Automatic response device, automatic response method, and computer program for automatic response

Abstract

To provide an automatic response device capable of making a more appropriate response to a occupant who speaks among a plurality of occupants in a vehicle. [Solution] The automatic response device 5 has a position estimation unit 31 that estimates the position of the occupant who spoke among the multiple occupants of the vehicle 1, and a response generation unit 33 that generates response information by inputting position information representing the position of the occupant who spoke and voice information representing the content of the utterance into a generation model that has been trained in advance to generate response information to the content of the utterance.

Description

[Technical Field]

[0001] The present invention relates to an automatic response device, an automatic response method, and a computer program for automatic response that automatically responds to speech from a vehicle occupant. [Background technology]

[0002] It has been proposed to recognize the speaking position and content based on the voice of a vehicle occupant, and based on the speaking position and content, select and output a portion of multiple pieces of information about facilities specified by the gestures of the person at the speaking position detected based on an image inside the vehicle cabin (see Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-90615 Summary of the Invention [Problem to be solved by the invention]

[0004] The content of the response required may differ depending on the location of the occupant who made the utterance.

[0005] SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide an automatic response device that can give an appropriate response to a occupant who speaks among a plurality of occupants in a vehicle. [Means for solving the problem]

[0006] According to one embodiment, there is provided an automatic answering device including: a position estimation unit that estimates a position of a occupant who has spoken among a plurality of occupants of a vehicle; and a response generation unit that generates response information by inputting position information indicating the position of the occupant who has spoken and voice information indicating the content of the utterance into a generation model that has been trained in advance to generate response information in response to the content of the utterance.

[0007] In one embodiment, the automatic response device further includes a control unit that controls an appliance provided at the position of the occupant who made the speech based on the response information.

[0008] In one embodiment, the automatic response device further includes an identification unit that identifies a partial area corresponding to the position of the occupant indicated by the position information in the interior image of the vehicle. The response generation unit then generates the response information by further inputting the partial area into the generative model.

[0009] According to another embodiment, there is provided an automatic response method, which includes estimating a position of a occupant who has spoken among a plurality of occupants of a vehicle, and generating response information by inputting position information representing the position of the occupant who has spoken and audio information representing the content of the utterance into a generative model that has been trained in advance to generate response information in response to the content of the utterance.

[0010] According to yet another embodiment, there is provided a computer program for automatic response, the computer program for automatic response including instructions for causing a computer to estimate a position of a occupant who has spoken among a plurality of occupants of a vehicle, and generate response information by inputting position information representing the position of the occupant who has spoken and audio information representing the content of the utterance into a generation model that has been trained in advance to generate response information in response to the content of the utterance. [Effects of the Invention]

[0011] The automatic response device according to the present disclosure has the effect of being able to provide an appropriate response to a occupant who speaks among multiple occupants in a vehicle. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a schematic diagram of a vehicle equipped with an automatic response device; [Figure 2] FIG. 2 is a diagram illustrating a hardware configuration of an automatic answering device. [Figure 3] FIG. 2 is a functional block diagram of a processor of an automatic answering device. [Figure 4]FIG. 10 is an explanatory diagram of the relationship between input and response information of a generative model according to this embodiment. [Figure 5] 10 is a flowchart illustrating the operation of an automatic answering device. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, an automatic answering device, an automatic answering method, and an automatic answering computer program executed by the automatic answering device will be described with reference to the drawings. This automatic answering device estimates the location of a occupant who has spoken among multiple occupants of a vehicle. Then, this automatic answering device generates response information by inputting location information indicating the location of the occupant who has spoken and voice information indicating the content of the utterance into a generation model that has been trained in advance to generate response information in response to the content of the utterance.

[0014] 1 is a schematic diagram of a vehicle equipped with an automatic answering device. In this embodiment, the vehicle 1 has a camera 2, a plurality of microphones 3-1 to 3-n (n is an integer of 2 or more, 4 in the illustrated example), a notification device 4, and an automatic answering device 5. The camera 2, the microphones 3-1 to 3-n, the notification device 4, and the automatic answering device 5 are connected to each other so that they can communicate with each other.

[0015] Camera 2 is an example of an interior imaging unit, and is mounted near the top edge of the windshield facing the interior of the vehicle so that all passengers in vehicle 1 are included in the imaging area. Camera 2 generates an image showing the interior of vehicle 1 at each predetermined imaging cycle and outputs the generated image to automatic response device 5. Hereinafter, the image generated by camera 2 will be referred to as an interior image.

