KR101799700B1 - Apparatus and the method for navigation using neuromorphic neural network model - Google Patents

Abstract

The present invention discloses a neuromorphic nerve network model based navigation device, and a method thereof. The neuromorphic nerve network model based navigation device comprises: an input current generation unit inputted with sensing information from a plurality of sensors in order to copy activity of neurons which process space information in the brain and generates input current; a neuromorphic nerve network model unit which includes a plurality of neuron models connected with a synapse model capable of learning by regulating synapse connection intensity, and generating navigation information for autonomous driving by firing a portion of neuron models in accordance with the input current generated by the sensing information. As such, a moving body equipped with the neuromorphic nerve network model based navigation device is able to avoid risk elements and obstacles in an environment based on learning, and is able to realize an autonomous driving navigation which finds a specific destination.

Description

[0001] The present invention relates to a neuromodern neural network model based navigation device and a neuromorphic neural network model,

The present invention relates to an autonomous navigation apparatus and method using a neuromotor neural network model based on brain location information processing principle.

Conventional navigation system based on GNSS (Global Navigation Satellite System) has been used to navigate by obtaining accurate position of moving means by using triangulation method using satellite. However, the conventional navigation system does not operate indoors and has a limit in which it is difficult to obtain an accurate current position due to an error in signal transmission.

The DGNSS (Differential GNSS) for correcting this is corrected through the base station having an absolute position on the ground surface. However, there is the same problem that the positioning error occurs due to a large cost for the base station installation.

In addition, the conventional autonomous navigation system has a disadvantage in that it is practically impossible to navigate in a non-search area. That is, the conventional autonomous navigation system has a disadvantage in that accurate map information must be constructed through a search process for a specific area for navigation.

SUMMARY OF THE INVENTION The present invention has been made in order to overcome the above-mentioned limitations of the prior art, and it is an object of the present invention to provide an autonomous navigation navigation apparatus based on a neuromotor neural network model that simulates the processing of place information processing in an animal (for example, Method.

According to the present invention, based on a learning-enabled neural network model capable of processing information about a moving direction, a moving speed, a distance to an obstacle, and a specific place, a new learning- Lomographic neural network model based navigation apparatus and method thereof can be provided.

According to an aspect of the present invention, there is provided a navigation device based on a neuromodic neural network model.

 According to an embodiment of the present invention, a plurality of sensors; An input current generator for receiving sensing information from at least one of the plurality of sensors and generating an input current; And a plurality of neuron models coupled based on a synapse model, wherein the plurality of neuron models are ignited according to the input current, and the synaptic connection strength with other neuron models connected through the synapse model is adjusted A neuromotor neural network model based navigation device including a neuromaric neural network model unit for generating navigation information can be provided.

The neuronal cell model is at least two of an interfacial cell model, a head-direction cell model, a sustained-activation cell model, a lattice cell model, a place cell model, and an exercise cell model.

The interfacial cell model, the head-direction cell model, and the place cell model are connected to each other through an exercise cell model and the synapse model, and the persistent speech cell model and the lattice cell model are connected through the synapse model. The motile cell model may be connected via the synapse model.

The neuromodern neural network model unit determines whether or not the head-directional cell model is ignited according to the input current according to the sensing information of the angle sensor, and determines whether or not the boundary cell model is ignited according to the input current according to the sensing information of the distance sensor And determines whether or not the continuous ignited cell model is ignited according to the input current according to the sensing information of the angle sensor and the velocity sensor.

Wherein the neuromodern neural network model unit determines whether or not to ignite some of the lattice cells of the lattice cell model in accordance with the utterance of a part of the persistent lenticular cells among the plurality of the persistent lenticular cells included in the persistent lenticular cell model, Activity control information of the placental cell model can be generated by lattice cell firing.

Wherein the neuromodern neural network model unit determines whether or not the motility cell model is ignited according to synaptic stimulation by any one of the interface cell model, the head cell model, and the place cell model, It is possible to determine the rotation direction by a certain angle in accordance with the ignition.

