Access control system, structure having the access control system, and method for access control

Abstract

The invention relates to an access control system (14) for one or more access-restricted spatial regions (10, 10', 10"), for example rooms or buildings, having at least one data evaluation unit (12) which is provided to identify access anomalies and / or location anomalies for at least one of the access-restricted spatial regions (10, 10', 10") by means of the access control system (14) on the basis of an analysis of one or more sensor data sets.

Description

[0001] Access control system, building with the access control system and procedure for access control

[0002] State of the art

[0003] The invention relates to an access control system according to claim 1, a structure according to claim 28 and a method according to claim 29.

[0004] Access control systems that monitor entry to and / or exit from restricted areas are known. However, for these systems, the process typically ends after identification and granting of entry and / or exit. Further use of the entry and / or exit data to record or monitor the behavior of the individuals concerned, or similar purposes, is generally not employed.

[0005] The object of the invention is, in particular, to provide a generic device with advantageous safety features. This object is achieved according to the invention by the features of the independent and dependent claims, while advantageous embodiments and further developments of the invention can be found in the dependent claims.

[0006] Advantages of the invention

[0007] An access control system for one or more restricted areas, such as rooms or buildings, is proposed. This system includes at least one data evaluation unit designed to detect access and / or presence anomalies for at least one of the restricted areas based on the analysis of one or more sensor data sets. This allows for the advantageous achievement of a high level of security, particularly regarding access to the restricted area and / or the safety and health of the individuals within it. For example, unauthorized presence, unusual behavior patterns, or sudden health problems of individuals within the restricted area can be advantageously detected.

[0008] The access control system is designed, in particular, as a system that is at least intended to control and / or regulate access to and / or exit from the restricted area. Preferably, the access control system is at least intended to control and / or manage at least one barrier, such as a door, gate, turnstile, barrier, etc., that makes access to the restricted area more difficult and / or physically controls access. Preferably, the access control system includes at least one locking system for locking and / or unlocking at least one of the barriers. The access control system can also form part of a higher-level locking system. The access control system can, in particular, include reading devices, such as card readers, biometric scanners, PIN keypads, etc.The access control system may, in particular, include a processing unit that verifies authorizations and, depending on the verification, grants or denies access to and / or exit from the restricted area. The processing unit may be a physical (central) server located on-site or remotely. Alternatively, the processing unit may be a cloud-based platform. Preferably, the data processing unit is at least part of the processing unit or identical to it. A "processing unit" is understood to be, in particular, a unit with information input, information processing, and information output. Advantageously, the processing unit includes at least a processor, memory, input and output devices, other electrical components, an operating program, control routines, and / or calculation routines.The term "intended" should be understood to mean, in particular, specifically programmed, designed, and / or equipped. The fact that an object is intended for a specific function should be understood to mean, in particular, that the object fulfills and / or performs this specific function in at least one application and / or operating state. Access and / or presence anomalies should also be understood to include, in particular, behavioral anomalies of persons within the access-restricted area.

[0009] The restricted area can be of many different types, for example, company premises and / or company buildings, a militarily or police-secured area, a building or site belonging to a government agency, ministry, embassy, ​​organization, health institution, production facility, industrial plant, data center, financial institution, infrastructure facility such as airports or train stations, educational institution, residential complex such as a gated community, sports ground, event venue, amusement park, holiday resort, shopping center, museum, historical site, construction site, etc. The sensor data sets can be time series from dedicated sensors.The sensor data sets may include data from components of the access control system that are not explicitly intended for monitoring environmental parameters, such as data (entry times, exit times, captured IDs, etc.) from the access control system's reader. The sensor data sets may also include data from sensors or other instruments primarily intended for other tasks unrelated to the access control system, such as smoke detectors, humidifiers and / or dehumidifiers, ventilation systems, heating systems, electricity meters, motion detectors for lights, carbon monoxide sensors, other smart home devices, etc.

[0010] An access and / or presence anomaly can be, in particular, any unusual, atypical, or irregular activity related to entering, remaining in, or leaving a restricted area, especially if it deviates from known (learned) patterns. Access and / or presence anomalies can indicate security problems, health issues, confusion, dangerous situations, attempted security breaches, or simply technical malfunctions.

[0011] Examples of access and / or residence anomalies include, but are not limited to:

[0012] - Unusual access times: If a person attempts to gain access to an area outside of its usual access times (e.g., an employee requesting access late at night),

[0013] - Multiple access attempts: Repeated attempts to gain access to a restricted area, especially if they fail,

[0014] - Attempts to gain access to unusual locations: Attempts to gain access to areas to which the person has no authorization or which are not normally entered by that person,

[0015] - Invalid access attempts: Use of invalid identification media or incorrect access codes,

[0016] Sudden changes in access behavior: Unusual changes in a person's access pattern, such as more or less frequent access, access attempts with stolen or lost identification media: Attempts to gain access with stolen or lost cards or ID cards,

[0017] - Anomalies in biometric data: Failures in biometric scans that could indicate attempts to deceive the systems (e.g., use of fake fingerprints),

[0018] - Suspicious movement patterns: Unexplained movements within the monitored area that could indicate unauthorized activity,

[0019] - Simultaneous access attempts: Attempts to gain access at different locations at the same time using the same login credentials, which could indicate duplicate use or misuse of the login credentials,

[0020] - Return to the area after a short time: Multiple access attempts in quick succession by the same person,

[0021] - Rapid movement through several access-restricted sub-areas of the access-restricted spatial area: flight-like movement of people,

[0022] - Frequent attempts with expired authorizations,

[0023] - frequent attempts with different identification media,

[0024] - Access to specially secured areas: Unusual attempts to gain access to high-security areas that are normally only accessible to a few people,

[0025] - Misuse of guest access: Unusually frequent or unexpected use of guest access or temporary authorizations; Attempts following security incidents: Access attempts immediately after known security incidents or alarms,

[0026] - Behavior after working hours: Attempts to access office buildings or work areas outside of regular working hours, especially on weekends or public holidays,

[0027] - Use of emergency exits: Use of emergency exits or other non-regular access points, especially without apparent reason,

[0028] - Changes in access patterns during holidays or absences: If access attempts occur while a person is officially on holiday or otherwise absent,

[0029] - Unusually long stay: If a person stays in an area for an unusually long time, especially if there is no obvious reason for doing so,

[0030] - Unauthorized access attempts by external parties: Attempts by external persons or service providers to gain access to protected areas without appropriate authorization,

[0031] - Unusual movement patterns within the building: movements that deviate from the person's normal route or usual patterns of behavior,

[0032] - Use of multiple identification media: If a person uses different access cards or media within a short period of time, which could indicate an attempt to circumvent security measures,

[0033] - Access attempts after permission changes: Attempts to gain access shortly after access permissions have been changed or restricted; unusual times for guest access: Use of guest access at unusual times that do not coincide with typical visiting times.

