KR20230017073A - A system and method for controlling smart ai kit - Google Patents

Abstract

A control system for a smart AI kit according to an embodiment of the present invention comprises: an AI moving object that includes an AI module, and performs an operation determined on the basis of a sensor signal and a learning result data set; a server that generates an AI algorithm via machine learning by receiving operation information and state information according to mission performance of the AI moving object, and transmits the AI algorithm to the AI moving object; and a mission performance application module that is installed in a user terminal, and connects the server and the AI moving object to each other. The server may determine whether or not an operation abnormality has occurred with respect to the AI moving object performing the determined operation. According to the present invention, immediate operations are possible.

Description

Smart AI kit control system and control method {A SYSTEM AND METHOD FOR CONTROLLING SMART AI KIT}

The present invention relates to a smart AI kit control system and control method, and more particularly, a smart AI kit control system and control method capable of learning various AIs used in real life as a user processes a mission using an AI mobile body. It is about.

The module assembly is a device capable of performing simple operations by assembling a plurality of modules having functions such as buttons, lamps, motors, and communication, and simply coding driving software and loading the module assembly. For example, as shown in FIG. 11, a module assembly as shown in FIG. 11 (b) is created by combining several modules of FIG. 11 (a), and a simple operation is performed by loading software written by a coding learner necessary for this can be performed. Or, for example, the module assembly described in Prior Document 1 is representative.

Recently, as AI (Artificial Intelligence) technology is popular, there is a demand from learners for coding incorporating AI, so attempts to introduce AI module assembly to the above-described module assembly are also being made. However, as an additional module of the existing module assembly, it was difficult to create a module assembly that can be operated according to various situations even if it is simply an AI module assembly, for example, an additional hardware module for matrix operation to perform calculations in deep learning. Since there was no system that could change the learned data set or AI algorithm to suit various situations, or reset the sensor or actuator and load it into the AI module assembly, simply manufacturing and supplying the AI module assembly could operate the AI module assembly was unable to provide.

Therefore, it is necessary to develop a smart AI kit control system that can be operated immediately by various algorithms using an AI module assembly.

[Related technical literature]

Prior Document 1: Republic of Korea Patent Registration No. 10-2038822

The present invention has been proposed to solve the above problems, and an object of the present invention is to provide an AI algorithm or basic data DB set change or calibration suitable for problem solving in various situations, for example, when solving various problems, It is to provide a smart AI kit control system that can be operated immediately.

For example, in the process of solving the problem of an AI moving object finding a cat in a picture or finding a traffic light, it is changed to a basic data DB for finding a cat or a basic data DB for finding a traffic light, or the camera (sensor) or actuator of the mobile body It may be an object of the present invention to provide an AI kit control system capable of repositioning.

In order to solve the above problems, the smart AI kit control system according to an embodiment of the present invention includes an AI module, and performs an operation determined based on a sensor signal and a learning result data set, an AI mobile body; A server that generates an AI algorithm through machine learning by receiving operation information and status information according to the mission performance of the AI mobile body, and transmits the AI algorithm to the AI mobile body; and a mission performance application module that is installed in the user terminal and performs a connection between the server and the mobile AI, and the server can determine whether an operation abnormality has occurred with respect to the mobile AI performing the determined operation.

In this case, the server may determine mission information including the type of mission provided to the AI mobile body according to module assembly information including the type of submodule included in the AI mobile body.

In addition, whether or not an operation abnormality has occurred is when the mission is not completed within a predetermined time for each mission, when the degree of mission processing does not meet the standard at a specific time, when the timeout of mission performance is repeated, or when the transmitted photo If the data is data unrelated to the mission, it may be determined according to perception.

In addition, the server may further include an anomaly judgment unit, and may change a DB set for determining an operation of an AI mobile body when there is an anomaly in operation.

In addition, the server may further include an anomaly determining unit; and may generate a correction algorithm for the AI algorithm when there is an operating anomaly.

Also, the correction algorithm may include at least one of a pre-processing algorithm, a learning result data set, a post-processing algorithm, and a machine learning type.

In addition, the server may further include a calibration unit that performs physical correction on the AI module when the mission is not processed based on an operation according to a correction algorithm.

In addition, the server includes an anomaly determination unit, and the anomaly determination unit may provide a user interface for manually modifying at least some of the AI algorithms or changing settings when there is an operation abnormality.

On the other hand, the smart AI kit control method according to an embodiment of the present invention includes an AI module, and transmits mission information including a type of mission to an AI mobile body by a mission transmission unit of a server; Determining the degree of mission processing of the AI moving object according to mission information using a mission determination unit; Receiving and machine learning operation information and state information of the AI mobile body using a machine learning unit; and determining whether an abnormal operation of the AI mobile mission has occurred by an abnormality determining unit.

