Procedures and control system for securing a surveillance area

Abstract

Method (100) for securing a monitoring area (1), comprising the following steps: - Subdividing (101) the monitoring area (1) into several segments (2), - Detecting (105) the entry of a person (5) into the surveillance area (1); characterized by: - Calculating (110) a time-dependent probability of residence (231) of the person (5) in a secured segment (2) of the several segments (2) of the surveillance area (1), - Adapting (120) a security measure relating to the secured segment (2) depending on the probability (231) of the person (5) being in the secured segment (2), wherein the monitoring area (1) includes intersection points (3) and path segments (4) along which the person (5) can move between the intersection points (3), where the calculation (110) of the probability of being located (231) includes a calculation (112) of transition probabilities between the path segments (4) connected at the intersection points (3), where the probability of residence (231) is calculated depending on a target position (7) sought by the person (5), where calculating (110) the probability of being located (231) includes calculating (114) a shortest path (31) along the path segments (4) from an initial position (6) to the target position (7), where, in calculating (110) the probability of being located (231), a set of possible paths (31, 32) is used, where the path set includes, in addition to the shortest path (31), a number of additional paths (32) between the starting position (6) and the target position (7), where the number of additional paths (32) includes another path (32, 33) if a path length of the further path (32, 33) exceeds a minimum path length specified by the shortest path (31) by a maximum of a specified upper limit.

Description

[0001] The present invention relates to a method and a control system for securing a monitored area.

[0002] Methods and control systems for safeguarding a monitored area are used in automation technology to reduce the risk to people or controlled processes from an automation process. Typically, the presence of people or objects in the monitored area is detected by one or more sensors, and, if necessary, the movement of an actuator that could endanger people or objects present in the monitored area is prevented. To safeguard the movement, the actuator can, for example, be stopped, or a safe parameter range can be defined for the movements to be performed by the actuator as soon as a person or object is detected in the monitored area.

[0003] The safety measures to be implemented to secure the monitored area are determined within the framework of a hazard assessment of the controlled system. This typically involves determining a safety level, usually quantitative, for each hazard emanating from the system. The safety level is typically defined based on the probability of occurrence and the potential severity of the hazard under consideration, such as the severity of possible injuries to a person at risk. The safety level then determines the requirements for the safety measures to be taken to reduce the hazard. Such safety measures are also referred to as safety functions. The methods to be used in the hazard assessment and the safety functions to be implemented to reduce the hazard are specified in various standards.

[0004] Control systems for securing monitored areas are used, among other things, to control automated guided vehicles (AGVs), for example, in automated warehousing. The monitored area typically includes at least those parts of a warehouse where the vehicles operate. If people are also present in the monitored area at the same time as the vehicles, they can be endangered by potential collisions. Safety measures to reduce this risk can include, among other things, driving the vehicles at a controlled speed, in particular by implementing controlled slow-speed operation.

[0005] The safety measures implemented to secure the monitored area typically reduce the availability or efficiency of the system controlled by the automation process. For example, slow-moving an automated guided vehicle (AGV) reduces the throughput of goods transported within the system. Therefore, only the safety measures necessary to reduce the required risk should be implemented, and all further measures should be avoided where possible.

[0006] The safety measures for securing the monitored area are generally statically defined and typically tailored to the greatest conceivable hazard that could originate from the controlled system. This can unnecessarily reduce the availability or efficiency of the controlled system, especially if the greatest conceivable hazard occurs only rarely and / or for a short duration. For example, when securing a monitored area traversed by a vehicle, this can lead to the vehicle slowing down whenever a person enters the monitored area, regardless of how close the person and vehicle are.

[0007] Publication DE 10 2016 209 704 A1 describes a method for controlling a personal protection device in a vehicle (paragraph

[0004] ), wherein the area around the vehicle is detected with an environment sensor and the sensor information is evaluated in a static occupancy grid and combined to form an environment map.

[0008] Publication DE 10 2008 062 916 A1 describes a method for determining the probability of a collision between a vehicle and a living being. To determine the probability of a collision, individual trajectories of the vehicle and the living being are calculated, and possible collisions are calculated based on the intersection of these trajectories.

[0009] The article Pellegrini, S., [et al.], “Wrong turn - No dead end: A stochastic pedestrian motion model”, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, San Francisco, CA, 2010 describes a stochastic model for modeling the social behavior of pedestrians in order to predict their future movements.

[0010] The object of the invention is to provide a method and a control system for securing a monitoring area in such a way that the security measures taken for this purpose are improved.

[0011] This task is solved by a method and a control system according to the independent claims. Further developments are specified in the dependent claims.

[0012] A procedure for securing a monitored area includes the following steps: - Dividing the monitoring area into several segments, - Detecting a person's entry into the monitored area, - Calculating a time-dependent probability of a person being in a secured segment of the multiple segments of the surveillance area, - Adapting a security measure concerning the secured segment depending on the probability of the person being in the secured segment.

[0013] Within the scope of the invention, it was recognized that securing the monitored area does not require constantly implementing the same security measures throughout the entire area. Rather, it suffices to implement a specific security measure only in those sub-areas or segments of the monitored area where the presence of a person cannot be ruled out with sufficient certainty. Conversely, if the probability of a person being present in the secured segment of the monitored area is sufficiently low, the security measure pertaining to that segment can be adapted to the reduced probability of presence, thereby increasing the availability of the controlled system. Adapting the security measure pertaining to the secured segment thus makes it possible to tailor the implemented security measure to the specific hazard situation present in the monitored area.

