A Robot Platform for Investigating a Defect of a Rail Based on an Artificial Intelligent Algorism

Abstract

The present invention relates to an artificial intelligence-based robot platform for inspecting rail defects. The artificial intelligence-based robot platform for inspecting rail defects comprises: an autonomous driving robot (12) capable of autonomously moving along a rail according to the operation of a central control module (11); a fusion sensor module (13) installed on the autonomous driving robot (12) to detect the condition of the rail; a data acquisition / processing module (14) that acquires and processes detection data by the fusion sensor module (13); a defect analysis server (15) that analyzes rail defects by analyzing data transmitted from the data acquisition / processing module (14); and an artificial intelligence learning algorithm (16) that learns the data acquired by the fusion sensor module (13) and provides the learning results to the defect analysis server (15).

Description

Technology Field

[0001] The present invention relates to an artificial intelligence-based robot platform for rail defect inspection, and specifically, to an artificial intelligence-based robot platform for rail defect inspection that enables the detection of rail defects through rail driving and rail condition detection by an autonomous driving robot equipped with a fusion composite sensor. Background Technology

[0002] Railways can develop defects due to various reasons. For example, wear of the rails, breakage of the rails, deformation due to deterioration of the rails, poor contact, cracks or similar defects can occur. Such defects can cause train accidents such as train derailment, and train accidents can cause human casualties and huge economic losses due to operation delays or recovery, so preventive measures need to be taken to prevent accidents. The types of track wear corresponding to rail defects can be classified into adhesive wear destroyed by adhesive defects in the contact part; abrasive wear caused by the cutting action of hard stones, protrusions or hard particles; corrosive wear caused by the corrosion action of the surface or lubricant; and fatigue wear corresponding to fatigue fracture near the surface due to repeated stress in the contact part. The types of track defects can also be classified into fractures with holes in the joints; rail partial wear where one part of the rail is worn; surface peeling where the rail surface peels off; welding part pitting where a part of the welding part is pitted; surface hardening; deterioration, scaly wear; chipping; joint part chipping; adhesive insulation crack; and rail defect. Such a wear state or type of track defect of the railway needs to be detected in advance and appropriate measures should be taken accordingly. For accident prevention of railways, the rails need to be inspected in advance and maintenance work should be carried out as necessary. However, it is difficult to detect rail defects through the patrol inspection of maintenance workers for inspection and maintenance of railway rails, and there are problems such as difficulty in application due to design changes of manufacturers and increased costs for inspection and maintenance or administrative problems. In addition, the method of visual inspection by workers has limitations in defect detection, can cause train accidents, and has the problem that workers have to work in a harsh environment.For example, if work is carried out at night, accidents involving collisions with trains may occur, and if work is performed during the summer when track deformation is likely to occur, workers may be put at risk. In order to address the issues regarding the effectiveness and reliability of railway track inspections, as well as the problems associated with manual inspection and maintenance, it is necessary to develop technology capable of automatically inspecting the condition of railway tracks. Regarding rail defect detection, Patent Registration No. 10-1734376 discloses a device for automatically detecting cracks in rails. Patent Publication No. 10-2019-0024447 discloses a real-time track defect inspection system. Patent Registration No. 10-2116890 discloses a system and method capable of analyzing and monitoring rail / track defects in real time while a train is in motion using a wireless accelerometer. Additionally, Patent Registration No. 10-1582250 discloses a track defect monitoring device. However, such prior art or known art does not disclose a technology capable of detecting and analyzing the condition of a track by a robot autonomously driving along the track, and inspecting track defects based thereon.