In this case, the server may determine mission information including the type of mission provided to the AI mobile body according to module assembly information including the type of submodule included in the AI mobile body.

In addition, whether or not an operation abnormality has occurred is when the mission is not completed within a predetermined time for each mission, when the degree of mission processing does not meet the standard at a specific time, when the timeout of mission performance is repeated, or when the transmitted photo If the data is data unrelated to the mission, it may be determined according to perception.

In addition, the server may change the DB set for determining the operation of the AI mobile body when there is an operation abnormality.

In addition, the server may generate a correction algorithm for the AI algorithm when there is an operational abnormality.

Also, the correction algorithm may include at least one of a pre-processing algorithm, a learning result data set, a post-processing algorithm, and a machine learning type.

In addition, the server may perform resetting or initialization of sensors or actuators of the AI moving object when the operation abnormality continues even by the correction algorithm.

In addition, the server may provide a user interface for manually modifying at least some of the AI algorithms or changing settings when there is an abnormal operation.

Other embodiment specifics are included in the detailed description and drawings.

Therefore, the smart AI kit according to the present invention is an optimal algorithm, for example, an optimal preprocessing algorithm or machine learning algorithm suitable for a learning mission, or an optimal learning data set so that the AI mobile body can immediately perform actions suitable for various situations. can be applied

In addition, according to the present invention, when the AI mobile body solves various situations, for example, various problems, it provides a change or calibration of an AI algorithm or basic data DB set suitable for problem solving, and can be operated immediately. A smart AI kit control system can be provided. there is.

For example, according to the present invention, in the process of solving a problem in which an AI moving object finds a cat in a picture or a traffic light, it is changed to a basic data DB for finding a cat or a basic data DB for finding a traffic light, or a camera (sensor) Alternatively, an AI kit control system capable of resetting the position of an actuator of a mobile body may be provided.

Effects according to the present invention are not limited by the contents exemplified above, and more various effects are included in the present specification.

1 is a schematic diagram of a smart AI kit control system according to an embodiment of the present invention.
2 is a block diagram for explaining detailed configurations of an AI mobile body, a user terminal, and a server according to an embodiment of the present invention.
3 is a flowchart for explaining the operation process of the smart AI kit control system according to an embodiment of the present invention.
4 is a diagram for illustratively explaining a mission according to an embodiment of the present invention.
5 is a flowchart of a maze finding mission performed by an AI mobile device according to an embodiment of the present invention.
6 to 8 are diagrams for explaining mission transfer between AI moving objects according to an embodiment of the present invention.
9 is a diagram illustrating a screen of an APP installed in a user terminal according to an embodiment of the present invention.
10 is a diagram for explaining transmission of a mission according to an embodiment of the present invention.
11 is a diagram for explaining the concept of module assembly in the prior art.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but can be implemented in various different forms. It is provided to fully inform you. In addition, for convenience of description, the size of components may be exaggerated or reduced in the drawings.

However, the following embodiments are provided to those skilled in the art to sufficiently understand the present invention, and may be modified in various forms, and the scope of the present invention is limited to the following examples. it is not going to be

Meanwhile, throughout the specification, when a certain part "includes" a certain component, it means that it may further include other components without excluding other components unless otherwise stated. In addition, the terms "~ unit" and "~ module" described in the specification mean a unit that processes at least one function or operation, which may be implemented as a hardware module, a software module, or a combination of hardware and software.

The above objects, features and advantages will become more apparent through the following detailed description in conjunction with the accompanying drawings, and accordingly, those skilled in the art to which the invention belongs will be able to easily implement the technical idea of the invention. .

Hereinafter, with reference to FIG. 1, a smart AI kit control system according to an embodiment of the present invention will be described. 1 is a schematic diagram of a smart AI kit control system according to an embodiment of the present invention.

In the smart AI kit control system 1000, the smart AI kit including the mission performance application installed in the AI mobile body 100 and the user terminal 200 controls the AI mobile body 100 according to the machine learning learning result data set suitable for the mission. By recognizing the situation, the AI moving object 100 can be controlled so that the AI moving object 100 can perform a mission.

In this embodiment, the AI movable body 100 is shown in the drawing as being implemented as a module assembly, but the present invention is not limited thereto. In the present invention, the AI mobile body 100 should be understood as a concept encompassing devices including moving means (eg, wheels, propellers, screws, etc.), calculation devices, network connection devices, sensors, and actuators.

Referring to FIG. 1 , the smart AI kit control system 1000 may include an AI mobile body 100, a user terminal 200, and a server 300.

The AI movable body 100 refers to a module assembly including an AI module, and may be assembled by combining an AI module and other sub-modules having other functions with a 'main module'. In this case, the main module may be a module capable of performing a computing function and a control function for the mobile AI 100 . Also, the main module may include AI hardware and separate AI modules may be coupled to the main module. In this specification, modules having functions other than the main module and the AI module will be referred to as submodules.

