Fall monitoring control device, fall monitoring system, fall monitoring control method, fall monitoring control program

Abstract

Using three-dimensional data, the system determines the movement and size of detected objects, and distinguishes between people who have fallen and other objects that trigger alarms. [Solution] Point cloud data is generated from three-dimensional image data captured by the 3D-LiDAR distance measuring device 18. Objects are detected from the difference between frames, and for the detected object, it is determined whether the length, width, or depth of the three-dimensional box 74 exceeds a threshold value X for multiple frames (N frames). Based on the result, a fall detection (e.g., threshold value X exceeds in all frames) or a near-miss alarm detection (e.g., threshold value X exceeds in any frame) is performed. In addition to fall detection, a near-miss alarm detection is performed as a check when the certainty of fall detection is low. This increases the reliability of fall detection and also enables the issuance of a near-miss alarm.

Description

[Technical Field]

[0001] The present invention relates to a fall monitoring control device, a fall monitoring system, a fall monitoring control method, and a fall monitoring control program for monitoring the presence or absence of an object falling from a station platform onto a train track. [Background Art]

[0002] Conventionally, there has been described a fall detection device that detects a person who has fallen from a station platform onto a track through which trains pass.

[0003] Patent Document 1 describes that a fall detection area for detecting a person who has fallen onto a track is set, a foreign matter detection area for detecting foreign matters other than a person present on the track is set, the detection size for foreign matters is set to be smaller than the detection size for fall detection, and the presence or absence of foreign matters in the foreign matter detection area is determined. More specifically, a plurality of scanner units are arranged in the height direction of the platform, scanning light output from a light emitting unit is deflected by a deflector to form a two-dimensional scanning range. When an object exists within the scanning range, a person who has fallen and weeds are distinguished and detected based on whether the light receiving unit can receive reflected light from the object (the fallen person or the weeds). [Prior Art Documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Unexamined Patent Publication No. 2023-177633 [Summary of the Invention] [Problem to be Solved by the Invention]

[0005] However, although the detection of foreign matters (weeds) employs time-series determination requiring a detection rate of 80% or more within one hour, fall detection is simply based only on the detection result of received light data. Furthermore, although the distance to the detected object is calculated, since detection is performed by a two-dimensional sensor (scanner unit), the size of the object (and changes in the size) is not taken into consideration.

[0006] Considering the above facts, the present invention aims to provide a fall monitoring control device, a fall monitoring system, a fall monitoring control method, and a fall monitoring control program that can use three-dimensional data to determine the fluctuation and size of a detected object, and to distinguish and detect a person who has fallen from an object that is not an object that triggers an alarm. [Means for solving the problem]

[0007] The fall monitoring control device according to the present invention includes an acquisition unit that acquires point cloud data generated based on distance and coordinate data of objects and space detected by a sensor; a detection unit that detects a specific object that exists within the detection range acquired by the acquisition unit; and a determination unit that determines whether the specific object is a falling object or an object other than a falling object, based on the detection maintenance state of the specific object detected by the detection unit.

[0008] The fall monitoring system according to the present invention comprises: a sensor that detects distance and coordinate data of objects and space from a detection area including the station platform and tracks; an acquisition unit that acquires point cloud data generated based on the distance and coordinate data of objects and space detected by the sensor; a detection unit that detects specific objects that exist within the detection range acquired by the acquisition unit; a determination unit that determines whether the specific object is a fall object or an object other than a fall object based on the detection maintenance state of the specific object detected by the detection unit; and a notification unit that notifies the determination result of the determination unit.

[0009] The fall monitoring and control method according to the present invention is characterized in that a sensor detects distance and coordinate data of objects and space from a detection area including the station platform and tracks, a fall monitoring and control device acquires point cloud data generated based on the distance and coordinate data of objects and space detected by the sensor, detects a specific object within the acquired detection range, determines whether the specific object is a fall object or an object other than a fall object based on the detection maintenance state of the detected specific object, and a notification unit notifies the determination result.