According to another aspect of the present invention, there is provided an autonomously navigable navigation method using a neuromotor neural network model.

According to an embodiment of the present invention, there is provided a method of sensing an input current, the method comprising: receiving at least one sensing information and generating an input current; Determining whether or not to ignite at least one neural cell model among a plurality of neural cell models connected through a synapse model based on the input current; And generating navigation information by adjusting the synaptic connection strength with the neural cell model, which has been uttered, and another neural cell model, which is connected via the synapse model through the synapse model, when the neural cell model is ignited Lt; / RTI > model-based navigation device may be provided.

By providing the navigation device based on the Neumochip neural network model and the method thereof according to the embodiment of the present invention, it is possible to navigate through self learning and judgment using the neuromotive neural network model simulating the processing of the place information processing in the brain There is an advantage to be able to do.

In addition, the present invention simulates the processing of place information processing in the brain of an animal (for example, a rat or a human), and autonomously explores a specific destination in a new environment or a new search area, There is an advantage that navigation can be performed.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram schematically illustrating an internal configuration of a neuromodern neural network model interface apparatus for navigation according to an embodiment of the present invention; FIG.
FIG. 2 is a flowchart illustrating an operation procedure of a neuromodern neural network model interface for navigation according to an embodiment of the present invention; FIG.

As used herein, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise. In this specification, the terms "comprising ", or" comprising "and the like should not be construed as necessarily including the various elements or steps described in the specification, Or may be further comprised of additional components or steps. Also, the terms "part," " module, "and the like described in the specification mean units for processing at least one function or operation, which may be implemented in hardware or software or a combination of hardware and software .

The present invention relates to a navigation apparatus and method for autonomous navigation through learning and judgment based on a neural network model including a plurality of neuron models. To this end, some of the neural cell models included in the neural network model of the present invention are unidirectionally connected through a synapse model, and specific neural cell models are ignited according to input currents based on at least one of the plurality of sensors, The synaptic connection strength to other connected neuronal models can be controlled to generate navigation information for autonomous navigation.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram schematically illustrating a configuration of a navigation device based on a Neuromotor Neural Network model according to an embodiment of the present invention.

Referring to FIG. 1, a navigation device 100 based on a Neuromotor neural network model according to an embodiment of the present invention includes a plurality of sensors 110, an input current generator 115, a neural network model unit 120, a memory 125 And a processor 130. As shown in FIG.

Each sensor 110 is a means for outputting various sensing information related to a moving object.

Here, each sensor 110 is a means for sensing various information related to a mobile body on which the navigation device 100 based on the neuromodern neural network model is mounted or communicatively connected.

Hereinafter, in order to facilitate understanding and explanation, it is assumed that the plurality of sensors 110 are at least one of an angle sensor, a velocity sensor, and a distance sensor. It is natural that each sensor 110 can be applied without limitation if it can sense at least one of the velocity and the angle of the moving object.

The input current generator 115 is a means for generating an input current according to sensing information of at least one of the plurality of sensors 110. Here, the input current generator 115 may generate the input current differently depending on the sensing information and type.

For example, the input current generator 115 may generate the input current according to the distance between the moving object and the interface. Here, the interface may be at least one of a wall, a cliff, and a cliff.

That is, when the distance information is input as the sensing information by the distance sensor, the input current generator 115 can generate the input current according to the distance information.

The sigmoid function is applied to the input current according to the distance information. The sigmoid function is expressed by Equation (1).

