Electronic device and control method thereof

Abstract

An electronic device and a control method thereof are provided. The electronic device obtains non-collision data corresponding to a non-collision state of a vehicle through at least one sensor, obtains generated data corresponding to the non-collision state through a neural network model trained on the basis of the non-collision data, obtains a loss related to a difference between the generated data and collected data when the collected data related to the state of the vehicle is obtained, and identifies whether the collected data indicates the non-collision state, on the basis of the obtained loss.

Description

Electronic device and control method thereof

[0001] The present disclosure relates to an electronic device and a method for controlling the electronic device, and specifically, to an electronic device capable of acquiring data corresponding to a non-collision state of a vehicle and a method for controlling the same.

[0002] Technology that detects whether a vehicle has collided or the likelihood of a collision through data analysis is gaining attention. Data indicating a vehicle's collision status (hereinafter referred to as "collision data") can be used to analyze the causes of accidents and prevent them, as well as to enhance vehicle design and safety. Furthermore, data indicating a vehicle's collision status can be used to train neural network models used in autonomous driving and to improve algorithms.

[0003] However, securing a sufficient amount of crash data is difficult. For example, crash data can be obtained by receiving user feedback after an actual vehicle collides with another vehicle or obstacle. However, this method not only requires user consent to receive feedback but can also cause inconvenience to users because feedback is requested only after an accident. Crash data can also be obtained through actual vehicle crash tests, but this method has limitations in that it consumes enormous costs and time.

[0004] Meanwhile, if a neural network model related to identifying whether a vehicle has collided is trained using a small amount of collision data without securing a sufficient amount, overfitting of the neural network model may occur due to the biased characteristics of the small amount of collision data, and consequently, the performance of the neural network model may degrade.

[0005] An electronic device and a method for controlling the same are provided, which can efficiently and effectively perform labeling regarding whether data is collision data or non-collision data.

[0006] Additional aspects will be disclosed in part in the following description, and in part will be apparent from the description or can be learned through the practice of the presented embodiments.

[0007] According to one aspect of the present disclosure, an electronic device comprises at least one sensor, at least one memory for storing instructions, and at least one processor including a processing circuit, wherein when the instructions are executed individually or collectively by the at least one processor, the electronic device acquires non-collision data corresponding to a non-collision state of a vehicle through the at least one sensor, acquires generated data corresponding to the non-collision state through a neural network model learned based on the non-collision data, and when collected data related to the state of the vehicle is acquired, acquires a loss related to the difference between the generated data and the collected data, and identifies whether the collected data indicates the non-collision state based on the loss.

[0008] The above at least one sensor includes an accelerometer and a gyroscope, and when the instructions are executed individually or collectively by the above at least one processor, the electronic device may acquire acceleration data through the accelerometer, and when an impact applied to the electronic device is detected based on the acceleration data, acquire angular velocity data through the gyroscope, and acquire non-collision data based on the acceleration data and the angular velocity data.

[0009] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may identify whether the collected data indicates the non-collision state based on the loss and the user feedback when user feedback is received regarding whether the collected data indicates the non-collision state.

[0010] The above at least one sensor further includes a barometric pressure sensor, and when the instructions are executed individually or collectively by the above at least one processor, the electronic device may acquire the non-collision data based on the acceleration data, the angular velocity data, and the barometric pressure data when barometric pressure data is acquired through the barometric pressure sensor.

[0011] The above at least one sensor further includes an audio sensor, and when the instructions are executed individually or collectively by the above at least one processor, the electronic device may acquire the non-collision data based on the acceleration data, the angular velocity data, and the audio data when audio data is acquired through the audio sensor.

[0012] The above at least one sensor further includes a GNNS sensor (Global Navigation Satellite System), and when the instructions are executed individually or collectively by the above at least one processor, the electronic device may acquire the non-collision data based on the acceleration data, the angular velocity data, and the position data when position data regarding the position of the electronic device is acquired through the GNNS sensor.

[0013] The electronic device may further include a communication interface, and when the instructions are executed individually or collectively by the at least one processor, the electronic device may acquire the non-collision data based on the acceleration data, the angular velocity data, and the external data when external data is received from an external device through the communication interface.

[0014] When the above instructions are executed individually or collectively by the at least one processor, the electronic device comprises a neural network model including a generator that generates the generated data and a discriminator that obtains a probability regarding whether the input data corresponds to the non-collision data, and when the above instructions are executed individually or collectively by the at least one processor, the electronic device obtains a first loss based on the difference between the generated data and the collected data, obtains a second loss based on the difference between the feature value obtained by inputting the generated data into the discriminator and the feature value obtained by inputting the collected data into the discriminator, and obtains the loss based on the weighted sum of the first loss and the second loss.

[0015] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may update the non-collision data based on the collected data when the collected data is identified as representing the non-collision state, and train the neural network model based on the updated non-collision data.

[0016] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may train the neural network model based on the loss.

[0017] According to one aspect of the present disclosure, a control method for an electronic device comprises: acquiring non-collision data corresponding to a non-collision state of a vehicle through at least one sensor; acquiring generated data corresponding to the non-collision state through a neural network model learned based on the non-collision data; acquiring a loss related to the difference between the generated data and the collected data when collected data related to the state of the vehicle is acquired; and identifying whether the collected data indicates the non-collision state based on the loss.

[0018] The above at least one sensor includes an accelerometer and a gyroscope, and the control method of the electronic device may further include the step of acquiring acceleration data through the accelerometer, the step of acquiring angular velocity data through the gyroscope when an impact applied to the electronic device is detected based on the acceleration data, and the step of acquiring non-collision data based on the acceleration data and the angular velocity data.

[0019] The step of identifying whether the above-mentioned non-collision state is indicated may include, when user feedback is received regarding whether the above-mentioned collected data indicates the above-mentioned non-collision state, the step of identifying whether the above-mentioned collected data indicates the above-mentioned non-collision state based on the loss and the user feedback.

[0020] The above at least one sensor further includes a barometric pressure sensor, and the step of acquiring the non-collision data may be to acquire the non-collision data based on the acceleration data, the angular velocity data, and the barometric pressure data when barometric pressure data is acquired through the barometric pressure sensor.

[0021] The above at least one sensor further includes an audio sensor, and the step of acquiring the non-collision data may be to acquire the non-collision data based on the acceleration data, the angular velocity data, and the audio data when audio data is acquired through the audio sensor.

[0022] The above at least one sensor further includes a GNNS sensor (Global Navigation Satellite System), and the step of acquiring the non-collision data may be to acquire the non-collision data based on the acceleration data, the angular velocity data, and the position data when position data regarding the position of the electronic device is acquired through the GNNS sensor.

[0023] The step of acquiring the above non-collision data may be performed such that, when external data is received from an external device, the non-collision data is acquired based on the acceleration data, the angular velocity data, and the external data.

[0024] When the above instructions are executed individually or collectively by the at least one processor, the electronic device, the neural network model includes a generator that generates the generated data and a discriminator that obtains a probability regarding whether the input data corresponds to the non-collision data, and the step of obtaining the loss may include a step of obtaining a first loss based on the difference between the generated data and the collected data, a step of obtaining a second loss based on the difference between the feature value obtained by inputting the generated data into the discriminator and the feature value obtained by inputting the collected data into the discriminator, and a step of obtaining the loss based on the weighted sum of the first loss and the second loss.

[0025] The control method of the electronic device may further include the steps of updating the non-collision data based on the collected data and training the neural network model based on the updated non-collision data when the collected data is identified as representing the non-collision state.

