JP5336896B2 - Objective behavior estimation apparatus and service providing system - Google Patents

Description

  The present invention relates to an objective behavior estimation device and a service providing system, and more particularly to a novel objective behavior estimation device and a service provision system that estimate, for example, what kind of behavior a human is performing on a specific object.

  2. Description of the Related Art Conventionally, a technique has been proposed for estimating what kind of behavior a person is performing based on accumulated personal position information and movement information. For example, in the technique of Patent Document 1, the position information of the target person is acquired using a GPS terminal, and “less interesting search behavior”, “hang around” based on the movement history and the time taken for movement. The behavioral state of the target person such as “searching behavior” and “searching behavior with a target of interest” is estimated.

JP 2008-152655 A [G06Q 50/00]

  However, in the conventional behavior estimation device such as the technique of Patent Document 1, the positional relationship between a specific object and a human is not considered, and what kind of action a human is performing on the specific object There is no technique to estimate

  Further, in the technique of Patent Document 1, since it is necessary to observe the target person for several minutes or more in order to estimate the action of the target person, it is not possible to provide a service that quickly corresponds to the action of the target person.

  Therefore, a main object of the present invention is to provide a novel objective behavior estimation apparatus and service providing system.

  Another object of the present invention is to provide an objective behavior estimation device and a service providing system that can estimate what kind of behavior a human is performing on a specific object.

  Still another object of the present invention is to provide an objective behavior estimating apparatus and a service providing system capable of estimating what behavior a human being is doing with respect to a specific object by short-time observation.

  The present invention employs the following configuration in order to solve the above problems. Note that reference numerals in parentheses and supplementary explanations indicate correspondence with embodiments described later in order to help understanding of the present invention, and do not limit the present invention.

1st invention is the objective action estimation apparatus which estimates what kind of action the person is performing with respect to a specific target object, Comprising: A target object is a mobile body which can move in an environment, and a human being And a position history acquisition unit that acquires a position history for a predetermined time of the object, and a first feature amount that indicates a feature amount related to a human and the object from the position history acquired by the position history acquisition unit. A feature quantity calculating means, a third feature quantity calculating means for calculating a third feature quantity indicating a feature quantity relating to the movement of the object from the position history acquired by the position history acquiring means, and a first feature quantity calculated by the first feature quantity calculating means. It is an objective behavior estimation apparatus provided with behavior estimation means for estimating behavior performed by a human on an object using the feature amount and the third feature amount calculated by the third feature amount calculation means .

In the first invention, the objective behavior estimation apparatus (10) includes a computer (30), and is detected by a position detection system (12) installed in the environment and an object movable in the environment . Based on the position information of ( 18b ), it is estimated what action a human being existing in the environment is performing on a specific object. The position history acquisition means (32, 36, S7, S15) acquires the position history for a predetermined time, for example, several seconds to several tens of seconds, of the human and the object from the position detection system. The first feature value calculating means (32, S19) calculates a first feature value that is a feature value related to the relationship between the person and the object from the position history data of the person and the object. For example, the distance between the human position coordinates and the position coordinates of the object, the relative speed of the person with respect to the object, and the like are calculated. Further, the third feature value calculating means (32, S17) calculates a third feature value that is a feature value related to the movement of the object from the position history data of the object. For example, a feature amount related to the moving distance, speed, angle, variance, etc. of the object is calculated. The behavior estimation means (32, 38, S21) estimates human behavior using the first feature value and the third feature value . For example, the first feature amount and the third feature amount are given to a discriminant defined using a threshold value, or a learning algorithm such as a support vector machine (SVM) is used to perform a human operation on an object. Is estimated (objective behavior).

According to the first aspect of the present invention, the position history of the person and the object is detected, the first feature quantity indicating the feature quantity between the person and the object, and the third feature quantity indicating the feature quantity related to the movement of the object. By using, it is possible to estimate in a short time the action that a person is performing on an object. Therefore, it is possible to quickly provide a service corresponding to the human objective behavior.

  A second invention is dependent on the first invention, and comprises second feature quantity calculation means for calculating a second feature quantity indicating a feature quantity related to human movement from the position history acquired by the position history acquisition means, and the behavior estimation The means further estimates an action performed by a human on the object using the second feature amount.

