Alcohol consumption detection device, alcohol consumption detection system, alcoholconsumption detection method, and alcoholconsumption detection program

Abstract

An alcohol consumption detection device (100) according to the present disclosure comprises: an alcohol consumption fluctuation information acquisition unit (110) that acquires alcohol consumption fluctuation information that indicates information about a subject, the information fluctuating due to a alcohol consumption of the subject; an over-detection factor information acquisition unit (120) that acquires over-detection factor information that indicates information about the factor of the fluctuation of the alcohol consumption fluctuation information and a factor other than the alcohol consumption of the subject; and an alcohol consumption detection unit (130) that estimates the alcohol consumption state of the subject on the basis of the alcohol consumption fluctuation information and the over-detection factor information. As a result, erroneous detection of alcohol consumption detection can be suppressed and the accuracy of alcohol consumption detection can be improved.

Description

Drinking Detection Device, Drinking Detection System, Drinking Detection Method, and Drinking Detection Program

[0001] The present disclosure relates to a drinking detection device, a drinking detection system, a drinking detection method, and a drinking detection program.

[0002] Since the sympathetic nerves of a living body such as a human are activated by drinking, the heart rate increases. Conventional drinking detection technologies that utilize this characteristic detect a drinking state when the heart rate increases (for example, the "drinking state detection device" described in Patent Document 1).

[0003] Japanese Patent Application Laid-Open No. 2009-219541

[0004] When the biological data value such as the heart rate increases, it may actually be due to factors other than drinking. When detecting a drinking state when the biological data value such as the heart rate increases, as in the drinking state detection device described in Patent Document 1, there is a problem of false detection where the biological data value increases due to factors other than drinking and is misdetected as being in a drinking state when it is not. <000001-0>The present disclosure has been made to solve the above problems, and an object thereof is to provide a drinking detection device that can suppress false detection of drinking and improve the accuracy of drinking detection.

[0006] The drinking detection device according to the present disclosure includes a drinking variation information acquisition unit that acquires drinking variation information indicating information of a subject that varies due to the subject's drinking behavior, an over-detection factor information acquisition unit that acquires over-detection factor information indicating a factor that causes the drinking variation information to vary and information of factors other than the subject's drinking behavior, and a drinking detection unit that estimates the subject's drinking state based on the drinking variation information and the over-detection factor information.

[0007] ~> According to the drinking detection device according to the present disclosure, false detection of drinking can be suppressed and the accuracy of drinking detection can be improved.

[0008] This is a configuration diagram showing a basic configuration example of an alcohol detection device according to Embodiment 1. This is a configuration diagram showing a basic configuration example of an alcohol detection system including the alcohol detection device according to Embodiment 1. This is a flowchart showing an example of a processing flow for an alcohol detection device according to Embodiment 1. This is a configuration diagram showing a detailed configuration example of the alcohol detection unit of an alcohol detection device according to Embodiment 1. This is a flowchart showing an example of a processing flow for an alcohol detection device having the alcohol detection unit shown in Figure 4. This is a configuration diagram showing a detailed configuration example of an alcohol detection system including the alcohol detection device according to Embodiment 1. This is a configuration diagram showing an example of a configuration when the alcohol detection device shown in Figure 6 is mounted on a vehicle. This is a configuration diagram showing a specific configuration example 1 of an alcohol detection system including the alcohol detection device shown in Figure 6. This is a flowchart showing an example of a processing flow for an alcohol detection device according to Configuration Example 1 of Embodiment 1. This is a configuration diagram showing a specific configuration example 2 of an alcohol detection system including the alcohol detection device shown in Figure 6. This is a flowchart showing an example of a processing flow for an alcohol detection device according to Configuration Example 2 of Embodiment 1. This is a configuration diagram showing a specific configuration example 3 of an alcohol detection system including the alcohol detection device shown in Figure 6. This is a flowchart showing an example of a processing flow for an alcohol detection device according to Configuration Example 3 of Embodiment 1. This is a configuration diagram showing a detailed configuration example of an alcohol detection system including an alcohol detection device according to Embodiment 2. This is a configuration diagram showing an example of the configuration when the alcohol detection device shown in Figure 14 is installed in a vehicle. This is a configuration diagram showing a specific example of the configuration of an alcohol detection system including the alcohol detection device shown in Figure 14. This is a flowchart showing an example of the processing flow of an alcohol detection device according to the configuration example of Embodiment 2. This is a configuration diagram showing a detailed example of the configuration of an alcohol detection system including the alcohol detection device according to Embodiment 3. This is a configuration diagram showing an example of the configuration when the alcohol detection device shown in Figure 18 is installed in a vehicle. This is a configuration diagram showing a specific example of the configuration of an alcohol detection system including the alcohol detection device shown in Figure 18. This is a flowchart showing an example of the processing flow of an alcohol detection device according to the configuration example of Embodiment 3. This is a configuration diagram showing a detailed example of the configuration of an alcohol detection system including the alcohol detection device according to Embodiment 4. This is a configuration diagram showing a detailed example of the configuration of the alcohol detection unit of the alcohol detection device according to Embodiment 4. This is a flowchart showing an example of the processing flow of an alcohol detection device having the alcohol detection unit shown in Figure 23.This is a schematic diagram showing example hardware configuration 1 of the alcohol detection device according to Embodiments 1 to 4. This is a schematic diagram showing example hardware configuration 2 of the alcohol detection device according to Embodiments 1 to 4.

[0009] The embodiments of this disclosure will be described below with reference to the accompanying drawings. Note that the drawings are schematic, and the relationships between sizes and positions shown in different drawings are not necessarily limited to those described and may be modified as appropriate. Furthermore, in the following description, similar components will be denoted by the same reference numerals, and their names and functions will be the same or similar. Therefore, detailed descriptions of them may be omitted.

[0010] Embodiment 1. Embodiment 1 will be described below with reference to the drawings. First, the basic form of the present disclosure will be described using Figures 1 and 2. Figure 1 is a configuration diagram showing a basic example of the configuration of the alcohol detection device 100 according to Embodiment 1. Figure 2 is a configuration diagram showing a basic example of the configuration of the alcohol detection system 1000 including the alcohol detection device 100 according to Embodiment 1.

[0011] Furthermore, in Embodiment 1, the subject may be an occupant of the moving object, for example, the driver of the moving object. The moving object may also be a vehicle.

[0012] The alcohol detection device 100 shown in Figure 1 includes an alcohol fluctuation information acquisition unit 110, an over-detection factor information acquisition unit 120, and an alcohol detection unit 130. The alcohol detection system 1000 shown in Figure 2 includes an alcohol detection device 100 and an output device 1100.

[0013] The output device 1100 outputs alcohol fluctuation information and false positive factor information to the alcohol detection device 100. Alcohol fluctuation information refers to information about subjects whose drinking behavior causes fluctuations. Details will be described later. False positive factor information refers to information about factors other than the subject's drinking behavior that cause fluctuations in alcohol fluctuation information. Details will be described later.

[0014] The alcohol consumption fluctuation information acquisition unit 110 acquires alcohol consumption fluctuation information that shows information about the subject that fluctuates due to the subject's drinking behavior. The alcohol consumption fluctuation information is information acquired from the subject's biological data, such as the subject's heart rate, pulse rate, blood pressure, level of alertness, and reaction time to the surroundings. The alcohol consumption fluctuation information acquisition unit 110 may acquire one type of alcohol consumption fluctuation information or multiple types of alcohol consumption fluctuation information. Although pulse rate and heart rate have strictly different meanings, they will be treated as synonymous in this explanation. Pulse wave rate will be referred to as heart rate from now on.

[0015] The over-detection factor information acquisition unit 120 acquires over-detection factor information that indicates factors other than the subject's drinking behavior that cause fluctuations in alcohol consumption fluctuation information. Here, "fluctuation in alcohol consumption fluctuation information" means that if the alcohol consumption fluctuation information is, for example, heart rate or blood pressure, then "increase in alcohol consumption fluctuation information," and if the alcohol consumption fluctuation information is, for example, alertness level or reaction time to the surroundings, then "decrease in alcohol consumption fluctuation information." When the alcohol consumption fluctuation information acquisition unit 110 acquires multiple types of alcohol consumption fluctuation information, the over-detection factor information acquisition unit 120 acquires over-detection factor information that indicates factors other than the subject's drinking behavior that cause fluctuations in at least one of the multiple types of alcohol consumption fluctuation information. For example, when the alcohol consumption fluctuation information acquisition unit 110 acquires heart rate and alertness level, which are alcohol consumption fluctuation information, the over-detection factor information acquisition unit 120 acquires over-detection factor information that indicates factors other than the subject's drinking behavior that cause fluctuations in at least one of heart rate or alertness level.

[0016] The alcohol detection unit 130 estimates the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110 and the over-detection factor information acquired by the over-detection factor information acquisition unit 120. For example, the alcohol detection unit 130 may estimate the subject's drinking status based on a rule when the value of the alcohol fluctuation information is above a threshold and the over-detection factor information is obtained. For example, the alcohol detection unit 130 may estimate that the subject is not drinking when the value of the first alcohol fluctuation information is above a threshold and the over-detection factor information A, which is a factor that causes the value of the first alcohol fluctuation information to fluctuate and is not related to the subject's drinking behavior, is acquired multiple times within a certain period, and the alcohol detection unit 130 may estimate that the subject is not drinking when the over-detection factor information B, which is part of the first over-detection factor information, is acquired at least once within a certain period. Furthermore, the alcohol detection unit 130 may stop estimating the subject's drinking status when the over-detection factor information acquisition unit 120 acquires over-detection factor information. Furthermore, when the alcohol consumption fluctuation information acquisition unit 110 acquires multiple types of alcohol consumption fluctuation information, the alcohol consumption detection unit 130 estimates the subject's drinking status based on the multiple types of alcohol consumption fluctuation information and false positive factor information, which indicates factors other than the subject's drinking behavior that cause fluctuations in at least one of the multiple types of alcohol consumption fluctuation information.

[0017] Furthermore, if the alcohol detection unit 130 determines that the subject is intoxicated, it may output a control signal to, for example, an alarm device 1200, which is an external device of the alcohol detection device 100, to issue an alarm. Alternatively, the alcohol detection unit 130 may output the result of determining the subject's intoxication status to the alarm device 1200, and the alarm device 1200 may issue an alarm based on the determination result. Note that the alarm device 1200 can be of any type as long as it can warn the subject that they are intoxicated.

[0018] Next, an example of the processing flow of the alcohol detection device 100 according to Embodiment 1 will be described with reference to Figure 3. Figure 3 is a flowchart showing an example of the processing flow of the alcohol detection device 100 according to Embodiment 1.

[0019] The alcohol detection device 100 shown in Figures 1 and 2 starts the process shown in Figure 3 when, for example, a control unit (not shown) receives a command from an external source to start processing. Specifically, for example, the alcohol detection device 100 starts the process shown in Figure 3 when the power source of the mobile vehicle starts up, or when it determines that an occupant has boarded the mobile vehicle. (Start)

[0020] In step ST1001, the alcohol consumption fluctuation information acquisition unit 110 acquires alcohol consumption fluctuation information that shows the subject's information that fluctuates due to the subject's drinking behavior. The alcohol consumption fluctuation information acquisition unit 110 outputs the acquired alcohol consumption fluctuation information to the alcohol consumption detection unit 130.

[0021] In step ST1002, the false detection factor information acquisition unit 120 acquires false detection factor information that indicates factors that cause fluctuations in alcohol consumption fluctuation information and are factors other than the subject's drinking behavior. The false detection factor information acquisition unit 120 outputs the acquired false detection factor information to the alcohol consumption detection unit 130.

[0022] Next, in step ST1003, the alcohol detection unit 130 estimates the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110 and the over-detection factor information acquired by the over-detection factor information acquisition unit 120.

[0023] Once the process reaches step ST1003 described above, the process shown in Figure 3 is terminated. (Termination) Note that the process shown in Figure 3 may be executed repeatedly.

[0024] As described above, the alcohol detection device 100 in this embodiment includes: an alcohol fluctuation information acquisition unit 110 that acquires alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior; an over-detection factor information acquisition unit 120 that acquires information about over-detection factors that are factors other than the subject's drinking behavior that cause the alcohol fluctuation information to fluctuate; and an alcohol detection unit 130 that estimates the subject's drinking state based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110 and the over-detection factor information acquired by the over-detection factor information acquisition unit 120. With the above configuration, the alcohol detection unit 130 estimates the subject's drinking state by taking into account not only the alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior, but also the over-detection factor information indicating information about factors other than the subject's drinking behavior that cause the alcohol fluctuation information to fluctuate. This suppresses the system from falsely detecting that the subject is drinking when they are not, and improves the accuracy of alcohol detection.

[0025] Furthermore, the alcohol detection system 1000 of this embodiment includes an alcohol detection device 100, and an output device 1100 that acquires alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior, and false detection factor information indicating information about factors other than the subject's drinking behavior that cause the alcohol fluctuation information to fluctuate, outputs the alcohol fluctuation information to the alcohol fluctuation information acquisition unit 110, and outputs the false detection factor information to the false detection factor information acquisition unit 120. With the above configuration, it is possible to suppress the system from falsely detecting that the subject is drinking when they are not, and to improve the accuracy of alcohol detection.

[0026] Furthermore, the alcohol detection method of this embodiment includes the steps of: the alcohol fluctuation information acquisition unit 110 acquiring alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior; the false detection factor information acquisition unit 120 acquiring false detection factor information indicating information about factors that cause fluctuations in alcohol fluctuation information but are other than the subject's drinking behavior; and the alcohol detection unit 130 estimating the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110 and the false detection factor information acquired by the false detection factor information acquisition unit 120. By performing the above steps, it is possible to suppress the system from falsely detecting that the subject is drinking when they are not, and to improve the accuracy of alcohol detection.