The AI mobile body 100 communicates with the user terminal 200 in which a mission performance application or program (hereinafter collectively referred to as a mission performance application) is installed, and the AI algorithm related to the mission from the user terminal 200 or the server 300, such as , the learning result data set, which is the learning result, and the pre-processing/post-processing algorithm can be received and the operation can be performed according to the AI algorithm.

In this specification, the AI algorithm and the correction algorithm to be described later in the future are preprocessing algorithms that enable driving of the AI mobile body 100, artificial intelligence algorithms such as CNN, RNN, ANN, DNN, LSTM, or the above-mentioned CNN, RNN, ANN , DNN, LSTM, etc. can encompass post-processing algorithms that determine actions after situation recognition through AI or learning result data sets generated by learning of artificial intelligence algorithms. Here, the preprocessing algorithm may refer to various processes such as data processing algorithms that are processed to be used as input data for artificial intelligence, for example, limiting a photo to a specific area or transforming sensor data into data of a specific area. . The post-processing algorithm may mean control according to the result of situational awareness of artificial intelligence.

The learning result data set may mean, for example, matrix operation coefficients generated by learning in a CNN or ANN. Alternatively, the learning result data set may mean a function for a curve when the AI algorithm is a curve fitting curve by regression analysis or SVM, etc., and may mean a random variable when the result of the AI algorithm is a random variable function. . That is, the learning result data set may mean an output generated as a result of machine learning. This learning result data set may be generated, for example, in the server 300 and loaded into the local AI mobile body 100 and utilized, or if the user terminal 200 or the AI mobile body 100 has sufficient resources, the user terminal ( 200) or the AI mobile body 100 may also be machine-learned and generated.

The user terminal 200 has a mission performance application module including information on missions to be transmitted to the AI mobile body 100 installed. The user terminal 200 may provide a connection between the server 300 and the moving AI 100, and display operation and status information of the moving AI. Therefore, the user terminal 200 is a device in which an application module providing a mission capable of performing coding learning by being connected to an AI mobile body can be installed. For example, the user terminal 200 is a smartphone, desktop, or laptop computer. , it may be a computing device such as a tablet PC.

The server 300 may transmit a mission to the AI mobile body 100 through a mission performance application module installed in the user terminal 200 . In addition, an AI algorithm for mission execution may be transmitted to the AI mobile body 100, and a mission performance application module may be executed to receive operation information and status information of the AI mobile body 100 according to the transmitted AI algorithm. In this case, the server 300 may determine mission information including a mission type to be transmitted to the AI mobile body 100 according to module assembly information including the sub-module type of the AI mobile body 100 . More detailed information about the server 300 will be described in detail with reference to FIG. 2 and the like.

2 is a block diagram for explaining detailed configurations of an AI mobile body, a user terminal, and a server according to an embodiment of the present invention.

Referring to FIG. 2 , the AI mobile body 100 may include a communication unit 110, a control unit 120, an execution unit 130, and a memory 140. Here, the communication unit 110, the control unit 120, the execution unit 130, and the memory 140 may be understood as a hardware module, a software module, or a combination of a hardware module and a software module.

The communication unit 110 is a component that receives a command for a mission from the server 300 and transmits operation information of the AI mobile body 100 based on the received mission to the server 300, via a USB port, LTE, or WiFi. Data can be transmitted and received between the AI mobile body 100 and the server 300 through a wired/wireless network such as ), Bluetooth, or the like.

More specifically, the communication unit 110 of the AI mobile body 100 may include an input/output terminal 110-1 and a Bluetooth receiving terminal 110-2. Here, the input/output terminal 110-1 is a USB port into which a USB cable disposed on one side of the main module is inserted so that mission information can be input from an external device, that is, the user terminal 200 in which the mission performance application module is installed. can However, if a mission signal can be received, a component terminal, a high definition multimedia interface (HDMI) terminal, and the like may be further included.

In addition, the Bluetooth receiver 110-2 may receive data from the user terminal 200 and transmit the data to the control unit 120 so as to perform short-range wireless communication with the user terminal 200 in which the mission performance application module is installed. Although the Bluetooth receiver 110-2 is regarded as a separate component, it may mean a network module among sub-modules assembled to the main module.

The control unit 120 is a component that controls the overall operation and state of the AI mobile body 100. In more detail, the control unit 120 controls the communication unit 110 to receive a mission from the mission performance application module when the input/output terminal 110-1 of the AI mobile body 100 is connected to an external device by wire/wireless. can

In addition, the controller 120 may further include a calibration unit that controls to calibrate the AI moving object 100 to find the cause when an error occurs during the operation of the AI moving object 100 according to the AI algorithm.

