WO2006030530A1

WO2006030530A1 - Imitated animal having stimulus priority judgment means

Info

Publication number: WO2006030530A1
Application number: PCT/JP2004/014034
Authority: WO
Inventors: Takanori Shibata; Kotarou Hayashi; Hidekazu Shimada
Original assignee: National Institute Of Advanced Industrial Science And Technology; Microjenics, Inc.
Priority date: 2004-09-17
Filing date: 2004-09-17
Publication date: 2006-03-23

Abstract

There is provided an imitated animal expressing a plurality of operation s by control of a computer. The imitated animal includes: perception means (1) for detecting a stimulus received by the imitated animal as an external parameter and generating an event; imitated feeling formation means (2) for expressing the imitated feeling as an internal parameter according to the detection state of the external parameter; operation execution means (5) for deciding the reaction operation for the combination of the external parameter and the internal parameter and receiving the event so as to realize the reaction operation as a movement of a predetermined portion. The perception means (1) includes: a sensor (7) for detecting the stimulus; an input interface unit (27) for detecting the external parameter from the output of the sensor; an event generation unit (25) for generating an event assigned to the combination of a single or a plurality of external parameters obtained via the input interface (27) from the sensor; and stimulus priority judgment means (26) for deciding the event to be employed by the operation execution means (5) when a conflict of events outputted from the event generation unit (25) is detected.

Description

Description Simulated Biological Device with Stimulus Priority Judgment Technology

The present invention relates to a simulated biological apparatus provided with a stimulus priority determination means that causes various operations to be performed by computer control, and particularly relates to optimization of sensitivity and response to externally received stimuli. Background art

Nowadays, various simulated biological devices have been devised that allow a plurality of operations to be expressed by computer control. For example, as described in Japanese Patent Application Laid-Open No. 10-34557, In order to control the robot equipped with the actuator unit, the content of the sensor information filtering in the sensor filter unit is determined by the control signal input from the control signal input unit, and the plurality of sensor units are determined by the filtering content. A method is disclosed in which sensor information extracted from the sensor is selected and narrowed down, and each actuator unit is controlled by a reflective control unit based on the narrowed sensor information.

However, there is an example in which the conventional simulated biological apparatus generates only one sensor control value corresponding to the sensor value by referring only to the sensor information of the sensor expression unit group having the activity of a predetermined value or more. As disclosed, the response control process based on filtering for new sensor information is limited, and the priority is individually compared and determined specifically between the currently executed action and the new sensor information. It has not yet been processed properly.

The present invention has been made in view of the above circumstances, and can perform efficient processing by reducing unnecessary processing for new sensor information, and can perform priority processing with the operation currently being executed. It is an object of the present invention to provide a simulated biological apparatus provided with a stimulus priority determination means for comparing and determining each of them individually and selectively expressing an appropriate action. Disclosure of the invention

The present invention is a simulated biological device that causes a plurality of operations to be generated by control by a computer, and a sensory means for detecting an stimulus received by the simulated biological device as an external parameter and generating an event, and a detection status of the external parameter To determine the reaction behavior for the combination of the external parameter and the internal parameter, and to realize the reaction motion in the movement of a predetermined part in response to the event. In the simulated biological apparatus, the sensory detection device includes: a sensor that detects the stimulus; an input interface unit that detects the external parameter from the output of the sensor; and the sensor. One external parameter obtained through the input interface part of A competition between an event generation unit that generates a reaction event assigned to a single stimulus or a combination of external parameters and a reaction event output from the event generation unit. It comprises stimulus priority determination means for determining a reaction event to be adopted by the action execution means when the reaction action is determined when it is known. The event detection process for verifying the presence or absence of a new reaction event occurring at the event generation unit and the previous reaction event during the execution of the reaction operation is performed, and a new reaction event or a previous reaction event is performed in the event detection process. When the reaction action is determined, the reaction event to be adopted by the action execution unit when the reaction action is determined is set to a priority given in advance for each category of the sensor that detects the external parameter applied to each reaction event. Stimulus priority determination means for determining via at least one of the priority determination steps between categories determined based on or the priority determination steps within categories determined based on priorities given in advance to each reaction event It is also possible to take a form with

According to the simulated biological apparatus of the present invention, efficient processing can be performed only by filtering between new sensor information, and priority is given to the currently executed operation and the new sensor information. It is possible to compare and judge the degree individually, and to compare and judge the priority with the action that is being executed efficiently, and to selectively express the appropriate action. Brief Description of Drawings

FIG. 1 is a principal functional block diagram showing an example of a simulated biological apparatus according to the present invention.

