DE102019202261A1 - Method for detecting an object within a monitored area - Google Patents

Abstract

The invention relates to a method for detecting an object (155) within an area monitored by means of a sensor system, in which the monitored area is subdivided into edge zones (131) and inner zones (132), and in which for the inner zones (132), respectively an active (Z _A ) and an inactive (Z _I ) state can be distinguished and in which the active state is used permanently for the edge zones (131), whereby, when an object (155) is in an active state by means of the sensor system (Z _A ) is detected, zones adjacent to this zone are set to the active (Z _A ) state, provided they are not already in the active state, and an object detected by the sensor system in an inner zone (132) only is then recognized as a real object when this zone is in the active state (Z _A ).

Description

The present invention relates to a method for recognizing an object within an area monitored by means of a sensor system, a computing unit and a computer program for its implementation, and an object recognition system.

State of the art

When monitoring an area for the presence of objects or people using sensors, in particular opto-electronic sensors (laser scanners, LiDAR sensors, cameras) and RADAR sensors, incorrect detections can occur, for example, the availability of the machine to be protected or the machine to be protected Can significantly reduce the vehicle.

Such false detections can have different reasons, for example environmental influences such as rain, snow, dust, hail and the like, a (false) detection of the ground or the roadway as an object (especially with RADAR sensors) or also systematic effects due to the physical measuring principle can occur (e.g. phantom objects or ghost objects with RADAR sensors).

So-called protection or safety areas are often defined, in which a recognized object or a recognized person causes the machine or vehicle to react. This can, for example, be a braking maneuver of a vehicle or also a shutdown of a press in order to protect people who are in the danger area of the machine.

However, false detections (so-called false positives) lead to reactions being initiated that are actually not necessary, since there is no real or no relevant object in this area.

In the DE 10 2009 034 848 B4 For example, a method is proposed for opto-electronic sensors that shifts an inner surveillance area to an outer surveillance area and tracks detected objects within the outer area (so-called object tracking). An object recognized in the inner monitoring area is only rated as a real object if an object was previously recognized in the outer monitoring area and the position of the object in the inner area can be plausibly predicted by the object tracking of the object in the outer area.

In the DE 10 2017 005 671 A1 For example, a method for determining incorrect detections of a sensor system of a vehicle is described, which is based on the fact that incorrect detection of a sensor can be detected by monitoring the area between the vehicle on which the sensor system is located and a vehicle driving in front of it. The vehicle in front is followed by a driver assistance system (tracking). If an object is now detected in the intermediate area between the two vehicles, it can be assumed with a high degree of probability that this is a false detection, since this intermediate area was driven over by the vehicle in front shortly before and should accordingly have been free.

Disclosure of the invention

According to the invention, a method for recognizing an object (including, in particular, a person), a computing unit and a computer program for carrying it out, and an object recognition system with the features of the independent patent claims are proposed. Advantageous configurations are the subject of the subclaims and the description below.

A method according to the invention is used to identify an object within an area monitored by means of a sensor system. A sensor system is expediently used that includes at least one sensor selected from: LiDAR sensors, laser scanners, time-of-flight (ToF) sensors, RADAR sensors, cameras, in particular stereo cameras, and ultrasound sensors. In general, (electro) optical sensors or sensors based on sound or radio can be considered.

The monitored area is divided into peripheral zones and inner zones. The total number of zones can be selected in particular depending on the computing power of an executing computing unit and / or the desired execution speed (for example in order to achieve real-time capability as far as possible). In particular, it should be noted that the area as a whole is monitored by means of the sensor system, i.e. The area does not have to be divided into different individual sensors (although this is of course possible), rather it is useful if the area is divided into zones in the processing unit when the information recorded or data supplied by the sensor system is evaluated.

Of the total available or used zones, those are then considered to be edge zones in particular which - in relation to an area covered by the monitored area - are located on the outside or on the outer edge. In the case of an (approximately) circular monitored area, the edge zones then form, for example, a ring on the outer edge of the area. The radial extension of this ring then corresponds approximately to the diameter or the width or length of a zone, but two or more zones in the radial direction can also be selected as the edge zone, if appropriate.