는 경계면 신경 세포 모델에 입력되기 위해 생성되는 입력 전류(nA)를 나타내고,

는 최대 입력 전류(nA) 상수를 나타내며,

는 입력 전류의 활성 조절 변수를 나타내고, distance는 이동체의 현재 위치와 특정 방향에 위치한 벽과의 거리를 나타내는 변수이며,

는 시그모이드 함수의 그래프 기울기 조절 상수를 나타내고,

는 시그모이드 함수의 최대값 오프셋 이동 상수를 나타낸다. here,

Represents the input current (nA) generated to be input to the interface neuron model,

Represents the maximum input current (nA) constant,

The distance is a variable indicating the distance between the current position of the moving object and the wall located in a specific direction,

Represents the graph slope control constant of the sigmoid function,

Represents the maximum value offset movement constant of the sigmoid function.

In addition, the input current generator 115 can generate an input current which changes in accordance with the moving direction of the moving object. That is, when the angle information of the moving object is inputted by the angle sensor, the input current generator 115 can generate the input current differently according to the angle information.

A Gaussian function is applied to the generation of the input current according to the angle information.

는 머리방향 신경 세포 모델에 입력되는 입력 전류(nA)를 나타내고,

는 입력 전류의 활성조절 변수를 나타내며,

는 최대 입력 전류(nA) 상수를 나타내고,

는 머리 방향 신경세포의 선호각도(°)와 현재 각도(°)간 차이의 절대값을 나타내며,

는 가우시안 함수의 표준편차를 나타내고,

는 입력전류가 가장 높아지는 머리 방향 신경 세포의 선호각도(°)를 나타내고,

는 이동체의 현재 이동 각도(°)를 나타낸다.here,

Represents the input current (nA) input to the head-oriented neuron model,

Lt; / RTI > represents the active regulating variable of the input current,

Represents the maximum input current (nA) constant,

Represents the absolute value of the difference between the preferred angle (°) and the current angle (°) of the head-oriented neuron,

Denotes the standard deviation of the Gaussian function,

Represents the preferred angle (°) of the head-oriented neuron with the highest input current,

Represents the current moving angle (deg.) Of the moving body.

Also, the input current generator 115 may generate an input current whose activity is controlled according to the moving direction and speed of the moving object.

For example, the input current generator 115 can generate the input current according to the angle information and the velocity information, when the angle information from the angle sensor and the velocity sensor are input as the sensing information, respectively.

This can be expressed by the following equation (3).

는 지속발화세포 모델 입력을 위한 입력 전류(nA)를 나타낸다. Amp는 최대 입력 전류(nA) 상수를 나타내고, VCO는 지속발화세포 활성조정 변수를 나타내고,

는 격자좌표 이동 변수를 나타내며,

는 지속발화세포의 선호 방향(radian)을 나타내고,

는 이동체의 현재 이동 각도(radian)를 나타내며, phase는 지속발화 세포 발화 위상 결정 변수를 나타내고, dt는 시뮬레이션 시간 단위(ms)를 나타내며,

는 로컬 영역 진동 빈도(local field oscillation frequency, Hz)를 나타내고,

는 행간(spacing) 크기 조절 변수를 나타내며, speed는 이동체의 움직임 속도(m/s)를 나타내고, x 및 y는 격자 세포 위치 조절 변수를 나타낸다. here,

Represents the input current (nA) for the sustained ignition cell model input. Amp represents the maximum input current (nA) constant, VCO represents the steady-

Represents a grid coordinate movement variable,

Represents the preferred direction (radian) of the persistent pyrogenic cells,

Phase represents the current moving angle (radian) of the moving object, phase represents the continuous firing cell firing phase determination parameter, dt represents the simulation time unit (ms)

Represents the local field oscillation frequency (Hz)

(M / s), and x and y represent the lattice cell position control variables, respectively.

In FIG. 1, the navigation device 100 based on the neuromodern neural network model according to an embodiment of the present invention includes one input current generator 115, and generates different input currents according to the sensing information of each sensor, And outputs it to the control unit 120. FIG. However, it is a matter of course that the input current generator for each sensor may be separately provided according to the implementation method.

The neural network model unit 120 includes a plurality of neuron models. Some of the plurality of neural cell models included in the neural network model unit 120 are unidirectionally connected through a synapse model.