[0014] The security measure affecting the protected segment can be adapted to the probability of a person being in the protected segment, for example, by only activating the measure if the probability of the person's presence exceeds a predefined threshold, and by not activating the measure if the probability of the person's presence falls below the predefined threshold. If the probability falls below the threshold, for example, no security measure, a reduced security measure, or a different security measure may be implemented. The reduced or additional security measure may, for example, have less of an impact on the availability of the controlled system than the security measure implemented when the threshold is exceeded.

[0015] The safety measure can, for example, serve to safeguard an actuator of a control system that could endanger a person in the protected segment. In particular, the safety measure can serve to safeguard a movement of the actuator that could endanger a person in the protected segment. For example, a robot located in the protected segment or a vehicle traveling through the protected segment might be driven by the actuator. The safety measure can, for example, consist of reliably preventing or stopping the dangerous movement, or of restricting the movement to a safe parameter range, such as with a safely reduced speed, force, acceleration, or similar parameters. The safety measure can, for example, consist of specifying a maximum speed for the movement of the actuator, especially of a vehicle driven by the actuator.A reduced safety measure can then consist of specifying a further maximum speed for the speed of movement of the actuator, whereby the further maximum speed is greater than the maximum speed when the safety measure comes into effect.

[0016] The monitored area can be located, for example, in a hall, such as a warehouse or a production hall. The monitored area can encompass the entire hall or a section thereof. The segments of the monitored area can include, in particular, aisles and / or intersections connecting multiple aisles. For example, each aisle and / or intersection within the monitored area can constitute a segment of the monitored area. The aisles can, for instance, be located between rows of racking in the hall.

[0017] The entry of a person into the monitored area can be detected, for example, by a sensor located at an entrance to the monitored area. The segments of the monitored area can cover all sub-areas where there is a risk to the person present. In particular, the segments of the monitored area can cover all sub-areas where a vehicle is traveling within the monitored area.

[0018] A controller that regulates the movement of the actuator can, for example, be configured to adjust the safety measure pertaining to the protected segment only when the controller receives a corresponding control signal. In all other cases, the controller can be configured not to adjust the safety measure. If the controller is operating a vehicle traveling through the segment, it can be configured to allow the vehicle to enter the protected segment at an unsafe speed only after receiving the control signal, and otherwise to allow the vehicle to enter the protected segment at a speed safely reduced to the maximum speed.

[0019] The time-dependent probability of a person being in the secured segment can be calculated, for example, using a stochastic model. When calculating this probability, uncertainties in detecting a person's position within the monitored area can be taken into account, such as the uncertainty of a sensor detecting the person, including measurement inaccuracies or sensor unreliability. Specifically, uncertainties in detecting the person's entry into the monitored area, or uncertainties in the sensor detecting this entry, can be considered when calculating the probability of presence. This allows the security measure to be adapted to the situation, even in cases where only partial information about the person's position is available.

[0020] To calculate the probability of a person's location, transition probabilities between individual segments can be calculated. These transition probabilities indicate the probability that a person will move from a given segment to a neighboring segment. The probability of a person's location in the secured segment can then be calculated from the transition probabilities, starting from a starting segment where the person's entry into the monitored area was detected. This calculation involves determining the transition probabilities for all segments connecting the starting segment to the secured segment.

[0021] The time-dependent probability of a person being located in the secured segment can be calculated in discrete time steps, for example, in equidistant discrete time steps. The transition probabilities between the individual segments can then each represent the transition probabilities for a transition during the time step.

[0022] The transition probabilities can be calculated based on a purely random movement of the person along the segments of the monitoring area (random walk). For each segment of the monitoring area, the transition probabilities for crossing to all adjacent neighboring segments can be the same. In particular, the transition probabilities for crossing to the neighboring segments can each be inversely proportional to the number of neighboring segments. However, the transition probabilities can also be calculated based on a directed movement of the person along the segments. The directed movement of the person can, for example, be determined by a target position to which the person is heading.

[0023] Transition probabilities can be specified for each segment under consideration, independently of any preceding segment. The preceding segment can be the segment from which the person transitioned to the segment under consideration during propagation. Alternatively, the transition probabilities can be specified for each segment under consideration, depending on the preceding segment. If the segment under consideration comprises a number of adjacent neighboring segments, each of these neighboring segments can be a preceding segment and / or a successor segment of the segment under consideration, and transition probabilities can be specified for all possible pairwise combinations of preceding and successor segments.These transition probabilities then each represent the probability that the person propagates from a specific predecessor segment, via the segment under consideration, to a specific successor segment.

[0024] When calculating the probability of a person being in the secured segment, the duration of their stay in each segment of the monitored area can also be taken into account. These durations indicate how long the person spends in each segment. In particular, the time dependency of the probability of a person being in each segment can be determined by the duration of their stay in that segment.

[0025] This procedure also allows for multiple individuals to be present within the monitored area simultaneously. In this case, analogous to the calculation of the probability of presence described below for a single individual, a time-dependent probability of presence in the secured segment of the monitored area can be calculated for each person. In particular, the entry of each individual into the monitored area can be recorded. The security measure pertaining to the secured segment can then be adjusted based on the probability of presence of all individuals within the secured segment, specifically based on an integration of these probabilities.

