EP2550420A1

EP2550420A1 - Safety apparatus for closing and opening a door

Info

Publication number: EP2550420A1
Application number: EP10719307A
Authority: EP
Inventors: Saïd EL FASSI; Clara Nogueira Alves
Original assignee: Siemens SAS
Current assignee: Siemens SAS
Priority date: 2010-03-26
Filing date: 2010-04-28
Publication date: 2013-01-30
Also published as: CA2794268A1; KR20130063492A; WO2011116836A1; US8869449B2; US20130125468A1; CN102859104A; BR112012024277A2

Abstract

The invention relates in particular to an apparatus for automatically closing and opening at least one door (P) in a frame (CAD), said door being controlled by a controller (CTRL) providing a safety mode for retaining the door in an open state and, if need be, a door reopening mode in the event of an obstacle blocking the complete closure of the door into the frame, characterized by: at least one camera (CAM) having a vision axis intersecting at least the plane of the frame and providing an image of a space divided by the plane of the frame; at least one image analysis unit (AN) that isolates the regions of interest of an obstacle in the space; a calculator (CALC) for extracting at least one geometrical characteristic of each region of interest, for classifying the same into at least one closest category of different safety scenarios of known characteristics, and for outputting, to the controller, a safety signal (SEC) which activates the safety mode and which is dependent on the evaluation of a safety alarm value calculated by a formula, in accordance with the category, using at least one weighing of dimensional criteria applied to the geometrical characteristics.

Description

Safe installation of closing and door opening The present invention mainly relates to a settled ^¬ closing and automatic opening of at least one door in a frame according to the preamble of claim 1.

The prior art of the present invention is based on an automatic locking system and opening at least one door in a frame, said gate being controlled by a controller to guarantee a safe mode hand ^¬ door tien in an open state and if necessary a re ^¬ door opening mode in the event of the presence of an obstacle obstructing the complete closure of the door in its frame.

As a concrete example, the invention will take as an example a so-called landing door of a public transport vehicle such as that provided in a subway train which may present real dangers during the transit of passengers through the train. door if the said installation is not equipped with a reliable safe mode. Of course, the invention is not restricted to this example, but any type of dessert boo ^¬ door and locking systems and auto- matic opening, such as a building.

When a passenger with reduced mobility (eg fau ^¬ teuil rolling) or an adult pushing a stroller into a train by the automatic door, a sudden closure door initially announced the train and if the passage is not released on time, causes compression at the edges of doors on the passenger stroller and any obstacle or ob ^¬ accompanying jet, which can be extremely dangerous in some cases. Another danger may arise when an object held by a passenger is outside the vehicle while the door has closed while the passenger ger is inside or vice-versa, the object is inside and the passenger is still outside the vehicle, both being secured. This is the case, for example, of a leash (dog) held by a person or a piece of luggage.

The invention thus proposes in the first case when identifying a passenger with reduced mobility or a stroller to prohibit closing the door by keeping it in an open state. In the second case (dog leash traver health ^¬ the door frame, luggage strap), the invention proposes to open the door after of course prohibited the train in the case of public transport. To date, the security of the closing of the doors is ensured by a sensitive edge able to measure a threshold pressure on an obstacle during the closing of the door and thus to warn of the presence of the obstacle preventing a complete closing. The sensitive edge two major faults, it must be put in compression so that the door opens again (this is very Deran ^¬ giant in the case of a wheelchair or a stroller caught in doors) and its sensitivity is very often in ^¬ sufficient to detect for example the straps of a backpack taken in the closed doors (bag outside, passenger inside or vice versa). This type of scenario can be very dangerous at the start of the train. On metros with driver, the driver is responsible for monitoring with visual means (video) that these situations are not present and that after closing doors nothing exceeds the doors. Here, a human error is not anyway not discarded. Finally, in the particular case of trans ^¬ Automatic called public ports as automatic driverless metros, management of gates is provided automati ^¬ cally, only the sensitive edges ensure the safety of the closing doors. Depending on the case, after detection of the safety edge, the door re-opens only or the effort mech ^¬ nique is removed for a short time. Others are available sitifs based infrared beams exist, for example for the management of elevator doors, but they are not used for subways, because in times of affluence not to ^¬ sager, it would be impossible to close the doors of vehicles.

