DE102019132151A1 - Method for calibrating an environment sensor of a vehicle on the basis of data from a reference environment sensor, taking into account occupancy cards, computing device and sensor system - Google Patents

Abstract

The invention relates to a method for calibrating an environment sensor of a vehicle, comprising the steps of: receiving sensor data from the environment sensor and determining an occupancy card based on the sensor data, the occupancy card describing a first area of an environment of the vehicle, receiving reference sensor data from a reference Surrounding sensor of the vehicle and determining a reference occupancy map based on the reference sensor data, the reference occupancy map describing a second area of the environment which corresponds at least in some areas to the first area, determining at least one feature in the occupancy map which is associated with at least one Feature in the reference occupancy map matches, determining a spatial displacement between the at least one feature in the occupancy map and the at least one feature in the reference occupancy map, and estimating a deviation of an installation angle of the environment sensor from one Target installation angle based on the shift.

Description

The present invention relates to a method for calibrating an environment sensor of a vehicle on the basis of reference sensor data of a reference environment sensor. The present invention also relates to a computing device for a sensor system of a vehicle and to such a sensor system. The present invention also relates to a computer program and a computer-readable (storage) medium.

Different environment sensors for sensor systems of vehicles are known from the prior art. Such environment sensors can be designed, for example, as ultrasonic sensors, radar sensors or cameras. In the present case, the interest applies in particular to environmental sensors which are designed as lidar sensors. Such lidar sensors (Lidar - light detection and ranging) are essential sensors for automated driving and driver assistance systems, since these sensors generate a detailed representation of the environment around the vehicle. In order to be able to reliably detect the surroundings of the vehicle with these lidar sensors, it is necessary that they are precisely calibrated.

Knowledge of the installation angle of the lidar sensor is essential here, since otherwise orientation errors will occur, which have a major influence, particularly with objects that are far away. Precise knowledge of the installation angle of the environment sensor is required for driver assistance systems and for autonomous driving. If objects are detected by several environment sensors, they are detected much wider or, in extreme cases, even as two separate objects in the event of a calibration error. In addition, virtual movements occur when a road user approaches the vehicle, since the angle error causes a significantly larger lateral error with objects that are far away than with objects in the immediate vicinity of the vehicle. For example, if there is an orientation error of 0.5 °, this can lead to a position error of 1.3 m for an object 150 m away. This can have the consequence that the assignment of objects or other road users to lanes is no longer possible. The installation position of the sensor can usually be determined more precisely on the basis of CAD data and has little influence on the accuracy of the measurement results.

A fundamental goal in driver assistance systems or sensor systems of vehicles is therefore to determine the installation angle of environment sensors as precisely as possible so that the following algorithms or the signal evaluation can be carried out as precisely as possible. The calibration of the environment sensor is to be carried out on the one hand after the sensor has been installed in the vehicle or after the vehicle has been manufactured. Due to thermal influences, vibrations and the like, it is also possible that the installation angle of the environment sensor changes during the operation of the vehicle and therefore a new calibration is necessary. Environment sensors and lidar sensors are mostly calibrated with so-called calibration targets. This means that several calibration targets or objects are set up, of which the position is known. It is often necessary that the environment sensor then has to be adjusted manually. This procedure is very complex and always requires a competent person. The vehicle is usually driven into a so-called calibration lane. Such a calibration method is not possible during testing or during operation of the vehicle.

Calibration methods that can be carried out online are also known from the prior art. This means that these are carried out while driving, without having to cover a specific route. In addition, these procedures can be carried out unsupervised, ie without trained personnel. For example, calibration methods are known from the prior art which use so-called point clouds. To this end, the US 9 297 899 B2 a method for determining extrinsic calibration parameters for a sensor. Here, data is received that describe the position of the vehicle. In addition, sensor data is received from the sensor. Corresponding point clouds are then determined on the basis of the sensor data, with each point in the point cloud being assigned a covariance based on the history of the movement of the vehicle. The extrinsic parameters are then calibrated on the basis of this point cloud.

In addition, calibration methods are known from the prior art in which sensors are automatically calibrated on the basis of detected objects. To this end, the US 6,202,027 B1 a method in which a movement path of the vehicle is predicted. In addition, objects in the vicinity of the vehicle are detected with the environment sensor and the calibration is carried out on the basis of the objects detected.

The object of the present invention is to provide a solution for how the calibration of environment sensors, in particular lidar sensors, of a vehicle can be carried out with less effort and yet reliably.

According to the invention, this object is achieved by a method, by a computing device, by a sensor system, by a computer program and by a computer-readable (storage) medium with the features according to the independent claims. Advantageous developments of the present invention are specified in the dependent claims.