  In the second invention, the second feature value calculating means (32, S9) calculates a second feature value that is a feature value related to the movement of the person from the position history data of the person (16). For example, a feature amount related to a human moving distance, speed, angle, variance, and the like is calculated. The behavior estimation means (32, 38, S21) further estimates the human objective behavior using the second feature amount.

According to the second invention, since the second feature value is calculated in addition to the first feature value and the third feature value and is used for estimating the human objective behavior, the human objective behavior can be estimated more accurately. In addition, more types of objective behavior can be estimated.

A third invention is a service providing system comprising the objective behavior estimating device according to claim 1 or 2 and a service providing means for providing a service according to a human behavior estimated by the objective behavior estimating device.

In the third invention, the service providing system (100) includes the objective behavior estimation device (10), and is movable within the environment and the human (16) detected by the position detection system (12) installed in the environment. Based on the position information of the object ( 18b ), it is estimated what action a human being existing in the environment is performing on the specific object. The service providing means (10, 14, S23) provides a service corresponding to the estimated human objective behavior to the human.

According to the third invention, similar to the first invention, it is possible to estimate in a short time the action that a human being in the environment is performing on a specific object, so that it can quickly respond to the human objective action. Service can be provided.

A fourth invention is dependent on the third invention, and the service providing means includes a communication robot having a function of executing communication using voice and body movement.

In the fourth invention, the service providing means (10, 14, S23) includes a communication robot (14, 18b). For example, when the communication robot is waiting at a certain point and the person (16) is approaching, the communication robot is controlled to move toward the person.

According to the fourth aspect of the invention, it is possible to control the communication robot according to the situation at the site by estimating how the human being is acting on the object. It eliminates the waste of movement and provides more advanced services more quickly.

According to the present invention, the position history of the person and the object is detected, and the first feature quantity indicating the feature quantity between the person and the object calculated from the position history and the feature quantity relating to the movement of the object are displayed. By using the three feature amounts , it is possible to estimate in a short time what kind of action a human being existing in the environment is performing on a specific object. Therefore, it is possible to quickly provide a service corresponding to a human action to the human.

  The above object, other objects, features, and advantages of the present invention will become more apparent from the following detailed description of embodiments with reference to the drawings.

It is an illustration figure which shows one Example of the service provision system of this invention. It is an illustration figure which shows schematically the mode of the environment where the position detection system 12 of FIG. 1 was applied. It is an illustration figure which shows the electrical structure of the position detection system 12 of FIG. It is an illustration figure which shows an example of the position history data which the position detection system of FIG. 1 detects. It is an illustration figure which shows an example of the human's objective action "approaching the target object." It is an illustration figure which shows an example of the human's objective action "going away from a target object." It is an illustration figure which shows an example of a human's objective action which "follows an object." It is an illustration figure which shows an example of the human's objective action "Waiting for the target object". It is an illustration figure which shows an example of the human's objective action "conversing with the target object". It is a flowchart which shows an example of operation | movement of the objective action estimation apparatus of FIG.

  Referring to FIG. 1, a service providing system (hereinafter referred to as “providing system”) 100 according to an embodiment of the present invention includes an objective behavior estimating device (hereinafter referred to as “estimating device”) 10 and a position detection. System 12 is included. The providing system 100 is applied to an environment such as an event venue, a shopping center, a museum, an amusement park, and the like, based on position information detected by the position detection system 12, with respect to an objective action of the human 16, that is, a specific object 18. Thus, it is estimated what action the human 16 is performing. Based on the estimated objective behavior of the human 16, a service suitable for the objective behavior is provided to the human 16 by giving an instruction to the service providing device 14 such as a communication robot or a mobile phone.

  The specific object 18 is not particularly limited as long as it is an object that exists in the environment to which the providing system 100 is applied, but preferably an object that the human 16 makes some kind of action or an object that makes some kind of action to the human 16 Is appropriately selected as the specific object 18. Further, the specific object 18 includes an installation object 18a that is fixedly provided in the environment and a moving body 18b that can move in the environment. Examples of the installation 18a include vending machines, display shelves, and guide signs (maps). Examples of the moving body 18b include mobile robots such as communication robots (humanoid robots) and animals such as dogs and cats. Can be mentioned.

