Method and device for online calibration of an environmental sensing system in a motor vehicle

Abstract

The invention relates to a method and a device for online calibration of an environmental sensing system in a motor vehicle, the method comprising the following steps: during the vehicle's journey, the environmental sensing system continuously senses positioning data of objects in its surroundings and transmits them wirelessly to a calibration service, together with motion data of the vehicle; the calibration service has access to a road signature in the form of a digital map in which the positions of permanent stationary signature objects that can be sensed by the environmental sensing system are recorded along the road traveled by the vehicle, and filters the positioning data transmitted by the environmental sensing system according to the signature objects; based on the signature object positioning data and on the motion data, the calibration service tracks the vehicle's motion on the digital map; by comparing the signature object positioning data with the positions of the signature objects relative to the vehicle on the digital map, the calibration service calculates correction data that are indicative of calibration errors of the environmental sensing system and transmits them to the vehicle; the environmental sensing system recalibrates based on the correction data.

Description

Technical Field

[0001] The present invention relates to a method for online calibration of an environmental sensing system in a motor vehicle and an apparatus for performing the method. Background Technology

[0002] In advanced driver assistance systems and autonomous driving systems, there are high requirements for the accuracy and reliability of the vehicle's environmental sensing system. The environmental sensing system of a motor vehicle consists of an increasing number of sensors (such as radar sensors, lidar sensors, and video cameras) used to locate and classify objects in the vehicle's surrounding environment. For angle-resolved sensors, calibration errors are prone to occur. These errors cause the sensor's optical axis to have an incorrect spatial relationship with the vehicle's horizontal direction of travel, resulting in systematic errors in angle measurements. Therefore, these sensors need to be calibrated before being put into operation to identify and computationally correct such calibration errors. For angle-resolved radar sensors, precise calibration of the antenna pattern is also required. This pattern illustrates the relationship between the amplitude and phase of the radar echoes received in multiple receiving channels and the azimuth angle of the located object.

[0003] Among known calibration methods, offline and online methods can be distinguished. In offline methods, vehicles pass through a calibration station at the end of the production line. This station has standardized reflectors, ensuring that the spatial relationship between the reflectors and the sensors in the vehicle is precisely known. This allows for comparison of sensor-obtained measurements with known reference values. Because calibration must be performed at relatively large reflector distances, and the calibration typically needs to cover a wide angular range of the sensors, the calibration station occupies a significant amount of space, and measurements require substantial time and labor. In high-productivity automotive factories, multiple calibration stations must be established to enable the necessary procedures to be performed on multiple vehicles in parallel. This results in enormous space requirements and high installation costs.

[0004] Once a vehicle is in operation, even minor parking collisions can cause sensor malfunctions and require recalibration. However, most repair shops lack the space to set up a large calibration station.

[0005] In online calibration methods, calibration is performed during operation by comparing and processing measurements from different sensors. In this way—at least under favorable conditions—calibration errors can be identified and eliminated. However, the accuracy of calibration is highly dependent on the local conditions in which it is performed. For example, in environments with a large number of metallic objects (such as guardrails or the like), online calibration becomes quite difficult, and therefore sufficient accuracy cannot always be guaranteed. Summary of the Invention

[0006] The objective of this invention is described as a precise calibration method that can be implemented at low cost and in a short time, without requiring a space-consuming calibration station.

[0007] According to the present invention, this task is solved by a method having the following steps: a) During vehicle operation, the environmental sensing system continuously senses the location data of objects in its surrounding environment and wirelessly transmits this location data, along with the vehicle's motion data, to a calibration service. b) The calibration service has access to road signatures in the form of a digital map and filters the location data sent by the environmental sensing system based on the signature object, in which the location of permanently stationary signature objects that can be sensed by the environmental sensing system is recorded along the road traveled by the vehicle. c) Based on the location data of the signed object and the motion data, the calibration service tracks the movement of the vehicle on the digital map; d) By comparing the location data of the signature object with the position of the signature object relative to the vehicle on the digital map, the calibration service calculates correction data and sends the correction data to the vehicle, the correction data (if present) indicating the calibration error of the environmental sensing system; e) The environmental sensing system is recalibrated based on the correction data.