[0026] As part of the procedure, a time-dependent probability of presence of a person or persons entering the monitored area is preferably calculated for each segment of the monitored area. This time-dependent probability of presence can be calculated, for example, on a control unit of the control system or on a separate risk control device. The calculation of the time-dependent probability of presence can also be performed as a server-based service, for example, as a service provided in a cloud.

[0027] In a further development of the procedure, the monitored area is traversed by a vehicle, and the safety measure includes safeguarding the vehicle's movement, in particular its slow-speed operation, within the secured segment. During slow-speed operation, the vehicle is controlled within a safe parameter range to ensure it does not exceed a maximum speed. The vehicle may be, in particular, a driverless vehicle, especially an AGV (Automated Guided Vehicle).

[0028] Adjusting the safety measure affecting the protected segment can then consist, in particular, of not protecting the vehicle's movement, specifically by allowing the vehicle to travel through the protected segment at increased or full speed. This allows for a higher average vehicle speed within the monitored area compared to continuous slow travel, and the availability of a system encompassing the entire vehicle is particularly high.

[0029] In a further development of the procedure, the probability of a person being in the secured segment of the surveillance area is calculated using a stochastic model. Calculating the probability of being in the surveillance area using a stochastic model allows for the consideration of uncertainties regarding the person's position, particularly uncertainties in detecting the person's entry into the surveillance area.

[0030] The stochastic model can, in particular, specify the transition probabilities between the individual segments. The stochastic model can, for example, be implemented as a Markov chain. The state space of the Markov chain can comprise, as individual states, the presence of the person in the individual segments of the monitoring area. The transition probabilities between the individual segments then constitute the transition probabilities between the states of the Markov chain.

[0031] The state space of the Markov chain can also include entry states assigned to individual segments, where the entry states of each segment indicate from which predecessor segment the transition to the respective segment occurred. For each entry state of the individual segments, transition probabilities can be specified for each successor segment reachable from the respective segment. The transition probability assigned to a successor segment for an entry state of a given segment indicates the probability that the individual propagates from the predecessor segment associated with the entry state to the successor segment in question, via the segment under consideration. For example, if a segment has four neighboring segments, then the segment can be assigned a total of four entry states, one for each possible predecessor segment.Each of these four entry states can be assigned four transition probabilities, one for each possible successor segment.

[0032] In a further development of the procedure, adjusting the security measure affecting the secured segment involves reducing, or preferably eliminating, the security measure when the probability of presence falls below a predetermined threshold. This significantly increases the availability and efficiency of the system secured by the security measure.

[0033] A further development of the procedure involves acquiring sensor data from a sensor, where the sensor data represents the presence of a person in a segment of the monitored area. The probability of the person's presence in the monitored segment is adjusted based on the sensor data, specifically set to one if the person's presence is detected in the monitored segment. Additionally, the probability of presence in all other segments can be set to zero. This allows the probability of the person's presence in the segments of the monitored area to be adjusted to the person's actual position during the securing of the monitored area, thus improving the accuracy of the probability calculation.

[0034] The monitored segment can be the secured segment or any other segment of the monitored area. The monitored segment can also be the segment where the person enters the monitored area. The sensor can be statically positioned within the monitored segment or dynamically moved within the monitored area. The sensor can be, for example, a light barrier, a light curtain, a laser scanner, or similar device. The sensor can also be a wireless sensor for detecting a transmitter worn by the person, such as an RFID reader or scanner. The sensor can be configured to monitor a single segment, multiple segments, or all segments of the monitored area. Multiple sensors can also be positioned at different locations, each monitoring a different segment of the monitored area.

[0035] When calculating the time-dependent probability of a person's location, uncertainties in the representation of the person's position by the position data, or knowledge of the person's position that can only be partially derived from the position data, can be taken into account. In particular, unreliability or measurement inaccuracy of the sensor can be considered. Such uncertainties can, for example, be fed as input data to a stochastic model used in calculating the probability of a person's location.

[0036] In a further development of the method, the sensor is arranged on a device that can move through the monitored area, and the segment monitored by the sensor is formed, depending on the time, by a segment located within a detection range of the sensor. The device can, in particular, be the vehicle traveling through the monitored area. The sensor can, in particular, be a collision sensor arranged on the device, especially a distance sensor or laser scanner.

[0037] The sensor data can include position information, which makes it possible to assign the sensor data to the segment of the monitoring area being monitored by the sensor. The position information can, for example, represent the sensor's position or a specific time during data acquisition.

[0038] In this procedure, the monitoring area comprises intersection points and path segments along which the person can move between the intersection points. Calculating the probability of a person's location involves calculating the transition probabilities between the path segments connected at each intersection point. This allows information about the paths the person is likely to take within the monitoring area to be considered when calculating the probability of their location.

[0039] The intersection points and path segments can each form individual segments of the monitored area. In particular, the segments of the monitored area can consist of the intersection points and the path segments. However, one or all of the intersection points and / or path segments can also encompass multiple segments of the monitored area. The intersection points and path segments can be designed in such a way that a person can only move through the monitored area along the path segments and intersection points. For example, the intersection points and / or the path segments can be physically delimited, such as by separating elements like shelves, walls, fences, or the like.

