Physique determination device and physique determination method

Abstract

Provided is a physique determination device (1) for determining the physique of a living body present in a target space on the basis of a moving body map in which a moving body present within the target space is represented as a three-dimensional space distribution, the moving body map being generated by performing signal processing of reflected waves produced by radio waves radiated into the target space by a radio wave sensor (2) being reflected by an object within the target space, said physique determination device (1) comprising a movement estimation unit (13) for estimating the movement of the living body on the basis of the moving body map, and a physique determination unit (14) for determining the physique of the living body on the basis of the moving body map and movement estimation information indicating the movement of the living body estimated by the movement estimation unit (13).

Description

Physique determination device and physique determination method

[0001] The present disclosure relates to a physique determination device and a physique determination method.

[0002] In recent years, efforts have been made to determine the physical size of a living organism present in a target space (hereinafter referred to as the "target space") based on a three-dimensional map (hereinafter referred to as the "dynamic body map") that represents the three-dimensional spatial distribution of moving objects present in the target space. The map is generated by signal processing of radio waves emitted by a radio wave sensor toward the space where a living organism may be present and reflected by objects in the target space. The results of determining the physical size of the living organism based on the dynamic body map are used for various control purposes. For example, if the target space is the interior of a vehicle, the physical size of the living organism in the vehicle is determined based on the dynamic body map, and the results of determining the physical size of the living organism are used to detect whether a child or pet has been left behind or to control airbag deployment. For example, Patent Document 1 discloses a technology in which a vehicle interior detection map (equivalent to a moving object map) is generated based on the radio waves emitted by a millimeter wave sensor toward the vehicle interior and reflected by reflecting objects present in the vehicle interior, and a trajectory is generated for the position in a predetermined direction estimated as the range of an object determined to be present in the vehicle interior, and a fluctuation component that can be estimated as chest pulsation contained in the waveform of the trajectory is compared with a threshold value to determine whether the occupant is an adult or a child.

[0003] Japanese Patent Application Laid-Open No. 2022-182342

[0004] In the moving object map, locations corresponding to moving objects moving in the target space appear as a point cloud. That is, when a living body is present in the target space, the moving object map has the property that the distribution trend of the point cloud that appears can change significantly depending on the movement of the living body. Because the above-mentioned conventional technology does not take this into consideration, depending on the movement of the living body, it is not possible to properly obtain a fluctuation component that can be estimated as chest pulsation contained in the waveform as a trajectory generated from the vehicle interior detection map, which may result in an erroneous determination of the living body's physique.

[0005] The present disclosure has been made to solve the above-mentioned problems, and aims to provide a physique determination device that determines the physique of a living organism present in a target space based on a moving body map that represents the three-dimensional spatial distribution of moving bodies present in the target space, generated by signal processing of radio waves irradiated toward the target space by a radio wave sensor and reflected by objects in the target space, and that takes the movement of the living organism into consideration to prevent erroneous determination of physique.

[0006] The physique determination device of the present disclosure is a physique determination device that determines the physique of a living organism present in a target space based on a moving body map that represents moving bodies present in the target space as a three-dimensional spatial distribution, generated by signal processing of radio waves irradiated toward the target space by a radio wave sensor and reflected by objects in the target space, and is equipped with a motion estimation unit that estimates the movement of the living organism based on the moving body map, and a physique determination unit that determines the physique of the living organism based on the moving body map and motion estimation information that indicates the movement of the living organism estimated by the motion estimation unit.

[0007] According to the present disclosure, erroneous determination of physique can be prevented by taking into consideration the movement of the living body.

[0008] 6A, 6B, 6C, and 6D are diagrams illustrating an estimated head and torso rectangle according to the first embodiment. FIG. 6B is a diagram illustrating an estimated foot region according to the first embodiment. FIG. 6C is a diagram illustrating a method in which a motion estimation unit estimates motion of a living organism based on a plurality of time-series motion maps according to the first embodiment. FIG. 6D is a diagram illustrating an example of a motion map generated by a motion map generation unit according to the first embodiment, the diagram illustrating an example of a motion map generated when an infant is seated in a rearward-facing child car seat installed in the left rear seat in the vehicle interior. FIG. 6C is a diagram illustrating an example of a motion map generated by a motion map generation unit according to the first embodiment. FIG. 6D is a diagram illustrating an example of a motion map generated by a motion map generation unit according to the first embodiment, the diagram illustrating an example of a motion map generated when an infant is seated in a rearward-facing child car seat installed in the left rear seat in the vehicle interior. FIG. 6C is a diagram illustrating an example of a motion map generated by a motion map generation unit according to the first embodiment. 11A and 11B are diagrams illustrating an example of a hardware configuration of a physique determination device according to Embodiment 1. FIG.

[0009] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. Embodiment 1. Fig. 1 is a block diagram showing an example configuration of a physique determination device 1 according to embodiment 1. The physique determination device 1 according to embodiment 1 is connected to a radio wave sensor 2, and the physique determination device 1 and the radio wave sensor 2 together constitute a physique determination system 100.

[0010] The physique determination device 1 acquires sensor data from the radio wave sensor 2 based on radio waves emitted by the radio wave sensor 2 toward a space (hereinafter referred to as the "target space") in which a living body may exist and reflected by objects in the target space. The acquired sensor data is then signal-processed to generate a three-dimensional map (hereinafter referred to as the "dynamic body map") representing a three-dimensional spatial distribution of moving bodies present in the target space. The physique determination device 1 then estimates the movement of the living body based on the generated dynamic body map, and determines the physique of the living body present in the target space based on information indicating the estimated movement of the living body (hereinafter referred to as "motion estimation information") and the dynamic body map. In the following first embodiment, as an example, the target space is a vehicle interior, and the physique determination device 1 determines the physique of an occupant present in the vehicle interior. In the first embodiment, the physique determination result of the occupant by the physique determination device 1 is output to, for example, a vehicle control device (not shown) and used by the vehicle control device to perform control such as abandoned body detection, seat belt control, or intrusion detection.

[0011] In the first embodiment, the physique determination device 1 and the radio wave sensor 2 are mounted on a vehicle (not shown). The radio wave sensor 2 is a sensor that emits radio waves toward a target space and observes a predetermined specific range of the target space based on the reflected waves, and includes a millimeter-wave radar. In the first embodiment, as an example, the radio wave sensor 2 is assumed to be a millimeter-wave radar. The radio wave sensor 2 detects objects within the vehicle cabin. The radio wave sensor 2 acquires, as sensor data, reflected waves that are generated when the radio waves emitted toward the vehicle cabin are reflected by objects within the vehicle cabin. The radio wave sensor 2 is equipped with, for example, an antenna with wide-angle directivity and is installed within the vehicle cabin so that the radio waves are irradiated to any occupants who may be present within the vehicle cabin. Note that the radio waves emitted from the radio wave sensor 2 also propagate outside the vehicle cabin. Therefore, the radio wave sensor 2 can also detect people and the like around the vehicle.

[0012] In the first embodiment, the radio wave sensor 2 is assumed to be a general radio wave sensor that detects objects. Here, an example of how the radio wave sensor 2 acquires sensor data will be described. While various modulation methods are available for the sensing signals of radio wave sensors, the following describes an example in which the FM-CW (Frequency Modulation - Continuous Wave) method, which is commonly used in automotive applications, is used as the modulation method. The radio wave transmitter / receiver (not shown) included in the radio wave sensor 2 periodically generates an FM signal (called a chirp wave) whose frequency increases and decreases. The radio wave transmitter / receiver amplifies the signal power to obtain the power required for radio wave emission and emits radio waves into the space within the vehicle cabin via a transmitting antenna (not shown). When the radio waves emitted into the space within the vehicle cabin reach a target object within the radio wave emission range of the radio wave transmitter / receiver, a portion of the waves is reflected by the surface of the target object and returns to the radio wave transmitter / receiver. Here, the target object refers to an object that reflects radio waves, such as a passenger in the vehicle or a vehicle structure.

[0013] The radio wave transmitting / receiving unit receives radio waves (reflected waves) reflected from the surface of the target object via a receiving antenna (not shown). A signal similar to the FM transmitted wave is input as a received FM signal to the radio wave transmitting / receiving unit. The received signal is input to the radio wave transmitting / receiving unit with a time lag corresponding to the time it takes for the radio waves to reach the target object and return. The radio wave transmitting / receiving unit extracts the frequency difference between the frequency of the generated FM signal and the frequency of the received signal, and generates an intermediate frequency (IF) signal having the frequency difference. An A / D conversion unit (not shown) included in the radio wave sensor 2 converts the intermediate frequency signal from an analog signal to a digital signal, acquires the digital signal as sensor data, and outputs the acquired sensor data to the physique determination device 1.

[0014] Although only one radio wave sensor 2 is illustrated in FIG. 1 , this is merely an example. The type and number of radio wave sensors 2 may be single or multiple. For example, multiple radio wave sensors 2 may be provided in the vehicle, and multiple radio wave sensors 2 may be connected to the body size determination device 1. For example, the radio wave sensor 2 may be provided in a position that allows a view of the entire vehicle interior, such as an overhead console, so that a single radio wave sensor 2 can cover the entire vehicle interior. Alternatively, the radio wave sensor 2 may be provided in each area that can be viewed by the radio wave sensor 2, such as each area corresponding to a seat, or in each area of ​​the vehicle interior that has been previously divided into multiple areas. Furthermore, the operation of the radio wave sensor 2 may be controlled by the radio wave sensor 2 alone using a trigger inside the radio wave sensor 2, or may be controlled based on a trigger external to the radio wave sensor 2.

[0015] A description will be given of an example of the configuration of the physique determination device 1. As shown in Fig. 1, the physique determination device 1 includes a sensor data acquisition unit 11, a data processing unit 12, a movement estimation unit 13, and a physique determination unit .

[0016] The sensor data acquisition unit 11 acquires sensor data from the radio wave sensor 2. The sensor data acquisition unit 11 outputs the acquired sensor data to the data processing unit 12.

[0017] The data processing unit 12 generates a motion map by performing signal processing on the sensor data for a predetermined time period output from the sensor data acquisition unit 11. The data processing unit 12 outputs the generated motion map to the movement estimation unit 13 and the physique determination unit 14. The data processing unit 12 may output the motion map to the physique determination unit 14 via the movement estimation unit 13, or may output it directly. Note that the arrow from the data processing unit 12 to the physique determination unit 14 is omitted in FIG. 1 .

