JP6478936B2 - Behavior estimation device, behavior estimation method, and behavior estimation program - Google Patents

Description

  Embodiments described herein relate generally to a behavior estimation device, a behavior estimation method, and a behavior estimation program.

  2. Description of the Related Art Conventionally, a technique for estimating behavior based on measurement values acquired from a sensor attached to a predetermined part of the body is known. In the past, pedometers that count the number of steps by attaching to the waist position of the body and detecting the vertical vibration that occurs with walking are widely known. In recent years, there are some devices that can measure a motion pattern with high accuracy using an acceleration sensor and estimate not only the number of steps but also a stationary time, a sleeping time, a standing motion, and a sitting motion.

  In general, in order to estimate behavior with high accuracy, it is preferable to attach a sensor to a trunk such as the waist. Behaviors such as standing and sitting are caused by movement in the direction of gravity in the trunk, and such movement is detected by a sensor and estimated from the time-series pattern. For example, there is a technique that measures acceleration in the gravitational direction and detects a standing motion or a sitting motion according to the order when upward and downward acceleration peaks are continuously observed. In addition to this, there is a technique for measuring a time-series pattern of acceleration in the gravitational direction and detecting that a standing motion has been performed by matching with a pattern corresponding to a standing motion prepared in advance.

Japanese Patent No. 3513632 Special table 2014-526913 gazette

  However, the conventional technique has a problem that it is difficult to estimate the behavior with high accuracy when the sensor is attached to a part other than the trunk. As described above, it is possible to estimate the behavior with high accuracy by attaching the sensor to the trunk such as the waist. For example, when a sensor is attached to a part other than the trunk, such as the wrist, there is a possibility that the wrist does not move in the direction of gravity even if standing or sitting is performed. Accuracy is reduced.

  The problem to be solved by the present invention is to provide a behavior estimation device, a behavior estimation method, and a behavior estimation program capable of estimating behavior with high accuracy even when a sensor is attached to a part other than the trunk. That is.

  The behavior estimation apparatus according to the embodiment includes an attitude information calculation unit, a first position information calculation unit, a second position information calculation unit, and a behavior estimation unit. The posture information calculation unit calculates posture information representing a posture of the first part or a change in posture from time-series information indicating the movement of the first part to which the sensor is attached. The first position information calculation unit calculates first position information representing the position of the first part or a change in the position from the time series information. The second position information calculation unit is a second position that represents the position of the second part or a change in the position from the posture information, the first position information, and the distance information indicating the distance between the first part and the second part. Calculate information. The behavior estimating unit estimates the behavior from the time series change of the second position information. The second part is a part whose position is uniquely determined with respect to the first part, and is a part closer to the trunk than the first part.

The figure which shows the system configuration | structure of the action estimation system which concerns on embodiment. The block diagram which shows the hardware constitutions of the action estimation apparatus which concerns on embodiment. The figure explaining mounting | wearing of the sensor which concerns on embodiment. The block diagram which shows the function structure of the action estimation apparatus which concerns on embodiment. The figure explaining estimation of the operation | movement which walks and drive | works which concerns on embodiment. It shows the change of the waveform representing the height H ₀ from the ground (floor) to the first part during the estimation operation for walking and running according to the embodiment. The figure which shows the change of the waveform showing L * sinP at the time of the estimation of the operation | movement which walks / runs concerning embodiment. The figure which shows the change of the waveform showing H at the time of the estimation of the operation | movement to walk / run according to embodiment. The figure explaining the estimation of the standing motion which concerns on embodiment. It shows the change of the waveform representing the height H ₀ from the ground (floor) to the first site in the time estimation operation stand according to the embodiment. The figure which shows the change of the waveform showing L * sinP at the time of estimation of the standing motion which concerns on embodiment. The figure which shows the change of the waveform showing H at the time of estimation of the standing motion which concerns on embodiment. The figure explaining the example which cancels the estimation of the action which concerns on embodiment. It shows the change of the waveform representing the height H ₀ from the ground (floor) during the cancellation of the estimated behavior of Embodiment until the first portion. The figure which shows the change of the waveform showing L * sinP at the time of cancellation of the estimation of the action which concerns on embodiment. The figure which shows the change of the waveform showing H at the time of cancellation of the estimation of the action which concerns on embodiment. The flowchart which shows the flow of the process by the action estimation apparatus which concerns on embodiment. The figure explaining estimation of the operation which crouches concerning a modification. It shows the change of the waveform representing the height H ₀ from the ground (floor) to the first portion when the estimation of squat according to a modification operation. The figure which shows the change of the waveform showing L * sinP at the time of the estimation of the squatting operation | movement which concerns on a modification. The figure which shows the change of the waveform showing H at the time of the estimation of the squatting operation | movement which concerns on a modification. The figure explaining the example which cancels the estimation of the action which concerns on a modification. It shows the change of the waveform representing the height H ₀ from the ground (floor) to the first portion when the cancellation of the estimated behavior according to a modification. The figure which shows the change of the waveform showing L * sinP at the time of cancellation of the estimation of the action which concerns on a modification. The figure which shows the change of the waveform showing H at the time of cancellation of the estimation of the action which concerns on a modification.

