JP2011093514A - Safety device for platform door - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect a passenger left between a platform door and a vehicle, or an object projected from the inside of the vehicle. <P>SOLUTION: A distance image sensor 14 detects image data of a distance to an object in an image pickup area 12 between a platform door 1 and a vehicle 8. A calculating unit 16 extracts data in a processing area 20 from the image data of the distance, and determines whether or not the extracted data is a predetermined value or more. When the extracted data is the predetermined value or more, a detection signal is output to the outside. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a safety device provided on a platform door installed on a platform, and particularly relates to one using a sensor capable of acquiring three-dimensional data.

  Examples of platform door safety devices include those disclosed in Patent Literatures 1 and 2.

  According to the technique of Patent Document 1, two arms are arranged at a predetermined distance from the vehicle on both sides of the door in the region between the platform door and the vehicle, and a light beam is drawn between the two arms. In order to pass through, the projector of the transmissive photoelectric sensor is provided at a predetermined height position from the platform of one arm, and the light receiver is provided at a predetermined height position from the platform of the other arm. According to the technique of Patent Document 2, the area between the platform door and the vehicle is photographed two-dimensionally by a camera, the photographed image is compared with a reference image, and a difference between the two is detected. The height of the different portion is calculated, and it is determined from the height whether there is an obstacle that causes the vehicle to travel.

JP 2006-224842 A JP 2003-146201 A

  According to the technique of Patent Document 1, it is possible to determine whether or not a person or an object exists on a line segment at a predetermined distance from the vehicle and at a predetermined height from the platform. However, a position closer to the vehicle than a position at a predetermined distance from the vehicle, a thing that does not exist at a predetermined height position from the platform, for example, a protruding object protruding from the vehicle door, specifically an umbrella or a bag It cannot be detected. According to the technique of Patent Document 2, since the image taken in two dimensions between the vehicle and the platform door is compared with the reference image, the presence or absence of a protrusion from the vehicle door can be detected. Since the size of the protrusion cannot be measured three-dimensionally, it cannot be determined whether or not the protrusion protrudes to the platform side so that there is a problem in the operation of the vehicle.

  The present invention is capable of detecting passengers left and left between the platform door and the vehicle, and when there is an object protruding from the inside of the vehicle, whether or not the object has a problem in vehicle operation. It is an object of the present invention to provide a platform door safety device that can be judged.

  The platform door safety device of one embodiment of the present invention includes a time-of-flight three-dimensional camera. The three-dimensional camera detects three-dimensional data in a three-dimensional area including a person or an object existing between the platform door and the vehicle. Whether the arithmetic unit extracts the three-dimensional data in the predetermined three-dimensional area from all the three-dimensional data in the three-dimensional area, and whether the extracted three-dimensional data is greater than or equal to a predetermined value. It is determined whether or not, and a detection signal is output to the outside based on the determination result.

  The predetermined three-dimensional area can be partitioned into a portion corresponding to the door tip of the vehicle door and a portion other than that. In this case, in the portion corresponding to the door, the tertiary of the portion partitioned by a plurality of points in which the rate of change of adjacent data in the extracted three-dimensional data is greater than or equal to the first predetermined value The calculation unit determines whether the original data is larger than a second predetermined value. When the magnitude is equal to or larger than the second predetermined value, the first detection signal is output to the outside. In addition, the calculation unit extracts three-dimensional data in a portion other than the portion corresponding to the door, and whether the calculated three-dimensional data has a size equal to or greater than a third predetermined value. Judgment is made, and if it is equal to or greater than the third predetermined value, a second detection signal is output to the outside.

  In the platform door safety device configured as described above, the size of the person or the object can be determined using the three-dimensional data from the three-dimensional camera. In the portion corresponding to the door end portion, the data of the portion partitioned by a plurality of points where the change rate of the adjacent portion of the three-dimensional data is changed to the first predetermined value or more is used.