[0026] According to one aspect of the present disclosure, a non-transient computer-readable recording medium comprising a program for executing a method of controlling an electronic device, wherein the method of controlling the electronic device comprises: acquiring non-collision data corresponding to a non-collision state of a vehicle through at least one sensor; acquiring generated data corresponding to the non-collision state through a neural network model learned based on the non-collision data; acquiring a loss related to the difference between the generated data and the collected data when collected data related to the state of the vehicle is acquired; and identifying whether the collected data indicates the non-collision state based on the loss.

[0027] Other aspects, features, and advantages of one embodiment according to the present disclosure will become more apparent from the following detailed description in conjunction with the accompanying drawings.

[0028] FIG. 1 is a block diagram showing the configuration of an electronic device according to one embodiment of the present disclosure,

[0029] FIG. 2 is a diagram for explaining the process of acquiring non-collision data according to one embodiment of the present disclosure,

[0030] FIG. 3 is a diagram illustrating the verification process of collected data and the learning process of a neural network model according to one embodiment of the present disclosure,

[0031] FIG. 4 is a block diagram showing the configuration of an electronic device according to one embodiment of the present disclosure,

[0032] FIG. 5 is a diagram showing acceleration data corresponding to the collision state of a vehicle,

[0033] FIG. 6 is a diagram showing audio data corresponding to the collision state of a vehicle,

[0034] FIG. 7 is a drawing showing a user interface according to one embodiment of the present disclosure,

[0035] FIG. 8 is a block diagram for explaining the specific configuration of an electronic device according to one embodiment of the present disclosure, and,

[0036] FIG. 9 is a flowchart illustrating a method for controlling an electronic device according to one embodiment of the present disclosure.

[0037] The embodiments described herein are subject to various modifications and may have various forms; specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the scope of specific embodiments and should be understood to include various modifications, equivalents, and / or alternatives of the embodiments of the present disclosure. In relation to the description of the drawings, similar reference numerals may be used for similar components.

[0038] In describing the present disclosure, if it is determined that a detailed description of related known functions or configurations could unnecessarily obscure the essence of the present disclosure, such detailed description is omitted.

[0039] Additionally, the following embodiments may be modified in various other forms, and the scope of the technical concept of the present disclosure is not limited to the following embodiments. Rather, these embodiments are provided to make the present disclosure more faithful and complete and to fully convey the technical concept of the present disclosure to those skilled in the art.

[0040] The terms used in this disclosure are used merely to describe specific embodiments and are not intended to limit the scope of the rights. The singular expression includes the plural expression unless the context clearly indicates otherwise.

[0041] In the present disclosure, expressions such as “have,” “may have,” “include,” or “may include” indicate the presence of such features (e.g., numerical values, functions, actions, or components such as parts) and do not exclude the presence of additional features.

[0042] In the present disclosure, expressions such as “A or B,” “at least one of A or / and B,” or “one or more of A or / and B” may include all possible combinations of items listed together. For example, “A or B,” “at least one of A and B,” or “at least one of A or B” may refer to cases including (1) at least one A, (2) at least one B, or (3) both at least one A and at least one B.

[0043] Expressions such as "first," "second," "first," or "second" used in this disclosure may modify various components regardless of order and / or importance, and are used only to distinguish one component from another and do not limit said components.

[0044] Where it is stated that a certain component (e.g., a first component) is "(operatively or communicatively) coupled with / to" or "connected to" another component (e.g., a second component), it should be understood that the said certain component may be directly connected to the said other component or connected through another component (e.g., a third component).

[0045] On the other hand, when it is stated that a certain component (e.g., a first component) is "directly connected" or "directly coupled" to another component (e.g., a second component), it may be understood that no other component (e.g., a third component) exists between said certain component and said other component.

[0046] As used in this disclosure, the expression “configured to” may be replaced, depending on the context, with, for example, “suitable for,” “having the capacity to,” “designed to,” “adapted to,” “made to,” or “capable of.” The term “configured to” may not necessarily mean only “specifically designed to” in hardware.

[0047] Instead, in some situations, the expression “device configured to do something” may mean that the device is “capable of doing something” together with other devices or components. For example, the phrase “processor configured (or set) to perform A, B, and C” may mean a dedicated processor for performing those operations (e.g., an embedded processor), or a generic-purpose processor (e.g., a CPU or application processor) capable of performing those operations by executing one or more software programs stored in a memory device.

[0048] In the embodiments, a 'module' or 'part' performs at least one function or operation and may be implemented in hardware or software, or a combination of hardware and software. Additionally, a plurality of 'modules' or a plurality of 'parts' may be integrated into at least one module and implemented by at least one processor, except for the 'module' or 'part' that needs to be implemented in specific hardware.

[0049] Meanwhile, the various elements and areas in the drawings are depicted schematically. Accordingly, the technical concept of the present invention is not limited by the relative sizes or spacing depicted in the attached drawings.

[0050] Hereinafter, embodiments according to the present disclosure are described in detail with reference to the attached drawings so that those skilled in the art can easily implement them.

[0051] FIG. 1 is a block diagram showing the configuration of an electronic device (100) according to one embodiment of the present disclosure.

[0052] The ‘electronic device (100)’ refers to a device capable of performing labeling on whether data is collision data or non-collision data. Specifically, the electronic device (100) can acquire non-collision data corresponding to the non-collision state of a vehicle based on sensing data, and can train a neural network model using the non-collision data. For example, the electronic device (100) can be implemented as various devices such as a smartphone, tablet PC, TV, server, etc., provided that the type of electronic device (100) according to the present disclosure is not limited thereto.

[0053] As illustrated in FIG. 1, the electronic device (100) may include a sensor (110), a memory (120), and a processor (130).

[0054] The sensor (110) can detect various information inside and outside the electronic device (100). The sensor (110) can be implemented as at least one sensor (110). That is, the sensor (110) may include one or more sensors (110). Specifically, the sensor (110) may include an accelerometer (111), a gyroscope (112), a barometer (113), an audio sensor (114), and a GNNS sensor (115) (Global Navigation Satellite System). In addition, the sensor (110) may include various types of sensors such as a lidar sensor, a motion sensor, a temperature sensor, a humidity sensor, an infrared sensor, etc.

[0055] In one embodiment, the processor (130) can obtain various information, such as the movement of the electronic device (100), changes in speed, and changes in the internal environment, through at least one sensor (110). In particular, the processor (130) can obtain data corresponding to the collision state or non-collision state of the vehicle (specific meanings will be described later) through at least one sensor (110). Various types of sensors (110) and related embodiments will also be described later.

[0056] At least one instruction regarding an electronic device (100) may be stored in the memory (120). Additionally, an operating system (O / S) for operating the electronic device (100) may be stored in the memory (120). Furthermore, various software programs or applications for operating the electronic device (100) may be stored in the memory (120) according to various embodiments of the present disclosure. Additionally, the memory (120) may include semiconductor memory such as flash memory or magnetic storage media such as a hard disk.

[0057] Specifically, various software modules for operating an electronic device (100) according to various embodiments of the present disclosure may be stored in the memory (120), and the processor (130) may control the operation of the electronic device (100) by executing the various software modules stored in the memory (120). That is, the memory (120) is accessed by the processor (130), and reading / writing / modifying / deleting / updating of data by the processor (130) may be performed.

[0058] Meanwhile, in the present disclosure, the term memory (120) may be used to include memory (120), ROM, RAM, or a memory card (e.g., micro SD card, memory stick) mounted in the electronic device (100) within the processor (130).

[0059] In one embodiment, non-collision data, generated data, and collected data may be stored in the memory (120). Acceleration data, angular velocity data, atmospheric pressure data, audio data, position data, etc. may be stored in the memory (120). Data for a neural network model may be stored in the memory (120). In addition, various information necessary within the scope of achieving the purpose of the present disclosure may be stored in the memory (120), and the information stored in the memory (120) may be updated as it is received from an external device or input by a user.

