Monitoring area setting device and monitoring area setting method

Abstract

To provide a monitoring area setting device that eliminates the need for cumbersome work of moving a fence for measurement on site, or of repeatedly entering numerical values while determining the appropriateness of a monitoring area using a jig, even when the installation accuracy of a safety fence has some dispersion.SOLUTION: A monitoring area setting device comprises: a point cloud data acquisition unit for acquiring point cloud data that shows movable fences and floor surfaces acquired by LiDAR; a surface acquisition unit for acquiring multiple surfaces based on the point cloud data; a labeling unit for distinguishing the movable fence from the floor surface by labeling the acquired multiple surfaces under a predetermined condition; and a monitoring area setting unit for setting a monitoring area for detecting an obstacle on the basis of the multiple surfaces including the labeled movable fence and floor surface. The monitoring area setting unit includes: matching a rising angle of each door surface with respect to the floor surface on the basis of a normal vector of the multiple door surfaces labeled by the labeling unit and a reference vector set on the floor surface; and setting the monitoring area on the basis of each matched door surface.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a monitoring area setting device and a monitoring area setting method. [Background technology]

[0002] Patent document 1 proposes a system which has a three-dimensional distance image sensor and generates a detection area (monitoring area) of an obstacle detection device which is installed at the platform doors of a platform where a train is stopped, the system comprising: the three-dimensional distance image sensor which is placed on the obstacle detection device and capable of acquiring a three-dimensional distance image which includes the boundary of the detection area; a memory unit which stores the three-dimensional distance image; and a calculation unit which generates the detection area based on the three-dimensional distance image stored in the memory unit.

[0003] Patent Document 2 proposes a detection area setting device that includes a point cloud data acquisition unit that acquires three-dimensional point cloud data indicating the coordinates of distance measurement points including reflection points on the surface of a reference object based on reflected light of measurement light irradiated onto an area including a reference object to be set as a detection area (monitoring area), a surface acquisition unit that acquires multiple surfaces including the surface of the reference object based on the point cloud data acquired by the point cloud data acquisition unit, a labeling unit that labels the multiple surfaces acquired by the surface acquisition unit under predetermined conditions, and an area setting unit that sets the detection area for detecting obstacles based on the multiple surfaces labeled by the labeling unit, wherein the surface acquisition unit executes a process for multiple attention points to acquire normals to attention surfaces generated at an attention point and a point near the attention point among the multiple distance measurement points included in the point cloud data as normal vectors at the attention points, and executes a surface acquisition process to acquire multiple surfaces including any of the attention points based on each normal vector. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2017-19380 A [Patent Document 2] JP 2022-187965 A Summary of the Invention [Problem to be solved by the invention]

[0005] The system for generating the detection area of ​​the obstacle detection device disclosed in Patent Document 1 requires that while a vehicle is running on the track, a three-dimensional distance image is acquired that includes the boundary on the track side, or a vehicle fixture that simulates the contour of the vehicle is run on the track to acquire a three-dimensional distance image that includes the boundary on the track side, and that the door panels of the platform doors are opened and closed to acquire a three-dimensional distance image that includes the boundary of the detection area on the platform door side, and there was a problem in that it was extremely cumbersome and time-consuming to acquire three-dimensional distance images individually for the many platform doors installed on the platform.

[0006] In the detection area setting device disclosed in Patent Document 2, the distance along the longitudinal direction of the platform along the track and the distance in the depth direction from the edge of the platform to the safety fence (platform doors) are input as numerical values ​​based on design information, greatly reducing the effort required to set the detection area.

[0007] However, many safety fences installed on platforms are retrofitted on existing platforms, so on-site alignment is required during construction, and installation accuracy is not very high. Therefore, the detection area, which is approximately a rectangular parallelepiped and generated based on numerical values ​​input based on design information, does not match the range of movement of the safety fence, especially the movable fence, and there is a risk that the movable fence will be erroneously detected as an obstacle, and if the detection area is made small, there is a risk of a wide blind spot being created. Therefore, it is necessary to repeatedly input numerical values ​​while judging the appropriateness of the detection area using a jig on-site, and as with Patent Document 1, a problem of requiring a very complicated effort has arisen.

[0008] For example, in the case of safety fences for the leading and trailing cars, the space between the edge of the platform and the safety fence may be set wider than others to ensure the space required for crew members to board and disembark from the vehicle. As a result, the ends (door tips) of the left and right movable fences are joined at an angle. Even for safety fences for cars in the middle, there is variation in the levelness of the platform, and the edge of the platform may be curved rather than straight, so even if the safety fence is installed according to the design values, the ends (door tips) of the left and right movable fences may not match up properly, resulting in misalignment. In such cases, it is necessary to make adjustments on site, which may result in the movable fence being tilted relative to the platform surface (floor surface).

[0009] The object of the present invention is to provide a monitoring area setting device and a monitoring area setting method that eliminate the need for the tedious task of measuring while operating the fence on-site, or repeatedly inputting numerical values ​​while using a jig to determine the appropriateness of the monitoring area (detection area), even if there is variation in the installation accuracy of the safety fence. [Means for solving the problem]

[0010] In order to achieve the above-mentioned object, a first characteristic configuration of a surveillance area setting device according to the present invention includes a point cloud data acquisition unit that scans a surveillance space in which a reference object is present, including a movable fence in a closed state that restricts entry to a boarding and disembarking area and a floor surface on which the movable fence is installed, using a LiDAR, and acquires point cloud data indicating a surface of an object including the reference object from an obtained distance image; a surface acquisition unit that executes a process for each of the points of interest to acquire, as a normal vector at each of the points of interest, a normal vector to a surface of interest generated at a predetermined point of interest and points adjacent to each of the points of interest from the point cloud data acquired by the point cloud data acquisition unit, and executes a surface acquisition process to acquire multiple surfaces including any of the points of interest based on the degree of coincidence of each of the acquired normal vectors; The movable fence has a labeling unit that identifies the floor surface from the multiple surfaces based on a predetermined reference vector and normal vectors of the multiple surfaces acquired by the surface acquisition unit, and identifies and labels door surfaces corresponding to the closed positions of the multiple doors that constitute the movable fence, and a monitoring area setting unit that sets a monitoring area for detecting whether or not an obstacle is present in the boarding and disembarking area based on the multiple surfaces labeled by the labeling unit, wherein the monitoring area setting unit is configured to align the rise angles of each door surface with respect to the floor surface based on the normal vectors of the multiple door surfaces labeled by the labeling unit and the reference vector, and to set the monitoring area based on each aligned door surface.