[0034] - Access by multiple people in unusual combinations,

[0035] - Access attempts during system maintenance: increased attempts to gain access during known maintenance times of the access control system, when the system may be more vulnerable,

[0036] - Increased access attempts shortly after staff changes: Unusually frequent access attempts immediately after a change in security or administrative personnel,

[0037] - Deviant use of parallel entrances: Use of parallel entrances that are not normally used by a person for access,

[0038] - Repeated false alarms: frequent triggering of false alarms by access attempts, which could indicate deliberate attempts to disrupt the system,

[0039] - Unexplained movements during alarm states: Movements or access attempts during or immediately after an alarm is triggered that do not match the known movement patterns of authorized persons,

[0040] - Use of identification media that have not been used for a long time: Experiments with access cards or media that have an unusual activity history and are suddenly used actively,

[0041] - Access attempts from unusual directions: Access requests made from a direction or position that is not normally used; unusual frequency of barrier or door openings: Barriers or doors that are normally rarely used are suddenly opened frequently;

[0042] - Increased use of spare keys or emergency access: Unusually frequent use of spare keys or emergency access that are normally only used in exceptional cases,

[0043] - unusual use of elevators,

[0044] - Attempts to gain access while wearing protective equipment: Attempts by individuals in unusual protective clothing or disguises to gain access, which makes their identification difficult,

[0045] - Irregular use of employee ID cards: Use of employee ID cards by people who are not normally in the area in question,

[0046] - Unusual movements in sensor-monitored areas: Detection of movements in sensor-monitored areas that are normally rarely or never frequented,

[0047] - Increased activity in the run-up to scheduled events: Attempts to gain access to or movement in areas related to upcoming scheduled events, which could indicate possible preparatory actions.

[0048] The above list is merely exemplary and should not be understood as a complete list of possible access and / or residence anomalies.

[0049] Furthermore, it is proposed that the analysis for detecting access and / or presence anomalies includes an evaluation of one or more sensor datasets by a trained machine learning algorithm of the data evaluation unit. This advantageously optimizes the detection of access and / or presence anomalies. The trained machine learning algorithm can advantageously detect new, previously unnoticed access and / or presence anomalies. The range of detectable access and / or presence anomalies can be significantly expanded. Various trained machine learning algorithms can be used for the detection of access and / or presence anomalies. For example, the trained machine learning algorithm could be based on a random forest method. Random forests are advantageously able to handle large amounts of data and are robust against noise.For example, the trained machine learning algorithm could be based on a Support Vector Machine (SVM) method. SVMs are particularly advantageous when it comes to separating normal and abnormal access behavior.

[0050] For example, the trained machine learning algorithm could be based on a neural network. Neural networks are advantageous because they can recognize complex patterns and relationships in data; deep learning networks, in particular, can be advantageously effective at processing large and complex datasets. For example, the trained machine learning algorithm could be based on a gradient boosting machine (GBM) method: GBMs are advantageously very powerful at detecting anomalies. For example, the trained machine learning algorithm could be based on a K-means clustering method. K-means clustering algorithms are advantageous because they can identify unusual behavioral patterns by grouping data points into clusters and defining anomalies as points that are far from the centroids of the clusters. For example, the trained machine learning algorithm could be based on an isolation forest method.Isolation forest methods, specifically designed for anomaly detection, are advantageous. For example, the trained machine learning algorithm could be based on an LSTM (Long Short-Term Memory) network. LSTM networks are particularly well-suited for sequential data and are very useful for detecting anomalies in data containing temporal dependencies. Alternatively, other well-known machine learning algorithms, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), autoencoders, label propagation, one-class SVM, deep Q-learning, or ARIMA (AutoRegressive Integrated Moving Average), could be used. Furthermore, the use or combination of several different types of machine learning algorithms, or the use of known hybrid models, is also conceivable.The trained machine learning algorithms could also be improved by integrating further techniques such as feature engineering, data preprocessing, and model assembly. The choice of the best algorithm depends on the specific requirements, the availability and quality of the data, and the desired accuracy and efficiency of anomaly detection, and lies primarily within the realm of the expert's knowledge.

[0051] Furthermore, it is proposed that the access control system include at least one sensor unit designed to capture one or more sensor data sets. This allows for particularly targeted detection of access and / or presence anomalies. Advantageously, sensor data sets specifically suited for access and / or presence anomaly detection can be generated. In particular, the sensor unit comprises one or more sensors or sensor modules. It is conceivable that the access control system includes several identical and / or several different sensor units, which are arranged in different positions, for example, in different sub-areas of the restricted area or assigned to different barriers within the restricted area.

[0052] If the sensor unit includes at least one indoor temperature sensor designed to acquire data from one of the sensor data sets, and the sensor data set is configured as an indoor temperature data set comprising temperatures measured by the indoor temperature sensor within the access-restricted area, then access and / or occupancy anomaly detection can advantageously be enabled, based at least partially on acquired indoor temperature time series. This allows for particularly effective access and / or occupancy anomaly detection. For example, an increase in indoor temperature can indicate an unusually high number of people in an area. Conversely, a decrease in indoor temperature can indicate an unusually low number of people in an area.For example, an increase in indoor temperature may indicate unusually high activity from machines or computing devices in a given area (computer overload, execution of very resource-intensive calculations / programs, etc.). Conversely, a decrease in indoor temperature may indicate unusually low activity from machines or computing devices in a given area (device failure or completion of a complex calculation, etc.). A decrease in indoor temperature may also indicate unauthorized access to a room (e.g., through a broken window, a security door left open, or similar means).

[0053] Furthermore, if the sensor unit includes at least one outdoor temperature sensor designed to acquire data from one of the sensor data sets, wherein the sensor data set is configured as an outdoor temperature data set comprising temperatures measured by the outdoor temperature sensor in an environment directly surrounding the access-restricted area or in a room of a building adjacent to the access-restricted area, the access and / or occupancy anomaly detection based on the indoor temperature data can be advantageously further optimized. The outdoor temperature data set can, for example, be used to evaluate the indoor temperature data from the indoor temperature data set or to compare it with simultaneously measured indoor temperature data.For example, it can be recorded whether a detected increase or decrease in indoor temperature correlates with or deviates from a rise or fall in temperatures in adjacent areas. A deviation can be a reinforcing indicator of an access and / or occupancy anomaly.