(Embodiment)
FIG. 1 is a diagram illustrating a system configuration example of a behavior estimation system 1 according to the embodiment. As shown in FIG. 1, the behavior estimation system 1 includes a behavior estimation device 100 and a sensor 200. In the behavior estimation system 1, each device can communicate wirelessly or by wire. The behavior estimation system 1 attaches a sensor 200 (for example, a motion sensor) to a specific part other than the trunk, and estimates the behavior of the user based on the measurement value measured by the sensor 200 according to the user's movement. To do.

  For example, the sensor 200 is a motion sensor that is worn on a part other than the trunk, such as a user's wrist, ankle, or head, and outputs measurement values that are measured according to the movement of these parts. The measurement value is output to the behavior estimation apparatus 100. In this embodiment, the case where the sensor 200 is mainly worn on the user's wrist is taken as an example. For example, the behavior estimation device 100 is an information processing device such as a PC (Personal Computer), a server device, or a tablet terminal. The behavior estimation apparatus 100 estimates the behavior of the user wearing the sensor 200 based on the measurement value output by the sensor 200.

  FIG. 2 is a block diagram illustrating a hardware configuration example of the behavior estimation apparatus 100 according to the embodiment. As illustrated in FIG. 2, the behavior estimation apparatus 100 includes a CPU (Central Processing Unit) 12, a ROM (Read Only Memory) 13, a RAM (Random Access Memory) 14, and a communication unit 15. The above units are connected to each other by a bus 11.

  CPU12 controls operation | movement of the action estimation apparatus 100 whole. CPU12 controls operation | movement of the action estimation apparatus 100 whole by running the program memorize | stored in ROM13 grade | etc., Using RAM14 grade | etc., As a work area. The RAM 14 temporarily stores information related to various processes, and is used as a work area when executing a program stored in the ROM 13 or the like. ROM13 memorize | stores the program for implement | achieving the process by the action estimation apparatus 100. FIG. The communication unit 15 communicates with an external device such as the sensor 200 via a wireless or wired network. The hardware configuration illustrated in FIG. 2 is an example, and in addition to the above, a display unit that outputs a processing result, an operation unit that inputs various information, and the like may be included.

  FIG. 3 is a diagram illustrating an example of mounting the sensor 200 according to the embodiment. As shown in FIG. 3, the sensor 200 is attached to a part other than the trunk, such as the user's wrist. Hereinafter, a part such as a wrist to which the sensor 200 is attached is referred to as a “first part”. In addition, a part whose position is uniquely determined with respect to the first part and is closer to the trunk than the first part is referred to as a “second part”. In the example shown in FIG. 3, when the wrist is the first part, the elbow is the second part. In general, the elbow has a linear positional relationship with the wrist and is a part where the position is uniquely determined with respect to the wrist and is closer to the trunk than the wrist. As will be described later, the position (change in position) of the elbow, which is the second part, can be calculated from the posture and position of the wrist, which is the first part, and the distance between the first part and the second part. When the ankle is the first part, the knee is the second part. When the head is the first part, the neck is the second part. In any case, the second part is a part whose position is uniquely determined with respect to the first part, and is a part closer to the trunk than the first part. The first part and the second part are not limited to the above example.