Accordingly, the neural network modeling unit 120 fires a neural cell model of a plurality of neural cell models based on the input current according to the sensing information by at least one of the plurality of sensors, The synaptic connection strength of other synaptic linked neuronal models can be controlled to generate navigation information for autonomous navigation.

According to one embodiment of the present invention, the neural network model unit 120 may include at least one of an interfacial cell model, a head direction cell model, a persistent speech cell model, a lattice cell model, a place cell model, and a motor cell model. Here, the interfacial cell model, the head cell model, and the place cell model are connected to the motor cell model through the synapse model in a unidirectional manner. The sustained ignition cell model and the lattice cell model are unidirectionally connected through the synapse model, The lattice cell model and the place cell model are connected unidirectionally through the synapse model.

Such a neuron model can be expressed by the following equation (4).

은 세포막의 정전 용량(

)을 나타내고, dt는 시뮬레이션 시간 단위(ms)를 나타내며,

는 시뮬레이션 시간의 흐름에 따른 세포막전위(mV) 변화를 나타낸다. I는 입력 전류(nA)를 나타내고, ion은 이온 개폐통로를 나타내며, inj는 입력 전류 생성기로부터 생성된 입력 전류(nA)를 나타내고, syn는 시냅스 자극을 통해 생성된 시냅스 전류(nA)를 나타내며, s는 시냅스를 통해 연결된 s번째 시냅스 신경세포를 나타내고(

), n은 시냅스를 통해 연결된 신경세포 전체 개수를 나타낸다. here,

The capacitance of the cell membrane (

), Dt represents the simulation time unit (ms)

Shows the change in cell membrane potential (mV) according to the simulation time. Wherein I represents an input current (nA), ion represents an ion switching path, inj represents an input current (nA) generated from an input current generator, syn represents a synapse current (nA) generated through a synapse stimulus, s represents the synaptic neuron connected through the synapse (

), and n represents the total number of neurons connected via the synapse.

For example, when the input current corresponding to the angle information is input, the neural network model unit 120 inputs the input current into the head-direction cell model, and determines whether the head-direction cell model is activated (active) have. That is, the head-direction cell model can determine whether or not to ignite (activity) according to the input current that is input differently according to the angle information. If the head-oriented cell model is ignited earlier than the motor cell model according to the input current, the neural network model unit 120 may generate synaptic stimulation to increase the synaptic connection strength between the head-oriented cell model and the motor cell model.

When the head-oriented cell model is fired after the speech of the exercise cell model according to the input current corresponding to the angle information, the neural network model unit 120 generates a synapse stimulus to lower the synaptic connection strength between the head- can do.

For example, the neural network modeling unit 120 may generate a synaptic stimulus using equation (5).

는 시냅스 자극을 나타내며, t는 시간(ms)을 나타내고, w는 시냅스 연결 강도를 나타내며,

는 시냅스 전 신경 세포가 발화한 시간(ms)을 나타내고,

는 시냅스 후 신경 세포가 발화한 시간(ms)을 나타내며, factor는 정규화 상수를 나타내고,

는 시간 상수를 나타내며, rise는 상승을 나타내고, decay는 하강을 나타낸다.

는 시냅스 연결 강도가 강화되는 조건을 나타내며,

는 시냅스가 강화될 때의 학습 정도 조절 상수를 나타내고, exp는 지수 함수(exponential function)를 나타내며,

는 시냅스가 강화될 때의 시간 상수를 나타낸다.

T represents the time (ms), w represents the synaptic connection strength,

Represents the time (ms) in which the synaptic neuron fires,

Represents the time (ms) after the stimulation of the post-synaptic neuron, factor represents the normalization constant,

Represents the time constant, rise represents the rise, and decay represents the fall.

Indicates a condition in which the synaptic connection strength is enhanced,

Represents the learning degree control constant when the synapse is strengthened, exp represents the exponential function,

Represents the time constant when the synapse is strengthened.