FIG. 2 is an overall functional block diagram showing an example of a simulated biological apparatus according to the present invention.

FIG. 3 is a partial functional block diagram showing an example of a simulated biological apparatus according to the present invention.

FIG. 4 is a partial functional block diagram showing an example of a simulated biological apparatus according to the present invention.

FIG. 5 is a flowchart showing an example of external parameter detection processing in the simulated biological apparatus according to the present invention.

FIG. 6 is a flowchart showing an example of external parameter analysis processing in the simulated biological apparatus according to the present invention.

FIG. 7 (A) is a flowchart showing an example of event detection processing and priority event determination processing in the simulated biological apparatus according to the present invention.

FIG. 7 (B) is a flowchart showing an example of priority event determination processing in the flowchart of FIG. 7 (A). BEST MODE FOR CARRYING OUT THE INVENTION

An embodiment of a simulated biological apparatus according to the present invention will be described with reference to the drawings.

The illustrated embodiment is a simulated biological apparatus that generates a plurality of operations under the control of a computer system, that is, a so-called robot system. The simulated biological apparatus includes a plurality of sensors 7 and a plurality of sensors as shown in FIG. The computer is composed of a computer having a switch, a memory, and a CPU that controls the sensor 7 and the sensor 7, and is housed in a casing having a predetermined outer shape and a movable structure. .

Then, as a functional module formed by them, a stimulus received from the environment where the simulated biological apparatus is placed is stored in the memory as an external parameter, respectively. Perceiving means 1 to generate events (reaction events) for detecting and reacting automatically, and spontaneously generating events (autonomous events) for developing autonomous behavior even when external parameters due to stimulation are not detected In order to express the simulated feeling according to the action determination means 4 for selecting the event of the action generated from either the reaction event or the autonomous event and the detection condition of the external parameter included in the reaction event. A plurality of motion patterns corresponding to combinations of simulated emotion forming means 2 for guiding internal parameters onto the memory, and external parameters included in the reaction event and internal parameters (including those included in an autonomous event; the same applies hereinafter). Database 3 in which the event assignments are stored, and the reaction event or autonomous event (hereinafter “event”). In response to this, the database 3 is referred to, and the reaction pattern or autonomous action pattern for the combination of the external parameter and the internal parameter included in the reaction event is determined. It is a simulated biological device having an action execution means 5 for receiving the action pattern of the reaction action or the autonomous action in response to the movement of a predetermined part and a timer 18 for outputting time information (see FIG. 2). .

The sensory means 1 includes a sensor 7 that detects sound, light, infrared, heat, acceleration, or pressure built in the simulated biological device, and performs an arithmetic process on the output of the sensor 7 and includes external parameters. A sensor processing unit 8 that outputs as a reaction event, and the simulated emotion forming means 2 uses an external parameter obtained from the perceptual means 1 to derive an emotion parameter as one of the internal parameters. 9 and from the calculated emotion parameters, the reaction holding time, the number of repetitions, the amount of movement (amplitude), the steady position, and the movement speed (hereinafter collectively referred to as “motion elements”) are changed. It is composed of an operation propagation amount calculation unit 10 for deriving an operation propagation parameter that is an internal parameter (see FIGS. 2 and 3). The emotion parameter is a combination of numerical data composed of a pleasant / discomfort parameter Kkh and an excitement / sedation parameter Kkt. The simulated emotion formation means 2 detects external parameters of various reaction events, and the pleasant / discomfort parameter Kkh Excitement Z Calorific parameter The numerical value of Kkt is increased or decreased as appropriate (for example, see Table 1). Based on the increase and decrease of those emotion parameters, as shown in Table 2, “joy”, “anger”, “sorrow” As a result of being recognized as one of the five simulated emotion zones consisting of ",""Easy", and "Neutral" and expressed as a reaction action by the action execution means 5, the sentiment of the simulated biological apparatus is simulated. It is expressed and its simulated emotion changes under certain conditions every time it receives a stimulus.

