JP2012133530A - On-vehicle device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a service suitable for each composition of passengers even when person(s) other than a driver is/are in a vehicle.SOLUTION: In a control device 11 of an on-vehicle device 10, an on-vehicle microphone 12 collects voices generated in a vehicle, a seat sensor 19 identifies seating positions, a speaker estimation part 31 estimates each speaker based on each voice and each seating position, a conversation content estimation part 32 estimates a conversation content based on the voices, a passenger composition estimation part 33 estimates a passenger composition based on the seating positions, the speakers and the conversation content, an action purpose estimation part 34 estimates an action purpose based on the conversation content and the passenger composition, and a recommended service determination part 35 determines a recommended service based on the passenger composition and the action purpose.

Description

  The present invention relates to an in-vehicle device having a function of providing various services such as music and video playback services.

In this type of in-vehicle device, it is desired to provide a more optimal service. For example, the device described in Patent Literature 1 estimates the driver's psychological state based on voices emitted by the driver of the vehicle. The music is automatically reproduced according to the psychological state. According to this configuration, an optimal service can be provided to the vehicle driver.
By the way, the occupant who gets into the vehicle is not limited to only one driver, and as the occupant configuration, various situations such as a couple, friends, family, and work associates are conceivable. In this way, when a person other than the driver is on board, if the optimum service is provided only to the driver by the configuration of the above-mentioned Patent Document 1, even if the driver can be satisfied, It is not always possible to satisfy passengers. In some cases, the provided service for the driver may cause discomfort to other passengers.

JP 2006-92430 A

  Accordingly, an object of the present invention is to provide an in-vehicle device that can provide an optimum service for the passenger configuration at that time even when a person other than the driver is on the vehicle.

According to the in-vehicle device of the first aspect, the voice collecting means collects the voice generated in the vehicle. The seating position specifying means specifies the position of the seat where the occupant is seated in the vehicle as the seating position. The speaker estimation unit estimates a speaker among passengers present in the vehicle based on the voice collected by the voice collection unit and the seating position specified by the seating position specification unit. The conversation content estimation means estimates the content of the conversation by the speaker based on the voice collected by the voice collection means. The occupant configuration estimating means estimates the configuration of the occupant existing in the vehicle based on the sitting position specified by the sitting position specifying means, the speaker estimated by the speaker estimating means, and the conversation content estimated by the conversation content estimating means. To do. The behavioral purpose estimation means estimates the behavioral purpose of the occupant existing in the vehicle based on the conversation content estimated by the conversation content estimation means and the occupant configuration estimated by the occupant configuration estimation means. The recommended service determining unit determines a recommended service as a recommended service based on the occupant configuration estimated by the occupant configuration estimating unit and the behavior purpose estimated by the behavior purpose estimation unit.
As a result, it is possible to provide services according to the configuration of the passengers present in the vehicle and the purpose of action of the passengers, even if a person other than the driver is on the vehicle. It is possible to provide the optimal service for the passenger configuration at the time.

According to the in-vehicle device of the second aspect, the voice print pattern storage means stores the voice print pattern of the passenger who has boarded the vehicle in the past. The speaker estimation unit estimates the speaker by comparing the voice print pattern collected by the voice collection unit with the voice print pattern stored by the voice print pattern storage unit.
Thereby, the estimation accuracy of the speaker by the speaker estimation means can be improved.

In this case, as in the in-vehicle device according to the third aspect, the voice print pattern storage means may be configured to store the voice print pattern having the highest frequency as the voice print pattern of the owner of the vehicle.
Thus, by providing the learning function in the voiceprint pattern storage function by the voiceprint pattern storage means, the estimation accuracy of the speaker is further improved when the speaker is a specific person (in this case, the owner of the vehicle). be able to.

According to the in-vehicle apparatus of the fourth aspect, the recommended service history accumulating unit accumulates the service determined by the recommended service determining unit in the past in association with the occupant configuration and the action purpose at that time. The recommended service determining means extracts the service corresponding to the occupant configuration estimated by the occupant configuration estimating means and the behavior purpose estimated by the behavior purpose estimating means from the recommended service history accumulating means, and the extracted service as the recommended service. decide.
As a result, the service actually determined and provided as the recommendation service in the past is provided again, and the possibility that an inappropriate service is provided can be reduced as much as possible.