  First, the position detection system 12 will be described. FIG. 2 is an illustrative view schematically showing an environment where the position detection system 12 is applied, and FIG. 3 is an illustrative view showing an electrical configuration of the position detection system 12. As shown in FIGS. 2 and 3, the position detection system 12 is a system for detecting the positions of a person 16 and an object 18, and includes a computer 20 that executes position detection processing, and a plurality of systems installed in the environment. It includes a laser range finder (LRF) 22 and a wireless ID tag reader 24, and a wireless ID tag 26 attached to the human 16 and a moving body 18b (movable object 18).

  The computer 20 of the position detection system 12 is a general-purpose computer such as a personal computer or a workstation. A plurality of LRFs 22 and a wireless ID tag reader 24 are connected to the computer 20 via a general-purpose interface such as RS-232C. Data including information detected by these devices 22 and 24 is transmitted to the computer 20 and is appropriately stored in a predetermined area of the memory of the computer 20 or an external database together with the detection time. Further, in the memory of the computer 20, the map data of the environment expressed in the XY two-dimensional plane coordinate system, the position data of the devices 22 and 24, the position data of the installation object 18a (the object 18 fixedly provided), etc. Is stored in advance. When the position coordinates (position history) of the installation object 18a are required, the position data stored in the memory of the computer 20 is appropriately read and used.

  The LRF 22 is an image sensor using a laser beam, and irradiates a pulse laser, and instantaneously measures the distance from the object from the time when the laser beam is reflected back to the object. As LRF22, LRF (model LMS 200) manufactured by SICK or the like can be used. The LRF 22 is installed, for example, on a pedestal having a height of about 90 cm, and is mainly used for detecting the position coordinates of the human 16 existing in the environment.

  In this embodiment, the computer 20 acquires sensor information from the LRF 22 at a frequency of, for example, 37.5 times / second, and estimates the position coordinates of the waist of the human 16 and the body direction using a particle filter and a human shape model. . The details of this estimation method are described in Japanese Patent Application No. 2008-6105 filed earlier by the present applicant.

  The wireless ID tag 26 is attached to each person 16 and each moving body 18b existing in the environment. As the wireless ID tag 26, for example, an RFID (Radio Frequency Identification) tag can be used. RFID is an automatic recognition technology using a non-contact IC tag using electromagnetic waves. Specifically, the RFID tag includes an IC chip, an antenna, and the like provided with a memory for identification information, a control circuit for communication, and the like. Information capable of identifying the human 16 and the moving body 18b is stored in advance in the memory of the RFID tag, and the identification information is output from the antenna by electromagnetic waves / radio waves having a predetermined frequency.

  The wireless ID tag reader 24 is installed, for example, on a rail laid on the ceiling in the environment, and detects output information from the wireless ID tag 26 described above. Specifically, the wireless ID tag reader 24 receives a radio wave superimposed with identification information transmitted from the radio ID tag 26 via an antenna, amplifies the radio signal, and separates the identification information from the radio signal. The information is demodulated (decoded). The wireless ID tag reader 24 detects the distance from the wireless ID tag 26 based on the detected radio wave intensity.

  In this embodiment, the computer 20 acquires sensor information (identification information and distance information) from the wireless ID tag 26 detected by the wireless ID tag reader 24 at a frequency of, for example, 5 times / second. Based on the acquired identification information, the person 16 and the moving body 18b wearing the wireless ID tag 26 are specified. Further, a person who possesses the wireless ID tag 26 based on the distance information acquired from the three wireless ID tag readers 24 for the same wireless ID tag 26 and the position data of the wireless ID tag reader 24 stored in advance. 16 and the position coordinates of the moving body 18b are specified.