[0040] At each intersection point, the transition probabilities for transitioning to the path segments following that intersection point can be equal. In particular, at each intersection point, the transition probabilities for transitioning to the path segments following that intersection point can be inversely proportional to the number of path segments connected via that intersection point. However, the transition probabilities can also be calculated based on the directed movement of the person along the path segments. This directed movement can, for example, be determined by a target position the person is aiming for.

[0041] The transition probabilities for each intersection point can also be specified depending on the path segment from which the person enters the intersection point. In this respect, multiple intersection states can be defined for each intersection point, one intersection state for each path segment adjacent to the intersection point that functions as a predecessor path segment or entry path segment. The intersection states of each intersection point can correspond to the entry states of the segments defined by that intersection point. The path segments adjacent to each intersection point can correspond to predecessor path segments and successor path segments of the segments defined by that intersection point.

[0042] For example, the directed transition probability that the person returns to the preceding path segment at the intersection point may be low, close to zero, or approximately zero. A further directed transition probability that the person transfers to a path segment that extends from the intersection point in a similar direction to the preceding path segment may also be low, close to zero, or approximately zero. This further directed transition probability may be higher than the directed transition probability for returning to the preceding path segment.

[0043] A second, further directed transition probability that the person transitions to a path segment opposite the preceding path segment at the intersection point, or forming an obtuse angle with the preceding path segment (e.g., an angle between 90° and 180°), may be elevated and thus greater than the directed transition probability for transitioning to the preceding path segment or the first further directed transition probability. A third further directed transition probability that the person transitions to a path segment connected to the target position at the intersection point may be particularly large, for example, close to one, approximately equal to one. In particular, the third further directed transition probability may be greater than the directed transition probability for transitioning to the preceding path segment and greater than the first and second further directed transition probabilities.

[0044] This method calculates the probability of a person's location within a specific target position. This allows for a particularly precise calculation of the person's location within the monitored area. When calculating the probability of location based on the target position, several or all paths leading to the target position along the segments of the monitored area can be considered, and the transition probabilities between the segments can be adjusted based on these paths. Specifically, the transition probabilities for individual segments can be adjusted depending on the paths leading through each segment. The target position can be predetermined based on the person entering the area or individually determined upon entry.

[0045] In a further development of the procedure, the target position is defined by a task to be performed by the person within the monitored area. Such a task can, for example, be stored in a control system implementing the safety measure or be entered into the control system. The task can also be determined by the control system, for example, based on an error message received by the control system. The error message could, for example, identify a segment of the monitored area where a repair measure is necessary, such as the repair of a device located in that segment, particularly a vehicle.

[0046] In this method, calculating the probability of finding a location involves determining the shortest path along the path segments from an initial position to the target position. This calculation uses a set of possible paths. Besides the shortest path, this set includes a number of additional paths between the initial and target positions. The number of additional paths includes another path if its length exceeds a minimum path length defined by the shortest path by no more than a predefined upper limit.

[0047] By specifying an upper limit, the number of paths considered can be restricted, and the possible paths within the path set can be calculated particularly efficiently. The upper limit can be defined such that the probability of a person taking a path to the target position longer than specified by the upper limit is sufficiently low. In particular, the upper limit can be defined based on a risk category of the mitigated hazard or on a safety level to be achieved through the safety measure. The upper limit can also be defined based on the minimum path length, for example, according to the formula L. max = L min * (1+x), where L max the upper limit, L min the minimum path length and x denote a predefined parameter, for example dependent on the risk category or the safety level.

[0048] Calculating the path lengths of all possible paths within the monitored area can be done using graphs. The monitored area can be viewed as a graph, where nodes represent the intersection points and edges represent the path segments. Edge weights can represent the distance or travel time between the intersection points connected by the individual edges. The shortest path can be calculated, for example, using Dijkstra's algorithm.

[0049] Only simple paths, i.e., paths that pass through each intersection point at most once, can be considered as possible or alternative paths. Furthermore, only paths that pass through a predetermined maximum number of intersection points can be considered as possible or alternative paths. This reduces the computational effort required to calculate the possible paths.

[0050] If the monitored area includes a blocked path segment along which the person cannot move temporarily or permanently, the blocked path segment can be disregarded when calculating possible or alternative paths. Only those paths that do not include the blocked path segment can then be considered as possible or alternative paths.

[0051] In a further development of the method, the transition probability from a first path segment to a second path segment is calculated at the intersection points as a function of the total number of paths in the path set that lead from the first path segment to the second path segment. The first path segment can, in particular, be the entry segment or a predecessor segment or predecessor path segment of the intersection point, and the second path segment can, in particular, be a successor segment or successor path segment of the intersection point.

[0052] In a further development of the procedure, the transition probability from the first path segment to the second path segment at an intersection connecting the first and second path segments includes a summand that is formed by multiplying a given probability for a directed choice of the next path segment by the total number of paths leading from the first path segment to the second path segment at the intersection point, and dividing by the total number of paths leading from the first path segment via the intersection point. This adds to the transition probability for the transition from the first to the second path segment a component of the given probability for the directed choice of the next path segment that is proportional to the number of paths leading from the first to the second path segment.