The block diagram which concerns on one Embodiment of this invention and shows the structure of a vehicle-mounted apparatus roughly The figure which shows an example of voiceprint pattern data Diagram showing an example of speaker pattern data Diagram showing an example of conversation content pattern data The figure which shows an example of crew member pattern data The figure which shows an example of action purpose pattern data The figure which shows an example of recommendation pattern data The figure which shows an example of a structure of recommendation service history database Flow chart showing the contents of speaker estimation processing Flow chart showing the content of conversation content estimation processing Flow chart showing the contents of passenger composition estimation processing Flow chart showing contents of action purpose estimation process The flowchart which shows the contents of recommendation service extraction decision processing

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
As shown in FIG. 1, the in-vehicle device 10 is mounted on a vehicle such as an automobile, for example, and includes a control device 11, an in-vehicle microphone 12 (corresponding to sound collection means), an in-vehicle speaker 13, a display unit 14, a navigation unit 15, and an operation switch. Unit 16, external memory 17, communication unit 18, vehicle information detection unit 20 including seat sensor 19 (corresponding to seating position specifying means), RTC 21 (Real Time Clock), pattern data storage unit 22, recommendation service history database 23, etc. I have.
The control device 11 is mainly composed of a microcomputer having a CPU, ROM and RAM (not shown). The in-vehicle device 10 executes a control program in the CPU of the control device 11 so that a speaker estimation unit 31 (corresponding to a speaker estimation unit), a conversation content estimation unit 32 (corresponding to a conversation content estimation unit), and an occupant configuration estimation Unit 33 (corresponding to occupant configuration estimating means), behavioral purpose estimating unit 34 (corresponding to behavioral purpose estimating means), recommended service determining unit 35 (corresponding to recommended service determining means), recommended service executing unit 36, pattern data accumulation processing unit 37 (corresponding to voice print pattern storage means) and a recommendation service history storage processing unit 38 (corresponding to recommendation service history storage means) are virtually realized by software.

The in-vehicle microphone 12 collects sound generated in the vehicle, converts the collected sound into a sound signal, and inputs the sound signal to the control device 11. The control device 11 is configured to be able to extract the voice print pattern of the speaker from the voice collected by the in-vehicle microphone 12 by executing processing by software (voice print pattern extraction processing described later). In addition, the control device 11 is configured to be able to extract a predetermined keyword from the voice collected by the in-vehicle microphone 12 by executing a process (a word extraction process described later) by software.
The in-vehicle speaker 13 is connected to the control device 11 via a sound controller (not shown), and outputs various sounds based on a sound output signal from the control device 11. The sound output from the in-vehicle speaker 13 is based on music data in addition to the sound related to guidance, the sound related to the operation explanation, the sound notifying that the antitheft function is operating, the talkback sound corresponding to the sound recognition result, and the like. Also included are music played back and audio played back based on video data.

The display unit 14 includes a color display such as a liquid crystal or an organic EL. The display unit 14 displays a route guidance screen, an operation explanation screen, various setting screens, a video screen reproduced based on video data, and the like.
The navigation unit 15 is a GPS receiver (not shown) that receives radio waves transmitted from GPS satellites to determine the current position of the vehicle by GPS (Global Positioning System), and a map stored in the map data storage unit. A map data input unit (not shown) for inputting data is provided, and the vehicle route guidance is executed based on the current position of the vehicle and the map data.

The operation switch unit 16 includes a mechanical switch provided near the screen of the display unit 14, a touch panel switch provided on the screen of the display unit 14, and the like. The user uses each switch of the operation switch unit 16 to switch the destination of the vehicle route guidance, the screen of the display unit 14 and the display mode (for example, change of the map scale, selection of the menu screen, route search, route search). Input various commands to start guidance, correct the current position and adjust the volume.
The external memory 17 is configured by a storage medium such as a removable hard disk drive or flash memory. The external memory 17 may be shared with, for example, a RAM or an EEPROM provided in the control device 11 of the in-vehicle device 10.

The communication unit 18 performs data communication with an information server (not shown) via, for example, a wireless communication line. That is, the communication unit 18 has various data transmission functions and reception functions.
The seat sensor 19 specifies the position of the seat where the passenger is seated in the vehicle as the seating position, and is provided in each seat (driver's seat, passenger seat, rear seat, etc.) in the vehicle. For example, when the driver is seated in the driver's seat, the seat sensor 19 provided in the driver's seat outputs a seating signal. The control device 11 detects “driver's seat” as the seating position based on the seating signal output from the seat sensor 19 of the driver's seat. For other seats (passenger seat, rear seat, etc.), the control device 11 similarly detects the seating position.