  Then, the computer 20 of the position detection system 12 integrates sensor information (positioning data) detected by the devices 22 and 24 or information estimated from the sensor information, and positions coordinates (x , Y) is stored in the memory or database in association with the detection time t. That is, the position history is stored as time-series data for each person 16 and the moving body 18b (identification information or ID number). For example, when the position at time tn is Ptn (xn, yn), the trajectory from time t0 to tn is represented as a point sequence Pt0, Pt1,..., Ptn. That is, the position history data (movement trajectory data) of each person 16 and each moving body 18b is represented by a sequence of n + 1 points on the xy plane as shown in FIG. 4, and each point (circled in FIG. 4). Includes information on the time at which the point was observed.

  The position detection system 12 is not limited to the above-described configuration as long as it can detect each person 16 and each moving body 18b and detect position coordinates or position history. A floor sensor, a GPS, or the like may be used as appropriate. In addition, when the moving body 18b moves while recognizing its own position with reference to the map data of the environment, for example, an autonomous moving robot, the moving body 18b sends the information to the computer 20 of the position detection system 12. Then, the own position history data may be transmitted.

  Based on the position history data of the person 16 and the object 18 transmitted from the position detection system 12 as described above, the estimation apparatus 10 estimates the action performed by the person 16 existing in the environment. In other words, the objective action of the human 16 is estimated based on the change in the positional relationship between the human 16 and the object 18.

  Returning to FIG. 1, the estimation device 10 includes a computer 30 that calculates a feature amount from position history data and estimates the objective behavior of the human 16. The computer 30 is, for example, a general-purpose personal computer or workstation including a CPU 32, a memory 34, a communication device 36, an input device, a display device, and the like. The memory 34 of the computer 30 includes an HDD, a ROM, and a RAM. A program for controlling the operation of the estimation apparatus 10 is stored in the HDD or ROM, and the CPU 32 executes processing according to this program. The RAM is used as a work area or buffer area for the CPU 32. The communication device 36 is for communicating with an external computer (such as the position detection system 12 or the service providing device 14) wirelessly or via a wire such as a wireless LAN or the Internet.

  A discriminant DB 38 is connected to the computer 30 of the estimation apparatus 10. The discriminant DB 38 may be provided outside the computer 30 as shown in FIG. 1, or may be provided in an HDD or the like in the computer 30, or other network on the network that can communicate with the computer 30. It may be provided in a computer.

  The discriminant DB 38 stores a discriminant for estimating the objective behavior performed by the human 16 existing in the environment based on the feature amount calculated from the position history data of the human 16 and the object 18. When estimating the objective behavior of the human 16, a feature amount described later is appropriately calculated according to the discriminant stored in the discriminant DB 38. A specific example of this discriminant will be described later.

  In the providing system 100 having such a configuration, as described above, the position detection system 12 detects data relating to the position coordinates or position history of the person 16 and the object 18 existing in the environment. Then, a feature amount is calculated from the position history data by the estimation device 10, and the objective action of the human 16 is estimated based on the feature amount.

  The feature quantity calculated from the position history data by the estimation device 10 includes a feature quantity (first feature quantity) between the human 16 and the object 18 and a feature quantity (second feature quantity) related to the movement of the human 16. And a feature amount (third feature amount) related to the movement of the moving body 18b. As for the installation object 18a, the position coordinates are always fixed, that is, the position history data always indicates the same position coordinates, so that the feature quantity relating to the movement of the installation object 18a is not calculated. The position history data (position coordinates) of the installation object 18a is used for calculating the first feature amount.

  The feature amount between the human 16 and the object 18 includes the distance between the position coordinate of the human 16 and the position coordinate of the object 18, the relative speed of the human 16 with respect to the object 18, or the relative speed of the object 18 with respect to the human 16. The speed, the angle formed by the line connecting the person 16 and the object 18 and the traveling direction of the person 16 or the object 18, the angle formed by the traveling direction of the person 16 and the traveling direction of the object 18, etc. is there.

  Further, the feature amount related to the movement of the human 16 or the moving body 18b includes, for example, a feature amount related to distance, speed, angle, and dispersion. Referring to FIG. 4, the feature quantity related to the distance is the Euclidean distance, the x-direction distance, the y-direction distance, and the like between the 0th point and the kth point. The feature quantity related to speed is the speed between the 0th point and the nth point, for example, the speed between the kth point and the k + 1st point is obtained from k = 0 to k = n. Mean and variance (variation of values). The feature quantity related to the angle is an angle formed by the 0th point, the kth point, and the nth point. The feature amount related to the variance includes the variance of the x coordinate and the variance of the y coordinate from the 0th point to the kth point.