[0003] The present invention aims to solve the problems of the prior art and has the following objectives. Prior art literature

[0004] Prior Art 1: Patent Registration No. 10-1734376 (Seyoung Communications Co., Ltd., published May 12, 2017) Railway rail crack detection device Prior Art 2: Patent Publication No. 10-2019-0024447 (Two ISIS Co., Ltd., published March 8, 2019) Real-time track defect inspection system Prior Art 3: Patent Registration No. 10-2116890 (GSG Co., Ltd., published May 209, 2020) System and method for real-time analysis and monitoring of mobile rail / track defects using a wireless accelerometer Prior Art 4: Patent Registration No. 10-1582250 (Luxco Co., Ltd., published January 6, 2016) Track defect monitoring device The problem to be solved

[0005] The objective of the present invention is to provide an AI-based rail defect inspection robot platform capable of inspecting track defects based on detection data acquired while an autonomous driving robot equipped with a fusion composite sensor module travels along a track and learns from the data using an AI learning algorithm. means of solving the problem

[0006] According to a suitable embodiment of the present invention, an artificial intelligence-based rail defect inspection robot comprises: an autonomous driving robot capable of autonomously moving along a rail according to the operation of a central control module; a fusion sensor module installed on the autonomous driving robot to detect the rail condition; a data acquisition / processing module that acquires and processes detection data by the fusion sensor module; a defect analysis server that analyzes rail defects by analyzing data transmitted from the data acquisition / processing module; and an artificial intelligence learning algorithm that learns data acquired by the fusion sensor module and provides the learning result to the defect analysis server.

[0007] According to another suitable embodiment of the present invention, an autonomous driving robot comprises: a pair of driving guidance blocks formed in positions facing each other; a plurality of driving wheels installed on the pair of driving guidance blocks; at least one motor for operating the plurality of driving wheels; and a driving guidance camera installed in the middle portion of the pair of driving guidance blocks.

[0008] According to another suitable embodiment of the present invention, a risk situation detection module for detecting a risk situation is further included.

[0009] According to another suitable embodiment of the present invention, the fusion sensor module includes a scanner for detecting rail breakage; a thermal imaging camera for detecting rail deterioration; and a vibration sensor for detecting vibrations occurring in the rail.

[0010] According to another suitable embodiment of the present invention, it further includes a train position / operation analysis module that analyzes the position and operation status of a train running along a rail, and a risk determination module that determines a risk level based on the analysis results of the train position / operation analysis module. Effects of the invention

[0011] The AI-based rail defect inspection robot platform according to the present invention replaces existing railway track inspection and maintenance work, which involves searching for defects in railway tracks manually or by using defect detection devices, thereby enabling inspection and maintenance work to be carried out economically and efficiently. Furthermore, the robot platform according to the present invention prevents railway accidents by detecting and repairing railway track defects in advance. The robot platform according to the present invention can be applied to inspect defects in railway tracks having various structures and has the advantage of not requiring separate operational proficiency. Additionally, the robot platform according to the present invention improves the reliability of judgment by verifying the presence of defects in the railway track through an AI learning algorithm. Brief explanation of the drawing

[0012] FIG. 1 illustrates an example of an artificial intelligence-based robot platform for rail defect inspection according to the present invention. FIGS. 2 and 3 illustrate embodiments of an autonomous driving robot for a robot platform according to the present invention. FIG. 4 illustrates an embodiment of an autonomous driving robot for a robot platform according to the present invention driving along a rail. FIG. 5 illustrates an example of the operating structure of a robot platform according to the present invention. FIG. 6 illustrates an example of a sensor installed on an autonomous driving robot for a robot platform according to the present invention. FIG. 7 illustrates an example of a rail defect detection method using a robot platform according to the present invention. Specific details for implementing the invention

[0013] The present invention is described in detail below with reference to embodiments shown in the attached drawings, but the embodiments are for a clear understanding of the invention and the invention is not limited thereto. In the description below, components having the same reference numeral in different drawings have similar functions and are not described repeatedly unless necessary for understanding the invention, and known components are described briefly or omitted, but should not be understood as being excluded from the embodiments of the present invention.

[0014] FIG. 1 illustrates an example of an artificial intelligence-based robot platform for rail defect inspection according to the present invention.

[0015] Referring to FIG. 1, the AI-based rail defect inspection robot platform includes: an autonomous driving robot (12) capable of autonomously moving along a rail according to the operation of a central control module (11); a fusion sensor module (13) installed on the autonomous driving robot (12) to detect the rail condition; a data acquisition / processing module (14) that acquires and processes detection data by the fusion sensor module (13); a defect analysis server (15) that analyzes rail defects by analyzing data transmitted from the data acquisition / processing module (14); and an AI learning algorithm (16) that learns the data acquired by the fusion sensor module (13) and provides the learning results to the defect analysis server (15).