A method according to the invention is used to calibrate an environment sensor of a vehicle. The method includes receiving sensor data from the environment sensor and determining an occupancy card on the basis of the sensor data, the occupancy card describing a first area of an area around the vehicle. In addition, the method includes receiving reference sensor data from a reference environment sensor of the vehicle and determining a reference occupancy map based on the reference sensor data, the reference occupancy map describing a second area of the environment, which at least partially coincides with the first area matches. In addition, the method comprises the determination of at least one feature in the occupancy card which corresponds to at least one feature in the reference occupancy card. In addition, the method includes determining a spatial shift between the at least one feature in the occupancy map and the at least one feature in the reference occupancy map. Finally, the method includes estimating a deviation of an installation angle of the environment sensor from a target installation angle on the basis of the displacement.

With the aid of the method, an environment sensor, in particular a lidar sensor, is to be calibrated for a vehicle. Such an environment sensor can be part of a sensor system or a driver assistance system of the vehicle. The method can be carried out with a corresponding computing device, which can be formed, for example, by a control unit of the vehicle or of the sensor system. When the vehicle is in operation, the sensor data can be recorded with the environment sensor. The sensor data describe the first part of the area around the vehicle. These sensor data can then be transmitted from the environment sensor to the computing device.

In addition, the reference sensor data are provided with the reference environment sensor. The reference environment sensor can be a radar sensor, a camera, a laser scanner or a lidar sensor. The reference sensor data, which are provided with the reference environment sensor, describe the second area of the environment of the vehicle. It is provided that the first area, which the first sensor data describe, at least in areas corresponds to the second area. Furthermore, it is preferably provided that the sensor data and the reference sensor data are recorded at the same time or with a predetermined time offset from one another. This means in particular that the sensor data and the reference sensor data describe the respective areas of the environment for the same point in time. It can also be provided that the sensor data and the reference sensor data describe the respective areas of the environment at two different points in time, the duration between these two points in time falling below a predetermined limit value. The sensor data and the reference sensor data can therefore be determined within a predetermined time range or during a measurement cycle.

Another possibility is to continuously determine the vehicle's own movement. The vehicle's own movement can be determined on the basis of odometry. This means, for example, that the data from wheel speed sensors, the data from a steering angle sensor and / or the data from an initial measuring unit are used. Alternatively or additionally, the vehicle's own movement can be determined using the data from a satellite-supported position determination system or using GPS data. If the intrinsic movement is known, the detections can be entered in a fixed occupancy card and reference occupancy card. A stationary occupancy card can therefore be determined using the sensor data and a stationary reference occupancy card can be determined using the reference sensor data. In this case, the duration has no influence.

Both the sensor data and the reference sensor data are transmitted to the computing device. The occupancy card can then be determined by means of the computing device on the basis of the sensor data. This occupancy map can also be referred to as a grid-based map or as a grid map. Occupancy cards of this type are used in most sensor systems or driver assistance systems for recording the vehicle surroundings in order to make the large number of sensor data manageable. These occupancy cards represent the surroundings of the vehicle, similar to an image from a camera. The occupancy map can have individual cells or points which are assigned to respective areas in the vicinity of the vehicle. Furthermore, the cells can indicate whether the area in the vicinity assigned to the cell is occupied by an object or not. Analogously to the determination of the occupancy card, the computing device opens the reference occupancy card Determined based on the reference sensor data. The occupancy map describes the first area of the environment and the reference occupancy map describes the second area of the environment, which at least partially corresponds to the first area. There is therefore an area in the vicinity that is mapped both by the occupancy card and by the reference occupancy card.

Furthermore, corresponding features can be recognized both in the occupancy card and in the reference occupancy card. These features can describe corresponding objects in the vicinity of the vehicle. It is provided that at least one feature is determined in the occupancy card, which is also present in the reference occupancy card. For example, several features can be recognized in the occupancy card. Several features can also be recognized in the reference occupancy card. It can also be checked whether the features that were recognized in the occupancy card can be assigned to the features in the reference occupancy card. In principle, it is sufficient if at least one feature is found that is present both in the occupancy card and in the reference occupancy card. A spatial offset or spatial shift is then determined for this at least one matching feature. For example, the occupancy map and the reference occupancy map can be determined with respect to a coordinate system, for example a vehicle coordinate system.

The position of the at least one feature in the occupancy card and the position of this at least one feature in the reference occupancy card can be determined. The spatial shift or offset between the features can then be determined on the basis of the respective positions. The spatial shift or the offset can be determined, for example, with respect to two spatial directions. An installation error in the environment sensor can then be determined on the basis of this spatial shift. In particular, a deviation of an installation angle of the environment sensor from the setpoint installation angle can be determined. In particular, it is provided here that the installation angle is determined with respect to an azimuth direction or about the vertical axis of the vehicle.

