Method and apparatus for diagnosing failure of freight car through image analysis based on reinforcement learning

Abstract

Embodiments relate to a method for diagnosing a failure of a freight car through image analysis based on reinforcement learning and may comprise the steps of: acquiring road image information and wheel image information through at least one sensor installed in a freight car; acquiring freight car-related information and environmental information; generating vehicle-related information on the basis of the road image information, the wheel image information, the freight car-related information, and the environmental information; and providing the generated vehicle-related information to a user.

Description

Method and device for diagnosing truck failures using reinforcement learning-based image analysis

[0001] The present invention relates to a method and device for diagnosing truck faults using reinforcement learning-based image analysis. More specifically, the present invention relates to a method and device for acquiring image information related to a truck and applying the acquired image information to reinforcement learning to diagnose truck faults.

[0002] Existing truck fault diagnosis has been performed through error code analysis based on diagnostic software. Specifically, when a fault occurs in the engine or other components of a truck, error codes are generated, and the presence of a truck fault can be confirmed through a system that analyzes and diagnoses these codes. The system for detecting truck faults can be an On-Board Diagnostics (OBD) vehicle diagnostic system, which monitors the engine, transmission, exhaust, and other components to identify and diagnose faults. However, vehicle diagnostic systems only diagnose abnormalities within each vehicle component, limiting their ability to comprehensively analyze the cause of the fault and thereby improve diagnostic accuracy.

[0003] Considering the above, the following describes a method and device for diagnosing truck failures using reinforcement learning-based image analysis.

[0004]

[0005] [Prior Art Literature]

[0006] [Patent Document]

[0007] Korean Patent Publication No. 10-2023-0128603

[0008] This specification relates to a method and device for diagnosing truck failures using reinforcement learning-based image analysis.

[0009] This specification relates to a method and device for diagnosing the replacement cycle of consumables in a truck and the magnitude of a vehicle impact through image analysis based on reinforcement learning.

[0010] This specification relates to a method and device for acquiring image information through multiple sensors attached to the bottom of a truck and analyzing the image information using reinforcement learning.

[0011] The present specification relates to a method and device for acquiring image information and controlling a truck to reduce causes of truck failure based on the acquired image information.

[0012] This specification relates to a method and device for acquiring image information and other information using multiple sensors and analyzing the information based on reinforcement learning.

[0013] According to one embodiment of the present specification, a method for diagnosing a truck failure through reinforcement learning-based image analysis may include a step of acquiring road image information and wheel image information through at least one sensor installed on a truck, a step of acquiring truck-related information and environment information, a step of generating vehicle-related information based on the road image information, wheel image information, truck-related information, and environment information, and a step of providing the generated vehicle-related information to a user.

[0014] In addition, according to one embodiment of the present specification, among at least one sensor installed on a truck, a first sensor is attached to a front end of a front wheel of the truck to obtain road image information, and the first sensor includes at least one of a photographing unit for obtaining road image information, a vertical angle control unit for controlling a vertical angle of photographing, and a horizontal angle control unit for controlling a horizontal angle of photographing, and can obtain road image information by taking into consideration truck-related information and environmental information.

[0015] Additionally, according to one embodiment of the present specification, the first sensor may further include at least one of a sensor attached to the rear end of the rear wheels of the truck to obtain a road image at the rear of the truck and a sensor attached to the side of the truck to obtain a road image at the side of the truck.

[0016] Additionally, according to one embodiment of the present specification, at least one sensor includes a second sensor for obtaining a wheel image, wherein the second sensor is attached between the front and rear wheels of the truck to obtain a wheel image, and the wheel image includes a still image based on a stationary state of the truck and a moving state image based on a driving state of the truck, wherein wheel wear information, wheel compression information, wheel shrinkage information, and other information can be derived based on the wheel state information.

[0017] Additionally, according to one embodiment of the present specification, the truck-related information may include truck sensing-based engine information, engine parts information, electrical system information, body information, external accessory information, transmission information, drive train information, braking system information, consumables information, and other information, and the environmental information may include at least one of road condition information, weather condition information, traffic condition information, and other information.

[0018] In addition, according to one embodiment of the present specification, vehicle-related information is derived based on a reinforcement learning-based analysis model that inputs road image information, wheel image information, truck-related information, and environment information, wherein the vehicle-related information includes at least one of obstacle avoidance information, vehicle impact amount information, and vehicle damage accumulation information, and the reinforcement learning-based analysis model can be learned based on pre-acquired information.

[0019] In addition, according to one embodiment of the present specification, when information on whether to avoid obstacles is generated from vehicle-related information, the reinforcement learning-based analysis model can generate information on whether to avoid a frontal obstacle of a truck and, if the truck passes through the frontal obstacle without avoiding it, information on wheels passing through the frontal obstacle.

[0020] In addition, according to one embodiment of the present specification, the reinforcement learning-based analysis model further generates vehicle impact information and vehicle damage accumulation information based on whether an obstacle is avoided, and the vehicle impact information and vehicle damage accumulation information are generated by reflecting the type of obstacle, the size of the obstacle, the speed of the truck, the degree of wheel contraction, and other information in the reinforcement learning-based analysis model, and the generated vehicle impact information and vehicle damage accumulation information can be reflected in consumable replacement cycle information.

[0021] Additionally, according to one embodiment of the present specification, the truck further includes a truck failure model together with a reinforcement learning-based analysis model, and vehicle impact information and vehicle damage accumulation information can be derived based on the truck failure model.