It goes without saying that a full ring does not have to be formed by the edge zones; rather, only a certain proportion of the zones located on the outer edge can be used as the edge zone. It is particularly useful here - as will also be shown on the basis of the following explanations - to cover those parts of the monitored area with edge zones into which moving objects can move from outside. In the case of monitoring at the front of a vehicle, on the one hand, the monitored area typically only covers approximately 180 °, and on the other hand, an edge zone is only necessary in this area at approximately 180 °.

For the sake of completeness, it should be mentioned again at this point that the above explanation of the monitored area and the edge zone is only an example.

Furthermore, an active and an inactive state are distinguished for the inner zones and the active state is used permanently for the edge zones. In this context, an active state is to be understood in particular to mean that the relevant zone is taken into account - within the framework of a detection of an object in the monitored area. A zone in the inactive state, however, is not taken into account.

If an object is now detected in a zone in the active state by means of the sensor system, zones adjacent to this zone (in which the object was detected) are set in the active state, provided they are not already in the active state. An object detected in an inner zone by means of the sensor system is only recognized as a real object if this zone is in the active state. While an object is determined by the sensor system, be it an actual object or just an object incorrectly determined due to incorrect measurements or the like, this object determined in this way is only seen as a real object in a zone that is in the active state Considered.

Since the edge zones are always in the active state, an object moving into the monitored area can be detected. In contrast, an object that suddenly appears in an inner inactive zone can be excluded as a real object.

The effect of the invention is that only an object moving into the monitored area from the outside is also recognized as a real object inside the area, but without complex object tracking or tracking being carried out for this purpose. This is done simply by activating zones. Whether an object detected in an active zone is the same object that was previously also detected in an adjacent active zone is irrelevant and is also not determined. A false detection of an object in the monitored area - at least in a relevant part thereof, i. the inner zones - can be avoided particularly reliably.

Since within the scope of the invention the entire area does not have to be monitored or checked for objects, but only those zones that are in the active state or have been set in this, the computing power required can be significantly reduced.

At this point it should be added that in practice an object in a certain zone is typically viewed as detected by means of the sensor system, in that the zones stored in the processing unit are evaluated in order to decide in which zone the detected object is located. In the border areas between zones, the object can be assigned to one or two, possibly even more, zones.

At least some of the inner zones are preferably defined as a security area, the detection of a real object in the security area triggering at least one security measure. Under such a measure, for example, a vehicle can be braked - for which the area is being monitored - or a machine - located in the area being monitored - can be switched off.

Advantageously, at least one further part of the inner zones lying outside the safety area is defined as the warning area, with at least one further predetermined measure being initiated if a real object is detected in the warning area. This further measure can be, for example, a weakened version of the measure mentioned above. It is therefore conceivable, for example, to initially brake the vehicle more slowly or to move or brake the machine more slowly.

It is particularly preferred if a size and / or shape and / or a location of the Security area is adapted as a function of at least one specifiable criterion, in particular constantly during the execution of the method. This can also apply to a possibly existing warning area. This can also increase the efficiency of the monitoring, since, for example, only ever a small, really relevant security area has to be monitored with increased attention. If the monitoring is used in a vehicle, the location of the safety area can be adjusted, ie pivoted, for example when cornering. In the case of a machine in the monitored area, the safety area can be selected larger or smaller, for example, depending on an activity currently being performed by the machine.

It is particularly expedient if an inner zone which is in the active state is set in the inactive state in accordance with at least one predetermined further criterion. The at least one further predetermined criterion includes in particular that no object was detected in this zone for a predetermined period of time and / or that within this or another predetermined period of time this zone does not go into the active state due to a detection of an object in an adjacent zone was set. In other words, an inner zone can always be excluded from the check with regard to the presence of an object if no object has been detected therein for a certain time. In this context, however, it is also particularly useful that all inner zones are set to the inactive state at the start of monitoring. Thus, inner zones are only considered for testing when an object is approaching from the outside. Alternatively, it is conceivable that all inner zones are set to the active state at the beginning and then gradually (by the mentioned criterion) are put into the inactive state.

