DE102022128864A1 - Object localization by a mobile robot - Google Patents

Abstract

The invention relates to a method and a device for locating people and/or recording a person's vital parameters using a mobile robot. To do this, parameters of the detection sensors are aligned accordingly, the robot follows a path, detects objects, localizes them and compares positions with stored rules in order to trigger events if necessary. It is described how the detection quality can be optimized by navigating the robot and varying the signal strength of the sensors.

Description

Field of the invention

The invention relates to robots for monitoring persons, in one aspect monitoring vital parameters.

Background of the invention and prior art

More and more tasks are now being taken over by technical systems. This also includes mobile robots. These can monitor objects or people (e.g. JP2019016339A ) and, in one aspect, also record vital parameters of people in this context. For example, it may be that close monitoring is necessary for people with health problems without the use of stationary sensors, for example, especially when several or even a large number of people are to be monitored.

One aspect of the technical challenge is the monitoring of patients in hospitals. It is known that mobile robots can carry out such monitoring. A mobile robot can patrol a corridor, drive into each room and then monitor the patients there. However, this requires opening the room doors. This can be implemented with additional opening mechanisms that can be controlled by the robot via radio, for example. However, the opening mechanisms are very complex to install. The challenge is therefore to enable robust patient monitoring when the doors are closed.

One approach pursued in the state of the art is the use of radar technology on a robot to detect people behind walls (e.g. CN113715032A or WO2005074422A2 However, robust approaches are needed to implement this in the clinical context. CN113715032A This is a description of a weapon system that enables very rough location of people behind walls. WO2005074422A2 merely describes in a very abstract way the possibility of detecting people through walls using a radar mounted on a robot.

In some situations it can be important to locate people precisely, for example in order to be able to measure vital parameters as accurately as possible. This is particularly the case if the measurement is to be contactless and the signals used may have to pass through several media. This may result in a corresponding attenuation of the signals and, if necessary, a fine adjustment of the signal strength. In some situations it may be necessary to locate the mobile robot carrying out the monitoring precisely. In the context of vital parameter recording or person localization it may be necessary to recognize critical situations. For this purpose, rules are necessary in order to classify the situation. It may also be necessary to raise an alarm, for example in order to help the monitored people in critical situations.

The object of the invention is now to address these issues accordingly in order to carry out a sufficiently precise localization of a person to be monitored or an area to be monitored or, in one aspect, to determine vital parameters of the person to be monitored.

If this document refers to already established state of the art in systems, devices or procedures, this may be referenced at these points with further sources.

General description of the invention

In one aspect, the invention comprises a computer-implemented method for monitoring an area by a mobile robot. The robot travels a path, detects an area to be monitored using at least one radar sensor unit, evaluates the detected radar signals and triggers an event. The event in turn can include a signal transmission to the cloud or to a terminal. The area to be monitored is detected through a wall and/or closed door. In one aspect, the area to be monitored is detected approximately perpendicular to the direction of travel of the robot. The signal evaluation in turn includes the identification of objects including people and enables, for example, a differentiation between moving and immobile objects, with moving objects preferably being people. In one aspect, the signal evaluation includes a determination of at least one vital parameter. Alternatively and/or additionally, the signal evaluation can include a localization of the object position. These object positions can, for example, be compared with object positions and/or object zones stored in the memory. This can also include, for example, so-called forbidden zones. This in turn means that no objects may be present in these zones, as can be defined in rules stored in the robot's memory. If the robot records vital parameters, a comparison is made between the determined vital parameters and the stored vital parameters. These can, for example, be recorded for a person at a local The robot can, for example, also hold comparison values for vital parameters for people that it is to record. A person can, for example, be defined by being assigned to a local position. The robot can determine this, for example, by being at a certain position, e.g. on its operating path, and by having corresponding vital parameters assigned to this position, possibly associated with at least one direction vector that describes in length and orientation where the person whose vital parameters are to be measured and compared is allowed to be. The actual signal evaluation of the radar signals, in turn, includes, for example, the comparison with patterns. In one aspect, a classifier that was created using imitation learning or that uses action recognition can be used for the signal evaluation. The position data of an object and/or a person used by the robot, which were determined by evaluating radar sensor data, are in one possible aspect standardized by data determined using a laser scanner, camera, ultrasound sensor and/or odometry unit. In a further, optional aspect, the robot's map is updated by incorporating the obstacles detected by the radar sensor unit into a map that was created based on data from a laser scanner, a camera or an ultrasonic sensor. In one aspect, the robot takes a scanning position in front of a possibly closed door at a defined distance in order to detect an area using the radar sensor unit on its path and performs a rotational movement. In an optional aspect, it can move along a convex path in front of the said door and maintain a defined minimum distance from the door.

In one aspect, the technical problem outlined at the beginning includes a sub-problem, namely the correct localization of objects in the area to be monitored, which is also beneficial, for example, for the precise recording of vital parameters.

For this purpose, the invention comprises a computer-implemented method for determining, correcting or synchronizing object position data, comprising the creation of a first map by a robot that shows obstacles or objects based on laser scanner, camera and/or ultrasonic sensor data, the creation of a second map by a robot with obstacles or objects based on radar sensor data and a comparison of the position data for obstacles or objects contained in the two maps. For example, the position data of objects or obstacles on the second map can be corrected at least partially by position data of objects or obstacles on the first map. For example, the corrected position data can be position data of stationary obstacles or objects. In one aspect, the second map is preferably created after the first map. The radar sensor unit used to determine the radar sensor data includes, for example, an area that cannot be simultaneously detected by the laser scanner, camera or ultrasonic sensor. In one aspect, the radar sensor data includes objects or obstacles behind a wall or closed door. For example, in one variant, the corrected position data of the second map can also be stored directly in the second map, i.e. the old position data is overwritten. The computer-implemented method can further include adding position data of moving objects from the second map to the first map and/or generating a third map by adopting stationary and moving object or obstacle positions from the first map and adopting additional moving object or obstacle positions from the second map. In one aspect, the coordinates of the first map are adopted for robot navigation, and the robot navigates according to the third map.

Based on these partially optional process steps, for example, a comparison can be made of the position data for moving objects determined using the radar sensor unit with zones for moving objects stored in a table, in one of the maps mentioned or in another map, with rules being assigned to each zone. Alternatively and/or in addition to this, the computer-implemented method includes triggering an event based on the rules for the respective zone, using the object position data to adjust the signal strength of the radar signal, classifying object data to detect people, and/or recording at least one vital parameter for a detected person.

In one aspect, the technical problem outlined at the beginning includes a sub-problem, namely the reliable detection of a person's vital parameters over a distance of several meters and through a wall or door while at the same time minimizing the radiation exposure to which this person is exposed by radar detection.

For this purpose, in one aspect, the robot has a computer-implemented method for adjusting the signal strength of a radar sensor, comprising the movement of the robot on a path, the detection of objects or obstacles in the radar signal detection area, the detection of persons in the radar signal detection area, the determination of the distance between the robot and the detected person and the determination of a signal strength for the radar sensor based on the determined distance between the robot and the person. The method can, for example, further comprise adjusting the signal strength of the radar sensor (e.g. by increasing the signal strength), for example then followed by a further adjustment of the signal strength such as a reduction. In addition, the method can be used for a radar data-based vital parameter measurement of the detected person. The person's position can in turn be stored in a map on the robot. In one aspect, the method can comprise determining the necessary detection depth of the radar sensor 31 by comparing the area to be detected with a map and/or further adjusting the signal strength of the radar sensor based on the necessary detection depth, which enables person detection in the area to be detected. In addition to the aforementioned, partially optional method steps, a signal can be transmitted to a terminal based on the evaluation of the determined vital parameters. The position data used in the method, which were determined by evaluating radar sensor data, can be standardized, for example, by data determined using a laser scanner, camera, ultrasound sensor and/or odometry unit. In one aspect, when collecting vital data, there is a wall or (closed) door between the robot and the detected person. The radar signal detection area can, for example, be aligned essentially perpendicular to the path of the robot.

In one aspect, the technical problem outlined at the beginning includes a sub-problem, namely the optimization of the robot's travel path in order to, on the one hand, efficiently detect anomalies in the surveillance area or in people to be monitored and, in doing so, to optimize the robot's running time with regard to the battery level, and, on the other hand, to enable close monitoring of people.

To address this problem, in one aspect the robot has a computer-implemented method for triggering a behavior of a mobile robot depending on anomalies detected by means of a radar sensor unit, comprising the movement of the robot on a path, the detection of objects or obstacles in the radar detection range, the position and/or time-related evaluation of the detection results, a comparison of detection results with previously stored detection results and the determination and evaluation of the deviation between the detection results and the previously stored detection results. The trigger event can, for example, include the adjustment of the path planning of the robot and/or the triggering of an event, e.g. the transmission of a signal via a wireless interface, e.g. to a terminal. An anomaly includes, for example, detection results that deviate from expected and/or stored detection results. The anomalies described in these steps can be associated with positions on a map, the detection of objects in the radar detection area can include the detection of people, the position- and/or time-related evaluation of the detection results can be based on defined sub-areas of the area covered by the radar sensor unit as a whole and these sub-areas can be evaluated individually, for example, and/or the position- and/or time-related evaluation of the detection results can be carried out, for example, in a multi-level map of the robot. Compared to the method steps described so far, further optional steps can include storing the position- and/or time-related evaluation of the detection results in a position and rule database, whereby the detection results can be, for example, vital parameters of a person. The radar signal detection area can, in one aspect, be aligned essentially perpendicular to the operating path of the robot. The position data used in the method, which were determined by evaluating radar sensor data, can be standardized, for example, by data determined using a laser scanner, camera, ultrasonic sensor and/or odometry unit. Overall, there can be a wall or closed door between the robot and the detected person.