That is, from the simulated emotion forming means 2, based on the above emotion parameters, for example, as shown in Table 2, when Kkh≥70 and Kkt≥70, "joy", Kkh≤-70, and Kkt≥70 In the case of "anger", Kkh≤-70, and in the case of Kkt≤-70, "sorrow", in the case of Kkh ≥70 and Kkt≤-70, "easy", and -70 If <Kkh <70 or -70 <Kkt <70, the simulated emotion is positioned as “neutral.” Each time a reaction event is detected, depending on the simulated emotion zone, for example, Table 3 As shown in the table, motion propagation parameters (repetition count increase / decrease parameter, steady position parameter, operation hold time increase / decrease parameter, speed parameter, amplitude increase / decrease parameter, etc.) are output and given to the following action execution means 5. Table 1

Table 2 Judgment conditions for simulated emotions

Pleasant / uncomfortable parameter Kkh ¾ Quiet parameter Kkt

-¾ Klch> 7 0 AND Kkt≥ 7 0

Angry Kkh≤- 7 0 AND Kkt≥ 7 0

Sad Kkh≤- 7 0 AND Kkt≤-7 0

Easy Kkh≥ 7 0 AND Kkt≤- 7 0

New 卜 ral 1 7 0 Kkh 7 0 OR 1 7 0 Kkt 7 0 Table 3

The motion determination means 4 generates the autonomous event including internal parameters (autonomous parameters) relating to autonomous motion that are spontaneously expressed even when no externally detectable stimulus is detected and no external parameter is detected. Autonomous motion generation unit 1 1 that receives the response event and the autonomous event, analyzes them, and determines either the response event or the autonomous event based on the priority order of the following event priority information set in advance. It is composed of the acceptance / rejection decision part 1 2 to output (see Fig. 2). The event includes event classification information for distinguishing between the autonomous event and the reaction event, and event priority information indicating the priority order of various events, and in the case of the reaction event, the external parameter is used as the external parameter. Event classification information that identifies one or more sensors 7 (acceleration sensors, pressure-sensitive (tactile) sensors, etc.) that detected the stimulus from the event, and event intensity that represents the amount of stimulation detected by the one or more sensors It has information. These external parameters For example, when the event classification information indicates a tactile sensor, a stimulus such as “striking”, “striking”, and “pressing strongly” can be expressed by the amount of the stimulus. In the case of the autonomous event, the autonomous parameter is included as the event classification information.

The operation execution means 5 includes various actuators 19 including motors, speakers, and the like, and receives the event and refers to the database 3 to generate operation patterns according to external parameters and autonomous parameters included in the various events. The operation selection unit 13 to be selected, and the control amount calculation unit for deriving the control parameter of the operation element related to the operation pattern selected by the operation selection unit 13 based on the operation propagation parameter given from the simulated emotion forming means 2 14 and an actuator controller 15 that adjusts control signals such as drive energy to the various actuators 19 in response to the control parameters (see FIG. 4).

The sensor processing unit 8 of the sensory means 1 includes an input interface unit 27 that detects the external parameter from the output of the sensor 7, and one case obtained from the sensor 7 via each input interface unit 27. When a competition is detected between an event generator 25 that generates a reaction event assigned to an external parameter or a combination of multiple external parameters, and a reaction event output from the event generator 25 That is, when a reaction event is continuously output from the event generation unit 25 with a relatively short time difference in a certain time zone, the reaction event to be adopted by the operation execution means 5 is changed to the old and new reaction events. This is provided with a stimulus priority determination means 26 that determines one from any of the events.

