JP6973351B2 - Sensor calibration method and sensor calibration device - Google Patents

Description

The disclosure according to this specification relates to a calibration technique for calibrating a plurality of in-vehicle sensors mounted on a vehicle.

Conventionally, for example, Patent Document 1 discloses a calibration method for calibrating a rider or a camera mounted on a vehicle by using an object existing around the own vehicle. Specifically, in the calibration method of Patent Document 1, GPS (Global Positioning System) information of another vehicle traveling in the vicinity of the own vehicle is acquired by vehicle-to-vehicle communication. Then, the rider or camera of the own vehicle is calibrated by the process of comparing the information perceived by the rider or camera with the information received from another vehicle.

U.S. Patent Application Publication No. 2018/0196127

The GPS information of other vehicles used in the calibration method disclosed in Patent Document 1 is prone to positioning errors, especially during traveling. Therefore, if GPS information of another vehicle whose reliability is unknown is used for calibration of an in-vehicle sensor such as a rider or a camera, it may be difficult to secure the accuracy of the calibrated parameter.

It is an object of the present disclosure to provide a sensor calibration method and a sensor calibration device capable of improving the calibration accuracy of an in-vehicle sensor.

In order to achieve the above object, one disclosed embodiment is a sensor calibration method performed by a computer (100) and calibrating a plurality of in-vehicle sensors (30) mounted on a vehicle (A). On at least one processor (61), the measurement information of the ground fixed object (RO) measured by the in-vehicle sensor is acquired (S102), and the known information of the ground fixed object by the information source (50) different from the in-vehicle sensor is obtained. Prepare (S105), evaluate at least the sensor reliability of each in-vehicle sensor that outputs the measurement information as the reliability of the measurement information (S106), and use the measurement information and known information for the same ground fixed object. External parameters set between multiple in-vehicle sensors are updated by comparing known information and measurement information, and the sensor reliability associated with each of the two in-vehicle sensors to which the external parameters are applied is used together to update the external parameters. It is a sensor calibration method including a step of changing the adjustment amount of (S107, S108).

Further, one aspect disclosed is a sensor calibration device used in a vehicle (A) and calibrating a plurality of in-vehicle sensors (30) mounted on the vehicle, and is a ground-fixed object measured by the in-vehicle sensors. The information acquisition unit (71) that acquires the measurement information of (RO), the information setting unit (84) that sets the known information of the ground fixed object by the information source (50) different from the in-vehicle sensor, and the reliability of the measurement information. The reliability evaluation unit (85) that at least evaluates the sensor reliability of each in-vehicle sensor that outputs measurement information, and the measurement information and known information about the same ground fixed object are used to obtain known information and measurement information. The external parameters set between multiple in-vehicle sensors are updated by the comparison of, and the adjustment amount of the external parameters in the update is changed by using the sensor reliability associated with each of the two in-vehicle sensors to which the external parameters are applied. It is a sensor calibration device including an update execution unit (86).

In these embodiments, the known information of the ground fixture is used to update the external parameters. Therefore, it is possible to update the external parameters between the in-vehicle sensors without being substantially affected by the positioning error that occurs in other vehicles. Based on the above, the accuracy of calibration of the in-vehicle sensor can be improved.

The reference numbers in parentheses are merely examples of the correspondence with the specific configuration in the embodiment described later, and do not limit the technical scope at all.

In one embodiment of the present disclosure, it is a figure which shows the whole picture of the structure related to calibration. It is a block diagram which shows the electric structure of an in-vehicle ECU, an infrastructure module and the like. It is a figure which shows an example of the reliability evaluation table used for quantifying the reliability of calibration. It is a flowchart which shows the detail of the sensor calibration process performed by the in-vehicle ECU.

In one embodiment of the present disclosure shown in FIGS. 1 and 2, the function of the sensor calibration device is mounted on the in-vehicle ECU (Electronic Control Unit) 100. The in-vehicle ECU 100 is one of a plurality of electronic control units mounted on the vehicle A, and is an in-vehicle computer that enables automatic driving or advanced driving support of the vehicle A. The vehicle-mounted ECU 100 is directly or indirectly electrically connected to the DCM41, the V2I communication device 43, the bus of the vehicle-mounted network 45, the plurality of vehicle-mounted sensors 30, and the like.

