JP6464992B2 - Operating data collection system - Google Patents

Description

  The present invention relates to a technique for collecting a large amount of operation data.

  NDS (Naturalistic) is an attempt to collect a large amount of driving data measured by in-vehicle sensors during normal natural driving and use it for analysis of driver characteristics, evaluation of driving risk, support for pre-reading driving, etc. Driving Study) is known (see Non-Patent Document 1).

  Unlike conventional probe car systems that collect only driving data for specific purposes, NDS limits general-purpose driving data for the entire driving, that is, the type of data so that it can be used for various applications. Without having to collect a lot of operation data under various circumstances.

S. Hallmark, D. McGehee, KM Bauer, JM Hutton, GA Davis, J. Hourdos, I. Chatterjee, T. Victor, J. Bargman, M. Dozza, H. Rootzen, JD Lee, C. Ahlstrom, O. Bagdadi, J. Engstrom, D. Zholud, and M. Ljung-Aust, "Initial Analyses from the SHRP2 Naturalistic Driving Study: Addressing Driver Performance and Behavior in Traffic Safety,” SHRP 2 Research Reports, SHRP 2 Safety Project S08, Transportation Research Board of the National Academies, 2013

  By the way, in a conventional approach in NDS, a data measurement / storage device is installed in a vehicle of a measurement cooperator, and after recording raw driving data, data is acquired by collecting the measurement / storage device. ing.

  However, in the future, when considering natural driving data that does not assume a specific situation for general vehicles, it is not possible to install and collect measurement and storage devices for all vehicles. Realistic.

Instead of collecting the measuring / accumulating device, it is conceivable that driving data measured while the vehicle is traveling is transmitted wirelessly and stored in an information center connected via a communication network.
However, an attempt by NDS that does not limit the type of operation data to be collected has a problem in that the amount of data collected by one vehicle becomes enormous and burdens the communication network.

  The present invention has been made in view of these problems, and an object thereof is to provide a technique for efficiently collecting general-purpose operation data relating to the entire operation.

The operation data collection system of the present invention includes an in-vehicle system and an information center capable of wireless communication with each other.
The in-vehicle system includes a data collection unit, a model storage unit, and a data discretization unit. The data collection unit repeatedly collects a plurality of measurement values representing the state of the host vehicle equipped with the in-vehicle system. Each model storage unit represents a discretization rule that is a rule used when discretizing the driving data into a plurality of partial data using the time series of measurement values collected by the data collecting unit as driving data. Model information that is common to each other. The data discretization unit discretizes the operation data into a plurality of partial data using model information.

The information center includes a data storage unit. The data accumulating unit accumulates discretized operation data, which is data discretized by the data discretizing unit, in the server.
The in-vehicle system transmits discretized operation data to the information center through communication.

  That is, in the present invention, the in-vehicle system converts the measured operation data into discretized operation data and transmits the discretized operation data to the information center, and the information center stores the discretized operation data in the server. For this reason, the amount of communication between the in-vehicle system and the information center and the storage capacity of the server can be suppressed, and general-purpose operation data can be efficiently collected.

  In addition, since each in-vehicle system uses common model information when discretizing operation data, the generated discretized data is structured according to the same standard, and data integration and analysis are performed. It can be done easily.

  In addition, the code | symbol in the parenthesis described in the claim shows the correspondence with the specific means as described in embodiment mentioned later as one aspect, Comprising: The technical scope of this invention is limited is not.

It is a block diagram which shows the structure of the driving | operation data collection system of 1st Embodiment. It is a block diagram which shows the structure of the vehicle-mounted system of 1st Embodiment. It is explanatory drawing which shows the outline | summary of a discretization process. It is a block diagram which shows the structure of the information center of 1st Embodiment. It is a block diagram which shows the structure of the driving | operation data collection system of 2nd Embodiment. It is a block diagram which shows the structure of the vehicle-mounted system of 2nd Embodiment. It is a block diagram which shows the structure of the information center of 2nd Embodiment. It is a block diagram which shows the structure of the driving | operation data collection system of 3rd Embodiment. It is a block diagram which shows the structure of the distributed processing apparatus of 3rd Embodiment.

Embodiments to which the present invention is applied will be described below with reference to the drawings.
[First Embodiment]
[Constitution]
The driving data collection system 1 shown in FIG. 1 includes an in-vehicle system 10 mounted on each vehicle and an information center 30 capable of wireless communication with the in-vehicle system 10.

