This application is based on and claims priority under 35 USC 119 from Japanese Patent Application No. 2018-195244 filed on Oct. 16, 2018.
An embodiment of the present disclosure relates to a data collection apparatus, a data collection system, and a data collection method.
Data collection apparatus are known that collect road information from vehicular devices installed in respective vehicles. These data collection apparatus collect road information at desired positions by selecting vehicles as road information collection targets on the basis of pieces of position information of the respective vehicles (refer to JP-A-2018-055581, for example).
However, the above-described conventional technique has room for improvement in allowing a user to recognize the certainty of information collection under a specified collection condition.
More specifically, in the above-described conventional technique, a user can connect a terminal to a center apparatus and specify a desired collection condition such as an upload condition through a user interface on the center apparatus. This contributes to, among other things, flexible information collection that is suitable for a situation and reduction of the amount of communication. However, on the other hand, it has problems that the user has difficulty recognizing at what probability information is collected under a specified collection condition and recognizing, on the basis of that probability, what number of pieces of information can be collected in a desired collection period.
One aspect of an embodiment has been made in view of the above, and hence its object is to provide a data collection apparatus, a data collection system, and a data collection method that allows a user to recognize the certainty of information collection under a specified collection condition.
A data collection apparatus according to the one aspect of the embodiment is equipped with a collection unit, a reception unit, an analysis unit, and a providing unit. The collection unit collects data relating to respective vehicles from vehicular devices installed in the respective vehicles. The reception unit receives a data collection condition specified by the user. The analysis unit calculates a probability of occurrence of an event for collection of data specified in the collection condition on the basis of past data collection results. The providing unit provides the user with information that is based on the probability of occurrence calculated by the analysis unit.
The one aspect of an embodiment allows a user to recognize the certainty of information collection under a specified collection condition.
A data collection apparatus, a data collection system, and a data collection method according to an embodiment of the present disclosure will be hereinafter described in detail with reference to the accompanying drawings. The present invention is not limited to the embodiment described below.
In the following description, the term “frequency” means a rate that is extracted statistically from past results or at which a certain event occurred. And the term “probability” means a rate at which a certain event will occur in the future.
First, the data collection method according to the embodiment will be outlined with reference to
As shown in
The data collection apparatus 10, which is implemented as, for example, a cloud server that provides a cloud service over a network N such as the Internet or a cellphone network, receives a vehicle data collection request from a data user, collects vehicle data from the respective vehicular devices 100 on the basis of the received collection request, and provides collected vehicle data to the data user.
Each vehicular device 100, which is, for example, a drive recorder having a camera, various sensors such as an acceleration sensor and a GPS (global positioning system) sensor, a storage device, a microcomputer, etc., picks up vehicle data that comply with a collection request received by the data collection apparatus 10.
Each vehicular device 100 uploads picked-up vehicle data to the data collection apparatus 10 as necessary. Using a drive recorder also as vehicular device 100 in this manner enables effective use of vehicular components installed in each vehicle V. Alternatively, a vehicular device 100 and a driver recorder may be used as separate devices.
The user terminal 200, which is a terminal that is used by the data user, is, for example, a notebook PC (personal computer), a desk-top PC, a tablet terminal, a PDA (personal digital assistant), a smartphone, or a wearable device such as glasses-type or wrist-watch-type information processing terminal.
For example, the data user is a developer who develops an autonomous drive technique on the basis of vehicle data provided by the data collection apparatus 10. The data collection apparatus 10 provides such a data user with user interface (hereinafter abbreviated as “UI”) pictures on the user terminal 200.
As shown in
As shown in
“Target vehicle” is identification information of a vehicle V as a collection target. “Data type” is a type of collection target data such as an acceleration position. “Collection trigger condition” is a condition as a trigger of collection and is, for example, a condition that the vehicle speed has exceeded a prescribed speed.
For example, in the example of
Each vehicular device 100 uploads vehicle data picked up in the vehicle V to the data collection apparatus 10 with proper timing, and the data collection apparatus 10 stores the received vehicle data. The data user browses or downloads vehicle data accumulated in the data collection apparatus 10, for example, via a UI picture mentioned above.