An object of the present invention is therefore to maximize the ^¬ safe caliber governed by a closing installation and automatic opening of at least one door in a frame, such as one or more landing doors of a public transport vehicle, especially if it is a vehicle without control or presence of a driver.

To this end, a facility is proposed through of claim 1 and an advantageous use of the ^¬ actual installation within public transport through claim 9.

A set of subclaims also has advantages of the invention.

From a self closing and opening ^¬ matic installation of at least one door in a frame, said gate being controlled by a controller ensuring mode secu- rity door holding in an open state and need a door reopening mode in the event of obs ^¬ obstacle obstructing the complete closure of the door in its frame, the invention provides that the controller comprises:

at least one camera provided with an axis of vision at least separating from the plane of the frame and providing an image of a volume cut by the plane of the frame,

at least one image analysis unit which isolates the re ^¬ gions of interest from an obstacle in the volume,

a calculator that extracts at least one geometrical characteristic from each region of interest, classifies it in least a category closest to different security scenarios of known characteristics and delivers to the controller a safe signal activating the safe mode and dependent on the evaluation of a security alarm value calculated by a formula (such as an algorithm) according to category involving at least one cry dimensional weighting ^¬ ^¬ teria applied to geometri cal characteristics. Advantageously, the regions of interest can be for my ^¬ geometric such as disks, squares, rec ^¬ tangles, bars. These are preferably chosen very simply and according to very distinctive geometries, to facilitate and accelerate any closure to avoid while ensuring a high level of geometric recognition security.

In order to maximize the desired security level, it is planned to define the dimensional criteria from at least one of the following elements:

- absolute dimensional values of characteristics,

- dimensional relative values between the charac ^¬ ticks

- dimensional value ratios of the scenario-based characteristics.

These elements can be selected and completed after a learning phase based on many scenarios, or even by the implementation of a neural network. The installation provides that the axis of vision is at least obli ^¬ that relative to the plane of the frame to be vertical in the plane of the frame or slightly offset thereof. This ^¬ priété is seen as providing increased security vision of regions of interest in the various cases of hazards described above. In order to further raise the security level many cameras can be arranged with triangulation axes of vision. This makes it possible to ensure that a masked area of an individual at risk will still be taken into account or will allow during busy periods to better distinguish an individual in danger among others. Similarly, in the event of a camera failure, redundant security protection is provided.

Similarly, the door frame may include a laser meter connected to the controller as a second safe mode of door retention in an open state and if necessary a door reopening mode. It is also possible to have additional means for keeping the door in an open state and, if necessary, a door reopening mode which are connected to the controller, such as at least one door edge with mechanical sensitivity (usual or variable) on a door. obstacle or at least one light barrier crossing the plane of the door frame. Finally, the installation according to the invention can provide that the calculator extracts dynamic characteristics of the re ^¬ gions of interest that are classified in a category of known scenarios and the security alarm value is calculated calcu ^¬ complementarily by security criteria based on these dynamic characteristics, ideally by a complementary pon ^¬ eration in formula based on the written ^¬ dimensional res. Thus, movements characteristic of a person with reduced mobility or a bag strap can also come to improve a hazard detection and thus provide a better level of security expected.

The invention thus provides that a use of

the installation according to the invention ideally has a high safe potential for doors with automatic closing of public transport vehicles, in particular vehicles automatic devices without driver. Also in order to achieve a maximum level of security, the camera image can also be transmitted to a central control station em ^¬ boat or on the ground, especially if security alarm. In this way, an operator can parallel

the installation according to the invention to note in real time a situation of danger emitted as well as the proper functioning of said installation. Examples of implementation and application are provided using the figures described:

Figure 1 Schematic diagram of the installation according to

The invention,

Figure 2 Principle of the invention in the case of a fau ^¬ teuil rolling,

Figure 3 Principle of the invention in the case of a breed or a dog leash.