[0022] In addition, according to one embodiment of the present specification, a first sensor among at least one sensor installed on a truck is attached to the front end of the truck front wheel, the rear end of the truck rear wheel, and the side of the truck to obtain a road image, wherein the first sensor includes at least one of a photographing unit for obtaining road image information, a vertical angle control unit for controlling a vertical angle of photographing, and a horizontal angle control unit for controlling a horizontal angle of photographing, and obtains road image information by considering truck-related information and environmental information, and a second sensor among at least one sensor installed on a truck is attached between the front and rear wheels of the truck to obtain a wheel image, wherein the wheel image includes a still image based on a truck stationary state and a moving state image based on a truck driving state, wherein wheel wear information, wheel compression information, wheel shrinkage information, and other information are derived based on the wheel state information, and the truck-related information includes truck sensing-based engine information, engine parts information, electrical system information, body information, external accessory information, transmission information, drive train information, braking system information, consumable information, and other information, and the environmental information includes at least one of road condition information, weather condition information, traffic condition information, and other information. Including any one, and vehicle-related information is derived based on a reinforcement learning-based analysis model that inputs road image information, wheel image information, truck-related information, and environment information, wherein the vehicle-related information includes at least one of obstacle avoidance information, vehicle impact amount information, and vehicle damage accumulation information, and the reinforcement learning-based analysis model is learned based on pre-acquired information, and when obstacle avoidance information is generated as vehicle-related information, the reinforcement learning-based analysis model can generate information on whether a truck avoids a frontal obstacle and, if the truck passes through the frontal obstacle without avoiding the frontal obstacle, information on wheels passing through the frontal obstacle.

[0023]

[0024] This specification can provide a method for diagnosing truck failures through image analysis based on reinforcement learning.

[0025] This specification can provide a method for diagnosing the replacement cycle of truck consumables and the magnitude of vehicle impact through reinforcement learning-based image analysis.

[0026] The present specification can provide a method for acquiring image information through multiple sensors attached to the bottom of a truck and analyzing the image information using reinforcement learning.

[0027] The present specification can provide a method for acquiring image information and controlling a truck to reduce causes of truck breakdown based on the acquired image information.

[0028] The present specification can provide a method for acquiring image information and other information using multiple sensors and analyzing the information based on reinforcement learning.

[0029] It should be understood that the effects of this specification are not limited to the matters described above, but can be expanded to various contents that can be derived from the detailed description of the embodiments of the invention below.

[0030]

[0031] FIG. 1 is a diagram illustrating an example of an operating environment of a system according to one embodiment of the present specification.

[0032] FIG. 2 is a block diagram for explaining the internal configuration of a computing device (200) in one embodiment of the present specification.

[0033] FIG. 3 is a drawing showing a truck with multiple sensors attached according to one embodiment of the present specification.

[0034] FIG. 4 is a diagram illustrating a method for acquiring data through multiple sensors in one embodiment of the present specification.

[0035] FIG. 5 is a diagram illustrating a method for performing truck failure diagnosis based on information sensed from outside the truck in one embodiment of the present specification.

[0036] FIG. 6 is a diagram illustrating a method for acquiring image information through multiple sensors in one embodiment of the present specification.

[0037] FIG. 7 is a diagram illustrating image analysis-based reinforcement learning in one embodiment of the present specification.

[0038] FIG. 8a is a diagram illustrating a method for controlling a vehicle through reinforcement learning-based image analysis in one embodiment of the present disclosure.

[0039] FIG. 8b is a diagram illustrating a method for controlling a vehicle through reinforcement learning-based image analysis in one embodiment of the present disclosure.

[0040] FIG. 9 is a diagram illustrating a method for performing training on a truck failure learning model based on at least one piece of information sensed through a truck in one embodiment of the present disclosure, and performing inference based on input information.

[0041] FIG. 10 is a diagram illustrating a method for diagnosing a truck failure based on vehicle sensing information in one embodiment of the present disclosure.

[0042] FIG. 11 is a flowchart illustrating a method for performing truck fault diagnosis through reinforcement learning-based image analysis in one embodiment of the present specification.

[0043]

[0044] In describing the embodiments of this specification, if a detailed description of a known configuration or function is judged to obscure the gist of the embodiments of this specification, a detailed description thereof will be omitted. In addition, parts of the drawings that are not related to the description of the embodiments of this specification have been omitted, and similar parts have been designated with similar drawing reference numerals.

[0045] In the embodiments of this specification, when a component is said to be "connected," "coupled," or "connected" to another component, this may include not only a direct connection, but also an indirect connection in which another component exists in between. Furthermore, when a component is said to "include" or "have" another component, unless otherwise specifically stated, this does not exclude the other component, but rather implies that the other component may be included.

[0046] In the embodiments of this specification, the terms first, second, etc. are used only for the purpose of distinguishing one component from another component, and do not limit the order or importance between components unless specifically stated otherwise. Therefore, within the scope of the embodiments of this specification, a first component in an embodiment may be referred to as a second component in another embodiment, and similarly, a second component in an embodiment may be referred to as a first component in another embodiment.

[0047] In the embodiments of this specification, distinct components are used to clearly illustrate their respective characteristics and do not necessarily imply separation. That is, multiple components may be integrated into a single hardware or software unit, or a single component may be distributed into multiple hardware or software units. Therefore, even if not specifically mentioned, such integrated or distributed embodiments are also included within the scope of the embodiments of this specification.

[0048] In this specification, the term "network" may encompass both wired and wireless networks. In this case, the term "network" may refer to a communications network that enables data exchange between devices, systems, and devices, and is not limited to a specific network.

[0049] Embodiments described herein may be entirely hardware, partially hardware and partially software, or entirely software. As used herein, "unit," "device," or "system" refers to a computer-related entity such as hardware, a combination of hardware and software, or software. For example, a unit, module, device, or system as used herein may be, but is not limited to, a running process, a processor, an object, an executable, a thread of execution, a program, and / or a computer. For example, both an application running on a computer and the computer itself may correspond to a unit, module, device, or system as used herein.