In addition, the control unit 120 further includes a memory controller for controlling the memory 140 to update and store operation information, operation result data, etc. on the memory for missions processed by the AI moving object 100 according to the AI algorithm. can do.

In addition, the controller 120 may control the memory so that the server 300 updates and stores the learning result data set generated by machine learning the operation information of the AI moving object 100 in the memory according to the machine learning logic.

The executor 130 may perform a calculation task of applying operation information of the AI moving object 100 generated according to the performance of the mission to the learning result data set stored in the memory 140 . Accordingly, the result data calculated by the execution unit 130 may be transmitted to the machine learning unit 130 of the server and used for machine learning, or the abnormality determination unit 340 determines whether the result data is abnormal. It can also be used for verification.

Referring to FIG. 2 , the AI mobile body 100 described above may be connected to the user terminal 200 through a wireless network communication network.

The user terminal 200 may include a communication unit 210 , a mission performance application module 220 and a display unit 230 . Meanwhile, the user terminal 200 may further include an input interface unit capable of transmitting a signal input by the user to the control unit 120 of the moving AI 100.

The communication unit 210 of the user terminal may transmit mission information provided from a mission performance application module installed in the terminal and an AI algorithm according to the mission information. In addition, the communication unit 210 may receive status information on whether or not the AI moving object 100 has processed a mission, the degree of processing, or whether there is a failure in real time and output it through the display unit 230 .

The mission performance application module 220 may provide mission information and an AI algorithm according to the mission information to the AI mobile body 100 through a communication unit.

Referring to FIG. 2 , the server 300 includes a mission transmission unit 310, a mission determination unit 320, a machine learning unit 330, an abnormality determination unit 340, a calibration unit 350, a mission transmission unit ( 360) and a storage unit 370.

The mission transmission unit 310 is a component that transmits a mission based on module information and sensing signals received from the AI mobile body 100 . In more detail, the mission transmission unit 310 may provide different types of missions according to module information received from the AI mobile body 100 . For example, among the submodules assembled in the main module of the AI mobile body 100, if wheels are included as means of transportation, missions that can be performed using wheels are provided, and missions that can be performed in the air are provided if propellers are included. can At this time, the mission may be provided when a specific DB and a specific AI algorithm reach an accurate AI judgment according to the processing result.

In addition, a submission may be transmitted according to a sensing signal sensed from the AI mobile body 100. Submissions perform certain actions when certain situations occur within the entire mission, such as taking a specific picture at a specific location or rotating three times when a specific signal is detected (when a specific sound is heard). means the mission Meanwhile, in this specification, missions such as finding a maze and finding an object may be referred to as overall missions. In addition, the mission transmission unit 310 may transmit an AI algorithm optimized for mission performance when transmitting a mission. Here, the AI algorithm optimized for mission performance may mean an algorithm capable of processing a mission in the shortest time within a standard time range. For example, an optimized AI algorithm recognizes a photo, a preprocessing algorithm that represents a specific sensing signal, such as a part of a photo that is most suitable for finding a path in a maze in the shortest time, for example, a 2cm above and below the center of a maze wall. At least one of the machine learning type information such as CNN used to do this, the learning result data set that is the result learned according to the above-mentioned machine learning type, or the post-processing algorithm that determines the operation according to the above-mentioned situational awareness by artificial intelligence can include At this time, different AI algorithms may be applied according to the type of mission.

The mission determination unit 320 is a component that determines whether or not the AI mobile body 100 has performed the transmitted mission. For example, the mission determination unit 320 determines that the mission has been performed if the mission is processed within a preset standard time for each mission, and sets a time-out if the mission is not processed within the standard time. It can be judged that the mission was not performed by processing it. In this specification, the reference time may mean a predetermined time for each mission.

The machine learning unit 330 is a component that performs machine learning according to the provided mission information and operation information for the mission information or sensor data of the AI moving object 100 . The AI moving object 100 may perform a mission based on the learning result data set generated through the machine learning unit 330. The machine learning unit 330 may include, for example, hardware modules and software modules for constructing neural network circuits such as CNNs, ANNs, and RNNs. However, it is not limited thereto, and may include hardware modules and software modules for performing other types of machine learning.

The abnormality determination unit 340 determines whether an abnormality has occurred in the operation of the AI movable body 100 while performing a mission, and when it is determined that there is an abnormality, it can transmit a supplemented AI algorithm, that is, a correction algorithm. According to another example, when it is determined that there is an abnormality, the abnormality determination unit 340 may change an optimal algorithm within a predetermined algorithm and transmit the changed optimal algorithm.