[0060] The processor (130) controls the overall operation of the electronic device (100). Specifically, the processor (130) may be connected to the configuration of the electronic device (100), which includes a communication interface (140) and a memory (120). The processor (130) may include a processing circuit and may be implemented as at least one processor (130). That is, the processor (130) may be implemented as one or more processors (130). The processor (130) may control the operation of the electronic device (100) by executing instructions stored in the memory (120) individually or collectively.

[0061] The processor (130) can be implemented in various ways. For example, the processor (130) can be implemented as at least one of an Application Specific Integrated Circuit (ASIC), an embedded processor, a microprocessor, hardware control logic, a hardware Finite State Machine (FSM), and a Digital Signal Processor (DSP). Meanwhile, in this disclosure, the term processor (130) may be used to include a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), and an MPU (Micro Processor Unit), etc.

[0062] In one embodiment, the processor (130) can acquire non-collision data corresponding to a non-collision state of the vehicle and can identify whether the collected data indicates a non-collision state. Before describing various embodiments implemented by the processor (130), the collision data and non-collision data according to the present disclosure will first be described.

[0063] 'Collision data' can collectively refer to data corresponding to the collision state of a vehicle. Specifically, collision data can represent a pattern of data change resulting from a collision between a vehicle and another vehicle, an obstacle, etc. For example, when a vehicle collides with an obstacle, acceleration data obtained through an accelerometer (111) of a smartphone placed inside the vehicle can indicate that acceleration increased rapidly in the direction of the collision within a very short period of time at the time of the collision, and can represent a pattern of vibration that occurred in the vehicle after the collision.

[0064] 'Non-collision data' can collectively refer to data corresponding to a vehicle's non-collision state. Specifically, non-collision data may indicate that the vehicle's movement did not change abruptly. Additionally, non-collision data may indicate that while the vehicle's movement changed abruptly, the degree or pattern of such change is distinguishable from cases where the vehicle collided.

[0065] For example, if a smartphone (electronic device (100)) placed inside a vehicle falls to the floor of the vehicle while the vehicle is driving normally, the data obtained through the accelerometer (111) of the smartphone may indicate that the smartphone has received a strong impact, but the degree or pattern of change may differ from that of a collision with a vehicle. As another example, if a user collides with another user while skiing with a smartphone (electronic device (100)), the user and the smartphone may receive a strong impact, but the degree of impact may be weaker compared to the impact caused by a collision with a vehicle.

[0066] Collision data and non-collision data may be a set of two or more types of sensing data. For example, collision data may be a set of acceleration data and angular velocity data corresponding to the collision state of the vehicle. As another example, non-collision data may be a set of acceleration data and barometric pressure data corresponding to the non-collision state of the vehicle.

[0067] Collision data and non-collision data may be data obtained through a sensor (110) included in the vehicle, or data obtained through a sensor (110) included in a device (e.g., electronic device (100)) placed inside the vehicle. For convenience of explanation, the following description is based on the premise that the processor (130) obtains non-collision data through at least one sensor (110) included in the electronic device (100).

[0068] The processor (130) can acquire non-collision data corresponding to a non-collision state of the vehicle through at least one sensor (110). Specifically, if the sensing data acquired through at least one sensor (110) corresponds to a non-collision state of the vehicle, the processor (130) can identify the sensing data as non-collision data.

[0069] For example, when a vehicle is driving normally but a smartphone placed inside the vehicle falls to the floor of the vehicle, or when a user rides an amusement ride or skis while holding the smartphone, the sensing data obtained through at least one sensor (110) may correspond to a non-collision state rather than a collision state of the vehicle.

[0070] In this case, the processor (130) may designate the sensing data as non-collision data and store it in memory (120). Designating the sensing data as non-collision data may involve identifying that the set of data or each of the data included in the sensing data is non-collision data, and storing it in memory (120) with added information indicating that it is non-collision data.

[0071] In particular, the processor (130) may designate only the sensing data obtained through at least one sensor (110) as non-collision data if the likelihood of the sensing data corresponding to a non-collision state of the vehicle is above a threshold level. This is because, as described below, non-collision data is used for training a neural network model, and for example, if the sensing data exhibits a pattern corresponding to a boundary region (gray region) between the collision state and the non-collision state of the vehicle, the learning effect of the neural network model may be reduced.

[0072] The processor (130) can obtain generated data corresponding to a non-collision state through a neural network model trained based on non-collision data. A 'neural network model' refers to a model trained to generate data corresponding to a non-collision state. The neural network model can be trained based on non-collision data rather than collision data. The structure of the neural network model and the training process will be explained in more detail with reference to FIG. 3.

[0073] 'Generated data' can collectively refer to data corresponding to a non-collision state. Generated data refers to fake data that simulates a non-collision state, which is data generated through a neural network model rather than data collected through a sensor (110). Generated data can be distinguished from collected data, which is real data obtained through a sensor (110), in that it is data generated through a neural network model.

[0074] When collected data related to the state of the vehicle is obtained, the processor (130) can obtain a loss related to the difference between the generated data and the collected data. 'Collected data' may collectively refer to data obtained through the sensor (110). Collected data can be distinguished from generated data generated through a neural network model in that it is data obtained, i.e., collected, through the sensor (110). Additionally, collected data can be distinguished from non-collision data in that it is data that has not been used for training the neural network model.

[0075] The collected data may be obtained through a sensor (110) included in an electronic device (100), or through a sensor (110) included in an external device (e.g., a server, multiple user terminals, etc.).

[0076] "Loss" may collectively refer to information related to the difference between generated data and collected data. Specifically, loss may represent the results of comparing how similar the distributions of collected data and generated data are, or it may represent the results of quantifying the difference between collected data and generated data. The types of loss and the acquisition process according to the present disclosure will be explained in more detail with reference to FIG. 3.

[0077] The processor (130) can identify whether the collected data indicates a non-collision state based on the acquired loss. Specifically, the processor (130) can identify whether the collected data indicates a non-collision state or a collision state based on whether the acquired loss is similar to the generated data by a certain level or more.

[0078] For example, the processor (130) can obtain a loss value (or score) corresponding to the acquired loss. If the loss value is greater than or equal to a preset threshold, the processor (130) can identify that the collected data indicates a non-collision state. If the loss value is less than a preset threshold, the processor (130) can identify that the collected data indicates a collision state.

[0079] According to the embodiments described above with reference to FIG. 1, the electronic device (100) trains a neural network model using non-collision data instead of collision data, which is difficult to obtain a large amount of data, and uses the trained neural network model to efficiently and effectively perform labeling on whether the collected data is collision data or non-collision data.

[0080] FIG. 2 is a diagram illustrating the process of acquiring non-collision data according to one embodiment of the present disclosure.

[0081] As illustrated in FIG. 2, at least one sensor (110) may include an accelerometer (111), a gyroscope (112), and a barometric pressure sensor (113), and a processor (130) may acquire sensing data through the accelerometer (111), the gyroscope (112), and the barometric pressure sensor (113).

[0082] The accelerometer (111) can measure the linear acceleration of the electronic device (100). The acceleration data obtained through the accelerometer (111) may include three-axis acceleration values ​​(in m / s² or g units) based on the x, y, and z axes. Based on the acceleration data, the processor (130) can obtain information about the motion state, movement, velocity change, etc. of the electronic device (100).

[0083] The gyroscope (112) can measure the angular velocity of the electronic device (100). The angular velocity data obtained through the gyroscope (112) may include rotational velocity values ​​(in units of ° / s) based on the x, y, and z axes. Based on the angular velocity data, the processor (130) can obtain information about the rotational motion, rotational speed, attitude change, etc. of the electronic device (100).

