Industrial vehicles, alarm sound volume estimation system, and alarm sound volume estimation program

Abstract

To prevent people from being surprised by notification sounds and falling over while exhibiting effect of the notification sound in also response to difference in each person's degree of surprise at sound volume.SOLUTION: An industrial vehicle comprises: a speaker 12; a learned model 20 having a sound volume estimation part 25; a sound volume change part 26 which changes sound volume of the speaker 12; a registration part 16 in which person information corresponding to a plurality of persons is registered; and a person specification part 17 which specifies person information corresponding to a notification object person. The sound volume estimation part 25 previously performs machine learning by teacher data that uses a notification distance from the speaker 12 to the person, person information of the person and sound volume of the notification sound as input data and uses a surprise degree of the person who has been notified as output data. When the person information and the notification distance is inputted to the sound estimation part, the sound estimation part estimates such a sound volume that the person corresponding to the inputted person information becomes a prescribed surprise degree or less. The sound volume change part 26 changes sound volume of the speaker 12 to the estimated sound volume or less.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a volume estimation system used in industrial vehicles, an industrial vehicle equipped with the system, and an alarm sound volume estimation program.

Background Art

[0002] In recent years, the proportion of workers aged 60 or older among the casualties due to industrial accidents has been increasing. Among the industrial accidents involving the elderly, there are falls during work. Therefore, it is necessary to construct a working environment that reduces the falls of the elderly.

[0003] Industrial vehicles such as forklifts are equipped with speakers that notify their own approach during travel, and avoid contact with pedestrians by the alarm sound (see, for example, Patent Document 1 and Patent Document 2). However, many of the elderly are hard of hearing and tend to have difficulty hearing the alarm sound. Therefore, it is conceivable to increase the alarm sound, but if the alarm sound is too loud, there is a risk of startling pedestrians and causing them to fall. In addition, the degree of surprise (hereinafter referred to as "surprise degree") with respect to the volume of the alarm sound varies among individuals.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] Therefore, the problem to be solved by the present invention is to provide an industrial vehicle that can prevent people from being startled by the alarm sound and falling, corresponding to the differences in the surprise degree with respect to the volume of each person.

Means for Solving the Problems

[0006] To solve the above problems, the industrial vehicle according to the present invention is A speaker that emits a notification sound to people, A trained model having a volume estimation unit, A volume control unit that changes the speaker volume, A registration section where person information corresponding to multiple people is registered, It includes a person identification unit that identifies person information corresponding to the person to be notified, The volume estimation unit has been pre-trained using training data, which includes the notification distance from the speaker to the person, the person's information, and the volume of the notification sound as input data, and the person's level of surprise as output data. When person information and notification distance are input, the unit estimates the volume at which the person corresponding to the input information will be below a predetermined level of surprise. The volume adjustment unit changes the speaker volume to a level below the estimated volume.

[0007] The above industrial vehicle is preferably, It also features a camera that photographs people and generates images of people, The registration section also contains facial images corresponding to each person's information. The person identification unit identifies the person information corresponding to the person to be reported by referring to the face image included in the person image and the face image corresponding to each person's information.

[0008] The above industrial vehicle is preferably, It also features a camera that photographs people and generates images of people, The trained model further includes a distance estimation unit, which has been pre-trained using training data that takes a person image as input data and the notification distance from the speaker to the person as output data. When a person image is input, it estimates the notification distance from the speaker to the person.

[0009] The above industrial vehicle is preferably, It also features a camera that photographs people and generates images of people, Speakers have directionality, The volume estimation unit identifies the person image closest to the speaker based on the notification distance corresponding to each person image, and outputs the estimated volume corresponding to that person image to the volume adjustment unit.

[0010] The above industrial vehicle is preferably, It also features a camera that photographs people and generates images of people, The trained model further includes a pose estimation unit, The posture estimation unit has been pre-trained using training data, which takes human images as input data and human posture types as output data. When a human image is input, it estimates the type of human posture. The volume estimation unit, upon receiving input about the type of person's posture, further estimates a volume level that will be below a predetermined level of surprise, based on that type of posture.