[0027] Furthermore, the alcohol detection program of this embodiment causes the computer to perform the following steps: an alcohol fluctuation information acquisition unit 110 acquires alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior; an over-detection factor information acquisition unit 120 acquires over-detection factor information indicating information about factors that cause fluctuations in alcohol fluctuation information but are other than the subject's drinking behavior; and an alcohol detection unit 130 estimates the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110 and the over-detection factor information acquired by the over-detection factor information acquisition unit 120. By performing these steps, it is possible to suppress the system from falsely detecting that the subject is drinking when they are not, and to improve the accuracy of alcohol detection.

[0028] Next, a detailed example of the configuration of the alcohol detection unit 130 of the alcohol detection device 100 shown in Figure 1 will be described using Figure 4. Figure 4 is a configuration diagram showing a detailed example of the configuration of the alcohol detection unit 130 of the alcohol detection device 100 shown in Figure 1.

[0029] As shown in Figure 4, the alcohol detection unit 130 may have an alcohol score calculation unit 131 that calculates an alcohol score indicating the degree of the subject's drinking state based on the alcohol fluctuation information acquired from the alcohol fluctuation information acquisition unit 110. If the alcohol fluctuation information acquisition unit 110 acquires multiple types of alcohol fluctuation information, the alcohol score calculation unit 131 may calculate each alcohol score based on each type of alcohol fluctuation information. For example, the alcohol score calculation unit 131 calculates a first alcohol score based on the first type of alcohol fluctuation information. Alternatively, if the alcohol fluctuation information acquisition unit 110 acquires multiple types of alcohol fluctuation information, the alcohol score calculation unit 131 may calculate one alcohol score based on each type of alcohol fluctuation information.

[0030] As shown in Figure 4, the alcohol detection unit 130 may have a feature calculation unit 132 that calculates effective features for estimating the drinking state based on the alcohol fluctuation information acquired from the alcohol fluctuation information acquisition unit 110, and the alcohol score calculation unit 131 may calculate the alcohol score based on the features calculated by the feature calculation unit 132. However, since the value of the alcohol fluctuation information during normal times when the person is not drinking (hereinafter referred to as the "reference value") varies from person to person, it may be difficult to estimate the state using the absolute value of the alcohol fluctuation information. For this reason, it is desirable for the feature calculation unit 132 to calculate the difference between the reference value of the alcohol fluctuation information and the value of the alcohol fluctuation information acquired at time t as a feature, and to estimate the state based on the feature. For example, the feature calculation unit 132 may calculate the median value from the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110 over a certain period of time at the beginning of the alcohol detection of the subject and use this median value as the reference value. However, during the period in which the reference value is calculated using the above method, the alcohol detection unit 130A does not start estimating the drinking state and remains in a waiting state. Therefore, calculating the reference value each time alcohol detection is performed on the same subject is inefficient. Thus, it is desirable to link the feature calculation unit 132 with a known personal authentication device. By linking the personal authentication ID and the reference value and storing them in a database device (hereinafter referred to as "DB"), the reference value can be retrieved from the DB for subsequent alcohol detections on the same subject, thus reducing the time required for the feature calculation unit 132 to calculate the reference value. Furthermore, if the alcohol fluctuation information is biometric data such as heart rate, the reference value of the biometric data may be obtained from a wearable sensor worn by the subject. This further reduces the time required to calculate the reference value.

[0031] Furthermore, there are no particular restrictions on the method for calculating the drinking score, but there are methods such as inputting the features calculated by the feature calculation unit 132 into a machine learning model to calculate the drinking score, or calculating the drinking score based on rules from the features calculated by the feature calculation unit 132. The drinking score is calculated in the range of 0 to 1.0, for example. When the drinking fluctuation information acquisition unit 110 acquires multiple types of drinking fluctuation information, the feature calculation unit 132 may calculate each feature based on each type of drinking fluctuation information, or it may combine all the drinking fluctuation information to finally calculate one type of feature. For example, the feature calculation unit 132 calculates a first feature based on the first drinking fluctuation information. Also, when the feature calculation unit 132 calculates multiple types of features, the drinking score calculation unit 131 may calculate each drinking score based on each feature, or it may combine all the features to finally calculate one type of drinking score.

[0032] Furthermore, as shown in Figure 4, the alcohol detection unit 130 may have an alcohol reliability calculation unit 133 that calculates an alcohol reliability score, which is the reliability of the alcohol score, based on the over-detection factor information acquired from the over-detection factor information acquisition unit 120. The alcohol reliability calculation unit 133 may also lower the alcohol reliability score based on the over-detection factor information. The alcohol reliability calculation unit 133 may further lower the alcohol reliability score if the over-detection factor information acquisition unit 120 acquires the same type of over-detection factor information A multiple times. For example, the alcohol reliability calculation unit 133 may hold an initial value of 100 for alcohol reliability, lower the alcohol reliability score to 80 when it acquires over-detection factor information A once, and lower it to 60 when it acquires over-detection factor information A again. In addition, if the over-detection factor information acquisition unit 120 acquires multiple types of over-detection factor information, the alcohol reliability calculation unit 133 may pre-set a reduction amount for the alcohol reliability score for each type of over-detection factor information and lower the alcohol reliability score based on the set reduction amount. For example, the alcohol confidence calculation unit 133 initially holds an alcohol confidence value of 100, and when it acquires over-detection factor information A from the over-detection factor information, it may lower the alcohol confidence value to 80, and when it acquires over-detection factor information B from the over-detection factor information, it may lower the alcohol confidence value to 60. When the alcohol fluctuation information acquisition unit 110 acquires multiple types of alcohol fluctuation information, the alcohol confidence calculation unit 133 calculates the alcohol confidence value based on over-detection factor information that indicates the factors that cause fluctuations in at least one of the multiple types of alcohol fluctuation information and are factors other than the subject's drinking behavior. When the alcohol score calculation unit 131 calculates an alcohol score based on each type of alcohol fluctuation information, the alcohol confidence calculation unit 133 calculates the alcohol confidence value, which is the confidence level of the alcohol score calculated based on the alcohol fluctuation information that fluctuates due to at least one of the over-detection factor information, based on the over-detection factor information that indicates the factors that cause fluctuations in at least one of the alcohol fluctuation information and are factors other than the subject's drinking behavior. In other words, the alcohol confidence calculation unit 133 calculates a first alcohol confidence score, which is the confidence score of the first alcohol score, based on first false positive factor information that indicates information on factors that cause fluctuations in the first alcohol fluctuation information and factors other than the subject's drinking behavior.

[0033] Furthermore, as shown in Figure 4, the alcohol detection unit 130 may also have an alcohol determination unit 134 that determines the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. For example, the alcohol determination unit 134 compares the alcohol score calculated by the alcohol score calculation unit 131 with a first threshold, and compares the alcohol confidence level calculated by the alcohol confidence level calculation unit 133 with a second threshold. If the alcohol score is equal to or greater than the first threshold and the alcohol confidence level is equal to or greater than the second threshold, it determines that the subject is intoxicated. As described above, the alcohol confidence level calculation unit 133 may lower the alcohol confidence level when over-detection factor information is acquired by the over-detection factor information acquisition unit 120. When the alcohol fluctuation information acquisition unit 110 acquires multiple types of alcohol fluctuation information, the alcohol determination unit 134 determines the subject's drinking status based on the alcohol score calculated based on the multiple types of alcohol fluctuation information and the alcohol confidence level calculated based on over-detection factor information, which indicates the factors that cause fluctuations in at least one of the multiple types of alcohol fluctuation information and are factors other than the subject's drinking behavior. When the alcohol score calculation unit 131 calculates an alcohol score based on each type of alcohol fluctuation information, the alcohol determination unit 134 determines the subject's drinking status based on at least one of the respective alcohol scores calculated based on each type of alcohol fluctuation information and the alcohol confidence level, which is the confidence level of the alcohol score calculated based on the alcohol fluctuation information that fluctuates due to at least one of the factors that cause fluctuations in the alcohol fluctuation information and are factors other than the subject's drinking behavior, based on the over-detection factor information. In other words, the alcohol determination unit 134 determines the subject's drinking status based on at least a first alcohol score and at least a first alcohol confidence level.

[0034] In the above description, the alcohol score calculation unit 131, the feature calculation unit 132, and the alcohol confidence calculation unit 133 are located inside the alcohol detection unit 130. However, the alcohol score calculation unit 131, the feature calculation unit 132, and the alcohol confidence calculation unit 133 may be located outside the alcohol detection unit 130. In other words, the alcohol determination unit 134 may determine the subject's drinking status based on the feature calculation unit, alcohol score, and alcohol confidence calculation unit calculated outside the alcohol detection unit 130.

[0035] Furthermore, in the above, the alcohol determination unit 134 determined the subject's drinking status based on the drinking score calculated by the drinking score calculation unit 131 and the drinking confidence level calculated by the drinking confidence level calculation unit 133. However, the alcohol determination unit 134 may also estimate the subject's drinking status based on the drinking score and over-detection factor information. For example, the alcohol determination unit 134 may compare the drinking score calculated by the drinking score calculation unit 131 with a threshold, and if the drinking score is equal to or greater than the threshold, it may determine that the subject is intoxicated. The drinking score calculation unit 131 may also lower the drinking score when over-detection factor information is acquired by the over-detection factor information acquisition unit 120.

[0036] Furthermore, in the above, the alcohol determination unit 134 compares the alcohol score calculated by the alcohol score calculation unit 131 with a first threshold, and compares the alcohol confidence calculated by the alcohol confidence calculation unit 133 with a second threshold. If the alcohol score is equal to or greater than the first threshold and the alcohol confidence is equal to or greater than the second threshold, the alcohol determination unit 134 determines that the subject is intoxicated. The alcohol confidence calculation unit 133 lowers the alcohol confidence when over-detection factor information is acquired by the over-detection factor information acquisition unit 120. However, the alcohol determination unit 134 may also stop estimating the subject's intoxication status if the alcohol score is equal to or greater than the first threshold and the alcohol confidence is equal to or less than the second threshold.

[0037] Next, an example of the processing flow of the alcohol detection device 100 having the alcohol detection unit 130 shown in Figure 4 will be explained using Figure 5. Figure 5 is a flowchart showing an example of the processing flow of the alcohol detection device 100 having the alcohol detection unit 130 shown in Figure 4.

[0038] The alcohol detection device 100, which has the alcohol detection unit 130 shown in Figure 4, starts the process shown in Figure 5 when, for example, a control unit (not shown) receives a command from an external source to start processing. (Start)

[0039] In step ST1101, the alcohol consumption fluctuation information acquisition unit 110 acquires alcohol consumption fluctuation information that shows the subject's information that fluctuates due to the subject's drinking behavior. The alcohol consumption fluctuation information acquisition unit 110 outputs the acquired alcohol consumption fluctuation information to the feature calculation unit 132.

[0040] Next, in step ST1102, the feature calculation unit 132 calculates features that are effective for estimating the drinking state based on the drinking fluctuation information acquired from the drinking fluctuation information acquisition unit 110.

[0041] Next, in step ST1103, the drinking score calculation unit 131 calculates the drinking score based on the features calculated by the feature calculation unit 132.

[0042] In step ST1104, the over-detection factor information acquisition unit 120 acquires over-detection factor information that indicates factors that cause fluctuations in alcohol consumption information and are factors other than the subject's drinking behavior. If the over-detection factor information acquisition unit 120 determines that there are over-detection factors (step ST1105 "YES"), proceed to step ST1107; if it determines that there are no over-detection factors (step ST1105 "NO"), proceed to step ST1106.

[0043] Next, in steps ST1106 and ST1107, the alcohol confidence level, which is the confidence level of the alcohol score, is calculated based on the over-detection factor information acquired from the over-detection factor information acquisition unit 120. If the process proceeds to step ST1106, the alcohol confidence level calculation unit 133 does not change the alcohol confidence level. If the process proceeds to step ST1107, the alcohol confidence level calculation unit 133 decreases the alcohol confidence level.

[0044] Next, in step ST1108, the drinking determination unit 134 estimates the drinking state of the subject based on the drinking score calculated by the drinking score calculation unit 131 and the drinking reliability calculated by the drinking reliability calculation unit 133. If the drinking score is greater than or equal to the threshold value and the drinking reliability is greater than or equal to the threshold value (step ST1108 “YES”), the process proceeds to step ST1110; otherwise (step ST1108 “NO”), the process proceeds to step ST1109.

[0045] When the process proceeds to step ST1109, the drinking determination unit 134 determines that the subject is in a non-drinking state (not drinking). When the process proceeds to step ST1110, the drinking determination unit 134 determines that the subject is in a drinking state (drinking).

[0046] When the process proceeds to step ST1109 or step ST1110, the process shown in FIG. 5 ends. (End) Note that the process shown in FIG. 5 may be repeatedly executed.

[0047] As shown in FIG. 5, the above steps ST1101 to ST1103 and the above steps ST1104 to ST1107 are performed in parallel processing. However, after the above steps ST1101 to ST1103, the above steps ST1104 to ST1107 may be performed. Specifically, when the drinking score calculated in step ST1103 is greater than or equal to the threshold value, the above steps ST1104 to ST1107 may be performed.

[0048] Next, a detailed configuration example of the drinking detection system 1000 including the drinking detection device 100 according to Embodiment 1 will be described with reference to FIG. 6. FIG. 6 is a configuration diagram showing a detailed configuration example of the drinking detection system 1000 including the drinking detection device 100 according to Embodiment 1. The drinking detection device 100 having this configuration example is referred to as a drinking detection device 100A, and the drinking detection system 1000 having this configuration example is referred to as a drinking detection system 1000A. [[ID=>

[0049] As shown in FIG. 6, the output device 1100A includes a biological data output device 1110. The biological data output device 1110 acquires the biological data of the subject and outputs it to the drinking detection device 100A.

[0050] The biometric data output device 1110 may be a biosensor capable of measuring the subject's heart rate, etc. The biosensor may be, for example, a wearable sensor that can be worn by the subject.