[0053] If the intersection point, as the connection point, connects the first path segment, the second path segment, and further path segments, then the individual transition probabilities for a transition from the first path segment to each of the further path segments can each comprise a summand. This summand is formed by multiplying the given probability for the directed choice of the next path segment by the total number of paths leading from the first path segment to each of the further path segments at the connection point, and dividing by the total number of paths leading from the first path segment via the connection point. Similarly, transition probabilities for a transition from the second path segment or for a transition from each of the further path segments to the remaining path segments connected to the connection point can also comprise a summand formed analogously.

[0054] In a further development of the procedure, the transition probability from the first path segment to the second path segment at the connection point includes an additional term, which is given by a predetermined probability of randomly selecting the next path segment divided by the total number of path segments connected to the connection point. This distributes the probability of randomly selecting the next path segment evenly among the path segments connected to the connection point.

[0055] If the intersection point, as the connection point, connects the first path segment, the second path segment, and further path segments, then the transition probabilities for a transition from the first path segment to the further path segments can each include the summand given by the given probability of randomly choosing the next path segment divided by the total number of path segments connected by the connection point. Similarly, the transition probabilities for a transition from the second path segment, or for a transition from each of the further path segments to the remaining path segments connected by the connection point, can also include the summand given by the given probability of randomly choosing the next path segment divided by the total number of path segments connected by the connection point.

[0056] A control system for securing a monitored area is described, wherein the monitored area is divided into several segments. The control system is configured to execute a security measure pertaining to the monitored area. The control system comprises a sensor and a risk control device. The sensor is configured to detect when a person enters the monitored area. The risk control device is configured to calculate a time-dependent probability of the person being in a secured segment of the monitored area and to adjust a security measure pertaining to the secured segment based on this probability.

[0057] The control system can be configured, in particular, to execute the method according to the invention. In particular, all further developments and advantages described in connection with the method according to the invention also relate to the control system according to the invention.

[0058] The control system may include a controller to operate an actuator of the control system that is protected by the safety measure. The risk control device may be configured to specify the safety measure to be taken by the controller, depending on the probability of the person being in the protected segment. The controller may be configured to take the safety measure by default, unless, during the process of adjusting the safety measure, the controller receives a control signal from the risk control device indicating that the safety measure needs to be adjusted, for example, lifted.

[0059] The risk control device can be implemented as a secure device, for example, on secure hardware and / or using secure software, whereby the secure device implements all security measures required by the relevant security standards, such as redundancy or diversity. Likewise, the sensor and / or the actuator and / or the controller can each be implemented as a secure device. The risk control device and the controller can each be implemented as logic units, for example, as microcontrollers, FPGAs, ASICs, or the like. The risk control device and the controller can be implemented as a single unit or separately and connected to each other via a data connection, in particular a secure one.

[0060] The invention is explained below with reference to figures. These figures are shown schematically: Fig. 1 a monitoring area divided into segments; Fig. 2 intersection points and path segments of the monitoring area; Fig. 3 a control system for securing the monitored area; Fig. 4 a connection point of the monitoring area with a first, second, third and fourth path segment and with paths leading via the connection point; and Fig. 5 a procedure for securing the surveillance area.

[0061] Fig. Figure 1 shows a monitoring area 1 subdivided into segments 2. The monitoring area 1 is located in a hall and structurally divided into segments 2. Separating elements 8, such as wall elements or shelves, are arranged between the segments 2, forming aisles 60 between the separating elements 8 and intersections 61 connecting the aisles 60. A vehicle 12, for example an AGV 12, and a person 5 can move along the segments 2.

[0062] To secure monitoring area 1, in particular to secure the movement of vehicle 12 along segments 2, the entry of person 5 into monitoring area 1 is detected, a time-dependent probability of person 5 being in a secured segment of monitoring area 1 is calculated, and a security measure relating to the secured segment is adjusted depending on the probability of person 5 being in the secured segment. In particular, several of the segments 2, for example, all segments 2 encompassing the intersections 61, or each of the segments 2, can be considered a secured segment, and the probability of being in each or all of the segments 2 of monitoring area 1 can be calculated, and a corresponding security measure for each or all of the segments 2 can be adjusted depending on the corresponding probability of person 5 being in the respective segment 2.The safety measures may in particular include a slow-speed driving of the vehicle 12 in the secured segments 5, especially in the segments 2 comprising the intersections 61.

[0063] Fig. Figure 2 shows intersection points 3 and the path segments 4 connecting the intersection points 3 within the monitoring area 1. The intersection points 3 are located in segments 2 comprising the intersections 61, and the path segments 4 are located in segments 2 comprising the aisles 60. The entry of person 5 into the monitoring area 1 is detected by a first sensor 20. The first sensor 20 detects an initial position 6 of person 5. The initial position 6 can be located at one of the intersection points 3. The initial position 6 is located in a segment 2 of the monitoring area 2 that forms a starting segment.

[0064] Starting from position 6, a shortest path 31 is calculated along which person 5 moves from position 6 to a target position 7. The target position 7 can be formed by one of the segments 2 or by one of the intersection points 3 of the monitoring area 1. A set of possible paths is also calculated. This set includes a first additional path 32, whose length does not exceed a minimum path length defined by the shortest path 31 by more than a predefined upper limit. If the shortest path 31 and the first additional path 32 have the same length, one of the paths 31, 32 is selected, for example, randomly, as the shortest path 31. A second additional path 33, whose length is greater than the predefined upper limit, is not included in the set of possible paths.

