JP7247560B2 - Robot, robot control method and program - Google Patents

Description

The present invention relates to a robot, a robot control method, and a program.

2. Description of the Related Art Conventionally, a robot disclosed in Patent Document 1, for example, is known as a robot configured to be able to execute each of a plurality of predetermined actions. In this robot, a program for controlling a plurality of predetermined motions of the robot is set, and a probabilistic automaton is used to generate a plurality of transition probabilities corresponding to each of the plurality of predetermined motions. One action is selected from the predetermined actions, and the selected action is executed by the machine itself. In addition, in this robot, sensors such as a pressure sensor and a visual sensor detect external input signals such as "hit" and "annoyed", and based on these external input signals, the robot's emotion parameters are set, The above transition probabilities can also be changed based on the set emotion parameters.

JP 2010-149276 A

In conventional robot control, such as the robot motion control described in Patent Document 1, transition probabilities used to select the type of robot motion are often predetermined. In addition, even if the transition probability is changeable, it is only changed based on the emotion parameter or the like. Therefore, among the plurality of predetermined motions of the robot, the execution frequency of some of the robot's predetermined motions with low transition probabilities becomes extremely low. , thereby reducing the diversity of robot motions.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and an object of the present invention is to increase the variety of motions of a robot.

In order to achieve the above object, the robot of the present invention
A robot configured to be able to execute each of a plurality of predetermined actions,
acquisition means for acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
selection means for selecting an action corresponding to a maximum required action value among the action required values set for each of the predetermined actions, from the plurality of predetermined actions as an action to be executed;
an operation control means for causing the machine to execute the action for execution selected by the selection means;
After the action for execution is executed by the action control means, among the action request values set for each of the predetermined actions, the execution frequency represented by the acquired execution frequency parameter is a predetermined action lower than the other predetermined actions. selection control means for controlling the selection means so that a predetermined action having a lower execution frequency than other predetermined actions is more likely to be selected as the action for execution by increasing an action request value corresponding to the action;
characterized by comprising

ADVANTAGE OF THE INVENTION According to this invention, the diversity of operation|movement of a robot can be improved.

It is a figure which shows the external appearance of the robot which concerns on Embodiment 1 of this invention. 2 is a diagram showing the functional configuration of the robot according to Embodiment 1; FIG. 4 is a diagram showing an example of an action request value table stored in a storage unit of the robot according to Embodiment 1; FIG. It is a figure which shows an example of the change of an action required value. 4 is a flowchart of operation control processing according to the first embodiment; 10 is a flowchart of a behavioral principle value update thread according to Modification 1; 10 is a flowchart of an action request value update thread according to Modification 1; 10 is a flowchart of an operation control thread according to Modification 1; FIG. 11 is a diagram showing an example of an action request value table stored in a storage unit of a robot according to Modification 2; 10 is a flowchart of operation control processing according to modification 2;

Hereinafter, embodiments of the present invention will be described with reference to the drawings. The same reference numerals are given to the same or corresponding parts in the drawings.

(Embodiment 1)
The robot 100 according to the first embodiment of the present invention is an electronic pet type robot that can autonomously perform each of a plurality of predetermined actions such as wandering around, resting around, and dozing off. As shown in FIG. 1, the robot 100 has, for example, the shape of a cute animal, with a head 10 having a nose 11, ears 12, a mouth 13, and eyes 14, and a body 20 having front legs 21, rear legs 22, A tail 23 is provided respectively. The head 10 and the body 20 are interconnected by a neck joint 15 indicated by broken lines.

Neck joint 15 includes a plurality of motors. A controller 110, which will be described later, controls these motors, so that the head 10 of the robot 100 can be rotated about a vertical axis, a horizontal axis, and a longitudinal axis. This allows the robot 100 to perform, for example, a nodding motion.

Although not shown, the robot 100 has a driving mechanism for changing the facial expression of the head 10 and moving the front legs 21, the hind legs 22, and the tail 23. The control unit 110 controls the driving mechanism. Emotions can be expressed through control. For example, the robot 100 can express its happiness by making a smile on its face, flapping its front legs 21 and hind legs 22 back and forth, and wagging its tail 23 .

In addition, as shown in FIG. 2, the robot 100 includes a control unit 110, a storage unit 120, an image input unit 131, a voice input unit 132, a sensor unit 133, a sound output unit 134, an emotion expression unit 135, a driving unit 135, and a driving unit 135. a portion 136;

The control unit 110 is configured by a CPU (Central Processing Unit) and the like. By executing programs stored in the storage unit 120, the control unit 110 functions as an acquisition unit 111, a setting unit 112, a selection unit 113, and an operation control unit 114, which will be described later. The control unit 110 also has a clock function and a timer function, and can acquire the current time and elapsed time.

The storage unit 120 includes a ROM (Read Only Memory), a RAM (Random Access Memory), and the like, and stores programs executed by the CPU of the control unit 110, various data, and the like. The storage unit 120 also stores an action request value table 121 as shown in FIG. Details of the action request value table 121 will be described later.

The image input unit 131 is an imaging device (camera) provided at the position of the nose 11 in the center of the front surface of the head 10 . The image input unit 131 acquires image data of the front of the robot 100 and inputs it to the control unit 110 .

The voice input unit 132 is composed of a plurality of microphones provided at the positions of the left and right ears 12 of the head 10, and constitutes an array microphone. The voice input unit 132 acquires the voice uttered by the user around the robot 100 as voice data and inputs the voice data to the control unit 110 .

The sensor unit 133 includes various sensors such as a human sensor, a temperature sensor, and a tactile sensor. The robot 100 has human sensors and temperature sensors at the positions of the eyes 14 of the head 10, and can acquire information about the surrounding environment. In addition, tactile sensors are provided at various positions on the surface of the head 10 and the body 20, and it is possible to detect being stroked, hit, or held by the user.

The sound output unit 134 is a speaker provided at the position of the mouth 13 of the head 10, and is controlled by the control unit 110 to output sounds, barks, and the like.

The emotion expression unit 135 has a driving mechanism such as a motor, and expresses the emotion of the robot 100 by changing the facial expression of the head 10 and moving the front legs 21, the rear legs 22, and the tail 23.