[0050] Additionally, in this specification, a device may be a mobile device such as a smartphone, tablet PC, wearable device, or HMD (Head Mounted Display), as well as a fixed device such as a PC or home appliance with display functions. Furthermore, as an example, a device may be an in-vehicle cluster or an IoT (Internet of Things) device. In other words, in this specification, a device may refer to any device capable of operating an application, and is not limited to a specific type. For convenience of explanation, the device on which an application operates is referred to as a device below.

[0051] In this specification, the network communication method is not limited, and connections between components may not be made using the same network method. The network may include not only communication methods utilizing communication networks (e.g., mobile communication networks, wired Internet, wireless Internet, broadcasting networks, satellite networks, etc.), but also short-range wireless communication between devices. For example, the network may include all communication methods that enable objects to network with each other, and is not limited to wired communication, wireless communication, 3G, 4G, 5G, or other methods. For example, wired and / or networks include Local Area Network (LAN), Metropolitan Area Network (MAN), Global System for Mobile Network (GSM), Enhanced Data GSM Environment (EDGE), High Speed ​​Downlink Packet Access (HSDPA), Wideband Code Division Multiple Access (W-CDMA), Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Bluetooth, Zigbee, Wi-Fi, VoIP (Voice over Internet Protocol), LTE Advanced, IEEE802.16m, WirelessMAN-Advanced, HSPA+, 3GPP Long Term Evolution (LTE), Mobile WiMAX (IEEE 802.16e), UMB (formerly EV-DO Rev. C), Flash-OFDM, iBurst and MBWA (IEEE 802.20) It may refer to a communication network using one or more communication methods selected from the group consisting of systems, HIPERMAN, Beam-Division Multiple Access (BDMA), Wi-MAX (World Interoperability for Microwave Access), and ultrasonic communication, but is not limited thereto.

[0052] The components described in various embodiments are not necessarily essential components, and some may be optional. Therefore, embodiments comprising a subset of the components described in the embodiments are also included within the scope of the embodiments of the present disclosure. Furthermore, embodiments including other components in addition to the components described in various embodiments are also included within the scope of the embodiments of the present disclosure.

[0053] Hereinafter, embodiments of the present specification will be described in detail with reference to the drawings.

[0054] FIG. 1 is a diagram illustrating an example of the operating environment of a system according to one embodiment of the present specification. Referring to FIG. 1, one or more user devices (110-1, 110-2) and one or more servers (120, 130, 140) are connected via a network (1). FIG. 1 is merely an example for explaining the invention, and the number of user devices or servers is not limited to that shown in FIG. 1.

[0055] One or more user devices (110-1, 110-2) may be fixed or mobile terminals implemented as a computer system. The one or more user devices (110-1, 110-2) may be, for example, a smart phone, a mobile phone, a navigation device, a computer, a laptop, a digital broadcasting terminal, a PDA (Personal Digital Assistants), a PMP (Portable Multimedia Player), a tablet PC, a game console, a wearable device, an IoT (Internet of Things) device, a VR (Virtual Reality) device, an AR (Augmented Reality) device, etc. For example, in the embodiments, the user device (110) may mean one of various physical computer systems that can communicate with other servers (120 to 140) through a network (1) using a wireless or wired communication method.

[0056] Each server may be implemented as a computer device or multiple computer devices that communicate with one or more user devices (110-1, 110-2) via a network (1) to provide commands, codes, files, contents, services, etc. For example, the server may be a system that provides each service to one or more user devices (110-1, 110-2) connected via the network (1). As a more specific example, the server may provide a service (e.g., provision of information, etc.) intended by an application to one or more user devices (110-1, 110-2) through an application as a computer program that is installed and run on one or more user devices (110-1, 110-2). As another example, the server may distribute a file for installing and running the above-described application to one or more user devices (110-1, 110-2) and receive user input information to provide a corresponding service.

[0057] FIG. 2 is a block diagram illustrating the internal configuration of a computing device (200) according to one embodiment of the present specification. This computing device (200) may be applied to one or more user devices (110-1, 110-2) or servers (120-140) described above with reference to FIG. 1, and each device and server may have the same or similar internal configuration by adding or excluding some components.

[0058] Referring to FIG. 2, a computing device (200) may include a memory (210), a processor (220), a communication module (230), and a transceiver (240). The memory (210) is a non-transitory computer-readable recording medium and may include a non-permanent mass storage device such as a random access memory (RAM), a read only memory (ROM), a disk drive, a solid state drive (SSD), a flash memory, etc. Here, the non-permanent mass storage device such as a ROM, an SSD, a flash memory, a disk drive, etc. may be included in the above-described device or server as a separate permanent storage device distinct from the memory (210). In addition, the memory (210) may store an operating system and at least one program code (for example, a browser installed and operated on a user device (110), or a code for an application installed on a user device (110) to provide a specific service). These software components may be loaded from a computer-readable recording medium separate from the memory (210). This separate computer-readable recording medium may include a computer-readable recording medium such as a floppy drive, a disk, a tape, a DVD / CD-ROM drive, a memory card, etc.

[0059] In another embodiment, the software components may be loaded into the memory (210) via a communication module (230) rather than a computer-readable recording medium. For example, at least one program may be loaded into the memory (210) based on a computer program (e.g., the application described above) that is installed by files provided by developers or a file distribution system (e.g., the server described above) that distributes the installation files of the application via a network (1).