[0010] The fall monitoring and control program according to the present invention is characterized by operating a computer as the fall monitoring and control device described above. [Effects of the Invention]

[0011] As described above, the present invention uses three-dimensional data to determine the movement and size of the detected object, and to distinguish and detect the person who fell and other objects that trigger an alarm. [Brief explanation of the drawing]

[0012] [Figure 1] (A) is a control block diagram showing an outline of the fall monitoring control device according to this embodiment, and (B) is a control block diagram classifying the control of the fall monitoring control device by function. [Figure 2] (A) to (C) are point cloud data images obtained by analyzing images captured by a distance measuring device. (A) shows the monitoring area, (B) shows the area of ​​focus in the case of near-miss alarm detection, (C) shows the area of ​​focus in the case of fall detection, and (D) shows the progress of fall detection using a three-dimensional box to identify the distance and coordinates of objects within the area of ​​focus. [Figure 3] This is a control flowchart showing the object detection processing routine executed by the fall monitoring control device according to this embodiment. [Figure 4] This is a control flowchart showing the fall detection processing routine executed by the fall monitoring control device according to this embodiment. [Figure 5] This is a timing chart according to this embodiment, where (A) is a timing chart showing the flow until the detected object is determined to have fallen, and (B) is a timing chart showing the flow until the near alarm is determined. [Figure 6] These are timing charts related to modified cases, where (A) is a timing chart showing the flow until the detected object is determined to have fallen, and (B) is a timing chart showing the flow until the near alarm is determined. [Modes for carrying out the invention]

[0013] Figure 1(A) is a control block diagram showing an outline of the fall monitoring control device 10 according to this embodiment. The fall monitoring control device 10 includes a microcomputer 12. The microcomputer 12 has a CPU (Central Processing Unit) 12A, RAM (Random Access Memory) 12B, ROM (Read Only Memory) 12C, input / output devices (I / O) 12D, and buses 12E such as a data bus and a control bus that connect these.

[0014] A hard disk 14 is connected to I / O12D. Additionally, a distance measuring device 18 is connected to I / O12D via I / F (interface) 16, an emergency notification device 22 via I / F20, and a train signal management device 26 via I / F24.

[0015] (Risting device 18) In this embodiment, a 3D-LiDAR (3D-Light Detection and Ranging) is used as the distance measuring device 18. LiDAR can be either 2D-LiDAR or 3D-LiDAR.

[0016] 2D LiDAR acquires distance information in a plane by rotating a distance-measuring light source with a motor, thereby scanning the light horizontally. In addition to plane distance information, 3D LiDAR acquires height distance information by adding vertical scanning using multiple light sources arranged vertically, mirrors, and various light bending techniques.

[0017] In this embodiment, a 3D-LiDAR is used as the distance measuring device 18 for explanation purposes, but instead of a 3D-LiDAR, multiple 2D-LiDARs or other sensors capable of acquiring three-dimensional information may be used.

[0018] In this embodiment, point cloud data is generated based on the distance and shape of an object acquired by 3D-LiDAR and used for fall detection (object detection, object type identification, etc.).

[0019] Point cloud data is expressed as a collection of a huge number of points based on distance and coordinate data of objects and spaces acquired by 3D-LiDAR.

[0020] (Emergency notification device 22) The emergency notification device 22 is a device for notifying station staff, train drivers, conductors or the like of an emergency when a passenger falls from a station platform or an obstacle is found on a track. The notification form may be any of visual devices (indicator lamps, bulletin boards, etc.), auditory devices (speakers, bells, buzzers, etc.), and combinations thereof. When the emergency notification device 22 is activated, an emergency brake device that issues an alarm to all trains operating within a certain distance from the station and activates an emergency brake for the receiving train may be interlocked therewith. For example, different usage may be adopted such that the emergency notification device 22 is interlocked with the emergency brake device when a fall is determined as described later, and only the emergency notification device 22 is activated when a weed (near alarm) is determined.