Drive unit 136 includes four wheels and a drive motor. Of the four wheels, two are arranged on the front side of the body 20 as front wheels, and the remaining two are arranged on the rear side of the body 20 as rear wheels. The front wheels may be arranged on the front legs 21 and the rear wheels on the rear legs 22, or four wheels may be arranged at positions different from the front legs 21 and the rear legs 22 (for example, directly under the body 20). . As the wheel, for example, an omni wheel, a mecanum wheel, or the like may be used. When the drive unit 136 is controlled by the control unit 110 to rotate the wheels, the robot 100 moves. In addition, the drive unit 136 includes a motor that moves the neck joint 15 , and the robot 100 nods, looks around restlessly, and sleeps by controlling the drive unit 136 with the control unit 110 . You can do the action you want.

Further, although not shown, the robot 100 may be provided with an operation button at a position corresponding to the back of the body 20, for example. The operation buttons are various buttons for operating the robot 100, and include a power button, a speaker volume control button, and the like.

Next, each function realized by the control unit 110 will be described. As described above, the control unit 110 functions as an acquisition unit 111, a setting unit 112, a selection unit 113, and an operation control unit 114 by executing programs stored in the storage unit 120. FIG. In addition, the control unit 110 supports a multithread function, and can execute multiple threads (different processing flows) in parallel.

The acquiring unit 111 acquires an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions executed by the robot 100 . Here, the predetermined action means that the robot 100 is acted upon by an external stimulus (a stimulus such as “stroked”, “struck”, “held”, etc. detected by the tactile sensor provided in the sensor unit 133). These are various actions performed by the robot 100 when the robot 100 is not in motion, such as "walking around", "looking around", "crying", "dejected", "crying", "drowsy", "sleep", and the like. . Further, the execution frequency parameter is a parameter representing the actual execution frequency of each predetermined operation. ), the number of executions (the number of executions: a small number of executions indicates a low execution frequency), the duration of the state in which the predetermined action is not executed (a long duration indicates that the execution frequency is low). low), etc. Acquisition unit 111 functions as an acquisition unit.

The setting unit 112 sets a selection parameter used when selecting a motion to be executed by the robot 100 (hereinafter referred to as an “execution motion”) from among a plurality of predetermined motions, based on the execution frequency parameter acquired by the acquisition unit 111. set. At this time, the setting unit 112 sets the selection parameter so that the selection unit 113 is likely to select the predetermined action with the lower execution frequency represented by the execution frequency parameter as the execution action. Here, the parameter for selection is a parameter used as a reference when selecting the predetermined action for each of the predetermined actions. Various parameters can be considered as the selection parameter, but in this embodiment, an action request value, which will be described later, is used as the selection parameter. The setting unit 112 functions as setting means.

The selection unit 113 selects a motion for execution of the robot 100 from a plurality of predetermined motions based on the selection parameter set by the setting unit 112 . Since the selection parameter is set based on the execution frequency parameter, the selection unit 113 selects the execution action from among a plurality of predetermined actions according to the execution frequency parameter acquired by the acquisition unit 111. Become. The selection unit 113 functions as selection means.

The motion control unit 114 controls the sound output unit 134 , the emotion expression unit 135 , and the drive unit 136 to cause the robot 100 to perform the execution motion (predetermined motion) selected by the selection unit 113 . The motion control unit 114 functions as motion control means.

Next, the action request value table 121 stored in the storage unit 120 will be described with reference to FIG. As shown in FIG. 3, a plurality of predetermined actions ("walking around", "looking around", etc.) executable by the robot 100 are based on a plurality of action principles ("boring", "lonely", "sleepy", etc.). assigned to belong to either Then, as shown in FIG. 3, the action request value of each predetermined action (a value indicating the strength of the request to perform the predetermined action) is the action principle value of the action principle to which the predetermined action belongs (time and its action principle). Various values are set depending on whether or not the operation is executed). The action request value table 121 is such a table in which the action request value for each predetermined action is determined for each action principle value for each action principle. In the robot 100 , default values for these action request values are set in advance in the action request value table 121 .

When selecting a motion for execution of the robot 100, the selection unit 113 refers to the action request value table 121 and selects the action request value of each predetermined motion based on the action principle value of each action principle at that time. Then, the predetermined action with the highest action requirement value is selected as the action for execution. Then, the motion control unit 114 controls the sound output unit 134, the emotion expression unit 135, and the driving unit 136 so that the predetermined motion selected by the selection unit 113 as the execution motion is executed. The predetermined operation (execution operation) is executed. Note that, for example, a predetermined value (for example, 1) is added to the action principle value every predetermined time, and when a predetermined action belonging to the action principle corresponding to the action principle value is executed, the predetermined value (for example, 1) is added. ) is subtracted.

For example, in the action request value table shown in FIG. 3, the action principle value of the action principle "boredom" is -2, the action principle value of the action principle "lonely" is -1, and the action principle of "sleepy". Let the value be 1. In this case, "hang around" (action request value 20) is selected as the predetermined action for the action principle "boredom", and "depressed" (action request value 20) is selected as the predetermined action for the action principle "lonely". , "drowsiness" (behavior request value 40) is selected as the predetermined action of the action principle "sleepy". Then, the selecting unit 113 finally selects the predetermined action with the highest action request value, "to fall asleep", from among these three selected predetermined actions, and the robot 100 selects "to fall asleep" as an execution action. ” will perform the action.

When the robot 100 performs the action of "drowsiness", the behavior principle value of "sleepy", which is the behavioral principle to which "drowsiness" belongs, is reduced (for example, by subtracting 1 to 0). Then, the action principle value of "boredom" or "lonely", which is the action principle to which the predetermined action that was not performed is increased (for example, by adding 1, the action principle value of "boredom" becomes -1, and the action principle value of "lonely" ” is set to 0). Then, at the time of the next selection, "cry" (action required value 50) is selected as the predetermined action of the action principle "boring", and "be depressed" (action required value 30) is selected as the prescribed action of the action principle "lonely". is selected, and "drowsiness" (action request value 20) is selected as the predetermined action of the action principle "sleepy". Therefore, this time, the selection unit 113 finally selects the predetermined action "cry" with the highest action requirement value among the three selected predetermined actions, and the robot 100 "cries". will take action.