[0016] The central control module (11) may have the function of controlling the rail defect detection process. Specifically, the central control module (11) may monitor the driving process while setting the driving of the autonomous driving robot (12) and control the operation of the fusion sensor module (13) installed on the autonomous driving robot (12). Additionally, it may receive detection data acquired and processed by the fusion sensor module (13) and transmit it to the defect analysis server (15). The autonomous driving robot (12) may have a structure capable of moving along the rail and may move along the rail autonomously or according to the setting or control of the central control module (11). When the driving conditions of the autonomous driving robot (11) are set by the central control module (11), the autonomous driving robot (11) may move along the rail based on this. The autonomous driving robot (11) may include a driving control module for independent driving control, and the autonomous driving robot (11) may move along the rail according to the control of the driving control module. For example, the autonomous driving robot (11) includes a situation detection camera that detects the condition of the rail, and can detect the condition of the rail ahead from an image obtained by the situation detection camera. By doing so, the autonomous driving robot (12) can check the condition of the rail ahead and drive accordingly. For example, it can check whether there is an obstacle on the rail or whether the rail corresponds to a curved section, and can stop driving or reduce the driving speed. The autonomous driving robot (12) can detect the speed and drive autonomously at a predetermined speed and drive along a predetermined section. A fusion sensor module (13) can be installed on such an autonomous driving robot (12) to detect the rail condition. The fusion sensor module (13) may include, but is not limited to, a laser scanner, a camera scanner, a thermal imaging camera, an ultrasonic sensor, a vibration sensor, or a noise sensor.As the autonomous driving robot (12) moves, the fusion sensor module (13) can detect the rail condition, and the information detected by the fusion sensor module (13) can be transmitted to the data acquisition / processing module (14). The data acquisition / processing module (14) can be mounted on the autonomous driving robot (12) or installed at a designated location adjacent to the rail. The data acquisition / processing module (14) can have the function of acquiring and processing detection data transmitted from the fusion sensor module (13). The data acquisition / processing module (14) can acquire detection data from the fusion sensor module (13) via wired or wireless means, and the data acquisition / processing module (14) can check for errors in the detection data. Additionally, the data acquisition / processing module (14) can classify and store the detection data, and the data acquisition / processing module (14) can include a rail map and correspond the acquired detection data to the rail map over time. Furthermore, the data acquisition / processing module (14) can convert the detection data into a form that can be transmitted via wireless communication means. Then, the converted detection data can be transmitted to the defect analysis server (15). The defect analysis server (15) can detect the defect status of the rail by analyzing the detection data transmitted from the data acquisition / processing module (14). For example, the degree of deterioration of each part of the rail, whether cracks have occurred, rail damage, or similar defects can be analyzed. An artificial intelligence learning algorithm (16) can be utilized during the process of analyzing rail defects by the defect analysis server (15). The artificial intelligence learning algorithm (16) can learn detection data acquired by the fusion sensor module (13) or detection data for the rail acquired by various methods, and can analyze the detection data transmitted from the data acquisition / analysis module (14) based on the learning results.If damage, cracks, or deterioration of the rail is confirmed from the analysis results by the artificial intelligence learning algorithm (16), repair or inspection can be carried out accordingly. In this way, rail defects can be automatically analyzed by the autonomous driving robot (12) traveling along the rail and acquiring and analyzing detection data through the fusion sensor module (13). Various types of artificial intelligence learning algorithms (16) can be applied to the learning and analysis of detection data, and the present invention is not limited by this.

[0017] FIGS. 2 and 3 illustrate embodiments of an autonomous driving robot for a robot platform according to the present invention.