  FIG. 4 is a block diagram illustrating a functional configuration example of the behavior estimation apparatus 100 according to the embodiment. As illustrated in FIG. 4, the behavior estimation apparatus 100 includes a posture information calculation unit 110, a first position information calculation unit 120, an acquisition unit 130, a second position information calculation unit 140, and a behavior estimation unit 150. . Some or all of these parts may be realized by software (programs), or may be realized by a hardware circuit.

  The posture information calculation unit 110 calculates posture information of the first part. More specifically, the posture information calculation unit 110 uses the time series information of the measurement values output from the sensor 200 via the communication unit 15 to determine the posture of the first part such as the wrist to which the sensor 200 is worn or Posture information representing a change in posture is calculated. As one aspect, the posture information is a pitch angle with reference to a plane orthogonal to the gravity direction axis. In addition, the posture information may be a change amount of the pitch angle per time.

  For example, the pitch angle can be calculated by a known posture estimation algorithm from the acceleration measured by the triaxial acceleration sensor and the angular velocity measured by the triaxial gyro sensor. Further, for example, the pitch angle may be calculated from the angle formed with the coordinate axis of the motion sensor using a known technique for estimating the direction of gravity from the acceleration measured by the three-axis acceleration sensor.

  The first position information calculation unit 120 calculates position information of the first part. More specifically, the first position information calculation unit 120 uses the time series information of the measurement values output from the sensor 200 via the communication unit 15 to determine the first part such as the wrist to which the sensor 200 is attached. First position information representing a position or a change in position is calculated. The first position information is a relative change amount in the direction of gravity from the initial position. For example, the first position information may be a relative change amount of coordinates from the initial position when the gravity direction is the coordinate axis. In addition, the first position information may be the speed and acceleration of the first part.

  For example, the first position information calculation unit 120 calculates the relative movement amount of the first part in the gravitational direction with reference to the startup time of the sensor 200 (initial position). This amount of movement is obtained by calculating the acceleration accompanying the movement of the first part by removing the gravitational acceleration component from the acceleration measured by the acceleration sensor and then integrating the gravitational direction component by the second order. For example, regarding the removal of the gravitational acceleration component from the acceleration measured by the acceleration sensor, the posture information calculation unit 110 calculates the roll angle R together with the pitch angle P as the posture information, and the gravitational acceleration component GX on each axis of the sensor 200. , GY and GZ are calculated by using (Equation 1). In (Expression 1), G represents the magnitude of gravitational acceleration.

GX = -G · sinP
GY = G · (sinR · cosP)
GZ = G · (cosR · cosP) (Equation 1)

  The gravity direction component AG of the acceleration accompanying the movement can be calculated from the measured values AX, AY, and AZ of each axis of the sensor 200 by (Equation 2).

  AG = (AX-GX) .GX + (AY-GY) .GY + (AZ-GZ) .GZ (Equation 2)

  The acquisition unit 130 acquires an actual measurement value of the distance between the first part and the second part or information for estimating the distance. More specifically, the acquisition unit 130 acquires a measured value of a distance between a first part such as a wrist and a second part such as an elbow or information used for estimating the distance. For example, the distance information between the first part and the second part is an actual measurement value of the length between the first part and the second part. In addition, since the distance between the wrist (first part) and the elbow (second part) can be statistically estimated by height, sex, etc., the distance information is the length between the wrist and the elbow. Instead of inputting the actual measured value, the user may input the height and gender and obtain the information for estimating the distance.

