On-device ai processing distributed processing apparatus and method

Abstract

To provide an on-device AI processing distributed processing apparatus and method capable of distributing an AI processing load of a terminal through other internal or external apparatuses.SOLUTION: An on-device AI processing distributed processing apparatus 500 is configured to: measure, when a command is input by a user 700, an AI processing amount required to execute the user command; select, when the measured AI processing amount exceeds a self-processing capacity, an AI processing apparatus 600 as an AI processing distributed processing target; request the selected AI processing distributed processing target to perform the AI processing distributed processing; and provide, when a first AI processing result value is received from the AI processing distributed processing target, a final result value to the user based on the first AI processing result value and a self-processed second AI processing result value.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present disclosure relates to an on-device AI processing distributed processing device and method that can distribute the AI processing load of a terminal via other internal or external devices. [Background technology]

[0002] Generally, artificial intelligence is a field of computer engineering and information technology that studies how to enable computers to think, learn, and develop themselves, which are functions of human intelligence, thereby enabling computers to imitate human intelligent behavior.

[0003] Furthermore, AI does not exist on its own, but is directly and indirectly connected to many other fields of computer science. In particular, in modern times, there are active attempts to incorporate AI elements into many fields of information technology and use them to solve problems in those fields.

[0004] Meanwhile, there has been active research into technologies that use artificial intelligence to recognize and learn about the surrounding environment, provide users with desired information in a desired format, and perform operations and functions that users desire.

[0005] An electronic device that provides such various operations and functions can be called an artificial intelligence device.

[0006] Recently, on-device AI, which can process information independently on terminal devices without the need to connect to a server or cloud, has been attracting attention.

[0007] On-device AI processes AI calculations independently within the user's own device without transmitting data to a server or cloud, which has the advantages of being fast, protecting personal information, and being cost-effective.

[0008] However, results processed by on-device AI have limitations in terms of completeness compared to results processed by AI processing on a server or cloud.

[0009] For example, when performing real-time translation in a noisy environment or when performing real-time translation of long sentences, on-device AI has had problems such as inaccurate processing results or not being able to provide processing results at all.

[0010] Therefore, it is necessary to develop on-device AI processing technology that can improve the speed of AI processing on terminals and the accuracy of resulting values. Summary of the Invention [Problem to be solved by the invention]

[0011] The present disclosure is directed to solving the above-mentioned problems and other problems.

[0012] The present disclosure aims to provide an on-device AI processing distributed processing device and method that can improve the AI processing speed and accuracy of result values of a device by selecting an external or internal AI processing distributed processing target and requesting distributed processing of AI processing when the AI processing volume for executing user commands exceeds the self-processing capacity. [Means for solving the problem]

[0013] An on-device AI processing distributed processing device according to one embodiment of the present disclosure includes an input unit to which a user command is input, and a processor that performs AI processing corresponding to the user command. When the user command is input, the processor measures the AI processing amount for executing the user command. If the measured AI processing amount exceeds the amount that the processor can process itself, the processor selects an AI processing distributed processing target and requests the selected AI processing distributed processing target to process the AI processing in a distributed manner. When a first AI processing result value is received from the AI processing distributed processing target, the processor can provide a final result value based on the first AI processing result value and a second AI processing result value that the processor processed itself.

[0014] An AI processing device according to an embodiment of the present disclosure is an AI processing device that is detachably attached to an on-device AI processing distributed processing device, and includes: a communication unit communicatively connected to the on-device AI processing distributed processing device; a memory that stores an AI model for AI processing; and a processor that performs AI processing in response to a distributed processing request from the on-device AI processing distributed processing device. When attached to the on-device AI processing distributed processing device, the processor communicatively connects with the on-device AI processing distributed processing device; when communicatively connected with the on-device AI processing distributed processing device, the processor checks whether a distributed processing request has been input from the on-device AI processing distributed processing device; and, when a distributed processing request has been input from the on-device AI processing distributed processing device, extracts distributed processing information from the distributed processing request, performs AI processing based on the distributed processing information to generate an AI processing result value, and provides the generated AI processing result value to the on-device AI processing distributed processing device.

[0015] An AI processing distributed processing system according to one embodiment of the present disclosure is an AI processing distributed processing system including an AI processing device that is detachably attached to an on-device AI processing distributed processing device, and includes an on-device AI processing distributed processing device that performs AI processing corresponding to a user command, and an AI processing device that performs AI processing in response to a distributed processing request from the on-device AI processing distributed processing device. When a user command is input, the on-device AI processing distributed processing device measures the AI processing amount for executing the user command, and if the measured AI processing amount exceeds the amount that can be processed by itself, selects an AI processing distributed processing target and requests the selected AI processing distributed processing target to perform distributed AI processing. When a first AI processing result value is received from the AI processing distributed processing target, the on-device AI processing distributed processing device can provide a final result value based on the first AI processing result value and a second AI processing result value that it has processed by itself.

[0016] An AI processing distributed processing method for an on-device AI processing distributed processing device according to one embodiment of the present disclosure may include the steps of: determining whether a user command is input; measuring an AI processing amount for executing the user command when the user command is input; selecting an AI processing distributed processing target when the measured AI processing amount exceeds a self-processable amount; requesting the selected AI processing distributed processing target to perform distributed processing of AI processing; self-processing AI processing for a self-processable amount of the AI processing amount; receiving a first AI processing result value from the AI processing distributed processing target; and providing a final result value based on the received first AI processing result value and the self-processed second AI processing result value.

[0017] A method for distributed AI processing of an AI processing device detachably attached to an on-device AI processing distributed processing device according to an embodiment of the present disclosure may include the steps of: determining whether the AI processing device is attached to the on-device AI processing distributed processing device; establishing a communication connection with the on-device AI processing distributed processing device when attached to the on-device AI processing distributed processing device; determining whether a distributed processing request is input from the on-device AI processing distributed processing device when the communication connection is established with the on-device AI processing distributed processing device; extracting distributed processing information from the distributed processing request when the distributed processing request is input from the on-device AI processing distributed processing device; performing AI processing based on the distributed processing information to generate an AI processing result value; and providing the generated AI processing result value to the on-device AI processing distributed processing device. [Effects of the Invention]

[0018] According to one embodiment of the present disclosure, when the AI processing volume for executing user commands exceeds the volume that the on-device AI processing distributed processing device can handle, the on-device AI processing distributed processing device selects an external or internally located AI processing distributed processing target and requests distributed processing of the AI processing, thereby improving the AI processing speed of the device, the accuracy of the result value, and the quality of service.

[0019] In addition, the present disclosure can minimize power consumption of a device by distributing AI processing together with externally or internally located distributed processing targets, thereby reducing heat generation and improving performance and lifespan. [Brief explanation of the drawings]

[0020] [Figure 1] FIG. 1 illustrates an artificial intelligence device according to one embodiment of the present disclosure. [Figure 2] FIG. 2 illustrates an artificial intelligence server according to one embodiment of the present disclosure. [Figure 3]FIG. 3 illustrates an artificial intelligence system according to one embodiment of the present disclosure. [Figure 4] FIG. 4 is a diagram illustrating an on-device AI processing distributed processing system according to an embodiment of the present disclosure. [Figure 5] FIG. 5 is a diagram illustrating an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 6] FIG. 6 is a diagram illustrating a process of measuring the AI processing amount of an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 7] FIG. 7 is a diagram illustrating a process of measuring the AI processing amount of an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 8] FIG. 8 is a diagram illustrating a communication connection process between an on-device AI processing distributed processing device and an external device according to an embodiment of the present disclosure. [Figure 9] FIG. 9 is a diagram illustrating a communication connection process between an on-device AI processing distributed processing device and an external device according to an embodiment of the present disclosure. [Figure 10] FIG. 10 is a diagram illustrating a process of selecting an AI processing distributed processing target in an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 11] FIG. 11 is a diagram illustrating a process of selecting an AI processing distributed processing target in an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 12] FIG. 12 is a diagram illustrating a notification process of an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 13] FIG. 13 is a diagram illustrating a notification process of an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 14] FIG. 14 is a diagram illustrating an AI processing distributed processing method of an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 15] FIG. 15 is a diagram illustrating an AI processing distributed processing method of an AI processing device that is detachably attached to an on-device AI processing distributed processing device according to an embodiment of the present disclosure. [Figure 16] FIG. 16 is a diagram illustrating the overall operation flow of an on-device AI processing distributed processing system according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0021] Hereinafter, the embodiments disclosed herein will be described in detail with reference to the accompanying drawings. Regardless of the reference numerals, identical or similar components will be designated by the same reference numerals, and redundant descriptions thereof will be omitted. The suffixes "module" and "unit" used in the following description are used to facilitate the preparation of the present specification and do not have any distinguishing meanings or functions. Furthermore, when describing the embodiments disclosed herein, if it is determined that a detailed description of known technology may obscure the gist of the embodiments disclosed herein, such a detailed description will be omitted. Furthermore, the accompanying drawings are provided to facilitate understanding of the embodiments disclosed herein, and the technical concepts disclosed herein should not be limited by the accompanying drawings. The accompanying drawings should be understood to include all modifications, equivalents, and alternatives within the concept and technical scope of the present disclosure.