[0011] In the point cloud data acquisition unit, the monitored space in which the reference object, including the closed movable fence and the floor surface, exists is scanned by the LiDAR, and point cloud data indicating the surface of the object, including the movable fence and the floor surface, existing in the monitored space is obtained. In the surface acquisition unit, multiple surfaces constituting the object existing in the monitored space are generated from multiple point cloud data. In the labeling unit, surfaces belonging to the same object, specifically, the floor surface and multiple door surfaces constituting the movable fence, are grouped based on the relationship between the normal vectors of the multiple surfaces and a preset reference vector. In the monitoring area setting unit, a monitoring area for detecting whether or not an obstacle exists in the boarding and disembarking area is set based on the grouped multiple surfaces.

[0012] In particular, the monitoring area setting unit aligns the rise angle of each door surface relative to the floor surface based on the normal vectors and reference vector of the multiple labeled door surfaces.As a result, even if the installation posture of each door surface is actually tilted during on-site adjustment, the monitoring area is set based on door surfaces in which the discrepancy in the tilt of the multiple constituent surfaces has been eliminated, eliminating the need for the complicated task of on-site adjustment.

[0013] The second characteristic configuration is that, in addition to the first characteristic configuration described above, the labeling unit identifies each door surface based on position information corresponding to the closed positions of the multiple doors that is input in advance.

[0014] For the multiple doors that make up the movable fence, the position of the joints of each door can be determined from position information indicating the closed position, and each door surface can be identified using the joints as boundaries.

[0015] The third characteristic configuration is that, in addition to the first characteristic configuration described above, the labeling unit identifies each door surface based on the relative bias of the multiple surfaces acquired by the surface acquisition unit.

[0016] Each door surface can be identified based on the relative deviation of each door surface obtained for the multiple doors that make up the movable fence, that is, the degree of relative inclination or deviation.

[0017] The fourth characteristic configuration is that, in addition to any of the first to third characteristic configurations described above, the labeling unit acquires a boundary line of the floor surface based on the point cloud data acquired by the point cloud data acquisition unit, generates an intersecting plane that passes through the boundary line and intersects with the floor surface, and identifies and labels the floor surface and the intersecting plane.

[0018] In the labeling section, a surface whose normal vector has a predetermined relationship with a preset reference vector, for example, a floor surface (platform), is identified. Then, based on point cloud data showing the floor surface, a boundary line (edge ​​on the track side, end) of the floor surface is acquired, and an intersecting surface (corresponding to the side of the vehicle) that passes through the boundary line and intersects with the floor surface is generated. Furthermore, the floor surface and the intersecting surface that intersects with the floor surface are labeled so as to be identifiable as two surfaces that define the monitoring area.

[0019] The fifth characteristic configuration is that, in addition to the fourth characteristic configuration described above, the surface acquisition process includes a clustering process in which, if the normal vectors of other points of interest that exist within a predetermined search range from a specific point of interest are within a predetermined allowable range of inclination with respect to the direction of the normal vector of the specific point of interest, the other points of interest are grouped together with the specific point of interest.

[0020] In the clustering process, other points of interest that exist within a specified search range are extracted based on a specific point of interest, and when the inclination of the normal vector of the other points of interest relative to the normal vector of the specific point of interest falls within a specified allowable range, the other points of interest are grouped into the same group as the specific point of interest as candidate points that constitute the same surface as the surface that contains the specific point of interest. The search range is a value that is appropriately set according to the reference object to be measured. Taking the example of a case where the reference object to be measured is a movable fence, the search range is set to a value that excludes points of interest in an area far away from the surface of the movable fence, while causing multiple points of interest corresponding to the surfaces on both sides of the transparent window of the movable fence to belong to the same group.

[0021] The sixth characteristic configuration is that, in addition to the fifth characteristic configuration described above, the surface acquisition process repeats the clustering process with each focus point as the specific focus point, and if any of the focus points to be grouped in a later clustering process have already been grouped in a previous clustering process, the surface acquisition process includes an integration process that integrates all of the focus points to be grouped in the later clustering process into the same group as the previous clustering process.

[0022] When the clustering process is repeated with each attention point as a specific attention point, a cluster including other attention points existing on the same surface is formed for each specific attention point. When any of the attention points constituting each cluster belongs to any of the previously formed clusters, the points are integrated into the same cluster as the previously formed cluster, thereby increasing the number of attention points that are candidates to be included in the same surface. In other words, the surface generated in the specified search range is expanded beyond the specified search range.

[0023] The seventh characteristic configuration is that, in addition to the sixth characteristic configuration described above, the surface acquisition process includes an integration process in which a representative normal vector representing each of the multiple groups generated by the clustering process is obtained, and groups whose respective representative normal vectors fall within a predetermined tolerance range are integrated into the same group.

[0024] The grouped attention points are a set of a plurality of attention points that exist in a predetermined search range based on at least a specific attention point. If the search range is set large, attention points that originally belong to different faces may belong to the same group, so the search range must be limited to a certain size. However, if the search range is set small, attention points that originally belong to the same face may belong to different groups. Therefore, in this configuration, when the representative normal vectors of each group fall within a predetermined allowable range, the groups are integrated into the same group. The predetermined allowable range can be defined, for example, by the direction of the representative normal vector, the distance between the starting points of the representative normal vectors relative to a predetermined reference plane, or the like.

[0025] The eighth characteristic configuration of the same is, in addition to the seventh characteristic configuration described above, that the surface acquisition process performs a filtering process on the points of interest grouped into the same group to remove outliers from the same group.

[0026] The point cloud data grouped into the same group by the clustering process is a collection of attention points whose distance between each ranging point and the inclination of the normal direction are within a predetermined allowable range, and the distance of each attention point from the normal direction of a specific attention point or the tangent plane of a specific attention point is not evaluated. Therefore, by performing a filtering process, outliers that cannot be evaluated as existing on the same plane are appropriately removed.