In one aspect, the technical problem outlined at the beginning includes a sub-problem, namely a simplified measurement of vital parameters without necessarily having to resort to reference values for the signal strength in a table, because this in turn would involve further steps to determine the reference value outside the actual measurement procedure, e.g. in the context of a calibration.

A method is therefore used that does not require such reference values. This computer-implemented method for recording at least one vital parameter by a mobile robot includes the movement of the robot on a path, the determination of the current position of the robot, the detection of objects or obstacles in the radar detection area, the detection of people in the radar detection area, the recording of vital parameters of a detected person in the radar detection area, a comparison of the vital parameter measurement results and adjustment of the signal strength depending on the comparison of the vital parameter measurement results and the implementation of a vital parameter evaluation. In this case, for example, the signal strength can be adjusted (in one direction) until the measured vital parameters deteriorate in quality after an adjustment of the signal strength. This improvement can be determined, for example, from one adjustment step to the next. If, for example, a deterioration is determined from one adjustment step to the next, the signal strength can be adjusted in the other direction. The method can also include a vital parameter assessment, for example a position- and/or time-related evaluation of the detection results. The method can alternatively and/or additionally include triggering an event such as a signal transmission to a cloud and/or to a terminal. In addition to the aforementioned steps, the person's position can be localized, if necessary supplemented by a comparison of the object position with object positions and/or object zones stored in memory 15. The vital parameters can, for example, be recorded for a person at a localized position. The signal evaluation of detected radar signals can, for example, represent a comparison with patterns. For example, the position data used, which were determined by evaluating radar sensor data, can be standardized by data determined using a laser scanner, camera, ultrasonic sensor and/or odometry unit. In one aspect, the robot takes a scanning position in front of a possibly closed door at a defined distance and performs a rotational movement in order to detect an area using a radar sensor unit on its path. In an optional aspect, it can move along a convex path in front of the said door and maintain a defined minimum distance from the door.

In one aspect, the technical problem outlined at the beginning includes a sub-problem, namely the detection of objects and/or vital parameters through walls, which cause a high attenuation of the radar signals and thus make it difficult to detect objects and/or vital parameters precisely.

This problem is addressed in the following embodiment. The robot has a computer-implemented method comprising moving the robot along a path, determining the current position of the robot, assuming a scanning position, and performing a rotational movement, wherein objects and/or vital parameters are detected at the scanning position by a radar sensor unit of the robot. The scanning position can be determined, for example, from a map or by detecting and evaluating patterns. The scanning position can, for example, have a defined distance from a door. The angle of the rotational movement mentioned can, for example, be at least 20°. The rotational movement can, in one aspect, be a rotational movement about a rotationally symmetrical axis of the robot. The rotational movement includes, for example, a rotation in one direction, optionally followed by a rotation in another direction. In one aspect, the robot moves parallel to walls when approaching a door, performs a rotational movement towards a door, followed by traveling along a convex path in front of the door (with the convexity seen from the door). Here, for example, the minimum distance of the convex path to the door corresponds to the defined distance to the door.

The derivation of rules based on the positions of objects, people and/or values associated with them, such as vital parameters, can, on the one hand, imply that the positions are positions within spaces defined in a map that are not entered, for example, during the robot's operational operation (i.e. during a patrol drive). However, it is also possible to define positions via the positions in areas that are entered/traveled by the robot during operational operation, if necessary in combination with a direction vector that has an orientation and optionally defined length and refers to coordinates that the robot takes or can take. The former means that a coordinate value from the robot's path is associated, for example, with a defined direction vector or direction vector range that can have a defined length. The latter means that, for example, certain areas of an area that the robot can travel through are associated with direction vectors or direction vector ranges. This approach allows rules to be established for the rooms in a map without a specific definition of the rooms, so that a position evaluation or assessment for persons or objects in the rooms is possible.

drawings

• 1 mobiler Roboter mit Cloud-Anbindung
• 2 Hardware-Komponenten im mobilen Roboter
• 3 Radarsensoreinheit
• 4 Software-Komponenten im mobilen Roboter
• 5 Ablauf Personenüberwachung
• 6 Ablauf Obj ektlokalisation
• 7 Initiale Kartierung
• 8 Anordnung Roboter bei Überwachungsvorgang
• 9 Ablauf Vitaldatenerfassung unter Anpassung der Radarsignalstärke
• 10 Ablauf Anomaliedetektion
• 11 Vitalparameter-Erfassung mit Regelkreis
• 12 Schritte der Vitalparameter-Messung
• 13 Einnehmen Abtastposition

Show it:

• 1 mobile robot with cloud connection
• 2 Hardware components in the mobile robot
• 3 Radar sensor unit
• 4 Software components in the mobile robot
• 5 Personal surveillance process
• 6 Object localization process
• 7 Initial mapping
• 8th Arrangement of robots during monitoring process
• 9 Vital data acquisition process with adjustment of radar signal strength
• 10 Anomaly detection process
• 11 Vital parameter recording with control loop
• 12 Steps of vital parameter measurement
• 13 Take scanning position

Detailed description of the invention

For robot 1 (see 1 ) is a mobile base 2, which can be commercially available, for example (e.g. Fetch100 OEM from Fetch Robotics, TIAGo from PAL Robotics or the TORY base from MetraLabs). There can be either a laser scanner 10 for obstacle detection and also mapping the environment and/or at least one 3D camera 11. These are preferably arranged in the direction of travel of the robot 1. The mobile base 2 has a battery 12 and a charging device 13 for this battery as well as at least one computer unit 14 with an integrated or connected memory 15 for controlling the robot 1 and a wireless interface 16 such as a WLAN module. Furthermore, the mobile base 2 can have an optional safety control 17 to slow the robot 1 to a safe speed when obstacles are detected in its environment. To brake in the event of a collision, the mobile base 2 can also have a circumferential protective contact strip 18. In one aspect, the robot also has two drive wheels 20, which are driven by a drive motor 21, and several support wheels 24. The robot 1 is connected to a cloud 4 via the wireless interface 16 and via this or directly to a terminal 5.

2 summarizes the important hardware components. Here, for example, it is also clear that the mobile base 2 has, in addition to the already described embodiments, an odometry unit 23, which is described elsewhere as an example (e.g. DE102020006337A1 ).

The mobile base 2 has a structure 3, with which the robot 1 reaches a height of, for example, at least 50 cm, preferably of, for example, at least 90 cm. The at least one radar sensor unit 30 is positioned essentially perpendicular to the direction of travel of the robot 1, for example at a height of 85 cm. It is located with an amplifier 33 in a radar sensor unit 30 (see also 3 ). For example, in order to correct movements of a person to be detected, two synchronized radar sensor units 30 can be used,

The robot has, as 4 can be seen, on the software level also via a path planning 40 and a map module 41 for creating and evaluating maps (42-44) as well as a person position determination module 45, a vital data acquisition module 46 and an object position determination module 47, which can, for example, be complementary to the map module 41. The object position determination module 47 uses data from the radar sensor unit 30 and possibly camera 11 or laser scanner 10 to detect obstacles or objects.

The core of the radar sensor unit 30 is the radar sensor 31, which can be, for example, a continuous wave Doppler radar, a frequency-modulated continuous wave Doppler radar (FMCW) or an (impulse radio) ultra-wideband radar, e.g. a UWB radar in the sub-10 GHz range from Novelda such as the X4 transceiver. In one aspect, the radar signal is emitted by the radar sensor unit 30 by means of at least one radar antenna 32, which is, for example, a directional antenna in order to obtain better noise reduction. However, several radar antennas 32 can also be used, e.g. antenna arrays. An overview of radar systems that can be used, for example, to measure heart rate, as well as corresponding measuring methods can be found in Maleševic et al. (2020, DOI:10.3390/s20082351).

The monitoring of objects or persons, especially within certain areas, is a process in 5 Some of the aspects referred to are also referred to 7 The robot 1 moves in step 110 in one aspect along a path, for example within a corridor 50. In doing so, radar signals are emitted and received essentially perpendicular to the direction of travel by the radar sensor unit 30, which penetrate at least one wall 60 and thus cover a room 53, thereby detecting an area to be monitored. Alternatively and/or additionally, the radar signals can also be used to directly monitor a person (total step 110 - robot follows path and monitors area using radar sensor). The signal strength is set in an optional aspect so that the area to be detected is covered by the radar signals, so that person detection is possible in the entire area to be detected in terms of depth, for example with a range up to the rear wall 61 of the room (e.g. through the radar signal detection area 59 in 8th In step 120, the radar signal is evaluated to identify, for example, objects where One aspect of this, the person detection, is carried out in step 340 in 10 described. The signal evaluation in step 120, in one aspect a classification of the signals, includes, for example, a comparison with signals or signal characteristics stored in the memory 15. The classification can be, for example, a pattern recognition (see, for example, EP2100153A1 or DOI: 10.1007/s00521-022-06886-2). In an optional step 130, a distinction is made between moving and stationary objects. Moving objects are identified, for example, by the fact that the received signals change in their intensity over time or that the signal propagation time changes here. In one aspect, the relative signal propagation time is understood, which takes into account movements of the robot 1. In step 140, the object position is localized. This is done, for example, by evaluating the relative detection angle from the robot, the signal propagation time or, if applicable, the signal strength and a comparison with the current position or pose of the robot 1. Pose here is understood to mean its orientation in space. In summary, the coordinates and the orientation in space form the robot position. This can be used to determine the position of the detected object or obstacle in space and in the robot's coordinate system. The robot 1 can determine its position, for example, via the odometry unit 23, in one aspect by including other sensors such as laser scanner 10 or camera 11 (for example as part of a visual SLAM method). Overall, steps 120, 130 and 140 take place in the object position determination module 47. In an optional step 145, the robot updates its map 42 by entering detected obstacles or objects in it that were determined via the radar sensor unit 30.