  Therefore, it is possible to detect not only whether there are people or objects between the platform door and the vehicle, but also how much the protruding object from the vehicle protrudes. If there is, a detection signal can be output to notify that. Further, it is possible to detect an elongated or thin object while suppressing the influence of noise. Furthermore, it is possible to determine whether or not there is a projecting object at the door end with only this data. Therefore, when taking a picture with a two-dimensional camera, it is necessary to prepare in advance data for reference necessary for taking the difference, that is, data after the vehicle stops and before the door of the vehicle is opened. No processing is ensured and easy. Since a time-of-flight 3D camera is used, 3D data can be obtained in an instant, processing is possible in a short time, and there is no mechanical scanning part, so the configuration is simple and durable. The nature is high.

  Furthermore, the calculation unit can make the first and second detection signals different. By comprising in this way, it can be discriminate | determined whether the passenger is left behind on a platform, or the protrusion from a vehicle protrudes so that it may become the obstruction of operation of a vehicle.

  The side surface on the vehicle side of the predetermined three-dimensional area can be defined along a vehicle limit where the vehicle is determined to be non-interfering with a building or a building limit where the building is determined to be non-interfering with a vehicle.

  When configured in this way, when the side of the vehicle side of the predetermined three-dimensional area is determined along the vehicle limit, there is a possibility that an object carried by the passenger or an object projected from the platform side may hit the vehicle. If the vehicle side of a predetermined 3D area is defined along the building limit, if the vehicle departs as it is, there is a possibility that an object protruding from the vehicle may hit a passenger or a building. Can be detected.

  Further, the side surface on the vehicle side of the predetermined three-dimensional area may be set based on a surface composed of data corresponding to the side wall of the vehicle among the three-dimensional data obtained from the three-dimensional camera. it can.

  With this configuration, even if a platform door is installed at a curved station where a part of the platform is curved, it is possible to get on and off accurately without being affected by the stop position of the vehicle. Customers and protrusions can be detected.

  The three-dimensional area may include a platform floor. In this case, the extracted three-dimensional data includes a portion corresponding to the floor. In the calculation unit, the change rate of the extracted three-dimensional data or the adjacent three-dimensional data in the extracted three-dimensional data is compared with a predetermined value for checking a three-dimensional camera attachment state.

  The three-dimensional camera may be attached to a place where a person getting on and off the vehicle may pass. In such a case, a person may come into contact with the 3D camera to change the mounting state of the 3D camera. If the comparison between the three-dimensional data extracted as described above and a predetermined value is made without being aware of the change in the mounting state, there is a possibility of erroneous detection. Therefore, if the platform floor is included in the 3D area, when a person touches the 3D camera, the 3D data for the floor is different from a predetermined value for checking the 3D camera mounting state. Or the rate of change between adjacent ones of the three-dimensional data is different from a predetermined value for checking the three-dimensional camera mounting state. By detecting this, the attachment state of the three-dimensional camera can be confirmed.

  The calculation unit is configured to determine the three-dimensional data determined based on the extracted three-dimensional data instead of the extracted three-dimensional data or a change rate of adjacent three-dimensional data in the extracted three-dimensional data. It is also possible to calculate a distance from the camera to a portion where the predetermined area intersects the floor, and compare this distance with a predetermined value for checking the three-dimensional camera mounting state. The distance can be obtained, for example, by determining a position where the change rate of the three-dimensional data is large as an intersection and obtaining a distance between the position and the three-dimensional camera.

  Thus, the distance between the intersection with the floor of the predetermined area and the three-dimensional camera is obtained, and this distance is compared with a predetermined value, so whether or not the mounting state of the three-dimensional camera has changed with high accuracy. It can be detected.

  As described above, according to the present invention, it is possible to detect a passenger who is left between the platform door and the vehicle, and if there is an object protruding from the inside of the vehicle, is the object problematic in vehicle operation? Can be determined.