  In addition, with the sensor 200 attached, the user is instructed to move only the tip of the elbow, and the length between the wrist and the elbow is estimated from the relationship between acceleration and angular velocity measured by the sensor 200. May be. Specifically, since the wrist moves on a spherical surface centered on the elbow, the angular velocity measured by the sensor 200 attached to the wrist is expressed as a function of acceleration and the length between the wrist and the elbow. can do. Therefore, for example, after mounting the sensor 200 so that the Z coordinate axis of the sensor 200 is perpendicular to the back of the hand, the user is instructed to bend only the elbow, and the acceleration and angular velocity of the sensor 200 in the Z coordinate axis direction at this time It can be calculated by measuring. When the differential value of the angular velocity is dωZ / dt, the length L between the wrist and the elbow can be estimated by (Equation 3).

  dωZ / dt = (AZ−GZ) / L (Equation 3)

  In other words, the distance between the first part and the second part may be a value input by the user as an actual measurement value as it is, may be obtained from information such as height and gender, You may estimate from the measured value measured by the sensor 200 after instructing.

  The second position information calculation unit 140 calculates position information of the second part. More specifically, the second position information calculation unit 140 includes the first part posture information calculated by the posture information calculation unit 110, the first part position information calculated by the first position information calculation unit 120, and Then, from the distance information between the first part and the second part acquired by the acquisition unit 130, the second position information representing the position of the second part or a change in the position is calculated. The second position information is a relative change amount in the direction of gravity from the initial position. For example, the second position information may be a relative change amount of coordinates from the initial position when the direction of gravity is the coordinate axis. That is, the second position information is a relative position in the gravity direction of the second part with reference to the gravity direction coordinates of the position information (first position information) of the first part when the sensor 200 is activated.

  For example, if the relative movement amount of the first part in the gravity direction is ΔG, the pitch angle of the first part is P, and the length between the wrist and the elbow is L, the relative position H of the second part is , (Equation 4).

  H = ΔG + L · sinP (Expression 4)

  The behavior estimation unit 150 estimates a user's behavior. More specifically, the behavior estimating unit 150 estimates the user's behavior based on the second position information calculated by the second position information calculating unit 140. For example, the behavior estimation unit 150 estimates a behavior that is an estimation target such as a motion of the user walking / running or a motion of the user standing / sitting. Hereinafter, an example of an action to be estimated will be described.

FIG. 5 is a diagram for explaining an estimation example of the walking / running operation according to the embodiment. FIG. 6A is a diagram illustrating a change example of a waveform representing a height H ₀ from the ground (floor) to the first part when estimating walking / running motion according to the embodiment. FIG. 6B is a diagram illustrating a change example of a waveform representing L · sinP when estimating walking / running motion according to the embodiment. FIG. 6C is a diagram illustrating a change example of a waveform representing H at the time of estimating the walking / running operation according to the embodiment. Note that FIG. 6A also shows a waveform (broken line) with respect to vertical movement caused by hand shaking.

As shown in FIG. 5, during walking or running, the states (A) and (B) are repeated. Specifically, as shown in FIG. 6A, H ₀ becomes small (takes a minimum value) in the state (A) and becomes large (takes a maximum value) in the state (B). Note that the time width between the state of (A) and the state of (B) is different between the time of walking and the time of running. Specifically, during walking, the time width between the state (A) and the state (B) is greater than during travel. For discrimination between walking and running, a cycle may be confirmed using a threshold value. Further, as shown in FIG. 6B, L · sinP increases (takes a maximum value) in the state (A) and decreases (takes a minimum value) in the state (B). During walking, the time width between the state (A) and the state (B) is greater than during travel. Further, as shown in FIG. 6C, H increases (takes a maximum value) in the state (A), and decreases (takes a minimum value) in the state (B). That is, it is understood that the vertical movement is repeated by walking or running. From these, the action estimation unit 150 estimates the action as a walking action or a running action.

FIG. 7 is a diagram for explaining an estimation example of the standing motion according to the embodiment. FIG. 8A is a diagram illustrating a change example of a waveform representing a height H ₀ from the ground (floor) to the first part at the time of estimating the standing motion according to the embodiment. FIG. 8B is a diagram illustrating a change example of a waveform representing L · sinP at the time of estimating the standing motion according to the embodiment. FIG. 8C is a diagram illustrating a change example of a waveform representing H at the time of estimating a standing motion according to the embodiment. In addition, when performing the operation | movement which stands | stands up from the state sitting on a chair etc., the operation | movement standing with a hand on the desk shall be made.