  Of course, feature quantities other than these can be calculated and used to estimate the objective behavior of the human 16. In addition to the feature amount calculated from the position history data, the objective behavior can be estimated by referring to other data. For example, when the object 18 is a communication robot, when the communication robot is performing a certain motion, the objective action of the human 16 may be estimated with reference to the motion being performed.

  In this embodiment, the above-described feature amount is appropriately used to “close to the object 18”, “go away from the object 18”, “follow the object 18”, “wait for the object 18”. The objective action (or the objective state or objective state of the human 16) performed by the human 16 such as “having” and “interacting with the object 18” is estimated. Hereinafter, with reference to FIG. 5 to FIG. 9, a method for estimating the objective behavior of the human 16 will be described.

  As shown in FIG. 5, the state in which the human 16 is “approaching the object 18” or “chasing the object 18” is in a certain time (for example, from time t0 to tn) while performing the action. It is considered that the object 18 exists in the moving direction of the person 16 and the distance between the person 16 and the object 18 is shortened in a few seconds to several tens of seconds). Therefore, the angle formed by the line α connecting the person 16 and the object 18 and the traveling direction β of the person 16 is smaller than the predetermined angle (DegA_TH), and the reduction amount of the distance between the person 16 and the object 18 is a predetermined amount ( When it is greater than (Decrese_TH), it can be determined that the human 16 is approaching the object 18 or is chasing the object 18. In this case, the discriminant DB 38 includes, for example, a discriminant representing “the angle between the line connecting the human and the object and the human traveling direction <DegA_TH and the amount of decrease in the distance between the human and the object> Decrese_TH”. An expression is described.

  As shown in FIG. 6, the state in which the human 16 is “away from the object 18” has no object 18 in the moving direction of the human 16 for a certain period of time during the action, In addition, it is considered that the distance between the human 16 and the object 18 is increasing. Therefore, the angle formed by the line α connecting the person 16 and the object 18 and the traveling direction β of the person 16 is smaller than a predetermined angle (DegL_TH), and the increase amount of the distance between the person 16 and the object 18 is a predetermined amount ( If it is greater than (Increse_TH), it can be determined that the person 16 is moving away from the object 18. In this case, the discriminant DB 38 includes, for example, a discriminant representing “the angle of the angle formed by the line connecting the human and the object and the human traveling direction> DegL_TH and the increase in the distance between the human and the object> Increse_TH”. An expression is described.

  As shown in FIG. 7, the state in which the human 16 “follows the object 18” has the same movement direction of the human 16 and the movement direction of the object 18 for a certain period of time while performing the action. It is considered that the distance between the human 16 and the object 18 does not change. Therefore, the angle formed by the moving direction β of the human 16 and the moving direction γ of the target 18 is smaller than the predetermined angle (DegF_TH), and the change amount of the distance δ between the human 16 and the target 18 is a predetermined amount (DistF_TH). When it is smaller than that, it can be determined that the human 16 is talking about the object 18. In this case, the discriminant DB 38 includes, for example, a discriminant representing “angle angle between the moving direction of the human and the moving direction of the object <DegF_TH and a change amount of the distance between the human and the object <DistF_TH”. Described.

  As shown in FIG. 8, the state in which the human 16 is “waiting for the object 18” has the human 16 in the moving direction of the object 18 for a certain period of time while performing the action, and It is considered that the distance between the human 16 and the object 18 is decreasing and the position of the human 16 does not change. Therefore, the angle formed by the line α connecting the person 16 and the object 18 and the traveling direction γ of the object 18 is smaller than a predetermined angle (DegW_TH), and the amount of decrease in the distance between the person 16 and the object 18 is a predetermined amount. When it is larger than (Decrese_TH) and the movement amount of the human is smaller than the predetermined amount (DistI_TH), it can be determined that the human 16 is waiting for the object 18. In this case, the discriminant DB 38 includes, for example, “the angle between the line connecting the person and the object and the traveling direction of the object <DegW_TH, and the decrease amount of the distance between the person and the object> Decrese_TH, and the person A discriminant representing the amount of movement <DistI_TH ”is described.