The DCM (Data Communication Module) 41 is a communication module mounted on the vehicle A. The DCM41 transmits and receives radio waves to and from base stations around the vehicle A by wireless communication in accordance with communication standards such as LTE (Long Term Evolution) and 5G. DCM41 enables the cooperation (Cloud to Car) between the cloud CLD and the in-vehicle device. The DCM41 shares various information such as map information and road information with the cloud database DBc installed on the cloud CLD. By installing the DCM41, the vehicle A becomes a connected car that can connect to the Internet.

The V2I (Vehicle to roadside Infrastructure) communication device 43 is a communication device for road-to-vehicle (V2I) communication. The V2I communication device 43 performs two-way communication with a roadside unit installed on the road. For example, on a road sign 10 or the like, an infrastructure module 10a capable of wirelessly communicating with a V2I communication device 43 is installed as a roadside device.

The infrastructure module 10a is composed of an infrastructure communication device 11, an infrastructure database (hereinafter, “infrastructure DB”) 12, an infrastructure controller 13 controlled by these, and the like. The infrastructure module 10a holds infrastructure information (details will be described later) related to the infrastructure such as the road sign 10 in the infrastructure DB 12. The infrastructure module 10a is connected to the Internet, and the infrastructure information of the infrastructure DB 12 can be updated to the latest information at any time. The infrastructure module 10a transmits the latest infrastructure information stored in the infrastructure DB 12 to the V2I communication device 43 through the infrastructure communication device 11.

A large number of in-vehicle devices are directly or indirectly electrically connected to the communication bus of the in-vehicle network 45. The in-vehicle network 45 can provide various vehicle information output to the communication bus to the in-vehicle ECU 100. The in-vehicle network 45 provides the in-vehicle ECU 100 with, for example, information indicating the traveling speed of the vehicle A (hereinafter, “vehicle speed information”) as information necessary for calibration described later.

The in-vehicle sensor 30 is mounted on the vehicle A and has a detection configuration for acquiring various information necessary for automatic driving or advanced driving support. The vehicle-mounted sensor 30 includes a GNSS receiver 31, an IMU 32, a millimeter-wave radar 33, a rider 34, a camera 35, and a sonar 36. The in-vehicle sensors 30 are installed at different positions and in different postures.

The GNSS (Global Navigation Satellite System) receiver 31 and the IMU (Inertial Measurement Unit) 32 are configured to specify the current position of the vehicle A. The GNSS receiver 31 receives positioning signals transmitted from a plurality of artificial satellites. The IMU 32 is composed of, for example, a 3-axis gyro sensor, a 3-axis acceleration sensor, or the like, and measures the inertial force acting on the vehicle A. The positioning signal received by the GNSS receiver and the measurement result of the IMU 32 are sequentially provided to the vehicle-mounted ECU 100 as own vehicle position information indicating the current position of the vehicle A.

The millimeter wave radar 33, the rider 34, the camera 35, and the sonar 36 are autonomous sensors that recognize the surrounding environment of the vehicle A. These autonomous sensors detect moving objects such as pedestrians and other vehicles, as well as stationary objects such as traffic signals, road signs 10 and road markings 20 such as lane markings and pedestrian crossings. Each autonomous sensor sequentially outputs measurement information including detection results of moving objects and stationary objects to the in-vehicle ECU 100. In addition, each autonomous sensor may be a plurality of each. Also, some autonomous sensors may be omitted.

The millimeter wave radar 33 irradiates the millimeter wave toward the traveling direction of the vehicle A, and acquires a detection result by a process of receiving the millimeter wave reflected by a moving object, a stationary object, or the like existing in the traveling direction. The rider 34 irradiates the laser beam in the traveling direction of the vehicle A or in the left-right front direction, and receives the laser beam reflected by the moving object, the stationary object, or the like existing in the irradiation direction, and acquires the detection result. The rider 34 may be a scanning type such as a rotating mirror system, a MEMS system, and a phased array system, or may be a non-scanning type such as a flash system. The camera 35 acquires a detection result by a process of analyzing a front image obtained by capturing a front region of the vehicle A. The sonar 36 acquires detection information by irradiating ultrasonic waves toward the periphery of the vehicle A and receiving ultrasonic waves reflected by moving objects, stationary objects, and the like existing in the irradiation direction.

The in-vehicle ECU 100 is an arithmetic unit that recognizes the traveling environment by combining the own vehicle position information and the measurement information acquired from each in-vehicle sensor 30. The in-vehicle ECU 100 continuously repeats the process of specifying the position of the own vehicle, the process of calculating the relative distance to the object around the own vehicle, and the like. The in-vehicle ECU 100 is mainly composed of a control circuit including a processor 61, a RAM 62, a memory device 63, an input / output interface 64, and the like.