[0021] (Signal management device 26) The signal management device 26 is a device that manages operations of home signals, starting signals, block signals, shunting signals, and governing signals (operation signal lines related to train operation).

[0022] In the present embodiment, information (signal) of a train that has passed through a yard exit from the signal management device 26 is used.

[0023] In the following description of the fall monitoring control device 10 of the present embodiment, the fall monitoring area is described as being fixed; however, for example, the fall monitoring area can be changed depending on whether a train enters within a predetermined time or not. In addition, it is also possible to grasp train operation status information using the signal management device 26, and control the timing of fall monitoring, the degree of strengthening of the monitoring system, and the like.

[0024] (Control functions of fall monitoring control device 10) Figure 1(B) is a control block diagram classifying the control of the fall monitoring control device 10 by function. The function of each block is operated by the microcomputer shown in Figure 1(A) based on a software program. IC chips (semiconductor integrated circuits) such as ASICs (Application Specific Integrated Circuits), FPGAs (Field Programmable Gate Arrays), and CPLDs (Complex Programmable Logic Devices) may be used to operate some or all of the functions of each block.

[0025] The distance measuring device 18 is connected to the data acquisition unit 50. The data acquisition unit 50 acquires the three-dimensional data detected by the distance measuring device 18 and sends it to the point cloud data generation unit 52. The point cloud data generation unit 52 generates a vast collection of points based on the three-dimensional data (distance and coordinate data of objects and space). Figure 2(A) is presented as a normal image, but it is point cloud data generated by the point cloud data generation unit.

[0026] On the other hand, although not essential, in this embodiment, the train operation signal line is received from the signal management device 26 by the signal receiving unit 54 and sent to the detection area setting unit 56. The detection area setting unit 56 grasps the train operation status based on the operation signal line, sets the fall detection area, and sends it to the point cloud data generation unit 52. As a result, the point cloud data generation unit 52 generates point cloud data limited to a detection area narrower than the area measured by the distance measuring device 18, thereby improving processing speed.

[0027] The point cloud data generation unit 52 is connected to the background data creation unit 58 and the object presence / absence determination unit 60. The background data creation unit 58 creates background data (still images) to be used as a criterion when determining whether or not an object has been displaced, and stores them in the background data storage unit 62.

[0028] The object presence / absence determination unit 60 calculates the difference between the point cloud data acquired in real time and the background data stored in the background data storage unit 62, and determines the presence or absence of an object based on the amount of displacement between data frames. A data frame is an image in single frames for which difference determination is performed, and the frame rate of one frame is set to 0.1 seconds.

[0029] Furthermore, it is preferable to update the background data periodically or irregularly so as to keep pace with changes in the surrounding environment.

[0030] The object presence / absence determination unit 60 is connected to the object tracking unit 64. When the object presence / absence determination unit 60 determines that an object is present, it outputs an activation signal to the object tracking unit 64. Upon receiving the activation signal, the object tracking unit 64 determines that the object may be a person who has fallen, and therefore performs a fall possibility determination (see Figures 2(B) and (C)).

[0031] The fall probability determination is performed based on the three-dimensional box 74 (so-called bounding box), which is the actual measurement result shown by the solid line in Figures 2(B), (C), and (D), and the threshold value stored in the fall determination threshold memory unit 66. The threshold value is represented by the three-dimensional box 76 with a dotted line frame, as shown in Figures 2(B), (C), and (D).

[0032] The threshold, or reference three-dimensional box 76, is based on the average height of a child old enough to ride a train alone, and corresponds to the size of a child of this average height when sitting in a so-called "gym class" position (let's call this the reference value X). (For example, the reference value X is a cube with a length, width, and depth of 60 cm each.) In this embodiment, a cube is used as an example under the above definition, but other shapes of thresholds based on other definitions may be used as long as they can detect a person falling.