[0027] A first characteristic configuration of the surveillance area setting method according to the present invention includes a point cloud data acquisition step of scanning a surveillance area in which a reference object exists, including a movable fence in a closed state that restricts entry to a boarding / disembarking area and a floor surface on which the movable fence is installed, by using a LiDAR, and acquiring point cloud data indicating a surface of an object including the reference object from an obtained distance image; a surface acquisition step of executing a process for each of the points of interest to acquire, as a normal vector at each of the points of interest, a normal to a surface of interest generated at a predetermined point of interest and a point adjacent to each of the points of interest from the point cloud data acquired in the point cloud data acquisition step, and acquiring a plurality of surfaces including any of the points of interest based on a degree of coincidence of each of the acquired normal vectors; the surface acquisition unit acquires normal vectors of the plurality of surfaces acquired by the surface acquisition unit, and identifies and labels the floor surface from the plurality of surfaces and the door surfaces corresponding to the closed positions of the plurality of doors constituting the movable fence; and a monitoring area setting step, based on the plurality of surfaces labeled by the labeling step, sets a monitoring area for detecting whether or not an obstacle is present in the boarding and disembarking area, wherein the monitoring area setting step is configured to align the rise angles of each door surface with respect to the floor surface based on the normal vectors of the plurality of door surfaces labeled by the labeling step and the reference vector, and to set the monitoring area based on each aligned door surface.

[0028] The second characteristic configuration is that, in addition to the first characteristic configuration described above, the labeling step identifies each door surface based on position information corresponding to the closed positions of the multiple doors that is input in advance.

[0029] The third characteristic configuration is that, in addition to the first characteristic configuration described above, the labeling step identifies each door surface based on the relative bias of the multiple surfaces acquired in the surface acquisition step.

[0030] The fourth characteristic configuration is that, in addition to any of the first to third characteristic configurations described above, the labeling step acquires a boundary line of the floor surface based on the point cloud data acquired in the point cloud data acquisition step, generates an intersecting plane that passes through the boundary line and intersects with the floor surface, and identifies and labels the floor surface and the intersecting plane that intersects with the floor surface. Effect of the Invention

[0031] As described above, according to the present invention, it is possible to provide a monitoring area setting device and a monitoring area setting method that eliminate the need for the tedious work of measuring while operating the fence on-site, or repeatedly inputting numerical values ​​while using a jig to determine the appropriateness of the monitoring area, even if there is variation in the installation accuracy of the safety fence. [Brief description of the drawings]

[0032] [Figure 1] (a) is an explanatory diagram of a front view of a safety fence installed on a platform, (b) is an explanatory diagram of a plan view showing the positional relationship between the safety fence installed on the platform and the track, and (c) is an explanatory diagram of a plan view of the safety fence installed at an angle. [Diagram 2] (a) is an explanatory diagram of the functional blocks of the obstacle detection system, and (b) is an explanatory diagram of the LIDAR. [Diagram 3] FIG. 1A is an explanatory diagram of a monitoring area, and FIG. 1B is an explanatory diagram showing the posture of a door that has been labeled before the monitoring area is set. [Figure 4] FIG. 4A is an explanatory diagram of a monitoring area setting device (setting unit), and FIG. 4B is a flowchart showing a procedure for a monitoring area setting process. [Diagram 5] 13 is a flowchart showing the procedure of a surface acquisition process. [Figure 6]FIG. 11 is an explanatory diagram of a surface obtained by the labeling process. [Figure 7] 13 is a flowchart showing a procedure of a labeling process. [Figure 8] 13 is a flowchart showing a procedure for a monitoring area setting process. [Figure 9] 10A and 10B are explanatory diagrams of a monitoring area setting process according to another embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0033] An example of a monitoring area setting device and a monitoring area setting method according to the present invention will be described below.

[0034] [Surveillance area set up on train platforms] As shown in Fig. 1(a), (b), and (c), a safety fence SF is installed on the edge of the track (railroad track) T side of a platform P installed in a railway station building to prevent accidents such as passengers coming into contact with a train. The safety fence SF is composed of a fence body SB (SB1, SB2) with a gate installed at a position corresponding to the door of the vehicle, and a door DR (DR1, DR2) which is a movable fence that opens and closes the gate. The fence body SB functions as a door pocket for each door DR. A side wall SW2 on which a LiDAR (Light Detection and Ranging) 10 is installed is protruding from one of the fence bodies SB2 on the vehicle side, and a side wall SW1 that defines the depth of the monitoring area DA is protruding from the other of the fence bodies SB on the vehicle side. The LiDAR 10 is a distance measuring device that irradiates a laser beam into a monitored space and measures the distance to the target object and the shape of the target object based on information on the reflected light.

[0035] The X, Y, and Z coordinate systems shown in each figure indicate three-dimensional coordinate systems with the installation position of one of the riders 10 as the origin. Fig. 1(a) is a front view of the safety fence SF seen from the track T side, and Figs. 1(b) and (c) are plan views showing the positional relationship between the platform P, the safety fence SF, and the track T. In Fig. 1(a), the hatched areas of the doors DR (DR1, DR2) are transparent acrylic resin plates DR10, DR20 provided to reduce the weight of the doors DR.

[0036] Safety fences SF, which are placed in correspondence with train doors, are often retrofitted onto existing platforms P, and if the safety fences are installed according to design values, the ends (door tips) DC of the left and right doors DR may not match properly, resulting in horizontal misalignment or vertical tilt. For example, this is because there may be variations in the flatness of the platform P, or the edge of the platform may be curved rather than straight. In such cases, on-site adjustment is required, and as a result, the left and right doors DR may be tilted relative to the platform surface (floor surface), and the straightness of the left and right doors DR may not be maintained in a plan view.

[0037] Figure 1(b) shows a good construction result where the straightness of the left and right doors DR is maintained in a plan view, while figure (c) shows an example where the left and right doors DR intersect at an angle. For example, in order for crew members to board and disembark, the fence body SB (SB1) near the crew compartment of the first and last cars is not installed in a parallel position to the edge of the platform, but may be intentionally installed at an angle to the edge of the platform.

[0038] In such a case, in order to generate the monitoring area DA, it is very cumbersome to perform measurements while operating the door DR on-site, or to repeatedly input values ​​that define the monitoring area DA while using a tool to determine the appropriateness of the monitoring area DA. The monitoring area setting device and monitoring area setting method according to the present invention eliminate the need for such cumbersome work.

[0039] 2(b), the LIDAR 10 employs a TOF (Time of Flight) system and includes a cylindrical casing 11 with an optical window W, a light projecting and receiving unit 13 housed in the casing 11 and equipped with a laser light source made of a semiconductor laser and a light receiving element, a scanning optical system 14, and a signal processing circuit 12 for controlling them. Note that the LIDAR 10 is not limited to the TOF system, and may use other measurement systems such as an AM system that calculates distance based on the phase difference between measurement light and reflected light.