[0018] 2 is a block diagram showing a detailed configuration example of the data processing unit 12 in the physique determination device 1 according to Embodiment 1. As shown in FIG. 2 , the data processing unit 12 includes a moving body extraction unit 121, a moving body analysis unit 122, and a moving body map generation unit 123.

[0019] The moving object extraction unit 121 extracts a moving object based on the sensor data acquired by the sensor data acquisition unit 11 from the radio wave sensor 2. In the first embodiment, the moving object extracted by the moving object extraction unit 121 is a living organism present in the target space and is the target for physique determination. Specifically, the moving object extracted by the moving object extraction unit 121 is an occupant present in the vehicle cabin. The moving object extraction unit 121 extracts the moving object by detecting movement within the vehicle cabin based on the sensor data. The movement within the vehicle cabin includes movement due to changes in the position of the occupant present in the vehicle cabin and body movement such as chest movement associated with breathing. One possible method for the moving object extraction unit 121 to extract a moving object is to apply a known MTI (Moving Target Indicator) filter to the sensor data, in this case, the received signal acquired by the radio wave sensor 2, more specifically, the intermediate frequency signal based on the received signal. By applying the MTI filter, signal components due to objects such as completely stationary structures in the vehicle cabin, such as seats, are removed, and signal components due to the moving object are extracted. In this case, the moving body extraction unit 121 may apply an MTI filter to movements with speeds above a certain level to remove signal components, thereby suppressing the influence of large movements that adversely affect the occupant's physique determination process performed in the physique determination device 1 and obtaining a moving body map closer to a normal state that facilitates physique determination. An example of a large movement that adversely affects the occupant's physique determination process is a movement such as raising an arm. For example, the moving body extraction unit 121 may apply an MTI filter that emphasizes only movements at a speed similar to the rate at which humans breathe (approximately 10 to 50 brpm). The moving body extraction unit 121 may also adjust the period of the radio waves emitted by the radio wave sensor 2 to prevent fast movements from being detected. In the physique determination device 1, the occupant's physique determination process is performed by the physique determination unit 14. The moving body map generation unit 123 generates the moving body map. Details of the physique determination unit 14 and the moving body map generation unit 123 will be described later. The moving object extraction unit 121 outputs the sensor data from which the signal components due to the moving object have been extracted to the moving object analysis unit 122 .

[0020] Based on the sensor data output from the moving object extraction unit 121, the moving object analysis unit 122 performs frequency analysis on the received signal indicating the moving object extracted by the moving object extraction unit 121 to extract information such as the three-dimensional position, speed, and signal strength of the moving object (hereinafter referred to as "moving object information"), and outputs the extracted moving object information to the moving object map generation unit 123. The moving object analysis unit 122 can extract the moving object information based on various known methods or procedures such as Fourier transform (FFT), integration processing, peak extraction, and beamforming.

[0021] The moving object map generation unit 123 generates a moving object map based on the moving object information output from the moving object analysis unit 122. The moving object map generated by the moving object map generation unit 123 represents the moving objects present in the vehicle cabin as a three-dimensional spatial distribution. In other words, the moving object map represents the distribution of the area in the vehicle cabin where the moving objects exist in three dimensions. More specifically, the moving object map represents the minute movements of the moving objects in the vehicle cabin, in other words, the target objects that reflected the radio waves emitted by the radio wave sensor 2, using multiple grids that correspond to the reflection points of the radio waves in three-dimensional space. In the first embodiment, each grid in the moving object map is also referred to as a point cloud. Note that the intermediate frequency signal is generated from the reflected waves of the radio waves reflected by the surface of the target object. The moving object analysis unit 122 can calculate the positions of the reflection points on the surface of the target object from the intermediate frequency signal and extract moving object information. Since there are multiple reflection points on the surface of the target object, the moving object map generation unit 123 can generate a three-dimensional spatial distribution for the target object based on the moving object information extracted by signal processing the reflected waves from the multiple reflection points.

[0022] 3A and 3B are diagrams showing an example of a moving object map generated by the moving object map generation unit 123 in the first embodiment. FIG. 3A shows an example of a top view of the moving object map, and FIG. 3B shows an example of a side view of the moving object map. The moving object map shown in FIGS. 3A and 3B is a moving object map generated when an adult is properly seated in the left rear seat in the vehicle interior. In the first embodiment, a properly seated state refers to a state in which the occupant's posture is not compromised and the occupant is seated deep in the seat along the backrest.

[0023] In a moving object map, for example, each point cloud is assigned a numerical value corresponding to the speed of a target object present at the point cloud's position. Specifically, the greater the speed of a target object present at the point cloud's position, the greater the numerical value assigned to that point cloud. In other words, in a moving object map, a point cloud included in an area where a moving object exhibiting slight motion is assigned a numerical value greater than a point cloud where no moving object is present. Note that moving object information includes information on the target object's position, signal strength (in other words, reflection intensity), and target object's speed. In a moving object map, each point cloud is assigned a numerical value corresponding to at least one of the target object's speed, reflection intensity, and moving object's position, and all of these numerical values ​​may be assigned. In the moving object maps shown in FIGS. 3A and 3B, point clouds with larger assigned numerical values ​​are displayed darker. In other words, in the moving object maps shown in FIGS. 3A and 3B, point clouds included in an area where a moving object is present are displayed darker. Note that stationary objects, in other words, reflected wave components with a speed of zero, do not appear in the moving object map.

[0024] Based on the moving object maps as shown in Figures 3A and 3B, the position and silhouette of a living body present in the vehicle cabin can be confirmed. That is, based on the moving object maps as shown in Figures 3A and 3B, the presence or absence, position, and approximate physique of a living body present in the vehicle cabin can be determined. The physique determination unit 14 determines the physique of the living body. The physique determination unit 14 will be described later.

[0025] In this way, in the data processing unit 12, the moving body map generation unit 123 generates a moving body map that shows only the presence of moving bodies based on the moving body information output from the moving body analysis unit 122, so that the moving body map generation unit 123 can exclude the influence of stationary structures such as seats from the moving body map and make it information that is limited to the living organism that is the subject of physique judgment.

[0026] The moving object map generating unit 123 outputs the generated moving object map to the motion estimating unit 13 .

[0027] The behavior of the moving object map generated by the moving object map generation unit 123 also changes depending on the specifications of the radio wave sensor 2. Figures 4A, 4B, and 4C are diagrams for explaining an example of a moving object map whose behavior changes depending on the specifications of the radio wave sensor 2. Figure 4B is a diagram showing an example of a moving object map generated based on sensor data obtained by a radio wave sensor 2 with a small number of antennas detecting an object present in the vehicle cabin. Figure 4A is a diagram showing an example of a moving object map generated based on sensor data obtained by a radio wave sensor 2 with a larger number of antennas than that shown in Figure 4B detecting an object present in the vehicle cabin. Figure 4C is a diagram showing a moving object map in which the distribution of moving objects on the moving object map shown in Figure 4B is superimposed. Figures 4A, 4B, and 4C are all side views of the moving object map.

[0028] For example, suppose the radio wave sensor 2 has a small number of antennas to reduce costs. In this case, as shown in FIG. 4B , the number of biological responses (moving bodies) detectable in a single sensing operation is smaller than when a radio wave sensor 2 with a large number of antennas is used, as shown in FIG. 4A . Therefore, it is difficult for the physique determination device 1 to obtain a moving body map capable of determining the physique of an occupant present in the vehicle cabin through a single sensing operation using the radio wave sensor 2. Therefore, for example, as shown in FIG. 4C , if the radio wave sensor 2 continuously acquires sensor data at short intervals, the moving body map generation unit 123 in the physique determination device 1 may generate a moving body map by overlaying the biological responses obtained at each interval based on the moving body information extracted by the moving body analysis unit 122 based on the sensor data continuously acquired at short intervals. For example, as shown in FIG. 4C , the moving body map generation unit 123 generates a moving body map by overlaying the biological responses obtained at each interval to clarify the silhouette of the occupant. This allows the physique determination device 1 to obtain a moving body map capable of determining the physique of an occupant present in the vehicle cabin through a single sensing operation using the radio wave sensor 2.

[0029] Furthermore, since only portions of a moving body are depicted on the moving body map within a certain period of time, the plot results on the moving body map can vary significantly depending on the movement of the living body, even for living bodies of the same physique. Figures 5A and 5B are diagrams illustrating an example of a moving body map in which plot results vary depending on the movement of the living body. Figure 5A shows an example of a moving body map generated when an adult is seated properly in the left rear seat of a vehicle and is awake, moving their arms, legs, or head. Figure 5B shows an example of a moving body map generated when an adult is seated properly in the left rear seat of a vehicle and is asleep, moving their chest only due to breathing. The physiques of the adult plotted in the moving body map shown in Figure 5A and the physiques of the adult plotted in the moving body map shown in Figure 5B are assumed to be the same. In Embodiment 1, "the same physique" does not necessarily mean exactly the same physique, but also includes approximately the same physique. Both Figures 5A and 5B are side views of the moving body map.

[0030] For example, even if two living organisms in a vehicle cabin have the same physique, the distribution of point clouds will be different between a motion map ( FIG. 5A ) generated when the person is performing awake movements and a motion map ( FIG. 5B ) generated when the person is performing sleep movements. The motion map generated during awake movements as shown in FIG. 5A shows the movement of the living organism, moving the entire body from the head to the feet. On the other hand, the motion map generated during sleep movements as shown in FIG. 5B shows only breathing movement near the chest. As such, because moving parts appear as responses in the motion map, the motion map generated during sleep movements tends to have fewer point clouds and weaker reflection intensity than the motion map generated during awake movements.

[0031] In determining the performance of physique determination, information about the silhouette of the entire body or height, such as the height when standing upright or the sitting height when seated, is particularly important. Therefore, when the physique determination device 1 performs physique determination based on a dynamic body map generated based on sensor data acquired from the radio wave sensor 2, it is desirable to perform the determination when the living body is in a normal state and the entire body, particularly the head area, is moving appropriately. In the first embodiment, the normal state refers to the basic state of the living body at the time of physique determination, and refers to a state in which the living body's posture is not disturbed. For example, when the living body is seated, the normal state is a normal sitting state, and when the living body is standing, the normal state is a normal standing state. In the first embodiment, the normal standing state refers to a state in which the living body's posture is not disturbed, the living body is standing upright with its back straight, that is, the body is standing perpendicular to the ground. In the first embodiment, the state of "moderate movement" refers to a state in which the living body is not completely still, such as being rigid, nor is it moving unnaturally, but is moving within a predetermined range that is expected to occur naturally, such as naturally occurring swaying. The physique determination device 1 obtains important information for determining physique from the moving body map, such as the silhouette of the body when the entire body of the living body is moving appropriately, and height or sitting height when the area around the head of the living body is moving appropriately.

