Data acquisition device, data acquisition method, and program

Abstract

To make it possible to effectively utilize data uploaded to a server.SOLUTION: A data acquisition device 17 includes: an acquisition unit 3A that acquires at least a result of detection from a first ambient situation sensor which detects an ambient situation of a vehicle 1 and has a first detection area, and acquires at least a result of detection from a second ambient situation sensor which detects the ambient situation of the vehicle 1 and has a second detection area partially overlapping the first detection area; and an upload processing unit 3C that, when an object of recognition located within the first detection area and the second detection area cannot be recognized based on a result of detection by the first ambient situation sensor but can be recognized based on a result of detection by the second ambient situation sensor, executes processing of uploading the result of detection by the first ambient situation sensor to a server.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a data collection device, a data collection method, and a program.

Background Art

[0002] Patent Document 1 describes a technique in which the result of object identification performed in a vehicle is uploaded to a server. In the technique described in Patent Document 1, when object identification is correctly performed and the identification score is higher than a threshold value, the result of the correctly performed object identification is uploaded to the server.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] If all of the various sensor data measured by a vehicle during travel is acquired and uploaded to a server, the communication load and the storage capacity will increase, and the cost will be wasted. Therefore, it is necessary to narrow down the sensor data for transmission and storage. On the other hand, when only the result of correctly performed object identification is uploaded to the server as in the technique described in Patent Document 1, there is a possibility that the uploaded data cannot be effectively utilized.

[0005] In view of the above points, an object of the present disclosure is to provide a data collection device, a data collection method, and a program that can effectively utilize the data uploaded to a server.

Means for Solving the Problems

[0006] (1) One aspect of the present disclosure is to obtain at least detection results from a first peripheral situation sensor that detects the peripheral situation of a vehicle and has a first detection area, and to detect the peripheral situation of the vehicle and obtain at least detection results from a second peripheral situation sensor that has a second detection area that partially overlaps with the first detection area. An acquisition unit, and a recognition target located within the first detection area and within the second detection area is unrecognizable based on the detection result of the first peripheral situation sensor, and is recognizable based on the detection result of the second peripheral situation sensor. In this case, it is a data collection device including an upload processing unit that executes a process of uploading the detection result of the first peripheral situation sensor to a server.

[0007] (2) In the data collection device of (1), the upload processing unit may execute a process of uploading time-series detection results of the first peripheral situation sensor including the detection result of the first peripheral situation sensor at the time when the recognition target is unrecognizable to the server.

[0008] (3) In the data collection device of (1), the acquisition unit acquires sensor information indicating the mounting position and orientation of the first peripheral situation sensor with respect to the vehicle and vehicle information indicating the vehicle speed and yaw rate of the vehicle, and the upload processing unit acquires the sensor information and the vehicle information acquired by the acquisition unit. A process of uploading to the server may be executed.

[0009] (4) One aspect of the present disclosure is that a data collection device obtains at least detection results from a first peripheral situation sensor that detects the peripheral situation of a vehicle and has a first detection area, and detects the peripheral situation of the vehicle and the first detection area. An acquisition step of obtaining at least detection results from a second peripheral situation sensor having a second detection area that partially overlaps with the above, and the data collection device is located within the first detection area and within the second detection area. When the recognition target is unrecognizable based on the detection result of the first peripheral situation sensor and recognizable based on the detection result of the second peripheral situation sensor, it is a data collection method including an upload processing step of executing a process of uploading the detection result of the first peripheral situation sensor to a server.

[0010] (5) One aspect of the present disclosure is a program for causing a processor to execute an acquisition step of acquiring at least detection results from a first peripheral situation sensor that detects the peripheral situation of a vehicle and has a first detection area, and from a second peripheral situation sensor that detects the peripheral situation of the vehicle and has a second detection area that partially overlaps with the first detection area, and an upload process step of executing a process of uploading the detection results of the first peripheral situation sensor to a server when a recognition target located within the first detection area and within the second detection area is unrecognizable based on the detection results of the first peripheral situation sensor and is recognizable based on the detection results of the second peripheral situation sensor.