Note that when there are a plurality of predetermined actions having the highest action requirement value, the selection unit 113 randomly selects one predetermined action from among the plurality of predetermined actions as an execution action.

By using the action request value table 121 and action principle values that are increased or decreased as appropriate, the robot 100 can perform predetermined actions belonging to various action principles, as described above. However, with respect to a plurality of predetermined actions that belong to the same action principle, there is a possibility that the execution frequency will be uneven. For example, among the predetermined actions belonging to the action principle "boredom", "hang around" is selected only when the action principle value of "boredom" is -2. Further, when the user adds a new predetermined action, the action request value of the predetermined action can be set to any value. It may become

Therefore, the setting unit 112 dynamically increases the action request value of the predetermined action that has not been executed. Then, when the predetermined action is finally executed as a result of increasing the action request value, the setting unit 112 restores the action request value of the executed predetermined action to the original value (to the default value).

A specific example will be described with reference to FIG. First, assuming that the action principle value of the action principle “boredom” is −1, the selection unit 113 selects “cry” with the highest action request value (action request value is 50) as the predetermined action of the action principle “boredom”. select. Here, it is assumed that the action requirement value belonging to another action principle is less than 50. Then, the selection unit 113 finally selects “crying” as the predetermined operation, and the operation control unit 114 performs the “crying” operation. Then, the setting unit 112 increases the action request values other than "crying" ("walking around" and "looking around") by a predetermined value (for example, 10). Also, since the action "crying", which is a predetermined action belonging to the action principle "boredom", has been performed, the action principle value of the action principle "boredom" is decreased (here, it is assumed to be -2).

Then, the action requirement values belonging to the action principle "boredom" are shown in the table in the middle of FIG. shown in bold). Assuming that the action principle value is -2, the selection unit 113 selects the predetermined action "hanging around" (action demand value is 30) with the highest action demand value as the prescribed action of the action principle "boredom". Here, it is assumed that the action requirement value belonging to another action principle is less than 30. Then, the selection unit 113 finally selects the predetermined action “wander around” as the action for execution, and the action control unit 114 performs the predetermined action “wander around”. Then, the setting unit 112 increases the action request value of the predetermined action (“look around” and “cry”) other than “walk around” by a predetermined value (for example, 10), and sets the action of the predetermined action “walk around”. Revert request values to default values. In addition, since the predetermined action "hanging around" belonging to the behavior principle "boredom" has been performed, the behavior principle value of the behavior principle "boredom" is decreased (here, it is assumed to be -3).

Then, the action demand values belonging to the action principle "boredom" are shown in the lower table of FIG. ). Assuming that the action principle value is -3, the selection unit 113 selects the predetermined action "look around" (action demand value is 30) with the highest action demand value as the action principle "boredom" prescribed action. Here, it is assumed that the action requirement value belonging to another action principle is less than 30. Then, the selection unit 113 finally selects the predetermined action “look around” as the action for execution, and the action control unit 114 performs the action “look around”.

In this way, even if there is a predetermined action whose execution frequency is extremely low with the values set in the action request value table 121 as they are, the setting unit 112 can dynamically increase the action request value of this predetermined action. As a result, the action requirement value of this predetermined action eventually becomes the highest, and as a result, this predetermined action is selected by the selection unit 113 as the action for execution. Further, even if there is a predetermined action whose execution frequency is extremely high, the action request value of this predetermined action will eventually become smaller than the action request value of the predetermined action that was not executed, and this frequently executed predetermined action is not selected by the selection unit 113 as an action for execution. Therefore, by dynamically changing the action request value in this way, it is possible to increase the diversity of actions even for a plurality of predetermined actions belonging to the same action principle.

Now, the motion control process for increasing the variety of motions of the robot 100 in this way will be described with reference to FIG. Execution of this process is started when the robot 100 is activated. It is also assumed that the action principle value of each action principle is initialized (to 0, for example) when the robot 100 is activated.

First, the control unit 110 resets the value T of the timer to 0 (step S101). Then, the control unit 110 determines whether or not there is an external stimulus (stroked, beaten, held, etc.) based on the detected value by the tactile sensor provided in the sensor unit 133 (step S102).

If there is an external stimulus (step S102; Yes), the motion control unit 114 causes the sound output unit 134 and the emotion expression unit 135 to perform a motion (a motion not included in the predetermined motion described above) in response to the external stimulus. and control the drive unit 136 (step S103), and return to step S101. The motion controlled by the motion control unit 114 in step S103 is a motion that is not the predetermined motion described above. For example, the expression unit 135 makes the robot 100 have a surprised expression, and the sound output unit 134 makes the robot 100 say “can”.

If there is no external stimulus (step S102; No), the controller 110 determines whether or not a predetermined period of time (for example, one minute) has elapsed using the timer T (step S104). If the predetermined time has not elapsed (step S104; No), the process returns to step S102.

If the predetermined time has passed (step S104; Yes), the obtaining unit 111 obtains an execution frequency parameter representing the actual execution frequency of each predetermined action (step S105). As the execution frequency parameter, for example, for each predetermined action, the elapsed time after the predetermined action was last performed (the duration of the state in which the predetermined action is not executed: the longer the duration, the lower the execution frequency. ), the number of executions of each predetermined action (for example, the cumulative number of executions of each predetermined action from a reference time (for example, the start time of the robot 100): the smaller the number of executions, the lower the execution frequency), The execution time of each predetermined action (for example, the cumulative execution time of each predetermined action from a reference time (for example, the start time of the robot 100): the shorter the execution time, the lower the execution frequency). As the execution frequency parameter, simply whether or not each predetermined action is actually executed (if executed, the execution frequency is high; if not executed, the execution frequency is low) may be used. For the sake of simplicity, the following description assumes that the execution frequency parameter is the presence or absence of actual execution.

Then, according to the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, the setting unit 112 determines that the execution frequency of all the predetermined actions belonging to the action principle is less than the reference value (all the predetermined actions are executed). The action principle value of the action principle is increased by a predetermined value (for example, 1) (step S106). Since there is an upper limit to the action principle value based on the size of the action request value table 121, the action principle value that has reached the upper limit is left at the upper limit value without being increased in step S106.