[0011] The above industrial vehicle is preferably, It also features a camera that photographs people and generates images of people, The trained model further includes a direction estimation unit, The orientation estimation unit has been pre-trained using training data that takes a person's image as input data and the orientation of the person's face as output data. When a person's image is input, it estimates the orientation of the person's face. The volume estimation unit, upon receiving the direction of a person's face, further estimates a volume level that will be below a predetermined level of surprise based on the direction of the person's face.

[0012] The above industrial vehicle is preferably, It is further equipped with an ambient sound acquisition unit that acquires ambient sounds, The volume estimation unit, upon receiving ambient sound input, further estimates a volume level that will be below a predetermined level of surprise, based on the ambient sound.

[0013] The above industrial vehicle is preferably, It also features a camera that photographs people and generates images of people, The trained model further includes a surprise level estimation unit, The surprise degree estimation unit performs machine learning in advance using teacher data that takes the person image of the notified person and the personal information of the notified person as input data and outputs the surprise degree as output data. When the person image of the notified person and the personal information of the notified person are input by the notification sound, it estimates the surprise degree corresponding to the personal information. The volume estimation unit relearns the volume that becomes below a predetermined surprise degree based on the person image of the person before notification, the personal information of the notified person, the notification distance, the volume at the time of notification in the notification sound, and the surprise degree of the notified person.

[0014] The industrial vehicle preferably has a tone color storage unit that stores a plurality of tone colors, and further includes a tone color control unit that determines the tone color of the speaker to be notified based on the personal information specified by the person specifying unit and switches to the determined tone color. The volume estimation unit is configured to perform machine learning in advance for each tone color and estimate the volume at which the person corresponding to the input personal information becomes below a predetermined surprise degree corresponding to the determined tone color.

[0015] The industrial vehicle preferably has a position detection unit that detects its own position, and further includes a tone color control unit that determines the tone color of the speaker to be notified based on the detected position and switches to the determined tone color. The volume estimation unit is configured to perform machine learning in advance for each tone color and estimate the volume at which the person corresponding to the input personal information becomes below a predetermined surprise degree for each tone color.

[0016] The industrial vehicle preferably The stored tone colors include buzzer sounds, beep sounds, and melodies.

[0017] In order to solve the above problems, the notification sound volume estimation system according to the present invention is a notification sound volume estimation system used for an industrial vehicle having a speaker that outputs a notification sound to a person, A trained model having a volume estimation unit, Volume adjustment section, A registration section where information on multiple individuals is registered, It includes a person identification unit that identifies person information corresponding to the person to be notified, The volume estimation unit has been pre-trained using training data, which includes the notification distance from the speaker to the person, the person's information, and the volume of the notification sound as input data, and the person's level of surprise as output data. When person information and notification distance are input, the unit estimates the volume at which the person corresponding to the input information will be below a predetermined level of surprise. The volume adjustment unit changes the speaker volume to a level below the estimated volume.

[0018] To solve the above problems, the notification sound volume estimation program according to the present invention is: A notification sound volume estimation program used in a notification sound volume estimation system, wherein the notification sound volume estimation system is a system used in an industrial vehicle having a speaker that outputs a notification sound to a person, and is equipped with a computer. The notification sound volume estimation program is pre-trained on a computer using training data that takes the notification distance from the speaker to the person, the person's information, and the volume of the notification sound as input data, and the person's level of surprise upon notification as output data. When the person's information and the notification distance are input, the program is executed as a volume estimation unit that estimates the volume at which the person corresponding to the input person's information will be below a predetermined level of surprise. [Effects of the Invention]

[0019] The industrial vehicle according to the present invention can provide an effective warning sound while also addressing the differences in how each person is startled by the sound, thereby preventing people from falling over due to being startled by the sound. [Brief explanation of the drawing]