[0040] The scanning optical system 14 includes a deflection mirror M. The scanning optical system 14 also includes a first deflection mechanism 14A that oscillates the deflection mirror M about a first axis P1 that is perpendicular to the paper surface of FIG. 2(b). The first deflection mechanism 14A includes an electromagnetic oscillation mechanism. The scanning optical system 14 also includes a motor that rotates the second deflection mechanism 14B about a second axis P2 that is parallel to the paper surface of FIG. 2(b).

[0041] The scanning optical system 14 repeatedly scans the pulsed measurement light emitted from the laser light source in a horizontal direction hd (see FIG. 1(b)) within a predetermined range via the second deflection mechanism 14B, while repeatedly scanning the scanning line in the horizontal direction hd in a vertical direction vd (see FIG. 1(a)) via the first deflection mechanism 14A. The signal processing circuit 12 calculates the distance and direction to the reflection point of the measurement light based on the time difference between the emission timing of the measurement light and the detection timing of the reflected light. That is, the LIDAR 10 detects the presence or absence of an obstacle based on the reflected light from an object present in the space surrounding the vehicle door, which is the monitoring area DA.

[0042] In the present invention, an obstacle monitoring area DA is set in the space scanned by the measurement light so as to prevent the platform P, the door DR, etc. from being erroneously determined to be an obstacle based on the reflected light from the surface of the platform P, the door DR, etc.

[0043] As shown in Fig. 3(a), the monitoring area DA corresponding to Fig. 1(b) and (c) is an area surrounded by the floor surface of the platform P, the side surface VW perpendicularly intersecting the edge of the platform P, the depth which is the distance between the side walls SW1 and SW2, the offset surfaces P', VW', SW1', SW2', DR1', and DR2' which are spaced inwardly by a predetermined width from each of the door surfaces which are the side surfaces of the doors DR1 and DR2, and the top surface TS' at a predetermined height. The predetermined width is not particularly limited and can be set appropriately. The angle formed by the offset surfaces DR1' and DR2' approximately corresponds to the angle of the doors DR1 and DR2.

[0044] [Configuration of Obstacle Detection System] As shown in FIG. 2(a), the obstacle detection system 100 includes a monitoring area setting unit 20 that sets a monitoring area for determining the presence or absence of an obstacle, a range sensor that scans a monitoring area DA set by the monitoring area setting unit 20 with measurement light and determines the presence or absence of an obstacle based on the presence or absence of reflected light from the monitoring area DA, and a system control unit 50 that controls the operation timing of the monitoring area setting unit 20 and the range sensor.

[0045] The above-mentioned LIDAR 10, monitoring area storage unit 30, and obstacle detection unit 40 constitute a range sensor, and the monitoring area setting unit 20 constitutes a monitoring area setting device of the present invention. The monitoring area storage unit 30 stores and manages the monitoring area. The obstacle detection unit 40 determines the presence or absence of an obstacle in the monitoring area DA, and outputs obstacle presence / absence information. The monitoring area storage unit 30 and the obstacle detection unit 40 may be incorporated inside the housing of the LIDAR 10.

[0046] The signal processing circuit 12 includes a PLC (programmable logic controller) that controls the laser light source, the light receiving element, and the scanning optical system, a dedicated DSP (digital signal processor) that performs distance measurement calculations on the reflected light detected by the light receiving element, etc. In this embodiment, an FPGA (Field Programmable Gate Array) is used as the signal processing circuit 12.

[0047] The monitoring area storage unit 30 is composed of a semiconductor memory, and the obstacle detection unit 40 is composed of a semiconductor memory in which a microcomputer and an application program executed by the microcomputer are stored, an input / output circuit, a communication circuit, etc. The monitoring area setting unit 20 is configured to be able to communicate with the range measurement sensor, and is composed of a personal computer or a laptop computer in which an application program for setting the monitoring area is installed.

[0048] The system control unit 50 is composed of a general-purpose computer installed with application programs for operation management that perform processing for controlling the opening and closing of the door DR of the safety fence SF, processing for detecting vehicle stoppage, processing for detecting the open / closed state of the vehicle door, and processing for determining whether the vehicle can depart or not.

[0049] The system control unit 50 activates the monitoring area setting unit 20 to set a monitoring area DA, for example, at night when the vehicle is not in operation. When the monitoring area DA is set by the monitoring area setting unit 20, the monitoring area DA is stored in the monitoring area storage unit 30 provided in the range sensor.

[0050] The obstacle detection unit 40 automatically starts obstacle detection when the monitoring area DA is set in the monitoring area setting unit 20. The obstacle detection unit 40 detects whether or not an obstacle is present in the monitoring area DA stored in the monitoring area storage unit 30, and outputs the result to the system control unit 50. When the vehicle starts to operate after the monitoring area DA is set, the system control unit 50 receives an output signal from the obstacle detection unit 40. That is, the system control unit 50 drives the LIDAR 10 to scan the monitoring area DA, and determines whether or not an obstacle is present in the monitoring area DA based on the presence or absence of reflected light from the monitoring area DA.

[0051] Until the vehicle enters platform P, when the system control unit 50 receives a signal from the obstacle detection unit 40 indicating the presence of an obstacle, it outputs an alarm signal to a pre-prepared alarm system, and when it receives a signal from the obstacle detection unit 40 indicating that no obstacle is present, it stops the alarm signal output to the alarm system.

[0052] When the system control unit 50 detects that a train has entered and stopped at platform P based on the output from the train entry detection sensor, it opens the door DR of the safety fence SF and also opens the train door, and monitors the movement of passengers getting on and off in the monitoring area DA via the obstacle detection unit 40. An imaging device, a lidar, or the like can be used as the train entry detection sensor, and the function thereof can also be incorporated into the obstacle detection system 100.

[0053] When it is time for the vehicle to depart, the system control unit 50 judges whether or not the vehicle can depart. The judgment of whether or not the vehicle can depart is based on the criterion that there is no risk of a serious accident occurring due to the departure of the vehicle, such as whether the vehicle doors are closed, and whether or not the obstacle detection unit 40 has received a signal indicating that no passengers or baggage have been detected in the monitoring area DA.

[0054] If no passengers or baggage are detected in the monitoring area DA, the system control unit 50 closes the door DR of the safety fence SF and permits the vehicle to depart.