  As shown in FIG. 9, the state in which the human 16 is “interacting with the object 18” does not change the position of the human 16 for a certain period of time while performing the action, and the human 16 is the target. It is considered that the object 18 faces the direction of the object 18 and that the operation state of the object 18 is in a conversation state. Therefore, the movement amount of the human 16 is smaller than a predetermined amount (DistI_TH), and the angle formed by the line α connecting the human 16 and the object 18 and the body direction ε of the human 16 is smaller than the predetermined angle (DegB_TH). When the operation state of the object 16 is in a conversation, it can be determined that the person 16 is interacting with the object 18. In this case, the discriminant DB 38 includes, for example, “the amount of human movement <DistI_TH, the angle between the line connecting the human and the object and the direction of the human body <DegB_TH, and the operation state of the object” A discriminant representing “in dialogue” is described.

  It should be noted that the above-described objective action of human 16 and its discriminant are merely examples, and the estimation device 10 (providing system 100) may estimate other objective actions or use other discriminants. The objective behavior may be estimated. Further, it is not limited to estimating the objective behavior of the human 16 by the discriminant defined using the threshold value as described above. For example, a known learning algorithm such as a support vector machine (SVM) or a neural network (NN) The objective behavior of the human 16 can also be estimated using the learning data. Also in this case, the above-described feature amount calculated from the position history data of the person 16 and the object 18 is appropriately used.

  As described above, the position history of the human 16 and the object 18 is detected, and the feature amount relating to the mutual relationship between the human 16 and the object 18, the feature amount relating to the movement of the human 16 and the movement of the object 18 (the moving body 18 b). By calculating the feature values and appropriately giving these feature values to the discriminant, it is possible to accurately estimate the objective action performed by the human 16 existing in the environment. Note that the objective behavior can be estimated using only the feature amount related to the relationship between the human 16 and the object 18, but at the same time, the feature amount related to the movement of the human 16 and the feature amount related to the movement of the target object 18. Alternatively, by using other data, more types of objective behavior can be estimated more accurately.

  In addition, since the estimation of the objective action of the human 16 can be executed only by detecting the position history of the human 16 and the object 18 for a short time of several seconds to several tens of seconds, the human 16 can respond quickly to the action of the human 16. Service can be provided.

  For example, the robot may be arranged in the environment as the service providing device 14 and the robot may be provided with a service according to the estimated human 16 object behavior. As the robot, an interaction-oriented robot (communication robot) having a function of executing communication using voice and body movement may be used. For example, Robbie (registered trademark) developed by the present applicant can be used. In addition, for example, a terminal such as a mobile phone or a PDA can be held by the person 16 as the service providing apparatus 14 and information corresponding to the estimated objective behavior of the person 16 can be presented to the terminal.

  Specifically, when the human 16 is approaching a guidance map or an exhibition shelf (installed object 18a) installed in the environment, the robot is moved to the place of the installed object 18a when the estimation is performed. It is recommended to explain how to read the guide map or explain the contents of the exhibits displayed on the display shelf. For example, if the robot is moved after the person 16 arrives at the place of the installation object 18a, it takes time for the robot to approach the person 16 and service provision is delayed. Further, if the robot is always present at the place of the installation object 18a, the robot cannot provide other services. However, as in this embodiment, by estimating (predicting) the objective behavior of the human 16 and controlling the robot accordingly, the time until the robot approaches the human 16 can be shortened, and the service can be performed more quickly. Can be provided. Also, when the robot's hand is free, other services can be provided to the human 16 so that the robot can work efficiently.

  Also, when the robot is waiting at a certain point and the human 16 is approaching, when the robot is moved toward the human 16 or when the robot is traveling around the environment, When the human 16 is chasing, it is good to stop the robot and make the human 16 wait. Furthermore, when the robot is moving toward the human 16 to provide a certain service, if the human 16 is waiting for the robot, the robot may approach the human 16 as it is. On the other hand, when the robot is moving toward the human 16, if the human 16 is moving away from the robot, it may be determined that no service is necessary and the movement is interrupted.