[0020] [Figure 1] This is a schematic diagram showing an industrial vehicle and its surroundings according to one embodiment of the present invention. [Figure 2] This is a block diagram of the control unit of an industrial vehicle. [Figure 3] This diagram shows the operation of each estimation unit in the trained model. [Figure 4] This figure shows examples of human postures. [Figure 5] This diagram shows the machine learning process for the volume estimation unit. [Figure 6] This diagram shows the operation of the volume estimation unit. [Figure 7] This flowchart illustrates how industrial vehicles alert people to their approach. [Modes for carrying out the invention]

[0021] Hereinafter, an embodiment of the present invention relating to an industrial vehicle, a notification sound volume estimation system, and a notification sound volume estimation program will be described with reference to the attached drawings.

[0022] Figure 1 is a schematic diagram showing an industrial vehicle 1 and its surroundings according to one embodiment of the present invention. As shown in Figure 1, the industrial vehicle 1 is a forklift and is moving forward along a passageway R. In front of the industrial vehicle 1 are pedestrians H (hereinafter sometimes referred to as "person H") 1, H2, and H3, and a worker H4 is working beside the passageway R.

[0023] Industrial vehicle 1 is an autonomously operating unmanned transport vehicle, but this is merely one example; the industrial vehicle 1 according to the present invention may also be a manned industrial vehicle 1. Furthermore, industrial vehicle 1 does not have to be a forklift; for example, it may be an in-plant transport vehicle without forks 11, an in-plant towing vehicle, a straddle carrier, etc., and is not particularly limited.

[0024] As shown in Figure 1, the industrial vehicle 1 comprises a body 10 and forks 11. The forks 11 are configured to be able to move up and down, thereby lifting and lowering loads.

[0025] As shown in Figure 2, the industrial vehicle 1 includes a speaker 12, a camera 13, and a control unit 15.

[0026] Speaker 12 is directional and outputs a notification sound V to person H. Speaker 12 is composed of multiple parametric speakers and is configured to expand or contract its directional range depending on the distance to person H. This prevents speaker 12 from outputting the notification sound V to anyone other than the target person H. Therefore, it is possible to prevent the notification sound V from being output to workers H4 who are not located in the direction of travel of the industrial vehicle 1, thereby preventing startling workers H4. Since the industrial vehicle 1 moves forward and backward, speaker 12 is configured to output the notification sound V at least in the forward and backward directions of the industrial vehicle 1. Furthermore, the number of speakers 12 and the configuration of the speaker 12's directionality are not particularly limited.

[0027] The specific sound of notification sound V is not limited. For example, notification sound V may be a buzzer sound, a beep sound, a specific melody, or a human voice.

[0028] Speaker 12 does not need to output the notification sound V unless a person H is detected by the industrial vehicle 1. Alternatively, speaker 12 may be configured to output the notification sound V at a predetermined volume only at intersections of the passage R, even if no person H is detected. In this embodiment, speaker 12 is configured to output the notification sound V in the direction of the detected person H when a person H is detected by camera 13 (i.e., when a person H is photographed). If multiple people H are detected simultaneously, speaker 12 is configured to output the notification sound V towards the person H closest to it.

[0029] Furthermore, the industrial vehicle 1 has a melody output unit (not shown) that emits a melody around the vehicle while it is in motion. The volume of the melody is set to a level that will not startle person H.

[0030] Camera 13 is configured to capture images of the area around the vehicle body 10, captures images of people H, and generates images of people HI. Camera 13 is composed of multiple cameras 13 and is configured to capture images of at least the front and rear of the industrial vehicle 1.

[0031] The control unit 15 has a computer comprising a storage device, memory, and an arithmetic unit, and the storage device stores a notification sound volume estimation program. The notification sound volume estimation program causes the computer to run as a person identification unit 17, a trained model 20, and a volume modification unit 26, which will be described later.