It is assumed here that the reference environment sensor is precisely calibrated. In particular, it is assumed that the reference environment sensor is precisely calibrated with regard to the installation angle. A corresponding calibration of the reference environment sensor can be carried out before the method is carried out. The calibration of the environment sensor is then carried out on the basis of the reference sensor data or the reference occupancy card. The calibration process can be carried out after the vehicle has been manufactured or after the environment sensor has been installed. It is not necessary here for appropriate calibration targets to be used to calibrate the environment sensor or for the vehicle to be maneuvered along a calibration lane.

The calibration process can therefore be carried out online or while the vehicle is in motion. In addition, the method can be used to check the calibration of the environment sensor. This can be carried out, for example, at predetermined times or at the beginning of a journey. Furthermore, the method can be used to carry out a recalibration. In addition, the process can be carried out unsupervised or without trained personnel. Furthermore, the method has an advantage over known methods from the prior art, which have a very high computational effort or require self-localization. Overall, the calibration method can thus be carried out with less effort, but nevertheless reliably.

It can also be provided that a combined occupancy card is determined. This combined occupancy card can be determined on the basis of the sensor data and the reference sensor data. This combined occupancy map can be determined for the area in the vicinity of the vehicle which is covered by both the sensor data and the reference sensor data. This combined occupancy card can thus represent a combination or intersection of the occupancy card and the reference occupancy card. Features that have been determined on the basis of the sensor data and features that have been determined on the basis of the reference sensor data can be present in the combined occupancy card. The spatial shift between these features can then be determined and the deviation in the installation angle can be derived from this.

The occupancy map and the reference occupancy map are preferably determined for a plurality of successive measurement cycles, a deviation value is determined for each measurement cycle, which describes the deviation in the installation angle, and the deviation in the installation angle is estimated based on the plurality of deviation values. With the environment sensor and the reference environment sensor, for example, several measurement cycles in succession can be carried out. The sensor data and the reference sensor data can then be used in each measuring cycle to be determined. Furthermore, the occupancy card and the reference occupancy card can be determined for the respective measuring cycles. The deviation value can then be determined for at least some of the measuring cycles on the basis of the at least one matching feature in the occupancy card and the reference occupancy card. This deviation value describes the deviation of the installation angle from the target installation angle for the respective measuring cycle. These deviation values can be determined, for example, for a predetermined period of time or for a predetermined number of times. For example, several hundred deviation values can be determined.

Based on this large number of deviation values or the individual estimates, the deviation of the installation angle or the most likely orientation can then be estimated by means of a robust method. In other words, the rotation error can be determined over a large number of measurements and the most likely orientation error can be determined using statistics. To estimate the deviation, for example, the median or mean value of the deviation values can be calculated. It can also be provided that a robust estimation method such as RANSAC or the like is used. It can also be provided that a quality value is also determined. The respective deviation values or estimated values can be provided in a one-dimensional array which indicates the deviation values which lie within a distribution, as well as corresponding outliers. Thus, a criterion can be provided which describes the accuracy of the estimation method.

In a further embodiment, in order to determine the matching features in the occupancy card and in the reference occupancy card, features are extracted and an assignment of the extracted features is carried out. The matching features or the at least one feature in the occupancy card which matches the at least one feature in the reference occupancy card should therefore be determined. The occupancy map and the reference occupancy map usually comprise a plurality of grids or cells which have a predetermined length and width. For example, the occupancy card and / or the reference occupancy card can have several hundred individual cells per row and column. These cells can, for example, be assigned to an area in the vicinity of the vehicle with a length of 0.1 m. On the basis of the measurements with the environment sensor or the reference environment sensor, it can then be determined for each cell whether it is occupied by an object in the environment. The occupancy card and the reference occupancy card are similar to an image that is provided with a camera. It is therefore provided in the present case that algorithms that are used in image processing are transferred to the occupancy card and the reference occupancy card. In order to be able to recognize the respective features in the occupancy card and the reference occupancy card, common methods such as the ORB algorithm (ORB - Oriented FAST and Rotated BRIEF) can be used. In particular, it is provided here that point-like features are determined. The features are typical signatures or structures in the occupancy cards. For example, these can be corners, edges of vehicles, posts, ends of crash barriers, plants or the like.

In addition, the features that were extracted in the occupancy map are assigned to the features that were extracted in the reference occupancy map. These distinctive features, which were extracted using common image processing methods, can therefore be associated between the two occupancy cards. Corresponding methods which are known from image processing can also be used for this purpose. The at least one matching feature can thus be determined in a reliable manner.