To describe the data collection method more specifically, a series of steps according to which vehicle data is provided to the data user in the data collection system 1 will be described with reference to
In this operation, the data collection apparatus 10 generates data for generation of tag data T having characteristics of index data to be attached to real data R to be collected and to be used for searching for or recognizing its outline. Data for generation of such tag data T is generated by manipulations by the data user using a program and data stored in the user terminal 200 or the data collection apparatus 10.
The specified collection condition and the generated data for generation of tag data T are stored in the data collection apparatus 10 and delivered to vehicles V as data collection targets and stored in their vehicular devices 100.
Each vehicular device 100 monitors output data of various sensors. If an event that satisfies the stored collection condition has occurred, the vehicular device 100 stores its real data R in a storage device. Furthermore, each vehicular device 100 generates tag data T corresponding to the real data R on the basis of the stored data for generation of tag data T and the real data, and stores the generated tag data T. Each vehicular device 100 uploads the generated tag data T to the data collection apparatus 10 and the data collection apparatus 10 stores the received tag data T. At this time, the real data R is not uploaded to the data collection apparatus 10.
When the data user connects the user terminal 200 to the data collection apparatus 10 to check a data collection status or collect real data R, meta information that is based on tag data T collected by the data collection apparatus 10 is displayed on the user terminal 200. At the same time, UI picture for allowing manipulations for collecting real data R corresponding to each piece of tag data T
When as shown in
Then, as shown in
From the viewpoint of the data capacity of each vehicular device 100, it is preferable that real data R and corresponding tag data R that have been uploaded to the data collection apparatus 10 be deleted from the vehicular device 100 after their uploading to the data collection apparatus 10.
It is preferable that tag data T not be data obtained by simply extracting part of real data R but be data that is converted into meta information to such an extent as to allow the data user to recognize an outline of the real data R and to judge whether the real data R is necessary when referring to the tag data T.
Incidentally, as described above, the data user can specify collection condition parameters in a desired manner. However, this means that it is difficult for the data user to recognize at what probability information is collected certainly under a specified collection condition and to recognize, on the basis of that probability, what number of pieces of information can be collected in a desired collection period.
That is, as shown in
In view of the above, in the data collection method according to the embodiment, when receiving a desired (specified) collection condition and a desired (specified) number of pieces of collected information from the data user, the data collection apparatus 10 extracts a frequency of occurrence of an event as a trigger of data collection on the basis of past similar collection conditions (hereinafter referred to as “similar conditions”). Furthermore, the data collection apparatus 10 calculates a probability of occurrence of the event under the collection condition specified by the data user on the basis of the extracted frequency, generates probability information relating to the calculated probability, and informs the user terminal 200 of the generated probability information.
More specifically, as shown in
At step S4, the data collection apparatus 10 generates probability information relating to the calculated probability. As shown in
At step S5, the data collection apparatus 10 informs the user terminal 200 of the generated probability information. If the collection will not be finished by a desired date specified by the data user or there is no similar condition, at step S6 the data collection apparatus 10 provides the data user with guidance on changing of the condition.
More specifically, as shown in
As shown in the pattern P1 in
The data collection apparatus 10 may search, in advance, past results etc. for similar conditions under which the collection will be completed by the desired date and time specified by the data user and display packages of recommendable condition formulae in the form of a list as a pattern P2. If the data user selects a desired package from the list, the collection condition can be changed easily so as to reflect the contents of the selected package.
As described above, in the data collection method according to the embodiment, when receiving, from the data user, a desired collection condition and a desired number of data to be collected specified by the data user, the data collection apparatus 10 extracts a frequency of occurrence of events to become triggers for data collection on the basis of past similar conditions. Furthermore, the data collection apparatus 10 calculates a probability of occurrence of such events under the collection condition specified by the data user, generates probability information relating to the calculated probability, and informs the user terminal 200 of it.
As a result, the data collection method according to the embodiment allows the data user to recognize the certainty of information collection under the specified collection condition. The configuration of the data collection system 1 according to the embodiment will be hereinafter described in more detail.
In other words, the constituent elements shown in
In the following description that will be made with reference to
As shown in
The data collection apparatus 10 will be described first, which is equipped with a communication unit 11, a storage unit 12, and a control unit 13.
For example, the communication unit 11 is implemented as an NIC (network interface card). Connected to a network N by wire or wirelessly, the communication unit 11 exchanges information with the vehicular devices 100 and the user terminal 200 over the network N.