Figure 1 shows a sliding hoistway door P in a ca ^¬ dre CAD and provided with a closing device and

automatic opening in its frame, said door being controlled by a CTRL controller ensuring a safe mode of keeping the door in an open state and if necessary a door reopening mode in the event of obs ^¬ obstacle obstructing the closure complete the door in its frame.

The installation finally includes:

at least one CAM camera equipped with an axis of vision AX at least intersecting with the plane of the CAD frame and providing an image of a volume VOL cut off by the plane of the frame,

at least one analysis unit AN of the image which isolates regions of interest from an obstacle in the volume, - a calculator CALC which extracts at least one characteristic ^¬ geometric each region of interest, classifies in at least one nearest category different security scenarios known characteristics and outputs to the controller a signal SEC safe activating Sécuri fashion ^¬ silent and dependent on the evaluation of a sé alarm value ^¬ curitaire calculated by a formula according to category using at least one weighting dimensional criteria applied to geometric characteristics.

If security alarm (= on SEC signal), the Contro ^¬ forbidden door closing or reopening the door, and of course sends a signal STOP asset with a vee ^¬ vehicle concerned. Figure 2 shows an image acquired from the top of the frame downwards for a first case where a person with reduced mobility (on a wheelchair) enters the frame of PI, P2, while the doors are activated in closing. The camera at the top of the frame acquires shortly before and during the closing of the successive images and by a usual means of image recognition extracted in our case two main regions of interest: the head 1 of the per ^¬ sonne and the square form 2 of his wheelchair (lined with two wheels characteristic of two edges of the square). The cal- culateur CALC then extracted a round one and a square 2 and clas ^¬ fied this result in the scenario category "wheelchair" in a first dimensional criteria "round com ^¬ caught in the square," as a 50% by example. A second dimensional criteria can then also be ex- trait from respective dimensions of the two characteris ^¬ geometric ticks 1, 2 (length, width, area) and in the case of increase allows for example the previous 20% probability. Finally, a third dimensional criterion can also be extracted from the ratio of the surfaces of the geometric characteristics 1,2, for example that the ratio of the areas "square divided by round" is greater than or equal to 3. Thus, under a suitable weighting of each of these three criteria, it is possible to conclude that Probabili ^¬ ty of 80% is reached, which exceeds a threshold prédéfinissa ^¬ ble to activate the safe signal and prohibit farm ^¬ ture or, if necessary, re-open the door.

Figure 3 presents, according to the same principle as figure 2, an image acquired from the top of the frame down for a second case where a passenger holds his dog on a leash and did not notice that his dog did not follow him. in or out of a vehicle while the two landing doors will close, are closing or are almost closed. The regions of interest are the head 3 (or / and the trunk / shoulders) of the passenger as well as the trunk and the head 5 of the dog. A stuck leash / shoulder scenario is then detected. Dimensional criteria are thus defined

s applying it to the dimensions of features of each of the regions of interest 3, 5. If it is possible troi ^¬ SIEME 4 region of interest such as a lead or a strap is sought, and also reinforces another dimensional criterion. Additional criteria of ^¬ face ratios between the geometric characteristics 3, 4, 5 can be calculated. Other dynamic dimensional criteria (panic or fear movement of the dog or the passenger) can be detected on the basis of movement of the geometrical characteristics 3, 5 (and 4). Finally, a weighting formula of these criteria thus makes it possible to trigger a security alarm if necessary.

Subsequently, the invention is described in other aspects allowing a commissioning of the installation in the case of automatic public transport vehicles.

Additionally or total substitution systems Sécuri ^¬ tary sensitive edges, the invention proposes using several cameras, optionally supplemented with a laser rangefinder, and installed around the door to identify dangerous Sce ^¬ nes. When a dangerous scene is present Vehicle dock, depending on the type of scene, the door will be open hand ^¬ held (for persons with reduced mobility) or will be reopened to release the stuck object (braces taken through a door).