At least one predetermined part of the inner zones is preferably set to the active state whenever at least one variable detected by means of the sensor system fulfills a predetermined measurement criterion, for example exceeds or falls below a predetermined threshold value. Combinations of different sizes and measurement criteria or sizes derived therefrom are also conceivable. This is based on the idea that certain objects, particularly those moving very quickly, could move from the outside into the monitored area, but are not (or not in time) detected as an object in the edge zones. This would result in the neighboring zones in the monitored area not being set to the active state towards the inside. If, however, certain (sufficiently large) changes in measured quantities are caused, for example a significantly increased reflected radiation in the case of a RADAR sensor, it can be assumed that it is a real object in any case, even if it - for whatever reasons always - has not been recognized as such in the peripheral zones. In addition, a vehicle speed can also be taken into account - in terms of several variables or measurement criteria. This ensures that objects are also recognized that are definitely real objects based on the measurement alone.

The proposed method is preferably carried out for a stationary or a mobile unit, in particular a vehicle. A machine or a robot in a factory hall, for example, can be considered as a stationary unit. In both cases, particularly reliable object recognition is desirable, in particular in the sense of increased protection of people from danger. In contrast, incorrect object recognition, as is reliably and efficiently avoided with the present method, would lead to unnecessary restrictions.

The method is particularly useful for autonomously driving mobile work machines in the outdoor area, for which there are so far only a few non-contact protective devices certified for recognizing people. Systems available to date have great availability problems in the event of external interference, such as dust, rain, snow and the like, so that the vehicles cannot drive or only to a limited extent under these conditions. The proposed method would significantly increase the availability of these vehicles. The method can also be used in conjunction with a corresponding sensor system in dusty environments, for example in road construction. The route of an autonomously driving tar machine could be secured with a corresponding sensor system and the method used to achieve robustness against blown dust and steam.

In general, static monitoring of security areas in the industrial environment (machine protection) and also in public facilities (access protection), especially in dusty environments and / or outdoors, comes into consideration, where conventional systems lead to too large a number of false detections, especially one Monitoring of routes of manual, partially automated and fully automated vehicles or machines such as autonomous industrial trucks, mobile work machines, cars, trucks and the like.

A computing unit according to the invention, for example a control device or another processing unit, preferably a motor vehicle, is set up, in particular in terms of programming, to carry out a method according to the invention when used with a sensor system.

The invention also relates to an object recognition system with a sensor system and a computing unit according to the invention.

The implementation of a method according to the invention in the form of a computer program or computer program product with program code for performing all method steps is advantageous, since this causes particularly low costs, especially if an executing control device is used for other tasks and is therefore available anyway. Suitable data carriers for providing the computer program are in particular magnetic, optical and electrical memories, such as e.g. Hard disks, flash memories, EEPROMs, DVDs, etc. It is also possible to download a program via computer networks (Internet, intranet, etc.).

Further advantages and configurations of the invention emerge from the description and the accompanying drawing.

The invention is shown schematically in the drawing using an exemplary embodiment and is described below with reference to the drawing.

Figure list

• 1 shows schematically an object recognition system according to the invention in a preferred embodiment in a vehicle.
• 2 shows schematically a sequence of a method according to the invention in a preferred embodiment.
• 3 shows a sequence of a method according to the invention in a further preferred embodiment using an example.

Embodiment (s) of the invention

In 1 is schematically an object recognition system according to the invention 100 in a preferred embodiment in a vehicle 150 shown. The object recognition system 100 has a sensor system 110 on, which in turn can have one or more suitable sensors, as already mentioned above, and a computing or processing unit 120 that come with the sensor system 110 is in communication.

By means of the sensor system 110 can be an area shown schematically and by way of example 130 in front of the vehicle 150 be monitored. A processing chain of the object recognition system 100 can be constructed in such a way that sensor data (relating to object information) from the sensor system 110 to the computing unit 120 be transmitted. From the computing unit 120 in turn, after processing, corresponding data (via a recognized object) can be transmitted, for example, to a desired control device in order to carry out a certain measure, for example.

It is also conceivable, however, that input data or input information are sent to the processing unit externally 120 transmitted and possibly to the sensor system 110 to get redirected. This is conceivable, for example, when individual sensors of the sensor system 110 can and should be specifically controlled and / or activated.

In 2 a sequence of an embodiment of a method according to the invention is shown schematically in a preferred embodiment, as it is, for example, with the object recognition system according to FIG 1 can be carried out.