[0054] Particularly in connection with the temperature sensors, it is also proposed that the trained machine learning algorithm be trained, at least, to detect and / or classify one or more of the access and / or occupancy anomalies in the indoor temperature dataset, especially in conjunction with the outdoor temperature dataset. This would advantageously enable reliable and / or precise detection of access and / or occupancy anomalies. Among other things, the trained machine learning algorithm could be trained using historical temperature data (indoor and outdoor) that includes typical temperature profiles under normal conditions in the areas monitored by the temperature sensors.This should enable the trained machine learning algorithm to distinguish between typical / normal temperature increases and / or decreases and / or temperature differences and unusual ones, particularly those indicating anomalies in access and / or occupancy. It is also conceivable that known potential anomalies in access and / or occupancy could be artificially generated and the resulting temperature data incorporated into the training of the machine learning algorithm. The indoor and / or outdoor temperature sensor can be a thermometer, specifically a digital thermometer.

[0055] Furthermore, it is proposed that the sensor unit include at least one acoustic sensor designed to acquire one of the sensor data sets, wherein the sensor data set is configured as a noise data set comprising noise levels, noise frequencies, and / or noise patterns measured by the acoustic sensor within the access-restricted area. This advantageously enables particularly good detection of access and / or occupancy anomalies. For example, a high overall noise level may indicate an unusually high number of people in an area. For example, a low overall noise level may indicate an unusually low number of people in an area. For example, an increase in overall noise level may indicate an event (accident, problem, etc.) in an area. For example, an abrupt cessation of all noise may indicate an event (accident, health problem, etc.).) in an area. For example, certain sound frequencies or patterns may indicate anomalies in access and / or presence (e.g., the recording of external sounds such as birdsong or insect stridulation may indicate open windows or security doors, or a broken window, etc.). For example, certain sound patterns may indicate exceptional emotional states, unusual emotional states, or other abnormal behavior of a person in the area (e.g., whispering, sobbing, sniffling, shouting, shrieking, rapid breathing, loud breathing, etc.). For example, certain sound patterns may indicate abnormal movement patterns of a person in the area (e.g., rapid or irregular footsteps, shuffling sounds, falling sounds, etc.).

[0056] If, in this context, the trained machine learning algorithm is trained to detect and / or classify one or more of the access and / or presence anomalies in the noise data set, reliable and / or precise detection of these anomalies can be advantageously achieved. Among other things, the trained machine learning algorithm could be trained using historical noise data, particularly noise levels, frequencies, and / or patterns, typical for normal operation / use of the restricted area. This should enable the trained machine learning algorithm to distinguish between typical / normal noise levels and / or unusual noise levels, especially those indicating access and / or presence anomalies.It is also conceivable that known potential access and / or presence anomalies could be artificially generated, and the resulting noise levels, frequencies, and / or patterns could be incorporated into the training of the machine learning algorithm. The acoustic sensor could be a single microphone or an array of microphones.

[0057] It is further proposed that the sensor unit include at least one optical sensor for capturing one of the sensor data sets, wherein the sensor data set is configured as an image data set comprising movements observed by the optical sensor within the access-restricted area and / or brightness levels captured by the optical sensor within the access-restricted area. This advantageously enables particularly good detection of access and / or presence anomalies. For example, unusually high or low movement activity, or an abrupt change in the detected movement activity, can indicate the presence of an access and / or presence anomaly. For example, an abrupt cessation of a person's movement activity can indicate an event (accident, health problem, etc.) in a certain area.For example, unusual behavior patterns of individuals included in the image data may indicate an access and / or location anomaly. It is conceivable that, to comply with data protection regulations, the optical sensor does not record individuals or faces, but only detects and records the degree of movement activity within the monitored field of view. Alternatively, however, the optical sensor could also record clear images of the monitored field of view.

[0058] For example, a deviation in brightness from the brightness typical for a given time period can indicate an access and / or occupancy anomaly (e.g., if it becomes bright in the middle of the night or dark in the middle of the day). Similarly, a significant variation / fluctuation in the recorded brightness (e.g., due to the use of a flashlight) can indicate an access and / or occupancy anomaly. Certain average directions of movement detected within the optical sensor's field of view can also indicate access and / or occupancy anomalies. Specifically, the data processing unit or another computing unit connected to the optical sensor is designed to perform image recognition based on the captured optical data, particularly images, to determine the parameters mentioned above.The optical sensor can be designed as a camera, a motion detector, or a light barrier.

[0059] If, in this context, the trained machine learning algorithm is at least trained to detect and / or classify one or more of the access and / or presence anomalies in the image data set, reliable and / or precise detection of these anomalies can be advantageously achieved. Among other things, the trained machine learning algorithm could be trained using historical image data, movement data, and / or brightness data typical for normal operation / use of the access-restricted area.

[0060] This should enable the trained machine learning algorithm to distinguish between typical / normal activities and unusual activities, particularly those indicating access and / or location anomalies. It is also conceivable that known potential access and / or location anomalies could be artificially generated and the resulting behavioral patterns, movement data, and / or brightness data incorporated into the training of the machine learning algorithm.

[0061] Furthermore, it is proposed that the sensor unit include at least one access sensor designed to acquire one of the sensor data sets, wherein the sensor data set is configured as an entry and / or exit data set, which comprises entries of persons into the access-restricted area and / or exits of persons from the access-restricted area as detected by the access sensor. This advantageously enables particularly good detection of access and / or presence anomalies. For example, an unusually long entry time, exit time, entry frequency, exit frequency, duration of stay, entry and exit sequence, or the like, can indicate the presence of an access and / or presence anomaly. For example, an unusually long duration of stay by a person can indicate an event (accident, health problem, etc.) in an area.For example, unusual behavior patterns and / or unfamiliar routes (sequences of entries) within the access-restricted area may indicate the presence of an access and / or presence anomaly. For example, a person entering or exiting the access-restricted area at an unusual time for that person may indicate an access and / or presence anomaly. For example, a person entering or exiting an access-restricted area where that person is not usually present, or not present at that time, may indicate an access and / or presence anomaly. For example, an unusually high number of people entering or exiting an access-restricted area may indicate an access and / or presence anomaly.For example, an unusually low number of people entering or exiting a restricted area may indicate an access and / or presence anomaly. The access sensor can be configured as a card reader, a biometric sensor, or the like. The access sensor preferably stores recorded entries and / or exits, particularly along with the respective recorded IDs. The access sensor preferably also stores recorded entry attempts and / or exit attempts, particularly along with the respective recorded IDs.