[0016] The microphones 3-1 to 3-n collect the voices of any of the passengers in the vehicle 1 and output audio signals representing the voices. To this end, one of the microphones 3-1 to 3-n is attached to each seat of the vehicle 1 at a position where it can collect the voices of the passengers sitting in that seat. In the example shown in FIG. 1, a microphone is attached to each of the driver's seat, passenger seat, right rear seat, and left rear seat in a position near the seat (e.g., on the instrument panel for the driver's seat and passenger seat, and on the back of the seat in front of the rear seat). The microphones 3-1 to 3-n may also be attached in an array at a position where it can collect the voices of any of the passengers in the vehicle cabin (e.g., near the ceiling at the front of the vehicle cabin or on the instrument panel). Each of the microphones 3-1 to 3-n outputs the generated audio signal to the automatic answering device 5.

[0017] The notification device 4 is provided in the passenger compartment of the vehicle 1 and notifies the occupant of the response content represented in the response information generated by the automatic response device 5. To this end, the notification device 4 has, for example, at least one of a speaker or a display device. When the notification device 4 receives a notification signal representing the response content to the occupant from the automatic response device 5, the notification device 4 notifies the driver of the response content by voice from the speaker or by displaying a message, image, or video on the display device. Note that the display device or speaker of the notification device 4 may be attached to each seat facing the occupant seated in that seat. In this case, the display device provided in the seat where the occupant who spoke may display the response content, or the speaker provided in the seat where the occupant who spoke may output a voice representing the response content.

[0018] The automatic response device 5 generates response information in response to the content of a speech made by any of the occupants of the vehicle 1, notifies the occupants of the vehicle 1 of the generated response information via the notification device 4, or controls any of the devices of the vehicle 1 in accordance with the response information.

[0019] Fig. 2 is a hardware configuration diagram of the automatic answering device 5. As shown in Fig. 2, the automatic answering device 5 has a communication interface 21, a memory 22, and a processor 23. The communication interface 21, the memory 22, and the processor 23 may each be configured as separate circuits, or may be configured integrally as a single integrated circuit.

[0020] The communication interface 21 has an interface circuit for connecting the automatic answering device 5 to other devices in the vehicle. The communication interface 21 passes the in-vehicle images received from the camera 2 and the audio signals received from each of the microphones 3-1 to 3-n to the processor 23. The communication interface 21 also outputs a notification signal received from the processor 23 to the notification device 4, or outputs a control command received from the processor 23 to one of the devices in the vehicle.

[0021] The memory 22 is an example of a storage unit and includes, for example, a volatile semiconductor memory and a non-volatile semiconductor memory. The memory 22 stores various data used in the automatic response process executed by the processor 23. Specifically, the memory 22 stores parameters that define a generation model for generating response information. Furthermore, the memory 22 may temporarily store images of the interior of the vehicle received from the camera 2 and audio signals received from each of the microphones 3-1 to 3-n.

[0022] The processor 23 includes one or more central processing units (CPUs) and their peripheral circuits. The processor 23 may further include other arithmetic circuits such as a logic unit, a numerical calculation unit, or a graphics processing unit. The processor 23 executes the automatic response process.

[0023] 3 is a functional block diagram of the processor 23 relating to the automatic response process. The processor 23 has a location estimation unit 31, an identification unit 32, a response generation unit 33, a notification processing unit 34, and a control unit 35. Each of these units in the processor 23 is, for example, a functional module realized by a computer program running on the processor 23. Alternatively, each of these units in the processor 23 may be a dedicated arithmetic circuit provided in the processor 23.

[0024] The position estimation unit 31 estimates the position of the occupant who has spoken among the multiple occupants. In this embodiment, in which a microphone is provided for each seat, if the average volume of any of the audio signals generated by the microphones 3-1 to 3-n over a recent predetermined period (e.g., several seconds) exceeds the speech detection threshold, the position estimation unit 31 estimates that one of the occupants has spoken during that predetermined period. The position estimation unit 31 then estimates the position of the seat where the microphone with the largest average volume over that predetermined period is provided as the position of the occupant who has spoken.