[0022] Terms including ordinal numbers such as "first," "second," etc. may be used to describe various components, but the components are not limited by the terms. The terms are used only to distinguish one component from another.

[0023] When a component is said to be "coupled" or "connected" to another component, it should be understood that it may be directly coupled or connected to the other component, but that there may be other components in between. Conversely, when a component is said to be "directly coupled" or "directly connected" to another component, it should be understood that there are no other components in between.

[0024] Furthermore, throughout this specification, the terms neural network, neural network network, and network function may be used interchangeably. A neural network is composed of a collection of interconnected computational units generally called "nodes." Such "nodes" may also be called "neurons." A neural network is composed of at least two or more nodes. The nodes (or neurons) that make up a neural network are interconnected by one or more "links."

[0025] FIG. 1 shows an AI device 100 according to one embodiment of the present disclosure.

[0026] The AI device 100 can be embodied as a fixed or mobile device such as a TV, projector, mobile phone, smartphone, desktop computer, laptop, digital broadcasting terminal, PDA (personal digital assistant), PMP (portable multimedia player), navigation, tablet PC, wearable device, set-top box (STB), DMB receiver, radio, washing machine, refrigerator, digital signage, robot, vehicle, etc.

[0027] Referring to FIG. 1, the AI device 100 may include a communication unit 110, an input unit 120, a learning processor 130, a sensing unit 140, an output unit 150, a memory 170, and a processor 180, etc.

[0028] The communication unit 110 can use wired or wireless communication technology to transmit and receive data to and from external devices such as other AI devices 100a to 100e and the AI server 200. For example, the communication unit 110 can transmit and receive sensor information, user input, learning models, control signals, etc. to and from external devices.

[0029] At this time, the communication technologies used by the communication unit 110 include GSM (Global System for Mobile communication), CDMA (Code Division Multi Access), LTE (Long Term Evolution), 5G, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Bluetooth, RFID (Radio Frequency Identification), Infrared Data Association (IrDA), ZigBee, and NFC (Near Field Communication).

[0030] The input unit 120 can acquire various types of data.

[0031] In this case, the input unit 120 may include a camera for inputting a video signal, a microphone for receiving an audio signal, a user input unit for receiving information input from a user, etc. Here, the camera or microphone may be treated as a sensor, and the signal obtained from the camera or microphone may be referred to as sensing data or sensor information.

[0032] The input unit 120 may acquire learning data for model learning, input data used when acquiring an output using a learning model, etc. The input unit 120 may also acquire raw input data, in which case the processor 180 or the learning processor 130 may extract input features from the input data as preprocessing.

[0033] The learning processor 130 can use the training data to train a model composed of an artificial neural network. Here, the trained artificial neural network can be referred to as a training model. The training model can be used to infer a result value for new input data that is not the training data, and the inferred value can be used as a basis for making decisions to perform certain actions.

[0034] At this time, the learning processor 130 can perform AI processing together with the learning processor 240 of the AI server 200 of FIG.

[0035] In this case, the learning processor 130 may include a memory integrated into or implemented in the AI device 100. Alternatively, the learning processor 130 may be implemented using the memory 170, an external memory directly coupled to the AI device 100, or a memory maintained in an external device.

[0036] The sensing unit 140 may acquire at least one of internal information of the AI device 100, information about the surrounding environment of the AI device 100, and user information using various sensors.

[0037] In this case, the sensors included in the sensing unit 140 include a proximity sensor, an illuminance sensor, an acceleration sensor, a magnetic sensor, a gyro sensor, an inertial sensor, an RGB sensor, an IR sensor, a fingerprint recognition sensor, an ultrasonic sensor, a light sensor, a microphone, a lidar, a radar, and the like.

[0038] The output unit 150 can generate an output related to the visual, auditory, or tactile senses, for example.

[0039] In this case, the output unit 150 includes a display unit for outputting visual information, a speaker for outputting auditory information, a haptic module for outputting tactile information, and the like.

[0040] The memory 170 may store data supporting various functions of the AI device 100. For example, the memory 170 may store input data acquired from the input unit 120, learning data, a learning model, a learning history, and the like.

[0041] Processor 180 can determine at least one executable action of AI device 100 based on information determined or generated using data analysis algorithms or machine learning algorithms, and processor 180 can control components of AI device 100 to perform the determined action.

[0042] To this end, processor 180 may request, retrieve, receive, or utilize data from learning processor 130 or memory 170 and control components of AI device 100 to perform a predicted or preferred action from among the at least one executable action.

[0043] At this time, if cooperation with an external device is required to perform the determined operation, the processor 180 can generate a control signal for controlling the external device and transmit the generated control signal to the external device.

[0044] The processor 180 can acquire intent information for the user input and determine the user's requirements based on the acquired intent information.

[0045] At this time, the processor 180 can acquire intention information corresponding to the user input by using at least one of an STT (Speech To Text) engine for converting voice input into a string of characters or an NLP (Natural Language Processing) engine for acquiring intention information of natural language.

[0046] In this case, at least one of the STT engines or the NLP engines may be configured with an artificial neural network trained at least in part by a machine learning algorithm, and at least one of the STT engines or the NLP engines may be trained by the learning processor 130, the learning processor 240 of the AI server 200, or a distributed processing thereof.

[0047] The processor 180 can collect history information including the operation details of the AI device 100 and user feedback on the operation, and store it in the memory 170 or the learning processor 130, or transmit it to an external device such as the AI server 200. The collected history information can be used to update the learning model.

[0048] The processor 180 can control at least some of the components of the AI device 100 to run the application stored in the memory 170. The processor 180 can also operate two or more components included in the AI device 100 in combination with each other to run the application.

[0049] FIG. 2 illustrates an AI server 200 according to one embodiment of the present disclosure.

[0050] Referring to Figure 2, the AI server 200 may refer to a device that trains an artificial neural network using a machine learning algorithm or uses a trained artificial neural network. Here, the AI server 200 may be configured with multiple servers to perform distributed processing and may be defined as a 5G network. In this case, the AI server 200 may be included as part of the AI device 100 and may jointly perform at least a portion of the AI processing.

[0051] The AI server 200 may include a communication unit 210, a memory 230, a learning processor 240, and a processor 260.

[0052] The communication unit 210 can send and receive data to and from external devices such as the AI device 100.

[0053] The memory 230 may include a model storage unit 231. The model storage unit 231 may store a model (or artificial neural network 231a) that is being trained or has been trained by the learning processor 240.

[0054] The learning processor 240 can use the training data to train the artificial neural network 231a. The training model may be installed in the AI server 200 of the artificial neural network, or may be installed in an external device such as the AI device 100.

[0055] The learning model may be implemented in hardware, software, or a combination of hardware and software. When a part or all of the learning model is implemented in software, one or more instructions constituting the learning model are stored in memory 230.

[0056] The processor 260 can utilize the learning model to infer outcome values for new input data and generate responses or control instructions based on the inferred outcome values.

[0057] FIG. 3 shows an AI system 1 according to one embodiment of the present invention.

[0058] 3, in the AI system 1, at least one of an AI server 200, a robot 100a, an autonomous vehicle 100b, an XR device 100c, a smartphone 100d, and a home appliance 100e is connected to a cloud network 10. Here, the robot 100a, the autonomous vehicle 100b, the XR device 100c, the smartphone 100d, and the home appliance 100e to which AI technology is applied may be referred to as the AI devices 100a to 100e.