[In-vehicle system]
As shown in FIG. 2, the in-vehicle system 10 includes a vehicle behavior sensor group 11, an additional information sensor group 12, an operation data generation device 13, and a communication device 14.

  The vehicle behavior sensor group 11 includes various devices for detecting information related to the behavior of the vehicle and driving operations that affect the behavior of the vehicle. Examples of the detection target of the vehicle behavior sensor group 11 include an operation amount of an accelerator pedal and a brake pedal, a steering angle, a vehicle speed, and a vehicle acceleration.

The additional information sensor group 12 includes various devices for detecting information related to the situation where the host vehicle is placed. The detection targets of the additional information sensor group 12 include, for example, position information of the own vehicle, image information obtained by imaging the surroundings and the interior of the own vehicle, destinations for navigation and automatic driving, route information to the destination, weather and road conditions Ambient environment information.

  The communication device 14 realizes communication with the information center 30, and may be configured to communicate directly with the information center 30, or may be configured to communicate via a public communication network or the like. Also good.

  The operation data generation device 13 includes a data collection unit 21, an additional information collection unit 22, a data discretization unit 23, a data addition unit 24, a data transmission unit 25, a model reception unit 26, a model update unit 27, a discretization model DB (database). ) 28. The operation data generation device 13 is mainly configured by a microcomputer including a CPU, a ROM, and a RAM, and the discretization model DB 28 is provided on the RAM. The other units 21 to 27 are realized by processing executed by the CPU according to a program stored in the ROM.

The data collection unit 21 repeatedly acquires the output of the vehicle behavior sensor group 11.
The additional information collection unit 22 repeatedly acquires additional information that is an output of the additional information sensor group 12.

  The data discretization unit 23 uses the series of data acquired by the data collection unit 21 as operation data, and uses the operation data as model data stored in the discretization model DB 28, and each of the operation discretions represents a certain operation scene. Discretize into multiple data series to represent.

  Here, a double articulation analyzer (DAA) that performs segmentation by an unsupervised driving scene division method using a double segment structure is used. That is, as shown in FIG. 3, first, a symbol string is generated by sequentially assigning symbols (C1, C2,...) To a multidimensional measurement value pattern that repeatedly appears in the operation data series, and the symbol string Are segmented into a plurality of partial series (W1, W2,...) Meaning a predetermined driving scene. This partial series is hereinafter referred to as discretized operation data.

  Specifically, in DAA, first, clusters representing various vehicle states grasped from driving data in a multidimensional space representing a range of driving data and transition probabilities between the clusters are defined in advance. Using these pieces of information, it is statistically processed to which cluster the operation data acquired from the data collection unit 21 belongs. As a result, the series of driving data is divided for each state of the vehicle (that is, for each cluster) as a division unit. By associating each cluster with a symbol (C1, C2,...) In advance, the time series of operation data is converted into a symbol string indicating which cluster it belongs to. In order to generate such a symbol string, for example, a hierarchical Dirichlet process hidden Markov model (HDP-HMM) which is one of models expressed by a hidden state and a stochastic transition between the states may be used. it can.

  Next, DAA uses a Nested Pitman-Yor Language Model (NPYLM), which is an example of an unsupervised chunking technique for discrete character strings using statistical information, to generate a predetermined operation scene. Segment into meaningful subsequences. At this time, the generation probability of the entire symbol string composed of the partial series is maximized. Thereby, it becomes possible to segment the driving data into driving scenes. However, transition probabilities between partial sequences and partial sequence generation probabilities are generated in advance by learning.

  That is, when such DAA is used, the cluster and the transition probability between the clusters used for generation of the symbol sequence, the transition probability between the partial sequences used when segmenting the symbol sequence, and the generation probability of the partial sequence are Used as model information.

  Note that DAA to which HDP-HMM or NPYLM is applied is, for example, a known one described in JP 2013-250663 A or the like. In addition, the method used for symbol string generation and symbol string segmentation is not limited to HDP-HMM or NYLM, and other techniques may be used.

  Returning to FIG. 2, the data adding unit 24 uses the data generated by the data discretizing unit 23 (here, a series of segmented symbol strings) as the discretized operation data. For the data, the additional information acquired by the additional information collecting unit 22 at the same timing as the data is added to the data. Specifically, an average value of the position information acquired at the same timing as the discretized driving data, an image feature amount acquired from the image information, destination and route information, surrounding traffic conditions, driving time, and the like can be added.