[0025] Even in an embodiment in which the microphones 3-1 to 3-n are arranged in an array, if the average volume of any of the audio signals generated by the microphones exceeds the speech detection threshold during a recent predetermined period, the position estimation unit 31 estimates that one of the occupants has spoken during that predetermined period. The position estimation unit 31 calculates the phase difference between the audio signals from each microphone during that predetermined period and estimates the direction from which the audio arrived based on the calculated phase difference. The position estimation unit 31 then estimates the position of the seat closest to the direction from which the audio arrived, as the position of the occupant who spoke. In this case, the direction from the mounting position of each microphone to each seat may be stored in advance in the memory 22. The position estimation unit 31 then identifies the seat closest to the direction from which the audio arrived, among the directions to each seat, and estimates the identified seat position as the position of the occupant who spoke.

[0026] Alternatively, the position estimation unit 31 may estimate the position of the occupant who spoke based on the vehicle interior image. In this case, the position estimation unit 31 inputs multiple vehicle interior images in the order in which they were generated to a classifier that has been trained in advance to estimate the position of the occupant who spoke. If any occupant has spoken within a recent predetermined period, the classifier for position estimation outputs the position of the occupant who spoke. The classifier for position estimation is configured as a deep neural network (DNN) having a recursive structure such as a recurrent neural network (RNN) or a long short-term memory (LSTM).

[0027] The position estimation unit 31 generates position information representing the estimation result of the position of the occupant who made the utterance, and outputs the generated position information to the identification unit 32 and the response generation unit 33. The position information includes a character string representing the position of the occupant who made the utterance (for example, a character string representing the seat of the occupant who made the utterance, such as the driver's seat or the passenger seat), or a vector representing the position of the occupant who made the utterance. When the position of the occupant who made the utterance is represented by a vector, for example, the vector includes different elements for each seat, and is generated so that the value of the element corresponding to the seat in which the occupant who made the utterance is seated is different from the value of the element corresponding to the other seats.

[0028] The identification unit 32 identifies a partial area in the vehicle interior image that corresponds to the position of the occupant who spoke, which is indicated by the position information. For example, for each seat, a position and range on the vehicle interior image that is set to include the occupant sitting in that seat are stored in advance in the memory 22. The identification unit 32 then reads from the memory 22 the position and range for the seat that corresponds to the position of the occupant who spoke, which is indicated by the position information, and identifies the area specified by the read position and range as the partial area that corresponds to the position of the occupant who spoke.

[0029] Alternatively, the identification unit 32 may detect areas in the vehicle interior image where each occupant is represented by inputting the vehicle interior image into a classifier that has been trained in advance to detect occupants. The identification unit 32 then identifies, among the areas in which each occupant is represented, an area corresponding to the position of the occupant who spoke, as a partial area corresponding to the position of the occupant who spoke, as indicated by the position information. In this case, the position of a reference point corresponding to each seat in the vehicle interior image is pre-stored in the memory 22. The identification unit 32 then identifies, among the areas in which each occupant is represented, an area closest to the reference point for the seat corresponding to the position of the occupant who spoke, as indicated by the position information, as a partial area corresponding to the position of the occupant who spoke. The classifier for occupant detection may be configured as a DNN with a convolutional neural network (CNN)-type architecture, such as a Single Shot MultiBox Detector, or a DNN with an attention mechanism, such as a Vision Transformer. Alternatively, the classifier for occupant detection may be configured as a classifier based on a machine learning method other than a DNN, such as an AdaBoost classifier.

[0030] The identification unit 32 notifies the response generation unit 33 of the position and range of the partial region in which the occupant who made the utterance is expressed.

[0031] The response generation unit 33 generates response information by inputting location information representing the position of the occupant who made the utterance, audio information representing the content of the utterance, and a partial area representing the occupant who made the utterance into a generation model that has been trained in advance to generate response information to the content of the utterance.

[0032] To generate speech information representing the content of an utterance, the response generation unit 33 inputs speech signals generated by each of the microphones 3-1 to 3-n, whose average volume value for a recent predetermined period exceeds a speech detection threshold, into a speech recognition model to recognize the content of the utterance represented in the speech signal and generate a character string representing the content of the utterance as speech information. Such a speech recognition model may be configured, for example, as a DNN with an attention mechanism, or a DNN with a recursive structure such as an RNN or LSTM. Alternatively, the speech recognition model may be configured as a GMM-HMM based on a Gaussian mixture distribution and a hidden Markov model, or a DNN-HMM based on a DNN and a hidden Markov model. The response generation unit 33 may recognize the content of the utterance represented in the speech signal by dividing the speech signal into frames having a predetermined time length, extracting speech features for each frame, and inputting the features for each frame to the speech recognition model in chronological order. The features for each frame may be, for example, a predetermined element of the cepstrum of that frame.