As shown in FIG. 7, when a standing operation is performed, the state transitions from the state (A) to the state (B). Specifically, as shown in FIG. 8A, H ₀ is not changed between the state of (A) and the state of (B) because an operation of standing with a hand on the desk is performed. Further, as shown in FIG. 8B, L · sinP is larger in the state of (B) than in the state of (A). Further, as shown in FIG. 8C, H is larger in the state (B) than in the state (A). From these, the behavior estimation unit 150 estimates the behavior as a standing motion. Regarding the sitting motion, since transition from (B) to (A) occurs, H ₀ does not change, and L · sinP decreases in time series from (B) to (A). Becomes smaller in time series from (B) to (A). In such a case, the behavior estimation unit 150 estimates the behavior as a sitting motion.

FIG. 9 is a diagram illustrating an example of canceling action estimation according to the embodiment. FIG. 10A is a diagram illustrating a change example of a waveform representing a height H ₀ from the ground (floor) to the first part at the time of canceling the behavior estimation according to the embodiment. FIG. 10B is a diagram illustrating an example of a change in a waveform representing L · sinP at the time of canceling the behavior estimation according to the embodiment. FIG. 10C is a diagram illustrating an example of a change in a waveform representing H at the time of canceling the behavior estimation according to the embodiment. Cancellation of action estimation means that the user is not performing the action to be estimated.

As shown in FIG. 9, when an operation of simply raising and lowering a hand is performed instead of a walking operation, a traveling operation, a standing operation, or a sitting operation, the states of (A) and (B) are repeated. . Specifically, as shown in FIG. 10A, H ₀ becomes smaller in the state of (A) due to the movement of the hand accompanying some work for raising and lowering the hand (takes a minimum value), and in the state of (B). Increases (maximum value). Further, as shown in FIG. 10B, L · sinP increases (takes a maximum value) in the state (A) and decreases (takes a minimum value) in the state (B). Further, as shown in FIG. 10C, H does not change between the state (A) and the state (B). In such a case, the behavior estimation unit 150 cancels the behavior estimation assuming that the behavior is not the estimation target behavior.

  As described above, the behavior estimation apparatus 100 can estimate a user's behavior in time series. For this reason, the behavior estimation apparatus 100 can output, as a processing result, what kind of behavior the user has performed, how many times he stood within a certain period of time, and how he sat down.

  FIG. 11 is a flowchart illustrating an example of a process flow performed by the behavior estimation apparatus 100 according to the embodiment. As illustrated in FIG. 11, the behavior estimation apparatus 100 calculates posture information of the first part based on the measurement value measured by the sensor 200 attached to the first part (step S101). Then, the behavior estimation apparatus 100 calculates the position information of the first part based on the measurement value measured by the sensor 200 attached to the first part (step S102).

  Subsequently, the behavior estimation apparatus 100 acquires distance information between the first part and the second part (step S103). Thereafter, the behavior estimation apparatus 100 calculates the position information of the second part using the posture information and position information of the first part, and the distance information between the parts (step S104). Then, the behavior estimation apparatus 100 estimates the behavior based on the position information of the second part (Step S105).

  For example, in behavior estimation, it may be determined whether or not behavior estimation processing is to be executed based on a change in the coordinate of the second part in the direction of gravity. That is, it is only necessary to execute the behavior estimation process when it is determined whether or not the user is performing any operation and then the state where the user is operating is shifted to the state where the user is not operating. In determining whether or not the user is operating, for example, a variance of relative coordinate values in a predetermined time is calculated, and the determination is made based on whether the variance value is greater than or less than a threshold value. Here, a case where a standing motion or a sitting motion is estimated is taken as an example.

  Specifically, the behavior estimation apparatus 100 identifies a standing motion or a sitting motion by evaluating the moving direction and the moving amount of the second part in a predetermined section. The predetermined section indicates a time during which it is determined that the user is operating, for example. When the position H moves upward by a predetermined length or more during a predetermined section, it is determined that a standing motion is being performed, and the standing motion is output as a behavior estimation result. Further, when the position H moves downward by a predetermined length or more during a predetermined section, it is determined that the sitting motion is being performed, and the sitting motion is output as an action estimation result.