In step 150, the object position is compared with object positions stored in memory 15, in one aspect a classification. For example, a position can be stored for an object type such as a bed 54. For example, zones can also be stored in a table or a map that describe within which coordinates which objects or people may be located and zones in which certain objects such as people may not be located (forbidden zones). Comparison at this point means, for example, that the signal pattern identified as bed 54 in step 120 is compared with a bed position stored in memory or a zone that is associated with a bed. In the same way, a signal identified as a person (e.g. 56-58) in step 120 is compared with positions stored for people in memory 15. If the object positions identified by the radar signal deviate from stored object positions in step 160, an event is triggered, which can be defined in stored rules, which can be located, for example, in a table stored in the memory 15 and/or in the position and rule database 48. This event can be a notification to an application in the cloud 4 or a terminal 5. This can include, for example, an alarm. For example, the position and rule database 48 can store that no person is allowed to be in certain zones of a room 53 at certain times, or conversely, that people are only allowed to be in certain zones at certain times (such as in bed 54).

The signal evaluation or pattern recognition, for example based on a classification in step 120, can be carried out by detecting people 57 and people's movements using a radar sensor, which does not necessarily have to be implemented on the robot 1, but can also be used in a laboratory environment. Various movement patterns can be trained here. In one aspect, imitation learning can be used here (see e.g. Wang, Borui, et al. "Imitation learning for human pose prediction." Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019, which, however, use skeletal data to classify body poses, whereas radar signals are used here). The classification method can be action recognition (see e.g. Sun et al. 2022, DOI: 10.1109/TPAMI.2022.3183112 or Wang, Mingyang, Yimin D. Zhang, and Guolong Cui. “Human motion recognition exploiting radar with stacked recurrent neural network.” Digital Signal Processing 87 (2019): 125-131) .

In one aspect, the signal evaluation in step 120 alternatively and/or additionally comprises a determination of at least one vital parameter such as the heart or pulse rate, respiration, etc. in the vital data acquisition module 41. Vital parameters in this document are understood to mean, for example, but not exclusively, pulse rate, heart rate variability, etc., the detection or derivation of which, for example, in WO2021038109A1 In the state of the art, corresponding methods for radar-based detection are known (e.g. Maleševic et al. (2020, DOI:10.3390/s20082351) . It will also be 12 The determination of at least one vital parameter can be carried out independently of the object identification and/or object position (where the objects can be people) and thus directly (see right dashed arrow in 5 ) lead to a comparison in step 125, in which a classification of the at least one determined vital parameter takes place. The classifier used here can be determined, for example, by means of machine learning methods. and indicate that the measured value is critical (e.g. a value that is too high or too low). A threshold comparison can also take place here, for example. In the case of this combination in the evaluation of at least one vital parameter and, for example, the object position, in this case, for example, the position of a person, at least one vital parameter can be determined for the person at the said position. This can then be used to make one or two comparisons (that of the position parameter in step 150 and that of the vital parameter in step 125), followed by step 160, or an event can be triggered directly in step 160. With regard to the event triggering in step 160 with reference to vital parameters, this can mean that an alarm is triggered if a vital parameter recorded for a person in the monitoring area is, for example, outside a value range defined in the position and rule database 48.

Example 1:

• A1. Computer-implementiertes Verfahren zur Überwachung einer Fläche durch einen mobilen Roboter 1, umfassend
  • • Abfahren eines Betriebspfads 63,
  • • Erfassung einer zu überwachenden Fläche mittels mindestens einer Radarsensoreinheit 30,
  • • Signalauswertung erfasster Radarsignale und
  • • Auslösen eines Ereignisses basierend auf der Auswertung,
  wobei die Erfassung der zu überwachenden Fläche durch eine Wand 60 und/oder geschlossene Tür 52 hindurch erfolgt.
• A2. Computer-implementiertes Verfahren nach A1, wobei die Erfassung der zu überwachenden Fläche in etwa senkrecht zur Fahrtrichtung des Roboters 1 erfolgt.
• A3. Computer-implementiertes Verfahren nach A1-A2, wobei die Signalauswertung die Identifikation von Objekten inkl. Personen umfasst.
• A4. Computer-implementiertes Verfahren nach A1-A3, wobei die Signalauswertung weiter eine Differenzierung in bewegliche und unbewegliche Objekte umfasst.
• A5. Computer-implementiertes Verfahren nach A1-A4, wobei die Signalauswertung eine Ermittlung von mindestens einem Vitalparameter umfasst.
• A6. Computer-implementiertes Verfahren nach A1-A3, weiter umfassend eine Lokalisierung der Objektposition.
• A7. Computer-implementiertes Verfahren nach A3-A6, weiter umfassend ein Vergleich der Objektposition mit im Speicher 15 hinterlegten Objektpositionen und/oder Objektzonen.
• A8. Computer-implementiertes Verfahren nach A7, wobei die im Speicher 15 hinterlegten Objektzonen verbotene Zonen sein können.
• A9. Computer-implementiertes Verfahren nach A5, weiter umfassend der Vergleich ermittelter Vitalparameter mit hinterlegten Vitalparametern und/oder eine Vitalparameterbewertung.
• A10. Computer-implementiertes Verfahren nach A1-A9, wobei das Auslösen eines Ereignisses eine Signalübermittlung in die Cloud 4 oder an ein Terminal 5 umfasst.
• A11. Computer-implementiertes Verfahren nach A5, wobei die Vitalparameter für eine Person an einer lokalisierten Position erfasst werden.
• A12. Computer-implementiertes Verfahren nach A1-A11, wobei die Signalauswertung erfasster Radarsignale einen Vergleich mit Mustern darstellt.
• A13. Computer-implementiertes Verfahren nach A1-A12, wobei für die Signalauswertung ein Klassifikator genutzt wird, der mittels Imitation Learning erstellt wurde oder der Action Recognition nutzt.
• A14. Computer-implementiertes Verfahren nach A6, wobei genutzte Positionsdaten, die durch Auswertung von Radarsensordaten ermittelt wurden, durch mittels Laserscanner 10, Kamera 11, Ultraschallsensor und/oder Odometrieeinheit 23 ermittelten Daten normiert werden.
• A15. Computer-implementiertes Verfahren nach A3 oder A6, weiter umfassend die Aktualisierung der Karte 42 des Roboters 1 durch Übernahme der mittels Radarsensoreinheit 30 detektierten Hindernisse.
• A16. Computer-implementiertes Verfahren nach A1, wobei der Roboter 1 eine Abtastposition vor einer Tür 52 in einem definierten Abstand einnimmt und eine Rotationsbewegung durchführt.
• A17. Computer-implementiertes Verfahren nach A1, wobei der Roboter 1 sich entlang eines konvexen Pfads vor einer Tür 52 bewegt und dabei einen definierten Mindestabstand zur Tür 52 einhält.
• A18. Vorrichtung zur Durchführung des Verfahrens nach A1-A19.

In summary, the computer-implemented procedure can be described as follows:

• A1. Computer-implemented method for monitoring an area by a mobile robot 1, comprising
  • • Driving along an operating path 63,
  • • Detection of an area to be monitored by means of at least one radar sensor unit 30,
  • • Signal evaluation of detected radar signals and
  • • Triggering an event based on the evaluation,
  wherein the detection of the area to be monitored takes place through a wall 60 and/or closed door 52.
• A2. Computer-implemented method according to A1, wherein the detection of the area to be monitored occurs approximately perpendicular to the direction of travel of the robot 1.
• A3. Computer-implemented method according to A1-A2, whereby the signal evaluation includes the identification of objects including persons.
• A4. Computer-implemented method according to A1-A3, wherein the signal evaluation further comprises a differentiation between moving and immobile objects.
• A5. Computer-implemented method according to A1-A4, wherein the signal evaluation comprises a determination of at least one vital parameter.
• A6. Computer-implemented method according to A1-A3, further comprising localizing the object position.
• A7. Computer-implemented method according to A3-A6, further comprising a comparison of the object position with object positions and/or object zones stored in the memory 15.
• A8. Computer-implemented method according to A7, wherein the object zones stored in the memory 15 can be prohibited zones.
• A9. Computer-implemented method according to A5, further comprising the comparison of determined vital parameters with stored vital parameters and/or a vital parameter evaluation.
• A10. Computer-implemented method according to A1-A9, wherein the triggering of an event comprises a signal transmission to the cloud 4 or to a terminal 5.
• A11. Computer-implemented method according to A5, wherein the vital parameters for a person are recorded at a localized position.
• A12. Computer-implemented method according to A1-A11, wherein the signal evaluation of detected radar signals represents a comparison with patterns.
• A13. Computer-implemented method according to A1-A12, whereby a classifier created by means of imitation learning or using action recognition is used for the signal evaluation.
• A14. Computer-implemented method according to A6, wherein used position data determined by evaluating radar sensor data are standardized by data determined by means of laser scanner 10, camera 11, ultrasonic sensor and/or odometry unit 23.
• A15. Computer-implemented method according to A3 or A6, further comprising updating the map 42 of the robot 1 by incorporating the obstacles detected by means of the radar sensor unit 30.
• A16. Computer-implemented method according to A1, wherein the robot 1 assumes a scanning position in front of a door 52 at a defined distance and performs a rotational movement.
• A17. Computer-implemented method according to A1, wherein the robot 1 moves along a convex path in front of a door 52 while maintaining a defined minimum distance from the door 52.
• A18. Device for carrying out the method according to A1-A19.

Example 2: Procedure for updating maps for surveillance purposes

In one aspect, the technical challenge in the context described so far is to correctly localize objects in the area to be monitored and, for example, to be able to accurately record vital parameters.

