Brain-Inspired Spiking Neural Network Controller for a Neurorobotic Whisker System

Abstract

It is common for animals to use self-generated movements to actively sense the surrounding environment. For instance, rodents rhythmically move their whiskers to explore the space close to their body. The mouse whisker system has become a standard model for studying active sensing and sensorimotor integration through feedback loops. In this work, we developed a bioinspired spiking neural network model of the sensorimotor peripheral whisker system, modeling trigeminal ganglion, trigeminal nuclei, facial nuclei, and central pattern generator neuronal populations. This network was embedded in a virtual mouse robot, exploiting the Human Brain Project's Neurorobotics Platform, a simulation platform offering a virtual environment to develop and test robots driven by brain-inspired controllers. Eventually, the peripheral whisker system was adequately connected to an adaptive cerebellar network controller. The whole system was able to drive active whisking with learning capability, matching neural correlates of behavior experimentally recorded in mice.

Variations on the theme: focus on cerebellum and emotional processing

“…In engineering, the cerebellum has been assimilated to a double forward and inverse controller, capable of predicting system states based on contextual information and previous memory (Ito, 1993(Ito, , 2008.The fact that our understanding of cellular phenomena is sufficient to explain the predictive capabilities of the system is demonstrated by the ability of neuro-robots, embedding a canonical cellular representation of the cerebellar circuit dynamics and mechanisms, to reproduce a wide set of sensory-motor control tasks (Casellato et al, 2015;Antonietti et al, 2016Antonietti et al, , 2018Antonietti et al, , 2022. This same system approach may be applied to cognitive and emotional processing, provided the controller, to which the cerebellar circuit is connected, is appropriately designed and implemented.…”

A theory of cerebellar learning as a spike-based reinforcement learning in continuous time and space

“…Thus, those computational studies considered the cerebellum as an SL machine. Spiking network models of the cerebellum as an SL machine have been used for realtime motor control and online learning of hardware robots [32,33]. Our present cerebellar model will help those robotics studies by providing the capability of RL over SL.…”

“…dp e (t) dt = 0.01a(t), (32) where a(t) ∈ {−1, 1} represents the action the agent takes at time t. Also, negative reward is defined as follows: r(t) = −5 when t = 500 and p e (t) > 0.1, 0 otherwise. (33) The grid size of the PF plane was 16 × 16. The parameters of the state value function, which defined in Equation ( 2), were ν = 200 and V 0 = 2.48.…”

Human Brain Project Partnering Projects Meeting: Status Quo and Outlook

“…CerebNEST , presented by Alessandra M. Trapani, is a bioinspired multiscale modeling of the cerebellar network ( Casellato et al, 2014 ; Geminiani et al, 2019 ), developed using the NEST-simulator in EBRAINS. Recent developments include robotic embodiment and a diffusive plasticity mechanism, making it more realistic and detailed ( Trapani et al, 2021 ; Antonietti et al, 2022 ).…”