[0032] However, as mentioned above, even if a living body is not completely still, the number of points appearing on the dynamic body map may be small due to small movements or limited body parts moving (for example, during sleep). The physique determination device 1 determines the physique of a living body by taking into consideration the fact that the distribution of points appearing on the dynamic body map changes depending on the movement of the living body.

[0033] Returning to the description of the example configuration of the physique determination device 1 shown in Fig. 1, the movement estimation unit 13 estimates the movement of the living body based on the motion map generated by the data processing unit 12.

[0034] The estimation of the motion of the living body by the motion estimator 13 will be described in detail with an example.

[0035] The biological motion estimated by the motion estimation unit 13 is expressed, for example, by a motion type or a motion amount of the biological motion. The motion type is a classification of motions that affect physique determination or motions specific to a specific physique of an infant or the like. The motion type is determined in advance by an administrator or the like, and is stored in a buffer inside the motion estimation unit 13 or in a storage unit (not shown) provided in a location that can be referenced by the physique determination device 1. The motion amount is a numerical value or the like that simply indicates how much the biological motion is occurring.

[0036] A specific example will be given below to explain the type of movement or the amount of movement estimated as a living body movement by the movement estimation unit 13. First, a specific example will be given below to explain the type of movement.

[0037] <Movement Type Example (A): Moderate Swaying Movement> The motion type of the living body estimated by the motion estimation unit 13 includes, for example, a "moderate swaying movement." In the first embodiment, a "moderate swaying movement" is assumed to be a sway (movement) within a predetermined range that is assumed to occur naturally throughout the body. Complete stillness, such as stiffness, or unnaturally large sway (movement) of the entire body is not included in the "moderate swaying movement." Specific examples of a "moderate swaying movement" include, for example, movement when a person is awake and seated, relaxed without being extremely stiff, or swaying that occurs throughout the body due to swaying, such as when a vehicle is moving. When a living body is performing a "moderate swaying movement," the silhouette of the living body becomes clear on the dynamic body map. In other words, when a living body is performing a "moderate swaying movement," it can be said that the living body is in a desirable state for performing a physique determination.

[0038] <Movement Type Example (B): Thrashing Movement> The movement type of the living body estimated by the movement estimation unit 13 may include, for example, a "thrashing movement." In the first embodiment, the "thrashing movement" is assumed to be a movement of vigorously moving the whole body, which is often seen in immature infants compared to adults. Note that the movement of the whole body in the "thrashing movement" is assumed to be a movement larger than the predetermined range assumed for the "moderate rocking movement."

[0039] <Movement Type Example (C): Head Swing> The movement type of the living organism estimated by the movement estimation unit 13 may include, for example, a "head swing." In the first embodiment, the "head swing" is assumed to be, for example, a movement of moving the head. Note that the head movement in the "head swing" is assumed to be a movement larger than the predetermined range assumed for the "moderate shaking movement." Because the head moves due to the "head swing," the height of the body can be determined on the moving body map, which is one of the important pieces of information used to determine the physique of the living organism.

[0040] <Movement Type Example (D): Posture Change Movement> The movement type of the living body estimated by the movement estimation unit 13 may include, for example, a “posture change movement.” In the first embodiment, the “posture change movement” is assumed to be a movement in which the living body assumes a posture other than a normal state (here, a normal sitting state), such as a posture disorder such as lying down or lying face down.

[0041] <Movement Type Example (E): Moving Movement> The movement type of the living body estimated by the movement estimation unit 13 may include, for example, a “moving movement.” In the first embodiment, the “moving movement” is assumed to be, for example, a movement in which the position in the vehicle cabin moves by a predetermined distance or more.

[0042] <Example of Motion Type (F): Leg-Flapping Movement> The motion type of the living body estimated by the motion estimation unit 13 may include, for example, "leg-flapping movement." In the first embodiment, "leg-flapping movement" is assumed to be, for example, a movement of violently moving the legs. Note that the leg movement in "leg-flapping movement" is assumed to be a movement larger than the predetermined range assumed for "moderate swaying movement." Like "thrashing movement," "leg-flapping movement" is estimated to be more frequently seen in immature infants than in adults.

[0043] <Example of Motion Type (G): Belt Fastening Action> The motion type of the living body estimated by the motion estimation unit 13 may include, for example, a “belt fastening action.” In the first embodiment, the “belt fastening action” is assumed to be the action of fastening a seat belt.

[0044] Note that, here, "moderate shaking motion," "violent motion," "head shaking motion," "posture change motion," "movement motion," "foot flapping motion," and "belt fastening motion" are given as specific examples of motion types, but these are merely examples, and the motion types may include types other than those mentioned above. The definition of the motion types is appropriately determined in advance, for example, by an administrator or the like.

[0045] Next, the amount of movement will be described using a specific example. For example, the amount of movement is expressed in three levels: "large," "medium," and "small." The movement estimator 13 estimates the amount of movement of the living organism as the movement of the living organism in three levels: "large," "medium," and "small." For example, if the amount of movement estimated by the movement estimator 13 is "large," it is estimated that the living organism is performing a large movement, such as a violent movement or a posture change. For example, if the amount of movement estimated by the movement estimator 13 is "medium," it is estimated that the living organism is performing a moderate movement. As described above, in the first embodiment, a state of "moderate movement" does not mean a completely still state, such as being rigid, and not a state of unnaturally large movement, but a state of movement within a predetermined range that is expected as a naturally occurring movement, such as a naturally occurring swaying. For example, if the amount of movement estimated by the movement estimator 13 is "small," it is estimated that the living organism is performing a small movement, nearly still state, such as a sleeping movement. Note that the state close to stillness referred to here does not mean complete stillness (e.g., zero speed), but rather movement that is smaller than the specified range of movement assumed as movement in a "moderately moving" state.

[0046] Although the example in which the amount of movement estimated by the movement estimating unit 13 is expressed in three levels, "large," "medium," and "small," has been given here, this is merely an example. For example, the amount of movement estimated by the movement estimating unit 13 may be expressed in five levels, from "1" to "5," or may be expressed as a numerical value, such as from 0% to 100%.

[0047] The movement estimation unit 13 may, for example, estimate the movement type of the living organism, estimate the movement amount, or estimate a combination of the movement type and movement amount of the living organism as the movement of the living organism. Combining the movement type and movement amount of the living organism means, for example, estimating the degree of the movement classified by the movement type from the movement amount, and estimating the estimated degree as the movement of the living organism. As a specific example, the movement estimation unit 13 may estimate that the movement of the living organism is a large posture change movement (the movement type is "posture change movement" and the movement amount is "large").

[0048] A specific example will be given to explain a method for estimating a biological movement by the movement estimator 13. Note that, in the following, the method for estimating a biological movement by the movement estimator 13 will be explained on the assumption that the movement estimator 13 estimates a movement type or a movement amount as a biological movement.

[0049] <Movement Estimation Method Example (A): Estimation from the Number of Point Clouds> For example, the movement estimation unit 13 can estimate the movement of a living organism based on the number of point clouds on a motion map. On a motion map, moving locations appear as point clouds. For example, when estimating the amount of movement of a living organism, the movement estimation unit 13 estimates the amount of movement as "small" when the number of point clouds appearing on the motion map is small (for example, when the number of point clouds is equal to or less than a predetermined first threshold), and estimates the amount of movement as "large" when the number of point clouds appearing on the motion map is large (for example, when the number of point clouds is equal to or greater than a predetermined second threshold (where the first threshold is less than the second threshold)).

[0050] <Movement Estimation Method Example (B): Estimation from Reflection Intensity of Point Clouds> For example, the movement estimation unit 13 may estimate the movement of a living organism based on information about the reflection intensity assigned to the point clouds on the moving body map. Point clouds with low reflection intensity on the moving body map are estimated to indicate small movement, while point clouds with high reflection intensity are estimated to indicate large movement. For example, when estimating the amount of movement of a living organism, the movement estimation unit 13 estimates the amount of movement as "small" if the reflection intensity assigned to the point clouds appearing on the moving body map is small (e.g., if the reflection intensity is equal to or less than a predetermined third threshold), and estimates the amount of movement as "large" if the reflection intensity assigned to the point clouds appearing on the moving body map is large (e.g., if the reflection intensity is equal to or greater than a predetermined fourth threshold (where the third threshold is less than the fourth threshold)). However, the reflection intensity tends to decrease depending on the distance from the radio wave sensor 2 to the object. Therefore, for example, when estimating the movement of a living organism from the reflection intensities assigned to the point clouds on the moving body map, the movement estimating unit 13 may take this into consideration and correct the reflection intensities according to the distance to the living organism before estimating the movement of the living organism. For example, when the moving body map generating unit 123 generates the moving body map, the reflection intensities assigned to the point clouds may be corrected according to the distance to the living organism.

[0051] <Movement Estimation Method Example (C): Estimation from Point Cloud Velocity> For example, the movement estimation unit 13 may estimate the movement of a living organism based on velocity information assigned to the point cloud on the moving body map. It is estimated that the greater the movement of the living organism, the faster the velocity tends to be. For example, when estimating the movement amount of a living organism, the movement estimation unit 13 estimates the movement amount as "small" if the velocity assigned to the point cloud appearing on the moving body map is small (e.g., if the velocity is equal to or less than a predetermined fifth threshold), and estimates the movement amount as "large" if the velocity assigned to the point cloud appearing on the moving body map is large (e.g., if the velocity is equal to or greater than a predetermined sixth threshold (where the fifth threshold is less than the sixth threshold)).

[0052] <Movement Estimation Method Example (D): Estimation from Distribution Pattern of Moving Body Map> For example, the movement estimation unit 13 may estimate the movement of a living organism based on the distribution pattern of a point cloud, such as the position or overall shape of the point cloud on the moving body map. For example, when estimating the movement type of a living organism, the movement estimation unit 13 estimates the movement type by comparing the distribution pattern of the point cloud on the moving body map with preset conditions (hereinafter referred to as "motion type determination conditions").