To achieve this goal, robot 1 creates according to 6 in step 210, as is known in the art, a first map 42 such as a so-called occupancy grid map for navigation purposes. This first map 42 is preferably created with laser and/or camera data (e.g. additionally or alternatively also using ultrasonic sensor data), in one aspect (only on the basis of) or preferably with the inclusion of odometry data. The coordinate system of the first map 42 is used by the robot 1 for navigation. A Monte Carlo localization can be used here, for example, as a method for mapping. For the purpose of mapping, the robot 1 moves in its area of operation and uses laser scanner 10 or camera 11 to detect stationary objects, such as walls (60, 61), in particular in the direction of travel (preferably only), which specifically function as obstacles. This can be done, for example, as part of a SLAM method. Ideally, the robot will cover its entire area of operation in order to create the first map 42. This is 7 where the robot 1 moves along a corridor 50, drives through open doors 51 into rooms 53. The exploration path 55, which is created, for example, during initial mapping of the area of operation, is selected in such a way that the areas of the area of operation are fully covered by the laser scanner detection area 62. It should be noted that the laser scanner detection area 62 in 6 smaller than in reality for many mobile robots. Typically, the detection depth is several meters. Instead of the laser scanner 10, a camera 11 with a slightly differently shaped detection area can be used, or a combination of several sensors of the same or different type or functional principle.

When creating the map in step 210, the robot 1 detects the obstacles in its surroundings and relative to its current position from its respective position and enters them as obstacles in the map. The current position can be determined using an odometry unit, for example, or as in WO2021254975A1 described as a type of SLAM method.

In step 220, a second map 43 is created using the data acquired by the radar sensor unit 30, which includes moving and/or stationary objects in the radar signal detection area 59. Such map creations are known from the prior art (e.g. Li, M., et al. 2018, High Resolution Radar-based Occupancy Grid Mapping and Free Space Detection., In Proceedings of the 4th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2018), pages 70-81, ISBN: 978-989-758-293-6) The data for this second map 43 is generated essentially perpendicular to the direction of travel of the robot 1. In terms of time, the creation of the second map 43 takes place at a later time than the creation of the first map 42. However, the first map 42 can be updated in parallel with the creation of the second map, in particular as part of a SLAM process.

In step 230, the determined position data of the moving and/or stationary objects or obstacles from the second map 43 are compared with position data of the stationary objects or obstacles from the first map 42. This can be done, for example, by means of data fusion. The position data of the stationary objects or obstacles on the first map 42 are used as reference values and the position of the objects or obstacles shown in the second map 43 is corrected (step 240). In fact, this means a synchronization of static object data or obstacle data. In practice, however, this only affects objects or obstacles that were already present previously and therefore primarily stationary obstacles or objects such as walls (60, 61) or, for example, beds 54, etc. This leads to a better relative position estimate of the detected moving obstacles or objects in the second map (43). In one aspect, this correction is only carried out for objects/obstacles that cannot be detected at the same time by the laser scanner 10 or camera 11. This means, for example, that as in 8th persons 57-58 are detected who were not in the rooms 53 when the first map 42 was created. The position of these persons is therefore estimated, for example, relative to the walls (60, 61), which in turn are located in the coordinate system that the robot 1 uses for navigation.

As an alternative to step 240, in which the position of the objects or obstacles in the second map 43 is corrected, the objects or obstacles from the second map 43 that were not included in the first map 42 can be transferred to the first map 42 (step 250). This applies, for example, primarily to moving objects or obstacles. As a result, the first map 42 can therefore contain moving objects or obstacles that were detected by data from the laser scanner 10 and/or the camera 11 as well as data that was recorded by the radar sensor unit 30. Alternatively, a third map 44 can also be created that contains the information functions of the first map 42 as a basis and adopts the additional obstacles or objects contained in the map 43 as described (step 260). In one aspect, the coordinate system of the first map 42 can be adopted and, for example, this third map 44 can be used as a basis for robot navigation. In summary, it can be stated that in steps 230, 240 and 250 the position data of objects or obstacles from the second map 43 are at least partially corrected. Furthermore, the method described here implies that the position data for objects or obstacles determined from the radar sensor data are normalized by position data determined by means of laser scanner 10 or camera 11 (possibly also ultrasonic sensor(s)) and possibly alternatively and/or additionally by means of odometry data, since the position correction is carried out for stationary objects or obstacles.

The map fusion or combination of the detected obstacles can be carried out, for example, as in Steinbaeck, Josef, et al. “Occupancy grid fusion of low-level radar and time-of-flight sensor data.” 2019 22nd Euromicro Conference on Digital System Design (DSD). IEEE, 2019.

In one aspect, object data is classified in order to detect persons in the detection range of the radar sensor unit 30 (or of its radar antenna(s) 32, as described by way of example elsewhere in this document, for example in order to evaluate the position of at least one person on the basis of rules, possibly in conjunction with the detection of at least one vital parameter of this person.

The position data thus determined for the objects detected by the radar sensor unit 30 can be used in step 140, which was already described above. Alternatively and/or additionally, the position data for moving objects determined by evaluating the radar sensor data are compared with position data from zones contained in one of the maps 42-44, a fourth map and/or a table. This corresponds, for example, to step 150. These zones are in turn assigned rules. For example, a rule can state that when an object is identified in a zone, an event is triggered by the robot 1. In a next step, such an event can then be triggered by the robot 1, e.g. the notification of a terminal 5.

• B1. Computer-implementiertes Verfahren zur Ermittlung, zur Korrektur oder zur Synchronisierung von Objekt-Positionsdaten, umfassend die folgenden Schritte:
  • • Erstellung einer ersten Karte 42 durch einen mobilen Roboter 1 mit Hindernissen bzw. Objekten basierend auf Laserscanner-, Kamera- und/oder Ultraschallsensor-Daten,
  • • Erstellung einer zweiten Karte 43 durch einen mobilen Roboter 1 mit Hindernissen bzw. Objekten basierend auf Radarsensor-Daten,
  • • Vergleich der Positionsdaten für in den Karten (42, 43) jeweils enthaltenen Hindernissen bzw. Objekten,
  wobei die Positionsdaten von Objekten bzw. Hindernissen der zweiten Karte 43 mindestens anteilig durch Positionsdaten von Objekten bzw. Hindernissen der ersten Karte 42 korrigiert werden.
• B2. Computer-implementiertes Verfahren nach B2, wobei es sich bei den korrigierten Positionsdaten um Positionsdaten von stationären Hindernissen bzw. Objekten handelt.
• B3. Computer-implementiertes Verfahren nach B1-B2, wobei die zweite Karte 43 zeitlich nach der ersten Karte 42 erstellt wird.
• B4. Computer-implementiertes Verfahren nach B1, wobei der für die Ermittlung der Radarsensor-Daten genutzte Radarsensoreinheit 30 eine Fläche erfasst, die nicht zugleich durch den Laserscanner 10, eine Kamera 11 oder einen Ultraschallsensor erfassbar ist.
• B5. Computer-implementiertes Verfahren nach B1, wobei die Radarsensor-Daten Objekte bzw. Hindernisse hinter einer Wand 60 oder Tür 52 umfassen.
• B6. Computer-implementiertes Verfahren nach B1-B5, wobei die korrigierten Positionsdaten der zweiten Karte 43 in der zweiten Karte 43 hinterlegt werden.
• B7. Computer-implementiertes Verfahren nach B1-B5, weiter umfassend das Hinzufügen von Positionsdaten beweglicher Objekte aus der zweiten Karte 43 in die erste Karte 42.
• B8. Computer-implementiertes Verfahren nach B1-B5, weiter umfassend die Erzeugung einer dritten Karte 44 durch Übernahme stationärer und beweglicher Objekt- bzw. Hindernispositionen der ersten Karte 42 und Übernahme zusätzlicher beweglicher Objekt- bzw. Hindernispositionen der zweiten Karte 43.
• B9. Computer-implementiertes Verfahren nach B8, wobei die Koordinaten der ersten Karte (42) übernommen werden und der Roboter (1) nach der dritten Karte 44 navigiert.
• B10. Computer-implementiertes Verfahren nach B 1-B9, weiter umfassend Vergleich der mit Hilfe der Radarsensoreinheit 30 ermittelten Positionsdaten für bewegliche Objekte mit in einer Tabelle, in einer der Karten (42-44) oder in einer weiteren Karte hinterlegten Zonen für bewegliche Objekte, wobei den Zonen jeweils Regeln zugeordnet sind.
• B11. Computer-implementiertes Verfahren nach B10, weiter umfassend das Auslösen eines Ereignisses basierend auf den Regeln für die jeweilige Zone.
• B12. Computer-implementiertes Verfahren nach B1 - B11, weiter umfassend Nutzung der Objekt-Positionsdaten zur Anpassung der Signalstärke des Radarsignals.
• B13. Computer-implementiertes Verfahren nach B1-B12, weiter umfassend die Klassifizierung von Objektdaten zur Detektion von Personen.
• B14. Computer-implementiertes Verfahren nach 13, weiter umfassend die Erfassung von mindestens einem Vitalparameter bei einer detektierten Person.
• B15. Vorrichtung zur Durchführung des Verfahrens nach B1-B14.