It is also advantageous if, when assigning the extracted features, a check is carried out to determine whether the features lie in a predetermined distance range and / or a predetermined angular range. It can therefore be checked, for example, whether the distance between the environment sensor and the at least one extracted feature in the occupancy map essentially corresponds to the distance between the reference environment sensor and the extracted feature in the reference occupancy map. It can also be provided that it is checked whether the difference between these two distances lies in the predetermined distance range. This distance range can be, for example, 1 m in relation to the surroundings. As an alternative or in addition, it can be provided that the angular range is specified with respect to the azimuth direction. Here, the angle between the environment sensor and the at least one feature in the occupancy card is compared with the angle between the environment sensor and the recognized feature in the reference occupancy map.

This angular range can be, for example, ± 5 °, in particular ± 2 °, in relation to the surroundings. The angular range can be determined as a function of the configuration of the environment sensor and a typical value for the decalibration. If the recognized or extracted features lie within the distance range and / or the angle range, these can be assigned Features in the occupancy cards are carried out to one another. The assignment can thus be carried out in a reliable manner.

In a further embodiment, a map area is selected to determine the matching features from the occupancy card and the reference occupancy card, the selection of the map areas being carried out on the basis of the respective detection areas of the environment sensor and the reference environment sensor. Both the occupancy card and the reference occupancy card can each cover a relatively large area in the vicinity of the vehicle. For the calibration of the environment sensor, it can therefore be provided that the area to be viewed is restricted or that map areas are selected from both the occupancy map and the reference occupancy map. The map areas can be selected from the occupancy map and the reference occupancy map in such a way that they describe the same area in the vicinity of the vehicle. In addition, the detection areas of the environment sensor and the reference environment sensor can be taken into account. The respective detection areas describe those areas in the surroundings of the vehicle in which objects can be detected with the environment sensor or the reference environment sensor. The computational effort can be significantly reduced through the selection of the respective map areas of the occupancy card and the reference occupancy card.

In addition, provision can also be made for the respective map areas to be adapted as a function of a speed of the vehicle. For example, the map areas can be enlarged as the speed of the vehicle increases. It can also be provided that the map areas are adapted as a function of the surroundings of the vehicle. If the vehicle is moving in a densely populated area or in an urban area, the respective map areas can be selected to be correspondingly smaller.

In a further embodiment, the occupancy card and the reference occupancy card are determined in such a way that they each describe static elements in the vicinity of the vehicle. For this purpose, so-called dynamic occupancy maps, for example, can be determined which describe dynamic or moving objects in the environment. These dynamic occupancy cards can be determined on the basis of the raw data from the environment sensor or the reference environment sensor. These dynamic occupancy cards can be used to mask the general occupancy card or the general reference occupancy card accordingly. In this way, the static objects or elements in the environment can be determined. As an alternative or in addition to this, it can be provided that a static occupancy card and a static reference occupancy card are determined. The specific occupancy map and the specific reference occupancy map can therefore in particular only describe static objects in the vicinity of the vehicle. Thus, the calibration of the environment sensor can be carried out on the basis of the static elements or objects in the environment. In this way, it can be prevented that the calibration is carried out on the basis of moving objects and thus cannot be carried out precisely.

In a further embodiment, linear features are recognized in the occupancy card and in the reference occupancy card and assigned to one another. Furthermore, it is preferably provided that an angle difference between the linear features is determined and the deviation of the installation angle is additionally determined on the basis of the angle difference. It is preferably provided here that the linear features are determined in two-dimensional spaces, which are determined on the basis of the occupancy map and the reference occupancy map by means of a Hough transformation.

The use of the Hough transformation or the consideration of the features in the Hough space is particularly suitable for linear elements or linear objects in the environment. If there are many linear objects in the vicinity of the vehicle, it can be difficult to extract the previously described point-like features, which are determined, for example, by means of ORB. Linear objects in the vicinity can be buildings, guard rails or the like, for example. It is therefore also provided in the present case that an analysis is carried out in the Hough space. The Hough space is examined both from the occupancy card and from the reference occupancy card. Common image processing methods can be used for this purpose. This two-dimensional space or Hough space describes the linear elements on the basis of their angle and the distance to the environment sensor or the reference environment sensor. For example, the Hough space can be determined with respect to the environment sensor which is to be calibrated. If several line-shaped features are recognized in the two-dimensional space, provision can be made for individual ones of these line-shaped features to be selected. Between the linear features, which were determined on the basis of the occupancy card, and the linear features Features that have been determined on the basis of the reference occupancy map, the angle difference with respect to the azimuth direction can then be determined in a reliable manner.