[0059] The cloud network 10 may refer to a network that constitutes a part of a cloud computing infrastructure or exists within a cloud computing infrastructure. Here, the cloud network 10 may be configured using a 3G network, a 4G or LTE (Long Term Evolution) network, a 5G network, or the like.

[0060] That is, the devices 100a to 100e, 200 that make up the AI system 1 are connected to each other via the cloud network 10. In particular, the devices 100a to 100e, 200 can communicate with each other via a base station, but can also communicate with each other directly without going through a base station.

[0061] The AI server 200 may include a server that performs AI processing and a server that performs calculations on big data.

[0062] The AI server 200 is connected to at least one of the AI devices constituting the AI system 1, namely, a robot 100a, an autonomous vehicle 100b, an XR device 100c, a smartphone 100d, or a home appliance 100e, via a cloud network 10, and can support at least a portion of the AI processing of the connected AI devices 100a to 100e.

[0063] At this time, the AI server 200 can train the artificial neural network using a machine learning algorithm instead of the AI devices 100a to 100e, and can directly store or transmit the learning model to the AI devices 100a to 100e.

[0064] At this time, the AI server 200 receives input data from the AI devices 100a to 100e, infers a result value for the received input data using a learning model, and generates a response or control command based on the inferred result value and transmits it to the AI devices 100a to 100e.

[0065] Alternatively, the AI devices 100a to 100e can infer a result value for input data using a direct learning model, and generate a response or control command based on the inferred result value.

[0066] FIG. 4 is a diagram illustrating an on-device AI processing distributed processing system according to an embodiment of the present disclosure.

[0067] As shown in FIG. 4, the on-device AI processing distributed processing system of the present disclosure includes an on-device AI processing distributed processing device 500 that performs AI processing corresponding to a user command, and an AI processing device 600 that performs AI processing corresponding to a distributed processing request of the on-device AI processing distributed processing device 500.

[0068] Here, the on-device AI processing distributed processing device 500 is an artificial intelligence device capable of performing on-device AI processing, and includes all of the following: stationary devices such as personal computers (PCs), network TVs, hybrid broadcast broadband TVs (HBTVs), smart TVs, and internet protocol TVs (IPTVs), as well as mobile or handheld devices such as smartphones, tablet PCs, notebooks, personal digital assistants (PDAs), smart watches, and smart glasses.

[0069] The AI processing device 600 may be an internal or external device of the on-device AI processing distributed processing device 500, and may be used to distribute a portion of the excess AI processing capacity of the on-device AI processing distributed processing device 500.

[0070] Here, the AI processing device 600, which is an internal device of the on-device AI processing distributed processing device 500, may be configured with one or more cores and may include processors for data analysis and deep learning, such as a neural network processing unit (NPU), a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), and a tensor processing unit (TPU).

[0071] The processor of the AI processing device 600 can read computer programs stored in memory and process data for machine learning, and can perform calculations for neural network learning, such as processing input data for learning in deep learning (DL), extracting features from input data, calculating errors, and updating neural network weights using backpropagation.

[0072] In addition, the AI processing device 600, which is an external device of the on-device AI processing distributed processing device 500, may include a battery pack 610 storing at least one AI model, earphones 620, smart glasses 630, a smart watch 640, etc., and may also include various peripheral devices such as external memory, a smart health band, an adapter, a Global Positioning System (GPS), a Personal Data Assistance (PDA), a barcode reader, a character recognition device, a voice recognition device, etc.

[0073] Here, the external device, AI processing device 600, may be a peripheral device of the on-device AI processing distributed processing device 500 and may be coupled to and communicatively connected to the on-device AI processing distributed processing device 500, or may be communicatively connected via wired and wireless means.

[0074] Meanwhile, when a command is input from a user 700, the on-device AI processing distributed processing device 500 measures the AI processing amount required to execute the user command, checks whether the measured AI processing amount exceeds the amount that can be processed by itself, and if the measured AI processing amount exceeds the amount that can be processed by itself, selects a target for distributed AI processing.

[0075] Here, the target of the AI processing distributed processing may be an AI processing device 600, which is an internal device or an external device.

[0076] Next, when at least one AI processing distributed processing target is selected from among the AI processing distributed processing targets connected to the on-device AI processing distributed processing device 500, the on-device AI processing distributed processing device 500 can request the selected AI processing distributed processing target to perform distributed processing of the AI processing.

[0077] Then, when the on-device AI processing distributed processing device 500 receives the first AI processing result value from the AI processing device 600 that is the target of the AI processing distributed processing, it can generate a final result value based on the first AI processing result value and the second AI processing result value that it processed itself and provide it to the user 700.

[0078] The on-device AI processing distributed processing device 500 stores at least one AI model and can provide various service results through the pre-trained AI model in response to a user's 700 command.

[0079] Here, the AI model may be a deep neural network (DNN), which is a neural network including multiple hidden layers in addition to an input layer and an output layer.

[0080] AI models can understand the latent structures of data such as photos, text, video, audio, music, etc.

[0081] The AI processing unit 600 may be installed inside or outside the device AI processing distributed processing device 500 as an internal or external unit of the AI processing distributed processing target 500 .

[0082] When the AI processing device 600 is installed in the on-device AI processing distributed processing device 500, it establishes a communication connection with the on-device AI processing distributed processing device 500. When the AI processing device 600 establishes a communication connection with the on-device AI processing distributed processing device 500, it checks whether a distributed processing request has been input from the on-device AI processing distributed processing device 500. When a distributed processing request has been input from the on-device AI processing distributed processing device 500, it extracts distributed processing information from the distributed processing request, performs AI processing based on the distributed processing information to generate an AI processing result value, and provides the generated AI processing result value to the on-device AI processing distributed processing device 500.

[0083] Here, the AI processing device 600 stores at least one AI model among the AI models pre-stored in the on-device AI processing distributed processing device 500.

[0084] FIG. 5 is a diagram illustrating an on-device AI processing distributed processing device according to an embodiment of the present disclosure.

[0085] As shown in FIG. 5, the on-device AI processing distributed processing device 500 of the present disclosure may include an input unit 510 to which a user command is input, and a processor 520 to perform AI processing corresponding to the user command.

[0086] Here, when a user command is input, the processor 520 measures the AI processing amount required to execute the user command, and if the measured AI processing amount exceeds the amount that the processor can process itself, it selects an AI processing device 530, 600 that is the target for distributed AI processing, requests the selected target for distributed AI processing to process the AI processing in a distributed manner, and when it receives a first AI processing result value from the target for distributed AI processing, it can provide a final result value based on the first AI processing result value and the second AI processing result value that it processed itself.

[0087] When measuring the AI processing amount, when a user command is input, the processor 520 can select at least one AI model to execute the user command, and measure the AI processing amount for executing the user command based on the computational amount of the selected AI model.

[0088] Here, if one AI model is selected, processor 520 can measure the AI processing amount based on the amount of calculation processed by that one AI model, and if multiple AI models are selected, processor 520 can measure the AI processing amount based on the total amount of calculation processed by the multiple AI models.

[0089] As an example, the processor 520 analyzes the selected AI model to determine whether there are branch points connecting one higher-level operator to multiple lower-level operators and junction points connecting multiple higher-level operators to one lower-level operator. If there are branch points and junction points, the processor 520 determines whether there is at least one parallel processing part based on the branch points and junction points. If there is a parallel processing part, the processor 520 can measure the AI processing amount based on the first calculation amount for the parallel processing part and the second calculation amount for the remaining part other than the parallel processing part.

[0090] When selecting an AI processing distributed processing target, if the AI processing amount is measured, the processor 520 checks whether the measured AI processing amount is an amount that it can process itself, and if it determines that the measured AI processing amount is an amount that it can process itself, it immediately executes AI processing corresponding to the user command, and if it determines that the measured AI processing amount exceeds the amount that it can process itself, it can select an AI processing distributed processing target.

[0091] Here, when processor 520 checks whether the measured AI processing amount is the amount it can process itself, if the AI processing amount is measured, it checks whether there is AI processing currently in operation, and if there is AI processing currently in operation, it calculates the amount it can process itself excluding the AI processing currently in operation, and can check whether the measured AI processing amount is the amount it can process itself based on the calculated amount it can process itself.

