System simulating a decisional process in a mammal brain about motions of a visually observed body

Patent metadata

Field	Value
Publication number	US-12561548-B2
Application number	US-202017632718-A
Country	US
Kind code	B2
Filing date	Aug 3, 2020
Priority date	Aug 5, 2019
Publication date	Feb 24, 2026
Grant date	Feb 24, 2026

Abstract

A system simulating a decisional process in a mammal brain about characteristics of motions related to body gestures of a visually observed body through a simulated visual path is provided. The system includes an interface toward simulated neuronal structures, the interface at least converting luminous information of the observed body to an optic flow data stream conveying information related to the visually observed body and that can be processed in the simulated neuronal structures, the system being a feed-forward system and comprising hierarchically from the visual observation to the decision: the simulated visual path and its interface, a simulated local motion direction detection neuronal structure for the detection of motion directions with receptive fields, a simulated opponent motions detection neuronal structure, a simulated complex patterns detection neuronal structure, and a simulated motion pattern detection neuronal structure.

First claim

The invention claimed is:

1 . A computer-based system simulating a mammal brain decisional process related to characteristics of body gesture motions of a visually observed body through a simulated visual path, the system comprising:
neuronal circuitry comprising hardware-simulated neurons, and the simulated visual path comprising an interface to the neuronal circuitry, the simulated visual path and the interface being configured to convert luminous information of the visually observed body to an optic flow data stream conveying visual observation information related to the visually observed body, the optic flow data stream being processed in the neuronal circuitry, the system being a feed-forward system and comprising hierarchically, from the visual observation to a decision, the following layers: the simulated visual path and the interface, which includes hardware simulating an eye and an eventual ophthalmic apparatus on said eye, simulated local motion direction detection neuronal circuitry detecting motion directions with receptive fields, and receiving a first optic flow data stream from the interface, simulated opponent motions detection neuronal circuitry detecting opponent motions related to at least expansion and contraction, and receiving a second optic flow data stream from the simulated local motion direction detection neuronal circuitry, simulated complex patterns detection neuronal circuitry detecting optic flow patterns globally on a whole visual observation and according to evolution during a time of the whole visual observation, and receiving a third optic flow data stream from the simulated opponent motions detection neuronal circuitry, the detectable patterns being prototypical patterns, and simulated motion pattern detection neuronal circuitry detecting motion patterns, receiving a fourth optic flow data stream from the simulated complex patterns detection neuronal circuitry, and providing decisions about characteristics of motions, wherein neurons of the simulated motion pattern detection neuronal circuitry implement a disremembering capability that is a function of a delay and of an activity of said simulated neuron, wherein the system further comprises an input-output interface allowing input of parameters values modifying at least the delay related to the disremembering capability.

2 . The computer-based system according to claim 1 wherein the simulated opponent motions detection neuronal circuitry further detects opponent motions related to rotation.

3 . The computer-based system according to claim 1 , wherein an internal noise of the system is further simulated in the system, and wherein a noise is added to the fourth optic flow data stream outputted from the simulated complex patterns detection neuronal circuitry.

4 . The computer-based system according to claim 1 , wherein the simulated local motion direction detection neuronal circuitry further detects motion directions in a two-dimensional space between receptive fields, the detectable motion directions being up, down and left, right.

5 . The computer-based system according to claim 2 , wherein the simulated opponent motions detection neuronal circuitry further detects expansions, contractions, clockwise rotations, and counter-clockwise rotations, in a two-dimensional space.

6 . The computer-based system according to claim 1 , wherein the simulated complex patterns detection neuronal circuitry comprises a set of the hardware-simulated neurons detecting prototypical patterns in a two-dimensional space within a global observation angle having a determined value, the simulated neurons of the simulated complex patterns detection neuronal circuitry being allocated to a number of groups of the set of the hardware-simulated neurons, the number of groups corresponding to a number of possible decisions the system is configured to provide, each group comprising simulated neurons asymmetrically and laterally connected together within the group.

7 . The computer-based system according to claim 1 , wherein the simulated motion pattern detection neuronal circuitry is further configured as a mutually inhibited simulated neurons network in which simulated neurons with a highest excitatory input suppress the activity of the other simulated neurons whose activities have not passed over corresponding defined thresholds in nonlinear and reciprocate operations.

8 . The computer-based system according to claim 7 , wherein the mutually inhibited simulated neurons of the simulated motion pattern detection neuronal structure have additional inputs through which the disremembering capability is executed, the additional inputs receiving data according to Dis T =u(t−τ a )*[T−S(P T (D)−K Dis T] for a primarily excited simulated neuron and Dis D =u(t−τ a )*[D−S(P D (D,T)−K Dis D] for the other simulated neurons, and where T is the activity of whichever simulated neuron gets excited first by the fourth optic flow data stream received from the simulated complex patterns detection neuronal structure and D is the activity of the rest of the neurons of the network, u( ) is the unit step function, τ a is a time constant, S( ) is a modified Michaelis-Menten function, or any other function with a same mathematical symmetry as Michaelis-Menten function, and k Dis a weighting coefficient.

9 . The computer-based system according to claim 1 , further comprising a learning device implemented with a risk-sensitive Q-learning algorithm.

10 . The computer-based system according to claim 9 , wherein the learning device is implemented in the simulated complex patterns detection neuronal circuitry.

11 . The computer-based system according to claim 1 , wherein the simulated opponent motions detection neuronal circuitry further detects opponent motions related to rotation,
wherein the simulated opponent motions detection neuronal circuitry further detects expansions, contractions, clockwise rotations, and counter-clockwise rotations, in a two-dimensional space, and wherein the simulated local motion direction detection neuronal circuitry further detects motion directions in a two-dimensional space between receptive fields, the detectable motion directions being up, down and left, right.

12 . The computer-based system according to claim 11 , wherein the simulated complex patterns detection neuronal circuitry comprises a set of the hardware-simulated neurons and is configured to detect prototypical patterns in a two-dimensional space within a global observation angle having a determined value, the simulated neurons of the simulated complex patterns detection neuronal circuitry being allocated to a number of groups of the set of the hardware-simulated neurons, the number of groups corresponding to a number of possible decisions the system is configured to provide, each group comprising simulated neurons asymmetrically and laterally connected together within the group.

13 . The computer-based system according to claim 12 , wherein the simulated motion pattern detection neuronal circuitry is further configured as a mutually inhibited simulated neurons network in which simulated neurons with a highest excitatory input suppress activity of the other simulated neurons whose activities have not passed over corresponding defined thresholds in nonlinear and reciprocate operations.

14 . The computer-based system according to claim 13 , wherein the mutually inhibited simulated neurons of the simulated motion pattern detection neuronal circuitry have additional inputs through which the disremembering capability is executed, the additional inputs receiving data according to Dis T =u(t−τ a )*[T−S(P T

Assignees

• Univ Montreal
• Essilor Int

Inventors

• Faubert Jocelyn
• Misaghian Khashayar
• Lugo Eduardo
• Tranvouez-Bernardin Delphine

Classifications

• G06V40/23

  Recognition of whole body movements, e.g. for sport training · CPC title

• G06N3/10

  Interfaces, programming languages or software development kits, e.g. for simulating neural networks · CPC title

• G06F18/2414 Primary

  Smoothing the distance, e.g. radial basis function networks [RBFN] · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N3/088

  Non-supervised learning, e.g. competitive learning · CPC title