[0084] The pressure sensor (113) can measure the atmospheric pressure of the surrounding air. Specifically, the pressure data obtained through the pressure sensor (113) may include air pressure values ​​(in hPa units). Based on the pressure data, the processor (130) can obtain information about changes in pressure due to movement or collision of the electronic device (100).

[0085] In the present disclosure, 'sensing data' may collectively refer to data or a set of data obtained through at least one sensor (110). For example, the sensing data may include at least some of acceleration data, angular velocity data, and atmospheric pressure data.

[0086] As illustrated in FIG. 2, when sensing data is acquired through at least one sensor (110), the processor (130) can input the sensing data into a non-collision data acquisition module to acquire non-collision data. The 'non-collision data acquisition module' can identify data corresponding to the non-collision state of the vehicle among the sensing data and designate it as non-collision data.

[0087] For example, information regarding acceleration data, angular velocity data, atmospheric pressure data, etc. corresponding to the collision state of a vehicle may be predefined and stored in memory (120). Based on the information stored in memory (120), the processor (130) can identify whether a 3-axis acceleration value based on the x, y, and z axes corresponds to the collision state of a vehicle, whether a rotational velocity value based on the x, y, and z axes corresponds to the collision state of a vehicle, and whether a change in atmospheric pressure according to atmospheric pressure data corresponds to the collision state of a vehicle. An embodiment related to identifying the collision / non-collision state of a vehicle using a plurality of sensors (110) will be described later in the description of FIGS. 4 to 7.

[0088] When the sensing data is identified as corresponding to a non-collision state of the vehicle, the processor (130) may designate the sensing data as non-collision data and store it in memory (120). The non-collision data stored in memory (120) can be used for training a neural network model as described below, and thus the non-collision data may be referred to as training data.

[0089] In one embodiment, the processor (130) can control the operating conditions of each sensor (110) based on the power consumption of each of at least one sensor (110). Specifically, the processor (130) can control the sensor (110) with low power consumption to operate at all times, and can control the sensor (110) with high power consumption to operate temporarily under specific conditions.

[0090] For example, the accelerometer (111) and the barometric pressure sensor (113) do not consume much power, so they can be operated at all times. Therefore, the accelerometer (111) and the barometric pressure sensor (113) can receive sensing data for a certain period of time before and after the point in time when an event with a possibility of a vehicle collision occurs. On the other hand, the gyroscope consumes a large amount of power, so it can be operated temporarily when an event with a possibility of a vehicle collision occurs. Therefore, the gyroscope can receive sensing data after the point in time when the event occurs.

[0091] 'Events with a possibility of collision of the vehicle' may be pre-set and may include the detection of an impact on the electronic device (100), a sudden change in the acceleration of the electronic device (100), a sudden change in the air pressure of the electronic device (100), etc.

[0092] For example, the processor (130) can acquire acceleration data through the accelerometer (111). Once acceleration data is acquired, the processor (130) can detect an impact applied to the electronic device (100) based on the acceleration data. When an impact is detected, the processor (130) can acquire angular velocity data through the gyroscope (112). Based on the acceleration data and angular velocity data, the processor (130) can acquire non-collision data. In addition, whether to operate always or conditionally can be determined based on the power consumption of each sensor (110).

[0093] In the above example, the processor (130) can obtain information about the time at which the impact was detected, the time at which the sensing data began to be received through the gyroscope (112) after the impact was detected (e.g., 20ms from the time the impact was detected), the time before the time the impact was detected among the data collected, the time after the time the impact was detected among the data collected, the sampling rate for each sensor (110), etc., and store it in memory (120).

[0094] According to the embodiments described above with reference to FIG. 2, the electronic device (100) can acquire sensing data in a manner that minimizes the power consumption of the electronic device (100), while also acquiring sensing data that can provide high accuracy regarding whether a vehicle has collided by using a plurality of sensors (110) immediately after an impact is detected.

[0095] FIG. 3 is a diagram illustrating the verification process of collected data and the learning process of a neural network model according to one embodiment of the present disclosure.

[0096] As mentioned above, a neural network model refers to a model trained to generate data corresponding to a non-collision state. A neural network model can be trained based on non-collision data rather than collision data.

[0097] Referring to FIG. 3, the neural network model may include a generator and a discriminator. For example, the neural network model may be a Generative Adversarial Network (GAN) model in which the generator and the discriminator compete and learn against each other, or it may be a model such as a Deep Convolutional Generative Adversarial Network (DCGAN) model or a Wasserstein GAN (WGAN) model, which have improved performance and stability compared to a GAN. However, the types of neural network models according to the present disclosure are not limited thereto.

[0098] The generator can generate generated data corresponding to actual data. Specifically, when a latent vector, which is a noise vector randomly sampled from a normal distribution or a uniform distribution, is input, the generator can transform the latent vector to learn complex patterns in the latent space and generate generated data corresponding to actual non-collision data. In other words, the generator can generate generated data that is difficult to distinguish from non-collision data.

[0099] The discriminator can obtain a probability regarding whether input data corresponds to non-collision data. Specifically, the discriminator can identify the probability regarding whether input data is real data, i.e., non-collision data, or fake data, i.e., generated data, created by the generator. For example, the discriminator can output a probability value closer to 1 as the likelihood of the input data being real data increases, and a probability value closer to 0 as the likelihood of the input data being fake data increases.

[0100] The processor (130) can obtain a loss based on the probability output by the discriminator. As described above, the loss can collectively refer to information related to the difference between the generated data and the collected data, and specifically, it can be calculated through the following process.

[0101] The processor (130) can obtain a first loss based on the difference between the generated data and the collected data. For example, the processor (130) can calculate the difference in absolute values ​​between each element of the generated data and the collected data, and calculate the first loss by summing the differences between all elements. As the difference between the generated data and the collected data becomes smaller, the first loss may be smaller.

[0102] The processor (130) can obtain a second loss based on the difference between the feature value obtained by inputting generated data into the discriminator and the feature value obtained by inputting collected data into the discriminator. For example, the processor (130) can obtain the output of the intermediate layer of the discriminator, the output of the feature extractor, or the final output of the discriminator as a feature value as a result of inputting the generated data and the collected data, respectively, into the discriminator. Then, the processor (130) can calculate the second loss by summing the differences between the feature values. The closer the difference between the features of the generated data and the features of the collected data becomes, the smaller the second loss may be.

[0103] The first and second losses may be defined as various losses, such as L1 loss, i.e., MAE (Mean Absolute Error) loss, or L2 loss, i.e., MSE (Mean Squared Error) loss.

[0104] The processor (130) can obtain a final loss based on the weighted sum of the first loss and the second loss. For example, the final loss can be obtained by multiplying each of the first loss and the second loss by a pre-set weight and then summing the results of the multiplication. The weight for each loss can be changed according to the settings of the user or developer.

[0105] In one embodiment, the processor (130) can identify whether the collected data indicates a non-collision state based on the acquired loss. For example, the processor (130) can acquire a loss value (or score) corresponding to the acquired loss. If the loss value is greater than or equal to a preset threshold, the processor (130) can identify that the collected data indicates a non-collision state. If the loss value is less than the preset threshold, the processor (130) can identify that the collected data indicates a collision state.

[0106] In one embodiment, if the collected data is identified as indicating a non-collision state, the processor (130) can update the non-collision data based on the collected data. For example, the processor (130) can add the collected data corresponding to the non-collision state to the training data set for training the neural network model by designating the collected data corresponding to the non-collision state as non-collision data. Then, the processor (130) can train the neural network model based on the updated non-collision data.

[0107] In one embodiment, the processor (130) can train a neural network model based on the acquired loss. For example, the generator can be trained to generate generated data that is as close as possible to non-collision data, and the discriminator can be trained to distinguish the generated data from the non-collision data as much as possible. Through such adversarial learning, the neural network model is able to generate generated data that represents a non-collision state.