[0033] The response generation unit 33 represents the combination of the content of the utterance and the position of the occupant who made the utterance in a single string by adding a string representing the position of the occupant who made the utterance, which is indicated by the position information, before or after the string representing the content of the utterance indicated by the audio information.

[0034] Furthermore, in order to input the partial area representing the occupant who spoke into the generative model, the response generation unit 33 crops the partial area indicated by the position and range notified by the identification unit 32 from the vehicle interior image. Alternatively, the response generation unit 33 may mask the area other than the partial area by rewriting the value of each pixel in the vehicle interior image other than the partial area indicated by the position and range notified by the identification unit 32 to a predetermined value.

[0035] In this embodiment, the generative model is configured as a Vision Language Model (VLM). The VLM serving as the generative model is configured, for example, as a combination of an image encoder that encodes an input image and a large-scale language model (LLM) in which multiple blocks, each including an attention layer and a feed forward layer, are stacked. The response generation unit 33 inputs a cropped partial region or an interior image of the vehicle in which areas other than the partial region are masked, into the image encoder. Furthermore, the response generation unit 33 inputs a character string representing the content of the utterance and the position of the occupant who made the utterance into the LLM. As a result, the generative model outputs text data representing the response content as response information. In this way, by inputting the position information along with the voice information representing the content of the utterance into the generative model, the generative model can generate response information representing the response content according to the position of the occupant who made the utterance. Furthermore, by using not only the voice information and the position information but also the partial region on the interior image in which the occupant who made the utterance is depicted to generate the response information, the generative model can determine the state of the occupant who made the utterance using the interior image. Therefore, the generative model can generate response information representing a more appropriate response content to the occupant who made the utterance. For example, even if the content of an utterance requests that a window or door be opened, if the occupant who made the utterance is a small child, the generative model can generate response information that does not unlock the window or door but includes a soothing message such as, "Please wait a moment." Furthermore, when generating response information to display an image on a display device installed at the position of the occupant who made the utterance, the generative model can determine the content of the image to be displayed taking into account the state of the occupant who made the utterance.

[0036] For example, if the occupant who made the utterance is seated in the rear right seat, a subregion representing the occupant seated in the rear right seat is input to the generative model. If the content of the utterance is "hot," a character string such as "rear right seat hot" is input to the generative model. The generative model then outputs text data such as "Turn up the air conditioning in the rear right seat" as response information.

[0037] The response information may include information for controlling the vehicle 1 itself or a device installed in the vehicle 1.

[0038] According to a modified example, the generative model may be provided with an input layer for inputting position information, separate from the image encoder and LLM, to which a vector representing the position of the occupant who spoke is input. In this case, only a character string corresponding to the speech information is input to the LLM, and the vector representing the position of the occupant who spoke is input to the input layer for inputting position information. The vector representing the position of the occupant who spoke, input to the input layer, is then incorporated into the LLM in a block among the multiple blocks of the LLM that has a cross attention mechanism that calculates cross attention between the vector and the output from the previous block, just like the subregion input to the image encoder.

[0039] The response generation unit 33 outputs the generated response information to the notification processing unit 34 and the control unit 35 together with the location information.

[0040] The notification processing unit 34 outputs the response information via the notification device 4. To do so, the notification processing unit 34 generates a notification signal representing the response content included in the response information and outputs the generated notification signal to the notification device 4 via the communication interface 21. For example, the notification processing unit 34 generates a voice signal representing the response content as the notification signal according to a predetermined voice synthesis method based on text data representing the response content included in the response information. The notification processing unit 34 then outputs the notification signal to a speaker included in the notification device 4, thereby causing the speaker to output a voice representing the response content. Alternatively, the notification processing unit 34 includes text data representing the response content in the notification signal. The notification processing unit 34 then causes a display device included in the notification device 4 to display the text data representing the response content.

[0041] Furthermore, if the notification device 4 has a display device or speaker provided for each seat, the notification processing unit 34 outputs a notification signal to the display device or speaker provided for the seat where the occupant who made the utterance is seated, as indicated by the position information, via the communication interface 21. The notification processing unit 34 then causes the display device provided for the seat where the occupant who made the utterance is seated to display text data representing the response content, or causes the speaker provided for that seat to output audio representing the response content. This allows the notification processing unit 34 to make an appropriate response even if the response content is intended only for the occupant who made the utterance.