[0053] The conditions for determining the type of movement include, for example, "Condition (1): When the ratio of the length in the sitting height direction to the length in the body thickness direction of the estimated head and body rectangle on the dynamic body map is approximately the same (when it is approximately square), it is estimated that there is no head swinging movement, and when the ratio of the length in the sitting height direction to the length in the body thickness direction is large (when it is a vertically long rectangle), it is estimated that there is head swinging movement." Note that the ratio of the length in the sitting height direction to the length in the body thickness direction that is deemed to be approximately the same is determined in advance. The movement estimation unit 13 estimates the presence or absence of a "head swinging movement" of the living body according to condition (1).

[0054] Here, the estimated head and torso rectangle under the above condition (1) will be described. Figures 6A, 6B, 6C, and 6D are diagrams for describing the estimated head and torso rectangle in embodiment 1. The moving object map shown using Figures 6A and 6B is a moving object map generated when an adult is seated in the left rear seat in the vehicle cabin, and the moving object map shown using Figures 6C and 6D is a moving object map generated when a child is seated in the left rear seat in the vehicle cabin. Figures 6A, 6B, 6C, and 6D are all side views of the moving object map. In embodiment 1, the "estimated head and torso rectangle" is a rectangle on the moving object map that, when viewed from the side, surrounds the point cloud located on the seat surface in the vehicle cabin and has its height direction parallel to the seat back. 6A, 6B, 6C, and 6D, the estimated head and body rectangles are indicated by 601 to 604, respectively. The estimated head and body rectangles are the sides of the area where the point clouds appearing there are estimated to be point clouds representing the head and body of a living body. Note that in the first embodiment, "parallel to the back of the seat" does not necessarily mean strictly parallel, but also includes approximately parallel.

[0055] The motion estimation unit 13, for example, sets an estimated head / torso rectangle in the dynamic body map and estimates the presence or absence of a "head-shaking motion" of the living body by comparing the set estimated head / torso rectangle with the above condition (1). Since the installation position of the radio wave sensor 2, the radio wave radiation range of the radio wave sensor 2, the position of the seat in the vehicle cabin, and the reclining angle of the seat are known in advance, the motion estimation unit 13 can determine which area in the dynamic body map is above the seat surface and which side is parallel to the seat back. The motion estimation unit 13 may, for example, acquire information about the seat position and reclining angle from a seat sensor (not shown) to set the estimated head / torso rectangle. For example, the motion estimation unit 13 sets the smallest rectangular parallelepiped or cube that surrounds the point cloud above the seat surface in the dynamic body map and is parallel to the seat back, and defines the side of the rectangular parallelepiped or cube as the estimated head / torso rectangle.

[0056] For example, if the subject is not performing a "head shaking" motion, head movement is unlikely to appear on the body motion map, and only chest movement may appear on the body motion map. In this case, the estimated head and torso rectangle on the body motion map will have a shape close to a square, with the ratio of the length in the sitting height direction to the length in the body thickness direction being approximately the same (see Figures 6A and 6C). On the other hand, if the subject is performing a "head shaking" motion, head movement will likely appear on the body motion map along with chest movement. In this case, the estimated head and torso rectangle on the body motion map will have a vertically long rectangle, with a large ratio of the length in the sitting height direction to the length in the body thickness direction (see Figures 6B and 6D).

[0057] The condition for determining the type of movement may include, for example, "Condition (2): if a number of points equal to or greater than a preset threshold (hereinafter referred to as "fluttering determination threshold") appears in the estimated foot area on the moving body map, it is estimated that there is a fluttering movement of the legs." The movement estimation unit 13 estimates the presence or absence of "fluttering movement of the legs" of the living body in accordance with condition (2).

[0058] Here, the estimated foot region under condition (2) will be described. FIG. 7 is a diagram illustrating the estimated foot region in the first embodiment. The dynamic object map shown in FIG. 7 is a dynamic object map generated when a child is seated in the left rear seat in the vehicle cabin. FIG. 7 is a side view of the dynamic object map. In the first embodiment, an area of ​​a predetermined size in front of the seat surface on the dynamic object map in the vehicle cabin is referred to as the "estimated foot region." The estimated foot region is, for example, a rectangular or cubic region with one side in the width direction of the vehicle the same length as the width of the seat surface, one side in the direction of travel of the vehicle a predetermined length toward the front of the seat surface and a length including a predetermined length from one end of the front side of the seat surface toward the direction of travel of the vehicle, and one side in the height direction of the vehicle a length from the footwell of the vehicle to approximately the center of the seat back. In FIG. 7, the estimated foot region is indicated by 701. The estimated foot region is a region in which a point cloud appearing therein is estimated to be a point cloud representing the feet of a living body.

[0059] The movement estimator 13 sets, for example, an estimated foot area in the moving body map, and estimates the presence or absence of "foot flapping" by comparing the set estimated foot area with the above condition (2). Note that the installation position of the radio wave sensor 2, the radiation range of the radio waves from the radio wave sensor 2, and the position of the seat in the vehicle cabin are known in advance, so the movement estimator 13 can determine which area in the moving body map will be the estimated foot area. The movement estimator 13 may set the estimated foot area by, for example, acquiring information about the position of the seat from a seat sensor (not shown).

[0060] For example, if a living body is "flapping its legs," it is estimated that a certain number of points or more will appear in the estimated foot area on the moving body map due to the flapping of the legs (see, for example, 701 in Figure 7).

[0061] <Movement Estimation Method Example (E): Estimation from Distribution Pattern of Dynamic Object Map and Vehicle Information> For example, the motion estimation unit 13 may estimate the motion of a living body based on the distribution pattern of the point cloud on the dynamic object map and vehicle information. The vehicle information includes, for example, information indicating whether a door is open or closed, the amount of seat belt withdrawal, and whether the seat belt is plugged into the buckle. The motion estimation unit 13 may acquire the vehicle information from a door sensor, a seat belt sensor, and a buckle sensor (not shown). For example, the motion type determination condition may include the following condition: "Condition (3): If the door is closed or the seat belt is withdrawn in a seat corresponding to an area where a point cloud appears on the dynamic object map, it is estimated that a belt fastening action has occurred. Also, if the seat belt is plugged into the buckle, it is estimated that a belt fastening action has occurred immediately before." The motion estimation unit 13 estimates whether a living body has performed a "belt fastening action" according to condition (3). Note that the "belt fastening action" also includes the "immediately preceding belt fastening action."

[0062] In the above example, the movement estimation unit 13 estimates the movement of the living organism based on the movement map and in accordance with preset conditions for determining the movement type. However, this is merely an example, and the movement estimation unit 13 may estimate the movement of the living organism using, for example, a trained model in machine learning (hereinafter referred to as a "machine learning model"). The machine learning model is generated in advance and stored in a location that the movement estimation unit 13 can refer to, such as an internal buffer of the movement estimation unit 13. The machine learning model is, for example, a model that receives the movement map as input and outputs information indicating the movement type. The movement estimation unit 13 estimates the movement of the living organism by inputting the movement map into the machine learning model and obtaining information regarding the movement type.

[0063] For example, based on a motion map obtained in advance in a vehicle cabin when a person is properly seated, thrashing about, shaking their head, losing their posture, moving, kicking their legs, and wearing a seatbelt, a machine learning model is generated that inputs the motion map and outputs information indicating the motion type, such as "moderate swaying motion," "thrashing motion," "head shaking," "posture change motion," "movement motion," "foot kicking," or "belt fastening motion." The motion estimation unit 13 inputs the motion map into the machine learning model and obtains information indicating the motion type, thereby estimating the "moderate swaying motion," "thrashing motion," "head shaking," "posture change motion," "movement motion," "foot kicking," or "belt fastening motion" performed by the living body.

[0064] For example, the machine learning model may be a model that receives a moving body map as input and outputs information indicating information for determining the distribution pattern on the moving body map (e.g., an estimated head / torso rectangle or an estimated foot region). In this case, the motion estimation unit 13, for example, inputs the moving body map to the machine learning model to obtain information for determining the distribution pattern on the moving body map, and estimates the motion of the living organism by comparing the obtained information for determining the distribution pattern on the moving body map with preset motion type determination conditions.

[0065] Furthermore, for example, the movement estimation unit 13 can estimate the movement amount of the living organism using a machine learning model. In this case, the machine learning model is, for example, a model that receives a motion map as input and outputs information indicating the movement amount. For example, the movement estimation unit 13 inputs the motion map into the machine learning model and obtains information indicating the movement amount, thereby estimating the movement amount of the living organism.

[0066] The above-described method of estimating the movement of a living body by the movement estimator 13 is one example. The movement estimator 13 may estimate the movement of a living body by, for example, a method that combines the above-described movement estimation methods, or may estimate the movement of a living body by using various other known methods. The movement estimator 13 may estimate the movement of a living body based on a motion map.

[0067] In the above-described method for estimating the movement of a living body by the motion estimator 13, it is assumed that the motion estimator 13 estimates the movement of the living body based on one motion map. However, this is merely an example, and the motion estimator 13 may estimate the movement of the living body based on, for example, multiple time-series motion maps generated within a set period in the past. Here, FIG. 8 is a diagram for explaining the method for estimating the movement of a living body by the motion estimator 13 based on multiple time-series motion maps in the first embodiment. FIG. 8 is a top view of the motion map. The motion map shown in FIG. 8 includes five time-series motion maps, from a state in which the occupant is outside the vehicle (see "Before Getting In" in FIG. 8 ), to a state in which the occupant gets into the left rear seat (see "While Getting In" in FIG. 8 ), moves around the rear seat (see "Moving in Seat" in FIG. 8 ), sits in the middle rear seat (see "Immediately After Seating" in FIG. 8 ), and several seconds after sitting (see "Several Seconds After Seating" in FIG. 8 ).

[0068] Even with each motion map alone, it is possible to estimate the type of movement by, for example, using the distribution pattern of the motion map. For example, in the case of a motion map showing a seat movement (see, for example, "Seat Movement" in Figure 8), responses are observed across positions corresponding to multiple seats on the motion map, in other words, a point cloud appears. Therefore, the motion estimation unit 13 can estimate that the living body is "moving" based on responses observed across positions corresponding to multiple seats on the motion map. However, in reality, the living body may not be moving, but may be sitting across multiple seats. Thus, for example, there are some living body states in which the motion estimation unit 13 may erroneously estimate the motion of the living body when attempting to estimate the motion of the living body based on a single motion map. For example, the motion estimation unit 13 can reduce erroneous estimation of the motion of the living body by estimating the motion of the living body based on multiple time-series motion maps generated within a set period in the past.