The processor 61 is hardware for arithmetic processing combined with the RAM 62, and can execute various programs. The memory device 63 has a configuration including a non-volatile storage medium, and stores various programs executed by the processor 61. In the memory device 63, at least a database update program, a sensor calibration program, and the like are stored in addition to the environment recognition program for recognizing the traveling environment. The database update program and the sensor calibration program are programs related to calibration for each in-vehicle sensor 30.

The in-vehicle ECU 100 includes functional units such as a sensor information acquisition unit 71, a parameter storage unit 72, and an object identification unit 73 by executing an environment recognition program. In addition, the in-vehicle ECU 100 includes a database update unit 77 as a functional unit by executing a database update program. Further, the in-vehicle ECU 100 includes functional units such as a stop determination unit 81, an environment estimation unit 82, a target selection unit 83, an infrastructure information setting unit 84, a reliability evaluation unit 85, and an update execution unit 86 by executing a sensor calibration program.

The sensor information acquisition unit 71 acquires the own vehicle position information provided by the GNSS receiver 31 and the IMU 32. In addition, the sensor information acquisition unit 71 acquires measurement information from each of the millimeter wave radar 33, the rider 34, the camera 35, and the sonar 36.

The parameter storage unit 72 stores external parameters for each in-vehicle sensor 30. The external parameter is a numerical group set between the two vehicle-mounted sensors 30, and is a numerical group that geometrically associates the two measurement information acquired by the two vehicle-mounted sensors 30. Specifically, the external parameters are defined in the form of 6 axes (x, y, z, roll, punch, yaw) corresponding to the mounting position and mounting posture (orientation) of each in-vehicle sensor 30. In the present embodiment, one of the plurality of vehicle-mounted sensors 30 is used as a reference master sensor for setting external parameters. The parameter storage unit 72 stores external parameters relative to the master sensor for each in-vehicle sensor 30 excluding the master sensor. According to the process of applying the external parameter to the measurement information of the vehicle-mounted sensor 30, the position coordinates of the detected object in the coordinate system of the vehicle-mounted sensor 30 can be converted into the position coordinates in the coordinate system of the master sensor.

The object identification unit 73 has a relative distance, direction, shape (size, height, area, etc.), type, etc. of the detected object detected from around the vehicle based on the measurement result acquired by the sensor information acquisition unit 71. To identify. In such a process, the object identification unit 73 carries out a process of combining the measurement information of a plurality of autonomous sensors, so-called sensor fusion. In the sensor fusion, the object identification unit 73 uses the external parameters stored in the parameter storage unit 72, and in the measurement results of two different autonomous sensors, the coordinates indicating the geometrically identical points are associated with each other for the detection accuracy. To improve.

The database update unit 77 updates the infrastructure information registered in the in-vehicle database (hereinafter, “in-vehicle DB”) 76 with the latest information. The in-vehicle DB 76 is a storage area secured in a memory device 63 or the like for storing infrastructure information, and functions as a local database. The database update unit 77 repeats the infrastructure information update process at a predetermined cycle to maintain the infrastructure information of the in-vehicle DB 76 in the latest state.

The infrastructure information is information associated with a specific infrastructure used in the calibration described later, specifically, a ground fixed object RO. The ground-fixed object RO is an object that is fixed to the ground and whose shape is predetermined by laws and regulations. The ground-fixed object RO includes, for example, a road sign 10 and a road marking 20. The infrastructure information includes shape information such as height and area from the ground in addition to absolute position information (latitude, longitude, altitude) of the ground fixed object RO. The road marking 20 drawn on the road surface is associated with, for example, infrastructure information indicating that the height is “0”.

The stop determination unit 81 determines the traveling state of the vehicle A based on the vehicle speed information of the vehicle A provided from the vehicle-mounted network 45. Specifically, the stop determination unit 81 determines that the vehicle A is stopped when the traveling speed indicated by the vehicle speed information is zero. The stop determination unit 81 may determine the stop of the vehicle A based on the own vehicle position information provided by the GNSS receiver 31.

The environment estimation unit 82 estimates the current environment around the vehicle based on the front image taken by the camera 35 and the communication information received by the DCM 41 or the V2I communication device 43. As an example, the environment estimation unit 82 determines the time (day / night or light / dark) and the weather.