[0018] Referring to FIGS. 2 and 3, the autonomous driving robot (12) comprises: a pair of driving guidance blocks (22a, 22b) formed at positions facing each other; a plurality of driving wheels (25_1 to 25_K) installed on the pair of driving guidance blocks (22a, 22b); at least one motor (26_1 to 26_K) that operates the plurality of driving wheels (25_1 to 25_K); and a driving guidance camera (23) installed in the middle portion of the pair of driving guidance blocks (22a, 22b). The autonomous driving robot may have a structure capable of moving along a rail and may have a structure capable of loading a fusion sensor module for acquiring detection data that detects the state of the rail. The autonomous driving robot may include a frame (F) having an overall rectangular planar shape, and a pair of receiving housings (21a, 21b) may be coupled to the upper part of the frame (F), and each of the pair of receiving housings (21a, 21b) may have a roof shape in which multiple planes are connected at an angle. A camera base (23a) may be installed between the pair of receiving housings (21a, 21b). Additionally, a driving guide block (22a, 22b) may be disposed at one end of each receiving housing (21a, 21b), and the pair of driving guide blocks (22a, 22b) may be disposed facing each other. The pair of driving guide blocks (22a, 22b) may be formed to be positioned on rails facing each other, and each driving guide block (22a, 22b) comprises a pair of polygonal planes facing each other; It may consist of a polygonal plane and a connecting plane that connects the sides excluding the bottom side of the polygonal plane. Each of the pair of polygonal planes may consist of a linear bottom side; a pair of vertical sides extending vertically from both ends of the linear bottom side; a pair of inclined sides extending inwardly from the upper ends of the pair of vertical sides; and a horizontal side connecting the upper ends of the pair of inclined sides.Each polygonal plane can be a hexagonal shape having a lower side with a larger length and an upper side with a smaller length, and the driving guidance blocks (22a, 22b) can be a polyhedral shape with an open bottom face. The connecting plane connecting the polygonal planes facing each other of the driving guidance blocks (22a, 22b) can have a width corresponding to the width of each rail, and each driving guidance block (22a, 22b) can be a hollow shape with an open bottom face. During the process of the autonomous driving robot driving on the rail, each driving guidance block (22a, 22b) can be positioned above each rail. A frame (F) can be arranged in a manner connecting the lower ends of a pair of driving guidance blocks (22a, 22b), and a pair of receiving housings (21a, 21b) can be arranged above the frame (F). Each receiving housing (21a, 21b) has a pair of vertical surfaces extending to face each other; A pair of inclined surfaces extending upward while sloping inward from the upper edge of a pair of vertical surfaces; and a horizontal surface connecting the upper surfaces of the pair of inclined surfaces may be formed, and the receiving housings (21a, 21b) may have a smaller height compared to the driving guide blocks (22a, 22b), but are not limited thereto, a camera base (23a) may be placed between the receiving housings (21a, 21b), and the camera base (23a) may be formed to have a smaller length and a smaller height while having a shape similar to each of the receiving housings (21a, 21b), but is not limited thereto. A pair of receiving housings (21a, 21b) may be connected to each other by the camera base (23a), and the camera base (23a) may be formed in the middle part in the width direction of the frame or in the middle part between two rails. The upper surface of the camera base (23a) may have a flat shape, and a driving guidance camera (23) may be placed on the upper surface of the camera base (23a).The driving guidance camera (23) can be positioned so that the autonomous driving robot faces forward in the driving direction, and an image or video of the front of the autonomous driving robot can be acquired by the driving guidance camera (23). The situation in front of the autonomous driving robot can be detected by the front image acquired by the driving guidance camera (23), and autonomous driving can be controlled based on this. A pair of sensor installation blocks (24a, 24b) can be installed in a manner that penetrates vertically through the upper surface of the receiving housing (21a, 21b) at the boundary portion between the driving guidance blocks (22a, 22b) and the receiving housing (21a, 21b), and one side of the sensor installation blocks (24a, 24b), which have a cuboid shape, can be installed on one side of the driving guidance blocks (22a, 22b). Various sensors related to rail condition detection, such as vibration sensors, noise sensors, infrared temperature sensors, or similar sensors, may be placed in the sensor installation blocks (24a, 24b), and the present invention is not limited thereto.