In addition, it can be provided that a correlation function, in particular a cross-correlation, is used to determine the offset or angle difference in the one-dimensional Hough spaces. Overall, a reliable calibration can therefore also be made possible in the case of linear objects in the vicinity of the vehicle.

A computing device according to the invention for a sensor system of a vehicle is designed or set up to carry out a method according to the invention with advantageous embodiments thereof. The computing device can in particular be formed by a control unit of a vehicle. The one-dimensional Hough space can only describe the azimuth angle. This can be achieved by reducing the distance axis to one value. For this purpose, for example, a maximum value can be determined over all distances per azimuth angle. Alternatively, the sum of the n-largest values can be determined, where n = 3 can be assumed, for example.

A sensor system according to the invention comprises a computing device according to the invention, an environment sensor and a reference environment sensor. The environment sensor and the reference environment sensor can be designed as a radar sensor or as a camera. In particular, however, it is provided that the environment sensor and / or the reference environment sensor are designed as a lidar sensor. The lidar sensor can be arranged in the area of the bumper of the vehicle. The reference environment sensor can be designed as a laser scanner or as a lidar sensor. For example, the reference environment sensor can be arranged in an area of a roof of the vehicle. The reference environment sensor can have a detection range of 360 ° in relation to the azimuth direction. The sensor system can also have a plurality of environment sensors.

The sensor system can be part of a driver assistance system of the vehicle. The vehicle can be maneuvered automatically or autonomously using the driver assistance system. A vehicle according to the invention comprises a sensor system according to the invention. The vehicle is designed in particular as a passenger car.

A further aspect of the invention relates to a computer program comprising instructions which, when the program is executed by a computing device, cause the computer to execute a method according to the invention and the advantageous refinements thereof. The invention also relates to a computer-readable (storage) medium, comprising instructions which, when executed by a computing device, cause the computing device to execute a method according to the invention and the advantageous embodiments thereof.

The preferred embodiments presented with reference to the method according to the invention and their advantages apply accordingly to the computing device according to the invention, for the sensor system according to the invention, for the driver assistance system according to the invention, for the vehicle according to the invention, for the computer program according to the invention and for the computer-readable (memory) medium.

Further features of the invention emerge from the claims, the figures and the description of the figures. The features and combinations of features mentioned above in the description, as well as the features and combinations of features mentioned below in the description of the figures and / or shown alone in the figures, can be used not only in the specified combination, but also in other combinations or on their own, without the frame to leave the invention.

• 1 eine schematische Darstellung eines Fahrzeugs, welches ein Sensorsystem mit einem Umfeldsensor und einem Referenz-Umfeldsensor aufweist, sowie eines Objekts, welches sich in einer Umgebung des Fahrzeugs befindet;
• 2 ein schematisches Ablaufdiagramm eines Verfahrens zum Kalibrieren des Umfeldsensors unter Verwendung einer Belegungskarte sowie einer Referenz-Belegungskarte des Referenz-Umfeldsensors;
• 3 eine schematische Darstellung der Belegungskarte sowie der Referenz-Belegungskarte, wobei die Belegungskarte und die Referenz-Belegungskarte jeweils Bereiche der Umgebung des Fahrzeugs beschreiben;
• 4 ein übereinstimmender Kartenbereich, welcher auf Grundlage der Belegungskarte, der Referenz-Belegungskarte sowie von Erfassungsbereichen des Umfeldsensors und des Referenz-Umfeldsensors bestimmt wird;
• 5 ein Merkmal in der Belegungskarte, welches einem Merkmal in der Referenz-Belegungskarte zugeordnet wird;
• 6 einen Bereich in der Belegungskarte, welcher für die Zuordnung des Merkmals in der Belegungskarte zu dem Merkmal in der Referenz-Belegungskarte verwendet wird;
• 7 eine kombinierte Belegungskarte, in welcher entsprechende Verschiebungen zwischen den Merkmalen der Belegungskarte und den Merkmalen der Referenz-Belegungskarte dargestellt sind; und
• 8 ein schematisches Ablaufdiagramm eines Verfahrens zum Kalibrieren des Umfeldsensors gemäß einer weiteren Ausführungsform.