[0008] The applicant provides an online service under the name "Bosch Straßensignatur" that provides reliable and robust vehicle positioning for precise lane navigation and automated driving. To this end, the positions of specific objects (such as traffic light poles, traffic signs, etc.) suitable as reference points for positioning using the vehicle's own sensor system are recorded in a constantly updated digital map of the area traversed by the vehicle; these objects are referred to herein as "signature objects." The basic idea of ​​this invention is to utilize this online service to precisely calibrate the vehicle's environmental sensing system during driving. In this way, interference from a large number of reflective objects (such as guardrails or the like, which are unsuitable as calibration references) can be avoided, and the accuracy of the calibration process is improved by the possibility of precisely locating the vehicle.

[0009] If a vehicle's environmental sensing system is to be calibrated, the system continuously sends its current location data to a calibration service over a certain period of time. This calibration service has access to or maintains the road signature itself. Then, in a filtering process, the location data of objects that are not signed are hidden. In this way, the amount of location data to be analyzed is reduced to a manageable size, and only objects whose exact location is known and therefore suitable as reference objects are considered. For example, in the first step, moving objects unsuitable as reference objects, such as pedestrians and moving vehicles, can be filtered out. This can be done within the environmental sensing system itself if necessary, so that the data does not need to be sent to the calibration service at all. Then, stationary objects that are not signed remain, such as extended objects unsuitable as location references, or objects that are only temporarily stationary (such as parked vehicles). As long as the vehicle's current location is at least roughly known, for example, based on GPS data, signed objects can be identified by comparing them with the road signature. Conversely, the vehicle can be precisely located based on previously identified signed objects. Since the calibration service also acquires vehicle motion data, it can extrapolate positioning based on this data, such as driving speed and steering wheel angle. The calibration error can then be identified and quantified by calculating the difference between the positioning data of the signed object measured by the environmental sensing system and the actual position of the signed object relative to the vehicle's current location. In this way, sufficiently accurate calibration of all sensors can be achieved even on relatively short driving sections.

[0010] Advantageous configurations and extensions of the invention are described below.

[0011] The filtering process performed by the calibration service may include, for example, the use of matched filter algorithms or artificial intelligence.

[0012] Road signatures do not need to cover the entire road network. Having road signatures for sections of the road network with limited space near vehicle manufacturers and repair shops is sufficient.

[0013] As the number of sensors in an environmental sensing system increases, the probability of identifying calibration errors based on differences in measurements from different sensors also increases, even if the data is insufficient for proper recalibration. In such cases, for example, an autonomous vehicle could automatically drive to the nearest area with a road signature and then recalibrate as it passes through that area.

[0014] On the other hand, the method proposed here can also be used to preventively check the calibration of environmental sensing systems during the time a vehicle travels through an area with road signatures.

[0015] "Road signatures" can be created not only for public transport roads, but also for test sections, such as those on production lines in motor vehicle manufacturers' factories or in repair shops. In this case, calibration may have already been performed during the manufacturing or testing process. Attached Figure Description

[0016] An embodiment will now be described in more detail with reference to the accompanying drawings. The drawings show: Figure 1 A diagram of a calibration system operating according to the method of the present invention; Figure 2 A portion of a digital map with road signatures, provided the vehicle is located; Figure 3 and Figure 2 The same local map, however, at a later moment when identifying calibration errors; Figure 4 Map sections with road signatures are used to illustrate the calibration process of the angle-resolved radar sensors; and Figure 5 A flowchart of the method according to the present invention. Detailed Implementation

[0017] exist Figure 1 The diagram schematically illustrates a vehicle 10 traveling on road 12 and equipped with an environmental sensing system 14. In the example shown, the environmental sensing system 14 consists of a long-range radar sensor 16 oriented forward in the direction of travel, four short-range radar sensors 18 at the corners of the vehicle, and an electronic control unit 20 that analyzes and processes data from all the radar sensors.

[0018] The control unit 20 is wirelessly connected to the communication interface 24 of the data processing device via interface 22. The data processing device provides calibration service 26 for online calibration of the environmental sensing system 14. It is understood that this calibration service can also be used by other vehicles.

[0019] The calibration service 26 has access to the road signatures 28. Roughly speaking, this is a data storage device containing a digital map of the road network, including road 12 traveled by vehicle 10. In this digital map, in addition to road 12, the locations of signature objects 30 are also recorded. These signature objects 30 are stationary objects that remain continuously in the same location and are therefore suitable as positioning reference points for vehicle 10. Examples of such signature objects include traffic light poles or traffic signs.