[0019] Referring to FIG. 3, two pairs of driving wheels (25_1 to 25_K) are placed inside driving guidance blocks (22a, 22b) to allow an autonomous driving robot to move along a rail. The driving wheels (25_1 to 25_K) can be rotated by motors (26_1 to 26_K), such as BLDC motors, placed inside a receiving housing (21a, 21b). An operating driver or motor operating driver that controls the operation of the motors may be placed inside the receiving housing (26_1 to 26_K), and an encoder may also be placed. A thermal imaging camera (27a, 27b) may be placed inside the driving guidance blocks (22a, 22b). The thermal imaging camera (27a, 27b) can acquire a thermal image of each rail through the open bottom surface of the driving guidance blocks (22a, 22b). A scan camera, laser scanner, ultrasonic sensor, or similar scanner or sensor may be placed using the sensor installation block, but is not limited thereto, and the scanner or sensor may be placed at various locations in the receiving housing (21a, 21b) or the driving guidance block (22a, 22b). A battery may be placed in the middle part of the frame (F) of the battery base (28) and may be placed on the top of the battery base (28). The battery may be various types of rechargeable batteries, and at least one battery may be placed at various locations inside the receiving housing (21a, 21b), and the present invention is not limited thereto. Various parts or components related to the driving of the autonomous driving robot, communication with the central control module, or operation of the fusion sensor module may be placed inside the receiving housing (21a, 21b), and the present invention is not limited thereto. Additionally, a detection data acquisition and processing module may be placed inside the receiving housing (21a, 21b), and the present invention is not limited by the parts, elements, or modules placed inside the receiving housing (21a, 21b).

[0020] FIG. 4 illustrates an embodiment of an autonomous driving robot for a robot platform according to the present invention driving along a rail.

[0021] Referring to FIG. 4, a pair of rails (41a, 41b) may be installed facing each other on the ground (E), and an autonomous driving robot (20) may move autonomously along the pair of rails. The condition of the rails ahead may be checked by a driving guidance camera (23), and based on this, the autonomous driving robot (20) may move along the rails (41a, 41b) at a predetermined speed. Scanners or sensors may be placed at various locations on the sensor installation blocks (24a, 24b), guidance blocks, or receiving housings to obtain detection data regarding the condition of the rails (41a, 41b), and the obtained detection data may be transmitted to a defect analysis server for analysis. The autonomous driving robot (20) may have various structures capable of autonomously driving along the rails (41a, 41b), and the present invention is not limited thereto.

[0022] FIG. 5 illustrates an example of the operating structure of a robot platform according to the present invention.

[0023] Referring to FIG. 5, the AI-based rail defect inspection robot further includes a danger situation detection module (56) for detecting danger situations. Driving conditions of the autonomous driving robot (12) can be set by the driving control module (51), for example, the driving speed or driving section can be set. The driving control module (51) can be installed in the central control module (51) and can have the function of monitoring the driving of the autonomous driving robot (12). A rail detection module (53) can be installed in the autonomous driving robot (12), and the rail detection module (53) can detect the track condition by analyzing an image acquired by a driving guidance camera. The central control module and the autonomous driving robot (12) can communicate via wired communication, but preferably can communicate via wireless communication (52). The rail detection module (53) can transmit track detection results to the driving control module (51) via wireless communication (52), and the driving control module (51) can set driving conditions for the autonomous driving robot (12) based on the transmitted track detection results. The state of the rail can be detected by the fusion sensor module (13) installed on the autonomous driving robot (12) and transmitted to the data acquisition / processing module (14), and the data acquisition / processing module (14) can process the received detection data according to a predetermined method and transmit it to the defect analysis server (15). Optionally, the data acquisition / processing module (14) and the defect analysis server (15) can be connected by a relay. In the defect analysis server (15), an artificial intelligence learning algorithm (16) is utilized to analyze the detection data and detect rail defects. According to one embodiment of the present invention, a dangerous situation detection module (56) can be installed on the autonomous driving robot (12), and a dangerous situation occurring on the rail can be detected by the dangerous situation detection module (56). For example, the danger situation detection module (56) can detect whether a train is approaching based on the detection result from a vibration sensor or a noise sensor, and detect whether it corresponds to a danger situation.Alternatively, the danger situation detection module (56) can detect a danger situation by determining whether there is an object or person on the rail or whether there is a damaged part of the rail from an image acquired by a driving guidance camera. The danger situation detection module (56) can detect the occurrence of a danger and transmit it to the danger situation propagation module (57). The danger situation propagation module (57) may include a siren, voice, or similar danger propagation means, and the danger situation propagation module (57) may be installed in a central control module or installed independently. The danger situation detection module (56) can detect various forms of danger that may occur on the rail or around the rail by various sensors, and the present invention is not limited thereto.