[0033] The object to be determined tracking unit 64 is connected to the tracking result analysis unit 68. When the object to be determined tracking unit 64 determines that there is a possibility of falling, it outputs an activation signal to the tracking result analysis unit 68. Upon receiving the activation signal, the tracking result analysis unit 68 performs a detailed analysis of the object that may fall. The analysis, as will be described in detail later using Figure 5, etc., involves obtaining information by monitoring the displacement of the object that may fall over a predetermined number of frames N. The information analyzed by the tracking result analysis unit 68 is sent to the fall / near alarm determination unit 70.

[0034] The fall / near-miss alarm determination unit 70 performs either a fall determination, which determines that an object that has the potential to fall is a person who has fallen, or a near-miss alarm determination, which determines that it is the result of weeds growing beside the tracks being blown by the wind or the like.

[0035] The fall / near-miss alarm determination unit 70 sends a signal to the emergency signal output unit 72, which instructs the emergency alarm device 22 to send different alert messages depending on whether a fall has been detected or a near-miss alarm has been detected.

[0036] The emergency signal output unit 72 is instructed to output a signal indicating maximum urgency in the event of a fall detection. On the other hand, in the event of a near alarm detection, even if the detection result is not a fall, it is instructed to output a signal at a less urgent level than maximum urgency, as it is a near alarm that also includes the possibility of a false detection (see Figure 2(D)).

[0037] The operation of this embodiment will be explained below with reference to the flowcharts in Figures 3 and 4.

[0038] (Object detection process) Figure 3 is a control flowchart showing the object detection processing routine executed by the fall monitoring control device 10 according to this embodiment.

[0039] In step 100, it is determined whether or not it is time to update the background data. If the determination is positive, the process moves to step 102 to update the background data and then returns to step 100. If the determination in step 100 is negative, it is not time to update the background data, so the process moves to step 104.

[0040] In step 104, image processing (generation of point cloud data) is performed, and the process proceeds to step 108.

[0041] In step 108, difference data between frames of the point cloud image is obtained, and the process proceeds to step 110 to determine whether or not it is an object based on the difference. If the determination in step 110 is affirmative (object determined), the process proceeds to step 112, where the object is identified as a tracking target, its object ID is obtained, stored on the hard disk 14, and the process returns to step 100, repeating the above steps. If the determination in step 110 is negative, it is not an object, so the process returns to step 100, repeating the above steps.

[0042] (Fall detection process) Figure 4 is a control flowchart showing the fall detection processing routine executed by the fall monitoring control device 10 according to this embodiment. The objects to be judged for falling are those assigned an object ID in Figure 3.

[0043] In step 150, the flags and counts are reset, and the process moves to step 152. In step 152, point cloud data is generated from the image captured by the rangefinder 18, and then the process moves to step 154 ​​to determine whether the length, width, or depth of the target object exceeds the reference value X.

[0044] If the result in step 154 ​​is positive, proceed to step 156, where, as part of the flag and counter processing, set the fall possibility flag F (1), increment the fall counter Ct (Ct←Ct+1), and increment the near alarm counter Cz1 (Cz1←Cz1+1), then proceed to step 162. If the result in step 154 ​​is negative (below the reference value X), proceed to step 158.

[0045] Step 158 determines whether the fall possibility flag F is set (1). If the result in step 158 is negative, proceed to step 152. If the result in step 158 is positive, proceed to step 160, where the near alarm counter Cz2 is incremented (Cz2←Cz2+1) as part of the counter processing, and proceed to step 162.

[0046] Step 162 determines whether the count has reached a default value (for example, Ct=N or Cz1+Cz2=1). If the result in step 162 is negative, the process returns to step 152 and the above steps are repeated. If the result in step 162 is positive, the process proceeds to step 164 to determine whether the fall condition has been met. The fall condition is Ct=N (the reference value X is exceeded in all N frames). If the result in step 164 is positive, the process proceeds to step 166 to perform a fall detection process, which involves issuing a fall notification. Specifically, the emergency notification device 22 is activated to notify staff that an object (for example, a person) has fallen, and the process proceeds to step 172. Alternatively, an alarm may be issued to all trains operating within a certain distance from the station, and an emergency braking device may be activated to apply the emergency brakes to the receiving trains.