Accordingly, when an abnormality occurs while performing a mission, the AI movable body 100 can perform the mission by supplementing the operation according to the correction algorithm received from the abnormality determining unit 340 or the modified optimal algorithm. For example, a timeout occurs for the same mission several times by the mission determination unit 320, or a photo transmitted to the server from a sensor of the AI moving object 100, for example, a camera that is a sensor, is related to the corresponding mission. When it is not present, it can be determined that an error has occurred in mission performance.

The calibration unit 350 is a component that physically calibrates the AI movable body 100 when an abnormality occurs by the abnormality determining unit 340 . For example, a command is given to the control unit 120 of the AI mobile unit 100 to reset (eg, Wi-Fi/Bluetooth reconnection, power on/off, firmware version update) or initialization of the AI mobile unit 100. can transmit. In particular, the calibration unit 350 may reset the settings of sensors or actuators constituting the moving AI 100. That is, it may be to perform physical correction such as correcting a value of a sensor or an actuator. In this case, the contents of resetting (calibration) and the AI algorithm or basic data DB to be changed may be received, for example, from another AI mobile body within a short distance.

The mission transmission unit 360 may determine whether the mission is performed smoothly by transmitting the mission to another AI moving object according to the mission transmission rule when the operation is determined to be abnormal by the abnormality determination unit 340 . For example, the mission delivery unit 360 may transmit the mission to another mobile AI located in a short distance or another mobile AI that is processing a mission having a lower mission priority. A more detailed description of the mission transfer unit 360 will be described later with reference to FIGS. 6 to 8 .

The storage unit 370 is a configuration in which data generated when the AI mobile body 100 performs a mission is stored. For example, the storage unit 370 includes mission data, which is information on missions that can be performed according to module assembly information of the AI movable object 100, AI algorithms necessary for the mission performance of the AI movable object 100, and a learning result data set. , an APP installation file of the user terminal 200, and the like.

Hereinafter, with reference to FIGS. 3 to 5 , the overall operation process of the AI moving object performing the mission according to the AI algorithm and the operation process of the AI moving object 100 according to an example of the mission will be described.

3 is a flowchart for explaining the operation process of the smart AI kit according to an embodiment of the present invention. 4 is an exemplary view in which an AI mobile body performs a maze finding mission according to an embodiment of the present invention. 5 is a flowchart illustrating an operation process of an AI moving object performing the maze finding mission of FIG. 4 .

Referring to FIG. 3, the control method of the smart AI kit includes the step of transmitting mission information to the AI moving object 100 connected to the server through the wired/wireless network (S100), and determining the degree of mission processing of the AI moving object 100. Step (S200), machine learning by receiving operation information and state information of the AI moving object 100 for mission information (S300), detecting an abnormal operation of the AI moving object 100 (S400), and determined correction Based on the algorithm, it may include a step (S500) of transmitting a DB set of the AI moving object 100 and an AI algorithm to the AI moving object 100.

Here, the DB set means a set of data to be learned. For example, if the mission is a maze, the ground truth data set related to the maze, or if the mission is to find a specific part in a photo, that It may be a ground truth data set for a specific part. Alternatively, it may be a ground truth data set collected during a specific period or a collected ground truth data set collected in a specific region.

Meanwhile, in the control method of the smart AI kit, when an abnormal operation of the AI movable object 100 is still detected after step S500, the AI movable object 100 or some submodules of the AI movable object 100 are controlled by the calibration unit 350. A calibration step of resetting, rebooting or initializing may be further included. Here, some sub-modules to be reset, rebooted, or initialized may be sensors or actuators among the components of the AI mobile body 100 as described above.

In step S100, the mission transmission unit 310 of the server 300 transmits mission information to the AI mobile body 100 through the network.

After step S100, the AI mobile body 100 may perform a mission. For example, in the case of performing a maze mission as shown in FIGS. 4 and 5, the AI mobile body 100 may receive image data in real time by a camera module as a sub-module (S1). Then, based on the learning result data set received from the server 300, situational awareness (eg, whether or not a wall is approaching and determining the position of the wall, etc.) is performed, and direction data of the AI moving object 100 is generated according to the situational awareness result You can (S2). In step S3, the direction data may determine the direction through rule base or machine learning according to the result according to situational awareness.

Subsequently, in step S200, the mission determination unit 320 determines the degree of mission processing for the AI moving object 100 performing the mission based on the received mission information. The degree of mission processing means the mission performance rate of all missions, and may be, for example, the number of processing sub missions or a position in a maze at a corresponding time.

Subsequently, in step S300, the machine learning unit 330 may receive operation information and status information of the AI moving object 100 for mission information and perform machine learning thereon. Here, the operation information of the AI moving object 100 may include whether or not the AI moving object 100 processes a given mission, and log information about movements such as going straight, backward, and rotating for processing the mission. In addition, the status information of the AI mobile unit 100 may include a Wi-Fi connection status, a Bluetooth connection status, a battery status (remaining capacity), a firmware version status, and the like of the AI mobile unit 100.