The storage unit 12 is implemented as a semiconductor memory device such as RAM (random access memory) or a flash memory or a storage device such as a hard disk drive or an optical disc. In the example of
Collection conditions that have been specified by the user terminal 200 and received by a reception unit 13b (described later) are accumulated in the collection condition information DB 12a. That is, the collection condition information DB 12a contains past results of collection conditions.
Collection data that have been collected from the vehicular devices 100 by a collection unit 13f (described later) are accumulated in the collection data DB 12b. That is, the collection data DB 12b contains tag data T and real data R (described above).
The control unit 13 is a controller and is implemented in such a manner that a CPU (central processing unit), an MPU (microprocessing unit), or the like runs various programs stored in an internal storage device of the data collection apparatus 10 using a RAM as a working area. The control unit 13 can be implemented by an integrated circuit such as an ASIC (application-specific integrated circuit) or an FPGA (field programmable gate array).
The control unit 13 is equipped with a UI providing unit 13a, the reception unit 13b, an analysis unit 13c, a guidance generation unit 13d, a distribution unit 13e, and the collection unit 13f and performs or realizes information processing functions and workings described later.
The UI providing unit 13a generates a collection condition setting picture and various UI pictures relating to data collection and provides them to the user terminal 200 via the communication unit 11. Specific examples of the UI pictures will be described later with reference to
The guidance information contains probability information relating to a probability calculated by the analysis unit 13c (described later) on the basis of a collection condition specified by the data user through the user terminal 200.
The reception unit 13b receives, via the communication unit 11, a desired number of data and a collection condition specified by the data user through the user terminal 200 and informs the analysis unit 13c of them. Furthermore, the reception unit 13b stores the collection condition specified by the data user in the collection condition information DB 12a.
When receiving, for example, a collection condition that has been changed by the data user so that the data collection will be completed by a desired date and time, the reception unit 13b updates a corresponding collection condition in the collection condition information DB 12a.
The analysis unit 13c analyzes past similar conditions on the basis of the desired number of data to be collected and the collection condition received by the reception unit 13b, the collection condition information DB 12a, and the collection data DB 12b. More specifically, the analysis unit 13c extracts past similar conditions that are similar to the collection condition received by the reception unit 13b from the collection condition information DB 12a. The term “similar conditions” include the same condition.
In doing so, the analysis unit 13c retrieves past similar conditions in descending order of similarity from the collection condition information DB 12a on the basis of priority ranks that are correlated with respective items of the collection condition. This will be described later with reference to
If a similar condition has been extracted, the analysis unit 13c extracts a frequency of occurrence of a trigger event of data collection under this similar condition on the basis of the collection data DB 12b. Then the analysis unit 13c calculates a probability of occurrence of the above event under the collection condition specified by the data user on the basis of the extracted frequency. The analysis unit 13c informs the guidance generation unit 13d of the calculated probability.
If no similar condition has been extracted, the analysis unit 13c informs the guidance generation unit 13d of an item of the collection condition that enables extraction of a similar condition(s) with a lower degree of alteration.
If a similar condition has been extracted but with the calculated probability the data collection will not be completed by the desired date and time, the analysis unit 13c searches for conditions under which the data collection will be completed by the desired date and time with lower degrees of alteration and informs the guidance generation unit 13d of a search result. In other words, if a scheduled date and time of completion of data collection under the similar condition is later than the desired date and time, the analysis unit 13c extracts other similar conditions under which data collection will be completed by the desired date and time in descending order of the degree of alteration and causes the UI providing unit 13a to inform the data user of them. The analysis unit 13c may list up plural such similar conditions and inform the data user of them in a form like the packages of recommendable condition formulae shown in
The guidance generation unit 13d generates guidance information relating to a collection condition on the basis of the analysis result communicated from the analysis unit 13c and informs the UI providing unit 13a of it (the guidance generation unit 13d does so also in a case of causing the data user to change the collection condition). For example, the guidance generation unit 13d generates guidance information so that probability information as shown in
The distribution unit 13e distributes a set collection condition stored in the collection condition information DB 12a to target vehicles V in, for example, a file form via the communication unit 11. The collection unit 13f collects, via the communication unit 11, vehicle data that are uploaded from the vehicular devices 100 and accumulates them in the collection data DB 12b.