[0047] Furthermore, if a negative result is obtained in step 164, the process proceeds to step 168 to determine whether the weed condition is met. The near alarm condition is Cz1>1 (the reference value X was exceeded at least once during N frames), and if a positive result is obtained in step 168, the process proceeds to step 170 to perform a fall notification as the near alarm determination process. Specifically, the emergency alarm device 22 is activated to notify staff, etc., of a near alarm indicating that an object is present, although it may be a weed, and the process proceeds to step 172. Also, if a negative result is obtained in step 168, it is determined that the reference value X was not exceeded even once during N frames, and since the object is not subject to fall determination or near alarm determination, the process proceeds to step 172.

[0048] In step 172, the object ID whose determination has been completed is deleted, and the process moves to step 174. In step 174, it is determined whether or not the object ID to be determined exists. If the determination is positive, the process returns to step 150 and the above steps are repeated. If the determination in step 174 is negative, this routine terminates.

[0049] While it is preferable that the flowcharts in Figure 3 and Figure 4 be processed in parallel, depending on the clock speed and number of cores of the CPU 12A, for example, each step may be processed using time-sharing, or the flowchart in Figure 3 may be executed once for every frame during the monitoring process of N frames in the flowchart in Figure 4.

[0050] (Example of timing chart for object detection and fall detection based on this embodiment) Figure 5 is a timing chart showing the process from when a detected object is determined to have fallen (see Figure 5(A)) to when a near-miss alarm is triggered (see Figure 5(B)). The horizontal axis represents time, and the vertical axis represents the size of the three-dimensional box. The bar graph shows the size of the three-dimensional box (the maximum value of any of the height, width, or depth) for each captured frame (e.g., 0.1 sec). A line representing the reference value X is set on the vertical axis as a threshold.

[0051] As shown in Figure 5(A), once processing begins, the difference in point cloud data is calculated, and at predetermined frames, the object is detected by comparing the difference with the background (the three-dimensional box 74, which is the actual measurement result of the object, with the three-dimensional box 76, which is a threshold).

[0052] Subsequently, the three-dimensional box 74 is tracked for three frames. If any of the length, width, or depth of the three-dimensional box 74 exceeds a reference value X, it is determined to be an object that may fall, and the process proceeds to fall detection.

[0053] In the fall detection process, if one or more frames exceeding a predetermined threshold (reference value X) are detected within a predetermined number of detection frames N set in advance, it is determined to be a weed (near alarm). If all of the predetermined number of detection frames exceed the predetermined threshold (reference value X), it is determined that there is a fallen object.

[0054] In other words, the maximum values ​​of the height, width, and depth of the three-dimensional box 74 are obtained for N frames (if one frame is 0.1 seconds, then 0.1N seconds) (a bar graph for N frames). When the final frame (the Nth frame) is reached, all bars in the bar graph (all N frames) exceed the baseline value X, and a fall is determined.

[0055] As shown in Figure 5(B), once processing begins, the difference in point cloud data is calculated, and at predetermined frames, the object is detected by comparing the difference with the background (the three-dimensional box 74, which is the actual measurement result of the object, with the three-dimensional box 76, which is a threshold).

[0056] Subsequently, the three-dimensional box 74 is tracked for a predetermined number of frames. If any of the length, width, or depth of the three-dimensional box 74 exceeds a reference value X, it is determined to be an object that may fall, and the process proceeds to fall detection.

[0057] In the fall detection process, the maximum values ​​of the height, width, and depth of the three-dimensional box 74 are obtained over N frames (0.1Nsec if one frame is 0.1sec) (bar graph for N frames).