는 시냅스 연결 강도가 약화되는 조건을 나타내며,

는 시냅스가 약화될 때의 학습 정도 조절 상수를 나타내고,

는 시냅스가 약화될 때의 시간 상수를 나타낸다. Also,

Indicates a condition in which the synaptic connection strength is weakened,

Indicates the learning level adjustment constant when the synapse is weakened,

Represents the time constant when the synapse is weakened.

For example, when the input current corresponding to the distance information is input to the neural network modeling unit 120, the neural network modeling unit 120 may input the input current corresponding to the distance information to the interface cell model. The interfacial cell model can be determined to be activated (active) according to the input current corresponding to the distance information. That is, the boundary cell model determines whether to ignite according to the input current that is input differently according to the distance information. That is, the interfacial cell model can be fired in response to an input current that increases with distance from the interface, such as a wall. When the interfacial cell model is ignited earlier than the motor cell model, the neural network model 120 generates a synaptic stimulus between the interfacial cell model and the motor cell model, while the synaptic stimulus increases the synaptic connection strength between the head- Lt; / RTI > Since the generation of the synapse stimulus is the same as the above, the duplicate description will be omitted.

In addition, when the interfacial cell model ignites after the speech of the motor cell model according to the input current, the neural network model unit 120 generates a synaptic stimulus between the interfacial cell model and the motor cell model, and the synaptic stimulus lowers the synaptic connection strength Lt; / RTI >

와 서로 동일한

를 가진 6개의 지속발화세포를 통한 시냅스 자극을 통해 활성화될 수 있다.For example, if the moving direction of the moving object (angle information) and the input current according to the speed information are inputted, the neural network model unit 120 inputs the input current corresponding to the angle information and the speed information to the continuous speech cell model can do. When the continuous ignited cells are ignited by the input current, the neural network model unit 120 activates (ignites) the lattice cell model. Here, one lattice cell model is defined as follows: < EMI ID =

And the same

Lt; RTI ID = 0.0 > 6 < / RTI >

In addition, the neural network modeling unit 120 can determine whether or not the place cell model is active according to the activity of various lattice cells included in the lattice cell model.

This can be expressed by the following equation (6).

는 장소 세포 모델과 격자 세포 모델간의 시냅스 강도를 나타내며, g는 격자세포 활성 정도를 나타내며, G는 격자 세포 집단을 나타낸다. Here, P represents a place cell activity regulation variable, t represents time,

Represents the synaptic strength between the place cell model and the lattice cell model, g represents the degree of lattice cell activity, and G represents the lattice cell population.

As already mentioned above, the neural network modeling unit 120 performs the activity (ignition) of the motor cell model according to the synaptic stimulation caused by the activation (ignition) of either the interfacial cell model, the head cell model or the place cell model Can be adjusted. The rotational direction of the moving object can be determined according to the activity (ignition) of the motility cell model.

The motile cell model includes rotational direction control information of independent moving objects. The motile cell model independently generates information based on the information about the environment acquired through the activity of the interface cell model, the head cell model, and the place cell model, Can be controlled.

For example, if a mobile object approaches a north (90 °) obstacle, it must be rotated in a direction other than north to avoid obstacles. At this time, a boundary cell model that fires at the northern interface of the obstacle, a head-direction cell model having a preferred angle in the head direction of the current mobile body, and an activation of a place- The activity of the motility cell model having the other rotation direction control information is induced and the moving object can continue the movement by avoiding the obstacle by generating the navigation information rotating in the corresponding direction. At this time, the rotation direction control angle of the exercise cell model is defined according to the number of exercise cell models, and the number of exercise cell models can be defined by the rotation direction of the exercise subject to be controlled.

This can be expressed by the following equation (7).

는 이동체의 현재 이동 각도(°)를 나타내고, t는 시간(ms)을 나타내며,

는 i번째 운동 세포 모델이 가진 특정 이동체 회전 방향 제어 각도(°)를 나타내고, n은 전체 운동 세포 모델의 개수를 나타낸다.