[0092] In some cases, when processor 520 checks whether the measured AI processing amount is a self-processable amount, if the AI processing amount is measured, it can check whether a preset self-processable amount exists, and if a preset self-processable amount exists, it can check whether the measured AI processing amount is a self-processable amount based on the preset self-processable amount.

[0093] As an example, the preset self-treatment amount may be a preset default value or a user value preset by the user.

[0094] In addition, when selecting an AI processing device 530, 600 as a target for AI processing distributed processing, the processor 520 checks whether it is connected to an external device. If an external device is connected, it obtains identification information of the external device from the external device and checks whether the external device is an AI processing device 600 based on the identification information. If the external device is an AI processing device 600, it can select the external device as a target for AI processing distributed processing.

[0095] For example, the on-device AI processing distributed processing device 500 of the present disclosure may further include a connection unit that is connected to and communicatively connected with an external device.

[0096] Here, the processor 520 may maintain a communication connection with the external device when the external device is connected to the connection unit, and may terminate the communication connection with the external device when the external device is disconnected from the connection unit.

[0097] As another example, the on-device AI processing distributed processing device 500 of the present disclosure may further include a communication unit that is communicatively connected to an external device via wire or wirelessly.

[0098] Here, the processor 520 can check whether or not there is a communication connection with an external device through the communication unit.

[0099] That is, the external device may include an AI processing device 600 having at least one AI model stored therein.

[0100] As an example, the AI processing device 600 of the external device may include at least one of a battery pack storing at least one AI model, earphones, external memory, a smart watch, smart glasses, a smart health band, an adapter, a Global Positioning System (GPS), a Personal Data Assistance (PDA), a barcode reader, a character recognition device, and a voice recognition device, but this is merely an example and is not limited thereto.

[0101] In addition, when processor 520 determines whether an external device is an AI processing device 600, it determines whether an AI model identifier is present in the identification information of the external device, and if an AI model identifier is present, it recognizes the external device as AI processing device 600 and can determine whether the device is capable of AI processing in response to user commands based on the AI model identifier.

[0102] Here, if the AI model identifier does not exist in the identification information of the external device, the processor 520 may recognize the external device as a general device and generate and provide a user notification requesting replacement of the external device.

[0103] For example, the processor 520 may generate and output a user notification in at least one of a text format and a sound format.

[0104] In addition, if there are multiple external devices capable of AI processing, the processor 520 can select all of the multiple external devices as targets for AI processing distributed processing, and assign priorities to the multiple external devices based on the processing performance indexes of the multiple external devices selected as targets for AI processing distributed processing.

[0105] Here, the processor 520 may assign the highest priority to an external device having the highest processing performance index among the plurality of external devices, and assign the lowest priority to an external device having the lowest processing performance index.

[0106] When the measured AI processing amount exceeds the amount that the processor 520 can process, the processor 520 checks the amount that the external device can process for the excess amount, and if multiple external devices are needed to process the excess amount, it can request the external devices to process the excess amount according to the priority assigned to the external devices.

[0107] Here, when requesting the processing of the excess amount, the processor 520 may request the processing of different excess amounts from a plurality of external devices.

[0108] As an example, the processor 520 may request an external device with a higher priority to process a first excess amount, and may request an external device with a lower priority to process a second excess amount, which is the remainder of the total excess amount excluding the first excess amount.

[0109] In some cases, when requesting the processing of the excess amount, the processor 520 may request the same excess amount to be processed by a plurality of external devices.

[0110] As an example, the processor 520 may distribute the total excess amount equally among multiple external devices, request an external device with a higher priority to process a first excess amount, and request an external device with a lower priority to process a second excess amount that is the same as the first excess amount.

[0111] Meanwhile, when processor 520 checks whether or not it is connected to an external device, if it is not connected to an external device, it checks whether AI processing device 530, which is an internal device capable of AI processing, exists, and if AI processing device 530, which is an internal device capable of AI processing, exists, it checks whether AI processing device 530, which is an internal device capable of AI processing, is currently in an inactive state, and if AI processing device 530, which is an internal device capable of AI processing, is currently in an inactive state, it can select AI processing device 530, which is an internal device capable of AI processing, as the target for AI processing distributed processing.

[0112] Here, if there are multiple AI processing devices 530, which are internal devices capable of AI processing, the processor 520 can select all of the multiple internal devices as targets for AI processing distributed processing and assign priorities to the multiple internal devices based on the processing performance indexes of the multiple internal devices selected as targets for AI processing distributed processing.

[0113] As an example, the processor 520 may assign the highest priority to an internal device having the highest processing performance index among the plurality of internal devices, and assign the lowest priority to an internal device having the lowest processing performance index.

[0114] In addition, when the measured AI processing amount exceeds the amount that the processor 520 can process, the processor 520 checks the amount that the internal devices can process for the excess amount, and if multiple internal devices are needed for the excess amount, it can request the internal devices to process the excess amount according to the priority assigned to them.

[0115] Here, when the processor 520 requests the processing of the excess amount, it may request the processing of different excess amounts from a plurality of internal devices.

[0116] As an example, the processor 520 may request an internal device with a higher priority to process a first excess amount, and may request an internal device with a lower priority to process a second excess amount, which is the remainder of the total excess amount excluding the first excess amount.

[0117] In some cases, when processor 520 requests the processing of an excess amount, it may request a plurality of internal devices to process the same excess amount.

[0118] As an example, the processor 520 may distribute the total excess amount equally among multiple internal devices, request an internal device with a higher priority to process a first excess amount, and request an internal device with a lower priority to process a second excess amount that is the same as the first excess amount.

[0119] In addition, when selecting a target for AI processing distributed processing, processor 520 checks whether AI processing device 530, an internal device capable of AI processing, exists, and if AI processing device 530, an internal device capable of AI processing, exists, processor 520 checks whether AI processing device 530, an internal device capable of AI processing, is currently in an inactive state, and if the internal device capable of AI processing is currently in an inactive state, processor 520 can select AI processing device 530, an internal device capable of AI processing, as a target for AI processing distributed processing.

[0120] Here, if the first AI processing device 530, which is an internal device capable of AI processing, is not present, the processor 520 checks whether it is connected to an external device. If an external device is connected, it obtains identification information of the external device from the external device and checks whether the external device is an AI processing device 600 based on the identification information. If the external device is an AI processing device 600, it can select the external device as the target for AI processing distributed processing.

[0121] Next, when processor 520 requests distributed processing of AI processing, it calculates the excess amount of the measured AI processing volume other than the amount it can process itself, checks the processable amount of the selected AI processing distributed processing target, and if the processable amount of the AI processing distributed processing target is greater than the excess amount, it can request distributed processing of AI processing for the excess amount from the AI processing distributed processing target.

[0122] Here, if the processable amount of the AI processing distributed processing target is less than the excess amount, the processor 520 can select another AI processing distributed processing target and request the multiple AI processing distributed processing targets to process the excess amount in a distributed manner.

[0123] Next, when processor 520 requests distributed processing of AI processing, it calculates the excess amount of the measured AI processing amount other than the amount it can process itself, extracts the distributed processing portion of the AI processing for executing the user command that corresponds to the excess amount, and requests distributed processing of AI processing from the selected AI processing distributed processing target using the extracted distributed processing portion.

[0124] As an example, when extracting a distributed processing portion corresponding to the excess amount, the processor 520 analyzes each AI model for executing user commands to determine whether there are branch points connecting one higher-level operator to multiple lower-level operators and confluence points connecting multiple higher-level operators to one lower-level operator.If branch points and confluence points are present, the processor 520 determines whether there is at least one parallel processing portion based on the branch points and confluence points, and extracts the parallel processing portion as a distributed processing portion.

[0125] As another example, when processor 520 extracts the distributed processing portion corresponding to the excess amount, if there are multiple AI models for executing user commands, it checks whether any of the multiple AI models are capable of parallel processing with each other, and if there are AI models capable of parallel processing, it can extract the processing portion performed by the AI models capable of parallel processing as the distributed processing portion.

[0126] When the processor 520 requests distributed processing of AI processing, it can provide a distributed processing request including AI model information, location information, distributed processing amount information, and input data corresponding to the distributed processing portion as a target for distributed processing of AI processing.