[0024] FIG. 6 illustrates an example of a sensor installed on an autonomous driving robot for a robot platform according to the present invention.

[0025] Referring to FIG. 6, the fusion sensor module (13) installed on the autonomous driving robot may include a laser scanner or a scan camera and scanner (131); and various types of sensors for detecting rail conditions, such as a thermal imaging camera (132), an ultrasonic sensor (133), a vibration sensor (134), or a noise sensor (135). A rail scan image may be acquired by the scanner (131), and the rail damage state may be detected by the rail damage detection module (61). A thermal image of the rail may be acquired by the thermal imaging camera (132), and the thermal image may be analyzed by the rail deterioration detection module (62) to detect the rail deterioration state. An ultrasonic image acquired by the ultrasonic sensor (133) may be transmitted to and analyzed by the internal crack detection module (63), and an internal crack of the rail may be detected based on the analysis result. Vibration or noise generated in the rail can be detected by a vibration sensor (134) or a noise sensor (135) and transmitted to a risk detection module (64), and the risk detection module (64) can analyze the vibration or noise to determine whether a risk has occurred. Optionally, the rail condition can be detected by the vibration sensor (134) or the noise sensor (135). An impact that causes vibration or noise to be applied to the rail can be applied, and vibration detection data or noise detection data can be acquired by the vibration sensor (134) or the noise sensor (135) and analyzed by a defect analysis server. Once the analysis of rail breakage, rail deterioration, or internal cracks is completed by the defect analysis server, rail condition data can be generated by the rail condition data generation module (65). Based on the generated rail condition data, the maintenance / inspection location determination module (66) can determine the maintenance or inspection location of the rail. The risk detection module (64) can detect a risk from vibration or noise transmitted through the rail. Additionally, a dangerous situation can be detected from an image acquired by a driving guidance camera.When a risk is detected by the risk detection module (64), the train location can be analyzed by the train location / operation analysis module (67), and at the same time, whether the train is approaching or moving away can be analyzed. The train location or the train operation pattern can be analyzed based on the vibration level or vibration change, and the risk level can be determined by the risk determination module (68). Then, an alarm can be generated according to the determined risk level to notify of the dangerous situation. The fusion sensor module (13) may include various sensors for detecting rail conditions or detecting the occurrence of a risk, and the present invention is not limited thereto.

[0026] FIG. 7 illustrates an example of a rail defect detection method using a robot platform according to the present invention.

[0027] Referring to FIG. 7, the rail defect detection method comprises the steps of: placing an autonomous driving robot equipped with a fusion sensor module on a track or rail (P71); setting driving conditions of the autonomous driving robot or detection conditions of the fusion sensor module (P72); autonomous driving of the autonomous driving robot along the rail according to the set conditions (P73); obtaining detection data regarding the state of the rail by the fusion sensor module (P74); processing the detected data and transmitting it to a defect analysis server (P75); analyzing the transmitted detection data by an artificial intelligence learning algorithm that has learned data regarding defects in various forms of the rail (P76); detecting a risk occurring on the rail by a risk detection means (P77); generating a risk signal (P78); detecting whether the risk level increases or decreases (P79); and generating an alarm according to the increase in the risk level (P80).