Represents the current moving angle (°) of the moving body, t represents the time (ms)

Represents the rotation angle (°) of a specific moving object rotation direction of the i-th movement cell model, and n represents the number of total movement cell models.

The memory 125 stores various applications, algorithms, various data derived from the process, and the like required for operating the navigation device based on the neuroma-based neural network model for destination search according to an embodiment of the present invention.

The processor 130 may include internal components (e.g., a sensor 110, an input current generator 115, a neural network model 120) of the novel Lomographic network model navigation device 100 according to an embodiment of the present invention ), The memory 125, and the like).

FIG. 2 is a flowchart illustrating an autonomous navigation navigation method using a Neumochick neural network model according to an embodiment of the present invention.

In step 210, the novel Lomographic model-based navigation device 100 receives at least one sensing information. Here, the sensing information may be at least one of distance information, angle information, and speed information, as described above.

In step 215, the novel Lomographic model-based navigation device 100 generates an input current according to the sensing information. In generating the input information according to the sensing information, the navigation device 100 based on the neuroma-based neural network model can generate input currents differently according to the type of sensing information as described above.

Since this is the same as described above, redundant description will be omitted.

In step 220, the neuromodern neural network model based navigation device 100 determines whether or not the neural cell model corresponding to the sensor type is ignited according to the input current.

More specifically, the neuromodic neural network model-based navigation device 100 can input the input current to the corresponding neuron model according to each sensor type.

For example, if the sensing information is distance information, the novel Lomographic model-based navigation device 100 can input the input current corresponding to the distance information to the boundary surface cell model. On the other hand, if the sensing information is angle information, the navigation device 100 based on the neuroma-pixel neural network model can input the input current corresponding to the angle information into the head-directional cell model. If the sensing information is angle information and velocity information, the navigation device 100 based on the neuroma-based neural network model can input the input current corresponding to the angle information and the velocity information to the persistent speech cell model.

Each neuron model included in the neuromodern neural network model-based navigation device 100 can be determined whether to ignite according to the input current. This is the same as that described with reference to FIG. 1, so duplicate descriptions will be omitted.

If the particular neural cell model is ignited, the neuromodern neural network model based navigation device 100, at step 225, determines the neural cell model associated with the neural cell model (e. G., Place cell model or exercise Lt; / RTI > cell model) to create synaptic stimulation for synaptic link strength regulation.

As described above, the interfacial cell model and the motor cell model are unidirectionally connected through a synapse model, and the head cell model and the motor cell model are unidirectionally connected through the synapse model.

In addition, the sustained ignition cell model and the lattice cell model are unidirectionally connected through the synapse model. The lattice cell model and the place cell model are unidirectionally connected through the synapse model. The place cell model and the exercise cell model are synapse They are connected in a single direction through the model.

For example, when the interfacial cell model is ignited, the neuromodern neural network model based navigation device 100 can generate a synaptic stimulus for controlling the synaptic connection strength between the interfacial cell model and the motor cell model. Since this is the same as that described with reference to Equation (4), redundant description will be omitted.

In addition, once the headed cell model has been ignited, the neuromodel neural network model based navigation device 100 can generate a synaptic stimulus for controlling the synaptic connection strength between the head-oriented cell model and the motor cell model.

In addition, when the sustained ignition cell model is ignited, a synaptic stimulus is generated for controlling the synaptic connection strength with the lattice cell model. When the lattice cell model is ignited, the neuromodern neural network model-based navigation device 100 generates a lattice cell model It is possible to generate place cell model activity modulation information corresponding to the firing of the various lattice cells involved. Since this is the same as that described with reference to Equation (5), the duplicated description will be omitted.

Once the place cell model is activated, the neuromodern neural network model based navigation device 100 can generate a synaptic stimulus for controlling the synaptic link strength between the place cell model and the motor cell model.

However, if the particular neuronal model is not ignited, proceed to step 210.