The invention will now be explained in more detail on the basis of preferred exemplary embodiments and with reference to the accompanying drawings. Show:

• 1 a schematic representation of a vehicle which has a sensor system with an environment sensor and a reference environment sensor, as well as an object which is located in an environment of the vehicle;
• 2 a schematic flow diagram of a method for calibrating the environment sensor using an occupancy card and a reference occupancy card of the reference environment sensor;
• 3rd a schematic representation of the occupancy card and the reference occupancy card, the occupancy card and the reference occupancy card each describing areas of the surroundings of the vehicle;
• 4th a matching map area, which is determined on the basis of the occupancy card, the reference occupancy map and detection areas of the environment sensor and the reference environment sensor;
• 5 a feature in the occupancy card which is assigned to a feature in the reference occupancy card;
• 6th an area in the occupancy card which is used for the assignment of the feature in the occupancy card to the feature in the reference occupancy card;
• 7th a combined occupancy card in which corresponding shifts between the features of the occupancy card and the features of the reference occupancy card are shown; and
• 8th a schematic flow diagram of a method for calibrating the environment sensor according to a further embodiment.

Identical or functionally identical elements are provided with the same reference symbols in the figures.

1 shows a vehicle in a schematic representation 1 , which is designed here as a passenger car, in a plan view. The vehicle 1 includes a sensor system 2 , which is part of a driver assistance system of the vehicle 1 can be. The sensor system 2 includes an environment sensor 4th , a reference environment sensor 5 as well as a computing device 3rd . It can also be provided that the sensor system 2 several environmental sensors 4th having. In the present simplified example, the sensor system comprises 2 only one environment sensor 4th , which is designed as a lidar sensor. In the example shown is the environment sensor 4th in a front area of the vehicle 1 arranged. With the environment sensor 4th or the lidar sensor can objects 7th in an environment 6th of the vehicle 1 are recorded. In the present example, an object is schematic 7th in the neighborhood 6th of the vehicle 1 shown.

With the environment sensor 4th sensor data can be provided which the object 7th in the neighborhood 6th describe. This sensor data can from the environment sensor 4th to the computing device 3rd be transmitted. The reference environment sensor 5 is also designed here as a lidar sensor or as a laser scanner. The reference environment sensor 5 is on a roof or an area of the roof of the vehicle 1 assembled. The reference environment sensor 5 comprises a circumferential range of 360 ° with respect to an azimuth direction. It is assumed here that the reference environment sensor 5 is precisely calibrated.

So that the object 7th with the environment sensor 4th or the lidar sensor can be reliably detected, it is necessary that the environment sensor 4th is precisely calibrated. In the present case, in particular, the environment sensor is calibrated 4th with regard to an installation angle Φ of interest. This installation angle should be preferred Φ with respect to an azimuth direction of the environment sensor 4th or with respect to a rotation about a vehicle vertical axis z. If the calibration of the environment sensor 4th regarding the installation angle Φ is not precise, orientation errors can occur, which are particularly evident in the case of objects that are far away 5 impact. When the environment sensor 4th is not sufficiently calibrated, this can lead to one of the environment sensor 4th detected object position from an actual position of the object 5 deviates. This can also have the consequence that the sensor data with the environment sensor 4th are provided, not with the reference sensor data, which with the reference environment sensor 5 can be provided, processed or merged together.

2 shows a schematic flow diagram of a method for calibrating the environment sensor 4th . In one step S1 are with the environment sensor 4th Sensor data provided. It will also be in one step S2 with the reference environment sensor 5 Reference sensor data provided. At the step S1 as well as at the step S2 the sensor data or the reference sensor data can be converted into a coordinate system, in particular a coordinate system of the vehicle 1 to be transformed. For example, the coordinate system can be determined with respect to a vehicle longitudinal direction x and a vehicle transverse direction y. The origin of the coordinate system can be a center point of the rear axle of the vehicle 1 be assigned. Alternatively, a fixed coordinate system can be selected. It is necessary here, the position of the environment sensor 4th or the sensor position is stored in this coordinate system at the respective measurement time.

Both the sensor data and the reference sensor data describe respective areas 11 , 12th the environment 6th of the vehicle 1 . An occupancy card is then created based on the sensor data M1 provided. This occupancy card M1 may have multiple cells representing respective areas in the area 6th of the vehicle 1 assigned. In addition, on the basis of the measurements with the environment sensor 4th determine whether the individual cells of the occupancy card M1 through an object 7th in the neighborhood 6th of the vehicle 1 are occupied or whether they are free of objects 7th are. Analogous to the determination of the occupancy card M1 becomes a reference occupancy card based on the reference sensor data M2 provided. In the present example, based on the occupancy card M1 and the reference occupancy card M2 additionally a combined occupancy card M3 generated.