[0069] For example, suppose the motion estimation unit 13 estimates the motion of a living body from five time-series motion maps as shown in FIG. 8 . In this case, the point cloud that appeared outside the vehicle on the motion map "before getting in" moves to the interior of the vehicle on the next motion map "during getting in." The point cloud that appeared on the next motion map "while moving from seat to seat" moves to a position corresponding to the middle rear seat on the next motion map "immediately after sitting down." Furthermore, on the next motion map "several seconds after sitting down," the point cloud also appears at a position corresponding to the middle rear seat. Based on these motion maps, the motion estimation unit 13 estimates the motion of a living body that includes a "getting in" motion, a "moving" motion, and then a "sitting down" motion (more specifically, a sitting down motion to sit down in the desired seat (the middle rear seat)). The motion estimation unit 13 can estimate the type of motion, such as a "getting in" or a "sitting down" motion, from the change in the position where the point cloud appears on the time-series motion map. Note that which ranges on the motion map correspond to which seats is known in advance based on the installation position of the radio wave sensor 2, the radio wave radiation range, etc., so the motion estimator 13 can estimate, for example, which seat the "seating motion" will be in. In this way, the motion estimator 13 can also estimate the progression of the motion of a living body based on the time-series motion map. In the above example, the motion estimator 13 can estimate a series of motions of a living body, such as an occupant outside the vehicle, getting into the left rear seat, moving around the rear seats, sitting in the middle rear seat, and then sitting down, based on the five time-series motion maps.

[0070] In the above example, the motion estimation unit 13 estimates the motion type of the living organism based on a plurality of time-series motion maps, but the motion estimation unit 13 can also estimate the amount of motion of the living organism based on a plurality of time-series motion maps. For example, if there is almost no change in the distance assigned to the point clouds appearing in the plurality of time-series motion maps (for example, the amount of change in distance is within a predetermined threshold) and the reflection intensity assigned to the point clouds is gradually decreasing, the motion estimation unit 13 can estimate that the amount of motion of the living organism is gradually decreasing.

[0071] For example, by estimating the motion of a living body based on multiple time-series motion maps, the motion estimation unit 13 can estimate the motion of a living body over a longer period of time or reduce erroneous motion estimation. As a result, the physique determination device 1 can further improve the accuracy of determining the physique of a living body.

[0072] When the movement estimation unit 13 estimates the movement of the living body, it outputs movement estimation information indicating the estimated movement of the living body to the physique determination unit 14. The movement estimation information includes, for example, a classification capable of identifying the movement of the living body (e.g., 01: moderate shaking movement, 02: thrashing movement, 03: head shaking movement, 04: posture change movement, 05: moving movement, 06: kicking movement, 07: belt wearing movement, 11: large amount of movement, 12: medium amount of movement, 13: small amount of movement, or 21: large posture change). The movement estimation information may be associated with information indicating the living body whose movement is estimated. The information indicating the living body whose movement is estimated is, for example, information capable of identifying the seat in which the living body is sitting. The movement estimation information may also include a moving body map.

[0073] Returning to the description of the exemplary configuration of the physique determination device 1 shown in Fig. 1, the physique determination unit 14 determines the physique of the living body based on the dynamic body map generated by the data processing unit 12 and the motion estimation information output from the motion estimation unit 13. The physique determination unit 14 determines the physique of the living body using a binary value, such as whether it is an adult or a child (e.g., 1: adult, 2: child). Note that this is merely an example, and the physique determination unit 14 may determine the physique of the living body using a more detailed classification, such as equivalent to a large adult, equivalent to a standard adult, equivalent to a small adult, or equivalent to an N-year-old child (N is an integer).

[0074] Here, an example of a method for determining the physique of a living organism by the physique determination unit 14 based on the dynamic body map and the motion estimation information will be described. The physique determination unit 14 determines the physique of the living organism based on the dynamic body map. In doing so, the physique determination unit 14 utilizes the motion estimation information. Examples of methods by which the physique determination unit 14 utilizes the motion estimation information in determining the physique of the living organism based on the dynamic body map include a method of utilizing the motion estimation information to determine whether or not to perform a determination of the physique of the living organism based on the dynamic body map (hereinafter referred to as "Application Method (A)"), and a method of utilizing the motion estimation information to switch the physique determination method of the living organism based on the dynamic body map or to change the conditions of the physique determination method (hereinafter referred to as "Application Method (B)"). Specific examples of Application Method (A) and Application Method (B) will be described. First, a specific example of Application Method (A) will be described.

[0075] <Application Method (A)> For example, the physique determination unit 14 first determines, based on the motion estimation information, whether the estimated motion of the living body is a preset motion for which physique determination is to be suspended (hereinafter referred to as a "physique determination suspended motion"). If the physique determination unit 14 determines that the estimated motion of the living body is a physique determination suspended motion, the physique determination unit 14 does not perform a physique determination of the living body. The physique determination suspended motion is set in advance by an administrator or the like, and information indicating the set physique determination suspended motion is stored in a location accessible by the physique determination unit 14, such as an internal buffer of the physique determination unit 14. The physique determination suspended motion may be, for example, a "thrashing motion," a "posture change motion," a "movement motion," a "motion amount equal to or greater than a preset threshold (hereinafter referred to as a "motion amount determination threshold")," or a "motion amount equal to or less than the motion amount determination threshold." Note that this is merely an example, and the physique determination suspended motion may be set in advance to a motion that is estimated to result in an erroneous determination if a physique is determined based on a motion map generated while the living body is performing such a motion.

[0076] For example, when a living body is "moving wildly," the number of point clouds on the dynamic body map generated in that state, or the range in which the point clouds appear, will be greater than when the living body is relaxed and sitting properly, and it is estimated that the silhouette of the area consisting of the point clouds on the dynamic body map will be different from the silhouette that would originally appear (for example, on a dynamic body map generated when the living body is sitting properly). It is estimated that determining the body size of a living body using such a dynamic body map is difficult, and that the body size of the living body will be misjudged. The silhouette of the area consisting of the point clouds that appears on the dynamic body map is the silhouette of the living body. Details of body size determination using a dynamic body map will be described later with an example.

[0077] Furthermore, for example, when a living organism is engaged in "postural fluctuation" or "large amount of movement," the silhouette of the area consisting of point clouds on the dynamic body map is estimated to be different from the silhouette that would normally appear, just as when the living organism is engaged in "violent movement." Furthermore, it is desirable to perform a physical assessment of a living organism when the living organism is relatively still, in other words, when the living organism is moving moderately. If the living organism's movement is too small or too large, determining the living organism's physical size based on a dynamic body map generated in such a state may result in an erroneous assessment. For example, when the amount of movement is "small," the number of point clouds appearing on the dynamic body map decreases, making it difficult to distinguish the living organism's silhouette from the silhouette of the point clouds appearing on the dynamic body map. Conversely, for example, when a living organism is engaged in "moving movement," determining the living organism's physical size based on a dynamic body map generated in such a state may result in an erroneous assessment.

[0078] On the other hand, for example, a "moderate shaking motion" or a "medium amount of motion" is estimated to be unlikely to be erroneously determined when a physique is determined based on a motion map generated while the subject is performing that motion, in other words, to be correctly determined. For example, a "head shaking motion" involves movement of the head, but not significant movement of the entire body. When only the head moves, it can be said that this makes it easier to determine the height of the subject on the motion map. In determining a physique, information on the height of the body (including sitting height) is one of the important pieces of information for making a correct determination. Therefore, for example, a "head shaking motion" is estimated to be unlikely to be erroneously determined when a physique is determined based on a motion map generated while the subject is performing that motion. Such motions, which are estimated to be unlikely to be erroneously determined when a physique is determined based on a motion map generated while the subject is performing that motion, are not included in the physique determination pending motions. Furthermore, when a person is seated in a vehicle, they typically fasten a seat belt. Although there is arm movement before and after fastening the seat belt, the movement of other body parts is considered to be similar to that of a normally seated person. Therefore, for example, "belt fastening" is included in the actions that are estimated to be unlikely to be erroneously determined when the physique of a living body is determined based on a dynamic body map generated when the living body is performing that action. It can be said that it is effective to determine the physique of a living body based on a dynamic body map generated when the living body is estimated to be performing the "belt fastening" action.

[0079] When the physique determination unit 14 determines, based on the movement estimation information, that the estimated movement of the living body is not a physique determination pending movement, it performs a physique determination of the living body based on the dynamic body map. An example of a method for determining the physique of a living body based on the dynamic body map by the physique determination unit 14 will be described below. In the example given below, it is assumed that the physique determination unit 14 determines the physique of the living body as either an "adult" or an "infant."

[0080] <Physique Determination Method (A): Determination from Point Cloud Near the Legs> For example, the physique determination unit 14 determines the physique of a living organism based on whether a range on the dynamic body map that is pre-set as a range corresponding to the area around the legs of the seat (hereinafter referred to as the "leg range") contains a number of point clouds equal to or greater than a pre-set threshold (hereinafter referred to as the "leg point cloud determination threshold"). The leg range is pre-set by an administrator or the like and stored in an internal buffer or the like of the physique determination unit 14. The administrator or the like sets the leg range, for example, based on the area around the legs of the seat in the target space. Since the relationship between the position in the target space and the position of the point cloud on the dynamic body map is known in advance, the administrator or the like can identify the leg range on the dynamic body map. This physique determination method (A) is effective when the living organism is seated, for example, inside a vehicle. When a living organism is in a seated position, for example, a living organism with a large build will have its feet on the ground, while a living organism with a small build will have its feet off the ground, so the point cloud in the leg range on the dynamic body map is likely to show a difference between an "adult" and an "infant." For example, if a point cloud with a number equal to or greater than the leg point cloud determination threshold appears in the leg range on the dynamic body map, the physique determination unit 14 determines that the living organism is an "adult," and if a point cloud with a number equal to or greater than the leg point cloud determination threshold does not appear in the leg range, the living organism is determined to be an "infant."

[0081] <Physique Determination Method (B): Determination from Point Cloud Near the Head> For example, the physique determination unit 14 may determine whether a living organism is an “adult” or an “infant” based on whether a point cloud appearing in a range on the dynamic body map that is pre-defined as a range corresponding to the range where the living organism's head is presumed to be located (hereinafter referred to as the “head range”) is above a pre-defined height (hereinafter referred to as the “head point cloud determination threshold”). The head range is pre-defined by an administrator or the like and stored in an internal buffer or the like of the physique determination unit 14. The point cloud appearing in the head range is presumed to correspond to the head, and the height information of the living organism can be determined based on the position of the point cloud, making it possible to determine the living organism's physique. For example, if the point cloud appearing in the head range on the dynamic body map is above the head point cloud determination threshold, the physique determination unit 14 determines that the living organism is an “adult,” and if the point cloud appearing in the head range is below the head point cloud determination threshold, the physique determination unit 14 determines that the living organism is an “infant.”