When a plurality of ground-fixed objects RO are detected by the millimeter-wave radar 33, the rider 34, the camera 35, and the sonar 36, the target selection unit 83 updates the external parameters from the plurality of ground-fixed objects RO. Select the ground fixed object RO to be used for. The target selection unit 83 refers to the identification result of the object identification unit 73, and grasps the type of the ground fixed object RO around the vehicle detected by the autonomous sensor. The target selection unit 83 sets the priority to be used for calibration based on the type of each ground fixed object RO. The target selection unit 83 sets a higher priority for the ground-fixed object RO, which is likely to ensure the reliability of calibration described later. Specifically, the ground-fixed object RO having a higher (closer to 1) feature reliability specified in the reliability evaluation table (see FIG. 3), which will be described later, has a higher priority.

The target selection unit 83 sets the priority order of each ground fixed object RO, and then determines the presence / absence of infrastructure information in order from the ground fixed object RO having the highest priority. The target selection unit 83 selects the ground fixed object RO having the highest priority among the ground fixed object ROs in which the infrastructure information exists, as the target to be used for updating the external parameters. The above-mentioned feature reliability is larger for the road sign 10 erected on the road surface than for the road marking 20 drawn on the road surface. Therefore, the target selection unit 83 preferentially selects the road sign 10 as the target of use over the road marking 20.

The infrastructure information setting unit 84 acquires the infrastructure information of the ground fixed object RO selected as the target to be used by the target selection unit 83, and sets it in a state that can be compared with the measurement information. The infrastructure information setting unit 84 can acquire infrastructure information from three information sources 50. The first information source 50 is a cloud database DBc. The map information and road information stored in the cloud database DBc are constantly updated with the latest information. The infrastructure information setting unit 84 acquires the infrastructure information included in the map information or the road information from the cloud CLD through the DCM41. The second information source 50 is the infrastructure module 10a. The infrastructure information setting unit 84 acquires the infrastructure information stored in the infrastructure database 12 through the V2I communication device 43. The third information source 50 is an in-vehicle DB 76. The infrastructure information setting unit 84 acquires the infrastructure information stored in the vehicle-mounted DB 76 by a process of referring to the memory device 63.

The reliability evaluation unit 85 evaluates the reliability of the measurement information by the in-vehicle sensor 30. Specifically, the reliability evaluation unit 85 quantifies the reliability of the measurement information and enables calibration according to the reliability. The higher the reliability, the larger the adjustment amount of the external parameter in the update. It is assumed that such reliability is affected by the measurement environment such as time and weather, the type and size of the ground fixed object RO targeted for measurement, and the type of the in-vehicle sensor 30 that outputs the measurement information. Therefore, a reliability evaluation table (see FIG. 3) that quantifies the influence of each factor on the reliability is defined in advance. By using the reliability evaluation table, the reliability evaluation unit 85 adjusts the reliability value according to the measurement environment, the type of the ground-fixed object RO, the type of the in-vehicle sensor 30, and the like.

As shown in FIGS. 2 and 3, evaluation criteria such as time reliability, weather reliability, feature reliability, and sensor reliability are preset in the reliability evaluation table. The time reliability is set to "1" in the daytime zone and "0.5" in the nighttime zone. The weather reliability is "1" in fine weather, "0.9" in cloudy weather, "0.8" in rainy weather, and "0.5" in the event of snowfall and fog. Further, if the wind speed is less than a specific wind speed (for example, 5 m / s), the weather reliability is set to "1", and if there is a wind exceeding the specific wind speed, the weather reliability is set to less than 1. 5 / wind speed (unit is m / s) ”.

The feature reliability is set for each type of ground fixed object RO. As an example, in the reliability evaluation table, types such as a road sign 10, a traffic light, a street light, a building, a road marking 20, a road edge, and a roadside plant are set as a ground fixed object RO. Then, the reliability corresponding to the time or the weather is assigned to each type of the ground-fixed object RO.

The sensor reliability is set for each autonomous sensor that outputs the measurement result. In this embodiment, the millimeter wave radar 33, the rider 34, the camera 35, and the sonar 36 are set as the types of the autonomous sensor. And, as with the feature reliability, the reliability corresponding to the time or the weather is assigned to each type of autonomous sensor. In an environment where object recognition by an autonomous sensor is difficult, such as heavy rain, heavy snow, and heavy fog, the weather reliability and the sensor reliability may be set to lower values.