The vehicle information detection unit 20 includes various devices (for example, an engine control device, a brake control device, a transmission control device, etc.), various sensors (for example, an acceleration sensor, a speed sensor, an accelerator sensor, etc.), and various switches. (For example, an ignition switch, a headlight switch, etc.). The seat sensor 19 described above is included in the sensors constituting the vehicle information detection unit 20. The vehicle information detection unit 20 inputs various information detected by these devices, sensors, switches, and the like to the control device 11 as vehicle information.
The RTC 21 is a time measuring means for acquiring information related to time such as date, time, and elapsed time. The RTC 21 counts the number of clock pulses output from a transmitter (not shown), and generates a unit time signal indicating a unit time when the counted value reaches a reference value. And RTC21 produces | generates terminal time based on the produced | generated unit time signal.

  The pattern data storage unit 22 is configured by a storage medium such as a hard disk drive or a flash memory, for example. In this case, the pattern data storage unit 22 stores the voiceprint pattern data PD1 shown in FIG. 2, the speaker pattern data PD2 shown in FIG. 3, the conversation content pattern data PD3 shown in FIG. 4, and the occupant configuration pattern data PD4 shown in FIG. Various pattern data such as action purpose pattern data PD5 shown in FIG. 6 and recommended pattern data PD6 shown in FIG. 7 are constructed.

  In the voiceprint pattern data PD1 (see FIG. 2), the voiceprint pattern extracted from the voice collected by the in-vehicle microphone 12 includes the number of detections of each seating position detected when the voiceprint pattern is obtained, and the voiceprint. It is stored in association with the total number of pattern extractions (the total number of seating position detections). For example, if “driver's seat” is detected as the seating position when “voice pattern A” is obtained, the seating position detection count in the driver's seat column corresponding to the voiceprint pattern A is incremented by one. Accordingly, the total number of extractions is also incremented by one. The same applies when “passenger seat” or “rear seat” is detected as the seating position. Note that the update process of the voiceprint pattern data PD1 is executed by the pattern data storage processing unit 37 by the control device 11. That is, the pattern data accumulation processing unit 37 accumulates voice print patterns of passengers who have previously boarded the vehicle. Further, the pattern data storage processing unit 37 can execute not only the update process of the voiceprint pattern data PD1, but also the update process of the other pattern data PD2 to PD6.

  In the speaker pattern data PD2 (see FIG. 3), the speaker candidates (for example, “owner” of the vehicle, “family” of the owner, “friends” of the owner, “others”, etc.) , And stored in association with each piece of information of the conversation keyword for speaker estimation. In this case, in the voiceprint pattern column corresponding to “owner”, the voiceprint pattern having the highest frequency in the voiceprint pattern data PD1 described above (the voiceprint pattern having the highest number of detections in the driver's seat column or the highest number of total extractions). A voiceprint pattern, in this case, voiceprint pattern A) is stored. In the voice print pattern column corresponding to “family” and the voice print pattern column corresponding to “friend”, a voice print pattern having a frequency (total number of extractions) of one or more in the above voice print pattern data PD1, that is, Of the voiceprint patterns stored in the voiceprint pattern data PD1, voiceprint patterns excluding the “owner” voiceprint pattern (in this case, the voiceprint pattern excluding the voiceprint pattern A and corresponding to the voiceprint patterns B and C) are stored. . In this case, any voice print pattern corresponding to “other person” is not stored. That is, the speaker pattern data PD2 is considered to be a voiceprint pattern of a person (including the owner himself) who has a closer relationship with the owner as the voiceprint pattern dataF1 has a higher frequency.

  In this case, no keyword is stored in the conversation keyword column for estimating the speaker corresponding to “owner”, “friend”, or “other”, but for estimating the speaker corresponding to “family”. In the conversation keyword field, family names (for example, family names, names related to the family (such as father and mother), family nicknames, etc.) are stored as keywords. It is naturally possible to store an appropriate keyword in the conversation keyword column for estimating a speaker corresponding to a speaker other than “family”.