[0032] The control unit 15, as shown in Figure 2, has a functional configuration that includes a registration unit 16, a person identification unit 17, a learned model 20, and a volume adjustment unit 26.

[0033] The registration unit 16 registers the person information PI and facial images of multiple people H. In this embodiment, the person information PI is an identifier corresponding to each person H.

[0034] The person identification unit 17 identifies a face image corresponding to a face image included in the person image HI from among multiple face images registered in the registration unit 16, and identifies a person information PI (identifier in this embodiment) corresponding to the identified face image. The person identification unit 17 transmits the person information PI (identifier) ​​corresponding to each person image HI to the volume estimation unit 25.

[0035] <Trained Model Configuration> The trained model 20 includes a surprise level estimation unit 21, a distance estimation unit 22, a posture estimation unit 23, a direction estimation unit 24, and a volume estimation unit 25.

[0036] The surprise level estimation unit 21 is, for example, a neural network that has been pre-trained by machine learning using training data in which the person image HI of the notified person H is input data and the surprise level Q is output data. As a machine learning method, for example, it may be a machine learning method that learns the correlation between the difference between the person image HI of person H before being notified and the person image HI of person H after being notified, and the surprise level Q. Alternatively, the surprise level estimation unit 21 may be configured to machine learn by using images of faces expressing emotions such as surprise or anxiety as training data, and to estimate the surprise level Q of the notified person H when the person image HI of the notified person H is input. Through such machine learning, as shown in Figure 3A, the surprise level estimation unit 21 estimates the surprise level Q of the notified person H when the person image HI of the notified person H is input by the notification sound V. The surprise level Q may be output as a numerical value between, for example, 0.01 and 1.0.

[0037] In this invention, the "predetermined degree of surprise RQ" is defined as a degree of surprise Q that is sufficient to prevent one from being so surprised that they fall over, drop something they are holding, or make a mistake in their work. This predetermined degree of surprise RQ may be a numerical value of the degree of surprise Q, such as 0.7.

[0038] The distance estimation unit 22 is, for example, a neural network that has been pre-trained by machine learning to recognize the correlation between human image HI as input data and the notification distance DN from speaker 12 to human H as output data. As a result, as shown in Figure 3B, when human image HI is input, the distance estimation unit 22 estimates the notification distance DN from speaker 12 to human H and the direction of human H from speaker 12. The estimated notification distance DN and direction from speaker 12 corresponding to each human image HI are output to the volume estimation unit 25. If no human image HI is input, the distance estimation unit 22 does not output the notification distance DN to the volume estimation unit 25.

[0039] The posture estimation unit 23 is, for example, a neural network that has been pre-trained by machine learning using training data in which the input data is a person image HI and the output data is the type of person H's posture AP. As a result, as shown in Figure 3C, when the posture estimation unit 23 receives a person image HI as input, it estimates the type of person H's posture AP (hereinafter sometimes simply referred to as "person H's posture AP"). The estimated person H's posture AP corresponding to each person image HI is output to the volume estimation unit 25.

[0040] Figure 4 shows examples of postures AP of person H. As shown in Figure 4, the types of postures AP of person H include, for example, an upright posture (see Figure 4A), a posture carrying luggage (see Figure 4B), a walking posture (see Figure 4C), and a working posture (see Figure 4D).

[0041] The orientation estimation unit 24 is, for example, a neural network that has been pre-trained using training data that takes a person image HI as input data and the orientation DF of the person H's face as output data. As a result, as shown in Figure 3D, when a person image HI is input to the orientation estimation unit 24, it estimates the orientation DF of the person H's face. The orientation DF of the person H's face can be, for example, the direction of the industrial vehicle 1, the direction opposite to the industrial vehicle 1, the direction of the load that the person H is carrying, or the direction of the work tool while working. The estimated orientation DF of the person H corresponding to each person image HI is output to the volume estimation unit 25.