Advantages of the Invention

[0011] According to the present disclosure, it is possible to effectively utilize the data uploaded to the server.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Modes for Carrying Out the Invention

[0013] Hereinafter, embodiments of the data collection device, data collection method, and program of the present disclosure will be described with reference to the drawings.

[0014] <First Embodiment> FIG. 1 is a diagram showing an example of a vehicle 1 to which a data collection device 17 according to the first embodiment is applied. In the example shown in FIG. 1, the vehicle 1 includes a surrounding situation sensor 11, a surrounding situation sensor 12, a vehicle state sensor 13, an HMI (Human Machine Interface) 14, a communication device 15, a vehicle control device 16, a steering actuator 16A, a braking actuator 16B, a driving actuator 16C, and a data collection device 17. The surrounding situation sensors 11 and 12 detect the surrounding situation of the vehicle 1 (for example, obstacles, surrounding vehicles, pedestrians, etc. existing in the surroundings), and transmit the detection results to the vehicle control device 16 and the data collection device 17. The surrounding situation sensors 11 and 12 include, for example, LiDAR (Light Detection And Ranging), radar, sonar, etc.

[0015] FIG. 2 is a diagram for explaining a first example of the relationship between the detection area AR1 of the surrounding situation sensor 11 and the detection area AR2 of the surrounding situation sensor 12. In the example shown in FIG. 2, the surrounding situation sensor 11 is constituted by a sonar arranged at the right rear part of the vehicle 1, the surrounding situation sensor 12 is constituted by a sonar arranged at the left rear part of the vehicle 1, and the detection results of the surrounding situation sensor 11 and the surrounding situation sensor 12 are used for parking support, automatic parking, etc. of the vehicle 1. The detection area AR1 of the surrounding situation sensor 11 and the detection area AR2 of the surrounding situation sensor 12 partially overlap in the overlapping area ARV.

[0016] FIG. 3 is a diagram for explaining a second example of the relationship between the detection area AR1 of the surrounding situation sensor 11 and the detection area AR2 of the surrounding situation sensor 12. In the example shown in FIG. 3, the surrounding situation sensor 11 is constituted by a LiDAR disposed at the right front part of the vehicle 1, the surrounding situation sensor 12 is constituted by a LiDAR disposed at the left front part of the vehicle 1, and the detection results of the surrounding situation sensor 11 and the surrounding situation sensor 12 are used for driving assistance, automatic driving, etc. of the vehicle 1. The detection area AR1 of the surrounding situation sensor 11 and the detection area AR2 of the surrounding situation sensor 12 partially overlap in the overlapping area ARV.

[0017] In the example shown in FIG. 1, the vehicle state sensor 13 detects the state of the vehicle 1 and transmits the detection result to the vehicle control device 16 and the data collection device 17. The vehicle state sensor 13 includes, for example, a vehicle speed sensor, a yaw rate sensor, and the like. The HMI 14 has functions such as receiving various operations of the driver of the vehicle 1, and transmits a signal indicating the operation of the driver of the vehicle 1 to the vehicle control device 16. The communication device 15 communicates with the outside of the vehicle 1 (for example, the server SV (see FIG. 4) etc.).

[0018] FIG. 4 is a diagram showing an example of the relationship between the vehicle 1 and the server SV etc. In the example shown in FIG. 4, the vehicle 1 and the server SV are configured to be communicable via the network NW. Also, each of a plurality of other vehicles OV and the server SV are configured to be communicable via the network NW. Also, in the example shown in FIG. 4, a data collection system SY is constituted by the server SV, the vehicle 1, and a plurality of other vehicles OV, and information such as sensor data obtained in the vehicle 1 and the plurality of other vehicles OV is collected by the server SV.