  Thus, by estimating what action the human 16 is performing on the object 18, it becomes possible to control the robot according to the situation at the site, and to eliminate the waste of movement of the robot and the human 16. And more advanced services can be provided more quickly. In this case, the robot functions as the service providing device 14 and also functions as the object 18.

  Below, an example of operation | movement of the above estimation apparatuses 10 is demonstrated using a flowchart. Specifically, the CPU 32 of the computer 30 of the estimation device 10 executes a process (behavior estimation process) for estimating the action performed by the human 16 on the specific object 18 according to the flowchart shown in FIG. . As shown in FIG. 10, the CPU 32 of the computer 30 determines whether or not the person 16 and the object 18 exist in the environment in step S1. For example, the identification information of the wireless ID tag 26 existing in the environment is acquired from the position detection system 12 via the communication device 36, and it is determined whether the human 16 and the object 18 exist in the environment. If “NO” in the step S1, that is, if the human 16 and the object 18 do not exist in the environment, the action estimation process is ended as it is. On the other hand, if “YES” in the step S1, that is, if the person 16 and the object 18 exist in the environment, the process proceeds to a step S3.

  In step S3, a variable i indicating the ID number of the person 16 and a variable k indicating the ID number of the object 18 are initialized (i = 0, k = 0). For example, when the presence of a plurality of persons 16 and a plurality of objects 18 is confirmed in step S1, the ID number i of each person 16 is set to 0 to L, and the ID number k of each object 18 is set to 0 to M. , I = 0 and k = 0.

  In a succeeding step S5, it is determined whether i = L. That is, it is determined whether or not the objective behavior has been estimated for all humans 16 existing in the environment. When “YES” in the step S5, that is, when the behavior estimation is performed for all the humans 16 existing in the environment, the behavior estimation processing is ended. On the other hand, if “NO” in the step S5, that is, if there is a person 16 who has not performed behavior estimation in the environment, the process proceeds to a step S7.

  In step S7, a position history for a predetermined time of the human 16 is acquired. Specifically, the position history of the person 16 is acquired from the position detection system 12 via the communication device 36 for several seconds to several tens of seconds from time t0 to tn, for example. At this time, the position coordinates when the time of the person 16 whose ID number is i (that is, the person i) is j is expressed by the expression Traj_ij (i = 0... L, j = t0... Tn), and is shown in FIG. As described above, the position history data of each person 16 is represented by n + 1 point sequences on the xy plane including time information.

  In the subsequent step S9, a second feature value, that is, a feature value related to the movement of the person 16 is calculated from the position history data of the person 16 acquired in step S7. For example, a feature amount related to a human moving distance, speed, angle, variance, and the like is calculated. Note that the second feature amount calculated in step S9 is appropriately determined according to the discriminant stored in the discriminant DB 38.

  In step S11, it is determined whether k = M. That is, it is determined whether or not the objective action of the person 16 for all the objects 18 existing in the environment has been estimated. If “YES” in the step S11, that is, if the action of the human 16 is estimated for all the objects 18 existing in the environment, the process proceeds to a step S13, the variable i is incremented, and the process returns to the step S5. On the other hand, if “NO” in the step S11, that is, if the action of the human 16 is not estimated for all the objects 18 existing in the environment, the process proceeds to a step S15.

  In step S15, a position history for a predetermined time of the object 18 is acquired. Specifically, the position history of the object 18 is acquired from the position detection system 12 through the communication device 36 for several seconds to several tens of seconds from time t0 to tn, for example. At this time, when the object 18 is the moving body 18b, the position coordinates when the time of the object 18 (object k) whose ID number is k is j are expressed by the formula Traj_kj (k = 0... M, j = t0... tn). Like the human 16, the position history data of the object 18 is represented by n + 1 point sequences on the xy plane including time information. When the object 18 is the installation object 18a, the position coordinates are the same at any time, and therefore the position history data of the object 18 is represented by one point on the xy plane.