In one step S3 becomes from the occupancy card M1 a map area 9 selected. Furthermore, in one step S4 from the reference occupancy card M2 the map area 9 selected which one with the map area 9 from the Occupancy card M1 matches. To determine the respective map areas 9 can also use the combined occupancy card 3rd can be used. In one step S5 becomes the map area 9 the occupancy card M1 masked. With this masking, those features of the occupancy card M1 does not take into account which dynamic objects or moving objects 7th in the neighborhood 6th of the vehicle 1 describe. Analogous to the step S5 is in one step S6 the map area 9 the reference occupancy card M2 masked. To carry out the steps S5 and S6 can each have information from the occupancy card M1 as well as from the reference occupancy card M2 can be used. It has to be taken into account that the steps S3 to S6 are optional, but improve the accuracy of the method and reduce the computing time.

In one step S7 are from the occupancy card M1 or the possibly masked map area 9 features F1 extracted. Similarly, in one step S8 from the reference occupancy card M2 features F2 extracted. To extract the features F1 , F2 from the occupancy cards M1 , M2 Corresponding image processing methods can be used. In one step S9 it is then checked whether at least one feature F1 in the occupancy card M1 with at least one feature F2 from the reference occupancy card M2 matches. In this case, it is checked whether the occupancy card M1 a feature F1 exists which is the same object 7th in the neighborhood 6th of the vehicle 1 describes how a feature F2 from the reference occupancy card M2 . If an appropriate assignment of the characteristics F1 , F2 can be done in one step S10 a spatial offset or a spatial shift between the features F1 , F2 to be determined.

It can be provided that the occupancy card M1 and the reference occupancy card M2 be determined for a plurality of chronologically successive measuring cycles. A deviation value can then be determined for each of the measuring cycles, which represents the deviation of the installation angle Φ describes from a target installation angle. This deviation value can be based on the spatial offset between the features F1 , F2 in the occupancy cards M1 , M2 be determined. The respective deviation values can be stored in a memory 10 of the sensor system 2 be deposited. The deviation of the installation angle of the environment sensor can then be determined on the basis of a plurality of deviation values 4th can be determined from a target installation angle (step S11 ). This result can then be used when acquiring the sensor data according to step S1 be used. A feedback can thus be provided, as it were.

3rd shows a schematic representation of a first area 11 the environment 6th which from the occupancy card M1 is mapped. There is also a second area 12th the environment 6th shown which of the reference occupancy card M2 is mapped. The area 13th is taken from the combined occupancy card M3 represents. There is also an area 8th shown which both from the occupancy card M1 as well as from the reference occupancy card M2 is mapped.

4th shows a further schematic representation of the first area 11 , the second area 12th as well as the matching area 9 . The reference environment sensor 5 has a detection range of 360 ° with respect to the azimuth direction. Thus, this reference environment sensor 5 no restrictions on the azimuth direction but for a distance. Here is the detection area of the reference environment sensor 5 within the second range 12th based on the lines 14th illustrated. The environment sensor 4th which is in the front of the vehicle 1 is arranged, has a detection area which is larger than the first area 11 is. This is shown schematically in the present case by the lines 15th indicated. This is done within the area 8th and the respective detection areas of the environment sensor 4th as well as the reference environment sensor 5 a map area 9 determines in which the matching characteristics F1 , F2 the occupancy card M1 and the reference occupancy card M2 be compared with each other.

5 shows a schematic representation of the occupancy card M1 . In the occupancy card M1 is also the position of the environment sensor 4th indicated. In the occupancy card M1 became the feature here F1 recognized which one to the environmental sensor 4th the distance d1 and the angle Φ1 having. This feature F1 was also considered a feature F2 in the reference occupancy card M2 as a feature F2 detected. To the environment sensor 4th exhibits this characteristic F2 a distance d2 as well as an angle Φ2 on. It can be checked whether an amount is a difference between the distance d2 between the environment sensor 4th and the feature F2 to a distance d1 between the environment sensor 4th and the feature F1 is within a predetermined distance range. A deviation in the distances d1 , d2 or the distance values can only be caused by an incorrect installation position of the environment sensor 4th or arise from measurement uncertainties. Both sources are for lidar sensors or laser scanners in the range from centimeters to a maximum of a few decimeters. Furthermore, it can be checked whether an amount is a difference between the angle Φ2 between the environment sensor 4th and the feature F2 and the angle Φ1 between the environment sensor 4th and the feature F1 is within a predetermined angular range. An area can then be derived from this 16 around the feature F1 around can be determined in which the feature F2 has to be located in order for it to be the characteristic F1 can be assigned. This is schematically shown in 6th illustrated.