[0062] If, in this context, the trained machine learning algorithm is trained to detect and / or classify one or more of the access and / or presence anomalies in the entry and / or exit data set, reliable and / or precise detection of these anomalies can be advantageously achieved. Among other things, the trained machine learning algorithm could be trained using historical access and / or exit data typical for normal operation / use of the access-restricted area. This should enable the trained machine learning algorithm to distinguish between typical / normal activities and unusual activities, particularly those indicative of access and / or presence anomalies.It is also conceivable that known possible entry and / or residence anomalies are artificially generated and the resulting entry and / or exit patterns and / or entry and / or exit schemes are included in the training of the machine learning algorithm.

[0063] Additionally, it is proposed that the trained machine learning algorithm is trained to detect and / or classify one or more of the access and / or presence anomalies from a combination of at least two, preferably at least three, of the various sensor data sets. This advantageously allows for particularly high precision and / or reliability in access and / or presence anomaly detection. Advantageously, a particularly high number of access and / or presence anomalies can be detected. In particular, the at least two, preferably at least three, sensor data sets to be considered by the trained machine learning algorithm are fed into the algorithm together. Specifically, the trained machine learning algorithm outputs information about the detected access and / or presence anomaly(s).The information can include a description of the access and / or presence anomaly and / or a classification of the access and / or presence anomaly. In addition to the sensor data sets from the access control system's sensor unit, further data could also be considered when executing the trained machine learning algorithm, e.g., external sensor data or further information about or from the restricted area.

[0064] Furthermore, it is proposed that the sensor unit be arranged in a barrier unit, in particular a barrier element / movable component such as a door, gate, barrier, turnstile, or the like, which is intended for opening and closing an entrance and / or exit to the access-restricted area, or in the immediate vicinity surrounding the entrance or exit to the access-restricted area. This allows for an advantageous design and / or reliable monitoring of the entrance and / or exit area. Advantageously, integration into a locking system encompassing the barrier unit is also possible. The immediate vicinity is, in particular, an area formed by all points in space, the distance of which from the barrier unit, in particular the barrier element, is at most as large as the maximum spatial extent of the barrier unit, in particular the barrier element.For example, the sensor unit could be located in a door frame or in an area around the barrier unit, from which the person can operate the barrier unit for immediate passage through the entry and / or exit.

[0065] Furthermore, it is proposed that the sensor unit form an integral part of the locking system of the barrier unit, in particular the barrier element, which is intended for locking and unlocking an entrance and / or exit to the access-restricted area. This enables an advantageous design and / or reliable monitoring of the entry and / or exit area. Advantageously, the access control system can be integrated into the locking system. Advantageously, existing components / sensors of the locking system can be used in the access control system. This can achieve, in particular, a reduction in compactness, cost-effectiveness, and / or complexity. Advantageously, a significant expansion of the functionality and usability of locking systems can be achieved.

[0066] Furthermore, it is proposed that the sensor unit form a sensor box permanently assigned to the barrier unit, in particular to the barrier element, and comprising the indoor temperature sensor, the acoustic sensor, the optical sensor, and the access sensor. This allows for increased compactness, cost-effectiveness, and / or reduced complexity. It is particularly conceivable that different barrier elements within an access-restricted area could each have their own sensor box. The sensor box preferably comprises a common housing. The sensor box is preferably permanently installed in a component, e.g., the barrier unit, or in a structure surrounding the barrier unit. This advantageously prevents unauthorized external access to the sensor unit.It is conceivable that the data evaluation unit is also integrated into the sensor box, or that the sensor box at least has a communication interface for (wired or wireless) communication with an externally arranged data evaluation unit.

[0067] Furthermore, it is proposed that the data evaluation unit, in particular the sensor unit, the locking system, or the sensor box, has at least one receiver module for receiving (wirelessly or via cable) at least one additional sensor data set from an external source. This additional sensor data set is intended to be used by the data evaluation unit, and in particular by the machine learning algorithm of the data evaluation unit, in addition to the sensor data set(s) for detecting access and / or presence anomalies and, in particular, for evaluation using the trained machine learning algorithm. This advantageously allows for high precision and / or reliability in the detection of access and / or presence anomalies. It also advantageously maximizes the number of detectable access and / or presence anomalies.Advantageously, the complexity of the access control system can be kept low, especially since not all sensors need to be provided by the sensor unit itself. It is conceivable that the access control system can be flexibly adapted to the type and number of existing external sensors. It is also conceivable that the trained machine learning algorithm can be flexibly adapted to the type and number of existing external sensors and trained accordingly. Preferably, the receiver module is a wireless receiver module capable of receiving data via a known radio protocol, such as WLAN, Bluetooth, or similar.

[0068] If the additional sensor data set originates from a device that is separate from and distinct from any type of access control system, and which is also assigned to and / or located within the restricted area, then a high degree of precision and / or reliability in the detection of access and / or presence anomalies can be advantageously achieved. This also allows for a maximum number of detectable access and / or presence anomalies. The device, separate from and distinct from the access control system, could be, for example, a fire alarm, a device for measuring Wi-Fi data usage, a heating controller, a ventilation system, a lighting system, an air conditioner, a smart home device, or similar.

[0069] Furthermore, it is proposed that the additional sensor data set is a data set originating from another sensor unit, another locking system, or another sensor box that is assigned to a different barrier unit, in particular a barrier element, intended for locking and unlocking an entrance and / or exit, or to a different access-restricted area, or to the same access-restricted area as the sensor unit, locking system, or sensor box with the receiver module. This advantageously allows for high precision and / or reliability in the detection of access and / or presence anomalies. Advantageously, the number of detectable access and / or presence anomalies can be maximized. In particular, the individual barrier units, especially individual barrier elements, and their assigned sensor boxes are networked together. This networking can be wireless or wired.Preferably, the interconnected sensor boxes form a sensor swarm. Sensor data sets from various barrier units / sensor boxes / access-restricted areas could be collected in a cloud or on a central server and processed there together. For example, the proposed network could advantageously track a person's path through a building or site with multiple access-restricted areas and monitor for anomalies / access and / or presence anomalies.

[0070] If the sensor unit, locking system, or sensor box has an integrated voice output unit, a high level of security can be advantageously achieved. For example, if an event is detected, such as a potential accident or health problem, a confirmation message can be issued audibly before any action is automatically taken. This can effectively reduce the number of false alarms. The voice output unit is specifically designed to output (computer-generated) speech messages generated by the data processing unit or another processing unit based on the analysis results of the data processing unit.