[0019] In the example shown in FIG. 1, the vehicle control device 16 controls the steering actuator 16A, the brake actuator 16B, and the drive actuator 16C based on information (data, signals) transmitted from, for example, the surrounding situation sensors 11, 12, the vehicle state sensor 13, and the HMI 14. The data collection device 17 collects data such as the detection results of the surrounding situation sensors 11 and 12 and the detection results of the vehicle state sensor 13. Further, the data collection device 17 causes the communication device 15 to perform communication for uploading the data collected from the surrounding situation sensors 11 and 12 and the vehicle state sensor 13 to the server SV. The data collection device 17 is constituted by a microcomputer including a communication interface (I / F) 171, a memory 172, and a processor 173. The communication interface 171 has an interface circuit for connecting the data collection device 17 to the surrounding situation sensors 11 and 12, the vehicle state sensor 13, the HMI 14, the communication device 15, and the vehicle control device 16. The memory 172 stores programs and various data used in the processes executed by the processor 173. Specifically, the memory 172 stores sensor information indicating the mounting positions of the surrounding situation sensors 11 and 12 with respect to the vehicle 1, sensor information indicating the postures of the surrounding situation sensors 11 and 12, and the like. The processor 173 has a function as an acquisition unit 3A, a function as a recognition processing unit 3B, and a function as an upload processing unit 3C.

[0020] In the example shown in FIG. 1, the acquisition unit 3A acquires the detection result of the surrounding situation sensor 11 from the surrounding situation sensor 11 and the detection result of the surrounding situation sensor 12 from the surrounding situation sensor 12. In the example shown in FIG. 2, the acquisition unit 3A acquires the detection result of the sonar (surrounding situation sensor 11) disposed at the right rear part of the vehicle 1 and the detection result of the sonar (surrounding situation sensor 12) disposed at the left rear part of the vehicle 1. In the example shown in FIG. 3, the acquisition unit 3A acquires the detection result of the LiDAR (surrounding situation sensor 11) disposed at the right front part of the vehicle 1 and the detection result of the LiDAR (surrounding situation sensor 12) disposed at the left front part of the vehicle 1.

[0021] In the example shown in FIG. 1, the acquisition unit 3A acquires from the memory 172 sensor information indicating the mounting positions of the surrounding situation sensors 11 and 12 with respect to the vehicle 1, sensor information indicating the postures of the surrounding situation sensors 11 and 12, and the like. Further, the acquisition unit 3A acquires the detection result of the vehicle state sensor 13 (for example, vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate) from the vehicle state sensor 13.

[0022] Based on the detection result of the surrounding situation sensor 11 acquired by the acquisition unit 3A, the recognition processing unit 3B recognizes the recognition target RT (see FIGS. 2 and 3) included in the detection area AR1 of the surrounding situation sensor 11, and based on the detection result of the surrounding situation sensor 12 acquired by the acquisition unit 3A, recognizes the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12. Specifically, the recognition processing unit 3B uses, for example, a model obtained by performing learning using teacher data that is a data set of the detection result of the surrounding situation sensor mounted on the learning vehicle and a label indicating the attributes of the recognition target detected by the surrounding situation sensor (a label indicating what the recognition target detected by the surrounding situation sensor is), and based on the detection result of the surrounding situation sensor 11, recognizes the recognition target RT included in the detection area AR1 of the surrounding situation sensor 11 (identifies the attributes of the recognition target RT, etc.). Further, the recognition processing unit 3B uses the model to recognize the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12 based on the detection result of the surrounding situation sensor 12.