[0042] The volume estimation unit 25 is, for example, a neural network, and as shown in Figure 5, it has been pre-trained by machine learning using training data that takes the notification distance DN from speaker 12 to person H, person information PI (identifier) ​​of person H, posture AP of person H, face orientation DF of person H, and the volume of notification sound V (hereinafter referred to as "notification volume") db as input data, and the degree of surprise Q as output data, to establish the correlation between these. As a result, as shown in Figure 6, when the person information PI (identifier), as well as the notification distance DN, person H's posture AP, and face orientation DF estimated by the distance estimation unit 22, orientation estimation unit 24, and posture estimation unit 23 as input, the volume estimation unit 25 estimates a notification volume db that is less than or equal to a predetermined degree of surprise RQ. Furthermore, the volume estimation unit 25 identifies the person image HI of the person H closest to the speaker 12 based on the notification distance DN corresponding to each person image HI, and outputs the estimated notification volume dB corresponding to that person image HI, and the direction of the person H closest to the speaker 12, to the volume adjustment unit 26.

[0043] In this embodiment, the volume estimation unit 25 does not output a notification volume dB to the volume adjustment unit 26 unless a notification distance DN is input (i.e., unless a person H is detected). The volume estimation unit 25 may be configured to estimate a notification volume dB of zero when the notification distance DN is less than or equal to a predetermined level of surprise RQ. This prevents the notification sound V from being output if the distance between the industrial vehicle 1 and person H is too close.

[0044] The volume adjustment unit 26 changes the notification volume dB of speaker 12 to a level below the notification volume dB estimated by the volume estimation unit 25, and also changes the output direction of speaker 12 towards the person H closest to speaker 12. By causing speaker 12 to output the notification sound V at a level below the estimated notification volume dB, the volume adjustment unit 26 can prevent person H from being startled and falling over by the notification sound V.

[0045] Furthermore, when a notification sound V is actually output, the volume estimation unit 25 uses the person information PI (identifier), notification distance DN, person H's posture AP, person H's face orientation DF, the actually output notification volume db, and the surprise level Q estimated by the surprise level estimation unit 21 as training data to relearn the correlations between these. In other words, in this embodiment, the volume estimation unit 25 takes the notification distance DN, person H's posture AP, face orientation DF, and the actually output notification volume db estimated by the person information PI (identifier), distance estimation unit 22, orientation estimation unit 24, and posture estimation unit 23 as input data, and uses the surprise level Q estimated by the surprise level estimation unit 21 as training data to relearn the correlations between these, thereby enabling more accurate estimation for each person information PI from now on.

[0046] The system comprising the aforementioned camera 13, registration unit 16, person identification unit 17, trained model 20, and volume adjustment unit 26 corresponds to the notification sound volume estimation system according to the present invention.

[0047] <Operation of industrial vehicles> Next, the operation of industrial vehicle 1 will be explained again, referring to Figure 7.

[0048] (1) While the industrial vehicle 1 is in motion, it outputs a melody using the melody output unit (see S1 (Step 1) in Figure 7). Until the industrial vehicle 1 photographs a person H with the camera 13, the notification volume dB is set to zero (see S2 in Figure 7).

[0049] (2) When the industrial vehicle 1 photographs a person H with the camera 13 (i.e., when it detects a person H) (Yes in S3 of Figure 7), the trained model 20 estimates the notification distance DN, the posture AP of person H, and the face orientation DF based on the person image HI (see S4 of Figure 7). Also, when the industrial vehicle 1 detects a person H, the person identification unit 17 identifies the face image and the corresponding identifier to determine which person the detected person H is (see S5 of Figure 7).

[0050] (3) Next, the industrial vehicle 1 uses the trained model 20 to estimate the alarm volume dB at which the identified person H will be less than or equal to a predetermined level of surprise RQ, based on the identifier of the identified person H, the estimated alarm distance DN, the posture AP of person H, and the orientation DF of the face (see S6 in Figure 7).