[0108] FIG. 4 is a block diagram showing the configuration of an electronic device (100) according to one embodiment of the present disclosure. FIG. 5 is a diagram showing acceleration data corresponding to a collision state of a vehicle. FIG. 6 is a diagram showing audio data corresponding to a collision state of a vehicle. FIG. 7 is a diagram showing a user interface according to one embodiment of the present disclosure.

[0109] As illustrated in FIG. 4, the electronic device (100) may further include a sensor (110), memory (120), and processor (130), as well as a communication interface (140), an input interface (150), and an output interface (160). Additionally, the sensor (110) may further include an accelerometer (111), a gyroscope (112), a barometric pressure sensor (113), as well as an audio sensor (114) and a GNNS sensor (115) (Global Navigation Satellite System).

[0110] The audio sensor (114) can receive audio data (signals). Specifically, the audio sensor (114) can receive sound (sound waves) and convert an analog signal corresponding to the received sound into a digital signal. The processor (130) can obtain information about the state and environment surrounding the electronic device (100) by analyzing the frequency, amplitude, duration, etc. of the audio signal. For example, the processor (130) can identify whether the audio data corresponds to a non-collision state of the vehicle by identifying whether the audio data corresponds to a unique frequency spectrum, sound pressure level, etc. of the sound generated during a collision of the vehicle.

[0111] The GNNS sensor (115) can receive signals transmitted from a satellite to obtain location data regarding the location of the electronic device (100). The location data may include information regarding longitude, latitude, and altitude. For example, the processor (130) can identify whether the location data corresponds to a non-collision state of the vehicle by identifying, based on the location data, cases where a sudden change in direction of the electronic device (100) is detected, cases where the vehicle suddenly stops at a specific location and that state persists, etc.

[0112] For example, a GNSS (Global Navigation Satellite System) sensor may include at least one of GPS (Global Positioning System), GLONASS (Global Navigation System), Galileo, Beidou, and KPS (Korean Positioning System). However, there are no special limitations on the type of GNSS sensor (115) according to the present disclosure.

[0113] As described above, various types of sensors (110) have been described, and the processor (130) can acquire non-collision data by comprehensively utilizing multiple sensors (110) of different types. For example, the processor (130) can identify whether each of two or more of the acceleration data, angular velocity data, barometric pressure data, audio data, and position data corresponds to a non-collision state of the vehicle, and by combining the identification results, determine whether to designate the sensing data as non-collision data.

[0114] Synthesizing identification results may mean that sensing data can be designated as non-collision data only when the identification results for each data point all match a non-collision state. Additionally, synthesizing identification results may mean calculating a score for the identification result of each data point and determining whether to designate the sensing data as non-collision data based on the weighted sum of the scores for each data point.

[0115] In one embodiment, the processor (130) can obtain non-collision data based on acceleration data, angular velocity data, and atmospheric pressure data. The processor (130) can obtain non-collision data based on acceleration data, angular velocity data, and audio data. And, the processor (130) can obtain non-collision data based on acceleration data, angular velocity data, and position data.

[0116] For example, referring to FIG. 5, when a vehicle collides, acceleration data obtained through the accelerometer (111) of the electronic device (100) placed inside the vehicle may show a rapid change around 200ms, which is the time when the vehicle collides. Referring to FIG. 6, when a vehicle collides, audio data obtained through the audio sensor (114) of the electronic device (100) placed inside the vehicle may show a rapid change around 200ms, which is the time when the vehicle collides.

[0117] In this way, when both acceleration data and audio data correspond to the collision state of the vehicle, the processor (130) may designate the sensing data including the acceleration data and audio data, or each of the acceleration data and audio data, as non-collision data. A criterion for identifying whether there is a sudden change may be determined by a threshold value set for each sensor (110).

[0118] As described above in the embodiment, if non-collision data is obtained using multiple sensors (110) of different types, the accuracy and reliability of the data can be significantly improved. Meanwhile, if non-collision data is obtained using a first type sensor (110) among multiple sensors (110) of different types, and a neural network model is trained based on the non-collision data, the processor (130) may additionally verify the output of the neural network model using a second type sensor (110) among multiple sensors (110).

[0119] For example, the processor (130) can acquire non-collision data corresponding to the vehicle's non-collision state through the accelerometer (111) and gyroscope (112), and acquire generated data corresponding to the non-collision state through a neural network model learned based on the non-collision data. When collected data related to the vehicle's state is acquired, the processor (130) can acquire a loss related to the difference between the generated data and the collected data.

[0120] In this case, the processor (130) may identify whether the collected data indicates a non-collision state based on the acquired loss, but if audio data corresponding to the time of acquiring the non-collision data is received through the audio sensor (114), additional verification may be performed using whether the audio data indicates a non-collision state. This embodiment may be applied particularly when the neural network model is implemented to receive only a specific type of data.

[0121] The communication interface (140) includes a circuit and can perform communication with an external device. Specifically, the processor (130) can receive various data or information from an external device connected through the communication interface (140) and can also transmit various data or information to the external device.

[0122] The communication interface (140) may include at least one of a WiFi module, a Bluetooth module, a wireless communication module, an NFC module, and an Ultra-Wide Band (UWB) module. Specifically, the WiFi module and the Bluetooth module can each perform communication using the WiFi method and the Bluetooth method. When using the WiFi module or the Bluetooth module, various connection information such as SSID is first transmitted and received, and then various information can be transmitted and received after establishing a communication connection using this information.

[0123] In addition, the wireless communication module can perform communication according to various communication standards such as IEEE, Zigbee, 3G (3rd Generation), 3GPP (3rd Generation Partnership Project), LTE (Long Term Evolution), and 5G (5th Generation). Furthermore, the NFC module can perform communication using the NFC (Near Field Communication) method, which utilizes the 13.56 MHz band among various RF-ID frequency bands such as 135 kHz, 13.56 MHz, 433 MHz, 860~960 MHz, and 2.45 GHz. Additionally, the UWB module can accurately measure the Time of Arrival (ToA), which is the time it takes for a pulse to reach a target, and the Angle of Arrival (AoA), which is the angle of arrival of the pulse at the transmitting device, through communication between UWB antennas. Accordingly, precise distance and location recognition within an error range of tens of centimeters indoors is possible.

[0124] In one embodiment, the processor (130) can receive external data from an external device through a communication interface (140). Here, 'external data' may be data obtained through a sensor (110) included in the external device. The processor (130) can obtain non-collision data using the external data together with the sensing data. For example, when external data is received from an external device through the communication interface (140), the processor (130) can obtain non-collision data based on acceleration data, angular velocity data, and the external data.

[0125] That is, the electronic device (100) can not only utilize a plurality of sensors (110) included in the electronic device (100) comprehensively, but also utilize sensors (110) of an external device connected to the electronic device (100) comprehensively. This embodiment may be applied, for example, when the electronic device (100) is a user's smartphone located inside a vehicle and the external device is the user's smartwatch.

[0126] The input interface (150) includes a circuit, and the processor (130) can receive user commands to control the operation of the electronic device (100) through the input interface (150). Specifically, the input interface (150) may be composed of components such as a microphone, a camera, and a remote control signal receiver. Additionally, the input interface (150) may be implemented in a form included in a display as a touch screen. In particular, the microphone can receive a voice signal and convert the received voice signal into an electrical signal.

[0127] The output interface (160) includes a circuit, and the processor (130) can output various functions that the electronic device (100) can perform through the output interface (160). Also, the output interface (160) may include at least one of a display, a speaker, and an indicator.