[0071] If the speech output unit is designed to initiate automated interaction with a person present in the restricted area upon detection of an access and / or presence anomaly, this can advantageously reduce the number of false alarms. Furthermore, it can be advantageous to stop any action by the person that triggered the access and / or presence anomaly. For example, the automated interaction could include a question about a person's health, a question about a need for assistance, a question about their intentions, a question about the reason for their behavior, an identification request, or similar inquiries.

[0072] Furthermore, if the interaction is a conversation based on computerized speech output and speech recognition, controlled by a language model such as a Large Language Model (LLM), with the person present in the access-restricted area, a particularly situation-specific follow-up question can be advantageously included. Additionally, any irritation of the person being addressed can be minimized, as the interaction can appear particularly natural.

[0073] Furthermore, it is proposed that the data evaluation unit be designed to issue a notification to a third party, such as a security service or emergency dispatch center, depending on the course or outcome of the interaction. This can advantageously achieve a high level of security, particularly for the individual concerned and / or for the access-restricted area. If, for example, the outcome of the interaction is that a health problem exists or that assistance is required, the appropriate service, such as an ambulance service or building maintenance service, can be alerted. If, for example, there is no response to the interaction, the security service and / or the ambulance service can be alerted, depending on the nature of the access and / or location anomaly.

[0074] Furthermore, it is proposed that the access and / or presence anomaly be an unauthorized entry or presence in the restricted area, an entry and / or exit pattern or presence pattern of a person indicating an accident or emergency, and / or a person's duration of presence in the restricted area indicating an accident or emergency. This can advantageously achieve a high level of security, particularly against health hazards and / or unauthorized access. The entry and / or exit pattern indicating an accident or emergency could, for example, be a person's failure to leave the restricted area within an expected timeframe.An entry and / or exit pattern indicating an accident or emergency could, for example, be the inability to reach a specific area or sub-area of ​​the restricted space. An entry and / or exit pattern indicating an accident or emergency could, for example, be the abrupt end of a recorded activity by a person. An entry and / or exit pattern indicating an accident or emergency could, for example, be a person rapidly moving between areas (suggesting a search for help). An entry duration indicating an accident or emergency could, for example, be significantly longer or shorter than expected.

[0075] Furthermore, it is proposed that the data evaluation unit be designed to categorize frequently or regularly recurring access and / or presence anomalies within the restricted area as normal in the future. This can advantageously reduce the number of false alarms. A high degree of flexibility and / or adaptability of the access control system can also be achieved. Preferably, the access control system learns over time. For example, if a monitored person changes their routine (e.g., arriving on different days or at different times than before), this will no longer be recognized as an access and / or presence anomaly after a while.

[0076] Furthermore, a building or part of a building is proposed to be equipped with an access control system, in particular a building with an access control system and an access-restricted room controlled by the access control system, or with several access-restricted rooms, each separately controlled by the access control system. This can advantageously achieve a high level of security, especially regarding access to the access-restricted area and / or the safety / health of the persons located in the access-restricted area. For example, unauthorized presence, unusual behavior patterns, or sudden health problems of persons located in the access-restricted area can be advantageously detected. The building or part of the building can, for example, be constructed of a building with a fire alarm system.only the barrier unit / only the barrier element, in particular only a door or the like, and / or a part of a wall or barrier system (e.g. barrier system or revolving door system) surrounding the barrier unit / barrier element, in particular the door or the like.

[0077] Furthermore, a method for access control for one or more restricted areas, such as rooms or buildings, is proposed, particularly by means of an access control system, wherein at least one data evaluation unit detects access and / or presence anomalies for at least one of the restricted areas, preferably those controlled by the access control system, based on an analysis of one or more sensor data sets. This advantageously allows for a high level of security, particularly with regard to access to the restricted area and / or the safety / health of the persons present in the restricted area.

[0078] Furthermore, it is proposed that, during the analysis for detecting access and / or presence anomalies, one or more of the sensor data sets be analyzed by a trained machine learning algorithm of the data evaluation unit. This can advantageously optimize the detection of access and / or presence anomalies.

[0079] The access control system, the structure, and the method according to the invention are not limited to the application and embodiment described above. In particular, the access control system, the structure, and the method according to the invention may, to achieve a functionality described herein, comprise a different number of individual elements, components, and units than the number specified herein. Specifically, all access and / or presence anomalies mentioned by way of example are to be understood as access and / or presence anomalies recognizable by the trained machine learning algorithm, for whose recognition the trained machine learning algorithm may be specifically trained.

[0080] Drawings

[0081] Further advantages will become apparent from the following description of the drawings. The drawings illustrate an embodiment of the invention. The drawings, the description, and the claims contain numerous features in combination. A person skilled in the art will expediently consider the features individually and combine them into meaningful further combinations.

[0082] They show:

[0083] Fig. 1 schematically shows a part of an exemplary building with restricted access areas and an access control system that controls the restricted access areas.

[0084] Fig. 2 schematically shows a sensor box of the access control system and Fig. 3 a schematic flowchart of a procedure for access control using the access control system.

[0085] Description of the exemplary embodiment

[0086] Figure 1 schematically shows a part of an exemplary structure 48. The structure 48 comprises several access-restricted areas 10, 10', 10". One of the access-restricted spatial areas 10 is, by way of example, designed as an access-restricted room of a building. Another of the access-restricted spatial areas 10' is, by way of example, designed as another access-restricted room of the same building. A second further of the access-restricted spatial areas 10" is, by way of example, designed as an enclosed area, in particular factory premises with the building. The structure 48 thus comprises, in the exemplary case, at least the building and the enclosure. The structure 48 has an access control system 14. The access-restricted spatial areas 10, 10', 10", in particular the rooms and the area, are each controlled separately by the common access control system 14.The access control system 14 regulates and / or controls at least one entry and / or exit to or from the individual access-restricted spatial areas 10, 10', 10”.

[0087] The structure 48 comprises devices 38 that are distinct from and separate from any type of access control system 14. The devices 38 are distinct from and separate from a locking system 32 of the structure 48. Each of the devices 38 is assigned to and / or located in one of the access-restricted areas 10, 10', 10'). By way of example, the device 38 shown in Figure 1 is configured as a smoke detector.

[0088] The structure 48 comprises several barrier units 28, 28', 28", 28'", 44. Each of the barrier units 28, 28', 28", 28'", 44 is designed to close and / or open, in particular to block and / or release, an associated entry and / or exit 30 to at least one of the access-restricted spatial areas 10, 10', 10". A first barrier unit 28, which controls the access-restricted spatial area 10, has, by way of example, a barrier element designed as an exterior door of the building. A second barrier unit 28', 44, which controls the further access-restricted spatial area 10', also has, by way of example, a barrier element designed as an exterior door of the building. A third barrier unit 28", which controls the second further access-restricted spatial area 10", It features, for example, a barrier element designed as a barrier to the perimeter fence of the site.A fourth barrier unit 28'”, which controls a connection between the access-restricted spatial area 10 and the further access-restricted spatial area 10', features, by way of example, a barrier element designed as an interior door of the building. Alternative designs of barrier elements and / or alternative arrangements and / or numbers of barrier units 28, 28', 28", 28'", 44 are of course also conceivable.