[0127] Next, when providing a final result value, the processor 520 can map the first AI processing result value received from the AI processing distributed processing target and the self-processed second AI processing result value to each other to generate a final result value corresponding to the user command.

[0128] In this way, when the amount of AI processing required to execute user commands exceeds the amount that the on-device AI processing distributed processing device disclosed herein can select an external or internal AI processing distributed processing target and request distributed processing of the AI processing, thereby improving the terminal's AI processing speed, accuracy of result values, and service quality.

[0129] In addition, the present disclosure can minimize power consumption of a device by distributing AI processing together with externally or internally located distributed processing targets, thereby reducing heat generation and improving performance and lifespan.

[0130] 6 and 7 are diagrams illustrating a process of measuring the AI processing amount of an on-device AI processing distributed processing device according to an embodiment of the present disclosure.

[0131] As shown in FIGS. 6 and 7, when a user command is input, the present disclosure can select at least one AI model 800 to execute the user command, and measure the AI processing amount for executing the user command based on the computation amount of the selected AI model 800.

[0132] As shown in FIG. 6, in the present disclosure, when one AI model 800 is selected, the AI processing amount can be measured based on the amount of calculations processed by the one AI model 800.

[0133] That is, the present disclosure can analyze the selected AI model 800 to determine whether there are branch points 810 connecting one higher-level operator to multiple lower-level operators and junction points 820 connecting multiple higher-level operators to one lower-level operator.

[0134] Then, in the present disclosure, if a branch point 810 and a junction point 820 exist, it is confirmed whether at least one parallel processing portion 830 exists based on the branch point 810 and the junction point 820, and if a parallel processing portion 830 exists, the AI processing amount can be measured based on a first calculation amount for the parallel processing portion 830 and a second calculation amount for the remaining portion other than the parallel processing portion.

[0135] Next, the present disclosure can determine whether to distribute the total AI processing amount for one AI model to the parallel processing portion 830 when the AI processing amount for executing user instructions exceeds the amount that can be processed by the AI model itself.

[0136] As another example, as shown in FIG. 7, when multiple AI models 800 are selected, the present disclosure can measure the AI processing amount based on the total amount of calculations processed by the multiple AI models 800.

[0137] Here, the present disclosure analyzes an AI model group to determine whether there is a branch point 850 connecting one higher-level AI model to multiple lower-level AI models and a junction point 860 connecting multiple higher-level AI models to one lower-level AI model; if there is a branch point 850 and a junction point 860, it determines whether there is at least one parallel processing part based on the branch point 850 and the junction point 860; if there is a parallel processing part, it can measure the AI processing amount based on a first calculation amount for the parallel processing part and a second calculation amount for the remaining part other than the parallel processing part.

[0138] Next, the present disclosure can determine whether to distribute the parallel processing portion of the total AI processing amount for an AI model group including multiple AI models when the AI processing amount for executing user instructions exceeds the amount that can be processed by the group itself.

[0139] In addition, the present disclosure checks whether there is a branch point connecting one higher-level operator to multiple lower-level operators and a confluence point connecting multiple higher-level operators to one lower-level operator for each AI model in the AI model group, and if a branch point and a confluence point exist, it checks whether there is at least one parallel processing part based on the branch point and the confluence point, and determines whether to perform distributed processing on the parallel processing part.

[0140] Furthermore, when selecting a target for distributed AI processing in the future, the present disclosure can request distributed processing of the parallel processing portion between the branching point and the confluence point within each AI model as the target for distributed AI processing, and can also request distributed processing of the entire processing portion performed by the parallel model between the branching point and the confluence point within an AI model group as the target for distributed AI processing.

[0141] 8 and 9 are diagrams illustrating a communication connection process between an on-device AI processing distributed processing device and an external device according to an embodiment of the present disclosure.

[0142] As shown in Figures 8 and 9, when the present disclosure selects an external device, AI processing device 600, as the target for AI processing distributed processing, it checks whether or not there is a connection with the external device, and if there is a connection, it obtains identification information of the external device from the external device and checks whether the external device is an AI processing device based on the identification information. If the external device is AI processing device 600, it can select the external device as the target for AI processing distributed processing.

[0143] As shown in FIG. 8, the on-device AI processing distributed processing device 500 of the present disclosure is coupled to and communicatively connected with an AI processing device 600, which is an external device.

[0144] The on-device AI processing distributed processing device 500 of the present disclosure may be connected to the AI processing device 600 via a wired connection 502, or may be configured to be connected to or separated from the AI processing device 600 via a connection on the body itself.

[0145] Here, the on-device AI processing distributed processing device 500 of the present disclosure can maintain a communication connection with the external device when the external device is connected to the connection unit 502, and can disconnect the communication connection with the external device when the external device is disconnected from the connection unit 502.

[0146] Also, as shown in FIG. 9, the on-device AI processing distributed processing device 500 of the present disclosure may be wirelessly connected to an AI processing device 600, which is an external device.

[0147] The on-device AI processing distributed processing device 500 of the present disclosure may include a communication unit that is communicatively connected to the AI processing device 600, which is an external device, via wired or wireless communication.

[0148] Here, the on-device AI processing distributed processing device 500 of the present disclosure can check whether or not there is a communication connection with the AI processing device 600, which is an external device, via the communication unit.

[0149] In this way, when the present disclosure is connected to an external device, that is, an AI processing device 600, it receives identification information from the external device and checks whether an AI model identifier is present in the received identification information of the external device. If an AI model identifier is present, it recognizes the external device as an AI processing device 600 and can check whether the external device is capable of AI processing in response to a user command based on the AI model identifier.

[0150] Thus, the external device of the present disclosure may include an AI processing device 600 having at least one AI model stored therein.

[0151] As an example, the AI processing device 600 of the external device may include at least one of a battery pack storing at least one AI model, earphones, external memory, a smart watch, smart glasses, a smart health band, an adapter, a Global Positioning System (GPS), a Personal Data Assistance (PDA), a barcode reader, a character recognition device, and a voice recognition device, but this is merely an example and is not limited thereto.

[0152] 10 and 11 are diagrams illustrating a process of selecting an AI processing distributed processing target in an on-device AI processing distributed processing device according to an embodiment of the present disclosure.

[0153] As shown in FIG. 10, the on-device AI processing distributed processing device 500 of the present disclosure checks whether or not it is connected to an external device. If an external device is connected, it acquires identification information of the external device from the external device, checks whether the external device is an AI processing device 600 based on the identification information, and if the external device is an AI processing device 600, it can select the external device as a target for AI processing distributed processing.

[0154] Here, when there are multiple external devices capable of AI processing, the on-device AI processing distributed processing device 500 of the present disclosure can select all of the multiple external devices as targets for AI processing distributed processing, and assign priorities to the multiple external devices based on the processing performance indexes of the multiple external devices selected as targets for AI processing distributed processing.

[0155] The present disclosure can assign the highest priority to an external device with the highest processing performance index among a plurality of external devices, and can assign the lowest priority to an external device with the lowest processing performance index.

[0156] Here, the processing performance index of the external device may be set based on the processing speed, processing time, and processing amount for executing AI processing.

[0157] For example, when the on-device AI processing distributed processing device 500 of the present disclosure measures the amount of AI processing that it can process, it checks the amount of processing that an external device can perform for the excess amount, and if multiple external devices are required to process the excess amount, it can request the external devices to process the excess amount according to the priority assigned to the external devices.

[0158] Here, in the present disclosure, when the excess amount of AI processing capacity that must be distributed is 100, the battery pack 610 with the highest processing performance index is selected from among the AI processing devices 600, and the amount of capacity that the battery pack 610 can process itself is checked. If the amount of capacity that the battery pack 610 can process itself is 60, the smartwatch 640 with the second highest processing performance index is additionally selected, and the amount of capacity that the smartwatch 640 can process itself is checked. If the amount of capacity that the smartwatch 640 can process itself is 40 or more, the excess amount can be requested to be processed by multiple AI processing devices 600 including the battery pack 610 and the smartwatch 640.

[0159] In this case, the present disclosure may additionally select the smart glasses 630 with the third highest processing performance index or the earphones 620 with the fourth highest processing performance index if the smart watch 640's self-processing capacity is less than 40.

[0160] In the present disclosure, when requesting processing of an excess amount, different excess amounts can be requested to be processed by a plurality of external devices.