It is a top view of the platform door which implemented the safety device of the platform door of a 1st embodiment of the present invention. It is a block diagram of the safety device of FIG. It is a top view which shows the process area | region of the distance image sensor in the safety device of FIG. It is a side view which shows the process area | region of the distance image sensor in the safety device of FIG. It is a perspective view which shows the process area | region of the distance image sensor in the safety device of FIG. It is a flowchart which shows a part of process which the calculating part in the safety device of FIG. 1 performs. It is a flowchart which shows the remaining part of the process which the calculating part in the safety device of FIG. 1 performs. It is a top view which shows the process area | region of the safety device of the platform door of the 2nd Embodiment of this invention. It is a side view which shows the change of the attachment state of the image distance sensor of the safety device of the platform door of the 3rd Embodiment of this invention. It is a flowchart which shows the process which the calculating part of the safety device of the platform door of the 3rd Embodiment of this invention performs.

  The platform door safety device according to the first embodiment of the present invention is attached to a platform door 1 as shown in FIG. The platform door 1 is provided at a predetermined distance from the edge of the platform 2. In the platform door 1, a pair of door pockets 4 are arranged along the platform 2 with an interval to be a door opening. A pair of sliding doors 6 advances and retracts in opposite directions from the door pocket 4 along the platform 2 so as to open and close the door opening between the door pockets 4. In the state where the sliding doors 6 and 6 close the door openings, the platform doors of the vehicle doors 10 and 10 of the vehicle 8 stopped on the platform 2 are positioned at the door ends of the sliding doors 6 and 6. The door 1 is installed. Reference numeral 11 indicates a side wall of the vehicle 8.

  In order to image a three-dimensional area, for example, an imaging space 12, indicated by a chain line in the space between the platform door 1 and the vehicle 8, in this embodiment, a three-dimensional camera For example, a TOF (Time-of-Flight) distance image sensor 14 is attached. The imaging space 12 is set to be inclined with respect to the vehicle 8 so that the vehicle doors 10 and 10 are included. The distance image sensor 14 includes a projector, for example, an LED and a plurality of two-dimensionally arranged image sensors. The distance image sensor 14 measures the arrival time of light that is irradiated from the projector and reflected and returned to each image sensor. The distance to the object can be measured for each image sensor. That is, two-dimensional distance data, that is, three-dimensional data can be measured. In this embodiment, this two-dimensional distance data is also expressed as a distance image.

  As shown in FIG. 2, the two-dimensional distance data from the distance image sensor 14 is transmitted to the calculation unit 16, where the processing is performed and passengers are left on the platform 2, or the vehicle 8 It is detected whether or not a protruding portion, for example, a passenger's umbrella or bag in the vehicle 8 protrudes beyond a predetermined length from between the door tips of the vehicle door 10, and notifies the host system 18.

  As shown in FIGS. 3 to 5, the calculation unit 16 does not use all the distance images obtained from the distance image sensor 14, that is, all of the two-dimensional distance data, but in a predetermined three-dimensional area, for example, the processing region 20. Only two-dimensional distance data is used. The side surface 20a on the vehicle 8 side of the processing area 20 is determined so that the vehicle 8 does not come into contact with the building on the platform 2 or a predetermined vehicle limit or the building on the platform 2 does not come into contact with the vehicle 8. It is predetermined to match the building limits. The end surfaces 20b and 20c on both sides in the direction along the platform 2 are set at positions slightly outside the door 6 side end of the side wall 11 of the vehicle 8 (opening direction side of the door 6). Further, the side surface 20d of the processing area 20 on the platform door side is determined at a position very close to the platform door, the lower surface of the processing area 20 is determined on the floor surface of the platform 2, and the upper surface is more than the upper surface of the vehicle door 10. It is set at a high position.

  The reason why the entire two-dimensional distance data is not used but is limited to the data in the processing area 20 is as follows. When all the data of the imaging region 12 is used, for example, in the data of the distance image sensor 14 using light and radio waves, not only the case where the light and radio waves radiated from the sensor are directly incident on the vehicle 8 or the like, but also the vehicle 8 Light and radio waves reflected many times between other parts, the platform door, or the platform 2 may enter the distance image sensor 14. As a result, as a result of the optical path length becoming longer, a distance value that cannot be detected by the distance image sensor 14, for example, a distance value indicating the inside of the vehicle 8 or a distance value indicating the floor under the platform 2 may be obtained. This is because the efficiency and accuracy of processing deteriorate.