  In the conversation content pattern data PD3 (see FIG. 4), the conversation content (for example, “work relationship”, “home relationship”, “hobby relationship”, “leisure relationship”, “others”) includes the conversation content estimation conversation. Stored in association with the keyword. In this case, in the conversation keyword column for estimating conversation contents corresponding to “work relation”, the title (for example, “president”, “department manager”, “section manager”), company name (for example, “XX Corporation”) , “ΔΔ office”), business words (for example, “transaction”, “delivery date”, “contract”, “meeting”) and the like are stored as keywords. The name of the family is stored as a keyword in the conversation keyword column for estimating the conversation content corresponding to “family relations”. In the conversation keyword column for estimating conversation contents corresponding to “hobby relation”, the name of a movie, the name of a song, sports terms, and the like are stored as keywords. In the conversation keyword column for estimating conversation contents corresponding to “leisure relation”, the name of a tourist attraction, the name of an entertainment facility, the name of an accommodation facility, and the like are stored as keywords. In this case, no keyword is stored in the conversation keyword column for estimating the conversation content corresponding to “others”, but it is naturally possible to store an appropriate keyword.

  In the occupant configuration pattern data PD4 (see FIG. 5), occupant configuration candidates (for example, “owner (1 person)”, “non-owner (1 person)”, etc.) In the case of two passengers, “Couple (two)”, “Couple (two)”, etc., “Family (three or more)”, “Friends (three or more) ) "," Work-related persons (three or more) ", etc.) are stored in association with each information of the sitting position, the speaker, and the conversation content. In this case, for example, “driver's seat” is stored in the seating position column corresponding to “owner (one person)”, “owner” is stored in the speaker column, and any condition is set in the conversation content column. Not stored. In addition, for example, “driver's seat”, “passenger seat”, and “rear seat” are stored in the seating position column corresponding to “family (three or more)”, and “family” is stored in the speaker column. In the content column, “family relations” is stored.

  In the action purpose pattern data PD5 (see FIG. 6), the action purpose candidates (for example, “return home”, “leisure”, “travel”, “pick-up”, etc.) include the conversation contents, crew composition, destination, day of the week, It is stored in association with each piece of information of driving time (elapsed time since the vehicle started traveling). In this case, for example, “Owner (1 person)”, “Couple (2 persons)”, “Family (3 persons or more)” are stored in the passenger composition field corresponding to “Return home”, and “ "Home" is stored, and no condition is stored in the conversation content column, day of week column, and driving time column. Further, for example, “couple (two people)” is stored in the passenger composition column corresponding to “pick-up”, “work relationship” and “family relationship” are stored in the conversation content column, and “ “Transportation facility” is stored, “Monday”, “Tuesday”, “Wednesday”, “Thursday”, and “Friday”, which are weekdays, are stored in the day of week column, and no conditions are stored in the driving time column.

  In the recommendation pattern data PD6 (see FIG. 7), candidates for recommendation services are stored in association with information on the occupant configuration and behavioral purpose. In this case, the recommendation service is divided into the highest priority recommendation service, the next priority recommendation service, and the next priority recommendation service. For example, the occupant configuration column is “owner (one person)” and the action When the purpose column is “Return home”, “Favorite recommendation service” is stored in the highest priority recommendation service, “Music playback” is stored in the next priority recommendation service, and “Video playback” is stored in the priority priority service one after another. Stored. Further, for example, when the occupant configuration column is “family (3 or more)” and the action purpose column is “return home”, “leisure”, or “travel”, the highest priority recommendation service is “ “Favorite recommendation service” is stored, “Video playback” is stored in the next priority recommendation service, and “Music playback” is stored in the priority priority recommendation service one after another. The “destination information provision” shown in FIG. 7 is information related to the destination set via the navigation unit 15 (for example, event information held at the destination, business hours of the destination, etc. ) Is output (provided) via the display unit 14 or the in-vehicle speaker 13. “Road information provision” is a service that outputs (provides) traffic information, traffic regulation information, accident information, and the like occurring on each road via the display unit 14 and the in-vehicle speaker 13.

  Next, the recommended service history database 23 will be described. The recommended service history database 23 is a database for accumulating recommended services determined by a recommended service extraction determination process described later. With this recommendation service history database 23, it is possible to grasp the history of the recommendation service provided in the vehicle in the past. Here, the configuration of the recommendation service history database 23 will be described with reference to FIG.