[0060] The processor (220) may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input / output operations. Instructions may be provided to the processor (220) by the memory (210) or the communication module (230). For example, the processor (220) may be configured to execute instructions received according to program code stored in a storage device such as the memory (210).

[0061] The communication module (230) can provide a function for the user device (110) and the server (120 - 140) to communicate with each other via the network (1), and can provide a function for each of the device (110) and / or the server (120 - 140) to communicate with other electronic devices.

[0062] The transceiver (240) may be a means for interfacing with an external input / output device (not shown). For example, external input devices may include devices such as a keyboard, mouse, microphone, camera, etc., and external output devices may include devices such as a display, speaker, haptic feedback device, etc.

[0063] As another example, the transmitter / receiver (240) may be a means for interfacing with a device that integrates input and output functions, such as a touch screen.

[0064] In addition, in other embodiments, the computing device (200) may include more components than the components of FIG. 2 depending on the nature of the device to which it is applied. For example, when the computing device (200) is applied to a user device (110), it may be implemented to include at least some of the above-described input / output devices, or may further include other components such as a transceiver, a Global Positioning System (GPS) module, a camera, various sensors, a database, etc. As a more specific example, when the user device is a smartphone, it may be implemented to further include various components that are generally included in a smartphone, such as an acceleration sensor or a gyro sensor, a camera module, various physical buttons, buttons using a touch panel, input / output ports, and a vibrator for vibration.

[0065] The operations described below are based on a computing device. For example, the computing device may be a device installed in a truck (or other vehicle). Specifically, a single computing device may be installed within the truck, and a system for diagnosing truck malfunctions may operate based on the computing device. In another example, the computing device may be another device or a server, and is not limited to a specific form.

[0066] As an example, the following description focuses on a truck equipped with a computing device. Data sensed by sensors attached to the truck can be acquired and processed based on the computing device installed on the truck, but may not be limited to this. For convenience of explanation, the description below focuses on a truck, but this may be an operation based on the computing device installed on the truck and may not be limited to a specific form.

[0067] FIG. 3 is a drawing illustrating a truck equipped with multiple sensors according to one embodiment of the present disclosure. Referring to FIG. 3 , a truck (300) may be equipped with multiple sensors (310, 320, 330). For example, the multiple sensors (310, 320, 330) may be installed on the underbody of the vehicle. The multiple sensors (310, 320, 330) may include cameras, lidar sensors, depth cameras, color cameras, and other image information acquisition sensors, and are not limited to specific embodiments.

[0068] For example, among the plurality of sensors (310, 320, 330), the first sensor (310) may be a sensor that captures image information by capturing a road on which the vehicle is driving. The first sensor (310) may be installed in front of the front wheels of the vehicle or on the bottom of the vehicle, and may not be limited to a specific location. The first sensor (310) may include a capturing unit for capturing road image information. In addition, as an example, the first sensor (310) may include a vertical angle control unit that controls a shooting angle in a vertical direction in consideration of a road image information acquisition range, and a horizontal angle control unit that controls a shooting angle in a horizontal direction. Here, the first sensor (310) may control the vertical angle control unit and the horizontal angle control unit based on driving status information or road information of the truck (300). As a specific example, when the truck (300) is driving on an uphill road or a downhill road, the vertical angle control unit may control an angle in consideration of a road slope, thereby capturing road image information. For example, the first sensor (310) may be controlled based on a control system within the truck (300). As another example, information regarding the first sensor (310) may be transmitted to a control system within the truck (300), and the control system may transmit the transmitted information to an external system. Here, control by the first sensor (310) may be controlled based on an external system. As another example, the external system may be linked to a device of the truck (300) owner, and is not limited to a specific form.

[0069] In addition, as an example, there may be a plurality of first sensors (310). Specifically, the first sensor (310) may be installed in front of the front wheels of the vehicle to obtain a road image in front of the truck. In addition, the first sensor (310) may be installed behind the rear wheels of the vehicle to obtain a road image at the back of the truck. As another example, the first sensor (310) may be installed on the side of the truck to obtain a road image, and the corresponding image information may be transmitted to the control system. In other words, the first sensor (310) may be a sensor that obtains a road image, and may be installed in each part of the truck to obtain a road image, and is not limited to a specific embodiment.

[0070] In addition, among the plurality of sensors (310, 320, 330), the second sensor (320, 330) may be a sensor that acquires wheel images. For example, the second sensor (320, 330) may be installed between the front and rear wheels of the truck (300) to acquire front and rear wheel images, respectively. Here, the images may be stationary images and rotational surface images based on the movement of the truck. Specifically, the second sensor (320, 330) may acquire wheel surface images when the truck is stopped. As another example, the second sensor (320, 330) may acquire wheel images when the truck is moving based on the driving of the truck. The second sensor (320, 330) may include a photographing unit, a vertical angle control unit, and a horizontal angle control unit, similar to the first sensor (310). The vertical angle control unit and the horizontal angle control unit of the second sensor (320, 330) can be controlled through the control system of the truck (300), thereby obtaining a target wheel image. For example, the control system can obtain wheel images from the second sensor (320, 330) to obtain information on the degree of wheel wear, degree of compression, tire contraction / relaxation information when shaking up and down, and other wheel status information, which will be described later.

[0071] Additionally, as an example, a plurality of second sensors (320, 330) may be included considering the front and rear wheels. That is, each of the plurality of second sensors (320, 330) can acquire images of the front and rear wheels, respectively, and transmit the corresponding information to the control system.

[0072] A truck control system can diagnose truck malfunctions. Specifically, the control system can be equipped with error code analysis and diagnostic software, similar to existing fault diagnosis systems, and can monitor vehicle durability. Furthermore, for example, the control system can monitor each component within the truck based on a sensor network and the Internet of Things (IoT), and obtain information based on each component. As another example, the control system can perform vibration and acoustic analysis of the truck.