[0055] If a passenger's body, bag, or the like is caught in the vehicle door and passengers getting on or off are detected in the monitoring area DA, the system control unit 50 reopens the vehicle door. The system control unit 50 controls the vehicle to prohibit departure while keeping the door DR of the safety fence SF open.

[0056] In this embodiment, an example has been described in which the system control unit 50 automatically controls the running of the vehicle and the opening and closing of the vehicle doors. However, in cases where the running of the vehicle and the opening and closing of the vehicle doors need to be performed by the vehicle crew, the system control unit 50 only needs to detect the stopping of the vehicle and the opening and closing state of the vehicle doors using a dedicated sensor, and can be configured to output a departure prohibition signal or permission signal to the vehicle based on the presence or absence of an obstacle in the monitoring area DA.

[0057] [Configuration of monitoring area setting device] 4(a) and (b), the monitoring area setting unit 20 includes the functional blocks of a point cloud data acquisition unit 21, a surface acquisition unit 22, a labeling unit 23, and a monitoring area setting unit 24. These functional blocks are embodied by being executed by a personal computer or the like in which an application program for setting the above-mentioned monitoring area is installed.

[0058] The monitoring area setting unit 20 sets a monitoring area DA for each Rider 10 installed near the door DR of each gate. The monitoring area setting unit 20 stores each set monitoring area DA in the monitoring area storage unit 30 in association with an ID previously set for the Rider 10. Thereafter, the monitoring area setting unit 20 outputs a signal to the system control unit 50 indicating that setting of the monitoring area DA has been completed. Below, the setting process of the monitoring area for one Rider 10 will be described, but it goes without saying that a similar setting process is performed for each Rider 10.

[0059] The point cloud data acquisition unit 21 starts the LIDAR 10 to scan the monitored space where a reference object exists, including a movable fence in a closed state that restricts entry to the boarding and disembarking area and the floor surface on which the movable fence is installed, and acquires distance image data including distances and directions of distance measurement points including reflection points on the surface of the reference object based on the reflected light of the scanned measurement light.Then, the point cloud data acquisition unit 21 executes a point cloud data acquisition step of acquiring three-dimensional point cloud data, which are the three-dimensional coordinates of the distance measurement points, by arithmetic processing based on the distance image data (FIG. 4(b), step S1).

[0060] In this embodiment, the monitoring area DA is an area defined between a safety fence SF provided on a platform P of a station building and a train parked on the platform P. The "reference object to be measured for setting the monitoring area" refers to the platform P, which is the floor surface, and the door DR, which is the movable fence, and the "monitored space in which the reference object is present" refers to the space between the door DR and the train door, including the platform P and the door DR, when there are no obstacles such as passengers getting on and off or luggage. In the point cloud data acquisition step, the presence of obstacles in the space is permitted as long as they are minimal.

[0061] The surface acquisition unit 22 executes a surface acquisition step to acquire multiple surfaces including the surface of the reference object to be measured, specifically the floor surface of the platform P and the side surface of the door DR, based on the point cloud data acquired by the point cloud data acquisition unit 21 (Figure 4(b), step S2).

[0062] The labeling unit 23 executes a labeling step of labeling the multiple surfaces acquired by the surface acquisition unit 22 under predetermined conditions (FIG. 4(b), step S3).

[0063] The monitoring area setting unit 24 executes a monitoring area setting step of setting a monitoring area for detecting an obstacle based on the multiple faces labeled by the labeling unit 23 (FIG. 4(b), step S4).

[0064] Each step is described in detail below. Each time the LIDAR 10 completes a unit scan in the horizontal direction hd (see FIG. 1(b)) and the vertical direction vd (see FIG. 1(a)) of the monitored space, the LIDAR 10 outputs direction and distance data to the reflection points to the point cloud data acquisition unit 21 via the signal processing circuit 12. When the point cloud data acquisition unit 21 receives this information from the signal processing circuit 12, it stores the point cloud data converted into coordinates in the three-dimensional coordinate space of X, Y, and Z of each reflection point together with a time stamp in an internal storage unit. The three-dimensional coordinate space may be a relative coordinate system with the installation position of each LIDAR 10 as the origin, or may be an absolute coordinate system that is not dependent on the installation position of each LIDAR 10. The point cloud data is not limited to three-dimensional coordinates of X, Y, and Z, and polar coordinates may also be adopted.

[0065] 5 shows the process executed by the plane acquisition unit 22. The plane acquisition unit 22 extracts a point of interest and a point adjacent to the point of interest from a plurality of distance measurement points included in the point cloud data, applies principal component analysis (PCA) to these extracted points to generate a plane of interest (SA1), and executes a process (SA2) for acquiring a normal vector to the plane of interest as a normal vector at the point of interest for all point cloud data (SA3).

[0066] The point of interest refers to point cloud data selected based on a predetermined rule in order to obtain a normal vector from all point cloud data. For example, a thinning rate can be set as the rule. When the thinning rate is set to 50%, each set of point cloud data extracted every other point cloud data in the horizontal and vertical directions from all point cloud data obtained by unit scans in the horizontal and vertical directions becomes a point of interest. The thinning rate is a value that is appropriately set based on the number of all point cloud data, taking into consideration the computation load. Alternatively, each point cloud data whose distance is distributed in a predetermined range from among all point cloud data may be set as a point of interest. A rule may be defined by combining both such distance and thinning rate.

[0067] The plane acquisition unit 22 searches for distance measurement points existing within a radius of about several mm as neighboring points centered on the point of interest using the kd-tree method. The plane acquisition unit 22 calculates the center of gravity of the searched point cloud data, and then performs principal component analysis of the point cloud data. The plane acquisition unit 22 sets the third principal component vector obtained by the principal component analysis as a normal vector (a basis vector with a magnitude of 1) and calculates a plane passing through the center of gravity of the point cloud data previously obtained.

[0068] In addition to the above-mentioned principal component analysis method, methods for finding a surface from a point of interest and nearby points include the three-dimensional Hough transform method, the least squares method, the least median method, etc. For example, the least squares method calculates a plane that passes through the center of gravity of each point cloud data and minimizes the squared distance of each point cloud data.