[0051] (4) Next, the industrial vehicle 1 changes the alarm volume dB of the speaker 12 using the volume change unit 26 so that it is less than or equal to the estimated alarm volume dB (see S7 in Figure 7).

[0052] (5) Next, the industrial vehicle 1 outputs a notification sound V towards the person H closest to the speaker 12 (pedestrian H1 in Figure 2) (see S8 in Figure 7).

[0053] When the industrial vehicle 1 is photographing a person H with the camera 13 (i.e., continuously detecting person H), it repeats steps S4 to S8 in Figure 7, automatically changing the alarm volume dB as the alarm distance DN, the posture AP of person H, and the face orientation DF change. Then, when the camera 13 stops photographing person H (i.e., when it stops detecting person H) (No. S3 in Figure 7), it changes the alarm volume dB back to zero (see S2 in Figure 7).

[0054] Industrial vehicle 1, with the above configuration, can prevent person H from falling due to being startled by the notification sound V. Furthermore, industrial vehicle 1, using a trained model 20, has learned not only the correlation between the notification distance DN, the notification volume db, and the degree of surprise Q, but also the correlation with person H's posture AP and face orientation DF. Therefore, it can prevent startling and causing fall of person H who is concentrating on work or person H who is looking in the opposite direction of industrial vehicle 1 by outputting an excessively loud notification volume db. Moreover, since industrial vehicle 1 has learned machine learning for each person information PI, it can estimate the notification volume db more appropriately. In addition, since industrial vehicle 1 estimates the notification distance DN, person H's posture AP, and face orientation DF based on the person image HI, there is no need to add special sensors.

[0055] Although one embodiment of the industrial vehicle, notification sound volume estimation system, and notification sound volume estimation program of the present invention has been described above, the present invention is not limited to the above embodiment. For example, the industrial vehicle, notification sound volume estimation system, and notification sound volume estimation program of the present invention may be implemented by appropriately combining the following modifications.

[0056] <Variation> The industrial vehicle 1 may be equipped with a known sensor capable of detecting the direction of a person H and the notification distance DN. In this case, the volume estimation unit 25 obtains the notification distance DN from this sensor, and the trained model 20 does not need to have a distance estimation unit 22.

[0057] • When each component of the trained model 20 performs machine learning, the human image HI used as input for each estimation may include multiple human image HIs. In other words, the human image HI according to the present invention is a concept that includes videos.

[0058] The input data for training the volume estimation unit 25 uses for machine learning only needs to include at least person information PI, notification distance DN, and notification volume db. In this case, when the volume estimation unit 25 receives person information PI and notification distance DN as input, it estimates a notification volume db that is less than or equal to a predetermined surprise level RQ.

[0059] The industrial vehicle 1 may further include an ambient sound acquisition unit for acquiring ambient sounds. In this case, the volume estimation unit 25 may further use training data that takes ambient sounds, person information PI, notification distance DN, person H's posture AP, person H's face orientation DF, and notification volume db as input data and surprise level Q as output data to pre-machine learn the correlation between these, and when person information PI, ambient sounds, notification distance DN, person H's posture AP, and face orientation DF are input, it estimates a notification volume db that is less than or equal to a predetermined surprise level RQ. Alternatively, in this case, the volume estimation unit 25 may pre-machine learn the correlation between at least ambient sounds, person information PI, notification distance DN, and notification volume db as input data and surprise level Q as output data, and estimate a notification volume db that is less than or equal to a predetermined surprise level RQ when ambient sounds, person information PI, and notification distance DN are input.

[0060] The industrial vehicle 1 may be configured to change the tone of the notification sound V to an appropriate tone for each person H to be notified. In this case, the industrial vehicle 1 further comprises a tone memory unit that stores multiple tones and a tone control unit. The tone control unit determines the tone of the notification speaker 12 based on the person information PI identified by the person identification unit 17, and switches the tone of the speaker 12 to the determined tone. The volume estimation unit 25 may also be pre-trained for each tone and configured to estimate the volume at which the person H corresponding to the input person information PI will be below a predetermined level of surprise RQ, corresponding to the determined tone. The tones stored in the tone memory unit may be, for example, a buzzer sound, a beep sound, or a melody, and the tone control unit may determine a buzzer sound for young people and a high-pitched beep sound for elderly people with hearing loss as the notification tone.