  The processing area 20 is further divided into first and second processing areas 22 and 24. The first processing region 22 is a region having a predetermined length along the length direction of the platform 2 from the door tips of the doors 10 and 10 of the vehicle 8, and the second processing region 24 is the first processing region. 22 are the remaining areas on both sides. The length of the first processing area 22 is shorter than the length of the second processing area 24.

  In the arithmetic unit 16, the two-dimensional distance data of the first and second processing areas 22, 24, and the first and third predetermined values respectively set for the first and second processing areas 22, 24, For example, a comparison is made with an inspection threshold value to determine whether or not an object exists in the processing regions 22 and 24. It is also conceivable to compare the distance data of the entire processing region 20 including the first and second processing regions 22 and 24 and one threshold value for inspection of the processing region 20. However, when the threshold value for inspection is increased, for example, it becomes impossible to detect an elongated protrusion. On the other hand, if it is reduced, for example, there is a possibility of erroneous detection if there is an elongated protrusion in the processing region 24 that is unlikely to protrude from the vehicle due to the influence of noise. Therefore, an inspection threshold value suitable for each of the first and second processing regions 22 and 24 is determined for each of the first and second processing regions 22 and 24, and an error is detected while performing reliable detection. Detection is kept to a minimum.

  Furthermore, the main purpose of the first processing area 22 is to detect the protruding portion A protruding from the door tip of the door 10 of the vehicle 8 in the first processing area 22 as shown in FIG. The change rate of the distance data of adjacent pixels among the processing data of one processing area 22, for example, the difference value of the distance data of adjacent images is equal to or greater than a second predetermined value, for example, a predetermined protrusion detection threshold value Only the data of the portion partitioned by a plurality of points having the size of is compared with the inspection threshold value for the first processing region 22. For example, when the projection A is to be detected from the two-dimensional image, the vehicle 10 enters the platform 2 and immediately after stopping, the door 10 is not yet opened (that is, the projection A does not exist). Is stored as a reference image, and whether the protrusion A is present by comparing the reference image with an image in which the door 10 is opened and a passenger gets on and off and the door 10 is closed. Whether or not will be detected. Therefore, it takes time and effort to store the image when the door 10 is not open. However, since it is determined that the door 10 is not open based on a signal from the host system of the station building, When there is a problem or delay in the signal from the wrong reference image is obtained, there is a problem. On the other hand, when the change rate of the distance data of the adjacent images is obtained, such a trouble of storing the images becomes unnecessary, the processing is simplified, and the above-described problem does not occur.

  FIG. 6 shows detection processing performed by the calculation unit 16 for the second processing region 24. First, a portion corresponding to the second processing region 24 is extracted from the distance image obtained from the distance image sensor 14 (step S2). Next, the extracted distance image is compared with an inspection threshold value of the second processing region 24. For example, whether the number or area of pixels exceeding the threshold value is equal to or greater than a predetermined value It is determined whether or not a person or an object having a predetermined size or more exists in the processing area 24 (step S4). If the answer to this determination is no, the detection process for the second processing area 24 is terminated. If the answer to the determination in step S4 is yes, the host system 18 is notified that there are passengers left behind (step S6), and the detection process for the second processing area 24 is terminated.

  FIG. 7 shows detection processing performed by the calculation unit 16 for the first processing region 22. First, a portion corresponding to the first processing region 22 is extracted from the distance image obtained from the distance image sensor 14 (step S8). The difference between the distance values of adjacent pixels in the extracted distance image of the first processing region 22 is taken to generate a difference image (step S10). It is determined whether all of the differences in the difference image are smaller than a threshold value for determining whether or not the edge of the protruding portion exists (the threshold value is two on the plus side and the minus side) (step S12). If the answer to this determination is yes, there are no protrusions or passengers from the vehicle 8 in the first processing area 22, so the detection process for the first processing area 22 is terminated.