  Further, the data to be added is not limited to the information acquired by the additional information collection unit 22 but may be information using the information acquired by the data collection unit 21 as additional information. Specifically, an accelerator pedal or brake pedal operation amount, a steering angle, an average value of vehicle speed or vehicle acceleration, a histogram, or the like acquired at the same timing as the discrete operation data can be added.

The data transmission unit 25 transmits the discrete operation data to which the additional information is added by the data addition unit 24 to the information center 30 via the communication device 14.
The model receiving unit 26 receives model information from the information center 30 via the communication device 14. The model updating unit 27 updates the stored contents of the discretized model DB 28 with the model information received by the model receiving unit 26.

[Information Center]
As shown in FIG. 4, the information center 30 includes a communication device 31, a data storage processing device 32, and an operation information DB 33.

  The communication device 31 realizes communication with the in-vehicle system 10, and may be configured to directly communicate with the in-vehicle system 10, or may be configured to communicate via a public communication network or the like. Also good.

The operation information DB 33 is composed of a large-capacity memory device such as a hard disk, and stores discretized operation data (hereinafter simply referred to as “discretized operation data”) to which additional information is added.
The data storage processing device 32 includes a data reception unit 41, a data storage unit 42, a model generation unit 43, and a model transmission unit 44. The data storage processing device 32 is mainly composed of a microcomputer comprising a CPU, a ROM, and a RAM. Each part 41-44 which comprises the data storage processing apparatus 32 is implement | achieved by the process which CPU performs according to the program memorize | stored in ROM.

The data receiver 41 receives discretized operation data from the in-vehicle system 10 via the communication device 31.
The data storage unit 42 stores the discretized operation data received by the data reception unit 31 in the operation information DB 33. At that time, additional information is used to classify and store each similar situation.

Each time a certain amount of information is accumulated in the driving information DB 33, the model generation unit 43 generates model information used for processing in the data discretization unit 23 based on the accumulated information.
The model transmission unit 44 distributes the model information generated by the model generation unit 43 to each of the in-vehicle systems 10 via the communication device 31. Thereby, the model information memorize | stored in each vehicle is updated.

[effect]
As described above, in the operation data collection system 1, the operation data detected by each vehicle is discretized by the in-vehicle system 10 and the discretized operation data is transmitted to the information center 30.

  That is, since the discretized operation data is compressed in comparison with the operation data before the discretization, transmission to the information center 30 using wireless communication can be realized without difficulty, and general-purpose including various information. Efficient operation data can be collected efficiently.

  The in-vehicle system 10 of each vehicle uses common model information when discretizing the driving data. As a result, the discrete operation data generated by each vehicle is structured according to the same standard, so that the discrete operation data collected from each vehicle can be easily integrated and analyzed.

  In addition, since the discretization of the driving data is performed based on information on the behavior of the vehicle and the driving operation that affects the behavior of the vehicle, the compression can be performed with an appropriate granularity according to the content of the driving. For example, driving sections with little change in vehicle behavior are discretized as a cluster, and driving sections with many changes in vehicle behavior are discretized finely.

  Further, since the additional information is associated with the discretized operation data, the data regarding the desired situation can be easily extracted by using the additional information, and the usability of the discretized operation data is improved. be able to. For example, it is possible to easily extract the discretized operation data that has been traveling in a specific time zone in a certain area. Furthermore, by using the associated additional information, it is possible to extract a distribution of driving behavior that often occurs under such specific conditions. By analyzing such information, it is possible to extract dangerous driving or near-miss under specific conditions, alert the driver, or improve road equipment.

[Second Embodiment]
Since the basic configuration of the second embodiment is the same as that of the first embodiment, the description of the common configuration will be omitted, and the description will focus on the differences.

[Constitution]
The driving data collection system 1a shown in FIG. 5 includes an in-vehicle system 10a mounted on each vehicle and an information center 30a capable of communicating with the in-vehicle system 10a.

[In-vehicle system]
As illustrated in FIG. 6, the in-vehicle system 10 a includes a vehicle behavior sensor group 11, an additional information sensor group 12, an operation data generation device 13 a, and a communication device 14. That is, it has the same configuration as the in-vehicle system 10 of the first embodiment except for the operation data generation device 13a.