[0073] For example, referring to FIG. 4, the truck (300) may further include other sensors (410, 420, 430) in addition to the plurality of sensors (310, 320, 330) described above. For example, the other sensors (410, 420, 430) may be installed on the wheels of the truck to measure the pressure information of the wheels and measure the degree of contraction and relaxation based on the pressure information. As another example, the other sensors (410, 420, 430) may be installed at other locations of the truck (300) and may be sensors that analyze the vibration of the truck (300). As another example, the other sensors (410, 420, 430) may be sensors that are installed at other locations of the truck (300) and sense acoustic information about each component inside the truck, and are not limited to a specific embodiment. For example, if an abnormality occurs in each component inside a truck, the sound generated from each component may be different, and a sensor that senses acoustic information may obtain the corresponding information. That is, the truck (300) may further include sensors (410, 420, 430) other than the sensors (310, 320, 330) that obtain image information, to obtain more truck-related information, and is not limited to a specific embodiment.

[0074] As another example, FIG. 5 is a diagram illustrating a method for performing truck fault diagnosis based on information sensed from the outside of a truck according to one embodiment of the present disclosure. Referring to FIG. 5, the control system of a truck (300) can further acquire truck-related sensing information from an external device. For example, the external device may be a device capable of acquiring image information from a roadside unit (RSU). As a specific example, the external device may be a device that acquires image information from streetlights (511, 512, 513) or traffic lights (520). For example, the external device may acquire image information and transmit it to the truck control system when the truck (300) enters within a preset distance. As another example, the truck control system may recognize an external device located within a preset distance and request acquisition of image information. In other words, image information may be acquired and transmitted based on a request from the truck control system. As another example, the external device may include a drone or other device, but is not limited to a specific embodiment.

[0075] Afterwards, the truck control system can diagnose the presence of a malfunction in the truck based on the acquired information. For example, the truck control system can diagnose the presence of a malfunction in each component within the truck, including the engine, engine components, electrical system, body, external components, transmission, drivetrain, braking system, and other components. In another example, the truck control system can diagnose the cause of a complex malfunction based on each component within the truck, as described below.

[0076] FIG. 6 is a diagram illustrating a method for acquiring image information through multiple sensors in one embodiment of the present specification. Referring to FIG. 6, a truck (300) can acquire a road image while driving on a road. For example, the road image can be acquired based on the first sensor (310) as described above. The road image can be a front image based on the driving direction of the truck (300), but is not limited thereto, and can also include side or rear images. The control system of the truck can acquire information on obstacles (610, 620) based on the image information acquired from the first sensor (310). As another example, the control system of the truck can acquire a front image of the truck through a black box as well as the first sensor (310). Here, the front image of the truck acquired through the black box can be an image from a longer distance than the road image acquired through the first sensor (310). The control system of the truck can improve the accuracy of obstacle recognition by using information about obstacles (610, 620) acquired through the first sensor (310) and information about obstacles (610, 620) in the front image of the truck acquired through the black box.

[0077] Figure 7 is a diagram illustrating image analysis-based reinforcement learning according to one embodiment of the present disclosure. Referring to Figure 7, the control system of a truck (300) may be equipped with a reinforcement learning-based analysis model (700). For example, reinforcement learning is a type of machine learning that operates in a dynamic environment rather than relying on a static data set and may be a method of learning through collected information. While supervised or unsupervised learning may require data collection, preprocessing, and labeling prior to learning, reinforcement learning may not require these operations. Reinforcement learning may be a model that operates based on appropriate incentives (or rewards) and initiates learning behaviors autonomously without intervention. For example, reinforcement learning may determine the environment to which it will be applied and establish an interface between the agent and the environment within the environment. Then, based on the actions of the agent, reward information may be set to suit the environment. Afterwards, it may be a learning method that derives an optimal conclusion based on the actions of each agent within the environment.

[0078] For example, a control system for a truck may include a reinforcement learning-based analysis model (700). The reinforcement learning-based analysis model (700) may operate based on image information, and as described above, may be updated by collecting information in a dynamic environment and providing incentives. For example, the reinforcement learning-based analysis model (700) may perform learning based on prior information. Here, the prior information may include at least one of road image information, wheel image information, truck-related information, environmental information, and vehicle failure information. In other words, the reinforcement learning-based analysis model (700) may have completed learning based on information acquired in advance. Thereafter, input information may be provided to the reinforcement learning-based analysis model (700), inference may be performed, and output information may be derived.

[0079] The input information of the reinforcement learning-based analysis model (700) may include a road image acquired from the first sensor (310) and a wheel image acquired from the second sensor (320, 330). As another example, the input information of the reinforcement learning-based analysis model (700) may further include, but is not limited to, a vehicle front image (black box image), road condition information, weather condition information, traffic condition information, vehicle sensing information, consumable information, and other information. The reinforcement learning-based analysis model (700) may derive at least one of obstacle avoidance, vehicle impact amount information, vehicle damage accumulation information, and other information as output information based on the above-described input information. Here, the reinforcement learning-based analysis model (700) may generate compensation information based on the above-described output information, and update the reinforcement learning-based analysis model (700) based on the compensation information.