[0023] In the example shown in FIG. 2, the recognition processing unit 3B can recognize the recognition target RT included in the detection area AR1 of the surrounding situation sensor 11 based on the detection result of the surrounding situation sensor 11, and can also recognize the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12 based on the detection result of the surrounding situation sensor 12. Thus, the detection area AR1 of the surrounding situation sensor 11 and the detection area AR2 of the surrounding situation sensor 12 are partially overlapped in the overlapping area ARV, and the safety of the vehicle 1 is enhanced. However, in reality, there may be a case where the recognition processing unit 3B cannot recognize the recognition target RT located within the detection area AR1 of the surrounding situation sensor 11 and within the detection area AR2 of the surrounding situation sensor 12 (i.e., located within the overlapping area ARV), even though it can be recognized based on the detection result of the surrounding situation sensor 12, for example. Therefore, in the example shown in FIG. 1, for example, in order to enable the recognition processing unit 3B to recognize the recognition target RT based on the detection result (only) of the surrounding situation sensor 11, the countermeasures described later are taken. In the example shown in FIG. 3, the recognition processing unit 3B can recognize the recognition target RT included in the detection area AR1 of the surrounding situation sensor 11 based on the detection result of the surrounding situation sensor 11, and can also recognize the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12 based on the detection result of the surrounding situation sensor 12. Thus, the detection area AR1 of the surrounding situation sensor 11 and the detection area AR2 of the surrounding situation sensor 12 are partially overlapped in the overlapping area ARV, and the safety of the vehicle 1 is enhanced. However, in reality, there may be a case where the recognition processing unit 3B cannot recognize the recognition target RT located within the detection area AR1 of the surrounding situation sensor 11 and within the detection area AR2 of the surrounding situation sensor 12 (i.e., located within the overlapping area ARV), even though it can be recognized based on the detection result of the surrounding situation sensor 12, for example. Therefore, in the example shown in FIG. 1, for example, in order to enable the recognition processing unit 3B to recognize the recognition target RT based on the detection result (only) of the surrounding situation sensor 12, the countermeasures described later are taken.

[0024] Specifically, in the example shown in FIG. 1, the upload processing unit 3C executes a process of uploading the detection result of the peripheral situation sensor 11 or the detection result of the peripheral situation sensor 12 to the server SV (see FIG. 4). Specifically, in the example shown in FIG. 2 (an example in which the recognition processing unit 3B can recognize the recognition target RT based on the detection result of the peripheral situation sensor 12 and cannot recognize the recognition target RT based on the detection result of the peripheral situation sensor 11), the upload processing unit 3C executes a process of uploading the detection result of the peripheral situation sensor 11 to the server SV. The server SV not only collects the detection result of the peripheral situation sensor 11 from the vehicle 1, but also collects the attributes of the recognition target RT (such as those used as labels in the additional learning of the model described later) recognized by the recognition processing unit 3B based on the detection result of the peripheral situation sensor 12 from the vehicle 1. Further, the server SV uses the detection result of the peripheral situation sensor 11 collected from the vehicle 1 and the attributes of the recognition target RT, etc., to perform additional learning of the model used by the recognition processing unit 3B to recognize the recognition target RT included in the detection area AR1 of the peripheral situation sensor 11. By using the model for which additional learning has been performed by the server SV, the recognition processing unit 3B can recognize the recognition target RT based only on the detection result of the peripheral situation sensor 11 that was unable to recognize the recognition target RT before the additional learning of the model (that is, without relying on the detection result of the peripheral situation sensor 12).

[0025] In the example shown in FIG. 3 (an example in which the recognition processing unit 3B can recognize the recognition target RT based on the detection result of the peripheral situation sensor 11 and cannot recognize the recognition target RT based on the detection result of the peripheral situation sensor 12), the upload processing unit 3C executes a process of uploading the detection result of the peripheral situation sensor 12 to the server SV. Server SV not only collects the detection results of the surrounding situation sensor 12 from vehicle 1, but also collects from vehicle 1 the attributes etc. of the recognition target RT (what is used as a label in additional learning of the model) recognized by the recognition processing unit 3B based on the detection results of the surrounding situation sensor 11. Further, server SV performs additional learning of the model used by the recognition processing unit 3B to recognize the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12 by using the detection results of the surrounding situation sensor 12 collected from vehicle 1 and the attributes etc. of the recognition target RT. By using the model for which additional learning has been performed by server SV, the recognition processing unit 3B can recognize the recognition target RT based only on the detection results of the surrounding situation sensor 12 (that is, without based on the detection results of the surrounding situation sensor 11), which were impossible to recognize the recognition target RT before the additional learning of the model was performed.