[0061] The industrial vehicle 1 may be configured to change its tone depending on its location within the facility. In this case, the industrial vehicle 1 further includes a location detection unit that detects its own position and a tone control unit. The tone control unit determines the tone of the speaker 12 to be used for notification based on the detected location information and switches the tone of the speaker 12 to the determined tone. In this case, the volume estimation unit 25 may have been pre-trained for each tone and be configured to estimate for each tone the volume at which a person H corresponding to the input person information PI will be below a predetermined level of surprise RQ.

[0062] • Person information PI may not only be an identifier but may also include gender, age, or other physical characteristics. In this case, the volume estimation unit 25 will estimate an appropriate volume level for the reported person H to be below a predetermined level of surprise RQ, not for each person H, but for each gender, age, and other physical characteristics or any of them.

[0063] The notification sound volume estimation system further comprises a server computer capable of communicating with the industrial vehicle 1, and the trained model 20 may be configured by the server computer. In this case, the industrial vehicle 1 communicates with the server computer to receive a notification sound volume db that is less than or equal to a predetermined surprise level RQ, and outputs a notification sound V at this notification sound volume db. [Explanation of Symbols]

[0064] 1. Industrial vehicles (forklifts) 10 car bodies 11 Forks 12 speakers 13 Cameras 15 Control Unit 16 Registration Department 17 Person identification part 20 Pre-trained Models 21. Estimation of Surprise Level 22 Distance Estimation Unit 23 Posture estimation section 24 Direction estimation unit 25. Volume estimation unit 26 Volume adjustment section AP Posture Types DF Face direction DN notification distance H people (pedestrians, workers) HI Human Image PI Person Information Q: Level of surprise R aisle RQ (Race Queen) - Determined level of surprise

Claims

1. A speaker that emits a notification sound to people, A trained model having a volume estimation unit, A volume control unit for changing the volume of the aforementioned speaker, A registration section where person information corresponding to multiple people is registered, It includes a person identification unit that identifies the person information corresponding to the person to be notified, The volume estimation unit has been pre-trained using training data in which the notification distance from the speaker to the person, the person's information, and the volume of the notification sound are input data, and the person's level of surprise upon notification is output data. When the person's information and the notification distance are input, the unit estimates the volume at which the person corresponding to the input person information will have a predetermined level of surprise or less. The volume adjustment unit adjusts the volume of the speaker to a level below the estimated volume in an industrial vehicle.

2. It also features a camera that photographs people and generates images of people, The aforementioned registration unit further registers facial images corresponding to each of the aforementioned person information. The industrial vehicle according to claim 1, wherein the person identification unit identifies the person information corresponding to the person to be notified by referring to the face image included in the person image and the face image corresponding to each of the person information.

3. It also features a camera that photographs people and generates images of people, The aforementioned trained model further comprises a distance estimation unit, The industrial vehicle according to claim 1, wherein the distance estimation unit is pre-trained using training data that takes the person image as input data and the notification distance from the speaker to the person as output data, and when the person image is input, it estimates the notification distance from the speaker to the person.

4. It also features a camera that photographs people and generates images of people, The speaker has directionality, The industrial vehicle according to claim 1, wherein the volume estimation unit identifies the person image closest to the speaker based on the notification distance corresponding to each person image, and outputs the estimated volume corresponding to that person image to the volume changing unit.