Further, the reliability evaluation unit 85 uses the size information of the ground fixed object RO, specifically, the area information of the ground fixed object RO for the reliability evaluation. The area information of the ground-fixed object RO may be a value based on the measurement result specified by the object identification unit 73, or may be a value included in the infrastructure information. The reliability evaluation unit 85 sets the size reliability by a process of dividing a specific coefficient (for example, 0.04) by the area (unit: m ^ 2) of the ground-fixed object RO.

The maximum value of the size reliability is set to "1". Further, the area of the ground-fixed object RO may be the projected area of the ground-fixed object RO when viewed from the vehicle-mounted sensor 30, or may be the projected area when the ground-fixed object RO is viewed from the front.

The reliability evaluation unit 85 has the above-mentioned time reliability, weather reliability, size reliability, minimum value of feature reliability, minimum value of first sensor reliability, and the above-mentioned minimum value of the first sensor reliability, as described in the following (Equation 1). The reliability of the comprehensive measurement information is set by multiplying the minimum value of the reliability of the second sensor.
(Equation 1) Reliability = Time reliability x Weather reliability x Size reliability x Minimum value of feature reliability
× Minimum value of first sensor reliability × Minimum value of second sensor reliability

As an example, the reliability when the road sign 10 having an area of "0.05 m ^ 2" is recognized by the rider 34 and the camera 35 under the condition that the time is "night" and the weather is "rain" is as follows (Equation 2). )become that way.
(Equation 2) Reliability = 0.5 × 0.8 × (0.04 / 0.05) × 0.8
× 0.7 × 0.7 = 0.125
Here, the minimum value of the feature reliability is one of the smaller one of "0.8" at night and "1.0" at rain. Similarly, the minimum sensor reliability of the lidar 34 is the smaller of "1.0" at night and "0.7" of rain, and the minimum sensor reliability of the camera 35 is "1.0" at night. It is one of the smaller of "0.7" and "0.7" of rain.

The update execution unit 86 continuously carries out iterative calculation to update the external parameters by comparing the measurement information and the infrastructure information using the measurement information and the infrastructure information about the same ground fixed object RO. The update execution unit 86 executes the update process of the external parameter while taking into consideration the reliability set by the reliability evaluation unit 85. The update process for optimizing such external parameters corresponds to the calibration of each in-vehicle sensor 30. This calibration is different from the calibration performed in a factory, a dealer, or the like, and is an on-road calibration performed during the user's use. In the calibration, the update execution unit 86 treats the absolute position (and height information, etc.) of the ground fixed object RO indicated by the infrastructure information as a substantially true value. The update execution unit 86 sets the external parameter after calibration so that the error between the value obtained by applying the external parameter to the measurement result and the infrastructure information is reduced.

The details of the processing of the sensor calibration method carried out by the in-vehicle ECU 100 by the cooperation of the above functional units will be described with reference to FIG. 2 with reference to FIG. The sensor calibration process shown in FIG. 4 is started by, for example, starting the vehicle-mounted ECU 100 based on the switching of the vehicle power supply to the on state, and is continuously started until the vehicle power supply is switched to the off state.

In S101, it is determined whether or not the vehicle A has stopped. When it is determined in S101 that the vehicle A is traveling, the determination in S101 is repeated. Then, if it is determined in S101 that the vehicle A is stopped, the process proceeds to S102.

In S102, the measurement information of each autonomous sensor is referred to, the ground fixed object RO existing around the vehicle A is grasped, and the process proceeds to S103. In S103, with respect to the ground-fixed object RO grasped in S102, the priority of selection is set in the order of increasing reliability in calibration, and the process proceeds to S104.

In S104, it is determined whether or not there is infrastructure information in order from the ground fixed object RO having the highest priority set in S103. When it is determined in S104 that there is no infrastructure information, it is determined whether or not there is infrastructure information for the ground fixed object RO having the next highest priority. Then, when it is determined in S104 that there is infrastructure information, the ground-fixed object RO is selected as a calibration target, and the process proceeds to S105.

In S105, the infrastructure information of the ground fixed object RO selected in S104 is set. Specifically, in S105, the absolute position, height, area, etc. of the ground-fixed object RO are acquired from any of the three information sources 50, and the process proceeds to S106. In S106, the reliability of the calibration this time is calculated using the reliability evaluation table (see FIG. 3) based on the information acquired in S102, S105, and the like, and the process proceeds to S107.