[0080] As a specific example, referring to FIG. 8A, the reinforcement learning-based analysis model (700) can derive, as output information, whether or not an obstacle is avoided based on the above-described input information. The control system of the truck can recognize an obstacle (810) in front of the truck (300) through the reinforcement learning-based analysis model (700). As an example, the reinforcement learning-based analysis model (700) can determine whether or not to avoid the obstacle based on the above-described input information. Specifically, if the obstacle (810) is not dangerous and does not cause a large impact to the vehicle, there is a need to pass over the obstacle (810) without avoiding it. On the other hand, if passing over the obstacle (810) without avoiding it would cause significant damage to the vehicle, the obstacle (810) can be avoided. As another example, even when passing over the obstacle (810) without avoiding it, there is a need to determine which of the left and right wheels (321, 322) to pass over.

[0081] For example, the reinforcement learning-based analysis model (700) can determine whether to avoid an obstacle (810) by using the above-described information as input. If the reinforcement learning-based analysis model (700) does not avoid the obstacle (810), the reward information generated based on the above-described operation can be reflected in the reinforcement learning-based analysis model (700) by considering the input information and can be updated based thereon. As another example, if the reinforcement learning-based analysis model (700) avoids the obstacle (810), the reward information generated based on the above-described operation can be reflected in the reinforcement learning-based analysis model (700) by considering the input information and can be updated based thereon. For example, the reward information can be reward information or penalty information, and the accuracy of the judgment based on the reinforcement learning-based analysis model (700) can be increased based on this. That is, the above-described information can be determined through recognition of the type, size, and surrounding situation of the obstacle (810) based on the reinforcement learning-based analysis model (700).

[0082] Also, as a specific example, referring to FIG. 8b, when a vehicle decides to pass through an obstacle, it can be determined which of the left and right wheels (321, 322) to use to pass through the obstacle (820). As a more detailed example, the reinforcement learning-based analysis model (700) can recognize whether a foreign substance is contained in the groove between the treads of the wheels (321, 322) based on the wheel image. As another example, the reinforcement learning-based analysis model (700) can recognize the wear condition of the wheels (321, 322) based on the wheel image. As another example, the reinforcement learning-based analysis model (700) can obtain information on the degree of contraction and relaxation of the wheels based on other sensors. In addition, the reinforcement learning-based analysis model (700) can obtain road condition information and other information. For example, a reinforcement learning-based analysis model (700) can determine which wheel (321, 322) to use to pass an obstacle on a road on which a vehicle is driving based on the above-described method.

[0083] As another example, if one of the wheels (321, 322) already contains foreign matter, the tread groove containing the foreign matter can be controlled to pass over the obstacle (820). That is, by allowing the tread groove containing the foreign matter to pass over the obstacle (820), additional foreign matter can be prevented from entering the other wheels, and smooth driving can be ensured through the above. As a specific example, in FIG. 8b, the truck (300) can be controlled to pass over a sand pile with smaller particles than the tread grooves of the wheels (321, 322) in the sand pile as an obstacle (820), thereby reducing the impact of the vehicle and enabling smooth driving.

[0084] As another example, the reinforcement learning-based analysis model (700) can derive output information for reducing vehicle impact information based on an obstacle (820). Here, the reinforcement learning-based analysis model (700) can control vehicle driving to minimize vehicle impact, thereby increasing vehicle driving efficiency and preventing vehicle breakdown. As another example, the reinforcement learning-based analysis model (700) can derive output information for minimizing vehicle damage accumulation information. For example, damage information may accumulate in a vehicle based on vehicle driving, and vehicle breakdown may occur due to the accumulated damage information. In consideration of the above, the reinforcement learning-based analysis model (700) can control the vehicle to minimize vehicle accumulated damage information based on the above-described input information, and is not limited to a specific embodiment.

[0085] As another example, referring to FIG. 9, the truck control system can control whether a truck is broken down through a truck breakdown model (920). For example, the truck breakdown model (920) may be a learning model that operates based on the reinforcement learning described above. For example, the truck breakdown model (920) may be a learning model that is included in the reinforcement learning-based analysis model (700) and derives impact information, vehicle cumulative damage information, failure information, and consumable information based on the obstacle avoidance described above. For another example, the truck breakdown model (920) may be a learning model that exists separately from the reinforcement learning-based analysis model (700) and derives impact information, vehicle cumulative damage information, failure information, and consumable information by reflecting the obstacle avoidance information of the reinforcement learning-based analysis model (700), and is not limited to a specific embodiment. Referring to FIG. 9, the truck failure model (920) can be trained through truck failure-related big data (910) including at least one of road image information, wheel image information, vehicle status information, and component failure information. For example, the truck failure model (920) can be trained together with the reinforcement learning-based analysis model (700) described above. That is, the truck failure model (920) can be included in the reinforcement learning-based analysis model (700) and trained together. However, the present invention is not limited thereto, and the truck failure model (920) and the reinforcement learning-based analysis model (700) can be trained separately.

[0086] Thereafter, the truck failure model (920) can obtain information sensed from the truck (300). Here, the sensed information may be the input information of FIG. 7. As another example, the sensed information may be input information different from the input information of FIG. 7, and is not limited to a specific embodiment. The truck failure model (920) can provide at least one of impact information, failure information, and consumable information as output information based on the sensed information and the above-described obstacle avoidance information. As a specific example, the truck failure model (920) can derive information on the degree of vehicle impact by reflecting whether the obstacle was avoided based on the above-described input information. Here, the truck failure model (920) can determine whether the vehicle impact is greater than a preset value as an impact that increases the accumulated damage of the vehicle and provide reward information or penalty information as compensation information. In addition, the truck control system (300) can provide output information based on the above-described truck failure model (920) to the user so that the user can manage the truck. As another example, the truck failure model (920) can derive vehicle failure information as output information based on the above-described input information. As a specific example, the truck failure model (920) can maintain the information if a vehicle failure does not occur based on the above-described input information. On the other hand, if a vehicle failure occurs based on the above-described input information and obstacle avoidance information, the truck failure model (920) can provide compensation information to the truck failure model (920) through the vehicle failure information. Here, the truck failure model (920) can derive information on the cause of the failure and complex information on the failure and provide them to the user. That is, the truck failure model (920) can derive whether or not a truck failure has occurred as output information, analyze information on the cause of the truck failure based on the output information, and provide the analyzed information to the user.