[0051] Furthermore, the biometric data output device 1110 may be an imaging device capable of capturing images of the subject's face, etc.

[0052] Furthermore, the biometric data output device 1110 may be a speech recognition device capable of recognizing the voice spoken by the subject. If there are people around the subject and it is desired to distinguish and recognize only the subject's voice, the speech recognition device may have a speech separation recognition function, and the imaging device and the speech recognition device may be combined to perform speech separation recognition.

[0053] As shown in Figure 6, the alcohol detection device 100A has biometric data acquisition units 111A and 121A that acquire the subject's biometric data output from the biometric data output device 1110. The biometric data is, for example, sensor data of the subject measured by a biosensor, image data of the subject captured by an imaging device, or voice data of the subject recognized by a voice recognition device. The biometric data acquisition units 111A and 121A may be sensor data acquisition units that acquire sensor data measured by a biosensor, image acquisition units that acquire image data captured by an imaging device, or voice acquisition units that acquire voice data recognized by a voice recognition device. As shown in Figure 6, the biometric data acquisition units 111A and 121A may be located inside the alcohol fluctuation information acquisition unit 110A and the over-detection factor information acquisition unit 120A, respectively, or outside the alcohol fluctuation information acquisition unit 110A and the over-detection factor information acquisition unit 120A.

[0054] As shown in Figure 6, the alcohol consumption fluctuation information acquisition unit 110A includes a biometric data acquisition unit 111A, which acquires the subject's biometric data from the biometric data output device 1110, which is an output device 1100A that is an external device of the alcohol consumption detection device 100A. The alcohol consumption fluctuation information acquisition unit 110A acquires alcohol consumption fluctuation information from the subject's biometric data. For example, it acquires heart rate, blood pressure, alertness level, reaction time to the surroundings, etc. from the subject's biometric data.

[0055] As shown in Figure 6, the false detection factor information acquisition unit 120A includes a biometric data acquisition unit 121A that acquires the subject's biometric data from the biometric data output device 1110, which is an output device 1100A that is an external device of the alcohol detection device 100A. The false detection factor information acquisition unit 120A acquires false detection factor information from the subject's biometric data. For example, it acquires specific emotions, specific behaviors, etc. from the subject's biometric data. Details will be described later.

[0056] Next, using Figure 7, we will describe an example configuration when the alcohol detection device 100A shown in Figure 6 is applied to an alcohol detection system that detects alcohol consumption in the driver of vehicle 1. Figure 7 is a configuration diagram showing an example configuration when the alcohol detection device 100A shown in Figure 6 is implemented in vehicle 1. The driver of vehicle 1 corresponds to the subject described earlier. Hereafter, the driver of vehicle 1 will also be simply referred to as "driver". Vehicle 1 is an example of a mobile body as described earlier.

[0057] As shown in Figure 7, the vehicle 1 is equipped with a control device 2. The control device 2 includes the components of the alcohol detection device 100A that have already been described. The control device 2 shown in Figure 7 is composed of a control unit 3 and an alcohol detection unit 130A. The control unit 3 is composed of an alcohol fluctuation information acquisition unit 110A and an over-detection factor information acquisition unit 120A.

[0058] A biometric data output device 1110 is provided outside the control device 2. The biometric data output device 1110 acquires the biometric data of the vehicle driver and outputs it to the control device 2 of the vehicle 1. The biometric data output device 1110 may be mounted on the vehicle 1, as shown in Figure 7, or it may be included in a driver monitoring system (hereinafter referred to as "DMS") or occupant monitoring system (hereinafter referred to as "OMS") mounted on the vehicle 1. If the biometric data output device 1110 is the biosensor described above, the biosensor may be installed in the front of the vehicle interior, inside the seat belt, inside the seat, inside the steering wheel, etc., or it may be included in a DMS or OMS mounted on the vehicle 1. If the biometric data output device 1110 is the biosensor described above, it may be attached to the subject themselves (wearable, etc.). Also, if the biometric data output device 1110 is the imaging device described above, the imaging device may be a monitoring camera used in a DMS or OMS mounted on the vehicle 1. The imaging device is installed, for example, in the front of the vehicle interior of the vehicle 1, and may be mounted on the instrument panel, steering column, A-pillar, rearview mirror, or ceiling of the vehicle 1. The imaging device may also consist of, for example, one visible light camera, multiple visible light cameras, one infrared camera, or multiple infrared cameras. If the imaging device consists of infrared cameras, a light source is provided that emits infrared light for imaging over an area including the subject's face. This light source may consist of, for example, an LED (Light Emitting Diode). Furthermore, if the biometric data output device 1110 is the voice recognition device described above, the voice recognition device may be included in the car navigation system (hereinafter referred to as "navigation") installed in the vehicle 1, or it may be included in the DMS or OMS installed in the vehicle 1.

[0059] The control unit 3 instructs the biological data output device 1110 to acquire biological data. Furthermore, the control unit 3 instructs the alcohol fluctuation information acquisition unit 110A, the over-detection factor information acquisition unit 120A, and the alcohol detection unit 130A in the control device 2 to control the timing of operation, control the exchange of information, etc.

[0060] Furthermore, the above-mentioned alarm device 1200 may be mounted on the vehicle 1, as shown in Figure 7, or it may be included in the DMS or OMS mounted on the vehicle 1. For example, the alarm device 1200 may be a speaker or buzzer that sounds through hearing, a light-emitting element such as a lamp or LED that sounds through sight, a monitor display that sounds through touch, airflow from the air conditioner, vibration of the steering wheel or seat, or an interlock that sounds through perception.

[0061] Furthermore, the personal authentication device and DB mentioned above may be included in the DMS or OMS installed in vehicle 1.

[0062] Next, using Figures 8 to 12, we will describe a specific configuration example of the alcohol detection system 1000A, including the alcohol detection device 100A shown in Figure 6, as well as an example of the processing flow.

[0063] First, using Figure 8, we will explain Configuration Example 1 of the alcohol detection system 1000A including the alcohol detection device 100A shown in Figure 6. Figure 8 is a configuration diagram showing Configuration Example 1 of the alcohol detection system 1000A including the alcohol detection device 100A shown in Figure 6. The alcohol detection device 100A equipped with Configuration Example 1 will be referred to as alcohol detection device 100Aa, and the alcohol detection system 1000A equipped with Configuration Example 1 will be referred to as alcohol detection system 1000Aa. In Configuration Example 1, the alcohol fluctuation information is "heart rate," and the false positive factor information is "specific emotion" and "specific behavior."

[0064] As shown in Figure 8, the alcohol consumption fluctuation information acquisition unit 110Aa has a heart rate acquisition unit 112 that acquires the subject's heart rate. The heart rate acquisition unit 112 may directly acquire the subject's heart rate from the biometric data output device 1110, or it may acquire biometric data such as the subject's electrocardiogram from the biometric data output device 1110 and calculate and acquire the heart rate based on the biometric data. The heart rate value may be an instantaneous heart rate value obtained from the reciprocal of the RR function, or it may be a value obtained by smoothing the instantaneous heart rate over a certain period of time in the past. If the biometric data output device 1110 is a biosensor, the heart rate measured by the biosensor may be acquired. If the biometric data output device 1110 is an imaging device, the heart rate of the subject may be acquired by image processing from the image captured by the imaging device. In summary, the heart rate acquisition unit 112 acquires the subject's heart rate from at least one of the biosensor, which is the biometric data output device 1110, or the imaging device.

[0065] As shown in Figure 8, the false detection factor information acquisition unit 120Aa has an emotion acquisition unit 122 that acquires the subject's specific emotions. The false detection factor information acquisition unit 120Aa also has an action acquisition unit 123 that acquires the subject's specific behaviors. In this embodiment, the emotion acquisition unit 122 and the action acquisition unit 123 acquire the subject's specific emotions and specific behaviors from the biometric data output device 1110.

[0066] When alcohol consumption fluctuation information is defined as "heart rate," a specific emotion is generally an emotion that can cause fluctuations in heart rate. Examples of emotions that can cause fluctuations in heart rate include "surprise," "tension," "anger," "impatience," and "excitement." There are no particular restrictions on the method of acquiring specific emotions, but for example, emotions may be inferred from the subject's facial expressions, voice, etc., or a known emotion recognition AI may be used to infer the subject's emotions. When inferring specific emotions from the subject's facial expressions, the emotion acquisition unit 122 may acquire the subject's facial expressions by detecting the face and facial parts from the face image acquired by the image acquisition unit, and then infer the subject's specific emotions from the acquired facial expressions. Face detection may be performed using a general algorithm classifier that combines a Haar-Like detector with AdaBoost or Casecade. Facial feature detection may be performed using a general algorithm detector, such as model fitting detection or a technique called Elastic Bunch Graph Matching, to detect the positions of facial feature points such as the outer corners of the eyes, inner corners of the eyes, nose, top of the head, and chin. Alternatively, the emotion acquisition unit 122 may not be configured to detect the face and facial features; an external device such as an imaging device may be configured to detect the face and facial features. If an external device is configured to detect the face and facial features, the image acquisition unit may acquire a face image and the results of face and facial feature detection from the external device. While there are no specific limitations on the method of acquiring facial expressions, examples include using FACS (Facial Action Coding System) to objectively quantify facial expressions based on the movement of facial muscles from the positions of the face and facial features detected by the above-described method, or inputting various facial expression data into a machine learning model for pre-training, and then inputting the face image acquired from the imaging device into the trained machine learning model to detect the subject's facial expressions. Furthermore, when inferring a specific emotion from the subject's voice, the emotion acquisition unit 122 may infer and acquire the subject's specific emotion from the voice data acquired by the voice acquisition unit. Alternatively, the emotion may be inferred and acquired from the subject's biological data such as brain waves and sweating.In that case, the emotion acquisition unit 122 may infer and acquire the subject's specific emotion from sensor data such as brain waves and sweating acquired by the sensor data acquisition unit. Alternatively, the emotion acquisition unit 122 may infer and acquire the subject's specific emotion by combining the above-described methods for inferring specific emotions. Furthermore, the emotion acquisition unit 122 may appropriately select and change the emotion inference method depending on the subject's condition. For example, if the subject is wearing a mask, sunglasses, etc., and the subject's facial expression cannot be recognized, the emotion acquisition unit 122 may select and change the emotion inference method to one that uses voice.

[0067] When alcohol consumption fluctuation information is defined as "heart rate," a specific behavior is generally any behavior that can cause fluctuations in heart rate. Examples of specific behaviors include "exercise," "eating and drinking," and "smoking." Note that "eating and drinking" excludes "drinking alcohol." The behavior acquisition unit 123 may, for example, infer the subject's specific behavior from the subject's skeletal structure, or from the subject's facial expression. There are no particular restrictions on the method of acquiring the skeletal structure, but one method is to input the positions of joints such as shoulders, elbows, wrists, and knees into a machine learning model for pre-training, and then input the captured images acquired from the imaging device into the trained machine learning model to acquire the subject's skeletal structure. For example, when inferring and acquiring the "eating and drinking" behavior among the specific behaviors, the behavior acquisition unit 123 may detect behaviors such as whether the wrist is positioned near the mouth or whether the subject is raising their wrist or arm as if drinking something, using the method described above, and then infer and acquire the subject's "eating and drinking" behavior from the detected behavior. Furthermore, the determination of whether the above-mentioned "eating and drinking" is "drinking alcohol" may be made, for example, by using an imaging device to capture an image of the label of the food or drink object, and the action acquisition unit 123 determining whether the food or drink object is "alcohol" and thus determining whether the above-mentioned "eating and drinking" is "drinking alcohol".

[0068] Next, an example of the processing flow of the alcohol detection device 100Aa shown in Figure 8 will be explained using Figure 9. Figure 9 is a flowchart of the example processing flow of the alcohol detection device 100Aa shown in Figure 8. Note that the alcohol detection unit 130Aa of the alcohol detection device 100Aa will be described assuming that it has the internal configuration of the alcohol detection unit 130 shown in Figure 4.

[0069] When the alcohol detection device 100Aa receives a command from an external source, for example, from a control unit (not shown), to start processing, it begins the process shown in Figure 9. (Start)

[0070] In step ST1201, the heart rate acquisition unit 112 acquires the heart rate. The heart rate acquisition unit 112 outputs the acquired heart rate to the feature calculation unit 132.

[0071] Next, in step ST1202, the feature calculation unit 132 calculates heart rate features based on the heart rate obtained from the heart rate acquisition unit 112.

[0072] Next, in step ST1203, the drinking score calculation unit 131 calculates the drinking score based on the heart rate features calculated by the feature calculation unit 132.

[0073] In step ST1204, the emotion acquisition unit 122 acquires specific emotions that cause fluctuations in heart rate. If the emotion acquisition unit 122 determines, for example, that the subject is "nervous" (step ST1205 "YES"), the process proceeds to step ST1209; if the subject is determined not to be nervous (step ST1205 "NO"), the process proceeds to step ST1208.

[0074] In step ST1206, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in heart rate. If the behavior acquisition unit 123 determines, for example, that the subject is "eating or drinking" (step ST1207 "YES"), the process proceeds to step ST1209; if it determines that the subject is "not eating or drinking" (step ST1207 "NO"), the process proceeds to step ST1208.

[0075] Next, in steps ST1208 and ST1209, the alcohol confidence score, which is the confidence level of the alcohol score, is calculated based on the specific emotions obtained from the emotion acquisition unit 122 and the specific behaviors obtained from the behavior acquisition unit 123. If the process proceeds to step ST1208, the alcohol confidence calculation unit 133 does not change the alcohol confidence score. If the process proceeds to step ST1209, the alcohol confidence calculation unit 133 decreases the alcohol confidence score.

[0076] Next, in step ST1210, the alcohol determination unit 134 estimates the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. If the alcohol score is above the threshold and the alcohol confidence level is above the threshold (step ST1210 "YES"), the process proceeds to step ST1212; otherwise (step ST1210 "NO"), the process proceeds to step ST1211.