[0082] In addition, when there are multiple point clouds appearing in the head area, the physique determination unit 14 may determine the point cloud corresponding to the head using an appropriate method. For example, the physique determination unit 14 may determine whether the living body is an "adult" or "infant" based on whether the position of the point cloud is above a head point cloud determination threshold, or may determine whether the living body is an "adult" or "infant" based on whether the position of the point cloud is above a head point cloud determination threshold. Furthermore, for example, when there are multiple point clouds appearing in the head area, the physique determination unit 14 may determine the point cloud corresponding to the head based on the height of the center of gravity of the multiple point clouds. The physique determination unit 14 may calculate the center of gravity of the point cloud from the average value of the coordinates of the multiple point clouds, or may calculate the center of gravity of the point cloud taking into account the reflection intensity assigned to the point cloud in addition to the positions of the multiple point clouds. It is assumed that a higher reflection intensity has a stronger influence and that the center of gravity is closer to the higher reflection intensity. For example, instead of setting the head range in advance, the physique determination unit 14 may determine the head range on the dynamic body map using various other known methods.

[0083] <Physique Determination Method (C): Determination from Point Cloud Near the Chest> For example, the physique determination unit 14 may determine whether a living body is an "adult" or "infant" based on whether a point cloud appearing in a range on the dynamic body map that is pre-defined as a range corresponding to the range where the living body's chest is presumed to be located (hereinafter referred to as the "chest range") is above a pre-defined height (hereinafter referred to as the "chest point cloud determination threshold"). The chest range is pre-defined by an administrator or the like and stored in an internal buffer or the like of the physique determination unit 14. For example, even if the living body is sleeping or the amount of movement is "low," movement near the chest associated with the living body's breathing will appear on the dynamic body map. The point cloud appearing in the chest range is presumed to correspond to the chest, and chest height information can be determined based on the position of the point cloud, making it possible to determine the living body's physique based on this information. For example, the physique determination unit 14 determines that the living body is an "adult" if the point cloud appearing in the chest area on the moving body map is above the threshold for chest point cloud determination, and determines that the living body is an "infant" if the point cloud appearing in the area corresponding to the chest area is below the threshold for chest point cloud determination.

[0084] In addition, when there are multiple point clouds appearing in the chest area, the physique determination unit 14 may determine the point cloud corresponding to the chest using an appropriate method. For example, the physique determination unit 14 may determine the point cloud corresponding to the chest from the multiple point clouds using a method similar to the method for determining the point cloud corresponding to the head from the multiple point clouds described above in <Physique Determination Method (B): Determination from Point Cloud Near the Head>. For example, instead of setting the chest area in advance, the physique determination unit 14 may determine the chest area on the dynamic body map using various other known methods.

[0085] For example, by viewing the dynamic body map over time, it is possible to see the movement of the chest area associated with breathing. In other words, the respiratory rate can be estimated from the periodic movement of at least one of the position, velocity, or reflection intensity information assigned to the point clouds appearing in the time-series dynamic body map. The physique determination unit 14 may estimate the respiratory rate based on the position, velocity, or reflection intensity information assigned to the point clouds appearing in the time-series dynamic body map, and determine whether the living body is an "adult" or an "infant" based on whether the respiratory rate is equal to or greater than a preset threshold (hereinafter referred to as the "breathing determination threshold"). The breathing determination threshold may be set to, for example, a commonly known average breathing rate for adults or a value that assumes a buffered value for the average. If the estimated breathing rate is equal to or greater than the breathing determination threshold, the physique determination unit 14 determines the living body to be an "infant," whereas if the estimated breathing rate is less than the breathing determination threshold, the living body is determined to be an "adult." In general, the breathing rate of an infant is faster than that of an adult.

[0086] <Physique Determination Method (D): Determination from the Size of the Area Consisting of Point Clouds> For example, the physique determination unit 14 may determine whether a living organism is an “adult” or an “infant” based on the size of the area composed of point clouds appearing on the dynamic body map. An example of the size of the area composed of point clouds appearing on the dynamic body map is the width of the area composed of point clouds appearing on the dynamic body map. For example, if the width of the area composed of point clouds appearing on the dynamic body map is equal to or greater than a predetermined threshold (hereinafter referred to as the “width determination threshold”), the physique determination unit 14 determines the living organism to be an “adult.” If the width of the area composed of point clouds appearing on the dynamic body map is less than the width determination threshold, the physique determination unit 14 determines the living organism to be an “infant.” The width determination threshold is set in advance by an administrator or the like and stored in a buffer or the like internal to the physique determination unit 14. The width determination threshold is set, for example, to the average width (e.g., shoulder width) of a typical adult, or a value obtained by adding a buffer to the average width. The physique determination method (D) is effective in a target space such as a vehicle interior, where the direction in which a living body faces when in a normal seated position is uniquely determined.

[0087] Other examples of the size of the area formed by the point cloud appearing in the dynamic body map include the vertical width, volume, or area of ​​the area formed by the point cloud appearing in the dynamic body map. For example, the physique determination unit 14 may determine that the living body is an "adult" if the vertical width of the area formed by the point cloud appearing in the range in the dynamic body map where upper body reactions are expected to occur (hereinafter referred to as the "estimated upper body range") is equal to or greater than a predetermined threshold (hereinafter referred to as the "vertical width determination threshold"). Alternatively, the physique determination unit 14 may determine that the living body is an "infant" if the vertical width is less than the vertical width determination threshold. The estimated upper body range and the vertical width determination threshold are set in advance by an administrator or the like and stored in a buffer or the like within the physique determination unit 14. The vertical width determination threshold is set, for example, to the average vertical width (e.g., sitting height) of a typical adult's body, or a value obtained by adding a buffer to the average.

[0088] Furthermore, for example, the physique determination unit 14 may determine that the living body is an "adult" if the volume of an area consisting of a cloud of points appearing on the dynamic body map is equal to or greater than a preset threshold (hereinafter referred to as the "volume determination threshold"), and may determine that the living body is an "infant" if the volume is less than the volume determination threshold. The volume determination threshold is set in advance by an administrator or the like and stored in a buffer or the like inside the physique determination unit 14. The volume determination threshold is set to, for example, the average value of the volume of the upper body of a typical adult, or a value obtained by adding a buffer to the average value.

[0089] Furthermore, for example, the physique determination unit 14 may compress a three-dimensional dynamic body map into two dimensions to generate a view of the dynamic body map from the side or front (hereinafter referred to as a "compressed two-dimensional map"), and determine whether the living organism is an "adult" or an "infant" based on the area of ​​an area formed by a point cloud in the generated compressed two-dimensional map (hereinafter referred to as a "compressed point cloud area"). For example, the physique determination unit 14 determines that the living organism is an "adult" if the area of ​​the compressed point cloud area is equal to or greater than a predetermined threshold (hereinafter referred to as an "area determination threshold"), and determines that the living organism is an "infant" if the area of ​​the compressed point cloud area is less than the area determination threshold. The area determination threshold is set in advance by an administrator or the like and stored in a buffer or the like internal to the physique determination unit 14.

[0090] The volume of a region consisting of a point cloud appearing in a moving body map or the area of ​​a compressed point cloud region may be the number of points included in the region, or the volume or area of ​​an ellipsoid or hexahedron (an ellipse or rectangle in the case of a compressed point cloud region) that defines a range within the region in which the point cloud exists. The volume of a region consisting of a point cloud appearing in a moving body map or the area of ​​a compressed point cloud region may also be the volume of a region consisting of a point cloud appearing in the estimated upper body range in the moving body map, or the area of ​​a compressed point cloud region in a compressed 2D diagram generated based on the region consisting of a point cloud appearing in the estimated upper body range in the moving body map.

[0091] Furthermore, if a person's physique is large, the reflection cross-sectional area of ​​the radio waves emitted by the radio wave sensor 2 is larger, resulting in stronger reflection of the radio waves, and the reflection intensity assigned to the corresponding point clouds also tends to be higher. Therefore, when determining the size of the area consisting of point clouds appearing in the dynamic body map, the physique determination unit 14 may determine the size based on the reflection intensity assigned to the point clouds rather than the number or position of the point clouds, or may combine both. For example, if the average or maximum value of the reflection intensity assigned to the point clouds in the estimated upper body area in the dynamic body map is equal to or greater than a predetermined threshold (hereinafter referred to as the "reflection intensity determination threshold"), the physique determination unit 14 may determine the living body to be an "adult," and if the average or maximum value of the reflection intensity is less than the reflection intensity determination threshold, the living body may be determined to be an "infant." For example, when the physique determination unit 14 determines the size of an area consisting of point clouds appearing on a moving body map based on the number, position, or combination of reflection intensities of the point clouds, it may determine whether the living body is an "adult" or an "infant" by comparing the sum of the reflection intensities assigned to each point cloud with a preset threshold value.

[0092] <Physique Determination Method (E): Determination from the Offset Amount of the Point Cloud from the Seat Back or Seat Surface> For example, the physique determination unit 14 may determine whether a living body is an “adult” or an “infant” based on the offset amount of the point cloud in the dynamic body map from a region deemed to be the seat back in the depth direction of the vehicle cabin. For example, the physique determination unit 14 may determine that the living body is an “infant” if the offset amount of the point cloud appearing in the dynamic body map from a region deemed to be the seat back in the depth direction of the vehicle cabin is equal to or greater than a predetermined threshold (hereinafter referred to as the “first offset amount determination threshold”), and may determine that the living body is an “adult” if the offset amount of the point cloud in the dynamic body map from a region deemed to be the seat back in the depth direction of the vehicle cabin is less than the first offset amount determination threshold. Furthermore, for example, the physique determination unit 14 may determine that the living body is an “adult” or an “infant” based on the offset amount of the point cloud in the dynamic body map from a region deemed to be the seat surface of the seat in the vertical direction of the vehicle cabin. For example, the physique determination unit 14 determines that the living body is a "child" if the offset amount of the point cloud appearing on the moving body map from an area considered to be the seat surface in the vertical direction inside the vehicle cabin is equal to or greater than a preset threshold (hereinafter referred to as the "second offset amount determination threshold"), and determines that the living body is an "adult" if the offset amount of the point cloud appearing on the moving body map from an area considered to be the seat surface in the vertical direction inside the vehicle cabin is less than the second offset amount determination threshold. The first offset amount determination threshold and the second offset amount determination threshold are set in advance by an administrator or the like and stored in a buffer or the like inside the physique determination unit 14. The first offset amount determination threshold is set to, for example, a value assuming a typical value for the thickness of a child car seat on the backrest side of the seat when the child car seat is placed on the seat, or a value obtained by adding a buffer to that value. The second offset amount determination threshold is set to, for example, a value that assumes a typical value for the thickness of the child car seat on the seat surface side when the child car seat is placed on the seat, or a value that adds a buffer to that value. The first offset amount determination threshold and the second offset amount determination threshold may be the same value or different values. In some cases, a child may be seated in a child car seat inside the vehicle.In this case, it is estimated that the entire point cloud is offset forward or upward in the vehicle interior from the area of ​​the moving object map corresponding to the backrest or seat surface by the thickness of the child car seat.