Next, each vehicular device 100 will be described, which is equipped with a communication unit 101, a storage unit 102, and a control unit 103. As described above, various sensors 150 such as a camera, an acceleration sensor, and a GPS sensor are connected to the vehicular device 100.
Like the communication unit 11, the communication unit 101 is implemented as an NIC, for example. The communication unit 101 is connected to the network N wirelessly, and exchanges information with the data collection apparatus 10 over the network N. Furthermore, the communication unit 101 receives output data of the various sensors 150.
Like the storage unit 12, the storage unit 102 is implemented as a semiconductor memory device such as RAM or a flash memory or a storage device such as a hard disk drive or an optical disc. In the example of
The collection condition information 102a is information including a collection condition delivered from the data collection apparatus 10. The vehicle data information 102b information including vehicle data picked up by a pickup unit 103c (described later). Each piece of vehicle data includes tag data T and real data R (described above).
Like the controller 13, the control unit 103 is a controller and is implemented in such a manner that a CPU, an MPU, or the like runs various programs stored in an internal storage device of the vehicular device 100 using a RAM as a working area. The control unit 103 can be implemented by an integrated circuit such as an ASIC or an FPGA.
Equipped with an acquisition unit 103a, a detection unit 103b, the pickup unit 103c, and an upload unit 103d, the control unit 103 performs or realizes information processing functions and workings described later.
The acquisition unit 103a acquires a collection condition delivered from the data collection apparatus 10 and stores it as part of the collection condition information 102a. The detection unit 103b monitors output data of the various sensors 150 and detects occurrence of an event that is defined as a trigger in a collection condition.
For example, when detecting occurrence of an event that is defined as a trigger for pickup of vehicle data in a collection condition, the detection unit 103b causes the pickup unit 103c to pick up vehicle data. For another example, when detecting occurrence of an event that is defined as a trigger for uploading of vehicle data to the data collection apparatus 10 in a collection condition, the detection unit 103b causes the upload unit 103d to upload vehicle data.
When a trigger for pickup of vehicle data is detected by the detection unit 103b, the pickup unit 103c picks up vehicle data from output data of the various sensor 150 and stores them as part of the vehicle data information 102b. When a trigger for a stop of pickup of vehicle data is detected by the detection unit 103b, the pickup unit 103c stops pickup of vehicle data.
When a trigger for uploading of vehicle data is detected by the detection unit 103b, the upload unit 103d uploads vehicle data of the vehicle data information 102b to the data collection apparatus 10.
Next, a specific example collection condition setting picture that is provided as a UI picture to the user terminal 200 by the UI providing unit 13a will be described will be described with reference to
As shown in part M1 of
In this item, as shown in
As shown in
As shown in
An item “desired number of data to be collected” is an item for specifying of a desired number of data to be collected (mentioned above). When the data user specifies, for example, a desired number in part M2 shown in
An item “desired date and time of completion” is an item corresponding to a collection period to be specified by the data user. A collection period desired by the data user is set when the data user inputs a desired date and time of completion into part M3 shown in
An item “sampling period” is an item for specifying of a sampling period. An item “extraction of difference” makes it possible to specify whether to pick up only data having a difference bit length with respect to previous data. In this manner, the data user can specify a collection condition in a flexible manner according to a purpose.
Next, a feature that the analysis unit 13c searches the collection condition information DB 12a for past similar conditions in descending order of similarity will be described with reference to
As a premise of the description, it is assumed that a condition formula shown in
The analysis unit 13c gives weights (degrees of relation) to respective items that are specified for each of “parameter,” “symbol,” and “threshold value” of the trigger block. In the following description, where degrees of relation or priority ranks are denoted by alphabetical characters such as “A” to “F,” it is assumed that the degree of relation or the priority rank is higher on the “A” side. Where degrees of relation are denoted by symbols “Δ” and “x,” it is assumed that the degree of relation of “Δ” is higher than that of “x.”
More specifically, where the condition formula shown in
With respect to the specified symbol (specified value) “>,” the analysis unit 13c sets a degree of relation “Δ” for a specified symbol “≥” and sets a degree of relation “x” for specified symbols “=,” “<,” and “≤.”