Each input interface unit 27 includes a level detection unit 20 that detects the amount of stimulus detected from the sensor 7 as an external parameter, and a level that distributes the level level of the external parameter based on one or more threshold values. The event generation unit 25 includes a determination unit 6, and the event generation unit 25 refers to an event table having contents as shown in the left part of Table 4 provided in the memory, for example. The tactile sensor is exemplified in FIG. 5), depending on the detection state of the sensor 7 and the result of the level determination (for example, the detection level determination of the tactile sensor corresponding to stroking, hitting, or pressing strongly). Generate various reaction events (first (See 囪 and Figure 2). Table 4

Μ 6

○ One -4.0 2.0 (strongly lake

M m¾¾p t

Beat, strong

(Fine Lake ff O--4,0 2.0 Tail Tap O One -4.0 3.0

Squeeze

○ One -4.0 3.0 (strongly lake

While picking up the sensor

〇--9.5 9.0

The processing in each event generation unit 25 in this example is based on the series of processing (external parameter detection) shown in Fig. 5 on the basis of the presence or absence of sensor information (external parameters) based on inputs from multiple sensors 7. Process), and the series of processes shown in Fig. 6 (external parameter analysis process), the process differs depending on whether the detected external parameters are none, only one, or multiple cases. To reduce the number of newly generated reaction events to one.

For example, when there is no sensor information in which the detected external parameter is detected, the external parameter detection process is further repeated to detect a new external parameter (non-detection process).

If there is only one case, as one case processing, check whether it is an external parameter from the tactile sensor in particular, and if it is an external parameter from other than the tactile sensor, refer to the event table. Then, the reaction event corresponding to the category of the external parameter and the amount of stimulation is adopted as the new event. On the other hand, if the external parameter is an external parameter from the tactile sensor, it has been placed in the vicinity of the tactile sensor in the latest external parameter detection history several times in the past (about 2 to 5 times). Check if there is an external parameter detected from another tactile sensor, and if it exists, determine the “stroking” stimulus and the relative position of the external parameter before and after detection from the tactile sensor. In addition to distinguishing between “sequential” and “reverse” from the relationship, one of the corresponding reaction events is adopted as a new event (inter-temporal complex processing).

In addition, when there are multiple detected external parameters, the detection of external parameters corresponding to the “lifting” stimulus by the acceleration sensor, The combination of external parameters that are the conditions for generating a specific composite event, such as generating another event (complex event) such as “easy” by competing with the detection of the external parameter corresponding to the “stroking” stimulus. Check if it exists, and if the combination exists, adopt the reaction event corresponding to the combination as a new event (simultaneous composite processing).

On the other hand, if the combination does not exist, the sensor with the highest priority is selected from the sensors that have detected the individual external parameters, and the inter-category priority table (based on the type of the sensor and the amount of stimulation) For example, refer to the priority between categories in Table 5 or Table 6) or the intra-category priority table (for example, refer to the intra-category priority in Table 5 or 6) and select one. When a single sensor is selected in this way, a reaction event to be adopted as a new event is determined through the above one case process.

Table 5

Stroke event

a Aya Reverse 撫 Push strongly (Shining back, head, rubbed, torso (carrying back ^ m + hitting + rosy ^) 0 Fin, out)

Body

The event that we were in

so

Event Li → 1

Category 1 2 2 2 3 3

1

Sound (My Outside ¾ 喑 (Up and Down 1 A C1

1

Excitation (up and down + ¾S 2

A C1

1

Remarks

1

1 A

Hate ¾¾ (¾> 1 1

Flash ^ ω 2

Beard: ^ Sinking 1 1 A 11

A Stops the current operation (according to the interrupt criterion) and moves to the operation for a new event. Al (A force different from the interrupt criterion) Interrupts the current operation and moves to a new event.

C (with interrupt criteria) Back force Cancels an incoming event.

C1 (A force different from the interrupt criterion) Cancel the incoming event.

― One: Undetectable combination.