[0161] As an example, the present disclosure may request an external device with a higher priority to process a first excess amount, and may request an external device with a lower priority to process a second excess amount, which is the remainder of the total excess amount excluding the first excess amount.

[0162] In other words, when the excess amount of AI processing capacity that must be distributed is 100, the present disclosure can request the battery pack 610, an external device with a higher priority, to process the first excess amount, 60, and can request the smart watch 640, an external device with a lower priority, to process the second excess amount, 40, which is the remainder of the total excess amount excluding the first excess amount, 60.

[0163] In some cases, when requesting the processing of an excess amount, the present disclosure may request a plurality of external devices to process the same excess amount.

[0164] As an example, the present disclosure can distribute the total excess amount equally among multiple external devices, request an external device with a higher priority to process a first excess amount, and request an external device with a lower priority to process a second excess amount that is the same as the first excess amount.

[0165] In other words, when the excess amount of AI processing that must be distributed is 100, the present disclosure can request the battery pack 610, an external device with a higher priority, to process the first excess amount, 50, and can request the smart watch 640, an external device with a lower priority, to process the second excess amount, 50, which is the same as the first excess amount.

[0166] As shown in FIG. 11, when an external device is not connected, the on-device AI processing distributed processing device 500 of the present disclosure checks whether an AI processing device 530 that is an internal device capable of AI processing exists. If an AI processing device 530 that is an internal device capable of AI processing exists, the on-device AI processing distributed processing device 500 checks whether the AI processing device 530 that is an internal device capable of AI processing is currently in an inactive state. If the AI processing device 530 that is an internal device capable of AI processing is currently in an inactive state, the on-device AI processing distributed processing device 500 can select the AI processing device 530 that is an internal device capable of AI processing as a target for distributed AI processing.

[0167] Here, if there are multiple AI processing devices 530, which are internal devices capable of AI processing, the processor 520 can select all of the multiple internal devices as targets for AI processing distributed processing, and assign priorities to the multiple internal devices based on the processing performance indexes of the AI processing devices 530 that are targets for AI processing distributed processing and the selected multiple internal devices.

[0168] As an example, the processor 520 may assign the highest priority to an internal device having the highest processing performance index among the plurality of internal devices, and assign the lowest priority to an internal device having the lowest processing performance index.

[0169] In addition, when the measured AI processing amount exceeds the amount that the processor 520 can process, the processor 520 checks the amount that the internal devices can process for the excess amount, and if multiple internal devices are needed for the excess amount, it can request the internal devices to process the excess amount according to the priority assigned to them.

[0170] Here, when the processor 520 requests the processing of the excess amount, it may request the processing of different excess amounts from a plurality of internal devices.

[0171] As an example, the processor 520 may request an internal device with a higher priority to process a first excess amount, and may request an internal device with a lower priority to process a second excess amount, which is the remainder of the total excess amount excluding the first excess amount.

[0172] In some cases, when processor 520 requests the processing of an excess amount, it may request a plurality of internal devices to process the same excess amount.

[0173] As an example, the processor 520 may distribute the total excess amount equally among multiple internal devices, request an internal device with a higher priority to process a first excess amount, and request an internal device with a lower priority to process a second excess amount that is the same as the first excess amount.

[0174] 12 and 13 are diagrams illustrating a notification process of an on-device AI processing distributed processing device according to an embodiment of the present disclosure.

[0175] As shown in FIG. 12, when an external device is communicatively connected, the on-device AI processing distributed processing device 500 of the present disclosure receives identification information from the external device and checks whether an AI model identifier is present in the received identification information of the external device. If an AI model identifier is present, the on-device AI processing distributed processing device 500 recognizes the external device as an AI processing device 600 and checks whether the external device is capable of AI processing in response to a user command based on the AI model identifier.

[0176] Here, if the on-device AI processing distributed processing device 500 of the present disclosure does not have an AI model identifier in the identification information of the external device, it recognizes the external device as a general device and can generate and provide a user notification 580 requesting a replacement of the external device.

[0177] That is, the present disclosure determines whether the external device is an AI processing device 600 that has at least one AI model stored therein and can process AI processing. If the currently connected external device is a general device that cannot process AI processing, it can generate a message stating "Please change the external device for distributed processing" and display it on the display screen as a user notification 580.

[0178] For example, the user notification 580 may be generated and output in at least one of a text format and a sound format.

[0179] As shown in FIG. 13, when the AI processing device 600 is attached to the on-device AI processing distributed processing device 500, it establishes a communication connection with the on-device AI processing distributed processing device 500. When the AI processing device 600 is connected to the on-device AI processing distributed processing device 500, it checks whether a distributed processing request has been input from the on-device AI processing distributed processing device 500. When a distributed processing request has been input from the on-device AI processing distributed processing device 500, it extracts distributed processing information from the distributed processing request and determines whether to perform AI processing based on the distributed processing information.

[0180] Here, if it is determined that the AI processing cannot be performed, the AI processing device 600 can generate and provide a notification to the on-device AI processing distributed processing device 500 informing it that the distributed processing is not possible.

[0181] The AI processing device 600 checks whether an AI model corresponding to the distributed processing request is pre-stored in memory, and if the AI model corresponding to the distributed processing request is not stored in memory, it can generate and provide a notification to the on-device AI processing distributed processing device 500 informing it that distributed processing is not possible.

[0182] For example, if the AI model corresponding to the distributed processing request is a real-time translation-related AI model, and the real-time translation-related AI model corresponding to the distributed processing request is not stored in memory, the AI processing device 600 may generate and provide a notification to the on-device AI processing distributed processing device 500 informing it that distributed processing is not possible.

[0183] Therefore, the on-device AI processing distributed processing device 500 can display on the display screen a user notification 590 received from the AI processing device 600 informing the user that distributed processing is not possible.

[0184] In some cases, when the on-device AI processing distributed processing device 500 receives a distributed processing impossible signal from the AI processing device 600, it can generate a message stating "Distributed processing of external device is impossible" and display it on the display screen as a user notification 590.

[0185] For example, the user notification 590 may be generated and output in at least one of a text format and a sound format.

[0186] Meanwhile, the AI processing device 600 of the present disclosure checks whether an AI model corresponding to the distributed processing request is pre-stored in memory, and if the AI model corresponding to the distributed processing request is not stored in memory, it can obtain the required AI model from an external server via a communication unit and store it in memory.

[0187] FIG. 14 is a diagram illustrating an AI processing distributed processing method of an on-device AI processing distributed processing device according to an embodiment of the present disclosure.

[0188] As shown in FIG. 14, the on-device AI processing distributed processing device of the present disclosure may check whether a user command is input (S110).

[0189] Then, when a user command is input, the present disclosure can measure the amount of AI processing required to execute the user command (S120).

[0190] Here, the present disclosure is capable of selecting at least one AI model for executing a user command when the user command is input, and measuring the AI processing amount for executing the user command based on the computation amount of the selected AI model.

[0191] Next, the present disclosure can check whether the measured AI processing amount exceeds the amount that can be processed by itself (S130).

[0192] Here, in the present disclosure, if the AI processing amount is less than the self-processable amount, the AI processing amount corresponding to the user command is self-processed (S190) and an AI processing result value can be generated and provided (S200).

[0193] Next, in the present disclosure, if the measured AI processing amount exceeds the self-processing capacity, a target for AI processing distribution can be selected (S140).

[0194] Here, the present disclosure checks whether or not there is a connection with an external device, and if the external device is connected, obtains identification information of the external device from the external device, checks whether the external device is an AI processing device based on the identification information, and if the external device is an AI processing device, selects the external device as a target for AI processing distributed processing.

[0195] Furthermore, the present disclosure can check whether an internal device capable of AI processing exists when an external device is not connected, and if an internal device capable of AI processing exists, check whether the internal device capable of AI processing is currently inactive, and if the internal device capable of AI processing is currently inactive, select the internal device capable of AI processing as the target for distributed AI processing.

[0196] Then, the present disclosure can request distributed processing of AI processing from the selected target for distributed processing of AI processing (S150).

[0197] Here, the present disclosure calculates the excess amount of the measured AI processing volume other than the self-processing capacity, checks the processable capacity of the selected AI processing distributed processing target, and if the processable capacity of the AI processing distributed processing target is greater than the excess amount, requests the AI processing distributed processing target to process the excess amount in a distributed manner.