[0069] Next, the surface acquisition unit 22 executes a clustering process. This clustering process includes the following steps SA4 to SA10. In the clustering process, first, a specific point of interest is extracted from a plurality of points of interest, and other points of interest existing within a predetermined search range are extracted from the extracted specific point of interest (SA4). Then, the normal vectors of the specific point of interest and the other points of interest are compared (SA5). If the normal vectors of the other points of interest are within a predetermined allowable range of inclination with respect to the direction of the normal vector of the specific point of interest (SA6), a clustering process is executed to group the other points of interest together with the specific point of interest (SA7). The other points of interest are grouped together with the specific point of interest as candidate points constituting the same surface as the surface containing the specific point of interest.

[0070] In this embodiment, the predetermined allowable range of inclination is set to 1°. The predetermined allowable range of inclination is a value that can be appropriately set depending on the object, etc. In addition, when grouping, the attention point and the distance measurement point that is a neighboring point of the attention point used to calculate the normal vector of the attention point may be grouped together, or only the attention point may be grouped. In the latter case, the normal vector may be included as an attribute value of the attention point.

[0071] The surface acquisition unit 22 repeats the clustering process with each attention point as a specific attention point (SA4 to SA10). If any of the attention points to be grouped in the later clustering process has already been grouped in the previous clustering process (SA8, Y), the surface acquisition unit 22 executes an integration process to integrate all the attention points to be grouped in the later clustering process into the same group as the previous clustering process (SA9).

[0072] In this embodiment, a space with a radius of about 2 m centered on a specific point of interest is set as the predetermined search range. For example, a transparent acrylic plate is installed in the center of a metal frame body to improve visibility and reduce weight of the door DR. In cases where there are few measurement points, such as above and below the frame body, and the estimation accuracy of the normal line is low, there is a risk that the side surfaces around the acrylic plate will be clustered as different surfaces. Even in such cases, by setting a space with a radius of about 2 m as the search range, the frame body surrounding the acrylic plate will be clustered as the same surface. The value of the predetermined search range can also be set appropriately depending on the object, etc.

[0073] Furthermore, after executing the integration process in step SA9, the surface acquisition unit 22 executes a second integration process (SA10'). The surface acquisition process obtains a representative normal vector representing each of the multiple groups generated by the clustering processes in steps SA4 to SA10, and integrates groups whose respective representative normal vectors fall within a predetermined allowable range into the same group.

[0074] The points of interest grouped by the clustering process of steps SA4 to SA10 are a set of a plurality of points of interest existing in a predetermined search range based on at least a specific point of interest. If the search range is set to a large value, points of interest that originally belong to different groups will end up in the same group, so the search range must be limited to a certain size, for example, a space with a radius of about 2 m as described above. If the search range is set to a smaller value than this, points of interest that originally belong to the same group may end up in different groups. Even in such a case, the points of interest can be integrated into the same group on condition that the representative normal vectors of each group fall within a predetermined allowable range.

[0075] A vector that is the median or average of the directions of the normal vectors of the points of interest that belong to the same group can be used as the representative normal vector. An approximate plane may be generated from multiple points of interest that belong to the same group, and a basis vector perpendicular to the plane may be used as the representative normal vector.

[0076] The specified tolerance range can be defined, for example, by both conditions that the direction of the representative normal vector falls within a specified tolerance range, and that the distance between an approximate plane generated from multiple focus points belonging to the same group and an approximate plane generated from multiple focus points belonging to another group falls within a specified range.

[0077] The second merging process in step SA10' is optional, and is not essential if the search range is set to an appropriate value. In addition, the merging processes in steps SA8 and SA9 are also optional, and if the merging processes in steps SA8 and SA9 are not performed, the second merging process effectively merges groups that belong to the same surface.

[0078] If the search range is increased in steps SA8 and SA9, or the allowable range of the distance between faces is increased in step SA10', then, provided that the condition that the inclination of the normal vector falls within a specified allowable range is met, a point of interest that is far away from a face that should be composed of points of interest included in the same group will be merged into the same group, which is an inconvenience.

[0079] Therefore, the surface acquisition unit 22 performs a filtering process on the point cloud data grouped into the same group by the clustering process to remove outliers from the same group (SA11). This process appropriately removes distance measurement points based on light reflected by the inner wall of the vehicle after passing through the window glass. Here, an outlier is a value that has an abnormally large residual from an estimated true value among the obtained observation values.

[0080] As the filtering process, RANSAC is preferably used. In this embodiment, three points are randomly selected from the points of interest included in the same group, and the distance between the plane created with the three points and each of the other points is calculated. If there are a certain number (1 / 4 of the total) or more of points whose calculated distance is equal to or less than a predetermined threshold (0.1 m), the points of interest are left as candidates, excluding outliers that deviate from the threshold. The surface acquisition unit 22 executes this process a predetermined number of times, and uses the point with the largest number of points that are equal to or less than the threshold to calculate the equation of the plane of the cluster. Note that the number of points selected to create the above-mentioned plane, the threshold, etc. are not particularly limited, and can be set as appropriate.

[0081] As a filtering process for removing outliers, in addition to RANSAC, M-estimation or the least median method can be adopted, and it is preferable to adopt a process that performs robust estimation including these.

[0082] Next, after applying a filtering process to each group, a predetermined number of faces are extracted in descending order of the number of distance measurement points (SA12). In this embodiment, by extracting a plurality of faces in descending order of the number of distance measurement points, the floor surface of the platform P, the inner surface of the door DR, the wall portion SW1, etc. are obtained.

[0083] 7 shows the process executed by the labeling unit 23. The labeling unit 23 identifies the floor surface of the platform P, the inner surface of the door DR, the side wall SW1 provided on the fence main body SB, etc. based on the relative relationship between a preset reference vector e and a normal vector for a cluster with a large number of measurement points among the multiple surfaces acquired by the surface acquisition unit 22 (SB1). Next, the labeling unit 23 detects the edge on the track side of the platform P, and generates an intersecting surface that intersects (is perpendicular to) the platform P as the side surface on the vehicle side (SB2).

[0084] In this embodiment, a vertical base vector is set as the reference vector e. Among the faces extracted in step SA2, a face whose normal vector is parallel to the reference vector e is identified as a face corresponding to the floor surface of the platform P, a face whose normal vector is orthogonal to the reference vector e and faces the rider 10 is identified as a face corresponding to the side wall SW1, and a face whose normal vector is orthogonal to the reference vector e and faces the track side is identified as a face along the inner side of the door DR.

[0085] Furthermore, a boundary line that is the edge of the platform P on the track T side is acquired according to the continuity based on the distance difference of adjacent point cloud data acquired by the point cloud data acquisition unit 21, and an intersecting plane VF that passes through the boundary line and is perpendicular to the floor surface of the platform P is generated and labeled. The intersection angle between the floor surface of the platform P and the intersecting plane VF is preferably 90°. However, if the platform P has a floor surface with a water gradient, it may be corrected so that the angle is 90° with respect to the horizontal plane.