[0020] In the environment surrounding vehicle 10, besides the signature object 30, there are other objects 32 that are not recorded in the road signature 28 because they are either not stationary, such as moving vehicles or pedestrians, or at least not permanently stationary, such as parked vehicles. Extended reflective objects (such as guardrails) are also not recorded in the road signature 28 because they are not suitable as reference points for positioning. The positioning data of stationary objects 32 are sent to the calibration service 26 in the same way as the positioning data of signature object 30.

[0021] Figure 2 A slightly larger map portion of road signature 28 is shown, in which a total of five signature objects are recorded, labeled here as 30-1, 30-2, 30-3, 30-4, and 30-5 for better distinction. Signature objects 30-1 and 30-2 are, for example, traffic light poles erected at the same height on both sides of road 12. At a given moment, the radar sensor 16 of vehicle 10 locates these two traffic light poles and measures their distances r1 and r2. These two radii, along with the azimuth angles of these two objects and the location data of other objects located by the environmental sensing system 14, are reported to the calibration service 26. In addition, the environmental sensing system reports the approximate position of vehicle 10, for example, determined by a GPS system. This approximate position of the vehicle on road 12 allows the calibration service to identify two of the located objects as signature objects 30-1 and 30-2, and to distinguish object 30, which has no corresponding counterpart in road signature 28, from the signature objects. The exact location of vehicle 10, and more precisely, the exact location of radar sensor 16, can now be determined by drawing a circle of radius r1 around signature object 30-1 and a circle of radius r2 around signature object 30-2. One of the two intersection points of these circles (the one located in front of the traffic light pole in the direction of travel) is the exact location of vehicle 10. At the same instant, radar sensor 16 also locates a more distant signature object 30-3, such as a traffic sign. Radar sensor 16 measures the azimuth angle α3 for this object and reports it to the calibration service as well. This allows the calibration service to determine the instantaneous direction of travel R1 of vehicle 10 based on road signature 28. The significant deviation of this direction of travel from the direction of road 12 indicates that the measured angle α3 is incorrect.

[0022] Figure 3The following describes a later point in time where vehicle 10 has approached two signature objects 30-4 and 30-5 (e.g., traffic light poles). The vehicle's new position is now determined using signature objects 30-4 and 30-5 in the same manner as before. As a result, the vehicle's true direction of travel is perfectly parallel to the direction of road 12. This information allows calibration service 26 to determine the offset angle Δα of radar sensor 16. Simultaneously, environmental sensing system 14 reports a new (also incorrect) azimuth angle of signature object 30-3. However, calibration service 26 feeds back the offset angle Δα to environmental sensing system 14, allowing it to now calculate the true angle. During recalibration, environmental sensing system 14 stores the offset angle Δα and subsequently uses it to correct all azimuth angles measured by radar sensor 16.

[0023] In addition to the object's location data, the environmental sensing system also reports the vehicle 10's motion data to the calibration service 26. This motion data specifically includes the vehicle's speed, steering angle, and / or longitudinal and lateral accelerations measured by inertial sensors. Once the precise location of the vehicle 10 in the digital map is known, the calibration service can continue to track the vehicle's trajectory based on the motion data, even if the location data of the signed object is not available. Therefore, in Figure 2 and Figure 3 In the example shown, the calibration service can be certain that no steering intervention occurred during vehicle motion. Typically, motion data enables micro-localization of the vehicle, while the location data of the signed object enables macro-localization. The comparison between the micro-localization and macro-localization then allows for the supervision of the calibration.

[0024] Figure 4 Another map section of road signature 28 is shown, where road 12 has a curved trajectory. Once the exact position of vehicle 10 is known, the calibration service can also track the vehicle along the curved trajectory based on its speed and steering angle. Throughout the movement, radar sensor 16 locates the individual signature object 30 and reports its (corrected) azimuth to the calibration service. However, based on the known vehicle trajectory, the calibration service can also calculate the azimuth independently of the radar sensor's measurement data and use this to check the radar sensor's calibration. If a discrepancy arises between the measured and calculated azimuth during this process, the calibration service feeds these discrepancies back to the environmental sensing system 14. Based on this information, the environmental sensing system 14 can correct the radar sensor's antenna pattern to the entire angular range swept by the signature object 30 during the curved driving process during recalibration.