[0198] Furthermore, the present disclosure allows for the selection of additional AI processing distributed processing targets when the processable amount of an AI processing distributed processing target is less than the excess amount, and requests the multiple AI processing distributed processing targets to process the excess amount in a distributed manner.

[0199] Next, the present disclosure can perform self-processing of AI processing for the self-processable amount of the AI processing amount (S160).

[0200] Next, the present disclosure can receive a first AI processing result value from the AI processing distributed processing target (S170).

[0201] Then, the present disclosure can provide a final result value based on the received first AI processing result value and the self-processed second AI processing result value (S180).

[0202] Here, the present disclosure can map the first AI processing result value received from the AI processing distributed processing target and the self-processed second AI processing result value to each other to generate a final result value corresponding to a user command.

[0203] FIG. 15 is a diagram illustrating an AI processing distributed processing method of an AI processing device that is detachably attached to an on-device AI processing distributed processing device according to an embodiment of the present disclosure.

[0204] As shown in FIG. 15, the AI processing device of the present disclosure can check whether it is installed in an on-device AI processing distributed processing device (S210).

[0205] Then, when the present disclosure is installed in the on-device AI processing distributed processing device, it can be confirmed whether it is communicatively connected to the on-device AI processing distributed processing device (S220).

[0206] Next, when the present disclosure is connected to the on-device AI processing distributed processing device, it can check whether a distributed processing request is input from the on-device AI processing distributed processing device (S230).

[0207] Next, when a distributed processing request is input from the on-device AI processing distributed processing device, the present disclosure can extract distributed processing information from the distributed processing request (S240).

[0208] Here, the present disclosure can extract distributed processing information including AI model information, location information, distributed processing amount information, and input data corresponding to the distributed processing portion from the distributed processing request.

[0209] Then, the present disclosure can perform AI processing based on the distributed processing information to generate an AI processing result value (S250).

[0210] Here, when performing AI processing, the present disclosure checks whether an AI model corresponding to the distributed processing request is pre-stored in memory, and if the AI model corresponding to the distributed processing request is not stored in memory, the required AI model can be obtained from an external server via a communication unit and stored in memory.

[0211] In some cases, the present disclosure may check whether an AI model corresponding to a distributed processing request is pre-stored in memory when performing AI processing, and if an AI model corresponding to the distributed processing request is not stored in memory, may generate and provide a notification to the on-device AI processing distributed processing device informing it that distributed processing is not possible.

[0212] Next, the present disclosure can provide the generated AI processing result value to an on-device AI processing distributed processing device (S260).

[0213] FIG. 16 is a diagram illustrating the overall operation flow of an on-device AI processing distributed processing system according to an embodiment of the present disclosure.

[0214] As shown in FIG. 16, the on-device AI processing distributed processing system of the present disclosure includes an on-device AI processing distributed processing device 500 that performs AI processing corresponding to a user command, and an AI processing device 900 that performs AI processing corresponding to a distributed processing request of the on-device AI processing distributed processing device 500.

[0215] First, the AI processing device 900 is installed in the on-device AI processing distributed processing device 500 and connected to it (S310).

[0216] Here, the on-device AI processing distributed processing device 500 stores at least one AI model, and the AI processing device 900 stores at least one AI model among the AI models pre-stored in the on-device AI processing distributed processing device 500.

[0217] Next, the AI processing device 900 requests a communication connection from the on-device AI processing distributed processing device 500, and the on-device AI processing distributed processing device 500 approves the communication connection in response to the communication connection request from the AI processing device 900, thereby allowing the AI processing device 900 and the on-device AI processing distributed processing device 500 to communicate with each other (S320).

[0218] Next, the on-device AI processing distributed processing device 500 may receive an input of a user command requesting a specific service, such as a real-time translation service (S330).

[0219] When a user command is input, the on-device AI processing distributed processing device 500 can measure the amount of AI processing required to execute the user command (S340).

[0220] Here, when a user command is input, the on-device AI processing distributed processing device 500 can select at least one AI model to execute the user command, and measure the AI processing amount for executing the user command based on the calculation amount of the selected AI model.

[0221] For example, when a user command is a request for a real-time translation service, the on-device AI processing distributed processing device 500 can select at least one AI model that performs real-time translation-related AI processing and measure the amount of AI processing based on the calculation amount of the AI model that performs the real-time translation-related AI processing.

[0222] Next, the on-device AI processing distributed processing device 500 can determine whether the measured AI processing amount exceeds its own processing capacity (S350).

[0223] Here, when the on-device AI processing distributed processing device 500 measures the AI processing amount, it checks whether there is currently running AI processing, and if there is currently running AI processing, it calculates the self-processing capacity excluding the currently running AI processing, and can check whether the measured AI processing amount is the self-processing capacity based on the calculated self-processing capacity.

[0224] In some cases, when the on-device AI processing distributed processing device 500 measures the AI processing amount, it checks whether a preset self-processable amount exists, and if a preset self-processable amount exists, it can check whether the measured AI processing amount is a self-processable amount based on the preset self-processable amount.

[0225] As an example, the preset self-treatment amount may be a preset default value or a user value preset by the user.

[0226] Next, if the measured AI processing amount exceeds the amount that can be processed by the on-device AI processing distributed processing device 500, the on-device AI processing distributed processing device 500 can select an AI processing distributed processing target (S360).

[0227] Here, the on-device AI processing distributed processing device 500 acquires the identification information of the AI processing device 900, and can select the AI processing device 900 as a target for AI processing distributed processing based on the identification information.

[0228] Then, the on-device AI processing distributed processing device 500 can request the AI processing distributed processing target 900 to perform distributed processing of the AI processing (S370).

[0229] Next, when a distributed processing request is input from the on-device AI processing distributed processing device 500, the AI processing device 900 can extract distributed processing information from the distributed processing request (S380).

[0230] Here, the AI processing device 900 can extract distributed processing information including AI model information, location information, distributed processing amount information, and input data corresponding to the distributed processing portion from the distributed processing request.

[0231] Then, the AI processing device 900 can perform AI processing based on the distributed processing information and generate an AI processing result value (S390).

[0232] Here, when performing AI processing, the AI processing device 900 checks whether an AI model corresponding to the distributed processing request is pre-stored in memory, and if the AI model corresponding to the distributed processing request is not stored in memory, it can obtain the required AI model from an external server via a communication unit and store it in memory.

[0233] Meanwhile, the on-device AI processing distributed processing device 500 can self-process the AI processing amount that can be processed by itself (S400).

[0234] Then, the on-device AI processing distributed processing device 500 can generate a self-processed AI processing result value (S410).

[0235] Next, the on-device AI processing distributed processing device 500 can receive the AI processing result value from the AI processing device 900 (S420).

[0236] Next, the on-device AI processing distributed processing device 500 can generate a final result value based on the AI processing result value received from the AI processing device 900 and the second AI processing result value processed by itself (S430).

[0237] Here, the on-device AI processing distributed processing device 500 can map the AI processing result value received from the AI processing device 900 and the second AI processing result value processed by itself to each other to generate a final result value corresponding to the user command.

[0238] Meanwhile, as yet another embodiment, the on-device AI processing distributed processing system of the present disclosure can simultaneously request distributed processing of AI processing from internal devices and external devices.

[0239] The on-device AI processing distributed processing system disclosed herein includes an on-device AI processing distributed processing device that performs AI processing corresponding to a user command, a first AI processing device that is disposed inside the on-device AI processing distributed processing device and performs first AI processing in response to a distributed processing request of the on-device AI processing distributed processing device, and a second AI processing device that is disposed outside the on-device AI processing distributed processing device and performs second AI processing in response to a distributed processing request of the on-device AI processing distributed processing device.

[0240] Here, when a user command is input, the on-device AI processing distributed processing device measures the AI processing amount required to execute the user command, and if the measured AI processing amount exceeds the amount that it can process itself, it requests the first AI processing device and the second AI processing device to simultaneously process the AI processing in a distributed manner.When it receives a first AI processing result value from the first AI processing device and a second AI processing result value from the second AI processing device, it can provide a final result value based on the first AI processing result value, the second AI processing result value, and the third AI processing result value that it has processed itself.

[0241] The on-device AI processing distributed processing device stores at least one AI model, and the first AI processing device and the second AI processing device store at least one AI model from among the AI models pre-stored in the on-device AI processing distributed processing device.