[0077] If the process proceeds to step ST1211, the alcohol detection unit 134 determines that the subject is not drinking alcohol (has not consumed alcohol). If the process proceeds to step ST1212, the alcohol detection unit 134 determines that the subject is intoxicated.

[0078] When the process reaches step ST1211 or step ST1212, the process shown in Figure 9 is terminated. (Termination) Note that the process shown in Figure 9 may be executed repeatedly.

[0079] Next, using Figure 10, we will describe Configuration Example 2 of the alcohol detection system 1000A including the alcohol detection device 100A shown in Figure 6. Figure 10 is a configuration diagram showing Configuration Example 2 of the alcohol detection system 1000A including the alcohol detection device 100A shown in Figure 6. The alcohol detection device 100A equipped with Configuration Example 2 will be referred to as alcohol detection device 100Ab, and the alcohol detection system 1000A equipped with Configuration Example 2 will be referred to as alcohol detection system 1000Ab. In Configuration Example 2, the biological data is referred to as "alertness level," and the information on factors causing false positives is referred to as "specific behavior."

[0080] As shown in Figure 10, the alcohol consumption fluctuation information acquisition unit 110Ab has an alertness acquisition unit 113 that acquires an alertness level indicating the subject's level of alertness. The alertness acquisition unit 113 acquires the alertness level from the biological data output device 1110. The biological data output device 1110 is, for example, a DMS, and the alertness acquisition unit 113 may acquire the alertness level directly from the DMS. Alternatively, the biological data output device 1110 is, for example, an imaging device that images the subject, and the alertness acquisition unit 113 may acquire the image captured of the subject and calculate and acquire the subject's alertness level based on the image. If the biological data output device 1110 is an imaging device, the biological data acquisition unit 111Ab is an image acquisition unit that acquires the image captured by the imaging device. The alertness acquisition unit 113 may, for example, estimate and acquire the subject's alertness level from the subject's facial expression. The alertness level acquisition unit 113 may acquire the subject's facial expression by detecting the face and facial parts from the face image acquired by the image acquisition unit using the method described above, and calculate the subject's alertness level from the acquired facial expression. The method for calculating the alertness level is not particularly limited, but for example, from the face and facial parts detected using the method described above, the degree of eye opening, eye closing time, and blinking interval may be calculated from the position of the upper eyelid, the degree of mouth opening may be detected from the position of the upper and lower parts of the mouth, and it may be determined whether the subject is yawning, and the subject's alertness level may be calculated from this information. When calculating the alertness level, this information may be input into a machine learning model to calculate the alertness level. Alternatively, the alertness level may be calculated using a rule-based method. For example, a rule may be set in advance to decrease the alertness level by 10 if the eye closing time is 3 seconds or more, and the alertness level may be calculated using this rule.

[0081] As shown in Figure 10, the false detection factor information acquisition unit 120Ab has an action acquisition unit 123 that acquires specific actions of the subject. In this embodiment, the action acquisition unit 123 acquires specific actions of the subject from the biological data output device 1110.

[0082] Furthermore, when alcohol fluctuation information is defined as "alertness," specific behaviors are generally those that can cause fluctuations in alertness. An example of a specific behavior is "conversation." As mentioned above, when detecting alertness, the degree of mouth opening is detected from the position of the upper and lower parts of the mouth. Since "conversation" is an action that involves opening the mouth, it may be difficult to distinguish it from other mouth-opening actions such as yawning. Therefore, when calculating alertness, if a subject is engaging in "conversation," their alertness may be judged as low. In that case, when estimating the drinking state based on alertness, the subject may be over-detected as being intoxicated when they are not. For these reasons, "conversation" is an action that can cause fluctuations in alertness.

[0083] The behavior acquisition unit 123 may, for example, infer and acquire a specific behavior of the subject from the subject's facial expressions, or infer a specific behavior of the subject from the voice the subject emits. For example, when inferring the "conversation" behavior among the specific behaviors, the behavior acquisition unit 123 may detect the degree of mouth opening from the face and facial parts detected by the method described above, and infer whether or not there is a conversation from the degree of change in the value of the degree of mouth opening. The degree of change in the degree of mouth opening is, for example, assuming that the degree of mouth opening is calculated in the range of 0 to 1.0, if the "standard deviation of the degree of mouth opening over a certain period of time in the past (e.g., 180 seconds) is 0.1 or more" or the "average value of the degree of mouth opening over a certain period of time in the past (e.g., 180 seconds) is 0.12 or more", then it is detected that there is a conversation. Alternatively, for example, when inferring the "conversation" behavior among the specific behaviors, the behavior acquisition unit 123 may acquire voice data emitted by the subject from the voice recognition device and infer the subject's "conversation" behavior from the voice data. In that case, the method described above can be used to distinguish between actions that are generally taken when the level of alertness is low, such as keeping the mouth open (mouth hanging open) or yawning, and actions that are generally taken when the level of alertness is high, such as "talking." It should be noted that even in the case of "talking," the speech may be slurred, and if the subject is intoxicated, slurred speech may appear. In that case, if the behavior acquisition unit 123 recognizes "talking" as a false positive, it may incorrectly detect that the subject is not intoxicated when they are. In that case, for example, the behavior acquisition unit 123 may analyze the voice data to distinguish between a state of slurred speech and a normal state, and may recognize "talking" in a normal state as a false positive.

[0084] Next, an example of the processing flow of the alcohol detection device 100Ab shown in Figure 10 will be explained using Figure 11. Figure 11 is a flowchart of the example processing flow of the alcohol detection device 100Ab shown in Figure 10. It will be explained assuming that the alcohol detection unit 130Ab of the alcohol detection device 100Ab has the internal configuration of the alcohol detection unit 130 shown in Figure 4.

[0085] When the alcohol detection device 100Ab receives a command from an external source, for example, from a control unit (not shown), to start processing, it begins the process shown in Figure 11. (Start)

[0086] In step ST1301, the arousal level acquisition unit 113 acquires the arousal level. The arousal level acquisition unit 113 outputs the acquired arousal level to the feature calculation unit 132.

[0087] Next, in step ST1302, the feature calculation unit 132 calculates the arousal level feature based on the arousal level acquired from the arousal level acquisition unit 113.

[0088] Next, in step ST1303, the drinking score calculation unit 131 calculates the drinking score based on the alertness feature calculated by the feature calculation unit 132.

[0089] In step ST1304, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in the level of arousal. If the behavior acquisition unit 123 determines, for example, that the subject is "talking" (step ST1305 "YES"), the process proceeds to step ST1307; if it determines that the subject is "not talking" (step ST1305 "NO"), the process proceeds to step ST1306.

[0090] Next, in steps ST1306 and ST1307, the alcohol confidence score, which is the confidence level of the alcohol score, is calculated based on the specific behavior acquired from the behavior acquisition unit 123. If the process proceeds to step ST1306, the alcohol confidence calculation unit 133 does not change the alcohol confidence score. If the process proceeds to step ST1307, the alcohol confidence calculation unit 133 decreases the alcohol confidence score.

[0091] Next, in step ST1308, the alcohol determination unit 134 estimates the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. If the alcohol score is above the threshold and the alcohol confidence level is above the threshold (step ST1308 "YES"), the process proceeds to step ST1310; otherwise (step ST1308 "NO"), the process proceeds to step ST1309.

[0092] If the process proceeds to step ST1309, the alcohol detection unit 134 determines that the subject is not drinking alcohol (has not consumed alcohol). If the process proceeds to step ST1310, the alcohol detection unit 134 determines that the subject is intoxicated.

[0093] When the process reaches step ST1309 or step ST1310, the process shown in Figure 11 is terminated. (Termination) The process shown in Figure 11 may be executed repeatedly.

[0094] Next, using Figure 12, we will describe Configuration Example 3 of the alcohol detection system 1000A including the alcohol detection device 100A shown in Figure 6. Figure 12 is a configuration diagram showing Configuration Example 3 of the alcohol detection system 1000A including the alcohol detection device 100A shown in Figure 6. The alcohol detection device 100A equipped with Configuration Example 3 will be referred to as alcohol detection device 100Ac, and the alcohol detection system 1000A equipped with Configuration Example 3 will be referred to as alcohol detection system 1000Ac. In Configuration Example 3, the alcohol fluctuation information will be "heart rate" and "alertness level," and the false positive factor information will be "specific emotion" and "specific behavior."

[0095] As shown in Figure 12, the alcohol consumption fluctuation information acquisition unit 110Ac has both a heart rate acquisition unit 112 that acquires the subject's heart rate and an alertness acquisition unit 113 that acquires the subject's alertness level.

[0096] As shown in Figure 12, the false positive factor information acquisition unit 120Ac has both an emotion acquisition unit 122 that acquires specific emotions of the subject and an arousal level acquisition unit 113 that acquires specific behaviors of the subject.

[0097] The processing flow example of the alcohol detection device 100Ac shown in Figure 12 will be explained using Figure 13. Figure 13 is a flowchart of the processing flow example of the alcohol detection device 100Ac shown in Figure 12. It will be explained assuming that the alcohol detection unit 130Ac of the alcohol detection device 100Ac has the internal configuration of the alcohol detection unit 130 shown in Figure 4.

[0098] When the alcohol detection device 100Ac receives a command from an external source, for example, from a control unit (not shown), to start processing, it begins the process shown in Figure 13. (Start)

[0099] In step ST1401, the heart rate acquisition unit 112 acquires the heart rate. The heart rate acquisition unit 112 outputs the acquired heart rate to the feature calculation unit 132. Also in step ST1401, the arousal level acquisition unit 113 acquires the arousal level. The arousal level acquisition unit 113 outputs the acquired arousal level to the feature calculation unit 132.

[0100] Next, in step ST1402, the feature calculation unit 132 calculates heart rate features based on the heart rate acquired from the heart rate acquisition unit 112. Also in ST1402, the feature calculation unit 132 calculates arousal level features based on the arousal level acquired from the arousal level acquisition unit 113.

[0101] Next, in step ST1403, the drinking score calculation unit 131 calculates the drinking score based on the heart rate feature and the alertness feature calculated by the feature calculation unit 132.

[0102] In step ST1404, the emotion acquisition unit 122 acquires specific emotions that cause fluctuations in heart rate. If the emotion acquisition unit 122 determines, for example, that the subject is "nervous" (step ST1405 "YES"), the process proceeds to step ST1410; if the subject is determined not to be nervous (step ST1405 "NO"), the process proceeds to step ST1409.

[0103] In step ST1406, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in heart rate. If the behavior acquisition unit 123 determines, for example, that the subject is "eating or drinking" (step ST1407 "YES"), the process proceeds to step ST1410; if the subject is determined to be "not eating or drinking" (step ST1407 "NO"), the process proceeds to step ST1409. Also in step ST1406, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in arousal level. If the behavior acquisition unit 123 determines, for example, that the subject is "talking" (step ST1408 "YES"), the process proceeds to step ST1410; if the subject is determined to be "not talking" (step ST1408 "NO"), the process proceeds to step ST1409.

[0104] Next, in steps ST1409 and ST1410, the alcohol confidence score, which is the confidence level of the alcohol score, is calculated based on the specific emotions obtained from the emotion acquisition unit 122 and the specific behaviors obtained from the behavior acquisition unit 123. If the process proceeds to step ST1409, the alcohol confidence calculation unit 133 does not change the alcohol confidence score. If the process proceeds to step ST1410, the alcohol confidence calculation unit 133 decreases the alcohol confidence score.

[0105] Next, in step ST1411, the alcohol determination unit 134 estimates the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. If the alcohol score is above the threshold and the alcohol confidence level is above the threshold (step ST1411 "YES"), proceed to step ST1413; otherwise (step ST1411 "NO"), proceed to step ST1412.

[0106] If the process proceeds to step ST1412, the alcohol detection unit 134 determines that the subject is not drinking alcohol (has not consumed alcohol). If the process proceeds to step ST1413, the alcohol detection unit 134 determines that the subject is intoxicated.

[0107] When the process reaches step ST1412 or step ST1413, the process shown in Figure 13 is terminated. (Termination) Note that the process shown in Figure 13 may be executed repeatedly.

[0108] Embodiment 2. Embodiment 2 will be described below with reference to the drawings. First, a basic configuration example of the alcohol detection system 1000B including the alcohol detection device 100B according to Embodiment 2 will be described using Figure 14. Figure 14 is a configuration diagram showing a basic configuration example of the alcohol detection system 1000B including the alcohol detection device 100B according to Embodiment 2. The same configuration as in Embodiment 1 will not be described. Also, in Figure 14, the same reference numerals as in Figures 1 to 13 indicate the same or corresponding parts. The following description will focus on the differences from Embodiment 1.

[0109] In the alcohol detection device 100A shown in Figure 6 of Embodiment 1, the alcohol fluctuation information acquisition unit 110A acquired alcohol fluctuation information from the subject's biological data. The alcohol detection device 100B of Embodiment 2 differs from the alcohol detection device 100A shown in Figure 6 of Embodiment 1 in that, similar to the alcohol detection device 100A of Embodiment 1, the alcohol fluctuation information acquisition unit 110B acquires alcohol fluctuation information from the subject's biological data, and also acquires alcohol fluctuation information from mobile information.

[0110] Furthermore, the output device 1100A of the alcohol detection system 1000A shown in Figure 6 of Embodiment 1 had a biometric data output device 1110 that acquires the subject's biometric data and outputs it to the alcohol detection device 100A. The alcohol detection system 1000B of Embodiment 2 differs from the alcohol detection system 1000A shown in Figure 6 of Embodiment 1 in that, similar to the alcohol detection system 1000A of Embodiment 1, the output device 1100B has a biometric data output device 1110, and further has a mobile information output device 1120 that acquires mobile information and outputs it to the alcohol detection device 100B.

[0111] In the second embodiment, the subject is the driver of the moving object. The moving object may be a vehicle.