The procedure is as follows:

• B1. A computer-implemented method for determining, correcting or synchronizing object position data, comprising the following steps:
  • • Creation of a first map 42 by a mobile robot 1 with obstacles or objects based on laser scanner, camera and/or ultrasonic sensor data,
  • • Creation of a second map 43 by a mobile robot 1 with obstacles or objects based on radar sensor data,
  • • Comparison of the position data for obstacles or objects contained in the maps (42, 43),
  wherein the position data of objects or obstacles of the second map 43 are at least partially corrected by position data of objects or obstacles of the first map 42.
• B2. Computer-implemented method according to B2, wherein the corrected position data are position data of stationary obstacles or objects.
• B3. Computer-implemented method according to B1-B2, wherein the second map 43 is created chronologically after the first map 42.
• B4. Computer-implemented method according to B1, wherein the radar sensor unit 30 used to determine the radar sensor data detects an area that cannot be simultaneously detected by the laser scanner 10, a camera 11 or an ultrasonic sensor.
• B5. Computer-implemented method according to B1, wherein the radar sensor data comprises objects or obstacles behind a wall 60 or door 52.
• B6. Computer-implemented method according to B1-B5, wherein the corrected position data of the second map 43 are stored in the second map 43.
• B7. Computer-implemented method according to B1-B5, further comprising adding position data of moving objects from the second map 43 into the first map 42.
• B8. Computer-implemented method according to B1-B5, further comprising the generation of a third map 44 by adopting stationary and moving object or obstacle positions of the first map 42 and adopting additional moving object or obstacle positions of the second map 43.
• B9. Computer-implemented method according to B8, wherein the coordinates of the first map (42) are adopted and the robot (1) navigates according to the third map 44.
• B10. Computer-implemented method according to B 1-B9, further comprising comparing the position data for moving objects determined with the aid of the radar sensor unit 30 with zones for moving objects stored in a table, in one of the maps (42-44) or in another map, wherein rules are assigned to each zone.
• B11. Computer-implemented method according to B10, further comprising triggering an event based on the rules for the respective zone.
• B12. Computer-implemented method according to B1 - B11, further comprising use of the object position data to adjust the signal strength of the radar signal.
• B13. Computer-implemented method according to B1-B12, further comprising the classification of object data for the detection of persons.
• B14. Computer-implemented method according to 13, further comprising the detection of at least one vital parameter of a detected person.
• B15. Device for carrying out the method according to B1-B14.

Example 3: Recording vital parameters through a wall over a distance of several meters

The technical task is, in the context of the general task, on the one hand to reliably record vital parameters (such as heart rate, pulse rate, etc.) of a person over a distance of several meters and through a wall or door, but on the other hand also to keep the radiation exposure caused by the radar sensor unit 30 used for this purpose for the person whose vital parameters are to be recorded as low as possible and at the same time to keep the energy consumption of the robot 1 low in order to optimize its range or minimize the operating time through battery charging processes.

To achieve this goal, the radar sensor unit 30 has an amplifier 33 that is power modulated. Power modulated means that its power is varied over time. The method for recording vital parameters is as follows and is explained using 8th and 10 illustrated:

In step 310, the robot 1 moves along a path, also referred to as the operating path 63, since, in contrast to the exploration path 55 (see 7 ) represents the path of the robot 1 in regular robot operation. In this case, this runs, for example, along a corridor 50, on the side of which there are, for example, several rooms 53. For the purpose of self-localization, the robot 1 accesses its position data via its map (eg 42), which also describes its position, e.g. on the operating path 63 in the corridor 50 (step 320). This position data is required in an optional aspect in order to determine the value for the signal strength with which the radar sensor unit 30 should monitor the detection area.

In a next optional step 325, the distance between the robot position and the edge area of the area to be monitored facing away from the robot 1 is determined, also referred to as determining the detection depth. This is in 8th the rear wall 61. Such information is optionally stored in the robot's memory 15, for example with reference to the map 42. This value is then optionally compared with a value in a table, which in turn has a suitable signal strength, which is then set by the robot (determination of corresponding signal strength for detection depth - (optional) step 328). This signal strength is selected so that it sufficiently enables person detection and the position associated with the person in the spatial area to be detected. It is usually lower than the signal strength that is necessary for recording the vital parameters. Steps 325-328 are therefore to be understood as optional steps.

The robot 1 emits and receives radar signals essentially perpendicular to the direction of travel by means of the radar sensor unit 30 (step 330 - detection of objects or obstacles in the radar signal detection area 59). These penetrate at least one wall 60 or door 52. This is possible, for example, through the signal cone (radar signal detection area 59) in 8th recognizable. The signal strength is set in an optional aspect so that the area to be detected is covered by the radar signals, which makes it possible to record vital parameters of a person in the area. For this purpose, the amplifier 33 is used, for example, to set the signal strength accordingly. For this purpose, in an optional aspect, there are values stored in a memory 15 that show the signal strength depending on the depth of the area to be detected.

In step 340 - person detection, it is evaluated whether a person (56-58) is located in the area to be detected. The radar sensor data for the detected objects or obstacles are classified and/or compared with stored values that are located in a memory 15 (e.g. an object identification database 49). The classifier used for person detection can be based on machine learning algorithms, as is well known from the state of the art. known (e.g. EP2100153A1 or DOI: 10.1007/s00521-022-06886-2). On the one hand, these can be (quasi-)stationary objects or obstacles such as a person in bed 56 or people outside the bed (57-58), which are present as stationary or moving objects in the radar sensor data.

In step 350, the position of the person is determined. To do this, the distance between robot 1 and the detected person is determined, for example via the signal propagation time, as is well known from the prior art, and the angle at which the person is located in the room is derived from the pose of the robot and the known angle from the radar antenna alignment on the robot. The associated position can be transferred to a map (for example 42-44) in an optional step 355. For this distance, which can also be determined from the map based on the optional step, in step 360 - determination of signal strength - a value for the signal strength is read out from a memory 15 (for example via a table stored there), which is necessary for recording the vital parameters of the detected person.

The radar amplifier 33 is then adjusted in step 370 so that its output power enables a vital parameter measurement of the person (referred to as adjustment of the signal strength). This measurement is then carried out accordingly, as referenced elsewhere in this document (step 380, elsewhere part of step 120). In many cases, this corresponds to an increase in the signal strength compared to the value used for person detection. The signal strength is then adjusted again in step 390 to the level that is necessary to detect people in the area to be evaluated, i.e. in the example of 8th in room 53, also known as adjusting the amplifier output. In many cases this is a reduction in the signal strength of the radar sensor 31.

This ensures that the higher power required to measure vital parameters is only used for a short time and only at a level that is necessary for the measurement. This means that, for example, the radiation exposure is lower if a person 58 is closer to the robot than further away (e.g. person 57). This ensures, for example, that people such as 58 who are close to robot 1 are not exposed to radiation as if the vital parameters were measured for a person far away from robot 1 (such as person 57).

• C1. Computer-implementiertes Verfahren zur Anpassung der Signalstärke eines Radarsensors 31 auf einem mobilen Roboter 1, umfassend
  • • Bewegung des Roboters 1 auf einem Pfad 62
  • • Ermittlung der aktuellen Position des Roboters 1
  • • Erfassung von Objekten bzw. Hindernissen im Radarsignal-Erfassungsbereich 59
  • • Detektion von Personen im Radarsignal-Erfassungsbereich 59
  • • Ermittlung der Distanz zwischen Roboter 1 und der detektierten Person und
  • • Ermittlung einer Signalstärke für den Radarsensor 31 auf Basis der ermittelten Distanz zwischen Roboter 1 und der Person.
• C2. Computer-implementiertes Verfahren nach C1, weiter umfassend die Anpassung der Signalstärke des Radarsensors 31.
• C3. Computer-implementiertes Verfahren nach C1-C2, weiter umfassend die Durchführung einer radardaten-basierten Vitalparametermessung bei der detektierten Person.
• C4. Computer-implementiertes Verfahren nach C3, weiter umfassend die erneute Anpassung der Signalstärke des Radarsensors 31.
• C5. Computer-implementiertes Verfahren nach C2, wobei die Signalstärke erhöht wird.
• C6. Computer-implementiertes Verfahren nach C4, wobei die Signalstärke reduziert wird.
• C7. Computer-implementiertes Verfahren nach C1, wobei die Personenposition in einer Karte (42, 43, 44) hinterlegt wird.
• C8. Computer-implementiertes Verfahren nach C1-C7, weiter umfassend die Ermittlung der notwendigen Erfassungstiefe des Radarsensors 31 durch Abgleich des zu erfassenden Bereich mit einer Karte 42.
• C9. Computer-implementiertes Verfahren nach C8, weiter umfassend die Anpassung der Signalstärke des Radarsensors 31 auf Basis der notwendigen Erfassungstiefe, welche eine Personendetektion im zu erfassenden Bereich ermöglicht.
• C10. Computer-implementiertes Verfahren nach C1-C9, weiter umfassend die Übermittlung eines Signals an ein Terminal 5 basierend auf der Auswertung der ermittelten Vitalparameter.
• C11. Computer-implementiertes Verfahren nach C1-C10, wobei genutzte Positionsdaten, die durch Auswertung von Radarsensordaten ermittelt wurden, durch mittels Laserscanner 10, Kamera 11, Ultraschallsensor und/oder Odometrieeinheit 23 ermittelte Daten normiert werden.
• C12. Computer-implementiertes Verfahren nach C1-C11, wobei sich zwischen Roboter 1 und detektierter Person eine Wand 60 oder geschlossene Tür 52 befindet.
• C13. Computer-implementiertes Verfahren nach C1-C11, wobei der Radarsignal-Erfassungsbereich 59 im Wesentlichen senkrecht zum Pfad des Roboters 62 ausgerichtet ist.
• C14. Vorrichtung zur Durchführung des Verfahrens nach C1-C13.