[0087] As another example, the truck failure model (920) can derive consumable information as output information based on the input information and obstacle avoidance information described above. Here, the consumable information may be information on parts that require replacement within the vehicle components. The truck failure model (920) can analyze and provide information related to whether the replacement period for vehicle components has arrived and the lifespan of consumables, and can provide the information as compensation information to the truck failure model (920). Thereafter, the truck failure model (920) can control the truck so that the lifespan of consumables can be extended based on the compensation information, or can provide the user with operation information related to truck control.

[0088] As a specific example, referring to FIG. 10, the control system of a truck can derive impact information based on the degree of contraction and relaxation of the wheel (320) according to the speed and type of obstacle based on the truck failure model (920) described above, and generate consumable information based thereon. For example, to derive the impact information described above, the truck failure model (920) can obtain vehicle speed information, obstacle type information, passenger information, and other input information of FIG. 7 as input information. Based on the above-described information, the truck failure model (920) can derive information related to the wheel (320) as a consumable as output information.

[0089] Here, as an example, referring to FIG. 10, the truck failure model (920) can derive consumable information based on real-time information. As a specific example, the vehicle-related information (1010) at time t1, the vehicle-related information (1020) at time t2, and the vehicle-related information (1030) at time t3 may be different, respectively. Here, the truck failure model (920) can derive consumable information based on the vehicle-related information at each time, provide the information to the user, and update the truck failure model (920).

[0090] For example, vehicle-related information and wheel wear information based on the degree of wheel contraction and relaxation may be correlated, and consumable information may be determined based on this. The truck failure model (920) may derive and provide users with information on whether consumables require replacement based on input information, as well as information on extending the lifespan of consumables, and is not limited to a specific embodiment.

[0091] Hereinafter, a flowchart illustrating a method for diagnosing a truck failure through reinforcement learning-based image analysis according to one embodiment of the present specification will be described. Referring to FIG. 11, image information and wheel information may be acquired through at least one sensor installed in a truck (S1110), and truck-related information and environmental information may be acquired (S1120). Thereafter, the truck may generate vehicle-related information based on road image information, wheel image information, truck-related information, and environmental information (S1130), and may provide the generated vehicle-related information to a user (S1140). For example, among the at least one sensor installed in the truck, a first sensor may be attached to a front end of a front wheel of the truck to acquire road image information. For example, the first sensor may include at least one of a photographing unit for acquiring road image information, a vertical angle control unit for controlling a vertical angle of the photographing, and a horizontal angle control unit for controlling a horizontal angle of the photographing, thereby acquiring road image information by taking into consideration the truck-related information and the environmental information.

[0092] In addition, as an example, the first sensor may be attached to the rear end of the rear wheel of the truck to obtain a road image of the rear of the truck, and may be attached to the side of the truck to obtain at least one of a road image of the side of the truck, as described above. As another example, the at least one sensor may include a second sensor to obtain a wheel image. Here, the second sensor may be attached between the front and rear wheels of the truck to obtain a wheel image. The wheel image may include a still image based on a stationary state of the truck and a moving state image based on a driving state of the truck, and based on the wheel state information, wheel wear information, wheel compression information, wheel shrinkage information, and other information may be derived.

[0093] In addition, as an example, the truck-related information may include truck sensing-based engine information, engine parts information, electrical system information, body information, external accessory information, transmission information, drivetrain information, braking system information, consumable information, and other information, and the environmental information may include at least one of road condition information, weather condition information, traffic condition information, and other information. In addition, the vehicle-related information may be derived based on a reinforcement learning-based analysis model that inputs road image information, wheel image information, truck-related information, and environmental information, wherein the vehicle-related information may include at least one of obstacle avoidance information, vehicle impact amount information, and vehicle damage accumulation information. Here, the reinforcement learning-based analysis model may be trained based on previously acquired information. In addition, as an example, when obstacle avoidance information is generated as vehicle-related information, the reinforcement learning-based analysis model may generate information on whether the truck avoids a frontal obstacle and, if it passes through the frontal obstacle without avoiding it, information on the wheels passing through the frontal obstacle. Here, the reinforcement learning-based analysis model may further generate vehicle impact amount information and vehicle damage accumulation information based on whether the truck avoids the obstacle. Vehicle impact information and accumulated damage information are generated by incorporating obstacle type, obstacle size, truck speed, wheel contraction, and other information into a reinforcement learning-based analysis model. The generated vehicle impact information and accumulated damage information can then be reflected in consumable replacement cycle information. Furthermore, trucks may further include a truck failure model in addition to the reinforcement learning-based analysis model. The vehicle impact information and accumulated damage information can be derived based on the truck failure model, as described above.

[0094] The embodiments described above can be implemented at least in part as computer programs and recorded on a computer-readable recording medium. A computer-readable recording medium on which a program for implementing the embodiments is recorded includes any type of recording device that stores data that can be read by a computer. Examples of computer-readable recording media include ROMs, RAMs, CD-ROMs, magnetic tapes, and optical data storage devices. In addition, the computer-readable recording medium can be distributed across network-connected computer systems, so that computer-readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present embodiments will be readily understood by those skilled in the art to which the present embodiments pertain.