[0042] In addition, if the response information generated by the generative model represents the control content of the device of the vehicle 1, the processing by the notification processing unit 34 may be omitted.

[0043] The control unit 35 controls the device specified in the response included in the response information in accordance with the response content. The control unit 35 determines the device to be controlled and the control command by referring to a control reference table that shows the correspondence between text data indicating the response content included in the response information, the device to be controlled (including the vehicle 1 itself), and the control command for executing the control. The control unit 35 then outputs the identified control command to the electronic control unit (ECU) of the device to be controlled via the communication interface 21.

[0044] The device to be controlled may be a device located at the position of the occupant who spoke. In this case, the control unit 35 identifies the position of the occupant who spoke by referring to the location information, and identifies the device to be controlled based on the text data representing the response content as described above. The control unit 35 then outputs a control command and a signal representing the position of the device to the ECU of the device to be controlled, thereby controlling the device located at the position of the occupant who spoke in accordance with the response information.

[0045] For example, suppose that the occupant who made the utterance is seated in the rear right seat, and in response to the utterance "It's hot," response information is generated indicating that the airflow to the rear right seat should be increased by a predetermined amount. In this case, the control unit 35 outputs a control command to the ECU that controls the air conditioning device to increase the airflow from the air outlet closest to the rear right seat by a predetermined amount.

[0046] The devices controlled according to the position of the occupant who spoke may include any of the window, door lock, interior light, and seat in addition to the air conditioning system. The control unit 35 controls the devices according to the response by opening or closing the window closest to the position of the occupant who spoke, locking or unlocking the door closest to the position of the occupant who spoke, turning on or off the interior light closest to the position of the occupant who spoke, or adjusting the seat position of the seat in which the occupant who spoke is sitting.

[0047] Furthermore, if the response content indicated in the response information relates to the driving control of vehicle 1, the control unit 35 may output a control command representing the response content to the ECU that controls the driving of vehicle 1 only if the position of the occupant who spoke is in the driver's seat, i.e., only if the occupant who spoke is the driver.

[0048] For example, suppose the content of the utterance is "Go faster." In this case, only when the position of the occupant who made the utterance, indicated by the position information, indicates the position of the driver's seat, the control unit 35 outputs a control command to increase the target speed of the vehicle 1 by a predetermined amount to the ECU that controls the driving of the vehicle 1. On the other hand, when the position of the occupant who made the utterance, indicated by the position information, indicates a seat other than the driver's seat, for example, the passenger seat, the control unit 35 does not output a control command according to the content of the response. This prevents the driving of the vehicle 1 from being inadvertently controlled by an occupant other than the driver, and as a result, prevents the behavior of the vehicle 1 from becoming unstable. Note that the control unit 35 may determine whether the device to be controlled, indicated by the text data representing the content of the response, is an ECU that controls the driving of the vehicle 1, by referring to a control reference table.

[0049] Furthermore, if the text data representing the response does not contain any of the words that identify the controlled device and are registered in the control reference table, the response information does not control the device, and in this case, the control unit 35 does not output a control signal.

[0050] 4 is an explanatory diagram of the relationship between input and response information of a generative model according to this embodiment. In this embodiment, text data 401 including the content of an utterance ("It's hot") and the position information of the occupant who made the utterance ("Rear right seat"), and a partial region 402 representing the occupant in the rear right seat extracted from an image of the interior of a vehicle are input to generative model 400. Then, with reference to text data 401 and partial region 402, generative model 400 outputs text data 403 representing the content of a response to the utterance ("I'll turn up the air conditioning in the rear right seat").

[0051] 5 is a flowchart showing the automatic response process according to this embodiment. The processor 23 executes the automatic response process according to this flowchart.

[0052] The position estimation unit 31 estimates the position of the occupant who has spoken and generates position information indicating the position (step S101). The identification unit 32 identifies a partial area corresponding to the position of the occupant who has spoken in the in-vehicle image (step S102).

[0053] The response generation unit 33 generates voice information representing the content of the occupant's speech based on the voice signal generated by any of the microphones 3-1 to 3-n (step S103).The response generation unit 33 then generates the response information by inputting the voice information, the position information, and a partial area corresponding to the position of the occupant who made the speech into a generative model (step S104).