The procedure described here can be represented as follows:

• C1. Computer-implemented method for adjusting the signal strength of a radar sensor 31 on a mobile robot 1, comprising
  • • Movement of robot 1 on a path 62
  • • Determining the current position of the robot 1
  • • Detection of objects or obstacles in the radar signal detection range 59
  • • Detection of persons in the radar signal detection area 59
  • • Determination of the distance between robot 1 and the detected person and
  • • Determination of a signal strength for the radar sensor 31 based on the determined distance between robot 1 and the person.
• C2. Computer-implemented method according to C1, further comprising adjusting the signal strength of the radar sensor 31.
• C3. Computer-implemented method according to C1-C2, further comprising carrying out a radar data-based vital parameter measurement of the detected person.
• C4. Computer-implemented method according to C3, further comprising readjusting the signal strength of the radar sensor 31.
• C5. Computer-implemented method according to C2, whereby the signal strength is increased.
• C6. Computer-implemented method according to C4, whereby the signal strength is reduced.
• C7. Computer-implemented method according to C1, wherein the person position is stored in a map (42, 43, 44).
• C8. Computer-implemented method according to C1-C7, further comprising determining the necessary detection depth of the radar sensor 31 by comparing the area to be detected with a map 42.
• C9. Computer-implemented method according to C8, further comprising adjusting the signal strength of the radar sensor 31 on the basis of the necessary detection depth, which enables person detection in the area to be detected.
• C10. Computer-implemented method according to C1-C9, further comprising the transmission of a signal to a terminal 5 based on the evaluation of the determined vital parameters.
• C11. Computer-implemented method according to C1-C10, wherein used position data, which were determined by evaluating radar sensor data, are transmitted by means of a laser scanner 10, Data determined by camera 11, ultrasonic sensor and/or odometry unit 23 are standardized.
• C12. Computer-implemented method according to C1-C11, wherein a wall 60 or closed door 52 is located between robot 1 and detected person.
• C13. The computer-implemented method of C1-C11, wherein the radar signal detection area 59 is oriented substantially perpendicular to the path of the robot 62.
• C14. Device for carrying out the process according to C1-C13.

Example 4: Adaptation of path planning based on an anomaly evaluation

In the context of monitoring people or patients, the travel path of robot 1 is to be optimized in order to efficiently detect anomalies in the monitoring area or in people to be monitored and to optimize the running time of robot 1 with regard to the battery level, and to enable close monitoring of people.

In one aspect, as in 8th and 10 illustrated, the robot 1 moves along a route such as an operating path 63 (step 310) through a building, preferably in a corridor 50 with rooms 53 on at least one side. In doing so, it detects its current position (step 320 - position determination). On its route, the robot 1 uses at least one radar sensor unit 30 to detect the areas that are preferably essentially perpendicular to the direction of travel of the robot 1. The radar signals are emitted through the walls 60, reflected behind the walls 60, for example on objects such as a bed 54, a person (56-58) or another wall 61, and detected by the radar sensor unit 30. The radar signal detection area 59, as shown in 8th outlined shows this in part. This means that objects or obstacles are detected in the radar detection area 59 (step 330). Person detection (step 340) optionally takes place. To evaluate the results of the detection, the robot 1 uses a map 43 in an optional aspect and evaluates measurements of the radar signals based on areas stored in a map, i.e. a position-related evaluation of the detection results takes place in step 430. Position-related means that the detection is associated with certain coordinates that are based on a map. These areas can, for example, be contained in one of the aforementioned maps 42-44 or simply be associated with one of these maps via a table that defines areas via coordinates and classifies them accordingly. In one aspect, these areas can be individual rooms 53 or sub-areas of the rooms 53, such as beds 54, that are located in these rooms 53. This means that the aforementioned rooms 53 can also be partitioned into different zones, which can then be individually evaluated with regard to the objects or obstacles detected therein using radar sensor data. A hierarchical structure is possible here, which leads, for example, to a multi-level map, with one map level covering the entire area that can be detected by the robot 1 using sensors, and further maps covering sub-areas and further subdivided zones. For example, it can then be defined in a position and rule database 48 in the memory 15 that, for example depending on the time of day, people are only allowed to stay in certain areas or zones.

The detected radar signals are evaluated after a detection process as described elsewhere in this document and optionally stored in a memory 15 and there, for example, in a position and rule database 48 (step 440). The evaluated results of the detection process are compared with previously stored detection process results (e.g. from a position and rule database 48) (step 450). If, for example, a different detection result is determined than in the stored detection process results, this is registered as an anomaly. For example, a moving object can be detected in a room (e.g. 53) or in a defined part of a room (i.e. a zone such as a bed 54) during detection. An anomaly would then exist if an earlier detection result for the room or the defined part of the room (i.e. the zone) had not detected any people. Alternatively, an anomaly can be defined by recorded vital parameters of a person analyzed using radar signals, where the measured vital parameter deviates from previously determined vital parameters of the person. An anomaly therefore includes detection results that deviate from expected and/or stored detection results. When detecting anomalies, rules can be stored that are based on a minimum difference between the detection results and the previously stored detection results. If this minimum difference is exceeded, the detection result is classified as an anomaly. In general, an anomaly is understood to mean one or more deviations of a detection result from a detection result defined (e.g. as a threshold value) (or previously determined, for example). The minimum difference can be determined, for example, using machine learning methods, as is known from the state of the art for event detection of traffic in computer networks, on the other hand, via rules defined in a position and rule database 48. In one aspect, more than one detection result can be taken into account, for example the current and the previous two (saved) detection results, which are compared with other previously saved detection results. For example, an anomaly can only be present if one or more detection results are above a threshold value, which results, for example, from a minimum number of calculated detection deviations, possibly per unit of time and/or evaluated zone. This is understood as a deviation determination and evaluation step 460.

Based on a detected anomaly, the path planning of the robot 1 is adjusted (step 470 - optional). The adjustment of the path planning causes the robot 1 to, for example, increase the frequency with which it passes through the area to be detected (e.g., a zone such as a bed 54 with a person 56 in a room 53). In one aspect, the robot 1 simultaneously triggers a signal via a wireless interface 16 to a terminal 5, which, for example, indicates the anomaly (step 160 - event triggering - optional).

• D1. Computer-implementiertes Verfahren zum Triggern eines Verhaltens eines mobilen Roboters 1 in Abhängigkeit von mittels eines Radarsensoreinheit 30 detektierten Anomalien, umfassend
  • • Bewegung des Roboters 1 auf einem Pfad 63,
  • • Erfassung von Objekten bzw. Hindernissen im Radar-Erfassungsbereich 59,
  • • Positions- und/oder zeitbezogene Auswertung der Erfassungsergebnisse,
  • • Vergleich Erfassungsergebnisse mit zuvor gespeicherten Erfassungsergebnissen und
  • • Ermittlung und Bewertung der Abweichung zwischen den Erfassungsergebnissen und den zuvor gespeicherten Erfassungsergebnissen.
• D2. Computer-implementiertes Verfahren nach D1, weiter umfassend die Anpassung der Pfadplanung des Roboters 1.
• D3. Computer-implementiertes Verfahren nach D1-D2, weiter umfassend das Auslösen eines Ereignisses durch den mobilen Roboter 1.
• D4. Computer-implementiertes Verfahren nach D1-D3, wobei die Anomalien mit Positionen auf einer Karte (42, 43 und/oder 44) assoziiert sind.
• D5. Computer-implementiertes Verfahren nach D1-D4, wobei die Erfassung von Objekten im Radar-Erfassungsbereich 59 die Detektion von Personen umfasst.
• D6. Computer-implementiertes Verfahren nach D1 - D5, wobei die positions- und/oder zeitbezogene Auswertung der Erfassungsergebnisse basierend auf definierten Teilbereichen des von der Radarsensoreinheit 30 insgesamt erfassten Bereichs erfolgt und diese Teilbereiche einzeln ausgewertet werden.
• D7. Computer-implementiertes Verfahren nach D1 - D6, wobei die positions- und/oder zeitbezogene Auswertung der Erfassungsergebnisse auf Basis einer Mehrebenenkarte des Roboters 1 erfolgt.
• D8. Computer-implementiertes Verfahren nach D1-D7, weiter umfassend Speichern der Positions- und/oder zeitbezogene Auswertung der Erfassungsergebnisse in einer Positions- und Regeldatenbank 48.
• D9. Computer-implementiertes Verfahren nach D1 - D8, wobei die Erfassungsergebnisse Vitalparameter einer Person umfassen.
• D10. Computer-implementiertes Verfahren nach D1-D9, wobei der Radarsignal-Erfassungsbereich 59 im Wesentlichen senkrecht zum Betriebspfad 62 des Roboters 1 ausgerichtet ist.
• D11. Computer-implementiertes Verfahren nach D1-D10, wobei genutzte Positionsdaten, die durch Auswertung von Radarsensordaten ermittelt wurden, durch mittels Laserscanner 10, Kamera 11, Ultraschallsensor und/oder Odometrieeinheit 23 ermittelte Daten normiert werden.
• D12. Computer-implementiertes Verfahren nach D1-D11, wobei sich zwischen Roboter 1 und detektierter Person eine Wand 60 oder geschlossene Tür 52 befindet.
• D13. Vorrichtung zur Durchführung des Verfahrens nach D1-D12.

With regard to the spatial or temporal-spatial position of objects or obstacles that are detected by the radar sensor unit 30, areas can be defined, for example on a map of the robot 1 or in a table with coordinates, which are each to be monitored individually by the robot 1 by means of the radar sensor unit 30 and which in turn can contain sub-areas or zones that are subject to other monitoring rules, such as different threshold values for the detection of anomalies, or which trigger different consequences in the event of an anomaly detection.