[0093] <Physique Determination Method (F): Determination from Distribution Pattern of Point Cloud in Dynamic Body Map> For example, the physique determination unit 14 may determine the physique of a living organism using a machine learning model (hereinafter referred to as a "physique determination model"). The physique determination model is generated in advance and stored in a buffer or the like inside the physique determination unit 14. The physique determination model is, for example, a machine learning model that receives the dynamic body map as input and outputs information indicating the physique of the living organism. The physique determination model may be, for example, a machine learning model that receives the dynamic body map as input and outputs information indicating the physique of each living organism present in an area corresponding to each seat in the vehicle interior. The physique determination unit 14 inputs the dynamic body map into the machine learning model and obtains information indicating the physique of the living organism, thereby determining the physique of the living organism. For example, when a living organism is seated in a rear-facing child seat, the silhouette of the point cloud corresponding to the living organism that appears on the dynamic body map will be a distinctive silhouette, unlike when the living organism is seated facing forward. Note that a living body seated facing forward is assumed to be seated facing forward, regardless of whether a child car seat is being used, such as in a forward-facing child car seat or directly on the seat rather than in a child car seat. Here, FIG. 9 shows an example of a moving body map generated by the moving body map generator 123 in embodiment 1, specifically, a moving body map generated when an infant is seated in a rearward-facing child car seat installed in the left rear seat of the vehicle interior. FIG. 9 shows an example of a side view of a moving body map. As shown in FIG. 9 , when a living body (infant) is seated in a rearward-facing child car seat, the moving body map, when viewed from the side, has a silhouette symmetrical with the angle of the seat back. Specifically, when the moving body map is viewed from the side, the position of the part of the point cloud appearing on the moving body map that is estimated to be the highest head is far away from the seat back, and the position of the part that is estimated to be the lower body or feet is close to the seat back. For example, by having the model for determining physique learn a moving body map in which such a characteristic silhouette point cloud appears, the physique determination unit 14 can accurately determine the physique of a living organism.

[0094] <Physique Determination Method (G): Determination from the Height of Point Cloud on Dynamic Body Map> For example, the physique determination unit 14 may determine whether a living organism is an “adult” or an “infant” based on whether the height of the point cloud appearing on the dynamic body map is above a predetermined threshold (hereinafter referred to as the “height determination threshold”). The height determination threshold is appropriately set by an administrator or the like and stored in an internal buffer or the like of the physique determination unit 14. For example, if the height of the point cloud appearing on the dynamic body map is above the height determination threshold, the physique determination unit 14 determines the living organism to be an “adult,” and if the height of the point cloud appearing on the dynamic body map is equal to or less than the height determination threshold, the physique determination unit 14 determines the living organism to be an “infant.” Note that, for example, the physique determination unit 14 uses the highest point cloud among the point clouds appearing on the dynamic body map as the point cloud to be compared with the height determination threshold.

[0095] The above-described "Physique Determination Method (A)" to "Physique Determination Method (G)" are merely examples, and the physique determination unit 14 may determine the physique of a living body using other methods based on the dynamic body map. For example, the physique determination unit 14 may determine the physique of a living body by combining the above-described "Physique Determination Method (A)" to "Physique Determination Method (G)."

[0096] Next, the application method (B) will be explained using a specific example.

[0097] <Application Method (B)> For example, the physique determination unit 14 may switch the physique determination method based on the estimated movement information. To give a specific example, if a "moderate shaking movement" or a "shaking movement" is estimated as the movement of a living organism, the physique determination unit 14 determines whether the living organism is an "adult" or an "infant" based on whether the point cloud appearing in the area corresponding to the head region on the dynamic body map is above a head point cloud determination threshold (see the above-mentioned <Physique Determination Method (B)>). If a "thrashing movement" is estimated as the movement of the living organism, the physique determination unit 14 switches the physique determination method to determine whether the living organism is an "adult" or an "infant" based on the distribution pattern of the point cloud on the dynamic body map (see the above-mentioned <Physique Determination Method (F)>). For example, the method of determining whether the living organism is an "adult" or an "infant" based on whether the point cloud appearing in the area corresponding to the head region on the dynamic body map is above a head point cloud determination threshold requires limited point cloud information for physique determination, and therefore has a low processing load for physique determination. On the other hand, a method for determining whether a living organism is an "adult" or an "infant" based on the distribution pattern of the point cloud in the dynamic body map requires a large amount of point cloud information for physique determination, resulting in a large processing load for physique determination. For example, if it is estimated that the living organism is "moving wildly," it is estimated that the shape of the point cloud in the dynamic body map is complex, and a physique determination method based on the distribution pattern of the point cloud in the dynamic body map is preferable, even if it involves a large processing load. The physique determination unit 14 can determine the physique of the living organism overall using a method with as little processing load as possible by changing the physique determination method depending on the estimated movement of the living organism.

[0098] For example, the physique determination unit 14 may change the conditions in the physique determination method based on the estimated movement information. The conditions in the physique determination method are, for example, thresholds. As a specific example, the physique determination unit 14 may use a method for determining whether a living body is an “adult” or an “infant” based on whether the point cloud appearing on the dynamic body map is above a height determination threshold (see the above-mentioned “Physique Determination Method (G)”). In this case, for example, if the physique determination unit 14 estimates that the living body is performing a “head shaking movement,” it may change the height determination threshold to a higher value assuming the head position (see the above-mentioned “Head Point Cloud Determination Threshold”), and if the living body is estimated to be performing a “sleeping movement,” it may change the height determination threshold to a lower value assuming the chest position (see the above-mentioned “Chest Point Cloud Determination Threshold”). The movement estimation unit 13 may also estimate the movement of the living body, more specifically, “sleeping movement” as a movement type. For example, when a predetermined number or less of points appear only in the chest area in the motion map, the motion estimation unit 13 estimates that the living body is making a "sleeping motion." For example, the motion estimation unit 13 may estimate that the living body is making a "sleeping motion" using a machine learning model.

[0099] Furthermore, for example, when it is estimated that a living body is "flapping its legs," the physique determination unit 14 may change the threshold value used in the physique determination method from the preset initial threshold value so that the physique of the living body is more likely to be determined to be that of a child. For example, in a vehicle, an adult's feet touch the floor, whereas a child seated in a child car seat has their feet in the air due to the influence of the seat surface and the thickness of the seat side of the child car seat. When it is estimated that a living body is "flapping its legs," it can be said that a child with their feet in the air is likely to be flapping their legs. Therefore, the physique determination unit 14 changes the threshold value used in the physique determination method so that the physique of the living body is more likely to be determined to be that of a child.

[0100] The above describes an example in which the physique determination unit 14 changes the threshold value as a condition in the physique determination method based on the estimated movement information, but this is merely an example. For example, the physique determination unit 14 may change the logic for physique determination from a preset initial logic as a condition in the employed physique determination method based on the estimated movement information. Note that how the logic for physique determination is changed depending on the type of movement estimated is set in advance by an administrator or the like.

[0101] Of course, the physique determination unit 14 may determine not to change the conditions in the physique determination method based on the estimated movement information, and determine the physique of the living body using a physique determination method that is a preset initial state. For example, if the amount of movement is estimated to be "medium," the physique determination unit 14 determines not to change the conditions in the physique determination method.

[0102] The physique determination unit 14 may combine the above-mentioned <Application Method (A)> and <Application Method (B)>.

[0103] The operation of the physique determination device 1 according to the first embodiment will be described. FIG. 10 is a flowchart for describing the operation of the physique determination device 1 according to the first embodiment. For example, when the physique determination device 1 detects that a vehicle door has been closed, the physique determination device 1 starts the operation shown in the flowchart of FIG. 10 . For example, a control unit (not shown) of the physique determination device 1 may acquire information about the opening and closing of the door from a door sensor (not shown) and detect that the vehicle door has been closed. When the control unit detects that the vehicle door has been closed, it outputs an instruction to start operation to each unit provided in the physique determination device 1. For example, the physique determination device 1 may end the operation after performing the operation shown in the flowchart of FIG. 10 once, or may repeatedly repeat the operation shown in the flowchart of FIG. 10 until the vehicle is powered off, or may end the operation shown in the flowchart of FIG. 10 when a preset time has elapsed.

[0104] The sensor data acquiring unit 11 acquires sensor data from the radio wave sensor 2 (step ST1). The sensor data acquiring unit 11 outputs the acquired sensor data to the data processing unit 12.

[0105] The data processing unit 12 generates a motion map (step ST2) by performing signal processing on the sensor data for the predetermined time period output from the sensor data acquisition unit 11 in step ST1. The data processing unit 12 outputs the generated motion map to the motion estimation unit 13 and the physique determination unit 14.

[0106] The motion estimation unit 13 estimates the motion of the living body based on the motion map generated by the data processing unit 12 in step ST2 (step ST3). After estimating the motion of the living body, the motion estimation unit 13 outputs the motion estimation information to the physique determination unit 14.

[0107] The physique determination unit 14 determines the physique of the living body based on the motion map generated by the data processing unit 12 and the motion estimation information output from the motion estimation unit 13 (step ST4).