[0128] The display can output image data under the control of the processor (130). Specifically, the display can output an image stored in the memory (120) under the control of the processor (130). In particular, the display according to one embodiment of the present disclosure may display a user interface stored in the memory (120). The display may be implemented as an LCD (Liquid Crystal Display Panel), OLED (Organic Light Emitting Diodes), etc., and the display may also be implemented as a flexible display, a transparent display, etc. depending on the case. However, the display according to the present disclosure is not limited to a specific type.

[0129] The speaker can output audio data under the control of the processor (130). The indicator can be lit under the control of the processor (130). Specifically, the indicator can be lit in various colors under the control of the processor (130). For example, the indicator can be implemented as an LED (Light Emitting Diodes), LCD (Liquid Crystal Display Panel), VFD (Vacuum Fluorescent Display), etc., but is not limited thereto.

[0130] In one embodiment, the processor (130) may provide a user interface for receiving user feedback regarding whether the collected data indicates a non-collision state, and may receive user feedback through the user interface. And, when user feedback is received, the processor (130) may identify whether the collected data indicates a non-collision state based on the loss and user feedback.

[0131] Referring to FIG. 7, the user interface may be displayed on at least a portion of the display of the electronic device (100) or the display of an external device. The user interface may include text for providing information, such as "Terminal impact occurred (710)" and "Has a collision occurred with the terminal (720)?". The user interface may include selection items, such as "Vehicle collision (730)" and "Phone dropped (740)". Additionally, the user interface may include a text input item (750) and a transmission item (760) for entering text about a situation that occurs other than "Vehicle collision (730)" or "Phone dropped (740)".

[0132] If a user interface as illustrated in FIG. 7 is provided, the processor (130) can receive user input selecting "vehicle collision (730)" or "phone drop (740)" through the user interface. The processor (130) can receive text about the situation through text input items (750) and transmission items (760). If text about the situation is received, since the user has not selected the selection item "vehicle collision (730)," the processor (130) can identify that the collected data indicates a non-collision state.

[0133] When text about a situation is received through a text input item (750) and a transmission item (760), the processor (130) inputs the received text into a natural language processing model to obtain information about the situation at the time the collected data was acquired, and may use the information about the acquired situation to identify whether the collected data indicates a non-collision state.

[0134] Meanwhile, the processor (130) may control the display to provide a user interface for obtaining user consent regarding the process of acquiring non-collision data using at least one sensor (110), the process of transmitting non-collision data to an external server, etc.

[0135] FIG. 8 is a block diagram for explaining the specific configuration of an electronic device according to one embodiment of the present disclosure.

[0136] Specifically, FIG. 8 is a block diagram of an exemplary electronic device (800) capable of performing the operations described in the description of the present disclosure. FIG. 8 is merely intended to explain the configuration of the electronic device according to the present disclosure in more detail, and even if the electronic device includes the configuration as shown in FIG. 8, various embodiments described above with reference to FIG. 1 can likewise be implemented.

[0137] Referring to FIG. 8, the electronic device (800) may be one of various forms of electronic devices, such as a notebook (890), smartphones (891) having various form factors (e.g., a bar-type smartphone (891-1), a foldable-type smartphone (891-2), or a sliderable (or rollable)-type smartphone (891-3)), a tablet (892), a cellular phone (not shown), and other similar computing devices (not shown). The components, their relationships, and their functions illustrated in FIG. 8 are illustrative only and are not intended to limit the implementations described or claimed herein. The electronic device (800) may be referred to as a mobile device, a user device, a multifunction device, a portable device, or a server.

[0138] The electronic device (800) may include components comprising at least one processor (810) (hereinafter referred to as processor (810)), at least one memory (820) (hereinafter referred to as memory (820)), at least one display (840) (hereinafter referred to as display (840)), at least one image sensor (850) (hereinafter referred to as image sensor (850)), at least one communication circuit (860) (hereinafter referred to as communication circuit (860)), and / or at least one sensor (870) (hereinafter referred to as sensor (870)). The components are merely exemplary. For example, the electronic device (800) may include other components (e.g., power management integrated circuitry (PMIC), audio processing circuit, antenna, rechargeable battery, or input / output interface). For example, some components may be omitted from the electronic device (800). For example, some components may be integrated into a single component.

[0139] The processor (810) may be implemented as one or more IC (integrated circuit (or circuitry)) chips and may perform various data processing operations. The processor (810) may include at least one electrical circuit and may process instructions (or programs, data, etc.) stored in memory (820) individually or collectively in a distributed manner. The processor (810) may include a processor assembly comprising one or more processing circuits. The processor (810) may include any processing circuit that is operative to control the performance and operations of one or more components of the electronic device (800) (e.g., memory (820), display (840), image sensor (850), communication circuit (860), and / or sensor (870)). For example, the processor (810) (e.g., application processor (AP)) may be implemented as a system on chip (SoC) (e.g., a single chip or chipset). For example, the processor (810) may be implemented with a plurality of cores (or at least one core circuit), a plurality of chips, or a plurality of chipsets. For example, the processor (810) may include one or more processing circuits. For example, the processor (810) may include one or more processing circuits configured to perform the various functions of the present disclosure individually and / or collectively. As an example without limitation, at least a portion of the processor (810) may be included in a first chip of the electronic device (800), and at least another portion of the processor (810) may be included in a second chip of the electronic device (800) different from the first chip of the electronic device (800).

[0140] For example, the processor (810) may include a central processing unit (CPU) (811), a graphics processing unit (GPU) (812), a neural processing unit (NPU) (813), an image signal processor (ISP) (814), a display controller (815), a memory controller (816), a storage controller (817), a communication processor (CP) (818), and / or a sensor interface (819). These components of the processor (810) are merely exemplary. For example, the processor (810) may include other components. For example, some components of the processor (810) may be omitted from the processor (810). For example, some components of the processor (810) may be included as separate components of the electronic device (800) outside the processor (810). For example, some components of the processor (810) (e.g., memory controller (816)) may be included in other components (e.g., at least part of memory (820), an interface (e.g. available for connection to at least one component of the electronic device (100)), a display (840) and / or an image sensor (850)).

[0141] The processor (810) may cause other components of the electronic device (800) to perform various operations by executing instructions stored in memory (820). The CPU (811) (or central processing circuit) may be configured to control the components of the processor (810) based on the execution of instructions stored in memory (820) (e.g., volatile memory (821) and / or non-volatile memory (822)). The GPU (812) (or graphics processing circuit) may be configured to execute parallel operations (e.g., rendering). The NPU (813) (or neural processing circuit, or AI (artificial intelligence) chip) may be configured to execute operations for an artificial intelligence model (e.g., convolution computation). An ISP (814) (or image signal processing circuit) may be configured to process a raw image acquired through an image sensor (850) into a format suitable for a component within an electronic device (800) or a component of a processor (810). A display controller (815) (or display control circuit, or DPU (display processing unit)) may be configured to process an image acquired from a CPU (811), GPU (812), ISP (814), or memory (820) (e.g., volatile memory (821)) into a format suitable for a display (840). A memory controller (816) (or memory control circuit) may be configured to control reading data from volatile memory (821) and writing data to volatile memory (821). A storage controller (817) (or storage control circuit) may be configured to control reading data from non-volatile memory (822) and writing data to non-volatile memory (822).The CP (818) (communication processing circuit) may be configured to process data obtained from a component of the processor (810) into a format suitable for transmitting to another electronic device via the communication circuit (860), or to process data obtained from another electronic device via the communication circuit (860) into a format suitable for processing by the component of the processor (810). For example, the communication circuit (860) may include one or more communication circuits. The sensor interface (819) (or sensing data processing circuit, sensor hub) may be configured to process data regarding the state of the electronic device (800) and / or the state around the electronic device (800), obtained through the sensor (870), into a format suitable for the component of the processor (810).