Next, according to the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, the setting unit 112 determines the action to which the predetermined action whose execution frequency is equal to or higher than the reference value (for example, executed for a certain period of time or a certain number of times) belongs to. The action principle value of the principle is decreased by a predetermined value (for example, 1) (step S107). Since there is a lower limit to the action principle value based on the size of the action request value table 121, the action principle value that has reached the lower limit is left at the lower limit value without being reduced in step S107. Note that in step S106, the behavior principle values of all behavior principles may be increased by a predetermined value (for example, 1) regardless of the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, but in this case In step S107, the action principle value of the action principle to which the predetermined action whose execution frequency is equal to or higher than the reference value belongs must be decreased by a value (for example, 2) to cancel the increase in step S106.

Then, the selection unit 113 selects the predetermined action with the largest action request value as the action for execution (step S108). If there are a plurality of predetermined motions with the maximum required action value, the selection unit 113 randomly selects one predetermined motion from among the predetermined motions with the maximum required action value as an action to be executed. Step S108 is also called a selection step.

Then, the motion control unit 114 controls the sound output unit 134, the emotion expression unit 135, and the driving unit 136 so as to execute the motion for execution selected by the selection unit 113 (step S109). Step S109 is also called an operation control step.

Next, the setting unit 112 determines that the predetermined action belonging to the action principle to which the predetermined action selected as the action to be performed by the selection unit 113 in step S108 belongs is not selected as the action to be performed by the selection unit 113 (the action that was not performed). The action request value of the predetermined action (which was not performed) is increased by a predetermined value (for example, 10) (step S110). The action request value may be increased without setting an upper limit. Then, the setting unit 112 sets the action request value of the predetermined action selected (executed) as the execution action by the selection unit 113 in step S108 to the default value (the value initially set in the action request value table 121). ) (step S111) and returns to step S101. Steps S110 and S111 are also called setting steps.

In steps S110 and S111, instead of the above process, the acquisition unit 111 acquires information on the actual execution state of the predetermined motion of the robot 100 (the duration of the state in which each predetermined motion is not executed, each predetermined At least one of information such as the execution time of the action and the number of executions of each predetermined action) is acquired as an execution frequency parameter, and the action request value is increased or returned to the default value according to the acquired execution frequency parameter. You can For example, if the duration of the non-execution state exceeds the continuation reference time, the action request value of the predetermined action is increased, and if the execution time exceeds the execution reference time or the number of executions exceeds the execution reference number of times, the predetermined action 's call-to-action values to their default values. In the processing in this case, the step in which the acquisition unit 111 acquires the execution frequency parameter is also called an acquisition step.

By performing the motion control processing described above, the robot 100 can increase the action request value of a predetermined motion for which the duration of a non-executed state exceeds the continuation reference time, thereby allowing the robot 100 to perform a predetermined motion with a low execution frequency. It can be made easy to be selected as an action. As a result, the robot 100 can perform predetermined motions based on various behavioral principles as described above, thereby increasing the diversity of motions. Then, when the user wishes to add a new predetermined action, the user simply assigns the predetermined action to be added under the relevant action principle in the action request value table 121 and sets the action request value to an arbitrary value. The robot 100 will surely execute the added predetermined motion. In other words, the user can easily add a new predetermined action without changing the values originally set in the action request value table 121 .

Note that in the above-described motion control process (FIG. 5), in step S109, the predetermined motion selected as the motion to be performed by the selection unit 113 is started and completed, and then the process proceeds to step S110. is assumed. However, in step S109, when the execution of the predetermined action as the execution action is started, the process may immediately proceed to step S110 without waiting for the end. In this case, when an external stimulus is detected in step S102, even if the predetermined action started in step S109 is in progress, the action is terminated and the action in step S103 is started.

If the predetermined action as the action for execution that has been started in step S109 ends in the middle, the process of step S111 may not be executed, or the process of step S111 may be executed depending on the operation time of the predetermined action. may decide whether or not For example, if the predetermined operation is completed (interrupted) after being performed for less than half of the original operation time, the process of step S111 is not executed, and if it is performed for more than half the original operation time, step S111. may be executed. Also, the setting unit 112 may set the action request value to a value obtained by internally dividing the current value and the default value by the ratio between the original execution time of the predetermined action and the time until it is interrupted. .

(Modification 1)
In the first embodiment, the action principle value is increased by a predetermined value each time the predetermined action is executed. may be increased by the value of When there is a difference in the execution time between the respective predetermined actions, updating the action principle value at a constant cycle in this way makes it possible to stabilize the update frequency of the action principle value. In addition, it is possible to change the amount or cycle of increasing the behavioral principle value for each behavioral principle. Such Modification 1 will be described.

In Modified Example 1, the robot 100 causes a plurality of threads, which will be described later, to operate in parallel in motion control. The first thread is an action principle value update thread that updates action principle values. The second thread is an action request value update thread that updates the action request value. The third thread is an action control thread that performs various action controls by referring to action principle values and action request values that are periodically updated. All of these threads start executing in parallel when the robot 100 is activated.

First, the action principle value update thread for updating the action principle value will be described with reference to FIG. The robot 100 needs to activate only one behavior principle value update thread when updating the behavior principle values for all behavior principles at a common cycle. In this case, the one action principle value update thread takes charge of updating the action principle values of all the action principles. Also, if there are action principle values to be updated in different cycles, action principle value update threads are activated for the number of types of update cycles. In this case, each behavioral principle value update thread takes charge of updating the behavioral principle value of the behavioral principle that is updated at the cycle of the behavioral principle value update thread. One action principle value update thread may be activated for each action principle value. In this case, each action principle value update thread is in charge of updating the corresponding action principle value.

When the behavioral principle value update thread is activated, the setting unit 112 first initializes the behavioral principle value to be updated by the behavioral principle value update thread (step S201). Next, control unit 110 waits for a predetermined period of time to elapse (step S202). This predetermined period of time is set corresponding to the cycle of updating the behavioral principle value that the behavioral principle value update thread takes charge of updating, and is, for example, one minute.

Next, the acquisition unit 111 acquires the execution frequency parameter of each predetermined action belonging to the behavioral principle updated by the behavioral principle value update thread (step S203). Here, the acquisition unit 111 acquires, as the execution frequency parameter, whether or not each predetermined action is executed during the predetermined time waited in step S202. Then, based on the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, the setting unit 112 determines whether the behavior principle value update thread belongs to the behavior principle to be updated during the predetermined time waited in step S202. If none of the actions have been executed, the action principle value responsible for updating is increased by a predetermined value (eg, 1) (step S204). However, the action principle value that has reached the upper limit is left at the upper limit value without being increased in step S204.