[0026] More specifically, in the example shown in FIG. 2 (an example where the recognition processing unit 3B can recognize the recognition target RT based on the detection results of the surrounding situation sensor 12 and cannot recognize the recognition target RT based on the detection results of the surrounding situation sensor 11), the upload processing unit 3C does not upload only the detection results of the surrounding situation sensor 11 at the time when the recognition processing unit 3B cannot recognize the recognition target RT to server SV, but executes a process of uploading the detection results of the surrounding situation sensor 11 in a time series including the time when the recognition processing unit 3B cannot recognize the recognition target RT to server SV. As a result, it is possible to make the recognition processing unit 3B recognize the recognition target RT based only on the detection results of the surrounding situation sensor 11 more reliably than when only the detection results of the surrounding situation sensor 11 at the time when the recognition processing unit 3B cannot recognize the recognition target RT are uploaded to server SV. In the example shown in FIG. 3 (an example in which the recognition processing unit 3B can recognize the recognition target RT based on the detection result of the surrounding situation sensor 11 and cannot recognize the recognition target RT based on the detection result of the surrounding situation sensor 12), the upload processing unit 3C does not upload only the detection result of the surrounding situation sensor 12 at the time when the recognition processing unit 3B cannot recognize the recognition target RT to the server SV, but executes a process of uploading the detection results of the surrounding situation sensor 12 in a time series including the time when the recognition processing unit 3B cannot recognize the recognition target RT to the server SV. As a result, it is possible to ensure that the recognition processing unit 3B can recognize the recognition target RT based only on the detection result of the surrounding situation sensor 12 more reliably than when only the detection result of the surrounding situation sensor 12 at the time when the recognition processing unit 3B cannot recognize the recognition target RT is uploaded to the server SV.

[0027] In a modified example of the example shown in FIG. 2 (an example in which the recognition processing unit 3B can recognize the recognition target RT based on the detection result of the surrounding situation sensor 12 and cannot recognize the recognition target RT based on the detection result of the surrounding situation sensor 11), the upload processing unit 3C not only uploads the detection result of the surrounding situation sensor 11 to the server SV, but also uploads sensor information indicating the attachment position of the surrounding situation sensor 11 to the vehicle 1 acquired by the acquisition unit 3A, sensor information indicating the attitude of the surrounding situation sensor 11, and vehicle information indicating the state of the vehicle 1 such as the vehicle speed and yaw rate to the server SV. Server SV not only collects the detection results of the surrounding situation sensor 11 from the vehicle 1, but also collects from the vehicle 1 the attributes of the recognition target RT (such as those used as labels in additional learning of the model) recognized by the recognition processing unit 3B based on the detection results of the surrounding situation sensor 12. Further, the server SV collects from the vehicle 1 sensor information indicating the mounting position of the surrounding situation sensor 11 with respect to the vehicle 1, sensor information indicating the attitude of the surrounding situation sensor 11, and vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate. Furthermore, the server SV uses the detection results of the surrounding situation sensor 11 collected from the vehicle 1, the attributes of the recognition target RT, etc., the sensor information indicating the mounting position of the surrounding situation sensor 11 with respect to the vehicle 1, the sensor information indicating the attitude of the surrounding situation sensor 11, and the vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate, to perform additional learning of the model used by the recognition processing unit 3B to recognize the recognition target RT included in the detection area AR1 of the surrounding situation sensor 11. By using the model for which additional learning has been performed by the server SV, the recognition processing unit 3B can recognize the recognition target RT based on the detection results of the surrounding situation sensor 11 that were impossible to recognize the recognition target RT before the additional learning of the model, the sensor information indicating the mounting position of the surrounding situation sensor 11 with respect to the vehicle 1, the sensor information indicating the attitude of the surrounding situation sensor 11, and the vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate (that is, without relying on the detection results of the surrounding situation sensor 12).