In addition, the said embodiment is an example and the operation | movement which can be estimated by the position H of an elbow is not limited to the operation | movement which stands or sits down. For example, it is possible to detect a user's fall based on a change in the position H of the elbow. The wrist height H ₀ , which is the first part, is close to the height of the ground when sitting with the hand on the ground (or the floor, etc.) or when the hand falls. At this time, the elbow as the second part is usually at least a position away from the ground when sitting with the hand on the ground (or the floor, etc.), so the position information of the second part calculated in step S104 is obtained. By using it, it is possible to distinguish between a fall state (or a state of lying with an elbow, etc.) and a state of sitting with a hand on the ground.

(Modification)
By using the first part as the ankle and the second part as the knee, it is also possible to estimate the action of the user squatting. Thus, for example, by using a sensor built-in band that can be worn on the ankle, or a sock with a built-in sensor 200 in the ankle, it is possible to estimate a squatting action without directly wearing the sensor on the knee that is the movable part. it can. As one aspect, it can be used to analyze the physical load of the user in the work.

  When the sensor 200 is attached to the ankle, it can be estimated that the squatting motion is performed also by detecting the motion of the ankle falling forward. However, since the angle of the ankle changes even when the knee is bent without squatting, it may be erroneously estimated as a squatting action in this case. According to the modification, the squatting motion can be estimated with higher accuracy by estimating the height of the knee more accurately from the angle of the ankle.

FIG. 12 is a diagram illustrating an estimation example of the squatting motion according to the modification. FIG. 13A is a diagram illustrating a change example of a waveform representing a height H ₀ from the ground (floor) to the first part at the time of estimating the squatting motion according to the modification. FIG. 13B is a diagram illustrating a change example of a waveform representing L · sinP at the time of estimating a squatting action according to the modification. FIG. 13C is a diagram illustrating a change example of a waveform representing H at the time of estimating a squatting action according to the modification.

As shown in FIG. 12, when a squatting action is performed, the state transitions from the state (A) to the state (B). Specifically, as shown in FIG. 13A, H ₀ changes between the state of (A) and the state of (B) because only the posture of the first part changes with the foot on the ground. There is almost no change. On the other hand, as shown in FIG. 13B, L · sinP becomes smaller because the angle of the ankle changes with the transition from the state (A) to the state (B). As a result, as shown in FIG. 12C, H is smaller in the state (B) than in the state (A). From these, the action estimation unit 150 estimates the action as a squatting action.

FIG. 14 is a diagram illustrating an example of canceling action estimation according to the modification. FIG. 15A is a diagram illustrating a change example of a waveform representing the height H ₀ from the ground (floor) to the first part when canceling the behavior estimation according to the modification. FIG. 15B is a diagram illustrating a change example of a waveform representing L · sinP at the time of canceling action estimation according to the modification. FIG. 15C is a diagram illustrating a change example of a waveform representing H at the time of canceling the action estimation according to the modification. Cancellation of action estimation means that the user is not performing the action to be estimated.

As shown in FIG. 14, when the operation of merely bending the knee while standing is performed instead of the operation of squatting, the state transitions from the state (A) to the state (B). Specifically, as shown in FIG. 15A, H ₀ increases as the ankle lifts with the movement of bending the knee, so that the state changes from the state (A) to the state (B). Also, as shown in FIG. 15B, L · sinP decreases as the angle of the ankle changes with the movement of bending the knee, and thus transitions from the state (A) to the state (B). As a result, as shown in FIG. 15C, H does not change between the state (A) and the state (B). In such a case, the behavior estimation unit 150 cancels the behavior estimation assuming that the behavior is not the estimation target behavior.

  In the above example of the estimation of the crouching motion, the first part has been described as the ankle. However, because of the structure of the foot skeleton, when the ankle is bent forward, the angle of the heel also changes at the same time. A crouching motion can be estimated even as a spear. Accordingly, for example, the squatting action can be estimated even by incorporating the sensor 200 in the heel portion of the shoe, so that the psychological burden on the user can be further reduced.