In one step 210 the data generated by the sensor system are received by the computing unit. In one step 220 it is then checked whether a possibly detected object is in an edge zone. If this is not the case (see n-decision), then in one step 230 checks whether the object - which is consequently located in an inner zone - is in an active inner zone or an inner zone which is in the active state. If this is also not the case (see n-decision), then according to step 260 assumed an error detection. Any measures or reactions will not be initiated. However, information about this error identification can be stored for later analysis of the system, for example.

According to an alternative embodiment it can be provided that in step 260 if certain measurement criteria are present for certain variables, for example a very high signal strength, zones are nevertheless activated and a real object is recognized, and an error recognition is only established if these measurement criteria are not present.

Will in step 220 however, it is determined that the detected object is located in an edge zone (cf. y-decision, an edge zone is always in the active state), then in step 240 a transfer of zones which are adjacent to the zone in question into the active state if they were not previously in the active and thus in the inactive state. In addition, in this step, those zones can be switched to the inactive state that are not such zones adjacent to a zone in which an object was detected. It should be noted that, due to a further detection in a further zone, such a zone may have to be switched to the active state. This always takes precedence over the inactive state. It should also be noted that this only affects inner zones, since edge zones are always in the active state or remain there.

To step 240 can also be reached when in step 230 it is determined that an object is detected in an inner zone which is in the active state (cf. y-decision).

After step 240 follows step 250 , in which it is checked whether the relevant (active) zone in which the object was detected lies in a safety area. If this is not the case (see n-decision), you can go back to step 210 returned or started. If, on the other hand, this is the case (cf. y decision), a violation of the security area, ie detection of a real object in the security area, is recognized and a corresponding measure is taken 270 initiated.

In 3 is a sequence of a method according to the invention in a further preferred embodiment using an example with a monitored area 130 in front of a vehicle 150 shown. The area is exemplary 130 , as in 3a to see, represented in a rectangular shape and divided into 14x9 zones (each square).

The ones at the top, left, and right of the range 130 lying zones form edge zones 131 . It should be noted that in this example the zones located at the lower edge (in practice on the vehicle) are not used as edge zones, since no real object can be found in the area during operation, ie while driving 130 will move. However, it can also be expedient to consider the zones located at the lower edge (in practice on the vehicle) as edge zones, for example when starting again after the vehicle has come to a standstill or when an object is overtaking. This can also be varied dynamically if necessary.

The ones within the fringes 131 the other zones (12x8 zones in total) thus form inner zones 132 within the meaning of the invention. A portion of these inner zones 132 again, for example the 8x6 inner zones located centrally on the inside and on the lower edge 132 , also forms a security area 133 .

The one between the security area 133 and inner zones located at the edge zones or a part thereof can, if necessary, also be assigned to an additional warning area or form one, as already explained above.

At the beginning of the process or the monitoring (or in the initial state), there are now the edge zones 131 - as well as otherwise permanently - in the active state, all inner zones 132 however, are in the inactive state.

In 3b are now three objects by way of example 155 or. 160 each shown in one of the zones that are detected by the sensor system. The objects 160 are detected in inner zones that are in the inactive state and are therefore not considered real objects. It is also irrelevant whether this object is 160 in the security zone 133 or another inactive inner zone. An inactive state of a zone or an inactive zone is denoted by Z _I.

The object 155 however, in a - (always) active - edge zone 131 detected. An active zone in which an object is detected can also be designated or set as occupied. As a result, all zones, those of the (occupied) zone (here an edge zone) in which the object 155 was detected, are adjacent, put into the active state, provided it is not already in the active state. Such active zones can also be designated or set as free.

As mentioned, edge zones are always in the active state. However, inner zones could also have already been switched to the active state due to another object. An active state of a zone or an active zone is denoted by Z _A.

In the example shown, all zones in the area come as adjacent zones 130 into consideration that with an edge or a corner to the zone in which the object 155 was detected. It goes without saying that the definition of what is to be understood by an adjacent zone depends on the division of the area 130 can be chosen in zones and / or the shape of the zones. However, it should be ensured that the area surrounding a zone in which an object is detected is covered to a certain extent by such neighboring zones.

In 3c a situation is shown in which the object 155 further onto the vehicle 150 has moved and therefore in an inner zone 132 is detected. Since the inner zone in question is in the active state (it has been put into this active state as explained above), the object 155 (still) recognized as a real object and in this case too, the neighboring zones are switched to the active state, provided they are not yet. The objects 160 however, they are still not recognized as real objects.