In step S400, the abnormal determination unit 340 may detect an abnormal operation of the AI moving object 100 performing the mission. If it is determined that there is an abnormality in step S400, the abnormality determination unit 340 may transmit a correction algorithm so that the operation of the AI moving object 100 can be changed. For example, whether or not an operation is abnormal is determined when the degree of mission processing does not meet the standard within a predetermined time, when the mission is timed out several times, or when sensor data, for example, in the case of a maze mission, photo data is converted into a maze. It is possible to check whether or not the operation is abnormal through a judgment such as transmitting data not related to the .

In step S500, the abnormality determining unit 340 changes the DB set and/or AI algorithm that determines the operation of the AI mobile body 100 and transmits the changed DB set and/or algorithm to the AI mobile body 100. can

In this case, according to FIG. 9, the APP installed in the user terminal 200 may display a DB set to be changed (basic data DB 910) and an AI algorithm 920 or reset content 930. On the other hand, if the APP installed in the user terminal 200 is changed to a DB set to be additionally changed (basic data DB 910) and an AI algorithm 920 or a reset content 930, whether or not it can actually be performed A simulation 940 may be provided.

On the other hand, when an abnormality in operation is detected, the abnormality determination unit 340 may change only the DB set or change only a part of the AI algorithm, or if the abnormality in operation is not resolved by changing the DB set after changing the DB set first, the algorithm is changed. They can also be changed sequentially. Additionally, the abnormality determining unit 340 may perform a calibration operation through the calibration unit 350 when abnormalities are continuously detected through the DB set and/or algorithm. Alternatively, a user interface for notifying the user terminal 200 of abnormal operation and manually modifying at least some of the AI algorithms or changing settings may be provided to the user terminal. In this case, a preprocessing algorithm, for example, manually changing the camera's part of interest may be performed.

In addition, in the present invention, a DB set may mean a 'basic data DB set' so that AI determination can be properly performed.

Regarding the overall operation, for example, when the AI movable body 100 performs a mission to find a Russian cat using a DB set of Asian cat photos, an abnormal operation is detected by the abnormality determination unit 340 You can change the DB set to a DB set consisting of European cat photos, or change the DB set to a DB set consisting of SNS links so that Russian cats can be found on SNS. At this time, the methods used to identify Russian cats in the DB set consisting of European cat photos or the comments (eg, posts, comments, photos, pictures, etc.) on SNS in the DB set consisting of SNS links and Russian cats and Relationship analysis methods of may mean the 'AI algorithm' of the present invention. Alternatively, when performing a mission to find a puppy using a DB set of foot shape photos, if a motion abnormality is detected by the abnormality determining unit 340, the DB set is changed to a DB set of gait images And, the AI algorithm can be changed by changing the methods for determining the gait based on the gait DB set. As such, the basic data DB set may mean changing the configuration of the databases that are determined so that the AI judgment is properly made when performing the mission (low-level AI). In addition, changing the AI algorithm may be a method of selecting an appropriate method for AI judgment in a state where various methods are pre-stored, or a method of creating and generating various methods themselves through AI learning (high level AI ).

Alternatively, when the AI movable body 100 performs a maze search mission using the maze DB set 'A' and an operation abnormality is detected by the abnormal determination unit 340, the DB set related to the maze 'B' You can change the DB set so that you can use it to perform the mission. Alternatively, when the AI movable body 100 encounters an obstacle and operates according to the AI algorithm set to move straight by rapidly controlling the speed, if an operation error occurs, image data is received from the camera module to measure the height of the obstacle, and the height of the obstacle is measured. You can also change the post-processing algorithm that redirects on a basis.

After the step S600, the mission can be continuously performed again, and the degree of mission processing and detection of whether or not the operation of the AI moving object 100 is abnormal can be continuously repeated (S200 to S500).

6 to 8 are diagrams for explaining mission transfer between AI moving objects according to an embodiment of the present invention. 9 is a diagram illustrating a screen of an APP installed in a user terminal according to an embodiment of the present invention. 10 is a diagram for explaining transmission of a mission according to an embodiment of the present invention. 11 is a diagram for explaining the concept of module assembly in the prior art. Meanwhile, the mission delivery process of the smart AI kit may be a process performed after step S500 or step S600 in FIG. 3, and may be implemented in a control method of a smart AI kit composed of a plurality of AI moving objects 100. In FIGS. 6 and 7 , each of the plurality of AI mobile bodies will be referred to as a 'resource r' for convenience of description.