[0025] In a similar manner, calibration service 26 can also check and, if necessary, correct the calibration of the remaining radar sensors 18. The key here is that calibration relies on a controlled number of signed object 30 location data, while all other location data is identified as unsigned objects and discarded by calibration service 26 at the latest. In this way, even with a large number of radar sensors to be calibrated, the amount of data to be processed remains within limits.

[0026] exist Figure 5 The flowchart illustrates the basic steps of the method according to the present invention.

[0027] In step S1, the environmental sensing system 14 checks, for example, based on GPS signals, whether a road signature exists for the road traveled by the vehicle. If so (J), the environmental sensing system begins transmitting positioning data from radar sensors 16 and 18 in step S2. Then, the calibration service 26 filters out signature objects from the positioning data in step S3. In another embodiment, only positioning data for stationary targets is transmitted. Then, the calibration service 26 filters out only stationary objects that are not recorded as signature objects in the digital map.

[0028] In principle, calibration services can also be implemented within the vehicle itself, with road signatures downloaded only from the cloud.

[0029] Then, in step S4, calibration service 26 tracks the vehicle's movement on a digital map using the transmitted positioning data and vehicle motion data. This allows the calculation of expected positioning data (distance, azimuth, and—in some cases where the height of the signing object is known—elevation angle) at any given time based on the relative positions of the signing object and the vehicle. Then, in step S5, the calculated data is compared with the positioning data transmitted by environmental sensing system 14. If a significant deviation occurs during this process, the necessary correction value is calculated in step S6 and reported to environmental sensing system 14, which then adjusts the calibration accordingly. Afterward, or if no significant deviation occurs, the process jumps back to step S1, and these steps are repeated cyclically.

[0030] If step S1 indicates that no road signature is found for the route being traveled, the program branches to step S7. Here, a check is performed to determine if a calibration error is indicated, for example, based on possible differences between measurement data from different radar sensors or between measurement data from a radar sensor and a video camera. If this is not the case (N), the program jumps back to step S1. Otherwise (J), an instruction is given to the driver or autonomous driving system to drive to the nearest area with a road signature, so that the calibration procedure can then be initiated in the subsequent execution of step S1.

Claims

1. A method for online calibration of an environmental sensing system (14) in a motor vehicle (10), Its characteristics include the following steps: a) During the driving of the vehicle (10), the environmental sensing system (14) continuously senses the location data of objects (30, 32) in its surrounding environment and wirelessly transmits the location data along with the motion data of the vehicle (10) to the calibration service (26). b) The calibration service (26) has access to road signatures (28) in the form of a digital map and filters the location data sent by the environmental sensing system (14) according to the signature object, in which the location of permanently stationary signature objects (30) that can be sensed by the environmental sensing system is recorded along the road (12) traveled by the vehicle in the digital map. c) Based on the location data of the signature object (30) and the motion data, the calibration service (26) tracks the motion of the vehicle (10) on the digital map; d) By comparing the location data of the signature object (30) with the position of the signature object relative to the vehicle (10) on the digital map, the calibration service (26) calculates correction data and sends the correction data to the vehicle (10), the correction data—if present—indicating the calibration error of the environmental sensing system (14); e) The environmental sensing system (14) is recalibrated based on the correction data.

2. The method according to claim 1, wherein, The matched filter algorithm is used to filter the transmitted location data.

3. The method according to claim 1 or 2, wherein, Artificial intelligence is used to filter the location data sent.

4. The method according to any one of the preceding claims, wherein, The environmental sensing system (14) also reports the location of the vehicle (10) as described by the GPS system, along with the positioning data and the motion data, to the calibration service (26).

5. The method according to any one of the preceding claims, wherein, Whenever the vehicle (10) is in an area where a road signature (28) exists, the calibration of the sensors of the environmental sensing system (14) is checked.

6. The method according to any one of the preceding claims, wherein, If there are signs of potential calibration errors, a driver prompt will be automatically output, or in the case of an autonomous vehicle, an instruction will be output to drive to the nearest area with a road signature (28).

7. The method according to any one of the preceding claims, wherein, The calibration is performed during the manufacture or testing of the motor vehicle, using road signatures created for the production line or test section.

8. An apparatus for online calibration of an environmental sensing system for motor vehicles, characterized in that... An electronic data processing system, the electronic data processing system being programmed to provide calibration services (26) for the method according to claim 1.

Images