[0112] The mobile information output device 1120 acquires mobile information and outputs it to the alcohol detection device 100B. Mobile information refers to information indicating the operation of the mobile vehicle by its driver or the movement of the mobile vehicle resulting from such operation.

[0113] The alcohol detection device 100B has a mobile body information acquisition unit 114 that acquires mobile body information output from the mobile body information output device 1120. As shown in Figure 14, the mobile body information acquisition unit 114 may be located inside the alcohol fluctuation information acquisition unit 110B or outside the alcohol fluctuation information acquisition unit 110B.

[0114] The alcohol consumption fluctuation information acquisition unit 110B acquires alcohol consumption fluctuation information from the subject's biological data as well as from the mobile information. The alcohol consumption fluctuation information acquisition unit 110B includes a biological data acquisition unit 111B that acquires the subject's biological data and a mobile information acquisition unit 114 that acquires mobile information. The alcohol consumption fluctuation information in Embodiment 2 may be, for example, the degree of operational disorder. Operational disorder indicates the degree of roughness of the subject's driving operations. The alcohol consumption fluctuation information acquisition unit 110B in Embodiment 2 acquires alcohol consumption fluctuation information from the biological data output device 1110 and the mobile information output device 1120.

[0115] Next, using Figure 15, an example configuration of the alcohol detection device 100B according to Embodiment 2 applied to an alcohol detection system for detecting alcohol consumption in a driver operating a vehicle 1 will be described. Figure 15 is a configuration diagram showing an example configuration when the alcohol detection device 100B according to Embodiment 2 is installed in a vehicle 1. The driver of vehicle 1 corresponds to the subject described earlier. Vehicle 1 is an example of a mobile body described earlier.

[0116] As shown in Figure 15, the vehicle 1 is equipped with a control device 2. The control device 2 includes the components of the alcohol detection device 100B that have already been described. The control device 2 shown in Figure 15 is composed of a control unit 3 and an alcohol detection unit 130B. The control unit 3 is composed of an alcohol fluctuation information acquisition unit 110B and an over-detection factor information acquisition unit 120B.

[0117] A biometric data output device 1110 and a vehicle information output device 1120 are provided outside the control device 2. The biometric data output device 1110 acquires the biometric data of the driver of vehicle 1 and outputs it to the control device 2 of vehicle 1. The vehicle information output device 1120 acquires vehicle information of vehicle 1 and outputs it to the control device 2 of vehicle 1.

[0118] Vehicle information corresponds to the moving object information already explained. Vehicle information is based on the driver's steering, braking, or acceleration operations, or at least one of the vehicle's movements resulting from each of these operations. Vehicle movements include, for example, changes in vehicle speed, changes in yaw rate, and changes in the vehicle's position within the lane.

[0119] The vehicle information output device 1120 corresponds to the mobile information output device 1120 already described. The vehicle information output device is the part that acquires data flowing on the signal lines of the CAN (Controller Area Network) bus, and acquires vehicle information such as vehicle speed, steering angle, and accelerator / brake depression (opening) in real time.

[0120] The control unit 3 instructs the biometric data output device 1110 to acquire biometric data. The control unit 3 also instructs the vehicle information output device 1120 to acquire vehicle information. Furthermore, the control unit 3 instructs the alcohol fluctuation information acquisition unit 110B, the over-detection factor information acquisition unit 120B, and the alcohol detection unit 130B in the control device 2 to control the timing of operation, control the exchange of information, etc.

[0121] Next, using Figures 16 and 17, a specific configuration example and a processing flow example of the alcohol detection system 1000B, including the alcohol detection device 100B according to Embodiment 2, will be described.

[0122] First, an example configuration of the alcohol detection system 1000B, including the alcohol detection device 100B, will be described using Figure 16. Figure 16 is a configuration diagram showing an example configuration of the alcohol detection system 1000B, including the alcohol detection device 100B. The alcohol detection device 100B equipped with this configuration example will be referred to as the alcohol detection device 100Ba, and the alcohol detection system 1000B equipped with this configuration example will be referred to as the alcohol detection system 1000Ba. In this configuration example, the alcohol fluctuation information will be "heart rate" and "operational disorder," and the false positive factor information will be "specific emotion" and "specific behavior."

[0123] As shown in Figure 16, the alcohol fluctuation information acquisition unit 110Ba includes a heart rate acquisition unit 112 for acquiring the subject's heart rate and an operation randomness acquisition unit 115. The operation randomness acquisition unit 115 may directly acquire operation randomness from the mobile information output device 1120, or it may acquire mobile information from the mobile information output device 1120 and calculate and acquire operation randomness based on the mobile information. The operation randomness acquisition unit 115 may also be composed of a machine learning model that takes mobile information as input and outputs the subject's operation randomness. Operation randomness may be, for example, the roughness of steering wheel operation, the variability of speed changes, the deviation of the vehicle position from the lane center, etc.

[0124] The alcohol consumption fluctuation information acquisition unit 110Ba may also include a heart rate acquisition unit 112, an alertness level acquisition unit 113, and an operation disorder level acquisition unit 115.

[0125] Next, an example of the processing flow of the alcohol detection device 100Ba shown in Figure 16 will be explained using Figure 17. Figure 17 is a flowchart showing the example of the processing flow of the alcohol detection device 100Ba shown in Figure 16. It should be assumed that the alcohol detection unit 130Ba of the alcohol detection device 100Ba has the internal configuration of the alcohol detection unit 130 shown in Figure 4.

[0126] When the alcohol detection device 100Ba receives a command from an external source, for example, from a control unit (not shown), to start processing, it begins the process shown in Figure 17. (Start)

[0127] In step ST2001, the heart rate acquisition unit 112 acquires the heart rate. The heart rate acquisition unit 112 outputs the acquired heart rate to the feature calculation unit 132. Also in step ST2001, the operation randomness acquisition unit 115 acquires the operation randomness. The operation randomness acquisition unit 115 outputs the acquired operation randomness to the feature calculation unit 132.

[0128] Next, in step ST2002, the feature calculation unit 132 calculates heart rate features based on the heart rate obtained from the heart rate acquisition unit 112. Also in ST2002, the feature calculation unit 132 calculates operation randomness features based on the operation randomness obtained from the operation randomness acquisition unit 115.

[0129] Next, in step ST2003, the drinking score calculation unit 131 calculates the drinking score based on the heart rate feature and the operation randomness feature calculated by the feature calculation unit 132.

[0130] In step ST2004, the emotion acquisition unit 122 acquires specific emotions that cause fluctuations in heart rate. If the emotion acquisition unit 122 determines, for example, that the subject is "nervous" (step ST2005 "YES"), the process proceeds to step ST2009; if the subject is determined not to be nervous (step ST2005 "NO"), the process proceeds to step ST2008.

[0131] In step ST2006, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in heart rate. If the behavior acquisition unit 123 determines, for example, that the subject is "eating or drinking" (step ST2007 "YES"), the process proceeds to step ST2009; if it determines that the subject is "not eating or drinking" (step ST2007 "NO"), the process proceeds to step ST2008.

[0132] Next, in steps ST2008 and ST2009, the alcohol confidence score, which is the confidence level of the alcohol score, is calculated based on the specific emotions obtained from the emotion acquisition unit 122 and the specific behaviors obtained from the behavior acquisition unit 123. If the process proceeds to step ST2008, the alcohol confidence calculation unit 133 does not change the alcohol confidence score. If the process proceeds to step ST2009, the alcohol confidence calculation unit 133 decreases the alcohol confidence score.

[0133] Next, in step ST2010, the alcohol determination unit 134 estimates the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. If the alcohol score is above the threshold and the alcohol confidence level is above the threshold (step ST2010 "YES"), the process proceeds to step ST2012; otherwise (step ST2010 "NO"), the process proceeds to step ST2011.

[0134] If the process proceeds to step ST2011, the alcohol detection unit 134 determines that the subject is not drinking alcohol (has not consumed alcohol). If the process proceeds to step ST2012, the alcohol detection unit 134 determines that the subject is intoxicated.

[0135] When the process reaches step ST2011 or step ST2012, the process shown in Figure 17 is terminated. (Termination) Note that the process shown in Figure 17 may be executed repeatedly.

[0136] As described above, the alcohol detection device 100B in Embodiment 2 includes: an alcohol fluctuation information acquisition unit 110B that acquires alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior; an over-detection factor information acquisition unit 120B that acquires information about over-detection factors that are factors that cause fluctuations in the alcohol fluctuation information but are other than the subject's drinking behavior; and an alcohol detection unit 130B that estimates the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110B and the over-detection factor information acquired by the over-detection factor information acquisition unit 120B. With the above configuration, it is possible to suppress the system from falsely detecting that the subject is drinking when they are not, and to improve the accuracy of alcohol detection.

[0137] Furthermore, in the alcohol detection device 100B of Embodiment 2, the alcohol fluctuation information acquisition unit 110B has both a biometric data acquisition unit 111B that acquires biometric data and a mobile body information acquisition unit 114 that acquires mobile body information, and acquires the subject's alcohol fluctuation information from the subject's biometric data as well as the mobile body information. With the above configuration, the alcohol detection unit 130B can estimate the subject's drinking state based on more alcohol fluctuation information than in Embodiment 1, and can further improve the accuracy of alcohol detection.

[0138] In the above configuration, the alcohol consumption fluctuation information acquisition unit 110B is configured to have both a biometric data acquisition unit 111B for acquiring biometric data and a mobile information acquisition unit 114 for acquiring mobile information, and acquires the subject's alcohol consumption fluctuation information from vehicle information in addition to the subject's biometric data. However, the alcohol consumption fluctuation information acquisition unit 110B may also have a mobile information acquisition unit 114 for acquiring mobile information, and acquire the subject's alcohol consumption fluctuation information from at least the mobile information. By doing so, even when biometric data such as the subject's heart rate cannot be acquired, alcohol consumption fluctuation information can be acquired from the mobile information, thus achieving the same effect as in Embodiment 1. If the alcohol consumption fluctuation information acquisition unit 110B acquires only the operational disorder level as the subject's alcohol consumption fluctuation information, the over-detection factor information acquisition unit 120B acquires over-detection factor information that indicates factors that cause the operational disorder level to fluctuate and are factors other than the subject's drinking behavior. In that case, the over-detection factor information is the subject's "specific behavior," and the "specific behavior" may be, for example, the subject's "avoidance behavior." "Avoidance behavior" refers to the action taken by the subject to avoid objects present in the vicinity of the moving object by manipulating the moving object. In this case, the false detection factor information acquisition unit 120B may be configured to infer and acquire the subject's "avoidance behavior" based on the moving object surrounding information described later.

[0139] Embodiment 3. Embodiment 3 will be described below with reference to the drawings. First, Figure 18 will be used to describe a basic configuration example of the alcohol detection device 100C and the alcohol detection system 1000C according to Embodiment 3. Figure 18 is a configuration diagram showing a basic configuration example of the alcohol detection system 1000C including the alcohol detection device 100C according to Embodiment 3. The same configuration as in Embodiments 1 and 2 will not be described. Also, in Figure 18, the same reference numerals as in Figures 1 to 17 indicate the same or corresponding parts. The following description will focus on the differences from Embodiments 1 and 2.

[0140] In the alcohol detection device 100A shown in Figure 6 of Embodiment 1, the over-detection factor information acquisition unit 120A acquired over-detection factor information from the subject's biological data. The alcohol detection device 100C of Embodiment 3 differs from the alcohol detection device 100A shown in Figure 6 of Embodiment 1 in that, similar to the alcohol detection device 100A shown in Figure 6 of Embodiment 1, the over-detection factor information acquisition unit 120C acquires over-detection factor information from the subject's biological data, and also acquires over-detection factor information from information about the surroundings of the moving object.

[0141] Furthermore, the output device 1100A of the alcohol detection system 1000A shown in Figure 6 of Embodiment 1 had a biometric data output device 1110 that acquires the subject's biometric data and outputs it to the alcohol detection device 100A. The alcohol detection system 1000C of Embodiment 3 differs from the alcohol detection system 1000A of Embodiment 1 in that, similar to the alcohol detection system 1000A shown in Figure 6 of Embodiment 1, the output device 1100C has a biometric data output device 1110, and further has a mobile body surrounding information output device 1130 that acquires mobile body surrounding information and outputs it to the alcohol detection device 100C.

[0142] Furthermore, in Embodiment 3, the subject is a passenger of the moving object, and may be, for example, the driver of the moving object. The moving object may also be a vehicle.

[0143] The mobile object surrounding information output device 1130 acquires mobile object surrounding information and outputs it to the alcohol detection device 100C. Mobile object surrounding information refers to information indicating the situation around the mobile object.

[0144] The alcohol detection device 100C has a mobile body surrounding information acquisition unit 124 that acquires mobile body surrounding information output from the mobile body surrounding information output device 1130. As shown in Figure 18, the mobile body surrounding information acquisition unit 124 may be located inside the overdetection factor information acquisition unit 120C, or it may be located outside the overdetection factor information acquisition unit 120C.

[0145] The overdetection factor information acquisition unit 120C acquires overdetection factor information from the subject's biological data as well as from information about the surroundings of the moving object. The overdetection factor information acquisition unit 120C includes a biological data acquisition unit 121C that acquires the subject's biological data and a moving object surrounding information acquisition unit 124 that acquires information about the surroundings of the moving object. In Embodiment 3, the overdetection factor information acquisition unit 120C acquires overdetection factor information from the biological data output device 1110 and the moving object surrounding information output device 1130.

[0146] Next, using Figure 19, an example configuration of the alcohol detection device 100C according to Embodiment 3 applied to an alcohol detection system for detecting alcohol consumption in a driver operating a vehicle 1 will be described. Figure 19 is a configuration diagram showing an example configuration when the alcohol detection device 100C according to Embodiment 3 is installed in a vehicle 1. The driver of vehicle 1 corresponds to the subject described earlier. Vehicle 1 is an example of a mobile body described earlier.