[0108] In this way, the physique determination device 1 estimates the movement of a living organism based on a dynamic body map generated by signal processing of radio waves emitted by the radio wave sensor 2 toward a target space where a living organism may exist and reflected by objects in the target space. The physique determination device 1 then determines the physique of the living organism based on the dynamic body map and movement estimation information indicating the estimated movement of the living organism. Therefore, the physique determination device 1 can prevent erroneous determination of the physique by taking the movement of the living organism into consideration. The conventional technology described above does not take into account the fact that the distribution trend of the point clouds appearing in the dynamic body map can change significantly depending on the movement of the living organism. Therefore, depending on the movement of the living organism, the fluctuation component that can be estimated as chest pulsation contained in the waveform as a trajectory generated from the vehicle interior detection map cannot be properly obtained, which may result in an erroneous determination of the living organism's physique. Furthermore, a possible method of determining the physique of a living organism based on a dynamic body map involves estimating the silhouette of the living organism from the dynamic body map and determining the physique of the living organism from the estimated silhouette. However, depending on the movement of the living organism, the silhouette appearing in the dynamic body map can change significantly, which may result in an erroneous determination of the living organism's physique. In contrast, as described above, the physique determination device 1 according to the first embodiment estimates the movement of a living organism based on a dynamic body map, and determines the physique of the living organism based on the dynamic body map and the movement estimation information indicating the estimated movement of the living organism. Therefore, the physique determination device 1 can prevent erroneous determination of the physique by taking the movement of the living organism into consideration.

[0109] 11A and 11B are diagrams illustrating an example of the hardware configuration of a physique determination device 1 according to embodiment 1. In embodiment 1, the functions of the sensor data acquisition unit 11, the data processing unit 12, the motion estimation unit 13, the physique determination unit 14, and a control unit (not shown) are realized by a processing circuit 101. That is, the physique determination device 1 includes the processing circuit 101 for controlling the determination of the physique of a living organism present in a target space based on a motion map generated by signal processing of radio waves irradiated toward the target space by the radio wave sensor 2 and reflected by objects in the target space. The processing circuit 101 may be dedicated hardware as shown in FIG. 11A or a processor 104 that executes a program stored in memory as shown in FIG. 11B.

[0110] When processing circuitry 101 is dedicated hardware, processing circuitry 101 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0111] When the processing circuit is the processor 104, the functions of the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and the control unit (not shown) 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 105. The processor 104 reads and executes the program stored in the memory 105 to perform the functions of the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and the control unit (not shown). In other words, the physique determination device 1 includes the memory 105 for storing a program that, when executed by the processor 104, results in the execution of steps ST1 to ST4 of FIG. 10 described above. It can also be said that the program stored in the memory 105 causes a computer to execute the processing procedures or methods of the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and the control unit (not shown). Here, the memory 105 may be, for example, a non-volatile or volatile semiconductor memory such as a RAM, a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable Read Only Memory), or an EEPROM (Electrically Erasable Programmable Read-Only Memory), or a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, or a DVD (Digital Versatile Disc).

[0112] Note that the functions of the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and the control unit (not shown) may be partially implemented by dedicated hardware and partially implemented by software or firmware. For example, the sensor data acquisition unit 11 may be implemented by a processing circuit 101 as dedicated hardware, and the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and the control unit (not shown) may be implemented by a processor 104 reading and executing programs stored in a memory 105. The storage unit (not shown) may be configured, for example, with the memory 105. The physique determination device 1 further includes devices such as the radio wave sensor 2, and an input interface device 102 and an output interface device 103 that perform wired or wireless communication.

[0113] In the first embodiment described above, it is assumed that all moving objects appearing on the moving object map are living organisms. However, for example, movement may occur in objects other than living organisms, such as objects shaking. In other words, the movement of objects other than living organisms may appear on the moving object map. Therefore, in the first embodiment described above, for example, the motion estimation unit 13 may consider the moving object appearing on the moving object map to be, for example, a shaking object, and estimate the movement of the living organism when the point cloud on the moving object map satisfies a predetermined condition (hereinafter referred to as the "living organism determination condition"). Furthermore, the physique determination unit 14 may determine the physique of the living organism when the point cloud on the moving object map satisfies the living organism determination condition. The living organism determination condition is a condition set by an administrator or the like for determining whether a point cloud represents a living organism, such as whether the point cloud does not appear in a position where a living organism cannot be present.

[0114] Furthermore, in the above-described first embodiment, the physique determination device 1 is described as including the sensor data acquisition unit 11 and the data processing unit 12, but this is merely an example, and the functions of the sensor data acquisition unit 11 and the data processing unit 12 may be provided by a device external to the physique determination device 1 and connected to the physique determination device 1, and the physique determination device 1 may be configured without the sensor data acquisition unit 11 and the data processing unit 12. For example, the radio wave sensor 2 may have the functions of the sensor data acquisition unit 11 and the data processing unit 12. In this case, the processes of steps ST1 and ST2 can be omitted from the operation of the physique determination device 1 described using the flowchart of FIG. 10 .

[0115] In the first embodiment described above, the physique determination device 1 is an on-board device mounted on a vehicle, and the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and a control unit (not shown) are provided in the on-board device. However, the present invention is not limited to this. A system may be configured by the on-board device and the server, with some of the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and a control unit (not shown) being mounted in the on-board device of the vehicle, and the others being provided in a server connected to the on-board device via a network. Furthermore, the sensor data acquisition unit 11, the data processing unit 12, the movement estimation unit 13, the physique determination unit 14, and a control unit (not shown) all being provided in the server.

[0116] Furthermore, in the above-described first embodiment, as an example, the target space is the interior of a vehicle, and the living body whose physique is determined by the physique determination device 1 is a vehicle occupant, but this is merely an example. The physique determination device 1 may determine the physique of an occupant of a moving body other than a vehicle or a person in a room, for example, in the interior of a moving body other than a vehicle, such as a commercial vehicle such as a bus, a train, or an airplane, or in a room. Furthermore, the target space may be outdoors instead of indoors, and the physique determination device 1 may determine the physique of a living body located outdoors. The physique determination device 1 according to the first embodiment can be applied to a physique determination device that determines the physique of a living body located in a target space, whether indoors or outdoors.

[0117] As described above, according to the first embodiment, the physique determination device 1 is configured to include the movement estimation unit 13 that estimates the movement of the living body based on the movement map, and the physique determination unit 14 that determines the physique of the living body based on the movement map and the movement estimation information that indicates the movement of the living body estimated by the movement estimation unit 13. Therefore, the physique determination device 1 can prevent erroneous determination of the physique by taking the movement of the living body into consideration.

[0118] Any of the components of the embodiments may be modified or omitted.

[0119] The physique determination device 1 of the present disclosure can be applied to a physique determination device that determines the physique of a living organism present in a target space based on a moving body map that represents the moving bodies present in the target space as a three-dimensional spatial distribution, generated by signal processing of the radio waves irradiated toward the target space by a radio wave sensor 2 and reflected by objects in the target space, thereby preventing erroneous determination of physique.

[0120] 1 Physique determination device, 2 Radio wave sensor, 11 Sensor data acquisition unit, 12 Data processing unit, 121 Moving body extraction unit, 122 Moving body analysis unit, 123 Moving body map generation unit, 13 Motion estimation unit, 14 Physique determination unit, 101 Processing circuit, 102 Input interface device, 103 Output interface device, 104 Processor, 105 Memory.

Claims

1. A physique determination device that determines the physique of a living organism present in a target space based on a moving body map that represents moving bodies present in the target space as a three-dimensional spatial distribution, generated by signal processing of radio waves irradiated toward the target space by a radio wave sensor and reflected by objects in the target space, the physique determination device comprising: a motion estimation unit that estimates the movement of the living organism based on the moving body map; and a physique determination unit that determines the physique of the living organism based on the moving body map and motion estimation information that indicates the movement of the living organism estimated by the motion estimation unit.

2. The physique determination device according to claim 1, wherein the biological movement estimated by the movement estimation unit is expressed as a movement type or a movement amount.

3. The physique determination device according to claim 1, characterized in that the movement estimation unit estimates the movement of the living body based on the distribution pattern of the point cloud on the moving body map.

4. The physique determination device according to claim 1, characterized in that the movement estimation unit estimates the movement of the living body based on the number of points on the moving body map.

5. The physique determination device according to claim 1, characterized in that the point cloud on the moving body map is provided with information indicating reflection intensity, and the movement estimation unit estimates the movement of the living body based on the reflection intensity information provided to the point cloud on the moving body map.

6. The physique determination device of claim 1, characterized in that the point cloud on the moving body map is assigned information indicating the speed of the moving body, and the movement estimation unit estimates the movement of the living body based on the speed information assigned to the point cloud on the moving body map.

7. The physique determination device according to claim 1, characterized in that the movement estimation unit estimates the movement of the living body based on the movement map and a machine learning model that inputs the movement map and outputs information indicating movement.

8. The physique determination device according to claim 1, characterized in that the movement estimation unit estimates the movement of the living body based on the distribution pattern of point clouds on multiple movement maps generated within a set period in the past.

9. The physique determination device of claim 1, characterized in that the physique determination unit determines whether or not to perform a physique determination of the living body based on the motion estimation information, and if it determines that a physique determination of the living body should be performed, determines the physique of the living body based on the motion map.

10. The physique determination device according to claim 9, wherein the physique determination unit determines the physique of the living body based on the distribution pattern of the point cloud on the dynamic body map.

11. The physique determination device of claim 1, characterized in that the physique determination unit switches the physique determination method or changes the conditions in the physique determination method based on the motion estimation information, and determines the physique of the living body based on the motion map in accordance with the conditions in the physique determination method after switching or the physique determination method after changing.

12. The physique determination device of claim 1, characterized in that the target space is a vehicle interior, and the physique determination unit determines the physique of the living body based on at least one of the distribution pattern of the point cloud in a leg range on the dynamic body map corresponding to the range around the legs of the seat, the distribution pattern of the point cloud in a head range corresponding to the range in which the head of the living body is estimated to be present, the distribution pattern of the point cloud in a chest range in which the chest of the living body is estimated to be present, the size of the area consisting of the point cloud, the offset amount of the point cloud in the depth direction of the vehicle interior from the area in the dynamic body map considered to be the backrest, or the offset amount of the point cloud in the vertical direction of the vehicle interior from the area in the dynamic body map considered to be the seat surface of the seat.

13. The physique determination device according to claim 1, characterized in that the target space is a vehicle interior, and the movement estimation unit estimates the movement of the living body based on the moving body map and vehicle information.

14. A physique determination method for determining the physique of a living organism present in a target space based on a motion map that represents a three-dimensional spatial distribution of moving objects present in the target space, the motion map being generated by signal processing of radio waves emitted by a radio wave sensor toward the target space and reflected by objects in the target space, the physique determination method comprising the steps of: a motion estimation unit estimating the motion of the living organism based on the motion map; and a physique determination unit determining the physique of the living organism based on the motion map and motion estimation information that indicates the motion of the living organism estimated by the motion estimation unit.

Images