Next, based on the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, the setting unit 112 determines whether the action principle value update thread belongs to the action principle to be updated during the predetermined time waited in step S202. When the predetermined action has been performed for a predetermined period of time or a predetermined number of times, the action principle value to be updated is decreased by a predetermined value (eg, 1) (step S205), and the process returns to step S202. However, the action principle value that has reached the lower limit is not decreased in step S205, but remains at the lower limit value.

By the action principle value update thread described above, each action principle value is continuously updated with a predetermined value at a predetermined cycle. Next, the action request value update thread for updating the action request value will be described with reference to FIG. Basically, the robot 100 activates one action request value update thread when activated. In this case, the one action request value update thread takes charge of updating the action request values of all the predetermined actions stored in the action request value table 121 . However, as with the action principle value update thread, if several types of action request values with different update cycles are prepared, the action request value update threads are activated as many times as the number of update cycle types. In this case, each action request value update thread is in charge of updating the action request value of a predetermined action that is updated at the cycle of the action request value update thread among the action request values stored in the action request value table 121. do.

When the action request value update thread is activated, first, the control unit 110 waits for a predetermined period of time to pass (step S301). This predetermined period of time is set corresponding to the cycle of updating the action request value that the action request value update thread takes charge of updating, and is, for example, one minute.

Next, the acquisition unit 111 acquires the execution frequency parameter of each predetermined action (step S302). As described above, various values can be used as the execution frequency parameter, but for the sake of simplicity, the execution frequency parameter will be the presence or absence of actual execution. That is, in step S302, the acquisition unit 111 acquires, as an execution frequency parameter, whether or not each predetermined action is executed during the predetermined time waited in step S301. Then, based on the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, the setting unit 112 sets the action request value stored in the action request value table 121 once during the predetermined time waited in step S301. The action request value of a predetermined action that was not executed (if multiple action request value update threads are running, the action request update thread is in charge of updating) is changed to a predetermined value ( For example, it is increased by 1) (step S303).

Next, based on the execution frequency parameter of each predetermined action acquired by the acquisition unit 111, the setting unit 112 updates the action request value stored in the action request value table 121 during the predetermined time waited in step S301. When a predetermined action that the action request value update thread is in charge of updating is executed for a certain period of time (for example, the time exceeding the execution reference time) or a certain number of times (for example, the number of times exceeding the execution reference number), the action request value of the predetermined action is returned to the default value (step S304), and the process returns to step S301.

Since it may not be desirable to interrupt the execution of a predetermined action that has already started, the action control thread (FIG. 8), which will be described later, multiplies the action request value by a predetermined weight. In order to more reliably prevent the interruption, in step S303, the action request value of the currently executed predetermined action (predetermined action that has not been completed to the end) is also increased by a predetermined value, and in step S304, the current For a predetermined action that is being executed (predetermined action that has not been completed to the end), the action request value of the predetermined action may not be returned to the default value.

In addition, the above-mentioned "processing for preventing the execution of a predetermined action from being interrupted" is a predetermined action with a lower execution frequency (for example, in the action request value table 121, a (predetermined actions whose number of action principle values is less than half of the total number of action principle values) are performed, and predetermined actions with a higher execution frequency (for example, in the action request value table 121, other predetermined actions A predetermined action for which the number of action principle values with a larger action requirement value is equal to or more than half of the total number of action principle values) may not be performed.

Each action request value is continuously updated at a predetermined cycle by the action request value update thread described above. Next, referring to FIG. 8, an action control thread that performs various action controls with reference to these periodically updated action principle values and action request values will be described.

First, the control unit 110 resets the value of the timer T to 0 (step S401). Then, the control unit 110 determines whether or not there is an external stimulus (stroked, beaten, held, etc.) based on the detected value by the tactile sensor provided in the sensor unit 133 (step S402).

If there is an external stimulus (step S402; Yes), the control unit 110 sets a motion corresponding to the external stimulus (a motion not included in the predetermined motion described above) as the next motion candidate (step S403), and step S407. proceed to

If there is no external stimulus (step S402; No), the controller 110 determines whether or not a predetermined period of time (for example, one minute) has elapsed using the timer T (step S404). If the predetermined time has not elapsed (step S404; No), the process returns to step S402.

If the predetermined time has elapsed (step S404; Yes), the selection unit 113 refers to the action request value table 121, extracts the maximum action request value, substitutes it for the variable M, and extracts the maximum action request value. The predetermined motion that the user has is set as a motion candidate for the next execution motion (step S405). Then, the selection unit 113 determines whether the variable M is larger than the value obtained by multiplying the value of the action request value of the currently executed predetermined action by a predetermined weight (for example, 1.2), or the currently executed predetermined action ends. It is determined whether or not to do so (step S406). The reason why the predetermined weight is multiplied is that it is considered better to continue the predetermined motion once started until the motion ends as much as possible. However, the weight may not be set (the weight may be set to 1), and the action may be immediately switched to the predetermined action with the maximum action request value. An operation time is set in advance for each predetermined action, and it is possible to determine whether or not the currently executing predetermined action will end from the difference between the time when the predetermined action started and the current time.

If the value obtained by multiplying the value of the action request value of the currently executed prescribed action by the prescribed weight is equal to or greater than the value of the variable M, and if the currently executed prescribed action does not end (step S406; No), step S401 back to

If the variable M is larger than the value obtained by multiplying the value of the action request value of the currently executed prescribed action by a prescribed weight, or if the currently executed prescribed action ends (step S406; Yes), the action control unit 114 terminates the action currently being executed (predetermined action or action in response to an external stimulus) (step S407). Then, the motion control unit 114 controls the sound output unit 134, the emotional expression unit 135, and the driving unit 136 so that the motion set as the “next motion candidate” in step S405 or S403 (predetermined motion or external stimulus) corresponding operation) is started (step S408). Then, the process returns to step S401.