[0028] In a modified example of the example shown in FIG. 3 (an example where the recognition processing unit 3B can recognize the recognition target RT based on the detection results of the surrounding situation sensor 11 and cannot recognize the recognition target RT based on the detection results of the surrounding situation sensor 12), the upload processing unit 3C not only uploads the detection results of the surrounding situation sensor 12 to the server SV, but also executes a process of uploading to the server SV the sensor information indicating the mounting position of the surrounding situation sensor 12 with respect to the vehicle 1 acquired by the acquisition unit 3A, the sensor information indicating the attitude of the surrounding situation sensor 12, and the vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate. The server SV not only collects the detection results of the surrounding situation sensor 12 from the vehicle 1, but also collects from the vehicle 1 the attributes of the recognition target RT (such as those used as labels in additional learning of the model) recognized by the recognition processing unit 3B based on the detection results of the surrounding situation sensor 11. Further, the server SV collects from the vehicle 1 sensor information indicating the mounting position of the surrounding situation sensor 12 with respect to the vehicle 1, sensor information indicating the attitude of the surrounding situation sensor 12, and vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate. Furthermore, the server SV uses the detection results of the surrounding situation sensor 12 collected from the vehicle 1, the attributes of the recognition target RT, etc., sensor information indicating the mounting position of the surrounding situation sensor 12 with respect to the vehicle 1, sensor information indicating the attitude of the surrounding situation sensor 12, and vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate, to perform additional learning of the model used by the recognition processing unit 3B to recognize the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12. By using the model for which additional learning has been performed by the server SV, the recognition processing unit 3B can surely recognize the recognition target RT based on the detection results of the surrounding situation sensor 12, which were impossible to recognize the recognition target RT before the additional learning of the model, sensor information indicating the mounting position of the surrounding situation sensor 12 with respect to the vehicle 1, sensor information indicating the attitude of the surrounding situation sensor 12, and vehicle information indicating the state of the vehicle 1 such as vehicle speed and yaw rate (that is, without relying on the detection results of the surrounding situation sensor 11).

[0029] FIG. 5 is a flowchart for explaining an example of the processing executed by the processor 173 of the data collection device 17 according to the first embodiment. In the example shown in FIG. 5, in step S10, the acquisition unit 3A acquires the detection results of the surrounding situation sensor 11 from the surrounding situation sensor 11. In step S11, the acquisition unit 3A acquires the detection results of the surrounding situation sensor 12 from the surrounding situation sensor 12. In step S12, the recognition processing unit 3B recognizes (specifically, attempts to recognize) a recognition target RT included in the detection area AR1 of the peripheral situation sensor 11 based on the detection result of the peripheral situation sensor 11 acquired in step S10. In step S13, the recognition processing unit 3B recognizes (specifically, attempts to recognize) a recognition target RT included in the detection area AR2 of the peripheral situation sensor 12 based on the detection result of the peripheral situation sensor 12 acquired in step S11.

[0030] In step S14, for example, the upload processing unit 3C determines whether the recognition processing unit 3B was able to recognize the recognition target RT based on the detection result of the peripheral situation sensor 11 in step S12. If YES, the process proceeds to step S15; if NO, the process proceeds to step S17. In step S15, for example, the upload processing unit 3C determines whether the recognition processing unit 3B was able to recognize the recognition target RT based on the detection result of the peripheral situation sensor 12 in step S13. If YES, the process shown in FIG. 5 ends; if NO, the process proceeds to step S16. In step S16, the upload processing unit 3C executes a process of uploading the detection result of the peripheral situation sensor 12 acquired in step S11 to the server SV.

[0031] In step S17, for example, the upload processing unit 3C determines whether the recognition processing unit 3B was able to recognize the recognition target RT based on the detection result of the peripheral situation sensor 12 in step S13. If YES, the process proceeds to step S18; if NO, the process shown in FIG. 5 ends. In step S18, the upload processing unit 3C executes a process of uploading the detection result of the peripheral situation sensor 11 acquired in step S10 to the server SV.