[0147] As shown in Figure 19, the vehicle 1 is equipped with a control device 2. The control device 2 includes the components of the alcohol detection device 100C that have already been described. The control device 2 shown in Figure 19 is composed of a control unit 3 and an alcohol detection unit 130C. The control unit 3 is composed of an alcohol fluctuation information acquisition unit 110C and an over-detection factor information acquisition unit 120C.

[0148] External to the control device 2, a biometric data output device 1110 and a vehicle surrounding information output device 1130 are provided. The biometric data output device 1110 acquires the biometric data of the driver of vehicle 1 and outputs it to the control device 2 of vehicle 1. The vehicle surrounding information output device 1130 acquires vehicle surrounding information of vehicle 1 and outputs it to the control device 2 of vehicle 1.

[0149] Vehicle surrounding information corresponds to the moving object surrounding information already described. Vehicle surrounding information includes, for example, information about objects present around vehicle 1, road information around vehicle 1, or weather information around vehicle 1.

[0150] The vehicle surrounding information output device 1130 corresponds to the mobile object surrounding information output device 1130 already described. The vehicle surrounding information output device 1130 may also be an object information output device that detects object information present around the vehicle 1. The object information output device consists of one or more ultrasonic sensors, millimeter-wave radar, near-infrared lasers, or external cameras. The object information output device is installed inside or outside the vehicle 1 so as to face outwards and detects the type, location, or type and location of objects such as pedestrians, bicycles, vehicles, traffic lights, and signs present around the vehicle 1. The vehicle surrounding information output device 1130 may also be a road information output device that acquires road conditions around the vehicle 1. The road information output device acquires road information, for example, via a navigation system or the cloud, and may be a network-connected terminal such as a navigation system or smartphone. The road information output device acquires information such as road congestion information and information on locations with a high incidence of near misses.

[0151] The control unit 3 instructs the biometric data output device 1110 to acquire biometric data. The control unit 3 also instructs the vehicle surroundings information output device 1130 to acquire vehicle surroundings information. Furthermore, the control unit 3 instructs the alcohol fluctuation information acquisition unit 110C, the over-detection factor information acquisition unit 120C, and the alcohol detection unit 130C in the control device 2 to control the timing of operation, control the exchange of information, etc.

[0152] Next, using Figures 20 and 21, a specific configuration example and a processing flow example of the alcohol detection system 1000C, including the alcohol detection device 100C according to Embodiment 3, will be described.

[0153] First, an example configuration of the alcohol detection system 1000C, including the alcohol detection device 100C, will be described using Figure 20. Figure 20 is a configuration diagram showing an example configuration of the alcohol detection system 1000C, including the alcohol detection device 100C. The alcohol detection device 100C equipped with this configuration example will be referred to as the alcohol detection device 100Ca, and the alcohol detection system 1000C equipped with this configuration example will be referred to as the alcohol detection system 1000Ca. In this configuration example, the alcohol fluctuation information is defined as "heart rate," and the false positive factor information is defined as "specific emotion" and "specific behavior."

[0154] As shown in Figure 20, the false positive factor information acquisition unit 120Ca has an emotion acquisition unit 122 that acquires the subject's specific emotions. The false positive factor information acquisition unit 120Ca also has an action acquisition unit 123 that acquires the subject's specific actions. In this embodiment, the emotion acquisition unit 122 acquires the subject's specific emotions from the biometric data output device 1110 and the mobile body peripheral information output device 1130. In this embodiment, the action acquisition unit 123 acquires the subject's specific actions from the biometric data output device 1110.

[0155] The emotion acquisition unit 122 of this embodiment infers the subject's specific emotion from information about the moving object's surroundings. In other words, the emotion acquisition unit 122 infers and acquires the subject's specific emotion from external factors, such as information about the moving object's surroundings. For example, when the emotion acquisition unit 122 acquires information such as a pedestrian suddenly running out as object information, information such as a frequent near-miss location as road information, or information such as a thunderstorm as weather information, it may infer that the subject is in a state of "surprise." Alternatively, for example, when the emotion acquisition unit 122 acquires information such as the presence of other vehicles as object information, or information such as traffic congestion as road information, it may infer that the subject is in a state of "anxiety" or "tension."

[0156] Furthermore, when the emotion acquisition unit 122 infers and acquires a first specific emotion from biometric data, it may increase the confidence that the subject is in a state of having the first specific emotion when it acquires external factors from the surrounding information of the moving object that could cause the subject to be in a state of having the first specific emotion. For example, when the emotion acquisition unit 122 infers from biometric data that the subject is in a state of "surprise," it may increase the confidence that the subject is in a state of "surprise" when it acquires information such as a pedestrian suddenly running out as object information of the surrounding information of the moving object, information such as a location where near misses frequently occur as road information, and information such as a thunderstorm as weather information. Also, for example, when the emotion acquisition unit 122 infers from biometric data that the subject is in a state of "anxiety" or "tension," it may increase the confidence that the subject is in a state of "anxiety" or "tension" when it acquires information such as the presence of other vehicles as object information of the surrounding information of the moving object, information such as the road being congested as road information, and information such as heavy rain as weather information.

[0157] In this embodiment, the alcohol consumption fluctuation information acquisition unit 110Ca has only a heart rate acquisition unit 112, as shown in Figure 20. However, the alcohol consumption fluctuation information acquisition unit 110Ca may further include an alertness level acquisition unit 113 and an operation disorder level acquisition unit 115.

[0158] Next, an example of the processing flow of the alcohol detection device 100Ca shown in Figure 20 will be explained using Figure 21. Figure 21 is a flowchart showing the example of the processing flow of the alcohol detection device 100Ca shown in Figure 20. It should be assumed that the alcohol detection unit 130Ca of the alcohol detection device 100Ca has the internal configuration of the alcohol detection unit 130 shown in Figure 4.

[0159] When the alcohol detection device 100Ca receives a command from an external source, for example, from a control unit (not shown), to start processing, it begins the process shown in Figure 21. (Start)

[0160] In step ST3001, the heart rate acquisition unit 112 acquires the heart rate. The heart rate acquisition unit 112 outputs the acquired heart rate to the feature calculation unit 132.

[0161] Next, in step ST3002, the feature calculation unit 132 calculates heart rate features based on the heart rate obtained from the heart rate acquisition unit 112.

[0162] Next, in step ST3003, the drinking score calculation unit 131 calculates the drinking score based on the heart rate features calculated by the feature calculation unit 132.

[0163] In step ST3004, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in heart rate. If the behavior acquisition unit 123 determines, for example, that the subject is "eating or drinking" (step ST3005 "YES"), the process proceeds to step ST3010; if it determines that the subject is "not eating or drinking" (step ST3005 "NO"), the process proceeds to step ST3009.

[0164] In step ST3006, the emotion acquisition unit 122 acquires specific emotions that cause fluctuations in heart rate from the subject's biometric data. If the emotion acquisition unit 122 determines, for example, that the subject is "nervous" (step ST3007 "YES"), the process proceeds to step ST3010; if the subject is determined not to be nervous (step ST3007 "NO"), the process proceeds to step ST3009. Also in step ST3006, the emotion acquisition unit 122 acquires specific emotions that cause fluctuations in heart rate from external factors, namely information about the moving body's surroundings. If the emotion acquisition unit 122 determines, for example, that there are "external factors that could cause nervousness" (step ST3008 "YES"), the process proceeds to step ST3010; if the subject is determined not to have "external factors that could cause nervousness" (step ST3008 "NO"), the process proceeds to step ST3009.

[0165] Next, in steps ST3009 and ST3010, the alcohol confidence score, which is the confidence level of the alcohol score, is calculated based on the specific emotions obtained from the emotion acquisition unit 122 and the specific behaviors obtained from the behavior acquisition unit 123. If the process proceeds to step ST3009, the alcohol confidence calculation unit 133 does not change the alcohol confidence score. If the process proceeds to step ST3010, the alcohol confidence calculation unit 133 decreases the alcohol confidence score.

[0166] Next, in step ST3011, the alcohol determination unit 134 estimates the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. If the alcohol score is above the threshold and the alcohol confidence level is above the threshold (step ST3011 "YES"), proceed to step ST3013; otherwise (step ST3011 "NO"), proceed to step ST3012.

[0167] If the process proceeds to step ST3012, the alcohol detection unit 134 determines that the subject is not drinking alcohol (has not consumed alcohol). If the process proceeds to step ST3013, the alcohol detection unit 134 determines that the subject is intoxicated.

[0168] When the process reaches step ST3012 or step ST3013, the process shown in Figure 21 is terminated. (Termination) The process shown in Figure 21 may be executed repeatedly.

[0169] As described above, the alcohol detection device 100C in Embodiment 3 includes: an alcohol fluctuation information acquisition unit 110C that acquires alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior; an over-detection factor information acquisition unit 120C that acquires information about over-detection factors that are factors other than the subject's drinking behavior that cause the alcohol fluctuation information to fluctuate; and an alcohol detection unit 130C that estimates the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110C and the over-detection factor information acquired by the over-detection factor information acquisition unit 120C. With the above configuration, it is possible to suppress the system from falsely detecting that the subject is drinking when they are not, and to improve the accuracy of alcohol detection.

[0170] Furthermore, in the alcohol detection device 100C of Embodiment 3, the over-detection factor information acquisition unit 120C has both a biometric data acquisition unit 121C that acquires the subject's biometric data and a mobile body surrounding information acquisition unit 124 that acquires subject surrounding information, which is information indicating the situation around the mobile body. The over-detection factor information acquisition unit 120C acquires over-detection factor information from the mobile body surrounding information in addition to the subject's biometric data. With the above configuration, the detection accuracy of the over-detection factor information acquisition unit 120C is improved, and the accuracy of alcohol detection can be further improved compared to Embodiments 1 and 2.

[0171] In the above configuration, the overdetection factor information acquisition unit 120C was configured to have both a biological data acquisition unit 121C for acquiring biological data and a mobile body surrounding information acquisition unit 124 for acquiring mobile body surrounding information, thereby acquiring the subject's overdetection factor information from the mobile body surrounding information in addition to the subject's biological data. However, the overdetection factor information acquisition unit 120C may also have at least a mobile body surrounding information acquisition unit 124 for acquiring mobile body surrounding information, thereby acquiring the subject's overdetection factor information from at least the mobile body surrounding information. By doing so, even when biological data such as the subject's facial expressions cannot be acquired, the overdetection factor information can be acquired from the mobile body surrounding information, thus achieving the same effects as in Embodiments 1 and 2.

[0172] In the above, the subject is assumed to be a occupant of a mobile vehicle, and the mobile vehicle surrounding information is defined as information indicating the situation around the mobile vehicle. However, the subject is not limited to the occupant of a mobile vehicle, and the mobile vehicle surrounding information may also be subject surrounding information indicating the situation around the subject. In that case, the emotion acquisition unit 122 may infer the subject's specific emotion based on the subject surrounding information.

[0173] Embodiment 4. Embodiment 4 will be described below with reference to the drawings. First, Figure 22 will be used to describe a basic configuration example of the alcohol detection device 100D and the alcohol detection system 1000D according to Embodiment 4. Figure 22 is a configuration diagram showing a basic configuration example of the alcohol detection system 1000D including the alcohol detection device 100D according to Embodiment 4. The same configuration as in Embodiments 1 to 3 will not be described. Also, in Figure 22, the same reference numerals as in Figures 1 to 21 indicate the same or corresponding parts. The following description will focus on the differences from Embodiments 1 to 3.

[0174] In the alcohol detection device 100Ac shown in Figure 12 of Embodiment 1, the alcohol detection unit 130Ac estimated the subject's drinking state based on multiple types of alcohol fluctuation information, such as heart rate and alertness level, and information on factors causing false positives. Similarly, in the alcohol detection device 100Ba shown in Figure 16 of Embodiment 2, the alcohol detection unit 130Ba estimated the subject's drinking state based on multiple types of alcohol fluctuation information, such as heart rate and operational disorder level, and information on factors causing false positives. The alcohol detection device 100D of Embodiment 4 differs from the alcohol detection device 100Ac of Embodiment 1 and the alcohol detection device 100Ba of Embodiment 2 in that, similar to the configuration of the alcohol detection device 100Ac of Embodiment 1 and the alcohol detection device 100Ba of Embodiment 2, the alcohol detection unit 130D estimates the subject's drinking state based on multiple types of alcohol fluctuation information and information on factors causing false positives, and further includes an estimation model changing unit 135 that changes the estimation model of the alcohol detection unit 130D.

[0175] Furthermore, in Embodiment 4, the subject may be an occupant of the moving object, for example, the driver of the moving object. The moving object may also be a vehicle.