Although omitted in the above description to avoid complication, if no operation is performed, such as immediately after the operation control thread is started, steps S406 and S407 are skipped after step S405. , the process proceeds to step S408. Alternatively, when the motion control thread is started, a special initial motion (which may only be internally recognized as a motion by the control unit 110 and may not be recognized as a motion by the user) is started. may

With the motion control thread described above, the robot 100 according to Modification 1 continues the motion as much as possible until the current predetermined motion is completed, and when a predetermined motion with a larger action request value is extracted, switches the action content to execution of a predetermined action with the maximum action request value. Also in Modified Example 1, the robot 100 can execute predetermined motions based on various behavioral principles. In the robot 100 according to Example 1, it is possible to independently set the update cycle of the action principle value, the update cycle of the action request value, and the switching cycle of the predetermined action. As a result, it is possible to improve the degree of freedom of variations in the appearance timing of each predetermined action. Also, in Modification 1, when the user wants to add a new predetermined action, the user can easily add the new predetermined action without changing the values originally set in the action request value table 121. .

In step S406 of the motion control thread (FIG. 8) described above, a constant of 1.2 was exemplified as the predetermined weight to be multiplied by the action request value of the predetermined motion currently being executed. may be changed to For example, for a predetermined action that is currently being executed, if the predetermined action is being executed for the first time on that day, the weight is set to 2.0, and the value of the weight is changed to 1.8, 1.5, 1.8, 1.5, 1.5, 1.8, 1.5, 1.5, 1.8, 1.5, 1.5, 1.8, 1.5, 1.5, 1.8, 1.5, 1.5, 1.8, 1.5, 1.5, 1.5, 1.8, 1.5, 1.5, 1.8, 1.5, 1.2. It is conceivable to decrease it like 2 (for example, decrease it by 0.1 each time one minute passes, and finally set it to 1.0). Also, in the case of a predetermined action that has already been performed on that day, the weight may be set to 1.2, and the value of the weight may be decreased to 1.0 as time passes.

Also, the weight value and the method of changing the weight value may be changed according to the type of the predetermined action. For example, one predetermined action is always unweighted (the weight value is always 1.0), and another predetermined action is always weighted 2.0 (2.0 from the beginning and over time). 2.0).

In addition, a predetermined action with a lower execution frequency (for example, a predetermined action whose number of action principle values with larger action requirement values than other predetermined actions in the action request value table 121 is less than half of the total number of action principle values) ) is multiplied by a weight (this weight value may be set to a larger value as the execution frequency is lower), and a predetermined action with a higher execution frequency (for example, in the action request value table 121, A predetermined action in which the number of action principle values that increase the action requirement value is equal to or more than half of the total number of action principle values) may be given no weight (the weight value is 1.0).

(Modification 2)
In the above-described first embodiment and modified example 1, all behavioral principles were treated equally, but the surrounding environment (quiet, noisy, cold, hot, user stared at, user stroked, user , etc.) or the internal state of the robot 100 (remaining battery level, amount of heat generated, etc.), the action principles to be selected for the predetermined action may be narrowed down.

For example, as shown in FIG. 9, the action request value table 121 includes the action principle "boredom" when the user does not care, the action principle "hungry" when the remaining battery level is below the reference value, and the ambient temperature. Alternatively, the robot 100 may define a behavioral principle of "hot" when the amount of heat generated by the robot 100 exceeds a reference value, a behavioral principle of "fun" when the user cares, etc., and perform a predetermined action according to each behavioral principle. Can be set with the requested value. Such a modification 2 will be described.

In Modified Example 2, the robot 100 selects an action principle based on the surrounding environment and the internal state of the robot 100, and selects and executes a predetermined action with the highest action requirement value among the selected action principles. Operation control processing for this will be described with reference to FIG. Execution of this motion control process is started when the robot 100 is activated. Note that in the second modification, the sensor unit 133 includes not only sensors (human sensor, touch sensor, temperature sensor, humidity sensor, etc.) for acquiring information on the surrounding environment, but also the internal state of the robot 100 (remaining battery level , calorific value, etc.). Further, it is assumed that the action principle value of each action principle is initialized when the robot 100 is activated.

First, the control unit 110 acquires the surrounding environment and the internal state of the robot from the image input unit 131, the voice input unit 132, and the sensor unit 133 (step S501).

Next, the control unit 110 selects an action principle based on the acquired ambient environment and internal state (step S502). For example, if the user has been ignoring the robot 100 for a while, the action principle "boredom" is selected. Also, when the user gazes at or holds the robot 100, the action principle "enjoyable" is selected. A plurality of action principles may be selected depending on the surrounding environment and internal state. For example, if the user keeps ignoring the robot 100, the remaining battery level of the robot 100 is 20% or less, and the ambient temperature is 30 degrees or higher, the action principle "boredom" and the action principle "hungry" , and action principle ``hot'' are selected.

Next, the selection unit 113 selects, as a motion for execution, a predetermined motion having the largest action requirement value among the predetermined motions included in the action principle selected in step S502 (step S503). If there are a plurality of predetermined motions with the maximum required action value, the selection unit 113 randomly selects one predetermined motion from among the predetermined motions with the maximum required action value as an action to be executed.

Next, the motion control unit 114 controls the sound output unit 134, the emotional expression unit 135, and the driving unit 136 so as to execute the predetermined motion selected by the selection unit 113 as the motion for execution in step S503 (step S504). .

Next, the setting unit 112 selects a predetermined action (not executed) other than the predetermined action selected by the selecting unit 113 from among the predetermined actions belonging to the action principle to which the predetermined action selected as the action to be performed by the selecting unit 113 in step S503 belongs. (predetermined action) is increased by a predetermined value (for example, 10) (step S505). Then, the setting unit 112 returns the action request value of the predetermined action (execution action) selected by the selection unit 113 in step S503 to the default value (the value initially set in the action request value table 121) (step S506). ). Note that in steps S505 and S506, the acquisition unit 111 obtains information on the actual execution state of the predetermined motion of the robot 100 (the duration of the state in which each predetermined motion is not executed, the execution time and execution time of each predetermined motion, etc.). information such as the number of times) as an execution frequency parameter, and based on the execution frequency parameter, the action request value may be increased or returned to the default value.

Then, the setting unit 112 sets a predetermined value (for example, 1 ) (step S507). Note that the action principle value that has reached the upper limit is left at the upper limit value without being increased in step S507.