[0142] Memory (820) may include one or more storage media (or one or more storage devices). For example, memory (820) may include a memory assembly comprising one or more storage media. For example, the one or more storage media may include a hard drive, a permanent memory such as flash memory, read-only memory (ROM) (e.g., non-volatile memory (822)), a semi-permanent memory such as random access memory (RAM) (e.g., volatile memory (821)), any other suitable type of storage (or storage assembly), or any combination thereof. Memory (820) may include a cache memory, which is one or more different types of memory used to temporarily store data for a function or feature of the electronic device (800). As an example not limited to, the cache memory may be included within the processor (810). The memory (820) may be fixedly embedded within the electronic device (800) or incorporated into one or more suitable types of components (e.g., a SIM (subscriber identity module) card and / or an SD (secure digital) card) that can be repeatedly inserted into and removed from the electronic device (800).

[0143] For example, memory (820) may store one or more software applications, such as operating system (or system) software applications, firmware software applications, driver software applications, plugin (e.g., add-in, add-on, and / or applet) software applications, and / or any other suitable software applications. For example, the one or more software applications may include instructions executable by the processor (810). For example, memory (820) may store instructions that can be called by an application programming interface (API). For example, memory (820) may store instructions within a library.

[0144] FIG. 9 is a flowchart illustrating a control method of an electronic device (100) according to one embodiment of the present disclosure.

[0145] The electronic device (100) can acquire non-collision data corresponding to a non-collision state of the vehicle through at least one sensor (110) (S910). If the sensing data acquired through at least one sensor (110) corresponds to a non-collision state of the vehicle, the electronic device (100) can identify the sensing data as non-collision data.

[0146] The electronic device (100) can obtain generated data corresponding to a non-collision state through a neural network model trained based on non-collision data (S920). A neural network model refers to a model trained to generate data corresponding to a non-collision state. The neural network model can be trained based on non-collision data rather than collision data.

[0147] When collected data related to the state of the vehicle is obtained, the electronic device (100) can obtain a loss related to the difference between the generated data and the collected data (S930). The collected data may be obtained through a sensor (110) included in the electronic device (100), or it may be obtained through a sensor (110) included in an external device (e.g., a server, multiple user terminals, etc.).

[0148] The electronic device (100) can obtain a first loss based on the difference between generated data and collected data. The electronic device (100) can obtain a second loss based on the difference between the feature value obtained by inputting generated data into a discriminator and the feature value obtained by inputting collected data into a discriminator. The electronic device (100) can obtain a final loss based on the weighted sum of the first loss and the second loss.

[0149] The electronic device (100) can identify whether the collected data indicates a non-collision state based on the acquired loss (S940). Additionally, if the collected data is identified as indicating a non-collision state, the electronic device (100) can update the non-collision data based on the collected data. Furthermore, the electronic device (100) can train a neural network model based on the acquired loss.

[0150] Meanwhile, the control method of the electronic device (100) according to the above-described embodiment may be implemented as a program and provided to the electronic device (100). In particular, the program including the control method of the electronic device (100) may be stored and provided on a non-transitory computer-readable medium.

[0151] Specifically, in a non-transient computer-readable recording medium comprising a program for executing a control method of an electronic device (100), the control method of the electronic device (100) may include: a step of acquiring non-collision data corresponding to a non-collision state of a vehicle through at least one sensor (110); a step of acquiring generated data corresponding to a non-collision state through a neural network model learned based on the non-collision data; a step of acquiring a loss related to the difference between the generated data and the collected data when collected data related to the state of the vehicle is acquired; and a step of identifying whether the collected data indicates a non-collision state based on the loss.

[0152] Although a method for controlling an electronic device (100) and a computer-readable recording medium including a program for executing the method for controlling the electronic device (100) have been briefly described above, this is merely to avoid redundant descriptions, and it is obvious that various embodiments of the electronic device (100) can also be applied to a method for controlling the electronic device (100) and a computer-readable recording medium including a program for executing the method for controlling the electronic device (100).

[0153] The artificial intelligence-related function according to the present disclosure is operated through the processor (130) and memory (120) of the electronic device (100).

[0154] The processor (130) may be composed of one or more processors (130). In this case, the one or more processors (130) may include at least one of a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), and an NPU (Neural Processing Unit), but are not limited to the examples of the processor (130) described above.

[0155] The CPU is a general-purpose processor (130) capable of performing not only general operations but also artificial intelligence operations, and can efficiently execute complex programs through a multi-layer cache structure. The CPU is advantageous for a serial processing method that enables organic linkage between previous and next calculation results through sequential calculations. The general-purpose processor (130) is not limited to the examples described above, except for cases where it is specified as the CPU described above.

[0156] A GPU is a processor (130) for large-scale computations, such as floating-point operations used in graphics processing, and can perform large-scale computations in parallel by integrating a large number of cores. In particular, a GPU may be advantageous for parallel processing methods such as convolution operations compared to a CPU. Additionally, a GPU can be used as a co-processor (130) to complement the functions of a CPU. The processor (130) for large-scale computation is not limited to the examples described above, except for cases where it is specified as the aforementioned GPU.

[0157] The NPU is a processor (130) specialized for artificial intelligence computation using an artificial neural network, and each layer constituting the artificial neural network can be implemented in hardware (e.g., silicon). At this time, since the NPU is designed to be specialized according to the specifications required by the manufacturer, it has a lower degree of freedom compared to a CPU or GPU, but it can efficiently process the artificial intelligence computation required by the manufacturer. Meanwhile, as a processor (130) specialized for artificial intelligence computation, the NPU can be implemented in various forms such as a TPU (Tensor Processing Unit), an IPU (Intelligence Processing Unit), a VPU (Vision Processing Unit), etc. The artificial intelligence processor (130) is not limited to the examples described above, except for cases specified as the aforementioned NPU.

[0158] Additionally, one or more processors (130) may be implemented as a System on Chip (SoC). In this case, the SoC may further include, in addition to one or more processors (130), a memory (120) and a network interface such as a bus for data communication between the processor (130) and the memory (120).

[0159] When a plurality of processors (130) are included in a System on Chip (SoC) included in an electronic device (100), the electronic device (100) can perform operations related to artificial intelligence (e.g., operations related to learning or inference of an artificial intelligence model) by using some of the processors (130) among the plurality of processors (130). For example, the electronic device (100) can perform operations related to artificial intelligence by using at least one of a GPU, NPU, VPU, TPU, or hardware accelerator specialized for artificial intelligence operations such as convolution operations or matrix multiplication operations among the plurality of processors (130). However, this is merely one embodiment, and it is obvious that operations related to artificial intelligence can be processed using a general-purpose processor (130) such as a CPU.

[0160] Additionally, the electronic device (100) can perform operations related to artificial intelligence functions using multi-cores (e.g., dual cores, quad cores, etc.) included in a single processor (130). In particular, the electronic device (100) can perform artificial intelligence operations such as convolution operations and matrix multiplication operations in parallel using multi-cores included in the processor (130).

[0161] One or more processors (130) control input data to be processed according to predefined operation rules or artificial intelligence models stored in memory (120). The predefined operation rules or artificial intelligence models are characterized by being created through learning.

[0162] Here, being created through learning means that a predefined rule of operation or an artificial intelligence model of desired characteristics is created by applying a learning algorithm to a number of learning data. Such learning may be performed on the device itself where the artificial intelligence according to the present disclosure is executed, or it may be performed through a separate server / system.