In S107, the absolute position of the ground fixed object RO in the infrastructure information set in S105 and the absolute position of the vehicle A based on the own vehicle position information are prepared, and based on the difference between these absolute positions, the vehicle is transferred from the ground fixed object RO. Calculate the relative distance to A. Then, the calculated relative distance is compared with the measurement information, and the adjustment value EPt for updating the current external parameter EP1 is set.

Specifically, when applied to the measurement results of the two in-vehicle sensors 30 (autonomous sensors), the value of the external parameter that minimizes the error with respect to the relative distance based on the difference in absolute position is searched for as the adjustment value EPt. NS. Then, the difference between the searched adjustment value EPt and the current external parameter EP1 is calculated as an offset OST, and the process proceeds to S108. The absolute position of the vehicle A shall be corrected to the content based on the mounting position of the master sensor by using the external parameter associated with the GNSS receiver 31.

In S108, the external parameter EP1 is adjusted in consideration of the reliability calculated in S106. In S108, the value of the current external parameter EP1 is adjusted so as to be close to the adjustment value EPt. For example, the value obtained by integrating the reliability with the offset OST is set as the adjustment amount of the external parameter EP1. As a result, when the reliability is low, the difference between the external parameter EP1 before adjustment and the external parameter EP2 after adjustment becomes small (see the white arrow). On the other hand, when the reliability is high, the difference between the external parameter EP1 before adjustment and the external parameter EP2 after adjustment becomes large (see the dot arrow).

In the present embodiment described so far, infrastructure information, which is known information of the ground-fixed object RO, is used for updating the external parameters. Therefore, it is possible to update the external parameters between the in-vehicle sensors 30 without being substantially affected by the positioning error that occurs in another vehicle. Based on the above, the accuracy of calibration of the vehicle-mounted sensor 30 can be improved.

In detail, since the mounting position of the GNSS receiver in the other vehicle is unknown, the position information of the other vehicle received from the other vehicle becomes a factor that causes an error due to the size of the other vehicle in the calibration. On the other hand, the ground-fixed object RO used for calibration in this embodiment is smaller than that of other vehicles. Therefore, the error in position accuracy due to the object size can be reduced.

Further, if the existing ground-fixed object RO is used for calibration as in the present embodiment, it is not necessary to install a special road object for calibration. As a result, it is possible to continuously update external parameters in parallel with normal use by the user.

In addition, communication delays inevitably occur in vehicle-to-vehicle communication used when acquiring other vehicle position information. Therefore, an error occurs between the time when the relative position of the other vehicle is measured and the time when the position information of the other vehicle is acquired. As a result, the accuracy of calibration tends to decrease. On the other hand, in the present embodiment, the measurement result in the state where the vehicle A is stopped and the relative distance between the vehicle A and the ground fixed object does not substantially change is used for updating the external parameter. The measurement result when the relative distance is constant tends to be more accurate than the measurement result when the relative distance is changing. Therefore, if the measurement result acquired under the establishment of the stop determination is used, the accuracy can be further improved even in the on-road calibration.

Further, in the present embodiment, the object position information of the ground-fixed object RO and the own vehicle position information of the vehicle A are used in updating the external parameters. In this way, if the information used as the true value can be set based on the two position information, it is possible to simplify the arithmetic processing for updating the external parameter as compared with the case where the position information is not used.

Further, in the present embodiment, the process of updating the infrastructure information is repeated in each information source 50 that is the source of the infrastructure information. Therefore, the infrastructure information used in the calibration is real-time information that reflects the latest state. Based on the above, highly reliable calibration of external parameters becomes possible.

In addition, the in-vehicle ECU 100 of the present embodiment can acquire infrastructure information from the cloud database DBc or the infrastructure DB 12 which is an information source 50 outside the vehicle. Such external information of vehicle A is regularly updated with high reliability. Therefore, it is possible to avoid adverse effects due to replacement of the road sign 10 and erroneous recognition.

Further, in the present embodiment, a process of selecting a ground-fixed object RO to be used for calibration is performed. Based on the above, in a situation where a plurality of ground-fixed objects RO exist around the vehicle, the infrastructure information of the ground-fixed object RO suitable for ensuring the accuracy of calibration can be preferentially used. As a result, the accuracy of calibration is more likely to be ensured.

Further, in the present embodiment, the road sign 10 erected on the road surface is preferentially selected as the object to be used for calibration rather than the road marking 20 drawn on the road surface. As described above, the higher the ground fixed object RO, the easier it is to secure the accuracy of the relative distance indicated by the measurement result. Therefore, if the road sign 10 is preferentially selected, the certainty of ensuring the accuracy of calibration can be further improved.