Next, the setting unit 112 decreases the action principle value of the action principle including the predetermined action selected by the selection unit 113 in step S503 by a predetermined value (for example, 1) among the action principles selected in step S502. (step S508) and returns to step S501. Note that the action principle value that has reached the lower limit is not decreased in step S508, but remains at the lower limit value.

In steps S507 and S508, the acquisition unit 111 obtains information on the actual execution state of the predetermined motion of the robot 100 (the execution time and number of executions of each predetermined motion, and the duration of the state in which each predetermined motion is not performed). etc.) is acquired as an execution frequency parameter, and among the behavior principles selected in step S502, the behavior principle value is increased or decreased based on the execution frequency parameter (for example, a predetermined action executed for a certain period of time or a certain number of times belongs to The action principle value of the action principle may be decreased by a predetermined value (eg, 1), and the other action principle values may be increased by a predetermined value (eg, 1).

By performing the motion control processing described above, the robot 100 can perform various predetermined motions from behavioral principles that match the ambient environment and internal state at that time, as described above. Then, the user can easily add a new predetermined action without changing the values originally set in the action request value table 121 .

(Modification 3)
In the first embodiment and each modified example described above, the default value of the action request value table 121 is fixed to a preset value. When the user adds a predetermined action, it is added to the action request value table 121 together with the default value of the action request value. Also, the default value of the action request value was not changed while the robot 100 was operating. Among the predetermined actions of the robot 100, the actions preferred by the user differ depending on the user. It can be made easier to appear. Such a modification 3 will be described.

In the third modification, the robot 100 performs the predetermined motion selected by the selection unit 113 by the motion control unit 114, and then the control unit 110 controls the image input unit 131, voice input unit 132, Based on the input from the sensor unit 133, the user's reaction to the predetermined action is acquired. If this reaction is a favorable reaction, the setting unit 112 increases the default value of the action demand value for the predetermined action in the action demand value table 121 by a predetermined value (for example, 10). Also, if this reaction indicates antipathy, the setting unit 112 decreases the default value of the action request value of the predetermined action in the action request value table 121 by a predetermined value (for example, 10).

In Modified Example 3, by changing the default values of the action request value table 121 according to the user's response, the robot 100 can increase the probability of performing a predetermined action that is favorable to the user. Also, it is possible to reduce the probability that the user will perform a predetermined action that is offensive to the user. In addition, since processing other than changing the default value can be performed in the same manner as in the above-described embodiment and modifications, various predetermined operations can be performed. Therefore, the robot 100 according to Modification 3 performs various predetermined actions while increasing the probability of performing a predetermined action that is favorable to the user, thereby improving the execution frequency of the predetermined action that is favorable to the user ( It is possible to achieve both a decrease in the execution frequency of a predetermined action that causes an antipathy to the user and an improvement in the variety of actions.

(Other modifications)
It should be noted that the first embodiment and each modified example described above are not limited to the contents described above, and various modifications are possible. For example, in the first embodiment and each modified example described above, each predetermined action performed by the robot 100 is defined in a small table divided for each action principle in the action request value table 121, but this is not the only option. can't For example, the action request value table 121 may be a single table in which action request values for all predetermined actions are defined without action principles (without dividing into small tables).

Further, among the above-described embodiments, the process of changing the action request value according to the execution frequency of each predetermined action cannot obtain a suitable execution frequency unless the data accumulates to some extent. Also, instead of time, for example, until the average value of the number of executions of each predetermined action for all the predetermined actions or some of the predetermined actions with the highest number of executions (for example, 10) exceeds the reference number of executions (for example, 5 times) ), the default action request value is used as it is, and after the predetermined period of time has elapsed, according to the execution frequency, the predetermined action with the lowest execution frequency is selected and executed, and accordingly, the predetermined action The execution frequency may be reset.

In addition, in the selection of the predetermined action, it is not always necessary to select according to the action request value. For example, the predetermined action may be selected at random. ) may be selected according to a set probability, or may be selected in a predetermined order.

Further, in the above-described embodiment and modification, the robot 100 has been described as including the image input unit 131 , the voice input unit 132 , the sensor unit 133 , the sound output unit 134 , the emotion expression unit 135 and the driving unit 136 . However, when the robot 100 does not need to acquire the surrounding environment (when the predetermined operation can be executed without information on the surrounding environment), the image input unit 131, the voice input unit 132, and the sensor unit 133 Either or all of them may not be provided.

Further, when the robot 100 does not need to output sound when performing a predetermined action, the sound output unit 134 may not be provided. When the robot 100 does not need to express an emotion by changing the facial expression or moving the front legs 21, the hind legs 22, or the tail 23, the emotion expression unit 135 is activated. You don't have to prepare. Further, when the robot 100 does not need to move, nod, look around, or doze off when performing a predetermined action, the drive unit 136 may not be provided.

Similarly, the robot 100 does not necessarily have the shape of a cute animal, nor does it need to have a structure in which the head 10 and the body 20 are connected by the neck joint 15 . Furthermore, the robot 100 may not have any or all of the nose 11, the ears 12, the mouth 13, the eyes 14, the front legs 21, the hind legs 22, and the tail 23; It may have wings or the like. In short, there are no particular restrictions on the appearance of the robot 100 .

Further, the robot 100 may perform user identification by the image input unit 131 or the voice input unit 132 and store a plurality of action request value tables 121 for each identified user. By doing so, in the above-described modified example 3, the default value of the action request value table 121 is set to a value suitable for each user. 100 makes it possible to perform various predetermined actions while increasing the probability of performing a predetermined action that a user who comes in contact with will feel favorably.

Each function of the robot 100 can also be implemented by a computer such as a normal PC (Personal Computer). Specifically, in the above embodiment, it is assumed that the programs such as the motion control process performed by the robot 100 are stored in advance in the ROM of the storage unit 120 . However, the program may be stored in a computer-readable storage medium such as a flexible disk, CD-ROM (Compact Disc Read Only Memory), DVD (Digital Versatile Disc), MO (Magneto-Optical Disc), memory card, USB (Universal Serial Bus) memory, etc. By storing and distributing the program in a recording medium, and reading and installing the program in the computer, a computer capable of realizing each of the functions described above may be configured.