A camera or a network of cameras and surround the enca ^¬ drement doors. The cameras are on board the vehicle and film the outside of the door or are docked parti¬ ^¬ particularly in the presence of landing doors. The background image is perfectly well known. It is proposed to implement algorithms for automatic identification of selected scenes. The network of cameras can be used to obtain a stereoscopic view of the scene and improve the reliability of the detection by appreciating the volume, the types of objects stuck by region of interest. An improvement of the detection is proposed by the addition of a laser range finder. A fusion of the data from the cameras with the laser rangefinder is proposed to improve the quality of the detection. The rangefinder will allow to appreciate with much more precision the precise location of the object and combined with the in ^¬ formations of the cameras the volume will be calculated with more precision. In the case where the installation according to the invention is used without edges sensitive to the doors, treatments are introduced on board the cameras or the laser range finder to achieve at least the known level of security SIL 3, that is to say one. probability of failure at least less than 10-7 / h. These treatments will make it possible to verify that the present detection system is still operational, and that its performance is guaranteed.

In particular, a camera or a camera network surrounding a train door a scene analysis being located on the vertical plane of the door and located in the passage of por threshold ^¬ you. The choice of the configuration of the cameras is made according to the accuracy of the detection of regions of interest that we intend to achieve. In particular, in the case of camera networks, it will be possible to make use of both to understand the volume of identified forms and American ^¬ improve the security level of the security function. These treatments can be embedded in the camera (s) or installed on dedicated equipment. The information from the cameras can be supplemented by the information received from a laser range finder. A data fusion is performed to accurately determine the location of the object (height and position relative to the door plane) and its volume. Three different treatments can then be implemented depending on the function to be performed.

1 Detection of persons with reduced mobility or strokes ^¬

The cameras (and possibly the laser) analyze the scene. A reference image of different types of chair rou ^¬ lant, strollers is stored and known. Top view we will see heads and a place a frame showing a free volume, or will see only the volume of the chair (case chair alone in the door). Dimensional detection criteria (associated analysis unit and calculator) can then be defined. From these gift ^¬ born, assessing the volume occupied by the wheelchair. By correlation with the known volumes and the identification of a head we can consider that the scene is detected. This volume is then located in relation to the vertical plane of the doors. If the intersection of this volume with the door plan is not empty then we consider that the scene is active. To improve the detection and to avoid false detections, a measurement of the height of the objects is carried out and their location with respect to the door plan determined. These me ^¬ safe are possible through the knowledge of the position and optical characteristics of the / cameras ^(¬ and Even tuellement laser rangefinder) used. The position of the ca ^¬ meras and the optical characteristics are also taken into account to identify the position of the objects with respect to the plane of the door. When the scene is declared active, a signal is transmitted to the system (= via the controller) managing the doors or the door to prohibit the closing of the door. When the door threshold is no longer occupied by this type of scene, the device disables the transmitted signal, no longer prohibiting the closing.

2 ° Detection of an object taken in the doors.

A bag with long straps, handles, belts of buckled clothing or any object attached to a person who can get stuck in the doors, can when closing the doors be found outside the vehicle (or inversely the ^per¬¬ rings outside and the object inside) without the door system reopening. An equally delicate case is a dog outside or inside the vehicle with the leash taken in the doors. This situation is dangerous if the vehicle were to start. When the doors are closed, the camera (s) analyze the vertical plane outside the door. A reference gate plan is known sys ^¬ tem. The system related to the installation according to the invention analyzes the closure of the door. An object is identi ^¬ fied by image processing. In the case of a cam network ^¬ ras (and laser rangefinder), the processing will evaluate the shape, height and volume of the object. A template ^de¬fining the size, the volume of the object is taken into account. If the object exceeds this template, the device prohibits by criteria defined on the above characteristics the departure of the train by sending to the automation systems or traction control a departure ban. After this prohibition, an order to open the door concerned is automatically controlled to release the jammed object.

3 ° Automatic reopening on a person's jamming

In the event that sensitive edges are not installed on the doors, the camera (s) (and the laser rangefinder) will analyze the door plan. A jam is identified by a nobody pressed between two leaves, for example sliding. The system detects the movement of the doors by analyzing the movement of the door edges. When, as a criterion comple ^¬ mentary, the edges of doors move more and do not always contiguous days, a crack analysis of your por ^¬ is performed. The height of the object, the volume and its shape are taken into account to make the decision to make the scene active. In the active scene, a reopening security command is transmitted to the gate controller system. The door immediately stops the closing force and reapplies the force after an adjustable time (for example 5 seconds) or reopens completely. It is also possible to send the acquired images to a train control center to confirm operationally an authorization to close or open the doors by an agent.