[0054] The notification processing unit 34 notifies all occupants or the occupant who spoke of the response content included in the response information via the notification device 4 (step S105). In addition, the control unit 35 controls the vehicle 1 itself or devices installed in the vehicle 1, particularly devices installed at the position of the occupant who spoke, in accordance with the response content (step S106).

[0055] As described above, this automatic answering device estimates the location of the occupant who spoke among multiple occupants of the vehicle. Then, this automatic answering device generates response information by inputting location information indicating the location of the occupant who spoke and voice information indicating the content of the occupant's speech into a generative model that has been trained in advance to generate response information to the content of the occupant's speech. Therefore, this automatic answering device can provide a more appropriate response to the occupant who spoke among multiple occupants in the vehicle.

[0056] According to a modified example, the response generation unit 33 may generate response information by inputting location information representing the location of the occupant who made the utterance and voice information representing the content of the utterance into the generative model, without inputting a partial region representing the occupant who made the utterance into the generative model. In this case, an LLM is used as the generative model. Furthermore, the processing of the identification unit 32 may be omitted. Even in this modified example, since location information is input into the generative model, the automatic response device can provide an appropriate response to the occupant who made the utterance. Furthermore, according to this modified example, since processing related to the in-vehicle image is omitted, the amount of calculation is reduced compared to the above embodiment.

[0057] In the above embodiment or each modification, a server capable of communicating via an in-vehicle wireless communication terminal (not shown) may have a generative model and generate response information by executing processing by the response generation unit 33. In this case, the automatic response device 5 generates a question signal including text data representing the position and speech content of the occupant who spoke and a partial region cropped from the in-vehicle image representing the occupant who spoke, and transmits the generated question signal to the server via the wireless communication terminal. The server then generates response information by inputting the text data and partial region included in the received question signal into the generative model and transmits the generated response information to the wireless communication terminal of the vehicle 1 that transmitted the question signal. Note that if the partial region is not input into the generative model, the automatic response device 5 does not need to include the partial region in the question signal. The automatic response device 5 simply executes processing by the notification processing unit 34 and the control unit 35 according to the response information received via the wireless communication terminal. According to this modification, the automatic response device 5 can use a larger-scale generative model than a generative model realized by in-vehicle hardware resources to generate response information, thereby generating more appropriate response information.

[0058] According to another modification, the processing of either the notification processing unit 34 or the control unit 35 may be omitted.

[0059] According to yet another variant, the generation model may be configured to directly input an audio signal that is estimated to contain the content of an utterance along with position information, and may be trained in advance to recognize the content of the utterance from the audio signal and generate response information according to the content of the utterance and the position of the occupant who made the utterance.

[0060] A computer program for realizing the automatic response process according to the above embodiment or modification may be provided in a form recorded on a computer-readable portable recording medium.

[0061] As described above, those skilled in the art can make various modifications to the embodiments within the scope of the present invention. [Explanation of symbols]

[0062] REFERENCE SIGNS LIST 1 vehicle, 2 camera, 3-1 to 3-n microphones, 4 notification device, 5 automatic answering device, 21 communication interface, 22 memory, 23 processor, 31 position estimation unit, 32 identification unit, 33 response generation unit, 34 notification processing unit, 35 control unit

Claims

1. a position estimation unit that estimates the position of a occupant who has spoken among a plurality of occupants of the vehicle; a response generation unit that generates response information by inputting position information indicating the position of the occupant who made the utterance and voice information indicating the content of the utterance into a generation model that has been trained in advance to generate response information in response to the content of the utterance; An automatic answering device having:

2. 2. The automatic answering device according to claim 1, further comprising a control unit that controls a device provided at the position of the occupant who made the speech based on the response information.

3. The vehicle interior image may further include an identification unit that identifies a partial area corresponding to the position of the occupant indicated by the position information, from within the vehicle interior image that represents the interior of the vehicle; The automatic answering device according to claim 1 , wherein the response generation unit generates the response information by further inputting the partial region into the generative model.

4. Estimating the position of the occupant who spoke among a plurality of occupants of the vehicle; generating response information by inputting position information indicating the position of the occupant who made the utterance and voice information indicating the content of the utterance into a generation model that has been trained in advance to generate response information in response to the content of the utterance; An automatic response method including:

5. Estimating the position of the occupant who spoke among a plurality of occupants of the vehicle; generating response information by inputting position information indicating the position of the occupant who made the utterance and voice information indicating the content of the utterance into a generation model that has been trained in advance to generate response information in response to the content of the utterance; An automatic response computer program that causes a computer to do the following:

Images