[0032] <Second Embodiment> The vehicle 1 to which the data collection device 17 of the second embodiment is applied is configured in the same manner as the vehicle 1 to which the data collection device 17 of the first embodiment described above is applied, except for the points described later.

[0033] As described above, in the example shown in FIG. 1 (an example of the vehicle 1 to which the data collection device 17 of the first embodiment is applied), the peripheral situation sensors 11 and 12 are not equipped with AI (artificial intelligence), and the processor 173 recognizes the recognition target RT included in the detection area AR1 of the peripheral situation sensor 11 based on the detection result of the peripheral situation sensor 11, and recognizes the recognition target RT included in the detection area AR2 of the peripheral situation sensor 12 based on the detection result of the peripheral situation sensor 12. It has a function as a recognition processing unit 3B. On the other hand, in an example of the vehicle 1 to which the data collection device 17 of the second embodiment is applied, the peripheral situation sensor 11 has a function of recognizing the recognition target RT included in the detection area AR1 of the peripheral situation sensor 11 based on the detection result of the peripheral situation sensor 11, and the peripheral situation sensor 12 has a function of recognizing the recognition target RT included in the detection area AR2 of the peripheral situation sensor 12 based on the detection result of the peripheral situation sensor 12.

[0034] In an example of the vehicle 1 to which the data collection device 17 of the second embodiment is applied, the acquisition unit 3A acquires from the peripheral situation sensor 11 the detection result of the peripheral situation sensor 11 and the recognition result of the recognition target RT included in the detection area AR1 of the peripheral situation sensor 11, and acquires from the peripheral situation sensor 12 the detection result of the peripheral situation sensor 12 and the recognition result of the recognition target RT included in the detection area AR2 of the peripheral situation sensor 12.

[0035] In an example of the vehicle 1 to which the data collection device 17 of the second embodiment is applied, when the peripheral situation sensor 12 can recognize the recognition target RT based on the detection result of the peripheral situation sensor 12 and the peripheral situation sensor 11 cannot recognize the recognition target RT based on the detection result of the peripheral situation sensor 11, the upload processing unit 3C executes a process of uploading the detection result of the peripheral situation sensor 11 to the server SV. The server SV not only collects the detection results of the surrounding situation sensor 11 from the vehicle 1, but also collects from the vehicle 1 the attributes etc. of the recognition target RT recognized by the surrounding situation sensor 12 based on the detection results of the surrounding situation sensor 12 (things used as labels in additional learning of the model). Further, the server SV uses the detection results of the surrounding situation sensor 11 collected from the vehicle 1 and the attributes etc. of the recognition target RT to perform additional learning of the model used by the surrounding situation sensor 11 to recognize the recognition target RT included in the detection area AR1 of the surrounding situation sensor 11. By using the model for which additional learning has been performed by the server SV, the surrounding situation sensor 11 can recognize the recognition target RT based only on the detection results of the surrounding situation sensor 11 which were impossible to recognize the recognition target RT before the additional learning of the model (that is, without relying on the detection results of the surrounding situation sensor 12).

[0036] In an example of the vehicle 1 to which the data collection device 17 of the second embodiment is applied, when the surrounding situation sensor 11 can recognize the recognition target RT based on the detection results of the surrounding situation sensor 11 and the surrounding situation sensor 12 cannot recognize the recognition target RT based on the detection results of the surrounding situation sensor 12, the upload processing unit 3C executes the process of uploading the detection results of the surrounding situation sensor 12 to the server SV. The server SV not only collects the detection results of the surrounding situation sensor 12 from the vehicle 1, but also collects from the vehicle 1 the attributes etc. of the recognition target RT recognized by the surrounding situation sensor 11 based on the detection results of the surrounding situation sensor 11 (things used as labels in additional learning of the model). Further, the server SV uses the detection results of the surrounding situation sensor 12 collected from the vehicle 1 and the attributes etc. of the recognition target RT to perform additional learning of the model used by the surrounding situation sensor 12 to recognize the recognition target RT included in the detection area AR2 of the surrounding situation sensor 12. The peripheral situation sensor 12 can recognize the recognition target RT based only on the detection result of the peripheral situation sensor 12 that could not recognize the recognition target RT before additional learning of the model, by using the model that has been additionally learned by the server SV (that is, without relying on the detection result of the peripheral situation sensor 11).