Safe treatment.

When the security feature requires a SIL3 (failure rate of less than 10-7 / hour), it is ne- necessary to add at camera functions per ^¬ setting to ensure that the images are from the scene real-time and the cameras are still operational.

Three types of treatment are carried out:

Two cameras are at least used. A correlation of the different images received from the cameras is ré ^¬ alized. Typically, this helps to identify if an optic is obstructed, if the camera no longer works properly, and / or if both cameras are facing the same scene. To improve the availability of the function, it is possi- ble to switch to a 3 camera network to achieve a re ^¬ dundancy and always have at least two cameras ^¬ tional operation.

This first analysis by correlation of camera images can be completed by tests of the injection acquisition chain at the level of each image electronics. known tests characterizing the door and following a determined sequence. The good reception of images at the uni ^¬ processing side will consider the acquisi ^¬ chain as always operational.

The image pickup is synchronized with the system time, so any time variation ^¬ informa tion transmission will be detected. Mastering this time is important to avoid having a dangerous time lag between the system and the analyzed scene. The reaction of your automatic Por- system linked to the installation of the invention must always be compatible with the sys tem ^¬ security level of public transportation. If a pre-defined threshold is exceeded, the automatic door system will be considered as faulty.

Claims

claims

Installation for closing and automatic opening of at least one door (P) in a frame (CAD), said door being controlled by a controller (CTRL) guaranteed ^¬ a safe mode of keeping the door in an open state and at require a door reopening mode in the event of an obstacle obstructing the complete closing of the door in its frame,

characterized by:

at least one camera (CAM) provided with an axis of vision at least secant to the plane of the frame and providing an image of a volume cut by the plane of the frame,

at least one analysis unit (AN) of the image which isolates regions of interest from an obstacle in the volume,

- a calculator (CALC) which extracts at least one geometric charac ^¬ teristic of each region of interest, classifies the at least one nearest category different security scenarios known characteristics and outputs to the controller a safe signal (SEC) activating the safe mode and dependency on evaluation of a security alarm value calculated by a formula according to category using at least one weighting dimensional criteria ^¬ c app to geometric characteristics.

Installation according to claim 1, wherein the regions of interest are geometric shapes such as disks, squares, rectangles, bars.

Installation according to one of Claims 1 to 2, for which the dimensional criteria comprise at least one of the following elements:

- absolute dimensional values of which

- dimensional relative values between the charac ^¬ tics,

- dimensional values of reports of charac ^¬ teristics based on a screenplay.

Installation according to one of claims 1 to 3, wherein the axis of vision is at least oblique rap ^¬ port to the plane of the frame to be vertical in the plane of the frame or slightly offset therefrom.

Installation according to one of the preceding claims, for which several cameras are arranged with triangulation axes of vision.

Installation according to one of the preceding claims, wherein the door frame comprises a laser range finder connected to the controller as a second SE mode ^¬ door holding curitaire in an open state and if necessary a door reopening mode.

Installation according to one of the preceding claims, wherein complementary holder means holding in an open state and need a way of Reu ^¬ door Verture are connected to the controller, such as at least one edge mechanical sensitivity relates to an obstacle or at least one light barrier crossing the plane of the door frame.

Installation according to one of the preceding claims, wherein the calculator extracts characteristic c ^¬ dynamic regions of interest that are conventionally relied to a category known scenarios and the value of security alarm is additionally calculated by security criteria based on these charac- c dynamic, ideally with a comple mentary ^¬ weighting in the formula based on the dimen- sional criteria.

9. Use of the installation according to one of the preceding claims for closing doors of automati ^¬ of public transport vehicles, in particular automatic driverless vehicles.

10. Use according to claim 9, for which the image of the camera is transmitted to an on-board control center ^¬ tral or on the ground, particularly in case of a security alarm.