  In the recommendation service history database 23, a recommendation service determined by a recommendation service extraction determination process, which will be described later, that is, a recommendation service (provided recommendation service) actually provided in the vehicle in the past, is displayed in the passenger configuration (the recommendation service is the recommendation service). Estimated occupant composition when provided), action purpose (estimated action purpose when the recommended service was provided), date (date when the recommended service was provided), time (provided the recommended service) Time) and the information on the favorability are stored in association with each other. The likability is information indicating how much the occupant of the vehicle has (liked) the recommendation service actually provided in the vehicle. For example, when the provided service is stopped immediately after provision (within a predetermined time after provision), the control device 11 stores information indicating “dislike” (in this case, the frequency “1”) in the likability column. For example, when the service is stopped at a stage where more than half of the service is provided, information indicating “normal” (in this case, frequency “2”) is stored in the preference field, and when the service is executed to the end, it is preferable. Information indicating “like” (in this case, frequency “3”) is stored in the sensitivity column. The update process of the recommended service history database 23 is executed by the recommended service history storage processing unit 38 by the control device 11. In other words, the recommended service history accumulation processing unit 38 corresponds to the service determined in the past by the recommended service determining unit 35 in accordance with the occupant configuration at that time (estimated occupant configuration at that time) and the action purpose (the estimated action purpose at that time). It will be added and accumulated.

  Next, the speaker estimation unit 31 will be described. The speaker estimation unit 31 estimates “speaker” among passengers in the vehicle based on the voice collected by the in-vehicle microphone 12 and the seating position specified by the seat sensor 19. Here, a speaker estimation process executed by the speaker estimation unit 31 by the control device 11 will be described with reference to FIG.

  The control device 11 extracts the voice print pattern of the speaker from the voice signal input from the in-vehicle microphone 12 by executing the voice print pattern extraction process (step A1). Next, the control device 11 executes voice print pattern matching processing (step A2). In this matching process, the control device 11 searches whether or not the voice print pattern extracted by the voice print pattern extraction process (see step A1) is present in the voice print pattern data PD1. Next, the control device 11 determines whether or not the corresponding voice print pattern exists in the voice print pattern data PD1 (step A3).

  When the corresponding voice print pattern exists in the voice print pattern data PD1 (step A3: YES), the control device 11 executes a keyword extraction process (step A4). On the other hand, when the corresponding voiceprint pattern does not exist in the voiceprint pattern data PD1 (step A3: NO), the control device 11 newly extracts the voiceprint pattern extracted by the process of step A1 (voiceprint pattern extraction process). Update processing (step A5) for storing (adding) in PD1 is executed, and then keyword extraction processing is executed (step A4).

  In the keyword extraction process, the control device 11 extracts a predetermined keyword (in this case, a keyword stored in the conversation keyword column for speaker estimation in the speaker pattern data PD2) from the voice collected by the in-vehicle microphone 12. Depending on the content of the voice collected by the in-vehicle microphone 12, no keyword may be extracted by this keyword extraction process.

  Next, the control device 11 executes speaker matching processing (step A5). In this speaker matching process, the control device 11 matches the voiceprint pattern stored in the voiceprint pattern field of the speaker pattern data PD2 with the voiceprint pattern extracted by the voiceprint pattern extraction process (see step A1), and The data stored in the conversation keyword column for speaker estimation is searched for data that matches the keyword extracted by the keyword extraction process (see step A4). And the control apparatus 11 estimates the person stored in the speaker column of the searched data as "speaker" which exists in the present vehicle. For example, when the extracted voiceprint pattern is the voiceprint pattern A and no keyword is extracted, the control device 11 estimates “owner” as “speaker”. When completing the speaker matching process, the control device 11 ends the speaker estimation process.

  Next, the conversation content estimation unit 32 will be described. The conversation content estimation unit 32 estimates the “conversation content” by the speaker estimated by the above-described speaker estimation process based on the voice collected by the in-vehicle microphone 12. Here, the conversation content estimation processing executed by the control content 11 by the conversation content estimation unit 32 will be described with reference to FIG.

  The control device 11 executes a keyword extraction process to obtain a predetermined keyword (in this case, a keyword stored in the conversation keyword for conversation content estimation in the conversation content pattern data PD3) from the voice collected by the in-vehicle microphone 12. Extract. Depending on the content of the voice collected by the in-vehicle microphone 12, no keyword may be extracted by this keyword extraction process.

  Next, the control device 11 executes a conversation content matching process (step B2). In this conversation content matching process, the control device 11 matches the keyword stored in the conversation keyword column for conversation content estimation of the conversation content pattern data PD3 with the keyword extracted by the keyword extraction process (see step B1). Search for data. And the control apparatus 11 estimates the content stored in the conversation content column of the searched data as "conversation content" in the present vehicle. For example, when the extracted keyword is a job title, a company name, a business word, or the like, the control device 11 estimates “work relation” as “conversation content”. When completing the conversation content matching process, the control device 11 ends the conversation content estimation process.