[0065] In addition to the first sensor 20, a second sensor 22 and a third sensor 24 are arranged within the monitoring area 1. The second sensor 22 is fixed within the monitoring area 1, and the third sensor 24 moves dynamically within the monitoring area 1. Specifically, the third sensor 24 is attached to the vehicle 12, which moves through the monitoring area 1, and is configured as a collision sensor for the vehicle 12. The first, second, and third sensors 20, 22, and 24 are configured to detect the presence of person 5 in a segment monitored by the first sensor 20, in a segment monitored by the second sensor 22, and in a segment monitored by the third sensor 24, respectively, within the monitoring area 1.If the presence of person 5 is detected in the segment monitored by the first sensor 20, in the segment monitored by the second sensor 22, or in the segment monitored by the third sensor 24, the probability of person 5 being in the relevant segment is adjusted, namely set to one.

[0066] Fig. Figure 3 shows a control system 200 for securing the monitored segments 2 of the monitoring area 1. The control system 200 comprises a controller 210, which transmits control information 212 to an actuator driving the vehicle 12. The control information 212 can be used, for example, to control the speed of the vehicle 12. The controller 210 is connected to a risk control device 220 of the control system 200. The risk control device 220 is configured to transmit a control signal 221 to the controller 210, which specifies the safety measures to be taken in the secured segments 2. In particular, the risk control device 220 is configured to adjust the safety measures by means of the control signal 221, for example, to cancel them.

[0067] The risk control device 220 is further configured to calculate, according to a model 230, a time-dependent probability of presence 231 of person 5 in the secured segments 2 of the monitoring area 1. The time-dependent probability of presence 231 is calculated depending on whether person 5 enters the monitoring area 1. For this purpose, the risk control device 220 is connected to the model described in Fig. The first sensor 20 (not shown) is connected and receives initial sensor data 201 from the first sensor 20, indicating that person 5 has entered the monitoring area 1. Based on this initial sensor data 201, the risk control device 290 adds presence information 224 to the model 230, representing person 5's entry into the monitoring area 1.

[0068] Will the presence of person 5 be confirmed by the in Fig. 3 not shown second sensor 22 monitored segment or in which the in Fig. If the third sensor 23 (also not shown) detects a monitored segment, the second sensor 22 transmits corresponding second sensor data 202, or the third sensor 23 transmits corresponding third sensor data 203, to the risk control device 220. Depending on the second or third sensor data 202 or 203, the risk control device 220 adds location information 223 to the model 230, representing the location of person 5 in the relevant monitored segment 2 of the monitoring area 1. Based on this location information 223, the time-dependent probability of person 5 being in the relevant monitored segment 2 is adjusted.

[0069] The risk control device 220 is configured to calculate the probability of person 5's location 231 depending on the target position 7 sought by person 5. For this purpose, the risk control device 220 adds target information 220, representing the target position 7, to the model 230. The target position 7 is specified based on a task to be performed by person 5 within the monitoring area 1. Task information 214, representing the task, is transmitted from the controller 210 to the risk control device 220. Based on the target position 7, the set of possible paths 31, 32 along which person 5 moves to the target position 7 can be determined using the model 230, and path information 232, representing the path set, can be transmitted to the risk control device 220.

[0070] Fig. Figure 4 shows one of the intersection points 3 of the monitoring area 1, which connects a first path segment 41, a second path segment 42, a third path segment 43 and a fourth path segment 44 as a connection point 40.

[0071] Model 230 assigns a crossing state to connection point 40 for each path segment 41, 42, 43, 44. Specifically, model 230 assigns a first crossing state, a second crossing state, a third crossing state, and a fourth crossing state to connection point 40. The first crossing state represents person 5 entering connection point 40 from the first path segment 41 (the entry path segment), the second crossing state represents entry from the second path segment 42 (the entry path segment), the third crossing state represents entry from the third path segment 43 (the entry path segment), and the fourth crossing state represents entry from the fourth path segment 44 (the entry path segment).

[0072] Transition probabilities are assigned to each intersection state, with one transition probability being assigned to each intersection state for each path segment 41, 42, 43, 44 connected to the connection point 40. For each intersection state, the transition probabilities represent the probability that the person will transition from the entry path segments assigned to the individual intersection states to the path segment 41, 42, 43, 44 assigned to the respective transition probability.

[0073] For example, the first crossing state is assigned a transition probability for a transition from the first path segment 41 back to the first path segment 41, a transition probability for a transition from the first path segment 41 to the second path segment 42, a transition probability for a transition from the first path segment 41 to the third path segment 43 and a transition probability for a transition from the first path segment 41 to the fourth path segment 44.

[0074] If model 230 merely represents a random, undirected propagation of person 5 along path segments 4, and if the transition probabilities are calculated independently of the target position 7, then the transition probabilities of the intersection states at connection point 40 all have the same value. This value is one divided by the number of path segments 41, 42, 43, 44 connected at connection point 40, i.e., 1 / 4.

[0075] If the transition probabilities are specified depending on the target position 7, then a probability p for a directed choice of the next path segment is included in model 230. A probability q for a random, undirected choice of the next path segment is then given by q = (1-p). The transition probabilities initially all comprise a summand, which corresponds to the probability q divided by the number of path segments 41, 42, 43, 44 connected to the connection point 40, i.e., q / 4. This summand represents an undirected movement of person 5 along the path segments 41, 42, 43, 44.