[0089] The access control system 14 comprises sensor units 16, 40. The sensor units 16, 40 are designed to acquire one or more sensor data sets. One sensor unit 16, 40 is located in each of the barrier units 28, 28', 28", 28'", 44. Alternatively, one sensor unit 16, 40 could also be located in the immediate vicinity of each of the entrances or exits 30 associated with each of the barrier units 28, 28', 28", 28'", 44. The structure 48, and in particular at least the barrier units 28, 28', 28", 28'", 44 of the structure 48, features the locking system 32. The locking system 32 is designed to lock and unlock the barrier units 28, 28', 28", 28'", 44. The locking system 32 is designed for the automatic opening and closing of the barrier units 28, 28', 28”, 28'”, 44.The locking system 32 comprises a reader or the like, which is intended for identifying an entity intending to pass through one of the entrances and / or exits 30 to one of the access-restricted areas 10, 10', 10" . The locking system 32 is intended to grant or prevent passage depending on authorizations and / or successful identification. The sensor units 16, 40 form integral components of the locking system 32. The sensor units 16, 40 are intended at least for generating sensor data sets.

[0090] The access control system 14 includes a data evaluation unit 12. The data evaluation unit 12 is designed to detect access and / or presence anomalies in at least one of the spatial areas 10, 10', 10" restricted by the access control system 14, based on an analysis of one or more sensor data sets, e.g., from sensor units 16, 40 and / or devices 38. The data evaluation unit 12 is exemplified as a cloud, specifically embedded in a cloud computing network. Alternatively, the data evaluation unit 12 could also be centrally located or decentrally located. The data evaluation unit 12 is intended for the execution of a trained machine learning algorithm.The machine learning algorithm is designed to analyze one or more of the sensor data sets from sensor units 16, 40 and / or devices 38 using the trained machine learning algorithm of the data evaluation unit 12 to detect access and / or presence anomalies. The access and / or presence anomaly can be unauthorized entry or presence in the access-restricted area 10, 10', 10". The access and / or presence anomaly can be an entry and / or exit pattern or presence pattern of a person indicating an accident or emergency. The access and / or presence anomaly can be a person's duration of stay in the access-restricted area 10, 10', 10" indicating an accident or emergency.The data evaluation unit 12 is designed to automatically categorize a frequently or regularly recurring access and / or presence anomaly in the restricted area 10, 10', 10” as normal in the future, especially if no reaction has been made to the recognized access and / or presence anomaly in the past.

[0091] Figure 2 schematically shows one of the sensor units 16. The sensor unit 16 is designed as a sensor box 34. The sensor box 34 comprises several sensor types, preferably in a common housing 50. The sensor box 34 is permanently assigned to one of the barrier units 28, 28', 28", 28'", 44. The sensor box 34 is integrated into the respective barrier element of the associated barrier unit 28, 28', 28", 28'", 44 or arranged in the immediate vicinity of the respective barrier element. The sensor unit 16, in particular the sensor box 34, has an indoor temperature sensor 18. The indoor temperature sensor 18 is designed as an indoor thermometer. The indoor temperature sensor 18 is designed to acquire one of the sensor data sets supplied to the data evaluation unit 12, in particular to the machine learning algorithm. The sensor data set of the indoor temperature sensor 18 is designed as an indoor temperature data set.The indoor temperature data set comprises temperatures measured by the indoor temperature sensor 18 within the respective access-restricted spatial area 10, 10', 10″ in which the indoor temperature sensor 18, in particular the sensor box 34, is located. The sensor unit 16, in particular the sensor box 34, includes an outdoor temperature sensor 20. The outdoor temperature sensor 20 is designed as an outdoor thermometer. The outdoor temperature sensor 20 is intended for acquiring one of the sensor data sets supplied to the data evaluation unit 12, in particular to the machine learning algorithm. The sensor data set of the outdoor temperature sensor 20 is designed as an outdoor temperature data set. The outdoor temperature data set comprises temperatures measured by the outdoor temperature sensor 20 outside the respective access-restricted spatial area 10, 10', 10″ in which the sensor box 34 is located, preferably outside the building.The outdoor temperature dataset alternatively comprises outdoor temperatures measured by the outdoor temperature sensor 20 in an environment directly surrounding the access-restricted spatial area 10, 10', 10" or in a room of a building adjacent to the access-restricted spatial area 10, 10', 10". The trained machine learning algorithm is trained to detect one or more of the access and / or occupancy anomalies in the indoor temperature dataset and / or to classify them based on the indoor temperature dataset. The trained machine learning algorithm is trained to detect one or more of the access and / or occupancy anomalies from a combination of the indoor and outdoor temperature datasets and / or to classify them based on the combination of the indoor and outdoor temperature datasets.

[0092] The sensor unit 16, in particular the sensor box 34, includes an acoustic sensor 22. The acoustic sensor 22 is configured as a microphone. The acoustic sensor 22 is designed to acquire sensor data sets that are fed to the data evaluation unit 12, in particular to the machine learning algorithm. The sensor data set of the acoustic sensor 22 is configured as a noise data set. The noise data set comprises noise levels, noise frequencies, and / or noise patterns measured by the acoustic sensor 22 within the respective access-restricted spatial area 10, 10', 10", in which the acoustic sensor 22, in particular the sensor box 34, is located. The trained machine learning algorithm is trained to recognize one or more of the access and / or presence anomalies in the noise data set and / or to classify them based on the noise data set.