Furthermore, with respect to the specified threshold value “60,” the analysis unit 13c sets a degree of relation “Δ” for specified values “50” and “70” that are close to it and sets a degree of relation “x” for specified values “40” and “80.”
In addition to the above settings, the analysis unit 13c sets priority ranks for the items “parameter,” “symbol,” and “threshold value.” It is assumed here that, for example, priority ranks “C,” “A,” and “B” are set for the items “parameter,” “symbol,” and “threshold value,” respectively.
Then the analysis unit 13c generates similar condition formulae indicating respective similar conditions in descending order of similarity while varying the specified value in descending order of the priority rank and the degree of relation. Then the analysis unit 13c searches the collection condition information DB 12a for past similar conditions in descending order of similarity.
For example, according to the priority ranks and the degree of relation shown in
If no similar condition is found, for example, the analysis unit 13c generates, as similar condition formula #2, a condition formula by changing the threshold value from “60 km/h” to “70 km/h.” The analysis unit 13c searches the collection condition information DB 12a for past similar conditions that satisfy the similar condition formula #2.
If no similar condition is found even if the above operation is performed certain times, the analysis unit 13c generates, as similar condition formula # n, a condition formula by changing the parameter from “vehicle speed” to “engine rotation speed” and changing the threshold value accordingly. The analysis unit 13c searches the collection condition information DB 12a for past similar conditions that satisfy the similar condition formula # n.
If no similar condition is found even if the above operation is performed repeatedly and only a condition formula without a degree of relation remains, the analysis unit 13c judges that there is no similar condition.
For example, the degree of relation shown in
In other words, the UI providing unit 13a provides guidance so as to recommend specified values of an item other than one item of a collection condition according to a specified value specified for the one item. This makes it possible to allow the data user to set another parameter easily according to the content of a parameter specified by the data user.
Next, a method for calculating a probability of occurrence of a collection condition by the analysis unit 13c will be described with reference to
The analysis unit 13c extracts, for each condition formula (for each of “condition A” and “condition B”), a frequency of occurrence of triggers under a similar condition that is similar to the specified collection condition and maps the triggers onto the necessary collection period.
Then the analysis unit 13c calculates a probability of occurrence of an event to become a trigger for each condition formula on the basis of a mapping result. The analysis unit 13c calculates a probability that the AND condition of the combination of the condition formulae (portions enclosed of dotted-line closed curves in
The analysis unit 13c calculates an expected sampling number, that is, the number of data to be collected as an expected number. Such a calculation result of the analysis unit 13c is communicated to the guidance generation unit 13d, and the guidance generation unit 13d generates guidance information including probability information as shown in
The product of probabilities (% values) of satisfaction of individual condition formulae is equal to a probability that the combination becomes true. However, in many cases, there exists a correlation and hence an actual probability is higher than the product. Thus, a simplified method may be employed in which a table showing correlation coefficients between threshold values (having proper intervals) of the parameter of each condition formula is set in advance and a probability is calculated taking into consideration a correlation coefficient corresponding to the product using the table.
Next, a process that is executed by the data collection system 1 according to the embodiment will be described with reference to
First, at step S101, the data user specifies a desired number of data to be collected through the user terminal 200. At step S102, the data user specifies a collection condition through the user terminal 200.
At step S103, the analysis unit 13c of the data collection apparatus 10 analyses past similar conditions. If the same condition is found (S104: yes), at step S105 the analysis unit 13c calculates a probability of occurrence under the specified condition on the basis of a frequency of occurrence of triggers under the same condition.
If the same condition is not found (S104: no), the analysis unit 13c searches for similar conditions (not the same condition) in descending order of similarity. If a similar condition is found (step S106: yes), at step S107 the analysis unit 13c calculates a probability of occurrence under the specified condition on the basis of a frequency of occurrence of triggers under the similar condition.
If no similar condition comes to be found after searching for similar conditions in descending order of similarity (S106: no), at step S108 the guidance generation unit 13d generates guidance information and the UI providing unit 13a provides guidance on changing of the condition on the basis of the generated guidance information. Then the process returns to step S102.
On the other hand, if a probability of occurrence is calculated at step S105 or S107, at step S109 the guidance generation unit 13d generates probability information relating to the probability and the UI providing unit 13a informs the data user of it.