  The operation data generation device 13a is different from the operation data generation device 13 described in FIG. 2 in that the model reception unit 26 and the model update unit 27 are omitted, instead of the data discretization unit 23 and the discretization model DB 28. Is different in that a lower discretization unit 23a and a lower discretization model DB 28a are provided.

  Of the processes executed by the data discretization unit 23 described in the first embodiment, the lower discretization unit 23a performs only the process of generating the first half symbol string. That is, the lower discretization unit 23a outputs the symbol string shown in FIG. 3 to the data addition unit 24 as discretized operation data.

  The lower discretization model DB 28a stores clusters and transition probabilities between clusters used when the lower discretization unit 23a generates symbol strings as lower model information. As the lower model information, a fixed value set mainly by the characteristics of the vehicle behavior sensor group 11 is used.

[Information Center]
As illustrated in FIG. 7, the information center 30 a includes a communication device 31, a data storage processing device 32 a, and an operation information DB 33. That is, it has the same configuration as the information center 30 of the first embodiment except for the data storage processing device 32a.

  The data storage processing device 32a is different from the data storage processing device 32 described in FIG. 4 in that the model transmission unit 44 is omitted, and an upper discretization unit 45, an upper discretization model DB 46, and a model update unit 47 are provided. Differences are added.

  The high-order discretization unit 45 performs only the process of segmenting the latter half symbol sequence among the processes executed by the data discretization unit 23 described in the first embodiment. That is, the discretization operation data transmitted from the in-vehicle system 10a corresponds to the symbol string shown in FIG. 3, and the data discretization unit 23 of the first embodiment outputs it by segmenting it. Discrete operation data equivalent to that is obtained and supplied to the data storage unit 42.

The upper discretization model DB 46 stores, as upper model information, the transition probability between partial sequences and the generation probability of the partial sequence used when the upper discretization unit 45 segments the symbol string.
The model update unit 47 updates the stored contents of the high-order discretization model DB 46 with the high-order model information generated by the model generation unit 43.

[effect]
As described above, according to the driving data collection system 1a, not only can the same effect as the driving data collection system 1 of the first embodiment be obtained, but also the processing load on the in-vehicle system 10a mounted on each vehicle. Can be reduced. Further, by dividing the discretized model into an upper model to be updated and a lower model that is a fixed value, the driving data collection system 1 of the first embodiment is not transmitted from the information center 30a to the in-vehicle system 10a. Since the same effect can be obtained, the amount of communication can be reduced and the cost of the in-vehicle system can be reduced.

[Third Embodiment]
Since the basic configuration of the third embodiment is the same as that of the second embodiment, the description of the common configuration will be omitted, and the description will focus on the differences.

[Constitution]
The driving data collection system 1b shown in FIG. 8 includes an in-vehicle system 10a mounted on each vehicle, a distributed processing device 50 capable of communicating with the in-vehicle system 10a, and an information center 30 capable of communicating with the distributed processing device 50. With. That is, the in-vehicle system 10a is the same as that of the second embodiment, the information center 30 is the same as that of the first embodiment, and a distributed processing device 50 is newly added.

[Distributed processing equipment]
As illustrated in FIG. 9, the distributed processing device 50 includes a communication device 51, a distributed processing unit 52, and a communication device 53.

The communication device 51 realizes communication with the in-vehicle system 10a, and the communication device 53 realizes communication with the information center 30. Each of the communication devices 51 and 53 may be configured to directly communicate with a communication partner or may be configured to communicate via a public communication network.

  The distributed processing unit 52 includes a data receiving unit 61, a higher discretization unit 62, a data transmission unit 63, a model receiving unit 64, a model update unit 65, and a higher discretization model DB 66. The distributed processing unit 52 is configured around a microcomputer including a CPU, a ROM, and a RAM. The upper discretization model DB 66 is configured on a RAM. The other units 61 to 65 are realized by processing executed by the CPU according to programs stored in the ROM.

The data receiving unit 61 receives discretized operation data (symbol string) from the in-vehicle system 10 a via the communication device 51.
The upper discretization unit 62 and the upper discretization model DB 66 are the same as the upper discretization unit 45 and the upper discretization model DB 46 (see FIG. 7) of the second embodiment.