[0076] For each crossing state of the connection point 40, the probability p for the directed transition is divided among the transition probabilities of the crossing state in such a way that the proportion allocated to each transition probability is proportional to the total number of paths leading from the entry path segment associated with the crossing state to the path segment associated with the transition probability, and inversely proportional to the total number of paths leading from the entry path segment associated with the crossing state via the connection point 40.

[0077] Through junction 40, starting from the first path segment 41 (the entry path segment), two first paths 51 of the path set lead to the third path segment 43, and one second path 52 leads to the fourth path segment 44. Furthermore, starting from the second path segment 42 (the entry path segment), three third paths 53 lead to the third path segment 43, and two fourth paths 54 lead to the fourth path segment 44. Since none of the paths 51, 52, 53, 54 of the path set lead through junction 40, starting from the third path segment 43 or the fourth path segment 44, no further summands are added to the transition probabilities associated with the third and fourth crossing states.

[0078] The transition probability of the first intersection state, representing a transition from the first path segment 41 to the third path segment 43, is augmented by a fraction of 2 / 3*p of probability p. The transition probability of the first intersection state, representing a transition from the first path segment 41 to the fourth path segment 44, is augmented by a fraction of 1 / 3*p of probability p. The transition probability of the second intersection state, representing a transition from the second path segment 42 to the third path segment 43, is augmented by a fraction of 3 / 5*p of probability p. The transition probability of the second intersection state, representing a transition from the second path segment 42 to the fourth path segment 44, is augmented by a fraction of 2 / 5*p of probability p.

[0079] The transition probabilities for the crossing states of the remaining crossing points 3 of the monitoring area 1 are calculated analogously to the described calculation of the transition probabilities of the crossing states of the crossing point 3 forming the initial position 6 and the crossing point 3 forming the target position 7, with the exception of the transition probabilities for the crossing states of the crossing point 3 forming the initial position 6 and the crossing point 3 forming the target position 7. Fig. The connection point 40 shown in Figure 4 is calculated. Since all paths begin at the initial position 6 and there is no directed entry to the initial position 6, all entry path segments of the intersection states of the intersection point 3 forming the initial position 6 are treated as entry path segments for all paths in the path set.

[0080] For the intersection point 3 forming the target position 7, model 230 includes a dwell parameter that specifies the probability that the person remains at the target position 7 during one of the time steps of model 230. In this respect, the dwell parameter parameterizes the duration of the person's stay at the target position 7. The transition probabilities of all those intersection states of intersection point 3 containing the target position 7, whose entry path segments are included in one of the possible paths, are reduced by means of the dwell parameter in order to account for the person's stay at the target position 7.

[0081] Fig.Figure 5 shows a procedure 100 for securing the monitoring area 1. The procedure 100 includes subdividing 101 the security area 1 into segments 2. Furthermore, the procedure 100 includes detecting 105 the entry of person 5 into the monitoring area 1, detecting 107 the presence of person 5 in the segments 2 of the monitoring area 1 monitored by sensors 20, 22, 24, calculating 110 the probability 231 of person 5 being in the secured segments 2 of the monitoring area 1, and adjusting 120 the security measures relating to the monitored segments 2.

[0082] Calculating the probability of presence (110) comprises calculating the transition probabilities between segments 2 of the monitoring area 1. This calculation involves first calculating the shortest path 31 from the initial position 6 to the target position 7, and then determining the set of possible paths 31, 32 from the initial position 6 to the target position 7. Adjusting the security measure (120) comprises comparing the probability of presence (231) of person 5 in the segments 2 of the monitoring area 1 secured by the security measure with a predetermined threshold, and reducing the security measure (124) if the probability of presence (231) falls below the predetermined threshold.Furthermore, adjusting the safety measure (120) includes maintaining the safety measure (125) if the probability of presence (231) does not fall below the specified threshold. Adjusting the safety measure (120) also includes outputting the control signal (130) to the controller (210). Reference symbol list 1 monitored area 2 segments 3 Intersection point 4 path segment 5 person 6 Starting position 7 Target position 8 separating element 9 hedged segment 10 devices 12 vehicles 20 first sensor 22 second sensor 24 third sensor Path 30 31 shortest path 32 first further path 33 second further path 40 connection point 41 first path segment 42 second path segment 43 third path segment 44 fourth path segment 51 first paths 52 second paths 53 third paths 54 fourth paths 60 gears 61 Intersection 100 procedures 101 Subdivisions 105 Recording an occurrence 107 Acquiring sensor data 110 Calculating a probability of residence 112 Calculating transition probabilities 114 Calculating a shortest path 116 Determining a path set 120 Adjust 122 Comparisons 124 Decrease 125 Retain Spend 130 200 control system 201 first sensor data 202 second sensor data 203 third sensor data 210 Control 212 Tax Information 214 Task Information 220 Risk Control Facility 221 Control signal 222 Destination Information 223 Accommodation Information 224 Attendance information 230 model 231 Probability of being there 232 Path information