  Next, the occupant configuration estimation unit 33 will be described. The occupant configuration estimation unit 33 is configured to determine whether the passenger is in the vehicle based on the seating position specified by the seat sensor 19, the speaker estimated by the speaker estimation process described above, and the conversation content estimated by the conversation content estimation process. The "occupant composition" that exists in is estimated. Here, an occupant configuration estimation process executed by the occupant configuration estimation unit 33 by the control device 11 will be described with reference to FIG.

  In this occupant configuration estimation process, the control device 11 executes an occupant configuration matching process (step C1). In this occupant configuration matching process, the control device 11 matches the seating position stored in the seating position column of the occupant configuration pattern data PD4 with the seating position specified by the input signal from the seat sensor 19, and the speaker column Is the same as the speaker estimated by the above-described speaker estimation processing, and the conversation content stored in the conversation content column is changed to the conversation content estimated by the above-described conversation content estimation processing. Search for matching data. Then, the control device 11 estimates the configuration stored in the occupant configuration column of the retrieved data as the “occupant configuration” in the current vehicle. For example, the specified seating positions are “driver's seat”, “passenger seat”, and “rear seat”, the estimated speaker is “family”, and the estimated conversation When the content is “family relation”, the control device 11 estimates “family (three or more)” as “occupant composition”. When completing the occupant configuration matching process, the control device 11 ends the occupant configuration estimation process.

  Next, the behavior purpose estimation unit 34 will be described. The behavior purpose estimation unit 34 determines the “action purpose” of the occupant existing in the vehicle based on the conversation content estimated by the conversation content estimation process described above and the occupant configuration estimated by the occupant composition estimation process. To be estimated. Here, the behavioral purpose estimation process executed by the behavioral purpose estimation unit 34 by the control device 11 will be described with reference to FIG.

  In this behavior purpose estimation process, the control device 11 executes a behavior purpose matching process (step D1). In this behavioral purpose matching process, the control device 11 matches the conversation content stored in the conversation content column of the behavioral purpose pattern data PD5 with the conversation content estimated by the above-described conversation content estimation process, and enters the occupant configuration field. The stored occupant configuration matches the occupant configuration estimated by the occupant configuration estimation process described above, and the destination stored in the destination column matches the destination set via the navigation unit 15. The day of the week stored in the day of the week column matches the day of the week at the present time, and the operation time stored in the operation time column is searched for data corresponding to the current operation time. And the control apparatus 11 estimates the content stored in the action purpose column of the searched data as "the action purpose" of the passenger | crew who exists in the present vehicle. For example, if there is no estimated conversation content and the estimated crew composition is “couple (2 people)”, “couple (2 people)”, “family (3 or more)”, “friends ( 3 ”or more), and the set destination is“ sightseeing spot ”or“ accommodation facility ”, and the current day of the week is“ Saturday ”,“ Sunday ”, or“ Holiday ” If the current driving time is “two hours or more”, the control device 11 estimates “travel” as an action purpose. When completing the action purpose matching process, the control device 11 ends the action purpose estimation process.

  Next, the recommendation service determination unit 35 will be described. The recommended service determination unit 35 recommends a service recommended as a service to be provided in the vehicle based on the occupant configuration estimated by the occupant configuration estimation process described above and the behavior purpose estimated by the behavior purpose estimation process described above. It is determined as “Recommendation Service”. Here, the recommended service extraction determination process executed by the recommended service determination unit 35 by the control device 11 will be described with reference to FIG.

  In this recommendation service extraction determination process, the control device 11 executes a recommendation service search process (step E1). In this recommendation service search process, the control device 11 matches the occupant configuration stored in the occupant configuration column of the recommendation service history database 23 with the occupant configuration estimated by the above occupant configuration estimation process, and the action purpose column. The behavioral purpose stored in is searched for data that matches the behavioral purpose estimated by the behavioral purpose estimation process described above. And the control apparatus 11 extracts the recommendation service stored in the provision recommendation service column of the searched data as "recommendation service". For example, when the estimated occupant configuration is “family (three or more)” and the estimated action purpose is “drive”, the control device 11 sets “road information provision” as a recommendation service. Extract as

  Next, the control device 11 determines whether or not a recommendation service has been extracted by this recommendation service search process, that is, whether or not a history of the corresponding recommendation service exists in the recommendation service history database 23 (step E2). . When the history of the corresponding recommended service exists in the recommended service history database 23 (step E2: YES), the control device 11 executes a recommended service determination process (step E3). On the other hand, when the history of the corresponding recommended service does not exist in the recommended service history database 23 (step E2: NO), the control device 11 executes the recommended service matching process (step E4), and then the recommended service determination process Is executed (step E3).