  If the answer to the determination in step S12 is no, there is a possibility that there is a projecting object from the vehicle 8 in the first processing area 22 or there is a possibility that passengers are left on the platform 2, so that the difference images are adjacent to each other. A closed region partitioned by pixels having a pixel difference equal to or greater than a predetermined value is extracted (step S14). The solid composed of the distance values of the closed region is a threshold value for inspection of the first processing region (determines whether or not the protruding object actually protrudes from the first processing region 22 by a predetermined amount or more. It is determined whether the size is greater than (threshold value) (step S16). The inspection threshold value is different from the predetermined value for the second processing region 22. Even when passengers are left behind in the vicinity of the door 10 of the vehicle 8, it can be detected by the threshold value for inspection.

  If the answer to the determination in step S16 is no, there are no protrusions or passengers from the vehicle 8 in the first processing area 22, so the detection process for the first processing area 22 is terminated. If the answer to the determination in step S16 is yes, the host system 18 is notified that a projecting object protrudes from the vehicle 8 in the first processing area 22 or that a passenger is left on the platform 2 (step S18). ). This notification is different from the notification that passengers are left behind in the second processing area 24 so that the host system 18 can distinguish them.

  The continuous operation time of the LED of the distance image sensor 14 is about 25,000 to 30,000 hours, and is further shortened by the influence of the ambient temperature. In order to extend the life of the product, it is only necessary to limit the light emission time of the LED. Therefore, the LED is not always turned on, but the LED is caused to emit light only when the distance image sensor 14 needs to be operated. For example, when a vehicle is in the platform and the vehicle door is closed and the vehicle is about to leave, there are no people or objects between the platform door and the vehicle, or there are protrusions protruding from the vehicle door. In the case of detecting a failure, the LED is turned on in synchronization with the presence of a standing line detection signal indicating that the vehicle is entering the platform, when the door controller starts to close the vehicle door. The lighting control of the LED is performed so that the lighting of the LED 3 is continued for a time set somewhat longer than the time required for the vehicle to start after the door is closed after the vehicle door starts to close.

  FIG. 8 shows a platform door safety device according to a second embodiment of the present invention. In the platform door safety device of the first embodiment, the position of the side surface of the processing region 20 on the vehicle 8 side is determined along the vehicle limit or the building limit. When the platform door is installed on a straight portion of the platform 2, the position of the side surface 20a on the vehicle 8 side of the processing area 20 is determined based on the vehicle limit or the building limit as described above. Even if the stop position on the platform 2 is slightly deviated, the distance from the distance image sensor 14 to the vehicle limit or the building limit is substantially constant, and no problem occurs. However, when the platform-door is installed on the curved portion of the platform 2, if the stop position of the vehicle 8 is slightly shifted, the angle of the vehicle 8 with respect to the distance image sensor 14 is shifted as shown in FIG. If the position of the side surface 20a on the vehicle 8 side of the processing region 20 is determined in advance based on the vehicle limit or the building limit, the undetectable region becomes large. Therefore, in the platform door safety device according to the second embodiment, the distance image sensor 14 detects the side walls 11 and 11 of the vehicle 8, and the platform 2 side is a predetermined distance from the side walls 11 and 11 in the normal direction, for example, 50 mm. Is the position of the surface 20a of the processing region 20 on the vehicle 8 side. Thereby, even if the stop position of the vehicle 8 is deviated, it is possible to prevent an undetectable region from becoming large. Since the other structure is the same as the platform door safety device of the first embodiment, the same parts are denoted by the same reference numerals and the description thereof is omitted.

  A platform door safety device according to a third embodiment is shown in FIGS. In the platform door safety device of this embodiment, detection of protrusions from the vehicle 8 and detection of people and objects are performed by the same method as in the first embodiment. However, as shown in FIG. 9 (a), the distance image sensor 14 is installed at the lower end of the door pocket 4 while being tilted slightly upward while facing the vehicle side so as to be inclined with respect to the vehicle. Has been. Since it is provided at the lower end of the door pocket 4, passengers may accidentally contact the distance image sensor 14, and the inclination of the distance image sensor 14 may change as shown in FIG. Alternatively, the inclination of the distance image sensor 14 may change due to vibration caused by vehicle traffic. That is, the attachment state of the distance image sensor 14 may change. When the mounting state of the distance image sensor 14 changes, the position of the processing region 20 changes, and a projecting object, person, or object from the vehicle cannot be detected accurately. Therefore, it is necessary to detect whether or not the attachment state of the distance image sensor 14 is maintained at a predetermined value.