[0037] As described above, the embodiments of the data collection device, data collection method, and program of the present disclosure have been described with reference to the drawings. However, the data collection device, data collection method, and program of the present disclosure are not limited to the above-described embodiments, and appropriate modifications can be made without departing from the spirit of the present disclosure. The configurations of the examples of the above-described embodiments may be combined as appropriate. In each example of the above-described embodiments, the processing performed in the data collection device 17 has been described as software processing performed by executing a program. However, the processing performed in the data collection device 17 may be processing performed by hardware. Alternatively, the processing performed in the data collection device 17 may be processing that combines both software and hardware. Further, the program stored in the memory 172 of the data collection device 17 (the program that realizes the functions of the processor 173 of the data collection device 17) may be recorded and provided, distributed, etc. on a computer-readable storage medium such as a semiconductor memory, a magnetic recording medium, an optical recording medium, etc.

Description of Reference Numerals

[0038] 1... Vehicle, 11... Peripheral situation sensor, 12... Peripheral situation sensor, 13... Vehicle state sensor, 14... HMI, 15... Communication device, 16... Vehicle control device, 16A... Steering actuator, 16B... Brake actuator, 16C... Drive actuator, 17... Data collection device, 171... Communication interface, 172... Memory, 173... Processor, 3A... Acquisition unit, 3B... Recognition processing unit, 3C... Upload processing unit

Claims

1. An acquisition unit that acquires at least detection results from a first peripheral situation sensor that detects the peripheral situation of a vehicle and has a first detection area, and from a second peripheral situation sensor that detects the peripheral situation of the vehicle and has a second detection area that partially overlaps with the first detection area; An upload processing unit that, when a recognition target located within the first detection area and within the second detection area cannot be recognized based on the detection result of the first peripheral situation sensor but can be recognized based on the detection result of the second peripheral situation sensor, executes a process of uploading the detection result of the first peripheral situation sensor to a server. A data collection device comprising:

2. The upload processing unit according to claim 1, wherein the upload processing unit executes a process of uploading to the server a time-series detection result of the first peripheral situation sensor including the detection result of the first peripheral situation sensor at a time when the recognition target cannot be recognized. The data collection device described.

3. The acquisition unit acquires sensor information indicating the mounting position and orientation of the first peripheral situation sensor with respect to the vehicle and vehicle information indicating the vehicle speed and yaw rate of the vehicle, The upload processing unit according to claim 1, wherein the upload processing unit executes a process of uploading the sensor information and the vehicle information acquired by the acquisition unit to the server. The data collection device described.

4. A data collection step in which a data collection device acquires at least detection results from a first peripheral situation sensor that detects the peripheral situation of a vehicle and has a first detection area, and from a second peripheral situation sensor that detects the peripheral situation of the vehicle and has a second detection area that partially overlaps with the first detection area; An upload processing step in which, when a recognition target located within the first detection area and within the second detection area cannot be recognized based on the detection result of the first peripheral situation sensor but can be recognized based on the detection result of the second peripheral situation sensor, the data collection device executes a process of uploading the detection result of the first peripheral situation sensor to a server. A data collection method comprising:

5. To the processor, An acquisition step of acquiring at least detection results from a first peripheral situation sensor that detects the peripheral situation of a vehicle and has a first detection area, and acquiring at least detection results from a second peripheral situation sensor that detects the peripheral situation of the vehicle and has a second detection area that partially overlaps with the first detection area; A program for executing an upload processing step of executing a process of uploading the detection result of the first peripheral situation sensor to a server when a recognition target located within the first detection area and within the second detection area cannot be recognized based on the detection result of the first peripheral situation sensor and can be recognized based on the detection result of the second peripheral situation sensor.

Images