[0095] Although the present specification has been described with reference to the embodiments illustrated in the drawings, these are merely exemplary, and those skilled in the art will understand that various modifications and variations of the embodiments are possible. However, such modifications should be considered within the technical protection scope of the present specification. Therefore, the true technical protection scope of the present specification should be determined to include other implementations, other embodiments, and equivalents to the claims, based on the technical spirit of the appended claims.

[0096] The present invention can be applied industrially by providing a method for obtaining image information and other information using a plurality of sensors and analyzing the information based on reinforcement learning.

Claims

1. In a method for diagnosing truck failures using image analysis based on reinforcement learning, A step of acquiring road image information and wheel image information through at least one sensor installed on a truck; A step of obtaining the above cargo vehicle-related information and environmental information; A step of generating vehicle-related information based on the road image information, the wheel image information, the truck-related information, and the environment information; and A method for diagnosing a truck failure, comprising the step of providing the generated vehicle-related information to a user.

2. In paragraph 1, Among the at least one sensor installed on the truck, the first sensor is attached to the front end of the front wheel of the truck to obtain the road image information. The first sensor includes at least one of a photographing unit for obtaining the road image information, a vertical angle control unit for controlling a vertical angle of photographing, and a horizontal angle control unit for controlling a horizontal angle of photographing. A method for diagnosing a truck failure, wherein the road image information is acquired by taking into consideration the truck-related information and the environmental information.

3. In paragraph 2, A method for diagnosing a truck failure, wherein the first sensor further includes at least one of a sensor attached to the rear end of the rear wheels of the truck to obtain the road image of the rear of the truck and a sensor attached to the side of the truck to obtain the road image of the side of the truck.

4. In paragraph 3, The at least one sensor comprises a second sensor for acquiring the wheel image, The second sensor is attached between the front and rear wheels of the truck to obtain the wheel image. The above wheel image includes a still image based on a stationary state of a truck and a moving state image based on a driving state of a truck. A truck failure diagnosis method in which wheel wear information, wheel compression information, wheel shrinkage information, and other information are derived based on the above wheel condition information.

5. In paragraph 1, The above truck-related information includes truck sensing-based engine information, engine parts information, electrical system information, body information, external accessory information, transmission information, drive train information, braking system information, consumables information, and other information. A method for diagnosing a truck failure, wherein the above environmental information includes at least one of road condition information, weather condition information, traffic condition information, and other information.

6. In paragraph 1, The above vehicle-related information is derived based on a reinforcement learning-based analysis model that uses the road image information, the wheel image information, the truck-related information, and the environmental information as inputs. The above vehicle-related information includes at least one of obstacle avoidance information, vehicle impact amount information, and vehicle damage accumulation information. A method for diagnosing truck failures, wherein the above reinforcement learning-based analysis model is learned based on pre-acquired information.

7. In paragraph 6, A method for diagnosing a truck failure, wherein when information on whether or not to avoid the obstacle is generated using the vehicle-related information, the reinforcement learning-based analysis model generates information on whether or not to avoid the obstacle in front of the truck and, if the truck passes through the obstacle without avoiding the obstacle in front, information on the wheels that pass through the obstacle in front.

8. In paragraph 7, The above reinforcement learning-based analysis model further generates the vehicle impact information and the vehicle damage accumulation information based on whether the obstacle is avoided. The above vehicle impact information and the above vehicle damage accumulation information are generated by reflecting the type of obstacle, the size of the obstacle, the speed of the truck, the degree of wheel contraction, and other information into the reinforcement learning-based analysis model. A method for diagnosing a truck failure, wherein the generated vehicle impact information and the vehicle damage accumulation information are reflected in consumable replacement cycle information.

9. In paragraph 8, The above truck further includes a truck failure model together with the reinforcement learning-based analysis model, A truck failure diagnosis method, wherein the vehicle impact information and the vehicle damage accumulation information are derived based on the truck failure model.

10. In paragraph 1, Among the at least one sensor installed on the truck, the first sensor is attached to the front end of the truck front wheel, the rear end of the truck rear wheel, and the side of the truck to obtain the road image, and the first sensor includes at least one of a photographing unit for obtaining the road image information, a vertical angle control unit for controlling the vertical angle of the photographing, and a horizontal angle control unit for controlling the horizontal angle of the photographing, and obtains the road image information by taking into consideration the truck-related information and the environmental information. Among the at least one sensor installed on the truck, the second sensor is attached between the front and rear wheels of the truck to obtain the wheel image, wherein the wheel image includes a still image based on the truck stationary state and a moving image based on the truck driving state, and based on the wheel state information, wheel wear information, wheel pressing information, wheel shrinkage information, and other information are derived. The above truck-related information includes truck sensing-based engine information, engine parts information, electrical system information, body information, external accessory information, transmission information, drive train information, braking system information, consumables information, and other information. The above environmental information includes at least one of road condition information, weather condition information, traffic condition information, and other information, The vehicle-related information is derived based on a reinforcement learning-based analysis model that inputs the road image information, the wheel image information, the truck-related information, and the environment information, wherein the vehicle-related information includes at least one of obstacle avoidance information, vehicle impact amount information, and vehicle damage accumulation information, and the reinforcement learning-based analysis model is learned based on pre-acquired information. A method for diagnosing a truck failure, wherein when information on whether or not to avoid the obstacle is generated using the vehicle-related information, the reinforcement learning-based analysis model generates information on whether or not to avoid the obstacle in front of the truck and, if the truck passes through the obstacle without avoiding the obstacle in front, information on the wheels that pass through the obstacle in front.

11. A computer program stored in a computer-readable medium that is combined with hardware and executes a method for diagnosing a truck failure through image analysis based on reinforcement learning according to any one of claims 1 to 10.

Images