In addition, in the form of receiving other vehicle position information from another vehicle, it is difficult to distinguish the reliability of the other vehicle position information because the environment in which the other vehicle position information is measured by the other vehicle is unknown. .. As a result, calibration using other vehicle position information may reflect unreliable information in external parameters. On the other hand, in the present embodiment, the reliability of the measurement result is possible, and after evaluating the reliability, the external parameters are updated according to the reliability. Therefore, it is possible to avoid the situation where the external parameter is calibrated to an inappropriate value based on the measurement result measured under adverse conditions. Further, if the measurement result can be obtained under favorable conditions, it becomes possible to effectively calibrate the external parameters.

Further, the reliability of the measurement result in the present embodiment is adjusted according to the time and weather, the type of the ground fixed object RO, the type of the in-vehicle sensor 30, and the like. As described above, by comprehensively incorporating the factors affecting the reliability into the evaluation of the reliability, the calibration of the external parameters can be performed more appropriately.

In the above embodiment, the infrastructure information corresponds to "known information", and the absolute position information of the ground fixed object RO corresponds to "object position information". Further, the road sign 10 corresponds to the "road building", the sensor information acquisition unit 71 corresponds to the "information acquisition unit", and the infrastructure information setting unit 84 corresponds to the "information setting unit". The in-vehicle ECU 100 corresponds to a "computer" and a "sensor calibration device".

(Other embodiments)
Although one embodiment of the present disclosure has been described above, the present disclosure is not construed as being limited to the above-described embodiment, and is applied to various embodiments and combinations without departing from the gist of the present disclosure. be able to.

In the above embodiment, the own vehicle position is specified by using the GNSS receiver and the IMU, but the method of acquiring the own vehicle position information may be appropriately changed. In the first modification of the above embodiment, the position of the own vehicle is specified by combining the measurement result of any one of the millimeter wave radar, the rider, the camera, and the sonar with the high-precision map acquired from the cloud. Further, the position of the own vehicle may be specified by the sensor fusion that combines a plurality of measurement results.

In the second modification of the above embodiment, the external parameters related to the GNSS receiver are calibrated. Further, in the modification 3 of the above embodiment, the external parameters related to the IMU are calibrated. As in the above modifications 2 and 3, the calibration of external parameters can be applied to in-vehicle sensors other than the autonomous sensor.

In the modified example 4 of the above embodiment, the external parameters are calibrated using the measurement results measured during traveling. In such a modification 4, the external parameters can be quickly converged by performing the calibration during traveling immediately after the replacement of the in-vehicle sensor.

In the fifth modification of the above embodiment, the DCM and the V2I communication device are omitted. Infrastructure information is acquired by referring to the in-vehicle DB. In the modification 6 of the above embodiment, the process of selecting the ground fixed object is omitted, and for example, calibration using the ground fixed object closest to the own vehicle is sequentially performed.

As in the above embodiment, the calculation method in which the reliability is added to the calibration may be changed as appropriate. Further, the specific reliability parameter value may be changed as appropriate. Further, in the modification 7 of the above embodiment, the evaluation of the reliability of the measurement information is omitted. Then, in the calibration, the adjustment value based on the measurement information is reflected in the current external parameter at a constant rate.

In the modified example 8 of the above embodiment, the height reliability is used instead of the size reliability or together with the size reliability. The height reliability is set based on the height of the ground fixed object, and a higher value is given to the ground fixed object having a height suitable for detection by the autonomous sensor. For example, if the height of the ground-fixed object is in the range of 1 to 10 m, the height reliability is set to "1". For ground-fixed objects with a height of less than 1 m or more than 10 m, the height reliability is set to "0.5", respectively.

The processor of the above embodiment is a processing unit including one or a plurality of CPUs (Central Processing Units). Such a processor may be a processing unit including a GPU (Graphics Processing Unit), a DFP (Data Flow Processor), and the like in addition to the CPU. Further, the processor may be a processing unit including an FPGA (Field-Programmable Gate Array) and an IP core specialized in specific processing such as learning and inference of AI. Each arithmetic circuit unit of such a processor may be individually mounted on a printed circuit board, or may be mounted on an ASIC (Application Specific Integrated Circuit), an FPGA, or the like.