  If the distance image sensor 14 is provided with an inclination detection sensor for this detection, the cost increases. Therefore, using the distance image obtained from the distance image sensor 14, the calculation unit 16 determines whether or not the attachment state has changed.

  In the state where the distance image sensor 14 is attached as shown in FIG. 9A, the position where the lower surface of the processing region 20 intersects the floor surface of the platform 2 is indicated by an arrow. When the inclination of the distance image sensor 14 changes as shown in FIG. 5B, the intersecting position changes as indicated by an arrow. Therefore, it is possible to determine whether or not the inclination of the distance image sensor 14 has changed by looking at the change in the intersection position. In a state where the distance image sensor 14 is normally attached without obtaining the intersection position, it is necessary to store a portion corresponding to the floor surface of the distance image of the distance image sensor 14 as a reference value and check the attachment state. It is also conceivable to compare a portion corresponding to the floor surface of the distance image of the distance image sensor 14 at a certain time with the reference value. In this case, the reference value must have a certain allowable range, and a subtle position change cannot be detected.

  Therefore, for example, a difference value between the distance image of the distance image sensor 14 and a pixel adjacent to each pixel in a portion corresponding to the floor surface is obtained, and a position where the change rate of these difference values is large is determined as an intersection position. When the difference value is used, since a subtle position change is emphasized, a subtle position change can also be detected. Only one point of intersection may be obtained. In this case, the storage amount can be reduced. However, the vertical rotation of the distance image sensor 14 cannot be detected. If the difference image on one line in the front and back direction of the paper surface of FIG. 9 is used, rotation in the vertical direction can be detected, but the storage amount increases. The intersection position may be obtained based on a predetermined number of difference images instead of all the difference images on one line, and the average value may be stored.

  In order to know the change of the crossing position, first, a reference value in a state where it is normally attached is acquired. That is, as shown in FIG. 10A, a portion corresponding to the floor surface is extracted from the distance image of the processing region 20 (step S20). Next, the difference between the distance values of adjacent pixels in the distance image is taken to generate a difference image (step S22). The distance between the position where the change rate of the difference image changes and the distance image sensor 14 is set as the floor end surface distance, and this floor end surface distance is stored as the reference floor end surface distance (step S24). Next, the difference image is also stored as a reference difference image (step S26).

  When it is necessary to confirm the attachment of the distance image sensor 14, a portion corresponding to the floor surface is extracted from the distance image of the processing region 20 as shown in FIG. Next, the difference between the distance values of adjacent pixels in the distance image is taken to generate a difference image, and the floor end surface distance is stored as the floor end surface distance based on this difference image (step S30). Then, it is determined whether the floor end surface distance acquired in step S30 and the reference floor end surface distance substantially match (step S32). If the answer to this determination is no, a signal requesting maintenance is output to the outside (step S34), and this process ends. If the answer to the determination in step S32 is yes, it is determined whether or not the difference image acquired in step S30 and the reference difference image substantially match (step S36). If the answer to this determination is no, step S34 is executed and the process is terminated. If the answer to the determination in step S36 is yes, there is no change in the attachment state of the distance image sensor 14, and thus this process is terminated.

  This attachment confirmation can be performed, for example, when performing object detection. As a result, it is possible to prevent erroneous detection simply due to a change in the position of the distance image sensor 14. Alternatively, it can be performed immediately before the opening operation of the platform door is performed, or can be performed when the vehicle has entered the platform 2 but the door is not yet opened. Alternatively, it can be performed every elapse of a certain period.

  Although the attachment position is confirmed using both the floor end surface distance and the difference image, the attachment position may be confirmed using only one of them. In addition, the attachment position can be confirmed using the distance image itself corresponding to the floor surface instead of the difference image.