[0176] The estimation model modification unit 135 modifies the estimation model of the alcohol detection unit 130D by changing the alcohol fluctuation information that the alcohol detection unit 130D uses to estimate the subject's drinking status. Initially, when the alcohol detection unit 130D estimates the subject's drinking status based on multiple types of alcohol fluctuation information and false positive factor information indicating factors that cause fluctuations in at least one of the multiple types of alcohol fluctuation information and factors other than the subject's drinking behavior, the estimation model modification unit 135 may modify the estimation model of the alcohol detection unit 130D so that the alcohol detection unit 130D estimates the subject's drinking status based on other alcohol fluctuation information excluding the first alcohol fluctuation information and false positive factor information indicating factors that cause fluctuations in the other alcohol fluctuation information and factors other than the subject's drinking behavior. For example, in the initial setting, when the alcohol detection unit 130D estimates the subject's drinking status based on drinking fluctuation information, namely heart rate and alertness level, and false positive factor information indicating factors that cause fluctuations in at least one of heart rate or alertness level, and factors other than the subject's drinking behavior, the estimation model modification unit 135 may modify the estimation model of the alcohol detection unit 130D so that the alcohol detection unit 130D estimates the subject's drinking status based on alertness level excluding heart rate, and false positive factor information indicating factors that cause fluctuations in alertness level, and factors other than the subject's drinking behavior. Furthermore, as an initial setting, when the alcohol detection unit 130D estimates the subject's drinking status based on drinking fluctuation information such as heart rate and operational disorder, and false positive factor information indicating factors that cause fluctuations in at least one of the heart rate or operational disorder, and factors other than the subject's drinking behavior, the estimation model modification unit 135 may modify the estimation model of the alcohol detection unit 130D so that the alcohol detection unit 130D estimates the subject's drinking status based on operational disorder excluding heart rate, and false positive factor information indicating factors that cause fluctuations in operational disorder, and factors other than the subject's drinking behavior.Furthermore, as an initial setting, when the alcohol detection unit 130D estimates the subject's drinking state based on drinking variability information such as heart rate, alertness level, and operational disorder level, and false positive factor information indicating factors that cause fluctuations in at least one of the heart rate, alertness level, or operational disorder level, and factors other than the subject's drinking behavior, the estimation model modification unit 135 may modify the estimation model of the alcohol detection unit 130D so that the alcohol detection unit 130D estimates the subject's drinking state based on alertness level excluding heart rate, operational disorder level, and false positive factor information indicating factors that cause fluctuations in at least one of the alertness level or operational disorder level, and factors other than the subject's drinking behavior. Note that the first drinking variability information excluded from the estimation model modification unit 135 may be other than heart rate, for example, operational disorder level or alertness level. Also, the first drinking variability information excluded from the estimation model modification unit 135 may be not just one type of drinking variability information, but two or more types of drinking variability information. Furthermore, as an initial setting, when the alcohol detection unit 130D estimates the subject's drinking status based on first alcohol fluctuation information and false positive factor information indicating factors that cause the first alcohol fluctuation information to fluctuate and are factors other than the subject's drinking behavior, the estimation model modification unit 135 may modify the estimation model of the alcohol detection unit 130D so that the alcohol detection unit 130D estimates the subject's drinking status based on second alcohol fluctuation information (which is not the first alcohol fluctuation information) and false positive factor information indicating factors that cause the second alcohol fluctuation information to fluctuate and are factors other than the subject's drinking behavior. The estimation model modification unit 135 may be located inside the alcohol detection unit 130D or outside the alcohol detection unit 130D.

[0177] Furthermore, the estimation model modification unit 135 may determine whether or not to change the estimation model of the alcohol detection unit 130D based on alcohol fluctuation information and false positive factor information. For example, as an initial setting, when the alcohol detection unit 130D estimates the subject's drinking status based on multiple types of alcohol fluctuation information and false positive factor information indicating factors other than the subject's drinking behavior that cause fluctuations in at least one of the multiple types of alcohol fluctuation information, if false positive factor information A, which is part of the first false positive factor information indicating factors other than the subject's drinking behavior that cause fluctuations in the first alcohol fluctuation information, is acquired continuously over a certain period of time rather than temporarily, the estimation model of the alcohol detection unit 130D may be modified to estimate the subject's drinking status based on other alcohol fluctuation information excluding the first alcohol fluctuation information and false positive factor information indicating factors other than the subject's drinking behavior that cause fluctuations in the other alcohol fluctuation information. Furthermore, the estimation model modification unit 135 may determine whether or not to change the estimation model of the alcohol detection unit 130D based on at least the alcohol fluctuation information. For example, in the initial setup, when the alcohol detection unit 130D estimates the subject's drinking status based on false positive factor information and multiple types of alcohol fluctuation information, if the first alcohol fluctuation information cannot be acquired properly, the estimation model of the alcohol detection unit 130D may be changed to estimate the subject's drinking status based on other alcohol fluctuation information excluding the first alcohol fluctuation information and false positive factor information indicating factors that cause the other alcohol fluctuation information to fluctuate and are factors other than the subject's drinking behavior.

[0178] Next, a detailed example of the configuration of the alcohol detection unit 130D will be described using Figure 23. Figure 23 is a configuration diagram showing a detailed example of the configuration of the alcohol detection unit 130D of the alcohol detection device 100D according to Embodiment 4. Note that the same configuration as the alcohol detection unit 130 shown in Figure 4 will be omitted from the explanation.

[0179] As shown in Figure 23, the alcohol detection unit 130D includes an inference model modification unit 135 that modifies the inference model of the alcohol score calculation unit 131 within the alcohol detection unit 130D. The inference model modification unit 135 modifies the inference model of the alcohol score calculation unit 131 by changing the alcohol fluctuation information used by the alcohol score calculation unit 131 to calculate the alcohol score indicating the degree of the subject's drinking. The inference model is, for example, a machine learning model. In this embodiment, the inference model modification unit 135 modifies the inference model of the alcohol score calculation unit 131 by changing the feature quantities used by the alcohol score calculation unit 131 to calculate the alcohol score indicating the degree of the subject's drinking. As an initial setting, when the alcohol score calculation unit 131 calculates the subject's drinking score based on multiple types of alcohol fluctuation information, the inference model modification unit 135 may modify the inference model of the alcohol score calculation unit 131 so that the alcohol score calculation unit 131 calculates the subject's drinking score based on other alcohol fluctuation information excluding the first type of alcohol fluctuation information. For example, if the initial setting is such that the drinking score calculation unit 131 calculates the subject's drinking score based on drinking variability information, namely heart rate and alertness level, the estimation model modification unit 135 may modify the estimation model of the drinking score calculation unit 131 so that the drinking score calculation unit 131 calculates the subject's drinking score based on alertness level excluding heart rate. Alternatively, if the initial setting is such that the drinking score calculation unit 131 calculates the subject's drinking score based on drinking variability information, namely heart rate and operation disorder level, the estimation model modification unit 135 may modify the estimation model of the drinking score calculation unit 131 so that the drinking score calculation unit 131 calculates the subject's drinking score based on operation disorder level excluding heart rate. Furthermore, as an initial setting, when the drinking score calculation unit 131 calculates the subject's drinking score based on drinking variability information such as heart rate, alertness, and operational disorder, the estimation model modification unit 135 may modify the estimation model of the drinking score calculation unit 131 so that the drinking score calculation unit 131 calculates the subject's drinking score based on alertness and operational disorder, excluding heart rate. Note that the first drinking variability information excluded from the estimation model may be something other than heart rate, for example, operational disorder or alertness.Furthermore, the first alcohol consumption fluctuation information excluded from the estimation model may not be just one type of alcohol consumption fluctuation information, but two or more types. Also, as an initial setting, when the alcohol consumption score calculation unit 131 calculates the subject's alcohol consumption score based on the first alcohol consumption fluctuation information, the estimation model modification unit 135 may modify the estimation model of the alcohol consumption score calculation unit 131 so that the alcohol consumption score calculation unit 131 calculates the subject's alcohol consumption score based on a second type of alcohol consumption fluctuation information that is not the first type of alcohol consumption fluctuation information. The estimation model modification unit 135 may be located inside the alcohol consumption detection unit 130D or outside the alcohol consumption detection unit 130D.

[0180] Furthermore, the estimation model modification unit 135 may determine whether or not to change the estimation model of the drinking score calculation unit 131 based on the features and the confidence level of drinking. For example, as an initial setting, when the drinking score calculation unit 131 calculates a subject's drinking score from multiple types of features calculated based on multiple types of drinking fluctuation information, if a first false positive factor information indicating a factor that causes the first drinking fluctuation information to fluctuate and is a factor other than the subject's drinking behavior is continuously acquired, and the confidence level of drinking calculated based on the first false positive factor information remains below a threshold for a certain period of time, the estimation model of the drinking score calculation unit 131 may be modified to calculate the subject's drinking score from features calculated based on other drinking fluctuation information excluding the first drinking fluctuation information. Furthermore, the estimation model modification unit 135 may determine whether or not to change the estimation model of the drinking score calculation unit 131 based at least on the drinking fluctuation information. For example, if the drinking score calculation unit 131 calculates a subject's drinking score from multiple types of feature quantities calculated based on multiple types of drinking fluctuation information as an initial setting, and the first drinking fluctuation information cannot be obtained successfully, the estimation model of the drinking score calculation unit 131 may be changed so that it calculates the subject's drinking score from feature quantities calculated based on other drinking fluctuation information excluding the first drinking fluctuation information.

[0181] Next, an example of the processing flow of the alcohol detection device 100D having the alcohol detection unit 130D shown in Figure 23 will be explained using Figure 24. Figure 24 is a flowchart showing the example of the processing flow of the alcohol detection device 100D having the alcohol detection unit 130D shown in Figure 23.

[0182] The alcohol detection device 100D, which has an alcohol detection unit 130D as shown in Figure 23, starts the process shown in Figure 24 when, for example, a control unit (not shown) receives a command from an external source to start processing. (Start)

[0183] In step ST4001, the heart rate acquisition unit 112 acquires the heart rate. The heart rate acquisition unit 112 outputs the acquired heart rate to the feature calculation unit 132. Also in step ST4001, the arousal level acquisition unit 113 acquires the arousal level. The arousal level acquisition unit 113 outputs the acquired arousal level to the feature calculation unit 132.

[0184] Next, in step ST4002, the feature calculation unit 132 calculates heart rate features based on the heart rate acquired from the heart rate acquisition unit 112. Also in ST4002, the feature calculation unit 132 calculates arousal features based on the arousal level acquired from the arousal level acquisition unit 113.

[0185] In step ST4003, the emotion acquisition unit 122 acquires specific emotions that cause fluctuations in heart rate. If the emotion acquisition unit 122 determines, for example, that the subject is "nervous" (step ST4004 "YES"), the process proceeds to step ST4009; if the subject is determined not to be nervous (step ST4004 "NO"), the process proceeds to step ST4008.

[0186] In step ST4005, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in heart rate. If the behavior acquisition unit 123 determines, for example, that the subject is "eating or drinking" (step ST4006 "YES"), the process proceeds to step ST4009; if the subject is determined to be "not eating or drinking" (step ST4006 "NO"), the process proceeds to step ST4008. Also in step ST4005, the behavior acquisition unit 123 acquires specific behaviors that cause fluctuations in arousal level. If the behavior acquisition unit 123 determines, for example, that the subject is "talking" (step ST4007 "YES"), the process proceeds to step ST4009; if the subject is determined to be "not talking" (step ST4007 "NO"), the process proceeds to step ST4008.

[0187] Next, in steps ST4008 and ST4009, the alcohol confidence score, which is the confidence level of the alcohol score, is calculated based on the specific emotions obtained from the emotion acquisition unit 122 and the specific behaviors obtained from the behavior acquisition unit 123. If the process proceeds to step ST4008, the alcohol confidence calculation unit 133 does not change the alcohol confidence score. If the process proceeds to step ST4009, the alcohol confidence calculation unit 133 decreases the alcohol confidence score.

[0188] Next, in step ST4010, the prediction model modification unit 135 determines whether or not to modify the prediction model of the drinking score calculation unit 131 based on the features calculated by the feature calculation unit 132 and the drinking confidence level calculated by the drinking confidence level calculation unit 133. If at least one of "tension" in step ST4004 or "eating and drinking behavior" in step ST4006 is obtained consecutively, and the drinking confidence level decreases and falls below the threshold, then in step ST4010, it is determined whether the state of drinking confidence level being below the threshold continued for a certain period of time. If it continued for a certain period of time (step ST4010 "YES"), proceed to step ST4012; if it did not continue for a certain period of time (step ST4010 "NO"), proceed to step ST4011.

[0189] In step ST4011, the drinking score calculation unit 131 calculates the drinking score using an estimation model that utilizes heart rate features and arousal features. In step ST4012, the drinking score calculation unit 131 calculates the drinking score using an estimation model that utilizes arousal features, excluding heart rate features.

[0190] Next, in step ST4013, the alcohol determination unit 134 estimates the subject's drinking status based on the alcohol score calculated by the alcohol score calculation unit 131 and the alcohol confidence level calculated by the alcohol confidence level calculation unit 133. If the alcohol score is above the threshold and the alcohol confidence level is above the threshold (step ST4013 "YES"), proceed to step ST4015; otherwise (step ST4013 "NO"), proceed to step ST4014.

[0191] If the process proceeds to step ST4014, the alcohol detection unit 134 determines that the subject is not drinking alcohol (has not consumed alcohol). If the process proceeds to step ST4015, the alcohol detection unit 134 determines that the subject is intoxicated.

[0192] When the process reaches step ST4014 or step ST4015, the process shown in Figure 24 is terminated. (Termination) Note that the process shown in Figure 24 may be executed repeatedly.

[0193] As described above, the alcohol detection device 100D in Embodiment 4 includes: an alcohol fluctuation information acquisition unit 110D that acquires alcohol fluctuation information indicating information about the subject that fluctuates due to the subject's drinking behavior; an over-detection factor information acquisition unit 120D that acquires information about over-detection factors that are factors other than the subject's drinking behavior that cause the alcohol fluctuation information to fluctuate; and an alcohol detection unit 130D that estimates the subject's drinking status based on the alcohol fluctuation information acquired by the alcohol fluctuation information acquisition unit 110D and the over-detection factor information acquired by the over-detection factor information acquisition unit 120D. With the above configuration, it is possible to suppress the system from falsely detecting that the subject is drinking when they are not, and to improve the accuracy of alcohol detection.