In step 230, the neuroma-based neural network model-based navigation apparatus 100 determines whether or not an exercise cell model has been ignited according to synaptic stimulation.

If the kinetic cell model is ignited, the neuromodern neural network model-based navigation apparatus 100 transmits navigation information for controlling the direction of rotation of the moving object on which the novel Lomographic model-based navigation apparatus 100 is mounted .

However, if the kinetic cell model has not been ignited, at step 240, the neuromodern neural network model based navigation device 100 determines whether the lattice cell model has been ignited according to the synaptic stimulation.

If the lattice cell model has not been ignited, proceed to step 210.

However, if the lattice cell model is ignited, the neuromodal neural network model based navigation device 100 in step 245 generates activity control information of the place cell model according to the lattice cell model utterance information.

In step 250, the novel Lomographic network model based navigation device 100 determines whether the place cell model has been ignited.

If the place cell model has been ignited, then at step 255, the neuromodern neural network model based navigation device 100 generates a synaptic stimulus between the place cell model and the motor cell model and proceeds to step 230.

On the other hand, the components of the above-described embodiment can be easily grasped from a process viewpoint. That is, each component can be identified as a respective process. Further, the process of the above-described embodiment can be easily grasped from the viewpoint of the components of the apparatus.

In addition, the above-described technical features may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

It will be apparent to those skilled in the art that various modifications, additions and substitutions are possible, without departing from the spirit and scope of the invention as defined by the appended claims. Should be regarded as belonging to the following claims.

100: Neuromotor neural network model-based navigation device
110: sensor
115: Input current generator
120: Neuromodic neural network model part
125: Memory
130: Processor

Claims (7)

A plurality of sensors;
An input current generator for receiving sensing information from at least one of the plurality of sensors and generating an input current; And
A plurality of neuron models connected based on a synapse model, wherein the plurality of neuron models are ignited according to the input current, and the synaptic connection strength with another neuron model connected through the synapse model is adjusted A novelogram network model based navigation device comprising a novel Lomographic neural network model section for generating navigation information.
The method according to claim 1,
Wherein the neuronal cell model is at least two of an interfacial cell model, a head direction cell model, a persistent speech cell model, a lattice cell model, a place cell model, and a motion cell model.
3. The method of claim 2,
The interfacial cell model, the head-direction cell model, and the place cell model are each connected through the synapse model with the exercise cell model,
Wherein the persistent speech cell model and the lattice cell model are connected via the synapse model,
Wherein the lattice cell model and the kinetic cell model are connected through the synapse model.
3. The method of claim 2,
Wherein the neuromodern neural network model unit comprises:
Determining whether or not the head-direction cell model is ignited according to the input current according to the sensing information of the angle sensor,
Determines whether or not the interface cell model is ignited according to the input current according to the sensing information of the distance sensor,
And determines whether or not the persistent ignition cell model is ignited according to an input current according to sensing information of the angle sensor and the velocity sensor.
3. The method of claim 2,
Wherein the neuromodern neural network model unit comprises:
Determining the ignition of some of the lattice cells of the lattice cell model according to the utterance of some persistent pyrogenous cells among the plurality of persistent pyrogenic cells included in the persistent pyroxus cell model,
Wherein the activity control information of the place cell model is generated according to the ignition of some lattice cells of the lattice cell model.
3. The method of claim 2,
Wherein the neuromodern neural network model unit comprises:
Determining whether or not the motor cell model is ignited according to synaptic stimulation by any one of the interfacial cell model, the head-direction cell model, and the place cell model,
Wherein the rotating direction of the moving object is determined according to the ignition of the moving cell model.
Receiving at least one sensing information and generating an input current;
Determining whether or not to ignite at least one neural cell model among a plurality of neural cell models connected through a synapse model based on the input current; And
And generating navigation information by adjusting the synaptic connection strength with other neuron models connected through the synapse model through the synaptic model and the excited neuron model when one of the neuron models is ignited A neuromotor neural network model based navigation method.

Images