5. It also features a camera that photographs people and generates images of people, The aforementioned trained model further comprises a posture estimation unit, The posture estimation unit has been pre-trained using training data in which the human image is input data and the type of human posture is output data, and when the human image is input, it estimates the type of human posture. The industrial vehicle according to claim 1, wherein the volume estimation unit, upon input of the type of posture of the person, further estimates a volume that will be less than or equal to a predetermined level of surprise based on the type of posture of the person.

6. It also features a camera that photographs people and generates images of people, The aforementioned trained model further includes a direction estimation unit, The orientation estimation unit has been pre-trained using training data in which the human image is input data and the orientation of the human face is output data, and when the human image is input, it estimates the orientation of the human face. The industrial vehicle according to claim 1, wherein the volume estimation unit, upon input of the direction of the person's face, further estimates a volume that will be less than or equal to a predetermined level of surprise based on the direction of the person's face.

7. It is further equipped with an ambient sound acquisition unit that acquires ambient sounds, The industrial vehicle according to any one of claims 1 to 6, wherein the volume estimation unit, upon input of the ambient sound, further estimates a volume that will be less than or equal to a predetermined level of surprise based on the ambient sound.

8. It also features a camera that photographs people and generates images of people, The aforementioned trained model further includes a surprise level estimation unit, The surprise level estimation unit has been pre-trained using training data that takes the person image and the person information of the person being notified as input data and the surprise level as output data. When the person image and the person information of the person being notified by the notification sound are input, the unit estimates the surprise level corresponding to the person information. The industrial vehicle according to claim 1, wherein the volume estimation unit relearns a volume that is less than or equal to a predetermined level of surprise, based on the person image of the person before notification, the person information of the person who was notified, the notification distance, the volume of the notification sound at the time of notification, and the level of surprise of the person who was notified.

9. A tone memory unit that stores multiple timbres, The system further includes a tone control unit that determines the tone of the speaker to be used for notification based on the person information identified by the person identification unit, and switches the tone of the speaker to the determined tone. The industrial vehicle according to claim 1, wherein the volume estimation unit has been pre-machine-trained for each tone and is configured to estimate, in accordance with the determined tone, the volume at which a person corresponding to the input person information will fall below a predetermined level of surprise.

10. A tone memory unit that stores multiple timbres, A position detection unit that detects its own position, The system further includes a tone control unit that determines the tone of the speaker to be used for notification based on the detected position, and switches the tone of the speaker to the determined tone. The industrial vehicle according to claim 1, wherein the volume estimation unit has been pre-machine-trained for each tone and is configured to estimate for each tone the volume at which a person corresponding to the input person information will fall below a predetermined level of surprise.

11. The industrial vehicle according to claim 9 or 10, wherein the stored sounds include buzzer sounds, beeps, and melodies.

12. A notification sound volume estimation system used in industrial vehicles equipped with speakers that emit notification sounds to people, A trained model having a volume estimation unit, Volume adjustment section, A registration section where information on multiple individuals is registered, It includes a person identification unit that identifies the person information corresponding to the person to be notified, The volume estimation unit has been pre-trained using training data in which the notification distance from the speaker to the person, the person's information, and the volume of the notification sound are input data, and the person's level of surprise upon notification is output data. When the person's information and the notification distance are input, the unit estimates the volume at which the person corresponding to the input person information will have a predetermined level of surprise or less. The volume adjustment unit is a notification sound volume estimation system that adjusts the volume of the speaker to a level below the estimated volume.

13. A notification sound volume estimation program used in a notification sound volume estimation system, The aforementioned notification sound volume estimation system is a system used in industrial vehicles that have a speaker that outputs a notification sound to people, and comprises a computer. The aforementioned notification sound volume estimation program is provided to the computer, A sound volume estimation program is executed by a sound volume estimation unit that has been pre-machine-trained using training data in which the notification distance from the speaker to a person, the person's information, and the volume of the notification sound are input data, and the person's level of surprise upon notification is output data, and when the person's information and the notification distance are input, the unit estimates the volume at which the person corresponding to the input person information will have a predetermined level of surprise or less.

Images