  In the recommendation service matching process, the control device 11 matches the occupant configuration stored in the occupant configuration column of the recommendation pattern data PD6 with the occupant configuration estimated by the above-described occupant configuration estimation process, and stores it in the action purpose column. Data for which the behavioral purpose that has been determined matches the behavioral purpose estimated by the behavioral purpose estimation process described above is searched. Then, the service stored in the highest priority recommended service column of the retrieved data is extracted as a recommended service. For example, when the estimated occupant configuration is “couple (two people)” and the estimated action purpose is “pickup”, the control device 11 is stored in the highest priority recommended service column. The “favorite recommendation” is extracted as a recommendation service. In this case, the “favorable recommendation” is specified as follows. That is, the control device 11 matches the occupant configuration stored in the occupant configuration column of the recommendation service history database 23 with the occupant configuration estimated by the above-described occupant configuration estimation process, and is stored in the action purpose column. Of the data whose behavioral purpose matches the behavioral purpose estimated by the behavioral purpose estimation process described above, data in which information indicating “like” (frequency “3”) is stored in the likability column is extracted. Then, the service stored in the provided recommended service column of the extracted data is specified as “favorable recommendation”. In addition, when there are a plurality of extracted data, for example, the date stored in the date column is selected to be close to the current date, or the time stored in the time column matches the current time. Alternatively, it is possible to specify any one of the plurality of extracted data as a “favorable recommendation” by selecting approximate data.

In the recommendation service determination process, the control device 11 determines the recommendation service extracted in step E1 or the recommendation service extracted in step E4 as a recommendation service to be provided in the vehicle. When completing the recommendation service determination process, the control device 11 ends the recommendation service extraction determination process.
The control device 11 executes the recommendation service determined by the recommendation service extraction determination process by the recommendation service determination unit 35 by the recommendation service execution unit 36 and provides it to the vehicle.

  As described above, according to the present embodiment, the control device 11 of the in-vehicle device 10 collects the sound generated in the vehicle by the in-vehicle microphone 12, and the position of the seat where the occupant is seated in the vehicle by the seat sensor 19. Is identified as the seating position. And the control apparatus 11 is a speaker among the passengers present in the vehicle based on the voice collected by the in-vehicle microphone 12 and the seating position specified by the seat sensor 19 by the speaker estimation process by the speaker estimation unit 31. Is estimated. Further, the control device 11 estimates the content of the conversation by the speaker based on the voice collected by the in-vehicle microphone 12 by the conversation content estimation process by the conversation content estimation unit 32. In addition, the control device 11 performs the occupant configuration estimation process by the occupant configuration estimation unit 33, the seating position specified by the seat sensor 19, the speaker estimated by the speaker estimation unit 31, and the conversation estimated by the conversation content estimation unit 32. Based on the content, the configuration of the occupants present in the vehicle is estimated. Further, the control device 11 exists in the vehicle based on the conversation content estimated by the conversation content estimation unit 32 and the occupant configuration estimated by the occupant configuration estimation unit 33 by the behavior purpose estimation process by the behavior purpose estimation unit 34. Estimate the behavioral purpose of the occupant. Then, the control device 11 recommends a service based on the occupant configuration estimated by the occupant configuration estimation unit 33 and the behavior purpose estimated by the behavior purpose estimation unit 34 by the recommendation service extraction determination process by the recommendation service determination unit 35. Is determined as a recommendation service. That is, the control device 11 determines a recommendation service to be provided in the vehicle according to what kind of occupant (occupant configuration) is going to do what (behavior purpose).

  As a result, it is possible to provide services according to the configuration of the passengers present in the vehicle and the purpose of action of the passengers, even if a person other than the driver is on the vehicle. It is possible to provide the optimal service for the passenger configuration at the time.

Further, the control device 11 causes the pattern data storage processing unit 37 to store the voice print pattern of the occupant who has boarded the vehicle in the voice print pattern data PD1. Then, in the speaker estimation process by the speaker estimation unit 31, the control device 11 collates the voice print pattern collected by the in-vehicle microphone 12 with the voice print pattern stored by the pattern data storage processing unit 37. Estimate the speaker.
Thereby, the estimation precision of the speaker by the speaker estimation process which the control apparatus 11 performs by the speaker estimation part 31 can be improved.