[0086] In Figures 6(a) and 6(b), the areas indicated by the thick solid lines indicated by the symbols P, SW1, DR1, DR2, and VF, and the hatched surfaces indicated by the symbols P, DR1, and DR2 are surfaces that have been labeled by the labeling process described above. Note that Figure 6(a) corresponds to Figure 1(a), and Figure 6(b) corresponds to Figure 1(c).

[0087] The labeling unit 23 determines that the normal vector is orthogonal when the inclination angle of the normal vector with respect to the reference vector is within the range of 90°±Δθ1, and determines that the normal vector is parallel when the inclination angle of the normal vector with respect to the reference vector is within the range of 0°±Δθ2. The values ​​of the allowable angles Δθ1 and θ2 are not particularly limited, but are set to Δθ1=20° and Δθ2=10° in this embodiment.

[0088] Also, as shown in FIG. 6(b), in the case where the door ends of the doors DR1 and DR2 are joined at an angle in plan view, the doors DR1 and DR2 are individually identified based on the relative deviation of the multiple faces obtained by the clustering process, that is, the degree of relative inclination or deviation. For example, among the faces obtained by the clustering process, faces that are orthogonal to the reference vector e within the range of the allowable angle, have normal vectors facing the track T side, and have normal vectors that differ by a predetermined angle Δθ3 or more can be configured to be identified as faces corresponding to the doors DR1 and DR2. The value of the predetermined angle Δθ3 is not particularly limited, but in this embodiment, it is set to θ3=10°. Also, in the case where the deviation of the starting point position of the representative normal vector of each face is larger than a predetermined allowable range, the faces can be configured to be identified as faces corresponding to the doors DR1 and DR2.

[0089] Furthermore, the monitoring area setting unit 24 is configured to align the rising angles of the faces of each of the doors DR1, DR2 relative to the floor surface based on the normal vectors of the faces of the multiple doors DR1, DR2 labeled by the labeling unit 23 and the reference vector e, and to set a monitoring area DA based on each aligned door surface.

[0090] As shown in Fig. 3(b), the surfaces of the multiple doors DR1, DR2 labeled by the labeling unit 23 may be inclined rather than perpendicular to the floor surface of the platform P. This is because when the safety fence SF is installed on-site, the doors DR1, DR2 may be inclined by adjusting the door edges to join. In such a case, if the monitoring area DA is generated based on the surfaces of the doors DR1, DR2, the monitoring area DA becomes a distorted space, and even if the offset adjustment described below is performed, there is a risk that the doors DR1, DR2 will be erroneously detected as an obstacle. Furthermore, if the offset adjustment value is increased, the blind spot area will become larger.

[0091] Therefore, the rising angle of the surface of each door DR1, DR2 with respect to the floor surface is adjusted to be 90°. If the thickness of each door DR1, DR2 increases toward the bottom, the rising angle may be adjusted to be the average value of the inclination of the surface of each door DR1, DR2. In addition to this adjustment, one or both of the surfaces of the doors DR1, DR2 may be translated in the normal direction so that the joints of the surfaces of the doors DR1, DR2 are aligned.

[0092] 8 shows the process executed by the monitoring area setting unit 24. The monitoring area setting unit 24 generates an area surrounded by the other surfaces, that is, the surfaces corresponding to the doors DR1 and DR2, the surface corresponding to the side wall SW1, and the side surface where the height of the intersecting surface VF is a predetermined height, with the floor surface of the platform P labeled by the labeling unit 23 as the bottom surface of the monitoring area, and also sets a depth, which is the distance between the rider 10 and the side wall SW1, and generates the side wall SW2 on the rider 10 side (SC1).

[0093] Next, a closed area that will be the reference monitoring area is generated (SC3) by setting the heightwise boundary of the monitoring area. An offset adjustment is performed (SC4) to move each surface that constitutes the reference monitoring area, excluding the heightwise boundary, inward from the reference monitoring area by a predetermined offset value. For example, a surface that is a predetermined distance higher than the floor surface of the platform P is set as the bottom surface of the monitoring area to avoid erroneously determining that the floor surface of the platform P is an obstacle. As a result of the offset adjustment, the generated area is stored in the monitoring area storage unit 30 as the final monitoring area. FIG. 3(a) shows the monitoring area DA generated in this manner.

[0094] Another embodiment will be described below. In the above-described embodiment, an example has been described in which the labeling unit 23 is configured to identify the faces of each of the doors DR1, DR2 based on the relative bias including the normal vectors of the multiple faces acquired by the surface acquisition unit 22, but the labeling unit 23 may also be configured to identify the faces of each of the doors DR1, DR2 based on position information corresponding to the closed positions of the multiple doors DR1, DR2 that is input in advance.

[0095] For example, as shown in Figures 6(a) and (b), the distance L (design value) from the rider 10 to the joint of the door edge in a planar view is input as position information corresponding to the closed position, and the position of the door edge is determined as the position of the door edge among the faces of the door DR acquired by the face acquisition unit 22 that is the distance L from the rider 10 in a planar view, and the faces of the doors DR1 and DR2 can be individually labeled with the door edge position as the boundary.

[0096] 9(a) to (d) show an example of a case where position information is input in advance. As shown in FIG. 9(a), first, the floor surface of the platform P, the intersecting surface that passes through the boundary line of the floor surface and intersects with the floor surface, and the side wall SW1 are labeled in the labeling process, and then, in the monitoring area setting process, a reference area DA1 is generated, which is a rectangular parallelepiped with a predetermined height higher than the height of the door DR, with the depth from the rider 10 to the side wall SW1.

[0097] Next, as shown in Figure 9(b), predetermined offset values ​​are set for the floor surface, intersecting surface, and side wall SW1, and a reference area DA2 divided in two is generated based on the position information, and further, as shown in Figure 9(c), the reference area DA2 is expanded to include doors DR1 and DR2 to generate a reference area DA3.