The data transmission unit 63 transmits the discretization operation data (a symbol string segmented) output from the higher discretization unit 62 to the information center 30 via the communication device 53.
The model receiving unit 64 receives the upper model information from the information center 30 via the communication device 53. The model update unit 65 updates the stored content of the higher discretization model DB 66 with the higher model information received by the model reception unit 64.

  Similar to the information center 30, the distributed processing device 50 may be installed outside the vehicle, or may be mounted on some vehicles having a sufficient processing capacity. Further, both of them may be mixed.

[effect]
As described above, according to the operation data collection system 1b, the same effects as those of the operation data collection systems 1 and 1a of the first and second embodiments can be obtained.

  Furthermore, according to the operation data collection system 1b, by providing the distributed processing device 50, the concentration of processing load on the information center 30 can be suppressed as compared with the case of the second embodiment.

[Other Embodiments]
As mentioned above, although embodiment of this invention was described, this invention can take a various form, without being limited to the said embodiment.

  (1) In the above embodiment, DAA is used as a method for discretizing operation data. However, the present invention is not limited to this, and a known method that uses some model information when discretizing is used. it can.

  (2) The functions of one component in the above embodiment may be distributed to a plurality of components, or the functions of a plurality of components may be integrated into one component. In addition, at least a part of the configuration of the above embodiment may be replaced with a known configuration having a similar function. Moreover, you may abbreviate | omit a part of structure of the said embodiment. Further, at least a part of the configuration of the above embodiment may be added to or replaced with the configuration of the other embodiment. In addition, all the aspects included in the technical idea specified only by the wording described in the claim are embodiment of this invention.

(3) In addition to the operation data collection system described above, a system including the operation data collection system as a constituent element, an in-vehicle system constituting the operation data collection system, an information center,
The present invention can also be realized in various forms such as a program for causing a computer to function as a distributed processing device, a medium on which the program is recorded, and an operation data collection method.

DESCRIPTION OF SYMBOLS 1,1a, 1b ... Driving data collection system 10, 10a ... In-vehicle system 11 ... Vehicle behavior sensor group 12 ... Additional information sensor group 13, 13a ... Driving data generation device 14, 31, 51, 53 ... Communication device 21 ... Data collection Unit 22: Additional information collection unit 23 ... Data discretization unit 23a ... Lower discretization unit 24 ... Data addition unit 25, 63 ... Data transmission unit 26, 64 ... Model reception unit 27, 47, 65 ... Model update unit 28 ... Discrete Modeling model DB 28a ... Lower discrete model DB 30, 30a ... Information center 31, 41, 61 ... Data receiving unit 32, 32a ... Data storage processing device 33 ... Operation information DB 42 ... Data storing unit 43 ... Model generating unit 44 ... Model transmission unit 45, 62 ... upper discretization unit 46, 66 ... upper discretization model DB 50 ... distributed processing device 52 ... Distributed processing part

Claims (2)

In a driving data collection system (1) comprising an in-vehicle system (10) mounted on a vehicle and an information center (30) capable of communicating with the in-vehicle system,
The in-vehicle system is
A data collection unit (21) for repeatedly collecting a plurality of measurement values representing the state of the host vehicle equipped with the in-vehicle system;
A model common to each vehicle representing a discretization rule, which is a rule used when discretizing the driving data into a plurality of partial data using the time series of measurement values collected by the data collecting unit as driving data A model storage unit (28) for storing information;
A data discretization unit (23) for discretizing the operation data into a plurality of partial data using the model information;
A model update unit (27) for updating the storage contents of the model storage unit;
With
The information center
A data storage unit (42) for storing discrete operation data, which is data discretized by the data discretization unit, in a server ;
A model generation unit (43) for generating the model information based on data stored in the server;
With
The in-vehicle system transmits the discretized operation data to the information center by communication ,
The information center distributes the model information to each of the in-vehicle systems by communication,
The model update unit of the in-vehicle system updates the storage content of the model storage unit with the model information generated by the model generation unit . The in-vehicle system is
An additional information acquisition unit (22) that acquires additional information that is information relating to a situation where the host vehicle on which the in-vehicle system is mounted is placed;
A data adding unit (24) for adding the additional information acquired by the additional information acquiring unit to data to be transmitted to the information center;
The operation data collection system according to claim 1 , further comprising:

Images