[0242] Next, when a distributed processing request is input from the on-device AI processing distributed processing device, the first AI processing device extracts distributed processing information from the distributed processing request, performs AI processing based on the distributed processing information to generate a first AI processing result value, and provides the generated first AI processing result value to the on-device AI processing distributed processing device.

[0243] In addition, the second AI processing device communicates with the on-device AI processing distributed processing device, and when communicatively connected to the on-device AI processing distributed processing device, it checks whether a distributed processing request is input from the on-device AI processing distributed processing device.When a distributed processing request is input from the on-device AI processing distributed processing device, it extracts distributed processing information from the distributed processing request, performs AI processing based on the distributed processing information to generate a second AI processing result value, and provides the generated second AI processing result value to the on-device AI processing distributed processing device.

[0244] Next, when the on-device AI processing distributed processing device requests distributed processing of AI processing, it checks whether the first AI processing device and the second AI processing device are both AI processing devices, and if both the first AI processing device and the second AI processing device are AI processing devices, it can simultaneously request distributed processing of AI processing from the first AI processing device and the second AI processing device.

[0245] Here, when the on-device AI processing distributed processing device determines whether a device is an AI processing device, it acquires identification information from each of the first and second AI processing devices and determines whether an AI model identifier is present in the identification information. If an AI model identifier is present, it recognizes the first and second AI processing devices as AI processing devices and can determine whether the devices are capable of AI processing in response to user commands based on the AI model identifier.

[0246] In addition, when requesting distributed processing of AI processing, the on-device AI processing distributed processing device can check the processing capacity of the first AI processing device and the second AI processing device, allocate the excess amount based on the processing capacity of the first AI processing device and the second AI processing device, and request distributed processing from the first AI processing device and the second AI processing device to each perform distributed processing of AI processing using the allocated allocated amount.

[0247] Here, when distributing the excess amount, the on-device AI processing distributed processing device can check the priorities assigned to the first AI processing device and the second AI processing device, and distribute the excess amount according to the priorities.

[0248] As an example, the on-device AI processing distributed processing device can obtain processing performance indexes from the first AI processing device and the second AI processing device, and assign priorities to the first AI processing device and the second AI processing device based on the processing performance indexes.

[0249] That is, the on-device AI processing distributed processing device can assign a higher priority to the device with a higher processing performance index between the first AI processing device and the second AI processing device.

[0250] Subsequently, when distributing the excess amount, the on-device AI processing distributed processing device can distribute the excess amount to the first AI processing device and the second AI processing device according to the priorities assigned to them, respectively.

[0251] Here, when distributing the excess amount, the on-device AI processing distributed processing device can allocate a first distribution amount of the total excess amount to the device assigned the highest priority among the first AI processing device and the second AI processing device, and allocate a second distribution amount remaining from the total excess amount excluding the first distribution amount to the device assigned the lowest priority.

[0252] In some cases, when distributing the excess amount, the on-device AI processing distributed processing device may distribute the excess amount equally, allocating a first allocated amount of the total excess amount to the first AI processing device and allocating a second allocated amount of the total excess amount that is the same as the first allocated amount to the second AI processing device.

[0253] In another case, when distributing the excess amount, the on-device AI processing distributed processing device may allocate a first distribution amount of the total excess amount to a first AI processing device, and allocate a second distribution amount of the total excess amount, which is different from the first distribution amount, to a second AI processing device.

[0254] Next, the second AI processing device establishes a communication connection with the on-device AI processing distributed processing device, and when it establishes a communication connection with the on-device AI processing distributed processing device, it checks whether a distributed processing request is input from the on-device AI processing distributed processing device. When a distributed processing request is input from the on-device AI processing distributed processing device, it extracts distributed processing information from the distributed processing request, performs AI processing based on the distributed processing information to generate a second AI processing result value, and provides the generated second AI processing result value to the on-device AI processing distributed processing device.

[0255] In this way, when the AI processing volume required to execute a user command exceeds the volume that can be processed by the device itself, the present disclosure selects an external or internal AI processing distributed processing target and requests distributed processing of the AI processing, thereby improving the AI processing speed of the device, the accuracy of the result value, and the quality of service.

[0256] In addition, the present disclosure can minimize power consumption of a device by distributing AI processing together with externally or internally located distributed processing targets, thereby reducing heat generation and improving performance and lifespan.

[0257] The present disclosure described above can be embodied as computer-readable code on a medium having a program recorded thereon. Computer-readable media include all types of storage devices that store data readable by a computer system. Examples of computer-readable media include hard disk drives (HDDs), solid-state disks (SSDs), silicon disk drives (SDDs), ROMs, RAMs, CD-ROMs, magnetic tapes, floppy disks, and optical data storage devices. The computer may also include a processor for an artificial intelligence device.

Claims

1. an input configured to accept user commands; a processor configured to perform AI processing responsive to the user instructions; The processor: When the user command is accepted, measuring the amount of AI processing required to execute the user command; If the measured AI processing amount exceeds the self-processing capacity, a target for AI processing distribution processing is selected, Request distributed processing of AI processing for the selected target of distributed processing of AI processing; After receiving a first AI processing result value from the AI processing distributed processing target, provide a final result value based on the first AI processing result value and the self-processed second AI processing result value; An on-device AI processing distributed processing device characterized by:

2. Measuring the amount of AI processing includes: Selecting at least one AI model in response to the received user command to execute the user command; measuring the amount of AI processing for executing the user instructions based on the computational complexity of the selected at least one AI model.

2. The on-device AI processing distributed processing device according to claim 1.

3. Measuring the amount of AI processing includes: Selecting one AI model and measuring the AI processing amount based on the amount of calculation that the one AI model can process; or selecting a plurality of AI models and measuring the AI processing amount based on the total amount of calculation that the plurality of AI models can process; 3. The on-device AI processing distributed processing device according to claim 2.

4. The processor is configured to determine whether the measured amount of AI processing is a self-processable amount; If it is determined that the AI processing amount exceeds the self-processing capacity amount, a target for the AI processing distribution processing is selected.

2. The on-device AI processing distributed processing device according to claim 1.

5. The processor: Determine whether or not there is an internal device capable of AI processing; If it is determined that the internal device is present, determining whether the internal device is currently in an inactive state; When it is determined that the internal device is currently in an inactive state, the internal device is selected as a target for the AI processing distributed processing. The on-device AI processing distributed processing device according to claim 4.

6. The processor: Calculating an excess amount, which is the difference between the measured AI processing amount and the self-processable amount, The selected AI processing distributed processing target confirms the processable amount, When the processable amount exceeds the excess amount, the system is configured to request the AI processing distribution processing target to perform distributed processing of the AI processing for the excess amount.

2. The on-device AI processing distributed processing device according to claim 1.

7. providing the final result value comprises: and generating the final result value corresponding to the user command by mapping the first AI processing result value received from the AI processing distributed processing target to a self-processed second AI processing result value.

2. The on-device AI processing distributed processing device according to claim 1.

8. An AI processing device that can be attached to an on-device AI processing distributed processing device, A communication unit communicably connected to the on-device AI processing distributed processing device; a memory for storing at least one AI model for AI processing; a processor that executes AI processing in response to a distributed processing request from the on-device AI processing distributed processing device; The processor: When attached to the on-device AI processing distributed processing device, a communication connection is established with the on-device AI processing distributed processing device; After being connected to the on-device AI processing distributed processing device, determining that a distributed processing request has been received from the on-device AI processing distributed processing device; After the distributed processing request is received from the on-device AI processing distributed processing device, distributed processing information is extracted from the distributed processing request; generating an AI processing result value by executing AI processing based on the distributed processing information; and providing the generated AI processing result value to the on-device AI processing distributed processing device. An AI processing device characterized by:

9. Extracting the distributed processing information includes: extracting the distributed processing information, including AI model information, location information, distributed processing amount information, and input data corresponding to the distributed processing portion, from the distributed processing request; The AI processing device according to claim 8 .

10. performing the AI processing includes: determining whether an AI model corresponding to the distributed processing request is pre-stored in the memory; When it is determined that the AI model is not stored in the memory, acquiring the AI model from an external server via the communication unit and storing the AI model in the memory. The AI processing device according to claim 8 .

Images