[0194] Furthermore, the alcohol detection device 100D of Embodiment 4 further includes an inference model changing unit 135 that changes the inference model of the alcohol detection unit 130D, and the inference model changing unit 135 changes the alcohol fluctuation information that the alcohol detection unit 130D uses to estimate the drinking state of the subject. With the above configuration, the alcohol fluctuation information acquisition unit 110D can acquire multiple types of alcohol fluctuation information, and for example, if the first alcohol fluctuation information among the multiple types of alcohol fluctuation information cannot be acquired normally, the alcohol detection unit 130D can estimate the drinking state of the subject without using the first alcohol fluctuation information, thereby improving the alcohol detection accuracy compared to continuing to use an inference model that uses the first alcohol fluctuation information. For example, when vehicle 1 is operating autonomously, the subject does not operate vehicle 1, so the operation randomness among the multiple types of alcohol fluctuation information cannot be acquired normally. Even in that case, the alcohol detection unit 130D can estimate the drinking state of the subject without using the operation randomness, thereby improving the alcohol detection accuracy compared to continuing to use an inference model that uses the operation randomness. Furthermore, with the above configuration, the alcohol fluctuation information acquisition unit 110D can acquire multiple types of alcohol fluctuation information. For example, if a first false positive factor information, which indicates a factor that causes the first alcohol fluctuation information to fluctuate and is a factor other than the subject's drinking behavior, is continuously acquired from among the multiple types of alcohol fluctuation information, and if the alcohol confidence level calculated based on the first false positive factor information remains below a threshold for a certain period of time, the alcohol score calculation unit 131 can calculate the subject's alcohol score from the feature quantities calculated based on the other alcohol fluctuation information excluding the first alcohol fluctuation information. In other words, since the alcohol detection unit 130D can estimate the subject's drinking state without using the first alcohol fluctuation information, the alcohol score calculation unit 131 can improve the accuracy of alcohol detection compared to continuing to use an estimation model that uses feature quantities calculated based on the first alcohol fluctuation information. For example, if at least one of "tension" or "eating and drinking behavior" is acquired consecutively, and the confidence level of drinking decreases and falls below a threshold, and if the confidence level of drinking remains below the threshold for a certain period of time, it means that the reliability of the heart rate information is low, making it difficult for the alcohol detection unit 130D to estimate the subject's drinking status using the heart rate.Even in that case, the alcohol score calculation unit 131 can calculate the subject's alcohol score without using heart rate features, thus improving the accuracy of alcohol detection compared to the alcohol score calculation unit 131 continuing to use an estimation model that uses heart rate features.

[0195] Next, the hardware configuration of the alcohol detection device 100 (100A, 100B, 100C, 100D) will be described. Figure 25 is a schematic diagram showing example 1 of the hardware configuration of the alcohol detection device 100 (100A, 100B, 100C, 100D) according to Embodiments 1 to 4. Figure 26 is a schematic diagram showing example 2 of the hardware configuration of the alcohol detection device 100 (100A, 100B, 100C, 100D) according to Embodiments 1 to 4. Each of the alcohol detection devices 100 (100A, 100B, 100C, 100D) is implemented by the hardware shown in Figure 25 or Figure 26.

[0196] Each of the alcohol detection devices 100 (100A, 100B, 100C, 100D) is composed of, for example, a processor 10001, a memory 10002, an input / output interface 10003, and a communication circuit 10004, as shown in Figure 25. The processor 10001 and memory 10002 are, for example, mounted on a computer. When the alcohol fluctuation information acquisition unit 110 (110A, 110B, 110C, 110D), the over-detection factor information acquisition unit 120 (120A, 120B, 120C, 120D), the alcohol detection unit 130 (130A, 130B, 130C, 130D), and the control unit (not shown) are all part of the processor 10001, the functions of each of the above components are realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 10002. In other words, the memory 10002 stores a program that causes the computer to function as an alcohol fluctuation information acquisition unit 110 (110A, 110B, 110C, 110D), an over-detection factor information acquisition unit 120 (120A, 120B, 120C, 120D), an alcohol detection unit 130 (130A, 130B, 130C, 130D), and a control unit (not shown). When the processor 10001 reads and executes the program stored in the memory 10002, the program and each piece of hardware cooperate to realize the functions of the alcohol fluctuation information acquisition unit 110 (110A, 110B, 110C, 110D), the over-detection factor information acquisition unit 120 (120A, 120B, 120C, 120D), the alcohol detection unit 130 (130A, 130B, 130C, 130D), and the control unit (not shown). The program causes the computer to execute the procedures or methods of each of the above components. Furthermore, a DB (including a dictionary database related to machine learning detectors such as the arousal level acquisition unit 113, the operation disorder level acquisition unit 115, the emotion acquisition unit 122, the behavior acquisition unit 123, and the drinking score calculation unit 131) and a storage unit (not shown) are realized by memory 10002 or other memory (not shown). Additionally, a communication unit (not shown) is realized by communication circuit 10004.

[0197] The processor 10001 uses, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a microprocessor, a microcontroller, or a DSP (Digital Signal Processor). The memory 10002 may be a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable Read Only Memory), or flash memory; it may be a magnetic disk such as a hard disk or flexible disk; it may be an optical disk such as a CD (Compact Disc) or DVD (Digital Versatile Disc); or it may be a magneto-optical disk. The processor 10001 and the memory 10002 or the communication circuit 10004 are connected in a manner that enables them to transmit data to each other. Furthermore, the processor 10001, the memory 10002, and the communication circuit 10004 are connected in a manner that allows them to mutually transmit data with other hardware via the input / output interface 10003.

[0198] Alternatively, the functions of the alcohol fluctuation information acquisition unit 110 (110A, 110B, 110C, 110D), the over-detection factor information acquisition unit 120 (120A, 120B, 120C, 120D), the alcohol detection unit 130 (130A, 130B, 130C, 130D), and the control unit (not shown) in the alcohol detection device 100 (100A, 100B, 100C, 100D) may be realized by a dedicated processing circuit 20001, as shown in Figure 26.

[0199] The processing circuit 20001 may, for example, be a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), an FPGA (Field-Programmable Gate Array), a SoC (System-on-a-Chip), or a system LSI (Large-Scale Integration). Alternatively, a combination of these may be used. Furthermore, a DB (including a dictionary database related to machine learning detectors such as the arousal level acquisition unit 113, the operation disorder level acquisition unit 115, the emotion acquisition unit 122, the behavior acquisition unit 123, the drinking score calculation unit 131, etc.) and a storage unit (not shown) are realized by memory 20002 or other memory (not shown). The memory 20002 may be a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable Read Only Memory), or flash memory; it may be a magnetic disk such as a hard disk or flexible disk; it may be an optical disk such as a CD (Compact Disc) or DVD (Digital Versatile Disc); or it may be a magneto-optical disk. Furthermore, a communication unit (not shown) is realized by the communication circuit 20004. The processing circuit 20001 and the memory 20002 or the communication circuit 20004 are connected in a manner that allows them to transmit data to each other. Furthermore, the processing circuit 20001, the memory 20002, and the communication circuit 20004 are connected in a manner that allows them to transmit data to each other with other hardware via the input / output interface 20003.In addition, some functions of the alcohol detection device 100 (100A, 100B, 100C, 100D), including the alcohol fluctuation information acquisition unit 110 (110A, 110B, 110C, 110D), the over-detection factor information acquisition unit 120 (120A, 120B, 120C, 120D), the alcohol detection unit 130 (130A, 130B, 130C, 130D), and the control unit (not shown), may be implemented by the processor 10001 and memory 10002, while the remaining functions are implemented by the processing circuit 20001. In this way, each of the functions of the above-mentioned components can be implemented by hardware, software, firmware, or a combination thereof.

[0200] In the embodiments described herein, the materials, dimensions, shapes, relative arrangements, or conditions for implementation of each component may be described, but these are all illustrative examples and are not limited to those described in each embodiment. Therefore, countless variations not illustrated are conceivable within the scope of each embodiment. For example, this includes modifying, adding, or omitting any component, or even extracting at least one component from at least one embodiment and combining it with a component from another embodiment.

[0201] It goes without saying that the objectives of the present invention can be achieved, and various design modifications are possible without departing from the spirit of the present invention.

[0202] The alcohol detection device described herein can be used, for example, to detect alcohol in the occupants of a vehicle.

[0203] 100 Alcohol detection device, 110 Alcohol fluctuation information acquisition unit, 111 Biometric data acquisition unit, 112 Heart rate acquisition unit, 113 Arousal level acquisition unit, 114 Moving body information acquisition unit, 115 Operational disorder level acquisition unit, 120 False detection factor information acquisition unit, 121 Biometric data acquisition unit, 122 Emotion acquisition unit, 123 Behavior acquisition unit, 124 Moving body surrounding information acquisition unit, 130 Alcohol detection unit, 131 Alcohol score calculation unit, 133 Alcohol confidence level calculation unit, 135 Prediction model modification unit, 1000 Alcohol detection system, 1100 Output device

Claims

A drinking fluctuation information acquisition unit acquires drinking fluctuation information that shows information about the subject that fluctuates due to the subject's drinking behavior, An over-detection factor information acquisition unit acquires over-detection factor information that indicates factors other than the drinking behavior that cause the aforementioned drinking fluctuation information to fluctuate, An alcohol detection device comprising: an alcohol detection unit that estimates the drinking status of a subject based on the alcohol fluctuation information and the over-detection factor information. 　 The system further comprises a drinking score calculation unit that calculates a drinking score indicating the degree of the subject's drinking based on the drinking fluctuation information acquired by the drinking fluctuation information acquisition unit, The alcohol detection device according to claim 1, wherein the alcohol detection unit estimates the drinking status of the subject based on the alcohol score and the false detection factor information. 　 The alcohol detection unit compares the alcohol score calculated by the alcohol score calculation unit with a threshold, and determines that the subject is intoxicated if the alcohol score is equal to or greater than the threshold. The alcohol detection device according to claim 2, wherein the alcohol score calculation unit reduces the alcohol score when the over-detection factor information is acquired by the over-detection factor information acquisition unit. 　 The system further includes an alcohol confidence calculation unit that calculates an alcohol confidence level, which is the confidence level of the alcohol score, based on the over-detection factor information acquired by the over-detection factor information acquisition unit, The alcohol detection device according to claim 2, wherein the alcohol detection unit estimates the drinking status of the subject based on the drinking score and the alcohol confidence level. 　 The alcohol detection unit compares the alcohol score calculated by the alcohol score calculation unit with a first threshold, and compares the alcohol confidence calculated by the alcohol confidence calculation unit with a second threshold. If the alcohol score is equal to or greater than the first threshold and the alcohol confidence is equal to or greater than the second threshold, the unit determines that the subject is intoxicated. The alcohol detection device according to claim 4, wherein the alcohol confidence calculation unit lowers the alcohol confidence when the over-detection factor information is acquired by the over-detection factor information acquisition unit. 　 The alcohol detection device according to claim 1, wherein the alcohol detection unit stops inferring the subject's drinking status when the over-detection factor information is acquired by the over-detection factor information acquisition unit. 　 The aforementioned alcohol consumption fluctuation information is the heart rate of the subject, The alcohol detection device according to any one of claims 1 to 6, wherein the alcohol fluctuation information acquisition unit has a heart rate acquisition unit that acquires the heart rate of the subject. 　 The aforementioned alcohol consumption fluctuation information is the level of alertness of the subject, The alcohol detection device according to any one of claims 1 to 7, wherein the alcohol fluctuation information acquisition unit has an alertness acquisition unit that acquires the alertness level of the subject. 　 The aforementioned false positive factor information is the subject's specific emotion, The alcohol detection device according to any one of claims 1 to 7, wherein the over-detection factor information acquisition unit has an emotion acquisition unit that acquires the specific emotion. 　 The alcohol detection device according to claim 9, wherein the emotion acquisition unit estimates the subject's specific emotion based on subject surrounding information, which is information indicating the circumstances surrounding the subject. 　 The aforementioned false positive factor information is the specific behavior of the subject, The alcohol detection device according to any one of claims 1 to 10, wherein the over-detection factor information acquisition unit has an action acquisition unit that acquires the specific action. 　 The subject is the driver of the vehicle, The aforementioned alcohol fluctuation information is a degree of driving disorder that indicates the degree of roughness in the driver's driving operations. The alcohol detection device according to any one of claims 1 to 11, wherein the alcohol fluctuation information acquisition unit has an operation randomness acquisition unit that acquires the operation randomness. 　 The aforementioned moving object is a vehicle, The alcohol detection device according to claim 12, wherein the degree of operational disorder is calculated based on at least one of the driver's steering, braking, or acceleration operations on the vehicle, or the vehicle motion resulting from each of these operations. 　 The aforementioned subject is a occupant of a mobile vehicle, The alcohol detection device according to claim 10, wherein the subject surrounding information is mobile body surrounding information, which is information indicating the situation around the mobile body. 　 The aforementioned alcohol consumption fluctuation information acquisition unit is capable of acquiring multiple types of alcohol consumption fluctuation information, The unit further comprises a prediction model changing unit that changes the prediction model of the alcohol detection unit, The alcohol detection device according to any one of claims 1 to 14, wherein the estimation model modification unit modifies the alcohol fluctuation information used by the alcohol detection unit to estimate the drinking status of the subject. 　 A drinking detection device according to any one of claims 1 to 15, An output device that acquires the aforementioned alcohol consumption fluctuation information and the aforementioned over-detection factor information, outputs the alcohol consumption fluctuation information to the alcohol consumption fluctuation information acquisition unit, and outputs the over-detection factor information to the over-detection factor information acquisition unit, A drinking detection system equipped with [a specific feature]. 　 The drinking fluctuation information acquisition unit acquires drinking fluctuation information that shows the subject's information which fluctuates due to the subject's drinking behavior, The process involves an over-detection factor information acquisition unit acquiring over-detection factor information that indicates factors other than the drinking behavior that cause the drinking fluctuation information to fluctuate, A method for detecting alcohol consumption, comprising the step of an alcohol consumption detection unit estimating the drinking status of a subject based on the alcohol consumption fluctuation information and the over-detection factor information. 　 The drinking fluctuation information acquisition unit acquires drinking fluctuation information that shows the subject's information which fluctuates due to the subject's drinking behavior, The process involves an over-detection factor information acquisition unit acquiring over-detection factor information that indicates factors other than the drinking behavior that cause the drinking fluctuation information to fluctuate, An alcohol detection program that causes a computer to perform the following steps: an alcohol detection unit estimates the drinking status of the subject based on the alcohol fluctuation information and the over-detection factor information.

Images