[0058] When the final frame (the Nth frame) is reached, three of the bar graphs for the N frames exceed the threshold value X, satisfying the condition Cz1>1 in step 168 of Figure 4, and a near-miss alarm is triggered.

[0059] In this embodiment, the probability of exceeding the baseline value X within N frames is 0% for no fall and 100% for a confirmed fall. In this embodiment, a fall is judged when the probability is 100%. Here, the threshold for determining a fall is such that a higher probability indicates higher judgment accuracy, while a lower probability indicates higher safety; these two are inversely related. Therefore, it is preferable to change the judgment criteria depending on various conditions. Examples of conditions include the station structure, environment, weather, and fall statistics.

[0060] As described above, according to this embodiment, point cloud data is generated from three-dimensional image data captured by the 3D-LiDAR distance measuring device 18, and an object is detected from the difference between frames. For the detected object, it is determined whether the length, width, or depth of the three-dimensional box 74 exceeds a threshold value X for multiple frames (N frames), and based on the result, a fall detection (e.g., threshold value X exceeds in all frames) or a near-alarm detection (e.g., threshold value X exceeds in any frame) is performed. In addition to fall detection, a near-alarm detection (including weeds) is performed when the reliability of fall detection is low. This increases the reliability of fall detection and also enables the issuance of a near-alarm.

[0061] (Example of a timing chart for object detection and fall detection based on a modified version) In this embodiment, the fall detection process always concludes that a fall has occurred, a near-miss alarm has been triggered, and no occurrences have occurred by the Nth frame. However, as a remedy, if unnatural behavior occurs in the Nth frame, the detection (remedy) may be performed in the next frame (from the N+1th frame onward).

[0062] Figure 6 is a timing chart showing the process in the rescue procedure, from when the detected object is determined to have fallen (see Figure 6(A)) to when a near-alarm is detected (see Figure 6(B)).

[0063] Up to the Nth frame, the flow is the same as the timing chart according to this embodiment (see Figure 5).

[0064] As shown in Figure 6(A), the fall detection process obtains the maximum values ​​of the height, width, and depth of the three-dimensional box 74 over N frames (0.1Nsec if one frame is 0.1sec) (bar graph for N frames).

[0065] Upon reaching the final frame (frame N), the bar graph shows that the bar values ​​for the other N-1 frames, excluding the final frame (frame N), exceed the baseline value X. In this case, the condition (Ct=N) in step 164 of the flowchart in Figure 4 is not met.

[0066] Therefore, as a remedy, the maximum values ​​of the height, width, and depth of the three-dimensional box 74 in the N+1th frame are obtained, and if the bar graph in that N+1th frame exceeds the reference value X, it is determined to have fallen.

[0067] As shown in Figure 6(B), the fall detection process obtains the maximum values ​​of the height, width, and depth of the three-dimensional box 74 over N frames (0.1Nsec if one frame is 0.1sec) (bar graph for N frames).

[0068] Upon reaching the final frame (frame N), the bar graph shows that the bar in the final frame (frame N) exceeds the baseline value X. In this case, according to the flowchart in Figure 4, the condition in step 164 (Ct=N) is not met, and the condition in step 168 (Cz1>1) is met. However, if the baseline value X is exceeded in the final frame (frame N), there is a possibility that it will continue to exceed the baseline value X in subsequent frames.

[0069] Therefore, as a remedy, the maximum values ​​of the height, width, and depth of the three-dimensional box 74 in the 16th frame are obtained, and if the bar graph in the N+1th frame is less than or equal to the reference value X, a near alarm is triggered. Furthermore, if the bar graph in the N+1th frame exceeds the reference value X, it is preferable to avoid the judgment and re-execute the judgment for N frames.