7th shows a schematic representation of a combined occupancy card M3 . The combined occupancy card M3 includes features F1 determined from the sensor data and features F2 which were determined based on the reference sensor data. Through the lines 17th is the assignment of the matching characteristics F1 , F2 shown. They are also through the lines 18th Outliers or incorrectly assigned characteristics F1 , F2 illustrated. As can be seen, the number of outliers is relatively small. After the outliers 18th detected, the angle from the displacements (lines 17th ) the deviation of the installation angle Φ estimated from the target installation angle. This twist was used to illustrate two viewing angles (lines 19th and 20th ).

8th shows a schematic flow diagram of a method for calibrating the environment sensor 4th according to a further embodiment. Here, too, the occupancy card is based on the sensor data M1 certainly. Furthermore, the reference occupancy card is based on the reference sensor data M2 certainly. In one step S12 is then based on the occupancy card M1 a two-dimensional Hough space is calculated. Furthermore, analog in one step S13 based on the reference occupancy card M2 also calculated a two-dimensional Hough space. An installation position of the environment sensor can be used to calculate this Hough space 4th must be taken into account.

Line-shaped features can be determined in the respective Hough spaces, which line-shaped objects 7th the environment 6th assigned. In one step S14 Corresponding maximum values for the azimuth angle can be determined in a one-dimensional Hough space. This can be done for the reference occupancy card M2 analog in one step S15 be performed. An angular offset can then be determined between the respective linear features. This can also be done in one step S16 to determine the deviation of the installation angle Φ of the environment sensor 4th can be taken into account by a target installation angle.

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

• US 9297899 B2 [0005]
• US 6202027 B1 [0006]

Claims (12)

Method for calibrating an environment sensor (4) of a vehicle (1), comprising the steps: - Receiving sensor data from the environment sensor (4) and determining an occupancy card (M1) on the basis of the sensor data, the occupancy card (M1) describing a first area (11) of an environment (6) of the vehicle (1), - Receiving reference sensor data from a reference environment sensor (5) of the vehicle (1) and determining a reference occupancy map (M2) based on the reference sensor data, the reference occupancy map (M2) having a second area (12) of the environment (6) describes which corresponds to the first area (11) at least in some areas, - Determination of at least one feature (F1) in the occupancy card (M1) which corresponds to at least one feature (F2) in the reference occupancy card (M2), - determining a spatial shift between the at least one feature (F1) in the occupancy card (M1) and the at least one feature (F2) in the reference occupancy card (M2), and - Estimating a deviation of an installation angle (Φ) of the environment sensor (4) from a target installation angle on the basis of the displacement. Procedure according to Claim 1 , characterized in that the occupancy card (M1) and the reference occupancy card (M2) are determined for a plurality of successive measurement cycles, a deviation value is determined for each measurement cycle, which describes the deviation of the installation angle (Φ), and the deviation of the installation angle (Φ) is estimated from the plurality of deviation values. Procedure according to Claim 1 or 2 , characterized in that in order to determine the matching features (F1, F2) in the occupancy card (M1) and in the reference occupancy card (M2) features (F1, F2) are extracted and the extracted features (F1, F2) are assigned becomes. Procedure according to Claim 3 , characterized in that when assigning the extracted features (F1, F2) it is checked whether the features (F1, F2) lie in a predetermined distance range and / or a predetermined angular range. Procedure according to Claim 3 or 4th , characterized in that a map area (9) is selected to determine the matching features (F1, F2) from the occupancy card (M1) and the reference occupancy card (M2), the selection of the map areas (9) based on respective detection areas the environment sensor (4) and the reference environment sensor (5) is carried out. Method according to one of the preceding claims, characterized in that the occupancy card (M1) and the reference occupancy card (M2) are determined in such a way that they each describe static elements in the surroundings (6) of the vehicle (1). Method according to one of the preceding claims, characterized in that in the occupancy card (M1) and in the reference occupancy card (M2) each line-shaped features are recognized and assigned to one another, an angle difference between the line-shaped features is determined and the deviation in the installation angle ( Φ) is also determined based on the angle difference. Procedure according to Claim 7 , characterized in that the linear features are determined in two-dimensional spaces, which are determined on the basis of the occupancy map (M1) and the reference occupancy map (M2) by means of a Hough transformation. Computing device (3) for a sensor system (2) of a vehicle (1), wherein the computing device (3) is designed to carry out a method according to one of the preceding claims. Sensor system (2) for a vehicle (1) comprising a computing device (3) according to Claim 9 , an environment sensor (4) and a reference environment sensor (5). Computer program, comprising instructions which, when the program is executed by a computing device (3), cause the latter, a method according to one of the Claims 1 to 8th to execute. Computer-readable (storage) medium, comprising instructions which, when executed by a computing device (3), cause them, a method according to one of the Claims 1 to 8th to execute.

Images