Further, in this case, the control device 11 causes the pattern data storage processing unit 37 to store the most frequently used voice print pattern in the voice print pattern data PD1 as the voice print pattern of the vehicle owner.
Thus, by providing the voice print pattern accumulation function by the pattern data accumulation processing unit 37 with the learning function, the estimation accuracy of the speaker when the speaker is a specific person (in this case, the owner of the vehicle) is further increased. Can be improved.

Further, the control device 11 causes the recommendation service history accumulation processing unit 38 to accumulate the service determined by the recommendation service determination unit 35 in the past in the recommendation service history database 23 in association with the occupant configuration and action purpose at that time. Then, the control device 11 recommends a service corresponding to the occupant configuration estimated by the occupant configuration estimation unit 33 and the behavior purpose estimated by the behavior purpose estimation unit 34 by the recommendation service extraction determination process by the recommendation service determination unit 35. Extracted from the history database 23 or the recommended pattern data PD6, and the extracted service is determined as a recommended service to be provided in the vehicle.
As a result, services that were actually determined and provided as recommended services in the past will be provided again, and inappropriate services (for example, they have never been determined as recommended services in the past, satisfying passengers in the car) Service that is highly likely not to be provided) can be reduced as much as possible.

In addition, this invention is not limited only to one embodiment mentioned above, It can apply to various embodiment in the range which does not deviate from the summary, For example, it can deform | transform or expand as follows. .
The pattern data storage unit 22 and the recommendation service history database 23 may be provided outside the in-vehicle device 10 (for example, an information server), and various pattern data and recommendation service history information may be acquired from the outside of the in-vehicle device 10.
Each of the pattern data PD1 to PD6 described above is an example, and the configuration can be changed as appropriate, and appropriate information can be stored in each column of the pattern data PD1 to PD6. The recommended service history database 23 described above is also an example, and the configuration can be changed as appropriate, and appropriate information can be stored in each column of the database 23.

  In the drawings, 10 is a vehicle-mounted device, 12 is a vehicle-mounted microphone (sound collecting means), 19 is a seat sensor (sitting position specifying means), 31 is a speaker estimation unit (speaker estimation unit), and 32 is a conversation content estimation unit (conversation). Content estimation means), 33 an occupant configuration estimation unit (occupant configuration estimation means), 34 an action purpose estimation unit (behavior purpose estimation means), 35 a recommendation service determination unit (recommendation service determination means), and 37 a pattern data storage process Reference numeral (voice print pattern storage means) 38 denotes a recommended service history storage processing section (recommended service history storage means).

Claims (4)

Voice collecting means for collecting voice generated in the vehicle;
Seating position specifying means for specifying the position of the seat on which the passenger is seated in the vehicle as the seating position;
Based on the voice collected by the voice collecting means and the seating position specified by the seating position specifying means, a speaker estimation means for estimating a speaker among passengers present in the vehicle;
Conversation content estimation means for estimating the content of the conversation by the speaker based on the voice collected by the voice collection means;
An occupant that estimates the configuration of the occupant present in the vehicle based on the sitting position specified by the sitting position specifying unit, the speaker estimated by the speaker estimating unit, and the conversation content estimated by the conversation content estimating unit Configuration estimation means;
Based on the conversation content estimated by the conversation content estimation means, and the occupant configuration estimated by the occupant configuration estimation means, the behavior purpose estimation means for estimating the behavior purpose of the occupant existing in the vehicle;
A recommended service determining unit that determines a recommended service as a recommended service based on the passenger configuration estimated by the passenger configuration estimating unit and the behavior purpose estimated by the behavior purpose estimating unit;
A vehicle-mounted device comprising: Voice pattern storing means for storing voice print patterns of occupants who have boarded the vehicle in the past;
2. The speaker estimation unit according to claim 1, wherein the speaker estimation unit estimates a speaker by comparing a voice print pattern collected by the voice collection unit with a voice print pattern stored by the voice print pattern storage unit. The in-vehicle device described.   The in-vehicle device according to claim 2, wherein the voice print pattern storage unit stores a voice print pattern having the highest frequency as a voice print pattern of a vehicle owner. The recommended service history accumulating means for accumulating the service determined by the recommended service determining means in the past in association with the occupant configuration and action purpose at that time,
The recommended service determining means extracts from the recommended service history accumulating means the service corresponding to the occupant configuration estimated by the occupant configuration estimating means and the behavior purpose estimated by the behavior purpose estimation means, and the extracted service is recommended. It determines as a service, The vehicle-mounted apparatus of any one of Claim 1 to 3 characterized by the above-mentioned.

Images