[0163] An artificial intelligence model may be composed of multiple neural network layers. At least one layer has at least one weight value and performs the layer's operation through the result of the operation of the previous layer and at least one defined operation. Examples of neural networks include CNN (Convolutional Neural Network), DNN (Deep Neural Network), RNN (Recurrent Neural Network), RBM (Restricted Boltzmann Machine), DBN (Deep Belief Network), BRDNN (Bidirectional Recurrent Deep Neural Network), Deep Q-Networks, and Transformers; however, the neural networks in this disclosure are not limited to the aforementioned examples except where specified.

[0164] A learning algorithm is a method of training a specific target device (e.g., a robot) using a number of learning data to enable the target device to make decisions or predictions on its own. Examples of learning algorithms include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, and the learning algorithms in this disclosure are not limited to the aforementioned examples except where specified.

[0165] A device-readable storage medium may be provided in the form of a non-transitory storage medium. Here, 'non-transitory storage medium' simply means that it is a tangible device and does not contain a signal (e.g., electromagnetic waves), and the term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily. For example, a 'non-transitory storage medium' may include a buffer in which data is stored temporarily.

[0166] According to one embodiment, the method according to the various embodiments disclosed herein may be provided by being included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)) or an application store (e.g., Play Store). TM It can be distributed online (e.g., downloaded or uploaded) through ) or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product (e.g., downloadable app) may be temporarily stored or temporarily created in a device-readable storage medium such as the memory (120) of the manufacturer's server, the application store's server, or the relay server.

[0167] Each component (e.g., module or program) according to the various embodiments of the present disclosure as described above may be composed of a single or multiple entities, and some of the aforementioned sub-components may be omitted, or other sub-components may be further included in the various embodiments. Generally or additionally, some components (e.g., module or program) may be integrated into a single entity to perform the same or similar functions as those performed by each of the respective components prior to integration.

[0168] Operations performed by a module, program, or other component according to various embodiments may be executed sequentially, in parallel, iteratively, or heuristically, or at least some operations may be executed in a different order, omitted, or other operations may be added.

[0169] Meanwhile, the terms “part” or “module” as used in this disclosure include a unit composed of hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit, for example. A “part” or “module” may be a component formed integrally, or a minimum unit or part thereof that performs one or more functions. For example, a module may be composed of an application-specific integrated circuit (ASIC).

[0170] Various embodiments of the present disclosure may be implemented as software comprising instructions stored on a machine-readable storage medium (e.g., a computer). The machine may include an electronic device (e.g., an electronic device (100)) according to the disclosed embodiments, which is a device capable of calling instructions stored from the storage medium and operating according to the called instructions.

[0171] When the above command is executed by the processor (130), the processor (130) may perform a function corresponding to the command using other components, either directly or under the control of the processor (130). The command may include code generated or executed by a compiler or an interpreter.

[0172] Although preferred embodiments of the present disclosure have been illustrated and described above, the present disclosure is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the present disclosure as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present disclosure.

Claims

1. In an electronic device (100), At least one sensor (110); At least one memory (120) for storing instructions; and At least one processor (130) including a processing circuit; comprising, When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), Through the above at least one sensor (110), non-collision data corresponding to the vehicle's non-collision state is obtained, and Through a neural network model trained based on the above non-collision data, generated data corresponding to the above non-collision state is obtained, and When collected data related to the state of the above vehicle is obtained, a loss related to the difference between the above generated data and the above collected data is obtained, and An electronic device (100) that identifies whether the collected data indicates the non-collision state based on the above loss.

2. In Paragraph 1, The above at least one sensor (110) includes an accelerometer (111) and a gyroscope (112), and When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), Acceleration data is obtained through the above accelerometer (111), and When an impact applied to the electronic device (100) is detected based on the above acceleration data, angular velocity data is obtained through the gyroscope (112), and An electronic device (100) for obtaining non-collision data based on the above acceleration data and the above angular velocity data.

3. In Paragraph 1, When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), An electronic device (100) that, upon receiving user feedback regarding whether the collected data indicates the non-collision state, identifies whether the collected data indicates the non-collision state based on the loss and the user feedback.

4. In Paragraph 2, The above at least one sensor (110) further includes a pressure sensor (113), and When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), An electronic device (100) that, when pressure data is obtained through the pressure sensor (113), obtains the non-collision data based on the acceleration data, the angular velocity data, and the pressure data.

5. In Paragraph 2, The above at least one sensor (110) further includes an audio sensor (114), and When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), An electronic device (100) that, when audio data is acquired through the audio sensor (114), acquires the non-collision data based on the acceleration data, the angular velocity data, and the audio data.

6. In Paragraph 2, The above at least one sensor (110) further includes a GNNS sensor (115) (Global Navigation Satellite System), and When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), An electronic device (100) that, when position data regarding the position of the electronic device (100) is obtained through the GNNS sensor (115), obtains non-collision data based on the acceleration data, the angular velocity data, and the position data.

7. In Paragraph 2, It further includes a communication interface (140), When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), An electronic device (100) that, when external data is received from an external device through the communication interface (140), obtains the non-collision data based on the acceleration data, the angular velocity data, and the external data.

8. In Paragraph 1, When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), The above neural network model includes a generator that generates the generated data and a discriminator that obtains a probability regarding whether the input data corresponds to the non-collision data. When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), A first loss is obtained based on the difference between the generated data and the collected data, and A second loss is obtained based on the difference between the feature value obtained by inputting the generated data into the discriminator and the feature value obtained by inputting the collected data into the discriminator. An electronic device (100) that obtains the loss based on the weighted sum of the first loss and the second loss.

9. In Paragraph 1, When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), If the above collected data is identified as indicating the above non-collision state, the above non-collision data is updated based on the above collected data, and An electronic device (100) that trains the neural network model based on the above-mentioned updated non-collision data.

10. In Paragraph 1, When the above instructions are executed individually or collectively by the at least one processor (130), the electronic device (100), An electronic device (100) that trains the neural network model based on the above loss.

11. A method for controlling an electronic device (100), A step of acquiring non-collision data corresponding to the non-collision state of the vehicle through at least one sensor (110); A step of obtaining generated data corresponding to the non-collision state through a neural network model trained based on the above non-collision data; When collected data related to the state of the vehicle is obtained, a step of obtaining a loss related to the difference between the generated data and the collected data; and A control method for an electronic device (100) comprising the step of identifying whether the collected data indicates the non-collision state based on the above loss.

12. In Paragraph 11, The above at least one sensor (110) includes an accelerometer (111) and a gyroscope (112), and The control method of the above electronic device (100) is, A step of acquiring acceleration data through the above accelerometer (111); When an impact applied to the electronic device (100) is detected based on the above acceleration data, a step of acquiring angular velocity data through the gyroscope (112); and A control method for an electronic device (100) further comprising the step of obtaining non-collision data based on the acceleration data and the angular velocity data.

13. In Paragraph 11, The step of identifying whether the above non-collision state is indicated is, A control method for an electronic device (100), comprising the step of identifying whether the collected data indicates the non-collision state based on the loss and the user feedback when user feedback is received regarding whether the collected data indicates the non-collision state.

14. In Paragraph 12, The step of acquiring the above non-collision data is, A control method for an electronic device (100) that, when external data is received from an external device, obtains the non-collision data based on the acceleration data, the angular velocity data, and the external data.

15. A non-transient computer-readable recording medium comprising a program for executing a method of controlling an electronic device (100), The control method of the above electronic device (100) is, A step of acquiring non-collision data corresponding to the non-collision state of the vehicle through at least one sensor (110); A step of obtaining generated data corresponding to the non-collision state through a neural network model trained based on the above non-collision data; When collected data related to the state of the vehicle is obtained, a step of obtaining a loss related to the difference between the generated data and the collected data; and A computer-readable recording medium comprising: a step of identifying whether the collected data indicates the non-collision state based on the above loss.

Images