• D1. Computer-implemented method for triggering a behavior of a mobile robot 1 depending on anomalies detected by a radar sensor unit 30, comprising
  • • Movement of robot 1 on a path 63,
  • • Detection of objects or obstacles in the radar detection range 59,
  • • Position and/or time-related evaluation of the recording results,
  • • Compare acquisition results with previously saved acquisition results and
  • • Determination and evaluation of the deviation between the acquisition results and the previously stored acquisition results.
• D2. Computer-implemented method according to D1, further comprising adapting the path planning of the robot 1.
• D3. Computer-implemented method according to D1-D2, further comprising triggering an event by the mobile robot 1.
• D4. Computer-implemented method according to D1-D3, wherein the anomalies are associated with positions on a map (42, 43 and/or 44).
• D5. Computer-implemented method according to D1-D4, wherein the detection of objects in the radar detection range 59 includes the detection of persons.
• D6. Computer-implemented method according to D1 - D5, wherein the position- and/or time-related evaluation of the detection results is carried out based on defined sub-areas of the area detected by the radar sensor unit 30 as a whole and these sub-areas are evaluated individually.
• D7. Computer-implemented method according to D1 - D6, wherein the position- and/or time-related evaluation of the detection results is carried out on the basis of a multi-level map of the robot 1.
• D8. Computer-implemented method according to D1-D7, further comprising storing the position and/or time-related evaluation of the detection results in a position and rule database 48.
• D9. Computer-implemented method according to D1 - D8, wherein the acquisition results comprise vital parameters of a person.
• D10. Computer-implemented method according to D1-D9, wherein the radar signal detection area 59 is oriented substantially perpendicular to the operating path 62 of the robot 1.
• D11. Computer-implemented method according to D1-D10, wherein used position data determined by evaluating radar sensor data are standardized by data determined by means of laser scanner 10, camera 11, ultrasonic sensor and/or odometry unit 23.
• D12. Computer-implemented method according to D1-D11, wherein a wall 60 or closed door 52 is located between robot 1 and detected person.
• D13. Device for carrying out the method according to D1-D12.

Example 5: Adaptive adjustment of radar signal strength for optimized vital parameter measurement

In one aspect, it is important to simplify the measurement of vital parameters without necessarily having to resort to reference values for the signal strength in a table, because this would in turn involve further steps to determine the reference value outside of the actual measurement procedure, e.g. as part of a calibration. The focus is therefore on a procedure that does not require such reference values.

This method is based on steps 310-340, as already described elsewhere in this document. This step is then followed by the vital parameter measurement 380 (explained in more detail elsewhere) and a comparison of the vital parameter measurement results (step 375). The latter means a comparison of the determined recording values with measured values stored in memory 15 with a view to qualitative quality criteria of the recording. The aim of this comparison is, for example, to reduce the noise in the recorded data. To do this, the recorded data is compared with data stored in memory 15 and deviations are determined. If these are, for example, above a threshold value, the signal strength is adjusted (step 370) via the amplifier 33. Here, for example, the signal strength is first increased, then another vital parameter measurement takes place (step 380) and a comparison of the vital parameter measurement results (step 375). If this last step results in an improvement in the measurement results, the signal strength is further adjusted (step 370) in the same direction until a further improvement in the measurement results in the loop (or control circuit) from steps 380 - 375 - 370 no longer leads to a further improvement in the measurement. For example, it is also possible to lower the signal strength in the first step and then gradually reduce it until at some point an increase in the signal strength produces a better measurement result. In step 385, a vital parameter evaluation takes place. This can optionally include, for example, using the best measurement value from the previous measurements or the second best. In the case of the best measurement value, the measurement values must be stored over time, for example in memory 15. The vital parameter evaluation includes a comparison (e.g. threshold-based) with data from the position and control database 48, for example to identify critical vital parameters. The vital parameter evaluation can also include a position- and/or time-related evaluation of the recording results. In one aspect, this means that the evaluation of the at least one recorded vital parameter is assessed over time, for example by comparing it with previously stored vital parameters, an evaluation is carried out on the basis of the position assigned to the vital data parameter, or a combination thereof. In the next step, an event can then optionally be triggered (step 160), as defined elsewhere, for example when a critical vital parameter has been determined.

The signal evaluation itself described above, e.g. for vital parameter recording in the vital data recording module 46, can be carried out as described in the prior art (DOI: 10.1155/2018/4832605): Based on a bandpass filtering of the signal (step 381), only signals in the frequency range that correspond to the frequency of the signal to be evaluated are allowed through, e.g. at the heart rate of a frequency of 1-3 Hz. In step 382, for example, a dimension reduction of the recorded time series data takes place in order to capture the underlying patterns, in one aspect with a principal component analysis (PCA). For example, the dominant frequency can then be determined using fast Fourier transformation (step 383) and outliers can be smoothed out using filtering such as low-pass filtering (step 384). The time course of these measurement results indicates, for example, the rate of the measured vital parameter (and inversely its frequency). However, adaptive moving target identification filters can also be used.

• E1. Computer-implementiertes Verfahren zur Erfassung von mindestens einem Vitalparameter durch einen mobilen Roboter 1, umfassend
  • • Bewegung des Roboters 1 auf einem Pfad 62,
  • • Ermittlung der aktuellen Position des Roboters 1,
  • • Erfassung von Objekten bzw. Hindernissen im Radar-Erfassungsbereich,
  • • Detektion von Personen im Radar-Erfassungsbereich,
  • • Erfassung von Vitalparametern einer detektierten Person im Radar-Erfassungsbereich.
  • • Abgleich der Vitalparameter-Messergebnisse und Anpassung der Signalstärke in Abhängigkeit vom Abgleich der Vitalparameter-Messergebnisse und
  • • Durchführung einer Vitalparameter-Bewertung.
• E2. Computer-implementiertes Verfahren nach E1, wobei die Anpassung der Signalstärke so lange vorgenommen wird, bis sich die gemessenen Vitalparameter nach einer Anpassung der Signalstärke qualitativ verschlechtern.
• E3. Computer-implementiertes Verfahren nach E1-E2, weiter umfassend eine Vitalparameter-Bewertung.
• E4. Computer-implementiertes Verfahren nach E1-E3, weiter umfassend das Auslösen eines Ereignisses eine Signalübermittlung in die Cloud 4 oder an ein Terminal 5 umfasst.
• E5. Computer-implementiertes Verfahren nach E1-E4, wobei das Auslösen eines Ereignisses eine Benachrichtigung eines Terminals umfasst.
• E6. Computer-implementiertes Verfahren nach E3, wobei die Vitalparameter-Bewertung eine Positions- und/oder zeitbezogene Auswertung der Erfassungsergebnisse umfasst.
• E7. Computer-implementiertes Verfahren nach E1-D6, weiter umfassend eine Lokalisierung der Personenposition.
• E8. Computer-implementiertes Verfahren nach E7, weiter umfassend ein Vergleich der Objektposition mit im Speicher 15 hinterlegten Objektpositionen und/oder Objektzonen.
• E9. Computer-implementiertes Verfahren nach E1-E8, wobei die Vitalparameter für eine Person an einer lokalisierten Position erfasst werden.
• E10. Computer-implementiertes Verfahren nach E1-E9, wobei die Signalauswertung erfasster Radarsignale einen Vergleich mit Mustern darstellt.
• E11. Computer-implementiertes Verfahren nach E7, wobei genutzte Positionsdaten, die durch Auswertung von Radarsensordaten ermittelt wurden, durch mittels Laserscanner 10, Kamera 11, Ultraschallsensor und/oder Odometrieeinheit 23 ermittelten Daten normiert werden.
• E12. Vorrichtung zur Durchführung des Verfahrens nach E1-E11.

In summary, the process can be described as follows:

• E1. Computer-implemented method for detecting at least one vital parameter by a mobile robot 1, comprising
  • • Movement of robot 1 on a path 62,
  • • Determination of the current position of robot 1,
  • • Detection of objects or obstacles in the radar detection area,
  • • Detection of persons in the radar detection area,
  • • Recording of vital parameters of a detected person in the radar detection area.
  • • Comparison of the vital parameter measurement results and adjustment of the signal strength depending on the comparison of the vital parameter measurement results and
  • • Conduct a vital sign assessment.
• E2. Computer-implemented method according to E1, wherein the adjustment of the signal strength is carried out until the measured vital parameters deteriorate qualitatively after an adjustment of the signal strength.
• E3. Computer-implemented procedure according to E1-E2, further comprising a vital parameter assessment.
• E4. Computer-implemented method according to E1-E3, further comprising the triggering of an event comprises a signal transmission to the cloud 4 or to a terminal 5.
• E5. The computer-implemented method of E1-E4, wherein triggering an event comprises notifying a terminal.
• E6. Computer-implemented method according to E3, wherein the vital parameter evaluation comprises a position- and/or time-related evaluation of the acquisition results.
• E7. Computer-implemented method according to E1-D6, further comprising localizing the person's position.
• E8. Computer-implemented method according to E7, further comprising a comparison of the object position with object positions and/or object zones stored in the memory 15.
• E9. Computer-implemented method according to E1-E8, wherein the vital parameters for a person are recorded at a localized position.
• E10. Computer-implemented method according to E1-E9, wherein the signal evaluation of detected radar signals represents a comparison with patterns.
• E11. Computer-implemented method according to E7, wherein used position data determined by evaluating radar sensor data are standardized by data determined by means of laser scanner 10, camera 11, ultrasonic sensor and/or odometry unit 23.
• E12. Device for carrying out the method according to E1-E11.

Example 6: Robot movement adjustment to avoid high signal attenuation

In one aspect, the walls 61 through which the detection of objects and/or vital parameters takes place have a high signal attenuation. To mitigate these effects, the robot 1 performs certain movements and assumes positions in order to achieve a lower signal attenuation and thus an increased detection quality.