[0093] The sensor unit 16, in particular the sensor box 34, comprises an optical sensor 24. The optical sensor 24 is configured as an image acquisition sensor. The optical sensor 24 is designed to acquire sensor data sets that are fed to the data evaluation unit 12, in particular to the machine learning algorithm. The sensor data set of the optical sensor 24 is configured as an image data set. The image data set comprises movements and / or changes in brightness measured by the optical sensor 24 within the respective access-restricted spatial area 10, 10', 10", in which the optical sensor 24, in particular the sensor box 34, is located. The trained machine learning algorithm is trained to recognize one or more of the access and / or presence anomalies in the image data set and / or to classify them based on the image data set.The sensor unit 16, in particular the sensor box 34, includes an access sensor 26. The access sensor 26 is configured as the reader. The access sensor 26 is designed to acquire sensor data sets that are fed to the data evaluation unit 12, in particular to the machine learning algorithm. The sensor data set of the access sensor 26 is configured as an entry and / or exit data set. The entry and / or exit data set comprises entries of persons into the associated access-restricted spatial area 10, 10', 10" detected by the access sensor 26. The entry and / or exit data set comprises exits of persons from the access-restricted spatial area 10, 10', 10" detected by the access sensor 26.The trained machine learning algorithm is trained to detect one or more of the entry and / or stay anomalies in the entry and / or exit data set and / or to classify them based on the entry and / or exit data set.

[0094] The trained machine learning algorithm is also specifically trained to detect and / or classify one or more of the access and / or presence anomalies from a combination of at least two, preferably at least three, of the different sensor data sets, in particular the indoor temperature data set, the outdoor temperature data set, the noise data set, the image data set, the entry and / or exit data set and / or one or more sensor data sets from external devices 38.

[0095] The data evaluation unit 12 comprises a receiver module 36. Alternatively or additionally, the sensor box 34 or the sensor unit 16 can also have a receiver module 36. The receiver module 36 is designed to receive at least one additional sensor data set from an external source. The additional sensor data set can be a data set originating from the device 38, which is designed separately from any type of access control system 14. Alternatively or additionally, the additional sensor data set can be a data set originating from the additional sensor unit 40, an additional locking system, or an additional sensor box 42, which is assigned to the other barrier unit 44 of the other access-restricted area 10' or of the same access-restricted area 10, and which is intended for locking and opening another entrance and / or exit 30, as / as the sensor unit 16, the locking system 32, or the sensor box 34.The data evaluation unit 12 is designed to use the additional sensor data set in addition to the one or more sensor data sets for the detection of access and / or presence anomalies.

[0096] The sensor unit 16, in particular the sensor box 34, includes an acoustic speech output unit 46. The acoustic speech output unit 46 could alternatively also be assigned to the locking system 32 or only connected to the sensor box 34 for data transmission purposes. The speech output unit 46 is designed to initiate automated interaction with a person present in the access-restricted area 10, 10', 10" upon detection of an access and / or presence anomaly. The interaction is a conversation with the person present in the access-restricted area 10, 10', 10" based on computerized speech output and computerized speech recognition, and controlled by a language model, e.g., a Large Language Model (LLM). The data evaluation unit 12 is designed to send a message to a third party, e.g., a third party, depending on the course or result of the interaction.to issue a security service or an emergency control center, etc.

[0097] Figure 3 shows a schematic flowchart of an exemplary procedure for access control for one or more of the access-restricted spatial areas 10, 10', 10” using the access control system 14. In at least one procedure step 52, a sensor data set is generated by one or more sensors of the sensor units 16, 40 and by one or more of the devices 38. In at least one further procedure step 54, several of the sensor data sets are received directly or via the receiver module 36 by the data evaluation unit 12. In at least one further procedure step 56, several of the sensor data sets are input into the trained machine learning algorithm of the data evaluation unit 12. In at least one further procedure step 58, the trained machine learning algorithm analyzes the input sensor data sets.In at least one further process step 60, the trained machine learning algorithm outputs analysis results that include access and / or location anomalies in at least one of the access-restricted spatial areas 10, 10', 10" insofar as these were detected by the trained machine learning algorithm in the input sensor data sets. For example, an access and / or location anomaly could occur if a person has entered one of the access-restricted spatial areas 10, 10', 10" but, contrary to expectations (e.g., based on that person's usual behavior), has not yet left. This access and / or location anomaly could indicate an accident or a problem affecting that person. Therefore, in at least one further process step 62, an interaction with the person is automatically initiated via the speech output unit 46.For example, the speech output unit 46 could issue a question such as "Are you all right?". If the person responds, this is recorded by a computerized speech recognition system in the speech output unit 46. The speech output unit 46 is designed for both speech output and speech recognition. Alternatively, the acoustic sensor 22 could also handle the speech recognition. Based on the responses, a message is then sent to the third party, e.g., the security service or the emergency control center, in at least one further step 64. Reference sign.

[0098] 10 area

[0099] 12 Data evaluation unit

[0100] 14 Access control system

[0101] 16 sensor units

[0102] 18 Indoor temperature sensor

[0103] 20 Outdoor temperature sensor

[0104] 22 Acoustic sensor

[0105] 24 Optical Sensor

[0106] 26 Access sensor

[0107] 28 barrier units

[0108] 30 Entry and / or Exit

[0109] 32 locking system

[0110] 34 Sensor Box

[0111] 36 Receiver module

[0112] 38 device

[0113] 40 Additional sensor units

[0114] 42 Additional sensor boxes

[0115] 44 Additional barrier units

[0116] 46 Speech output unit

[0117] 48 Building

[0118] 50 cases

[0119] 52nd process step

[0120] 54th process step

[0121] 56th process step

[0122] 58th process step

[0123] 60th process step

[0124] 62nd process step

[0125] 64th process step

Claims

Claims 1. An access control system (14) for one or more access-restricted spatial areas (10, 10', 10"), for example, rooms or buildings, comprising at least one data evaluation unit (12) designed to detect access and / or presence anomalies for at least one of the spatial areas (10, 10', 10") restricted by the access control system (14) based on an analysis of one or more sensor data sets.

2. An access control system (14) according to claim 1, characterized in that the analysis for detecting the access and / or presence anomaly comprises an evaluation of one or more of the sensor data sets by a trained machine learning algorithm of the data evaluation unit (12).

3. An access control system (14) according to claim 1 or 2, characterized by at least one sensor unit (16) designed to acquire one or more of the sensor data sets.

4. Access control system (14) according to claim 3, characterized in that the sensor unit (16) has at least one indoor temperature sensor (18) provided for recording one of the sensor data sets, wherein the sensor data set is designed as an indoor temperature data set which comprises temperatures measured by the indoor temperature sensor (18) within the access-restricted spatial area (10, 10', 10”).

5. Access control system (14) according to claim 4, characterized in that the sensor unit (16) has at least one outdoor temperature sensor (20) provided for recording one of the sensor data sets, wherein the sensor data set is designed as an outdoor temperature data set which includes temperatures that are measured by the outdoor temperature sensor (20) in an environment directly surrounding the access-restricted spatial area (10, 10', 10”) or in a room of a building adjacent to the access-restricted spatial area (10, 10', 10”).