If the collection will be completed by the desired date and time (S110: yes), the process is finished. On the other hand, if the collection will not be completed by the desired date and time (S110: no), at step S111 the analysis unit 13c searches for a condition formula under which the collection will be completed by the desired date and time.
At step S112, the guidance generation unit 13d generates guidance information on the basis of a result of the search and the UI providing unit 13a provides guidance on changing of the condition on the basis of the generated guidance information. Then the process returns to step S102.
As described above, the data collection apparatus 10 according to the embodiment is equipped with the collection unit 13f, the reception unit 13b, the analysis unit 13c, and the UI providing unit 13a (an example of the “providing unit”). The collection unit 13f collects data relating to respective vehicles V from vehicular devices 100 installed in the respective vehicles V. The reception unit 13b receives a data collection condition specified by a data user (an example of the “user”). The analysis unit 13c calculates a probability of occurrence of an event for collection of data specified in the collection condition on the basis of past data collection results. The UI providing unit 13a provides the user with information that is based on the probability of occurrence calculated by the analysis unit 13c.
Configured as described above, the data collection apparatus 10 according to the embodiment allows the data user to recognize the certainty of information collection under a specified collection condition.
The reception unit 13b receives, together with the collection condition, a desired number of data to be collected specified by the user. The analysis unit 13c derives a data collection period required for collection of the desired number of data on the basis of the probability of occurrence and causes the UI providing unit 13a to provide information relating to the collection period.
With these features, the data collection apparatus 10 according to the embodiment allows the data user to recognize information relating to a data collection period required for collection of a desired number of data.
The UI providing unit 13a provides the data user with at least information relating to a scheduled date and time of completion of collection of the desired number of data or information relating to the number of data to be collected by a date and time specified by the user.
With this feature, the data collection apparatus 10 according to the embodiment allows the data user to recognize a scheduled date and time of completion of collection of a desired number of data or the number of data to be collected by a date and time specified by the data user.
The analysis unit 13c extracts, from the collection results, a similar condition that is a past collection condition similar to the collection condition received by the reception unit 13b, derives a frequency of occurrence of the event under the similar condition, and calculates a probability of occurrence on the basis of the frequency of occurrence.
With this feature, the data collection apparatus 10 according to the embodiment can inform the data user with a probability of occurrence that is calculated with high accuracy on the basis of past collection results.
The analysis unit 13c extracts a similar condition according to priority ranks that are set for respective items of the collection condition and degrees of relation of values specified for respective items specified by the data user.
With this feature, the data collection apparatus 10 according to the embodiment can extracts a similar condition that is substantially equivalent to the collection condition specified by the data user because it extracts a similar condition according to priority ranks and degrees of relation rather than a mere degree of coincidence with the collection condition.
In the above-described embodiment, if a similar condition that is similar to a specified collection condition is found, the analysis unit 13c extracts a frequency of occurrence of triggers under the similar condition and calculates a probability of occurrence under a specified collection condition on the basis of the extracted frequency. However, the invention is not limited to this.
For example, the analysis unit 13c may generate a learning model of performing machine learning such as deep learning on the basis of past results in a prescribed period existing in the collection condition information DB 12a and the collection data DB 12b and, when a desired collection condition is input, outputting a probability of occurrence of triggers and a predicted date and time of completion corresponding to the collection condition.
Furthermore, although in the above-described embodiment the data user is, for example, a developer of an autonomous drive technique, this is just an example; the data user may be a corporation such as a service provider or a general individual.
Those skilled in the art would easily conceive other advantages and modifications. Thus, broader modes of the invention are not limited to the above-described particular, detailed, and typical embodiment. Various modifications are therefore possible without departing from the spirit and scope of a generalized inventive concept that is defined by the claims and their equivalents.
Number | Date | Country | Kind |
---|---|---|---|
JP2018-195244 | Oct 2018 | JP | national |
Number | Name | Date | Kind |
---|---|---|---|
5748098 | Grace | May 1998 | A |
8437958 | Krause | May 2013 | B2 |
20130345961 | Leader | Dec 2013 | A1 |
20150243166 | Yoshioka | Aug 2015 | A1 |
Number | Date | Country |
---|---|---|
2018-055581 | Apr 2018 | JP |
Number | Date | Country | |
---|---|---|---|
20200118020 A1 | Apr 2020 | US |