[0098] Among the point cloud data previously acquired, the point cloud data included in each divided reference area DA3 corresponds to the doors DR1 and DR2, so the surfaces of the doors DR1 and DR2 are individually labeled in each divided reference area DA3. For the surfaces of the doors DR1 and DR2 obtained in this way, the detection surfaces of the doors DR1 and DR2 in the reference area DA3 are adjusted by the above-mentioned matching of the rising angle and parallel movement in the normal direction, and a reference area DA3 with offset-processed detection surfaces DR1' and DR2' is generated. As shown in FIG. 9(d), the divided reference area DA3 is integrated to generate the final monitoring area DA.

[0099] In the above-described embodiment, an example was described in which the rise angles of each surface are aligned and parallel translation in the normal direction is performed to align the detection surfaces of doors DR1, DR2. However, this is not limited to alignment of the rise angles and parallel translation in the normal direction. When the intersection angle of doors DR1, DR2 is small in a planar view, a single plane close to both planes may be generated and offset processing may be performed on the single plane.

[0100] The above-described embodiments are each an explanation of one example of the present invention, and the scope of the present invention should not be construed as being limited by the description. It goes without saying that the present invention can be appropriately modified and designed within the scope that achieves the effects of the present invention. [Explanation of symbols]

[0101] 10: Rider 12: Signal processing circuit 20:Monitoring area setting section 21: Point cloud data acquisition unit 22: Surface acquisition part 23: Labeling Department 24:Monitoring area setting section 30:Monitoring area information storage unit 40: Obstacle detection unit 50: System control unit BA: Standard monitoring area DA: Monitoring area DR: Door SF: Safety fence

Claims

1. a point cloud data acquisition unit that scans a monitored space in which a reference object is present, the reference object including a movable fence in a closed state that restricts entry to the boarding and disembarking area and a floor surface on which the movable fence is installed, using a LiDAR, and acquires point cloud data indicating the surface of an object including the reference object from the obtained distance image; a surface acquisition unit that executes a process for acquiring, for each of the points of interest, a normal to a surface of interest generated at a predetermined point of interest and a point adjacent to each of the points of interest from the point cloud data acquired by the point cloud data acquisition unit as a normal vector at each of the points of interest, and executes a surface acquisition process that acquires a plurality of surfaces including any of the points of interest based on a degree of coincidence of each of the acquired normal vectors; a labeling unit that identifies the floor surface from the multiple surfaces acquired by the surface acquisition unit based on a preset reference vector and normal vectors of the multiple surfaces acquired by the surface acquisition unit, and identifies and labels door surfaces corresponding to closed positions of the multiple doors that constitute the movable fence; a monitoring area setting unit that sets a monitoring area for detecting whether or not an obstacle is present in the boarding and disembarking area based on the plurality of faces labeled by the labeling unit; Equipped with The surveillance area setting unit is configured to align the rise angles of each door surface relative to the floor surface based on the normal vectors of the multiple door surfaces labeled by the labeling unit and the reference vector, and to set the surveillance area based on each aligned door surface.

2. 2. The monitoring area setting device according to claim 1, wherein the labeling unit identifies each door surface based on previously input position information corresponding to the closed positions of the plurality of doors.

3. The surveillance area setting device according to claim 1 , wherein the labeling unit identifies each door surface based on the relative deviation of the plurality of surfaces acquired by the surface acquisition unit.

4. A monitoring area setting device as described in any one of claims 1 to 3, wherein the labeling unit acquires a boundary line of the floor surface based on the point cloud data acquired by the point cloud data acquisition unit, generates an intersecting plane that passes through the boundary line and intersects with the floor surface, and identifies and labels the floor surface and the intersecting plane.

5. 5. The monitoring area setting device of claim 4, wherein the surface acquisition process includes a clustering process in which, if the normal vectors of other points of interest that exist within a specified search range from a specific point of interest are inclined within a specified allowable range with respect to the direction of the normal vector of the specific point of interest, the other points of interest are grouped together with the specific point of interest.

6. 6. The monitoring area setting device of claim 5, wherein the surface acquisition process includes a process of repeating the clustering process with each of the points of interest as the specific point of interest, and if any of the points of interest to be grouped in a subsequent clustering process have already been grouped in a previous clustering process, integrating all of the points of interest to be grouped in the subsequent clustering process into the same group as the previous clustering process.

7. 7. The monitoring area setting device of claim 6, wherein the surface acquisition process includes a consolidation process for obtaining a representative normal vector representing each of the multiple groups generated by the clustering process, and consolidating into the same group those groups whose respective representative normal vectors fall within a predetermined tolerance range.

8. 8. The monitoring area setting device according to claim 7, wherein the surface acquisition process performs a filtering process on the points of interest grouped into the same group to remove outliers from the same group.

9. a point cloud data acquisition step of scanning a monitored space in which a reference object is present, the space including a movable fence in a closed state that restricts entry to a boarding and disembarking area and a floor surface on which the movable fence is installed, using a LiDAR, and acquiring point cloud data indicating the surface of an object including the reference object from the obtained distance image; a surface acquisition step of executing a process for acquiring, for each of the points of interest, a normal to a surface of interest generated at a predetermined point of interest and a point adjacent to each of the points of interest as a normal vector at each of the points of interest, and executing a surface acquisition process for acquiring a plurality of surfaces including any of the points of interest based on a degree of coincidence of each of the acquired normal vectors; a labeling step of identifying the floor surface from the multiple surfaces acquired in the surface acquisition step based on a preset reference vector and normal vectors of the multiple surfaces acquired by the surface acquisition unit, and identifying and labeling door surfaces corresponding to closed positions of the multiple doors constituting the movable fence; a monitoring area setting step of setting a monitoring area for detecting whether or not an obstacle is present in the boarding and disembarking area based on the plurality of faces labeled in the labeling step; Equipped with A monitoring area setting method in which the monitoring area setting step is configured to align the rising angles of each door surface relative to the floor surface based on the normal vectors of the multiple door surfaces labeled in the labeling step and the reference vector, and set the monitoring area based on each aligned door surface.

10. 10. The method for setting a monitored area according to claim 9, wherein the labeling step identifies each door surface based on position information corresponding to a closed position of the plurality of doors that is input in advance.

11. 10. The surveillance area setting method according to claim 9, wherein the labeling step identifies each door surface based on a relative deviation of the plurality of surfaces acquired in the surface acquisition step.

12. A monitoring area setting method according to any one of claims 9 to 11, wherein the labeling step acquires a boundary line of the floor surface based on the point cloud data acquired in the point cloud data acquisition step, generates an intersecting plane that passes through the boundary line and intersects with the floor surface, and identifies and labels the floor surface and the intersecting plane that intersects with the floor surface.

Images