  In the subsequent step S17, a third feature value, that is, a feature value related to the movement of the moving body 18b is calculated from the position history data of the object 18 acquired in step S15. The third feature amount calculated in step S17 is also appropriately determined according to the discriminant stored in the discriminant DB 38. When the object 18 is the installation object 18a, this step S17 is not executed and the process proceeds to step S19 as it is.

  In step S19, a first feature value, that is, a feature value related to the relationship between the person 16 and the object 18 is calculated from the position history of the person 16 and the object 18 acquired in steps S7 and S15. For example, the distance between the position coordinates of the person 16 and the position coordinates of the object 18, the angle between the line connecting the person 16 and the object 18 and the traveling direction of the person 16 or the object 18, and the like are calculated. The first feature amount calculated in step S19 is also appropriately determined according to the discriminant stored in the discriminant DB 38.

  In subsequent step S21, the first to third feature values calculated in steps S9, S17, and S19 are given to the discriminant read from the discriminant DB 38, and the action that the person 16 is performing on the object 18 is estimated. . For example, when the angle between the line connecting the person 16 and the object 18 and the traveling direction of the person 16 <DegA_TH and the amount of decrease in the distance between the person 16 and the object 18> Decrese_TH, the person 16 It is determined that the object 18 is approaching.

  In step S23, the behavior pattern of the human 16 is set to the objective behavior calculated in step S21, for example, “approaching the object”, and in step S25, the variable k is incremented and the process returns to step S11.

  The computer 30 or another computer gives an instruction to the service providing apparatus 14 such as a communication robot or a mobile phone according to the action pattern of the person 16 set in step S23. That is, a control instruction process for giving a control instruction to the service providing apparatus 14 is executed, and a service suitable for the objective action of the person 16 is provided.

  According to this embodiment, the position history of the person 16 and the object 18 is detected, the feature amount relating to the mutual relationship between the person 16 and the object 18, the feature amount relating to the movement of the person 16, and the object 18 (moving body 18b). By giving the discriminant a feature amount related to the movement of the object 16, it is possible to estimate the action performed by the human 16 on the object 18 in a short time. Therefore, a service corresponding to the objective behavior of the person 16 can be provided quickly.

  In the above-described embodiment, the position detection system 12 and the estimation device 10 perform different processing (position detection processing, behavior estimation processing, control instruction processing, and the like) using different computers 20 and 30. However, one computer can perform these processes in a lump, and three or more computers can further share the processes.

DESCRIPTION OF SYMBOLS 10 ... Objective action estimation apparatus 12 ... Position detection system 14 ... Service provision apparatus 16 ... Human 18, 18a, 18b ... Object 20 ... Position detection system computer 22 ... Laser range finder 24 ... Wireless ID tag reader 26 ... Wireless ID tag 30 ... computer of objective behavior estimation device 38 ... discriminant DB
100 ... service providing system

Claims (4)

An objective behavior estimation device that estimates how a human being is acting on a specific object,
The object is a movable body that is movable in the environment,
Position history acquisition means for acquiring a position history for a predetermined time of the human and the object;
First feature amount calculating means for calculating a first feature amount indicating a feature amount between the person and the object from the position history acquired by the position history acquiring means;
From the position history acquired by the position history acquisition means, third feature amount calculation means for calculating a third feature amount indicating a feature amount related to the movement of the object, and the first feature amount calculated by the first feature amount calculation means. 1. An objective behavior estimation apparatus comprising behavior estimation means for estimating behavior performed by the human with respect to the object using one feature value and the third feature value calculated by the third feature value calculation means . A second feature amount calculating means for calculating a second feature amount indicating a feature amount related to the movement of the person from the position history acquired by the position history acquiring means;
The objective behavior estimation apparatus according to claim 1, wherein the behavior estimation unit further estimates the behavior performed by the human with respect to the object using the second feature amount. A service providing system comprising: the objective behavior estimation device according to claim 1 or 2 ; and service provision means for providing a service according to a human behavior estimated by the objective behavior estimation device. The service providing system according to claim 3 , wherein the service providing unit includes a communication robot having a function of executing communication using voice and body movement.

Images