Referring to FIG. 6, the mission delivery method of the smart AI kit is the step of controlling the operation of the AI moving object 100 for the mission based on the AI algorithm according to the DB set that is the basic data for AI discrimination (S610), AI moving object (100) Determining whether the operation is abnormal (S620), displaying the display of the abnormal operation situation of the AI mobile body and confirming whether the mission is delivered (S630), selecting another AI mobile body based on the mission delivery rule (S640 ), determining whether other AI moving objects have required modules for the transmitted mission (S650), mission delivery step (S660), mission performance confirmation step of other AI moving objects (S670), and mission execution conditions of other AI moving objects A receiving step (S680) may be included.

After the operation of the moving AI 100 is controlled according to step S610, the mission determining unit 320 checks whether the moving AI 100 has an abnormal operation situation (S620). In this case, the occurrence of an abnormal operation situation may determine, for example, whether the AI movable body 100 has completed the mission within a predetermined time for the corresponding mission, or whether the AI movable body 100 does not have essential modules. . Alternatively, it may be determined that the operation is abnormal even when the degree of progress of the mission does not change within a predetermined time for the corresponding mission.

Then, in step S630, the display unit 230 of the user terminal may display the status information of the moving AI 100, and may provide a UI for inquiring whether or not to perform mission delivery from the user. 7 and 8, in the smart AI kit according to an embodiment of the present invention, four resources (r ₁ , r ₂ , r ₃ , r ₄ ) capable of delivering a mission are displayed on a user terminal, and each resource is It is assumed that four missions have reference times (t ₁ , t ₂ , t ₃ , t ₄ ). At this time, as shown in FIG. 7 , when an abnormal operation situation occurs when the first resource (r ₁ ) performs the third mission having a reference time t ₃ , the abnormality determination unit 340 detects this and the user terminal 200 ), and outputs a UI asking whether or not to transmit the mission on the screen of the APP. In this case, the user terminal 200 may output information (resource, reference time of the mission to be performed) on the abnormal occurrence 804 to the display unit 230 .

Referring to FIG. 8 , the display unit 230 displays a maze 801 in which a plurality of AI mobile bodies (a, b, c, and d) perform a maze finding mission and points A, B, C, and D displayed on the maze. An abnormality detection table 802 may be displayed by distinguishing between a case in which the operation is performed normally (803) and a case in which an operation error occurs (804) while performing missions a, b, c, and d.

After step S630, it is possible to select another AI mobile body to transmit the mission to another AI module based on the mission delivery rule (S640). The mission delivery rule may be, for example, 1) whether another AI mobile object to be delivered is located in the shortest distance, 2) the degree of excellence in mission performance, 3) essential modules required to perform the mission, and 4) the number of times the corresponding mission is performed. Meanwhile, in the present invention, the transmission process of missions between mission modules (resources) has been described as one-way, but missions may be traded in both directions between both modules. The excellent degree of mission performance capability may mean, for example, that the mission performance time for the corresponding mission is the fastest or the mission performance completion rate is high.

Referring to FIG. 7 , for example, a mission c (①) that the first resource (r ₁ ) failed to process in time t ₃ is a second resource located at the shortest distance (x) from the first resource (r ₁ ). r ₂ ).

It can be delivered even when other AI modules to which the mission is delivered are processing a mission with a lower priority than the mission to be delivered. For example, even if the closest AI moving object b is performing mission b, mission c has a higher priority, and if an abnormal operation occurs in AI moving object a, mission c may be delivered to AI moving object b. there is.

According to another example, the mission may be transmitted to another mobile AI that is processing a mission having a lower priority than the mission performed by the mobile AI (r ₂ ), but the mission may be delivered based on the mission processing rate of the mobile AI. For example, if a mission processed by another AI moving body has a low priority and the mission processing rate is less than 50% at the time when the mission is to be delivered, the mission is delivered as it is. However, at the time when the mission is to be delivered, if the mission processing rate of the low priority mission processed by other AI moving objects is 50% or higher, the mission can be delivered in the next time zone if the priority is low according to the priority of the mission in the next time zone. . However, if the priority of the mission in the next time zone is higher than that of the original AI moving object, the mission with a lower priority of the other AI moving object is different from the other AI moving object (however, the priority of the other AI moving object is Even if the priority is higher than the priority of the mission performed in

For example, referring to FIG. 10, when the original AI moving object r2 is trying to deliver the mission b', and the mission processing rate of the other AI moving object r3 exceeds 50%, the other AI moving object ( r3) Instead, the mission is transferred to another AI moving object (r1) that is performing a mission (c) whose priority is lower than the mission (b') of the original AI moving object (r2) and higher than the mission (c'') of the next time period. can

Accordingly, the storage unit of the server 300 may further store information about location information about moving AI objects, difficulty of missions, priority of mission processing, and the like.

At this time, it may be transmitted after checking whether essential modules for the mission to be delivered to other resources are provided (S650). In this case, whether or not essential modules are provided may be confirmed based on module assembly information of other AI movable bodies.