[0070] In the relief measures, the conclusion may be delayed not only based on the result of the final frame (N frames), but also on the progress of the entire N frames. However, limitations are necessary to prevent delays in judgment. These limitations may be the number of frames delayed or the number of retries. [Explanation of symbols]

[0071] 10. Fall monitoring control device 12 Microcomputers 12A CPU 12B RAM 12C ROM 12D input / output device (I / O) 12E Bus 14 Hard disk 16 I / F 18. Distance measuring device (sensor) 20 I / F 22 Emergency alarm system 24 I / F 26 Signal management device 50 Data acquisition unit (acquisition unit) 52 Point Cloud Data Generation Unit 54 Signal receiving section 56 Detection Area Setting Unit 58 Background Data Creation Section 60. Object presence / absence determination unit (detection unit) 62 Background data storage unit 64. Target Tracking Unit 66 Fall detection threshold storage unit 68 Tracking Results Analysis Department 70 Fall / Near Alarm Judgment Unit (Judgment Unit) 72 Emergency signal output section 74 Three-dimensional box (distance) 76 Three-dimensional box (threshold)

Claims

1. An acquisition unit that acquires point cloud data generated based on coordinate data indicating an object detected by a sensor, the spatial distance to the object, and the position of the object, A detection unit that detects a specific object located within a detection range acquired based on the point cloud data acquired by the acquisition unit, A determination unit determines whether the specified object is a fallen object or an object other than a fallen object, based on the detection maintenance state of the specified object detected by the detection unit. A fall monitoring and control device having the following features.

2. The fall monitoring control device according to claim 1, wherein the detection unit detects a specific object based on the difference between background point cloud data acquired as the background of the detection range and point cloud data of the detection range acquired by the acquisition unit.

3. The specific object detected by the detection unit is represented as a three-dimensional box demarcated by the smallest boundary, In the determination unit, the detection maintenance state is the amount of displacement of the size of the three-dimensional box and the duration of the displacement. The fall monitoring and control device according to claim 1, which determines whether the specified object is the falling object or the object to be determined, based on the amount of displacement of the three-dimensional box and the duration of the displacement.

4. In the determination unit, the detection maintenance state is the distance to the specific object detected by the detection unit and the duration for which that distance is maintained. If the distance is greater than or equal to the threshold, or if the duration of maintaining the distance satisfies the conditions, the specified object is determined to be the fallen object. The fall monitoring control device according to claim 1, wherein the specified object is determined to be the object to be determined when the distance is less than a threshold or when the duration of maintaining the distance does not meet the conditions.

5. The fall monitoring control device according to claim 1, wherein the sensor is a standalone 3D-LiDAR.

6. The fall monitoring control device according to claim 1, further comprising a notification unit that notifies in different notification forms depending on whether the determination unit determines that the specified object is a fallen object or the object to be determined.

7. A sensor that detects an object, the spatial distance to the object, and coordinate data indicating the position of the object from a detection area including the station platform and tracks, A fall monitoring and control device comprising: an acquisition unit that acquires point cloud data generated based on an object detected by the sensor, the spatial distance to the object, and coordinate data indicating the position of the object; a detection unit that detects a specific object located within a detection range acquired based on the point cloud data acquired by the acquisition unit; and a determination unit that determines whether the specific object is a falling object or an object other than a falling object based on the detection maintenance state of the specific object detected by the detection unit, A notification unit that notifies the determination result of the determination unit, A fall monitoring system having the following features.

8. The sensor detects an object, the spatial distance to the object, and coordinate data indicating the position of the object from a detection area including the station platform and tracks. The fall monitoring control device acquires point cloud data generated based on the object detected by the sensor, the spatial distance to the object, and coordinate data indicating the position of the object. Based on the acquired point cloud data, it detects a specific object that exists within the acquired detection range, and based on the detection maintenance state of the detected specific object, it determines whether the specific object is a falling object or an object other than a falling object. The news department will announce the results of the judgment. A method for monitoring and controlling falls.

9. Computers, To be operated as a fall monitoring control device according to any one of claims 1 to 6, Fall monitoring and control program.

Images