Table 6

First event

Event

Sound (Microphone left beard sound ^, m m phase phase fuss: name vs. ran light light change change event after detection m output event event; * ^ = ί 1 1 1 1

Category

1 1 2 1 1 1 2 1 Sound (Microphone ¾, Sound (Up / Down + ¾) 1 1 A C

CI A A

Sound (up and down A A

Resignation K name etc.)

If the boat is in the middle, the key if 1 A CI A A drops ¾¾ 细 1 A A C

1 1 A A A A A Flush ^ a A A A Beard 1 1 A A A A Power! DC water

Happei 1

Degree 脑 K supine na fl

Test 1

Out

及 '麟 1 水水稍 Steel 1

Flat) → 1 1 A A A A

AC weak

(Vibration 鹏彌彌 1

Movement 《Sena, avoidance Ww l ^ j 1

Chestnut

(Water 1

Strong

m Thigh, supine, avoiding, 2

Water ¥ ϋ

Touching, touching, head or back

Sense (Hi ra Yode 1

SEA ^ TQSid f + stroking) 2

I¾t 顿 or back 1 2 A A A A 2

Mouth IX (fin J¾1P # 3

Force (_Bf, + mouth P0 3

A Stops the current operation (according to the interrupt criterion) and moves to the operation for a new event.

A1 Interrupts the current operation (although different from the interrupt judgment criteria) and moves to the operation for a new event.

C (According to interrupt judgment criteria) Rear force Cancels an incoming event.

C1 (Different from the interrupt judgment criteria) Cancels the incoming event.

One: A combination that cannot be detected.

The priority determination unit 26 periodically verifies the presence or absence of a new reaction event sent from the event generation unit 25 and the presence or absence of a previous reaction event currently being executed. In the event detection process, a new reaction event is detected from the event generation part 25 to which the interface part 27 of each sensor is connected, or a previous reaction event is detected. As the priority event determination process, the reaction event to be adopted by the action execution means 5 when determining the reaction action is, for example, as shown in Table 5 or Table 6, The step of determining the priority between categories determined based on the priority given in advance for each category of the sensor 7 that has detected the external parameter related to the response event, or the organism estimated from the stimulus that is the source of each reaction event In view of the influence on the response, the determination is made through at least one of the priority determination steps in the power category that is determined based on the priority given in advance to each reaction event.

If the previous reaction event does not exist in the event detection process, the new reaction event detected by the event detection process is sent to the action determination means 4, and the action determination means 4 sends the new event event. When 卜 is adopted as an event sent to the operation executing means 5, the operation executing means 5 starts to execute an operation pattern for the new event. On the other hand, if there is a previous reaction event in the event detection process, that is, if the operation pattern by the previous event is continuing at that time, first, the destination of the ongoing operation pattern is expressed. Are compared with new events, and it is determined whether or not they belong to the same category, that is, the same type of sensor. In the case of the same category, the intra-category priority determination step is performed, and in the case of different categories, the inter-priority priority determination step is performed. In any of the intra-category priority determination step and the inter-category priority determination step, if the priority of the previous event wins, the new event that occurred later is discarded and the ongoing operation pattern is changed. If the priority of the new event is higher, the execution of the ongoing action pattern is stopped and a new event is sent to the action determining means 4. When the new event is adopted as an event to be sent to the action execution means 5 by the action execution means 4 ′, the action execution means 5 starts executing the action pattern for the new event. (See Fig. 6). When events with the same priority compete, the latest reaction event is sent to the action determining means 4.

Example

Here, to give an example where the simulated biological device is a seal-type robot, in this system, by setting events for a large number of events in detail, priority of superiority or inferiority depends on the type of sensor 7 alone. It is appropriate to provide Detection from sensor 7 associated with system operation, and sensor (heat sensor, current sensor, potential sensor, etc.) 7 that detects contents associated with system abnormalities such as battery exhaustion and abnormal overheating. The priority of the event by is set higher than the event by the detection from other types of sensors 7, and all other types of sensor 7 have the same priority (see Table 5 and Table 6).

The contents of the judgment criterion table are basically set so that the event with a strong stimulus has a higher priority for the event with a weak stimulus regardless of whether the sensor 7 is the same or not. Examples of these criteria are shown in Table 5 and Table 6 above, and the relationship between the previous event and the new event that occurred later is shown in the above criteria table. Is saved.

