networks. 2020. 121, 242-258. preprint ⟨hal-02140049⟩

Pitti A, Gaussier P, Quoy M (2017) Iterative free-energy optimization for recurrent neural networks (INFERNO). PLOS ONE 12(3): e0173684. doi: 10.1371/journal.pone.0173684

Pitti A., Braud R., Mahé S., Quoy M. & Gaussier P., Neural Model for Learning-to-Learn of Novel Task Sets in the Motor Domain, Frontiers in Psycholology, August 2013, vol.4, no.771, doi: 10.3389/fpsyg.2013.00771.

Pitti A. & Kuniyoshi Y. (2011) Modeling the Cholinergic Innervation in the Infant Cortico-Hippocampal System and its Contribution to Early Memory Development and Attention. Proc. of the International Joint Conference on Neural Networks IJCNN 2011, pp.1409-1416.

The sense of touch is considered as an essential feature for robots in order to improve the quality of their physical and social interactions.

For instance, tactile devices have to be fast enough to interact in real time, robust against noise to process rough sensory information as well as adaptive to represent the structure and topography of a tactile sensor itself – i.e. the shape of the sensor surface and its dynamic resolution. In this paper, we conducted experiments with a self-organizing map neural network that adapts to the structure of a tactile sheet and spatial resolution of the input tactile device; this adaptation is faster and more robust against noise than image reconstruction techniques based on electrical impedance tomography. Other advantages of this bio-inspired reconstruction algorithm are its simple mathematical formulation and the ability to self-calibrate its topographical organization without any a priori information about the input dynamics. Our results show that the spatial patterns of simple and multiple contact points can be acquired and localized with enough speed and precision for pattern recognition tasks during physical contact.

current PhD student Mehdi Abdelwahed

Mehdi Abdelwahed;Alexandre Pitti;Olivier Romain;Fethi Ben Ouezdou (2020) Use of Multi-frequency Electrical Impedance Tomography as Tactile Sensor for Material Discrimination. 5th International Conference on Advanced Robotics and Mechatronics (ICARM).

former PhD student Ganna Pugach

Spatio-Temporal Tolerance of Visuo-Tactile Illusions in Artificial Skin by Recurrent Neural Network with Spike-Timing-Dependent Plasticity Scientific Reports 7, article number 41056 (2017)

Pugach, Melnyk, Tolochko, Pitti, Gaussier Touch-Based Admittance Control of a Robotic Arm Using Neural Learning of an Artificial Skin IEEE IROS 2016, 3374-3380

Pugach G., Pitti A., Gaussier P., Neural learning of the topographic tactile sensory information of an artificial skin through a self-organizing map, Advanced Robotics, 2015, 29, 21, 1393-1409

Pugach, G. Khomenko V., Pitti A., Melnyk A., Henaff P. & Gaussier P. (2013) Electronic hardware design of a low cost tactile sensor device for physical Human-Robot Interactions. IEEE XXXIII International Scientific Conference Electronics and NanoTechnology, Kiev : Ukraine, pp.297-302

<- Body image construction using Gain-Fields neurons GIF Files at: https://perso-etis.ensea.fr/alexpitt/PPS_files/

We are working on the neural mechanisms underlying visuomotor integration for reaching and body schema, affordances learning, the emergence of tool-use in infants development for robots.

current Phd student Julien Abrossimoff

former Phd student Raphael Braud

Abrossimoff, J. Pitti, A. & Gaussier, P. Visual Learning for Reaching and Body-Schema with Gain-Field Networks, IEEE ICDL EPIROB conference, Tokyo 2018

Gaussier, P. & Pitti, A. Reaching and Grasping: what we can learn from psychology and robotics, book chapter 14, ed. Corbetta and Santello, Reach-to-Grasp Behavior, Brain, Behavior, and Modelling Across the Life Span, Collection: Frontiers of Developmental Science, Routledge, 2018

Braud, R. Pitti, A. Gaussier, P. A modular Dynamic Sensorimotor Model for affordances learning, sequences planning and tool-use IEEE Transactions on Cognitive and Developmental Systems vol 10, number 1, 2018

Mahé S., Braud R., Gaussier P., Quoy M. & Pitti A., Exploiting the gain-modulation mechanism in parieto-motor neurons: Application to visuomotor transformations and embodied simulation, Neural Networks, Jan 2015, vol.62, no.1, pp.102-111

Pitti A., Mori, H., Kouzuma, S. & Kuniyoshi Y., Contingency Perception and Agency Measure in Visuo-Motor Spiking Neural Networks, IEEE Trans. on Autonomous Mental Development, Jan 2009, vol.1, no.1, pp. 86-97.

Pitti A., Alirezaei H. & Kuniyoshi Y., Cross-modal and scale-free action representations through enaction, Neural Networks. Special issue "What it means to communicate", Feb 2009, vol.22, no.2, pp. 144-154.

Advanced Robotics 2018

"Feedback-Driven Oscillatory Control of a Tensegrity-based Vertebral Column Robot for Postural Balance"

Artem Melnyk, Alex Pitti

Topic: In comparison to animals, current robots are not very soft and their structure is still fragile.

We propose to exploit the architectural design principle known as Tensegrity to build Soft Robots lightweight, robust, compliant and fun to build!

We started with a Vertebral Column architecture controlled by Central Pattern Generators that can stand upward and balance to absorb shocks. [more info on the control part in the publications]

DIY:

Tensegrity structures are nice educational tools too, we provide the 3D components in Tinkercad, https://www.tinkercad.com/things/6iBMNx721aK-tensegrity10

enjoy and please share with us your comments, prototypes and ideas!

Melnyk, A. & Pitti, A. Synergistic Control of a Multi-Degrees of Freedom Vertebral Column Robot based on Tensegrity for Postural Balance Advanced Robotics, 2018

Pitti A., Niiyama, R. & Kuniyoshi Y., Creating and modulating rhythms by controlling the physics of the body, Autonomous Robots, 2010, 28, 3, 317-329

Pitti A., Lungarella, M. & Kuniyoshi Y., Generating Spatiotemporal Joint Torque Patterns from Dynamical Synchronization of Distributed Pattern Generators, Frontiers in NeuroRobotics, Feb 2009, vol.3, no.2, pp. 1-14

Pitti, A. Lungarella M. & Kuniyoshi Y. (2006) Exploration of natural dynamics through resonance and chaos. Proc. of the 9th Int. Conf. on Intelligent Autonomous Systems, pp.558-565.

Pitti, A. Lungarella M. & Kuniyoshi Y. (2005) Quantification of emergent behaviors induced by feedback resonance of chaos. Recent Advances in Artificial Life: Advances in Natural Computation, vol.3, chap.15, pp.199-213

AI at the service of Society to improve learning capabilities of cognitively impaired infants.

En collaboration avec l'Institut Médico-Educatif d'Ennery.

Nasir, M. Fellus, L. & Pitti, A. SPEAKY Project: Adaptive Tutoring System based on Reinforcement Learning for Driving Exercizes and Analysis in ASD Children IEEE ICDL EPIROB conference, Tokyo 2018, Workshop on “Understanding Developmental Disorders: From Computational Models to Assistive Technologies”