Claims

[1] Method (100) for securing a monitoring area (1), comprising the following steps: - Subdividing (101) the monitoring area (1) into several segments (2), - Detecting (105) the entry of a person (5) into the surveillance area (1); characterized by : - Calculating (110) a time-dependent probability of residence (231) of the person (5) in a secured segment (2) of the several segments (2) of the surveillance area (1), - Adapting (120) a security measure relating to the secured segment (2) depending on the probability (231) of the person (5) being in the secured segment (2), wherein the monitoring area (1) includes intersection points (3) and path segments (4) along which the person (5) can move between the intersection points (3), where the calculation (110) of the probability of being located (231) includes a calculation (112) of transition probabilities between the path segments (4) connected at the intersection points (3), where the probability of residence (231) is calculated depending on a target position (7) sought by the person (5), where calculating (110) the probability of being located (231) includes calculating (114) a shortest path (31) along the path segments (4) from an initial position (6) to the target position (7), where, in calculating (110) the probability of being located (231), a set of possible paths (31, 32) is used, where the path set includes, in addition to the shortest path (31), a number of additional paths (32) between the starting position (6) and the target position (7), where the number of additional paths (32) includes another path (32, 33) if a path length of the further path (32, 33) exceeds a minimum path length specified by the shortest path (31) by a maximum of a specified upper limit. [2] Method (100) according to claim 1, wherein the monitoring area (1) is traversed by a vehicle (12), wherein the safety measure comprises securing a movement of the vehicle (12), preferably a slow movement of the vehicle (12), within the secured segment (2). [3] Method (100) according to one of the preceding claims, wherein the probability (200) of the person (5) being in the secured segment (2) of the monitoring area (1) is calculated using a stochastic model (230). [4] Method (100) according to one of the preceding claims, wherein the adjustment (120) of the safety measure relating to the secured segment (2) comprises a reduction, preferably a cancellation, of the safety measure when the probability of presence (231) falls below a predetermined limit value. [5] Method (100) according to one of the preceding claims, wherein the method (100) comprises acquiring (107) sensor data (201, 202, 203) from a sensor (20, 22, 24), wherein the sensor data (201, 202, 203) represent the presence of the person (5) in a segment (2) of the monitoring area (1) monitored by the sensor (20, 22, 24), wherein the probability of being (231) of the person (5) in the monitored segment (2) is adjusted on the basis of the sensor data (201, 202, 203), in particular set to one when the presence of the person (5) in the monitored segment (2) is detected. [6] Method (100) according to claim 5, wherein the sensor (20, 22, 24) is arranged on a device (12) that can be moved through the monitoring area (1), wherein the segment (2) monitored by the sensor (20, 22, 24) is formed in a time-dependent manner by a segment (2) arranged within a detection range of the sensor (20, 22, 24). [7] Method (100) according to one of the preceding claims, wherein the target position (7) is specified by a task to be performed by the person (5) in the monitoring area (1). [8] Method (100) according to one of the preceding claims, wherein at the intersection points (3) the transition probability from a first path segment (4, 41) to a second path segment (4, 42) is calculated as a function of a total number of paths (30) contained in the path set that lead from the first path segment (4) to the second path segment (4). [9] Method (100) according to claim 8, wherein the transition probability from the first path segment (4, 41) to the second path segment (4, 42) at a junction point (3) connecting the first and the second path segment (4, 41, 42) as a junction point (40) comprises a summand formed by a given probability for a directed choice of a next path segment (4) multiplied by the total number of paths (30) leading from the first path segment (4, 41) to the second path segment (4, 42) at the junction point (30), and divided by a total number of paths (30) leading from the first path segment (4, 41) via the junction point (40). [10] Method (100) according to claim 9, wherein the transition probability from the first path segment (4, 41) to the second path segment (4, 42) at the connection point (40) comprises a further summand which is given by a predetermined probability for a random choice of the next path segment (4) divided by a total number of path segments (4) connected to the connection point (40). [11] Control system (200) for securing a monitoring area (1), wherein the monitoring area (1) is divided into several segments (2), wherein the control system (200) is configured to execute a safety measure relating to the monitoring area (1), wherein the control system (200) comprises a sensor (20) and a risk control device (220), wherein the sensor (20) is designed to detect the entry of a person (5) into the monitoring area (1), characterized by , that the risk control device (220) is trained to calculate a time-dependent probability of the person (5) being in a secured segment (2) of the several segments (2) of the surveillance area (1), that the risk control device (220) is trained to adapt a security measure relating to the secured segment (2) depending on the probability (231) of the person (5) being in the secured segment (2), that the monitoring area (1) includes intersection points (3) and path segments (4) along which the person (5) can move between the intersection points (3), that the calculation (110) of the probability of being located (231) includes a calculation (112) of transition probabilities between the path segments (4) connected at the intersection points (3), that the probability of residence (231) is calculated depending on a target position (7) sought by the person (5), that calculating (110) the probability of being located (231) includes calculating (114) a shortest path (31) along the path segments (4) from an initial position (6) to the target position (7), that in calculating (110) the probability of being at the location (231) a set of possible paths (31, 32) is used, that the path set includes, in addition to the shortest path (31), a number of additional paths (32) between the starting position (6) and the target position (7), and that the number of additional paths (32) includes another path (32, 33) if a path length of the further path (32, 33) exceeds a minimum path length specified by the shortest path (31) by a maximum of a specified upper limit.

Images