As a memory device for storing a sensor calibration program or the like, various non-transitory tangible storage media such as a flash memory and a hard disk can be adopted. The form of such a storage medium may also be changed as appropriate. For example, the storage medium may be in the form of a memory card or the like, and may be inserted into a slot portion provided in an in-vehicle ECU and electrically connected to a control circuit.

The control unit and method thereof described in the present disclosure may be realized by a dedicated computer constituting a processor programmed to perform one or more functions embodied by a computer program. Alternatively, the apparatus and method thereof described in the present disclosure may be realized by a dedicated hardware logic circuit. Alternatively, the apparatus and method thereof described in the present disclosure may be realized by one or more dedicated computers configured by a combination of a processor for executing a computer program and one or more hardware logic circuits. Further, the computer program may be stored in a computer-readable non-transitional tangible recording medium as an instruction executed by the computer.

A vehicle, RO ground fixed object, 10 road sign (road sign), 20 road sign, 30 in-vehicle sensor, 50 information source, 61 processor, 71 sensor information acquisition unit (information acquisition unit), 84 infrastructure information setting unit (Information setting unit), 86 update execution unit, 100 in-vehicle ECU (computer, sensor calibration device)

Claims (11)

It is a sensor calibration method performed by a computer (100) and calibrating a plurality of in-vehicle sensors (30) mounted on a vehicle (A).
On at least one processor (61)
The measurement information of the ground fixed object (RO) measured by the in-vehicle sensor is acquired (S102),
Prepare known information of the ground-fixed object by an information source (50) different from that of the in-vehicle sensor (S105).
As the reliability of the measurement information, at least the sensor reliability of each in-vehicle sensor that outputs the measurement information is evaluated (S106).
Using the measurement information and the known information about the same ground-fixed object, the external parameters set between the plurality of in-vehicle sensors are updated by comparing the known information with the measurement information, and the external parameters are set. The adjustment amount of the external parameter in the update is changed by using the sensor reliability associated with each of the two applied in-vehicle sensors (S107, S108).
Sensor calibration method including the step. Further including the step of determining the stop of the vehicle (S101),
The sensor calibration method according to claim 1, wherein in the step of updating the external parameter, the measurement information measured in a state where the vehicle is stopped is used. The known information includes object position information indicating the position of the ground fixing object.
The vehicle-mounted sensor provides vehicle position information indicating the position of the vehicle.
The sensor calibration method according to claim 1 or 2, wherein the object position information and the own vehicle position information are used for updating the external parameters. The sensor calibration method according to any one of claims 1 to 3, wherein in the step of preparing the known information, the known information is provided from the information source that repeatedly updates to the latest information. The sensor calibration method according to any one of claims 1 to 4, wherein in the step of preparing the known information, the known information is provided from the information source outside the vehicle via wireless communication. One of claims 1 to 5, further comprising the step of selecting the ground fixing object to be used for updating the external parameter from the plurality of ground fixing objects located around the vehicle (S103). The sensor calibration method described in the section. In the step of selecting the ground fixture, the ground fixture (10) erected on the road surface is used for updating the external parameter rather than the road marking (20) drawn on the road surface. The sensor calibration method according to claim 6, which is preferentially selected. The sensor calibration method according to any one of claims 1 to 7, wherein the sensor reliability is adjusted according to at least one of the time and weather at which the measurement information is measured. The sensor calibration method according to any one of claims 1 to 8, wherein the sensor reliability is adjusted according to the type of the ground fixed object for which the measurement information is measured. The sensor calibration method according to any one of claims 1 to 9 , wherein the sensor reliability is adjusted according to the type of the vehicle-mounted sensor that outputs the measurement information. A sensor calibration device used in a vehicle (A) that calibrates a plurality of in-vehicle sensors (30) mounted on the vehicle.
An information acquisition unit (71) that acquires measurement information of a ground fixed object (RO) measured by the in-vehicle sensor, and
An information setting unit (84) that sets known information of the ground fixed object by an information source (50) different from the in-vehicle sensor, and
As the reliability of the measurement information, a reliability evaluation unit (85) that at least evaluates the sensor reliability of each in-vehicle sensor that outputs the measurement information, and
Using the measurement information and the known information about the same ground-fixed object, the external parameters set between the plurality of in-vehicle sensors are updated by comparing the known information with the measurement information, and the external parameters are set. An update execution unit (86) that changes the adjustment amount of the external parameter in the update by using the sensor reliability associated with each of the two applied in-vehicle sensors.
A sensor calibration device.

Images