6. Access control system (14) according to claims 2 and 4 or according to claims 2 and 5, characterized in that the trained machine learning algorithm is at least trained to detect and / or classify one or more of the access and / or presence anomalies in the indoor temperature data set, in particular in conjunction with the outdoor temperature data set.

7. Access control system (14) according to one of claims 3 to 6, characterized in that the sensor unit (16) has at least one acoustic sensor (22) provided for recording one of the sensor data sets, wherein the sensor data set is designed as a noise data set which comprises noise levels, noise frequencies and / or noise patterns measured by the acoustic sensor (22) within the access-restricted spatial area (10, 10', 10”).

8. Access control system (14) according to claims 2 and 7, characterized in that the trained machine learning algorithm is at least trained to detect and / or classify one or more of the access and / or presence anomalies in the noise data set.

9. Access control system (14) according to one of claims 3 to 8, characterized in that the sensor unit (16) has at least one optical sensor (24) provided for capturing one of the sensor data sets, wherein the sensor data set is designed as an image data set which comprises movements observed by the optical sensor (24) within the access-restricted spatial area (10, 10', 10”) and / or brightness levels captured by the optical sensor (24) within the access-restricted spatial area (10, 10', 10”).

10. Access control system (14) according to claims 2 and 9, characterized in that the trained machine learning algorithm is at least trained to detect and / or classify one or more of the access and / or presence anomalies in the image data set.

11. Access control system (14) according to one of claims 3 to 10, characterized in that the sensor unit (16) has at least one access sensor (26) provided for recording one of the sensor data sets, wherein the sensor data set is designed as an entry and / or exit data set, which includes entries of persons into the access-restricted spatial area (10, 10', 10”) and / or exits of persons from the access-restricted spatial area (10, 10', 10”) detected by the access sensor (26).

12. Access control system (14) according to claims 2 and 11, characterized in that the trained machine learning algorithm is at least trained to detect and / or classify one or more of the access and / or presence anomalies in the entry and / or exit data set.

13. Access control system (14) according to at least two, preferably at least three, of claims 6, 8, 10 and 12, characterized in that the trained machine learning algorithm is trained at least to detect and / or classify one or more of the access and / or presence anomalies from a combination of at least two, preferably at least three, of the different sensor data sets.

14. Access control system (14) according to one of claims 3 to 13, characterized in that the sensor unit (16) is arranged in a barrier unit (28, 28', 28”, 28'”), which is provided for closing and opening an entry and / or exit (30) to the access-restricted spatial area (10, 10', 10”), or in a close area surrounding the entry or exit (30) to the access-restricted spatial area (10, 10', 10”).

15. Access control system (14) according to one of claims 3 to 14, characterized in that the sensor unit (16) forms an integral part of a locking system (32) of a barrier unit (28, 28', 28”, 28'”), which is provided for closing and opening an entry and / or exit (30) to the access-restricted spatial area (10, 10', 10”).

16. Access control system (14) at least according to claims 3, 4, 7, 9 and 11 and according to at least one of claims 14 or 15, characterized in that the sensor unit (16) forms a sensor box (34) permanently assigned to the barrier unit (28, 28', 28”, 28'”) and comprising the indoor temperature sensor (18), the acoustic sensor (22), the optical sensor (24) and the access sensor (26).

17. Access control system (14) according to one of claims 14 to 16, characterized in that the data evaluation unit (12), the sensor unit (16), the locking system (32) or the sensor box (34) has at least one receiving module (36) for receiving at least one further sensor data set from an external source, which is intended to be used by the data evaluation unit (12) in addition to the sensor data set(s) for detecting access and / or presence anomalies.

18. Access control system (14) according to claim 17, characterized in that the further sensor data set is a data set which originates from a device (38) that is different from and separate from any type of access control system (14), which is also assigned to and / or arranged in the access-restricted area (10, 10', 10”).

19. Access control system (14) according to claim 17, characterized in that the further sensor data set is a data set which originates from a further sensor unit (40), a further locking system or a further sensor box (42) which is assigned to a different barrier unit (44) for closing and opening an entry and / or exit (30) of a different access-restricted spatial area (10, 10', 10”) or the same access-restricted spatial area (10, 10', 10”) than the sensor unit (16), the locking system (32) or the sensor box (34) with the receiver module (36).

20. Access control system (14) according to one of claims 3 to 19, characterized in that the sensor unit (16, 40), the locking system (32) or the sensor box (34, 42) has an acoustic voice output unit (46).

21. Access control system (14) according to claim 20, characterized in that the speech output unit (46) is provided to carry out an automated interaction with a person present in the access-restricted spatial area (10, 10', 10”) when an access and / or presence anomaly is detected.

22. Access control system (14) according to claim 21, characterized in that the interaction is a conversation with the person present in the access-restricted spatial area (10, 10', 10”) based on a computerized speech output and on a computerized speech recognition and controlled by a language model, e.g. a Large Language Model (LLM).

23. Access control system (14) according to claim 22, characterized in that the data evaluation unit (12) is provided to issue a message to a third party, e.g. a security service or an emergency control center, etc., depending on a course or result of the interaction.

24. Access control system (14) according to one of the preceding claims, characterized in that the access and / or presence anomaly is unauthorized access or presence in the access-restricted spatial area (10, 10', 10”).

25. Access control system (14) according to one of the preceding claims, characterized in that the access and / or residence anomaly is an entry and / or exit scheme or residence scheme of a person indicating an accident or emergency.

26. Access control system (14) according to one of the preceding claims, characterized in that the access and / or residence anomaly is an access duration of a person in the access-restricted spatial area (10, 10', 10”) indicating an accident or an emergency.

27. Access control system (14) according to one of the preceding claims, characterized in that the data evaluation unit (12) is provided to categorize the same access and / or presence anomaly as normality in the future if it is frequently or regularly detected in the access-restricted spatial area (10, 10', 10”).

28. Building (48) or building component with the access control system (14) according to one of the preceding claims, in particular buildings with the access control system (14) according to one of the preceding claims and with an access-restricted room controlled by the access control system (14) or with several access-restricted rooms each separately controlled by the access control system (14).

29. Method for access control for one or more access-restricted spatial areas (10, 10', 10”), for example rooms or buildings, in particular by means of an access control system (14) according to one of the preceding claims, wherein access and / or presence anomalies are detected by at least one data evaluation unit (12) at least for one of the access-restricted spatial areas (10, 10', 10”) based on an analysis of one or more sensor data sets.

30. Method according to claim 29, characterized in that, in the analysis for detecting the access and / or presence anomaly, one or more of the sensor data sets are analyzed by a trained machine learning algorithm of the data evaluation unit (12).

Images