  According to the embodiment, a part that is uniquely determined from the position of the first part based on posture information and position information of the first part to which the sensor 200 is attached, and that is closer to the trunk than the first part Since the position information of the second part is calculated and the behavior of the user wearing the sensor 200 is estimated, the behavior can be estimated with high accuracy even when the sensor 200 is worn on a part other than the trunk. Can do. Further, according to the embodiment, since the sensor 200 is attached to a part such as a wrist, the psychological burden at the time of attachment is small, and it is easy for more users to accept. In addition, according to the embodiment, the behavior is estimated from the position information of the second part, which is a part whose position is uniquely determined with respect to the first part and is closer to the trunk than the first part. The degree of freedom of the place where 200 is mounted is high.

  Moreover, the behavior estimation apparatus 100 according to the above embodiment can be realized by using a general-purpose computer device as basic hardware, for example. The program to be executed has a module configuration including each function described above. Further, the program to be executed is an installable or executable file that is recorded on a computer-readable recording medium such as a CD-ROM, CD-R, DVD, etc. It may be provided by being incorporated in advance.

  The above-described embodiments are presented as examples, and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. Moreover, each embodiment can be combined suitably as long as the contents do not contradict each other. Each embodiment and its modifications are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof. For example, in the embodiment, the case where the behavior estimation device 100 and the sensor 200 are separate casings has been described as an example, but each function of the behavior estimation device 100 described above may be included in the sensor 200.

DESCRIPTION OF SYMBOLS 100 Behavior estimation apparatus 110 Posture information calculation part 120 1st position information calculation part 130 Acquisition part 140 2nd position information calculation part 150 Action estimation part 200 Sensor

Claims (7)

A posture information calculation unit for calculating posture information representing a posture of the first part or a change in posture from time-series information indicating movement of the first part to which the sensor is attached;
A first position information calculation unit for calculating first position information representing a position of the first part or a change in position from the time series information;
Second position information representing the position of the second part or a change in the position is calculated from the posture information, the first position information, and distance information indicating a distance between the first part and the second part. A second position information calculation unit;
An action estimation unit that estimates an action from a time-series change of the second position information,
The behavior estimation apparatus, wherein the second part is a part whose position is uniquely determined with respect to the first part and closer to the trunk than the first part. An actual value of the distance, or an acquisition unit for acquiring information for estimating the distance;
The second position information calculation unit uses the measured value of the distance or the distance between the first part and the second part calculated from the information for estimating the distance, to calculate the second position. The behavior estimation apparatus according to claim 1 which calculates information.   The behavior estimation according to claim 1, wherein the posture information calculation unit calculates the posture information that is a pitch angle based on a plane orthogonal to the gravity direction axis or a change amount of the pitch angle per time. apparatus.   The behavior estimation apparatus according to any one of claims 1 to 3, wherein the first position information calculation unit calculates the first position information that is a relative change amount in a direction of gravity from an initial position.   The behavior estimation apparatus according to any one of claims 1 to 4, wherein the second position information calculation unit calculates the second position information, which is a relative change amount in a gravity direction from an initial position. A behavior estimation method executed by a behavior estimation device,
Calculating posture information representing a posture of the first part or a change in posture from time-series information indicating movement of the first part to which the sensor is attached; and
Calculating first position information representing a position of the first part or a change in position from the time-series information;
Second position information representing the position of the second part or a change in the position is calculated from the posture information, the first position information, and distance information indicating a distance between the first part and the second part. Steps,
Estimating a behavior from a time-series change of the second position information,
The second part is a part whose position is uniquely determined with respect to the first part, and is a part closer to the trunk than the first part. In the behavior estimation device,
Calculating posture information representing a posture of the first part or a change in posture from time-series information indicating movement of the first part to which the sensor is attached; and
Calculating first position information representing a position of the first part or a change in position from the time-series information;
Second position information representing the position of the second part or a change in the position is calculated from the posture information, the first position information, and distance information indicating a distance between the first part and the second part. Steps,
Estimating a behavior from a time-series change of the second position information, and
The behavior estimation program, wherein the second part is a part whose position is uniquely determined with respect to the first part and closer to the trunk than the first part.

Images