Moving the robot 1 along a path (step 310) can, in one aspect, proceed as follows (see 13 ): While the robot follows an operating path 63 (step 110 or 310) with knowledge of its respective position, the robot 1 assumes at least one scanning position (step 510). In one aspect, a scanning position is understood to be a fixed position of the robot 1 (defined by an area), which is stored, for example, on a map (42, 44) of the robot 1 and is located at a defined distance from a door 52, which is stored, for example, in a memory 15. At the scanning position, the robot uses the radar sensor unit 30 to detect objects and/or vital parameters. In one aspect, the defined distance to the door 52 is less than 1 m, for example less than 80 cm, and in another preferred aspect less than 30 cm. If the scanning position is not stored on the map (42, 44), it can be detected by the robot 1 using stored rules when traveling along the operating path 63 by comparing stored patterns with patterns on (stationary) obstacles detected by a laser scanner 10, a camera 11 or a radar sensor.

In step 520, the robot 1 performs a rotational movement at the scanning position, particularly if it is a fixed point and not a series of points, which takes place, for example, around a vertical axis. The rotation either initially takes place in one direction. The robot rotates through an angle of at least 20°, in one aspect at least 45°. In an alternative aspect, the robot 1 rotates around an axis (which does not necessarily have to be an axis that runs through the robot 1). If the robot 1 has not performed a (at least almost complete) rotation around an axis, the robot 1 rotates in the opposite direction (optional step 530).

In one aspect, assuming a scanning position can comprise a rotational movement at a point for a robot 1 that is capable of rotationally symmetrical rotation. If, for example, the wheel arrangement of the robot 1 does not permit such a rotational movement, it can be a rotational movement about an axis that lies outside the robot 1. In a further aspect, the sequence of steps 510-530 comprises a scenario in which the robot 1, when approaching a door 52 in the corridor, moves parallel to walls 61, a rotational movement towards the door 52, followed by traveling along a convex operating path 63 in front of the door 52 (i.e. traveling along an arc with a curvature away from the door) again perpendicularly away from the door, with the defined distance of the robot 1 from the door in an area that is, for example, approximately in the middle of the door.

• F1. Computer-implementiertes Verfahren zur Steuerung eines Roboters 1, umfassend
  • • Bewegung des Roboters 1 auf einem Pfad,
  • • Ermittlung der aktuellen Position des Roboters 1,
  • • Einnehmen einer Abtastposition,
  • • Durchführung einer Drehbewegung,
  wobei an der Abtastposition eine Erfassung von Objekten und/oder Vitalparametern durch eine Radarsensoreinheit 30 des Roboters 1 erfolgt.
• F2. Computer-implementiertes Verfahren nach F1, wobei die Ermittlung der Abtastposition aus einer Karte (42, 44) oder über eine Erfassung und Auswertung von Mustern erfolgt.
• F3. Computer-implementiertes Verfahren nach F1 - 2, wobei die Abtastposition einen definierten Abstand zu einer Tür 52 aufweist.
• F4. Computer-implementiertes Verfahren nach F1 - 3, wobei der Winkel der Drehbewegung mindestens 20° umfasst.
• F5. Computer-implementiertes Verfahren nach F1 - 4, wobei die Drehbewegung eine Drehbewegung um eine rotationssymmetrische Achse des Roboters 1 umfasst.
• F6. Computer-implementiertes Verfahren nach F1 - 5, wobei die Drehbewegung eine Rotation in eine Richtung, gefolgt von einer Rotation in eine andere Richtung umfasst.
• F7. Computer-implementiertes Verfahren nach F1 - 6, wobei sich der Roboter 1 beim Annähern an eine Tür 52 sich parallel zu Wänden 61 bewegt, eine Drehbewegung hin zu einer Tür 52 vornimmt, gefolgt vom Abfahren eines konvexen Pfads 63 vor der Tür 52.
• F8. Computer-implementiertes Verfahren nach F7, wobei der minimale Abstand des konvexen Pfads zur Tür dem definierten Abstand zur Tür 52 entspricht.
• F9. Vorrichtung zur Durchführung des Verfahrens nach F1-F8.

In summary, the process described here, which can also be combined with the processes from the previous examples, can be represented as follows:

• F1. Computer-implemented method for controlling a robot 1, comprising
  • • Movement of robot 1 on a path,
  • • Determination of the current position of robot 1,
  • • Taking a scanning position,
  • • Performing a rotary movement,
  wherein objects and/or vital parameters are detected at the scanning position by a radar sensor unit 30 of the robot 1.
• F2. Computer-implemented method according to F1, wherein the determination of the scanning position is carried out from a map (42, 44) or by detecting and evaluating patterns.
• F3. Computer-implemented method according to F1 - 2, wherein the scanning position has a defined distance to a door 52.
• F4. Computer-implemented method according to F1 - 3, wherein the angle of the rotational movement comprises at least 20°.
• F5. Computer-implemented method according to F1 - 4, wherein the rotational movement comprises a rotational movement about a rotationally symmetric axis of the robot 1.
• F6. Computer-implemented method according to F1 - 5, wherein the rotational movement comprises a rotation in one direction followed by a rotation in another direction.
• F7. Computer-implemented method according to F1 - 6, wherein the robot 1 moves parallel to walls 61 when approaching a door 52, makes a rotational movement towards a door 52, followed by traveling along a convex path 63 in front of the door 52.
• F8. Computer-implemented method according to F7, wherein the minimum distance of the convex path to the door corresponds to the defined distance to the door 52.
• F9. Device for carrying out the method according to F1-F8.

Reference designations

1

robot

2

mobile base

3

Construction

4

Cloud

5

terminal

10

Laser scanner

11

camera

12

accumulator

13

Charging device

14

Computer unit

15

Storage

16

Wireless interface

17

Safety control

18

Protective contact strip

20

drive wheel

21

Drive motor

22

cover

23

Odometry unit

24

Support wheel

30

Radar sensor unit

31

Radar sensor

32

Radar antenna

33

amplifier

40

Path planning

41

Map module

42

first card

43

second card

44

third card

45

Person position determination module

46

Vital data recording module

47

Object position detection module

48

Position and rule database

49

Object identification database

50

corridor

51

open door

52

closed door

53

Space

54

Bed

55

Exploration path

56

Person in bed

57, 58

Person out of bed

59

Radar signal detection range

60

Wall

61

rear room wall

62

Laser scanner detection range

63

Operational path

QUOTES INCLUDED IN THE DESCRIPTION

This list of documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions.

Cited patent literature

• JP 2019016339 A [0002]
• CN113715032A [0004]
• WO 2005074422 A2 [0004]
• DE 102020006337 A1 [0023]
• EP 2100153 A1 [0027, 0047]
• WO 2021038109 A1 [0030]
• WO 2021254975 A1 [0034]

Cited non-patent literature

• Wang, Borui, et al. "Imitation learning for human pose prediction." Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019, which, however, use skeletal data to classify body poses, whereas radar signals are used here). The classification method can be action recognition (see e.g. Sun et al. 2022, DOI: 10.1109/TPAMI.2022.3183112 [0029]
• Wang, Mingyang, Yimin D. Zhang, and Guolong Cui. “Human motion recognition exploiting radar with stacked recurrent neural network.” Digital Signal Processing 87 (2019): 125-131) [0029]
• Maleševic et al. (2020, DOI:10.3390/s20082351) [0030]
• Li, M., et al. 2018, High Resolution Radar-based Occupancy Grid Mapping and Free Space Detection., In Proceedings of the 4th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2018), pages 70-81, ISBN: 978-989-758-293- 6) [0035]

Claims (18)

Computer-implemented method for monitoring an area by a mobile robot 1, comprising • traveling along a path, • detecting an area to be monitored by means of at least one radar sensor unit (30), • evaluating detected radar signals and • triggering an event, wherein the detection of the area to be monitored takes place through a wall (60) and/or closed door (52). Computer-implemented method according to Claim 1 , whereby the detection of the area to be monitored takes place approximately perpendicular to the direction of travel of the robot 1. Computer-implemented method according to Claim 1 - 2 , whereby the signal evaluation includes the identification of objects including persons. Computer-implemented method according to Claim 3 , whereby the signal evaluation further includes a differentiation between moving and immobile objects. Computer-implemented method according to Claim 1 - 4 , wherein the signal evaluation comprises a determination of at least one vital parameter. Computer-implemented method according to Claim 3 , further comprising a localization of the object position. Computer-implemented method according to Claim 3 - 6 , further comprising a comparison of the object position with object positions and/or object zones stored in the memory (15). Computer-implemented method according to Claim 7 , whereby the object zones stored in the memory (15) can be prohibited zones. Computer-implemented method according to Claim 5 , further comprising the comparison of determined vital parameters with stored vital parameters and/or a vital parameter evaluation. Computer-implemented method according to Claim 1 - 9 , wherein the triggering of an event comprises a signal transmission to the cloud (4) or to a terminal (5). Computer-implemented method according to Claim 5 , where vital parameters for a person are recorded at a localized position. Computer-implemented method according to Claim 1 , whereby the signal evaluation of detected radar signals represents a comparison with patterns. Computer-implemented method according to Claim 1 - 12 , whereby a classifier created using imitation learning or action recognition is used for signal evaluation. Computer-implemented method according to Claim 6 , wherein used position data, which were determined by evaluating radar sensor data, are standardized by data determined by means of laser scanner 10, camera 11, ultrasonic sensor and/or odometry unit 23. Computer-implemented method according to Claim 3 or 6 , further comprising updating the map (42) of the robot (1) by incorporating the obstacles detected by means of the radar sensor unit (30). Computer-implemented method according to Claim 1 , wherein the robot 1 takes a scanning position in front of a door 52 at a defined distance and performs a rotational movement. Computer-implemented method according to Claim 1 , wherein the robot 1 moves along a convex path in front of a door 52 while maintaining a defined minimum distance from the door 52. Device for carrying out the method according to Claim 1 - 17 .

Images