In 3d a situation is shown in which the object 155 even further on the vehicle 150 has moved. The security area now also belongs here 133 corresponding inner zones to the neighboring zones, which are brought into the active state. The objects 160 however, they are still not recognized as real objects.

In 3e a situation is now shown in which the object 155 even further on the vehicle 150 has moved. The zone in which the object 155 was detected, belongs to the security area 133 on and is in the active state. A real object located in the security area is thus recognized and a suitable measure can be initiated. The objects 160 however, they are still not recognized as real objects.

In addition, in 3e to see that inner zones that were previously set to the active state, but are no longer considered to be adjacent zones, are set to the inactive state. A lapse of a predetermined period of time can also be used as a criterion.

In this way it can therefore be ensured that only those detected objects are viewed or recognized as real objects in the effect of the invention, which are located from the outside via the edge zones into the monitored area 130 move without actually having to be followed or tracked. This also means that only a certain number of zones need to be checked, which saves additional computing power.

As already mentioned, it can also be provided that under certain conditions (if a variable detected by means of the sensor system exceeds a predetermined threshold value) all inner zones (or at least a certain part thereof) are set to the active state. In this way it can be avoided, for example, that particularly fast moving objects are overlooked.

QUOTES INCLUDED IN THE DESCRIPTION

This list of the 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 assumes no liability for any errors or omissions.

Patent literature cited

• DE 102009034848 B4 [0006]
• DE 102017005671 A1 [0007]

Claims (14)

Method for recognizing an object (155) within an area (130) monitored by means of a sensor system (110), in which the monitored area is divided into edge zones (131) and inner zones (132), and in which for the inner zones (132 ) a distinction is made between an active (Z _A ) and an inactive (Z _I ) state and in which the active state is used permanently for the edge zones (131), wherein, if an object (155) in a zone in the active state (Z _A ) is detected, zones adjacent to this zone are set in the active (Z _A ) state, provided they are not already in the active state, and a by means of the sensor system (110) in a inner zone (132) detected object is only recognized as a real object if this zone is in the active state (Z _A ). Procedure according to Claim 1 , wherein at least a part of the inner zones (132) is defined as a safety area (133), and wherein if a real object is detected in the safety area (133), at least one predetermined measure (270) is initiated. Procedure according to Claim 2 , wherein at least one further part of the inner zones (132) lying outside the safety area (133) is defined as a warning area, and when a real object is detected in the warning area, at least one further predetermined measure is initiated. Procedure according to Claim 3 , wherein a size and / or a shape and / or a location of the warning area is adapted as a function of at least one specifiable criterion. Method according to one of the Claims 2 to 4th , wherein a size and / or a shape and / or a location of the security area (133) is adapted as a function of at least one specifiable criterion. Method according to one of the preceding claims, wherein an inner zone (132) in the active state (Z _A ) is set in the inactive state (Z _I ) in accordance with at least one predetermined further criterion. Procedure according to Claim 6 , wherein the at least one predefined criterion comprises that no object was detected in this zone for a predefined period of time and / or that within this or a further predefined period of time this zone is not set into the active state due to a detection of an object in an adjacent zone has been. Method according to one of the preceding claims, wherein at least a predetermined part of the inner zones (132) is furthermore set to the active state (Z _A ) when at least one variable detected by means of the sensor system (110) fulfills a predetermined measurement criterion. Method according to one of the preceding claims, wherein a sensor system (110) is used which comprises at least one sensor selected from: LiDAR sensors, laser scanners, time-of-flight sensors, RADAR sensors, cameras and ultrasonic sensors. Method according to one of the preceding claims, which is carried out for a stationary or a mobile unit, in particular a vehicle (150). Computing unit (120) which is set up to carry out all method steps of a method according to one of the preceding claims when used with a sensor system (110). Object recognition system (100) with a sensor system (110) and a computing unit (120) according to Claim 11 . Computer program that causes a processing unit (120) to perform all method steps of a method according to one of the Claims 1 to 10 to be performed when it is executed on the computing unit (120). Machine-readable storage medium with a computer program stored thereon Claim 13 .

Images