As for the priority among sensors of the same type, for sound sensor input, (priority of event due to detection of relatively weak sound) <(priority of event due to detection of relatively strong sound), For optical sensor input, (priority of event due to detection of daily light intensity change) <(priority of event due to detection of extraordinary light intensity change (camera flashlight, etc.)), tactile sense (pressure) In the case of sensor input, (priority of event by detection of stroking) <(priority of event by detection of strong press) <(priority of event by detection of hitting), or In the case of acceleration sensor input, the priority of the event is relatively weak stimulus, such as (priority of event by detecting relatively weak vibration) <(priority of event by detecting relatively strong vibration). Who by that Iben me detection of relatively strong stimulus is set outweigh the event by the detection. The rationale is that while the senses are paralyzed by strong stimuli in ecology, weak stimuli will be difficult to perceive.

Also, in order to give attention to the words spoken by the user, it is necessary to react to speech that includes meanings such as words rather than speech that does not include meanings. Priority) <(Priority by event based on detection of speech recognition input using microphone and language recognition means (speech recognition chip, etc.)). In order to express, to react preferentially to posture changes over voice (Priority of event due to detection of voice direction input, voice recognition input, and tactile sensor input) <(priority of event due to detection of acceleration input), and strong such as earthquake In order to express instinct to detect danger when vibration is detected (priority of event by detecting voice recognition input, voice direction input, optical sensor input, sensor input, and relatively weak acceleration sensor input) The priority of each event is set in the form of (degree) <(priority of event due to detection of relatively strong acceleration sensor input).

Also, for example, when stroking an animal's body, the reaction difference between when it is boiled without striking its fur and when it is swirled, and when it is suddenly boiled, it is boiled after stroking without striking its fur. In order to express the difference from the case where the tactile sensor is input, the tactile sensor input is detected by the time difference from the back to the front, so that the boiled event (in reverse fist Event priority) and set as (event on reverse touch) <(boiled event (front order event) without countering the fur due to time difference detection from front to back at multiple locations of the tactile sensor input) However, it is possible to set a strong tendency to disregard against repulsion. Table 4 is an example of the relationship between the sensor location where a stimulus is detected, the event that occurs when the stimulus is detected, and the change in emotion / parameter. Industrial applicability

If the reaction continuation means, which is a feature of the simulated biological device according to the present invention, is adopted, the ecology of the simulated organism can be expressed more precisely, and communication that transcends a simple stimulus pole becomes possible.

Claims

The scope of the claims

1. A simulated biological apparatus that causes a plurality of actions to be generated by computer control, each of which detects a stimulus received by the simulated biological apparatus as an external parameter and generates an event, and the detection status of the external parameter To determine the reaction action for the combination of the external parameter and the internal parameter, and the reaction action in response to the event. In the simulated biological apparatus having the action execution means (5) embodied in the movement of

From the sensor (7) for detecting the stimulus, an input interface unit (2 7) for detecting the external parameter from the output of the sensor (7), and the sensor (7) An event generation unit (25) that generates a reaction event assigned to one external parameter or a combination of multiple external parameters obtained through each input interface unit (27), and the event Stimulus priority determination means (2) for determining a reaction event to be adopted by the action execution means (5) when the reaction action is determined when competition of reaction events output from the generation section (25) is detected. 6) A simulated biological apparatus characterized by comprising:

2. Performs event detection processing for verifying the presence or absence of a new reaction event occurring in the event generation section (25) and the previous reaction event during the reaction operation, and a new reaction in the event detection processing. When the event or the previous reaction event is detected, the reaction event to be adopted by the action execution means (5) when determining the reaction action is detected, and the external parameter related to each reaction event is detected. An inter-category priority determination step that is determined based on priorities given in advance for each category of sensors (7), or an intra-category priority determination step that is determined based on priorities given in advance to each reaction event The simulated biological apparatus according to claim 1, further comprising stimulation priority determination means (26) that is determined via at least one of the following.