[0028] The fusion sensor module may include a scan camera, a laser scanner, a thermal imaging camera, an ultrasonic sensor, a vibration sensor, a noise sensor, or similar sensors, and may be installed on an autonomous robot. The autonomous robot may have various structures capable of autonomously driving along a rail and may be placed on the rail (P71). As the robot drives along the rail under predetermined conditions (P73), detection data regarding the rail condition may be acquired by the fusion sensor module (P74). The detection data may be analyzed by a defect analysis server to detect rail defects (P76), and during the rail condition detection process, hazards occurring on the rail may be detected by a hazard detection sensor, such as a vibration sensor (P77). For example, a hazard may occur such as a train approaching the track, and a hazard signal may be generated accordingly (P78). When a hazard signal is generated, the level of risk may be determined by a control module installed on the autonomous robot (P78). If the train moves away from the detection area and the risk is reduced accordingly (P79, N), risk detection can continue (P77). Conversely, if the risk increases (P79, Y), an alarm can be generated and the risk can be propagated (P80). The alarm can be generated in various forms, and the present invention is not limited by this.

[0029] Although the present invention has been described in detail above with reference to the presented embodiments, those skilled in the art may make various modifications and variations without departing from the technical spirit of the invention by referring to the presented embodiments. The present invention is not limited by such modifications and variations, but is limited only by the claims appended below. Explanation of the symbols

[0030] 11: Central Control Module 12: Autonomous Driving Robot 13: Fusion Sensor Module 14: Data Acquisition / Processing Module 15: Defect Analysis Module 16: AI Learning Algorithm 22a, 22b: Driving guidance blocks 23: Driving guidance camera 25_1 to 25_K: Driving wheels 26_1 to 26_K: Motors 56: Hazardous Situation Detection Module 67: Train Location / Operation Analysis Module 68: Risk Determination Module 131: Scanner 132: Thermal imaging camera 134: Vibration sensor

Claims

Claim 1 The autonomous driving robot (12) capable of autonomously moving along a rail according to the operation of a central control module (11); a fusion sensor module (13) installed on the autonomous driving robot (12) to detect the rail condition; a data acquisition / processing module (14) that acquires and processes detection data by the fusion sensor module (13); a defect analysis server (15) that analyzes rail defects by analyzing data transmitted from the data acquisition / processing module (14); and an artificial intelligence learning algorithm (16) that learns the data acquired by the fusion sensor module (13) and provides the learning results to the defect analysis server (15), wherein the autonomous driving robot (12) comprises a pair of roof-shaped housings (21a, 21b) that are coupled to the upper part of a square planar frame (F) and each have multiple planes connected to be inclined; and a driving guidance camera (23) installed on the upper surface of a camera base (23a) formed between the pair of housings (21a, 21b) to connect the pair of housings (21a, 21b) to each other. A pair of driving guide blocks (22a, 22b) having a hollow polyhedral shape with an open bottom face, formed to be positioned on rails facing each other and disposed at one end of each receiving housing (21a, 21b); a plurality of driving wheels (25_1 to 25_K) installed on the pair of driving guide blocks (22a, 22b); at least one motor (26_1 to 26_K) for operating the plurality of driving wheels (25_1 to 25_K); and a pair of sensor installation blocks (24a, 24b) installed in a manner that penetrates vertically through the upper surface of the receiving housing (21a, 21b) at the boundary portion between the driving guide blocks (22a, 22b) and the receiving housing (21a, 21b), and the fusion sensor module (13) includes a scanner (131) installed on the pair of sensor installation blocks (24a, 24b) for detecting rail damage; A thermal imaging camera (132) installed inside the driving guide block (22a, 22b) for detecting rail deterioration;An artificial intelligence-based rail defect inspection robot platform characterized by including a vibration sensor (134) installed in a pair of sensor installation blocks (24a, 24b) for detecting vibrations occurring in the rail.; Claim 2 delete Claim 3 An artificial intelligence-based rail defect inspection robot platform according to claim 1, further comprising a risk situation detection module (56) for detecting a risk situation. Claim 4 delete Claim 5 The artificial intelligence-based rail defect inspection robot platform according to claim 1, further comprising a train position / operation analysis module (67) that analyzes the position and operation status of a train running along a rail, and a risk determination module (68) that determines a risk level based on the analysis results of the train position / operation analysis module (67).

Images