  In each of the embodiments described above, the TOF type distance image sensor is used as the three-dimensional camera. However, the present invention is not limited to this, and for example, the striped light is irradiated and reflected on the vehicle 8 side. A scanning method that detects the three-dimensional data by reading the shape with a CCD or the like, a stereo vision method that detects the three-dimensional data by calculating the distance by the principle of triangulation using two cameras, or the vehicle 8 It is also possible to use a method in which a specific pattern projected on the side is captured by a single camera and three-dimensional data is detected from deformation of the projected pattern. In each of the above embodiments, the processing area 20 is divided into the first and second processing areas 22 and 24 to detect whether there are passengers or protrusions, but the detection accuracy is somewhat reduced, but the processing area It is also possible to detect whether there are passengers or protrusions in 20 as a whole. In the above embodiment, the distance image sensor 14 is installed on the platform door 1 side, but can also be installed on the vehicle 8 side. In the first and second embodiments, the distance image sensor 14 is provided in the middle of the height direction of the door pocket 4, but is not limited thereto, and is provided at the lower end of the door pocket 4 as in the third embodiment. Further, in the third embodiment, the distance image sensor 14 is provided at the lower end of the door pocket 4, but the present invention is not limited to this, and the height direction of the door pocket 4 as in the first and second embodiments. It may be provided in the middle.

1 Platform door 2 Platform 8 Vehicle 14 Distance image sensor (3D camera)
16 Calculation unit

Claims (6)

A time-of-flight 3D camera that detects 3D data in a 3D area including a person or object existing between the platform door and the vehicle;
Extracting the three-dimensional data in a predetermined three-dimensional area out of all the three-dimensional data in the three-dimensional area, and determining whether the extracted three-dimensional data is greater than or equal to a predetermined value. Based on the determination result, a calculation unit that outputs a detection signal to the outside,
A platform door safety device comprising:
The predetermined three-dimensional area is divided into a portion corresponding to the door tip of the vehicle door and a portion other than that, and the portion corresponding to the door tip is adjacent in the extracted three-dimensional data. The calculation unit determines whether or not the rate of change of the three-dimensional data is greater than or equal to a second predetermined value for the three-dimensional data of a portion partitioned by a plurality of points having a magnitude greater than or equal to the first predetermined value When the magnitude is equal to or larger than the second predetermined value, the first detection signal is output to the outside, the three-dimensional data in a portion other than the portion corresponding to the door is extracted, and the extracted three-dimensional data is A safety device for a platform door that determines whether or not the magnitude is greater than or equal to a third predetermined value, and outputs a second detection signal to the outside when it is greater than or equal to the third predetermined value.   The platform door safety device according to claim 1, wherein the calculation unit makes the first and second detection signals different from each other.   3. The platform safety device according to claim 1, wherein a side surface of the predetermined three-dimensional area on the vehicle side has a vehicle limit or non-interference in which the vehicle is determined not to interfere with the building. Platform door safety device defined along the architectural limits set forth for interference.   4. The platform door safety device according to claim 1, wherein a side surface of the predetermined three-dimensional area on the vehicle side corresponds to a side wall of the vehicle in the three-dimensional data obtained from the three-dimensional camera. Platform door safety device set on the basis of the surface made up of data.   5. The platform door safety device according to claim 1, wherein the three-dimensional area includes a floor of the platform, and the extracted three-dimensional data includes a portion corresponding to the floor. A safety device for a platform door, wherein a rate of change of the extracted three-dimensional data or adjacent three-dimensional data in the extracted three-dimensional data is compared with a predetermined value for checking a three-dimensional camera mounting state. The platform door safety device according to claim 5, wherein the calculation unit replaces the extracted three-dimensional data or a change rate of adjacent three-dimensional data in the extracted three-dimensional data with the extracted three-dimensional data. A platform that calculates a distance from the three-dimensional camera determined based on three-dimensional data to a portion where the predetermined area intersects the floor, and compares this distance with a predetermined value for checking the mounting state of the three-dimensional camera Door safety device.

Images