If other AI moving objects for the transmitted mission after step S650 do not have essential modules, other AI moving objects may be selected by applying the mission transfer rule again.

On the other hand, if the other AI moving object for the transmitted mission after step S650 has an essential module, the server 300 may transmit the mission to the selected other AI moving object (S660).

In addition, the server 300 may check the performance of the mission of the selected other AI mobile object. For example, whether or not the selected other AI mobile object performs the mission can be known by whether an operation abnormality occurs during the reference time (S670).

When the mission performance of the selected other AI mobile body is confirmed, the confirmed mission performance conditions, such as DB set, AI algorithm, and calibration setting values, etc. The mission can be performed again (S680).

Therefore, according to the present invention, the smart AI kit is an optimal algorithm so that the AI mobile body can immediately perform actions suitable for various situations, such as an optimal preprocessing algorithm or machine learning algorithm suitable for a learning mission, or an optimal learning data set, Since the optimal post-processing algorithm can be applied, it is possible to maintain learner's interest in coding and maximize the learning effect.

In addition, the smart AI kit according to the present invention can enhance learners' understanding of AI algorithms by directly checking the operation process of AI moving objects operated by various algorithms by interlocking AI modules with mobile devices.

As a result, it is possible to remarkably increase the learning effect of AI training for users using the smart AI kit.

Although the embodiments of the present invention have been described in more detail with reference to the accompanying drawings, the present invention is not necessarily limited to these embodiments, and may be variously modified without departing from the technical spirit of the present invention. . Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain, and the scope of the technical idea of the present invention is not limited by these embodiments. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. The protection scope of the present invention should be construed according to the claims below, and all technical ideas within the equivalent range should be construed as being included in the scope of the present invention.

1000: Smart AI control system
100: AI mobile body 200: user terminal
300: server 400: network

Claims (15)

An AI mobile body that includes an AI module and performs an operation determined based on a sensor signal and a learning result data set;
a server for generating an AI algorithm through machine learning by receiving operation information and status information according to mission performance of the AI moving object, and transmitting the AI algorithm to the AI moving object; and
It includes a mission performance application module installed in a user terminal and performing a connection between the server and the AI mobile body,
The smart AI kit control system, wherein the server determines whether an operation abnormality has occurred with respect to the AI mobile body performing the determined operation. According to claim 1,
The server determines mission information including the type of mission provided to the AI mobile body according to module assembly information including the type of submodule included in the AI mobile body. According to claim 1,
Whether or not the above operation abnormality has occurred,
If the mission is not completed within the predetermined time for each mission, if the degree of mission processing does not meet the standard at a specific time, if the mission execution timeout is repeated, or if the transmitted photo data is data that is not related to the mission A smart AI kit control system determined by cognition. According to claim 1,
The server includes an anomaly determination unit, and the anomaly determination unit changes a DB set for determining the operation of the AI mobile body when there is an operation abnormality. According to claim 1,
The server includes an anomaly determination unit, and the anomaly determination unit generates a correction algorithm for the AI algorithm when there is an operation abnormality. According to claim 5,
The correction algorithm includes at least one of a pre-processing algorithm, the learning result data set, a post-processing algorithm, and a machine learning type, smart AI kit control system. According to claim 5,
The server,
A calibration unit for performing physical correction on the AI module when the mission is not processed based on the operation according to the correction algorithm; further comprising a smart AI kit control system. Transmitting a mission command to an AI module connected to a server through a network using a mission transmission unit;
Determining the degree of mission processing of the AI mobile body using a mission determination unit;
Receiving and machine learning operation information and state information of the AI mobile body using a machine learning unit; and
, Smart AI kit control method comprising: determining whether an abnormal operation of the AI mobile mission has occurred by an abnormality determining unit. According to claim 8,
Wherein the server determines mission information including the type of mission provided to the AI mobile body according to module assembly information including the type of submodule included in the AI mobile body. According to claim 8,
Whether or not the above operation abnormality has occurred is when the mission is not completed within a predetermined time for each mission, when the degree of mission processing does not meet the standard at a specific time, when the timeout of mission execution is repeated, or when the transmitted photo data A smart AI kit control method, which is determined by recognition if is data unrelated to the mission. According to claim 8,
Wherein the server changes a DB set for determining the operation of the AI mobile body when there is an operation abnormality. According to claim 9,
Wherein the server generates a correction algorithm for the AI algorithm when there is an operation abnormality. According to claim 12,
The correction algorithm includes at least one of a preprocessing algorithm, the learning result data set, a postprocessing algorithm, and a machine learning type, smart AI kit control method. According to claim 12,
Wherein the server resets or initializes a sensor or an actuator of the AI moving body when the operation abnormality continues even by the correction algorithm. According to claim 12,
The server provides a user interface for manually modifying at least some of the AI algorithms or changing settings when there is an operation abnormality.

Images