Alternatively, a computer such as a normal PC may be prepared as a control device for controlling the robot 100 separately from the robot 100, and the functions of the control unit 110 and the storage unit 120 may be realized by the control device. In this case, by connecting the robot 100 and the control device wirelessly or by wire, the control device can receive input data from the image input unit 131, the voice input unit 132, and the sensor unit 133 of the robot 100. Also, the sound output unit 134, the emotion expression unit 135, and the driving unit 136 of the robot 100 can be controlled.

Although the preferred embodiments of the present invention have been described above, the present invention is not limited to such specific embodiments, and the present invention includes the invention described in the claims and their equivalents. be The invention described in the original claims of the present application is appended below.

(Appendix 1)
A robot configured to be able to execute each of a plurality of predetermined actions,
acquisition means for acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
According to the acquired execution frequency parameter, selection of an execution action to be executed by the device from the plurality of predetermined actions is performed so that a predetermined action with a lower execution frequency is more likely to be selected as the execution action. , a selection means that performs
an operation control means for causing the device to execute the predetermined operation selected as the execution operation;
A robot characterized by comprising:

(Appendix 2)
The robot according to appendix 1, wherein the acquiring means acquires a duration of a state in which each of the plurality of predetermined actions is not performed as the execution frequency parameter.

(Appendix 3)
The robot according to Supplementary Note 1, wherein the obtaining means obtains, as the execution frequency parameter, the number of executions, which is the number of times each of the plurality of predetermined actions is executed.

(Appendix 4)
The selecting means selects the execution motion from the plurality of predetermined motions based on comparison results between the plurality of selection parameters corresponding to the plurality of predetermined motions;
Based on the execution frequency parameter of each of the plurality of predetermined actions, the selection of the action for execution from the plurality of predetermined actions is performed so that a predetermined action with a lower execution frequency is more likely to be selected as the action for execution. 4. The robot according to any one of appendices 1 to 3, further comprising setting means for setting each of the plurality of selection parameters.

(Appendix 5)
The action control means controls an execution frequency parameter corresponding to a predetermined action currently being executed as the action to be executed, and a predetermined action other than the predetermined action as the action to be executed among the plurality of predetermined actions. According to the comparison result with the corresponding execution frequency parameter, when the execution frequency of the predetermined action as the action for execution is lower, the predetermined action currently being executed is completed without interruption. 5. The robot according to any one of the appendices 1 to 4, characterized in that it is executed by a machine.

(Appendix 6)
A robot control method configured to be able to execute each of a plurality of predetermined actions, comprising:
an acquisition step of acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
According to the execution frequency parameter obtained in the obtaining step, when selecting an action to be executed by the robot from among the plurality of predetermined actions, a predetermined action with a lower execution frequency is more likely to be selected as the action to be executed. a selection step performed so that
a motion control step of causing the robot to perform the predetermined motion selected as the execution motion in the selection step;
How to control the robot, including

(Appendix 7)
A computer of a control device for controlling a robot configured to be able to execute each of a plurality of predetermined actions,
an acquisition step of acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
According to the execution frequency parameter obtained in the obtaining step, when selecting an action to be executed by the robot from among the plurality of predetermined actions, a predetermined action with a lower execution frequency is more likely to be selected as the action to be executed. a selection step performed so that
a motion control step of causing the robot to perform the predetermined motion selected as the execution motion in the selection step;
program to run the

10 Head 11 Nose 12 Ears 13 Mouth 14 Eyes 15 Neck Joint 20 Torso 21 Front Legs 22 Hind Legs 23 Tail 100 Robot 110 Control Unit , 111... acquisition unit, 112... setting unit, 113... selection unit, 114... operation control unit, 120... storage unit, 121... action request value table, 131... image input unit, 132... voice input unit, 133... sensor unit , 134... Sound output unit, 135... Emotion expression unit, 136... Driving unit

Claims (5)

A robot configured to be able to execute each of a plurality of predetermined actions,
acquisition means for acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
selection means for selecting an action corresponding to a maximum required action value among the action required values set for each of the predetermined actions, from the plurality of predetermined actions as an action to be executed;
operation control means for causing the machine to execute the execution operation selected by the selection means;
After the action for execution is executed by the action control means, among the action request values set for each of the predetermined actions, the execution frequency represented by the acquired execution frequency parameter is a predetermined action lower than the other predetermined actions. selection control means for controlling the selection means so that a predetermined action having a lower execution frequency than other predetermined actions is more likely to be selected as the action for execution by increasing an action request value corresponding to the action;
A robot characterized by comprising: 2. The robot according to claim 1, wherein said obtaining means obtains, as said execution frequency parameter, a duration of a state in which each of said plurality of predetermined actions is not executed. 2. The robot according to claim 1, wherein said obtaining means obtains, as said execution frequency parameter, the number of times each of said plurality of predetermined actions has been executed. A robot control method configured to be able to execute each of a plurality of predetermined actions, comprising:
an acquisition step of acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
a selection step of selecting, as an action for execution, an action corresponding to a maximum required action value among the action required values set for each of the predetermined actions, from the plurality of predetermined actions;
an operation control step of causing the self-machine to execute the execution operation selected in the selection step;
After the action for execution is executed in the action control step, among the action request values set for each of the predetermined actions, the execution frequency represented by the acquired execution frequency parameter is lower than that of other predetermined actions. a control step of increasing the action request value corresponding to the predetermined action so that the predetermined action whose execution frequency is lower than that of other predetermined actions is more likely to be selected as the action for execution;
How to control the robot, including a computer of a control device for controlling a robot configured to be able to execute each of a plurality of predetermined actions;
Acquisition means for acquiring an execution frequency parameter representing the actual execution frequency of each of the plurality of predetermined actions;
selection means for selecting an action corresponding to a maximum required action value among the action required values set for each of the predetermined actions as an execution action from the plurality of predetermined actions;
Operation control means for causing the machine to execute the execution operation selected by the selection means ;
After the action for execution is executed by the action control means, among the action request values set for each of the predetermined actions, the execution frequency represented by the acquired execution frequency parameter is a predetermined action lower than the other predetermined actions. selection control means for controlling the selection means so that a predetermined action having a lower execution frequency than other predetermined actions is more likely to be selected as the action for execution by increasing an action request value corresponding to the action;
A program to function as

Images