JP2013239209A - Communication robot, robot control program, and robot control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a new area-dividing device for extracting an area that can be transmitted to humans, and to provide an area division program, an area division method and a communication robot.SOLUTION: A robot 10 comprises a CPU 60, and the CPU 60 extracts an area which is not in contact with an object, or extracts an area which is not in contact with not only the object but also a pattern of a floor by using a grid map. The robot 10 selects one area corresponding to a request of a human from among the extracted (divided) areas, and determines utterance content and determines a gesture, according to size of the selected area or positional relation with the selected area. Then, the robot 10 produces utterance with relative language expression set with respect to a directive or the specific object, and instructs the selected area to the human by executing a finger-pointing movement. The robot 10 can divide an area like the human and instruct one of the divided areas like the human.

Description

  The present invention relates to an area dividing device, an area dividing program, an area dividing method, and a communication robot, and more particularly to, for example, an area dividing apparatus, an area dividing program, an area dividing method, and an area dividing apparatus. Related to communication robots.

  An example of this type of area dividing device is disclosed in Patent Document 1. The area dividing device disclosed in Patent Document 1 divides the color space according to the division width in the axial direction of the color section determined based on the color difference between the pixels of the color image, and then the divided color space. Each pixel of the color image is labeled based on the color classification number set for each color space division area, and in each labeled pixel, an area where pixels of the same color classification number are continuous is defined as the same color area. The color image is divided every time.

  An example of this type of communication robot is disclosed in Patent Document 2. The communication robot disclosed in Patent Document 2 detects its own position, a human position, and an object position from a motion capture system. Further, the robot determines the instruction word determination model based on the distance from the human, the line of sight of the robot with respect to the base line connecting the robot and the human, and the line of sight of the human. Then, the robot performs a pointing operation so as to indicate a desired object, and utters an instruction word (“this”, “it”, “that”) determined according to the instruction word determination model.

JP 2000-67240 [G06T 7/00] JP 2006-231497 [B25J 13/00, B25J 5/00]

  For example, in an interaction that conveys a certain area, it is necessary to convey the area that can be communicated to a human by an utterance / gesture. However, in the communication robot shown in Patent Document 2, a specific object is transmitted by an utterance / gesture, and the transmission of the area itself is not handled. In addition, this communication robot cannot automatically extract a region that can be transmitted to a person from its own sensor information.

  Here, the region that can be transmitted to humans has a characteristic that is determined by an object or a pattern existing around. However, there are cases where there is no clear boundary on these areas, and areas that can be transmitted to humans simply by dividing areas based on color difference information, as in the area dividing apparatus of Patent Document 1. It is difficult to extract.

  Therefore, a main object of the present invention is to provide a novel area dividing apparatus, area dividing program, area dividing method, and communication robot.

  Another object of the present invention is to provide an area dividing device, an area dividing program, an area dividing method, and a communication robot that can extract an area that can be transmitted to a human.

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

  The first invention is a storage means for storing map information in which a certain environment where an obstacle including at least an object is registered is viewed from directly above, a plurality of map information having a center along an arbitrary straight line Drawing means for drawing a circle, enlargement means for enlarging each of a plurality of circles drawn by the drawing means until just before contacting the obstacle, and centering along the same straight line among the plurality of circles enlarged by the enlargement means A region generating unit that generates one or a plurality of regions by combining circles having a region, a region extracting unit that extracts one region from one or a plurality of regions generated by the region generating unit according to a predetermined rule, and a region An area dividing apparatus comprising an execution means for repeatedly executing an enlargement means, an area generation means, and an area extraction means for map information excluding the area extracted by the extraction means A.

  In the first invention, the area dividing device includes a storage means (64), a drawing means (60, S111, S113, S131, S155, S175), an enlargement means (60, S113, S115, S117, S131, S133, S135, S155, S157, S159), an area generation means (60, S167), an area extraction means (60, S85, S87) and an execution means (60, S93). The storage means stores map information in which a certain environment in which an obstacle including at least an object is registered is viewed from above. The drawing means draws a plurality of circles having centers along arbitrary straight lines in the map information. The enlargement means enlarges each of the plurality of circles drawn by the drawing means until just before contacting the obstacle. For example, the circle is expanded by extending the radius by a certain length, and when the circle contacts an obstacle, the radius of the circle is returned to the previous length. The region generation unit generates one or a plurality of regions by combining circles having centers along the same straight line among the plurality of circles expanded by the expansion unit. The area extracting unit extracts one area from one or a plurality of areas generated by the area generating unit according to a predetermined rule. The executing means repeatedly executes the enlarging means, the area generating means, and the area extracting means for the map information excluding the area extracted by the area extracting means.

  According to the first invention, since the circles that do not contact the obstacles are joined together, a certain range surrounded by the obstacles can be extracted as one region. That is, it can be divided into areas that can be transmitted to humans.

  A second invention is dependent on the first invention, and the obstacle includes a floor surface or a ground pattern in the environment.

  In the second invention, the obstacle includes a floor surface or a ground pattern in the environment. For example, a line appearing at a tile joint is detected. This is because it is considered that the region may be divided by such a joint when a human recognizes the region. Here, the obstacle means an object or pattern (line) that becomes an obstacle to expanding a circle.

  According to the second invention, the pattern of the floor and the ground is also included in the obstacle, so that the area can be divided with a sense close to humans.

  A third invention is according to the first or second invention, further comprising a dividing unit that divides the map information in a grid pattern, and the drawing unit is provided for each frame of the map information divided in a grid pattern by the dividing unit. Next, a plurality of circles having centers along straight lines with different angles are drawn.

  In the third invention, the dividing means (60, S41) divides the map information into a grid pattern. The drawing means draws a plurality of circles having centers along different straight lines for each frame of the map information divided in a lattice shape. That is, straight lines with different angles are set for each frame, and a plurality of circles are drawn along the straight lines.

  According to the third invention, since a plurality of circles are drawn along a straight line set for each frame obtained by dividing the map information into a lattice shape, it is possible to extract a region over the entire map information.

  The fourth invention is dependent on the first to third inventions, and the predetermined rule includes that the distance to the specific point is the shortest.

  In the fourth invention, the predetermined rule includes that the distance to the specific point is closest. This is to express the knowledge that a certain region near the speaker is recognized more accurately than a region far away. Thus, for example, a particular point may mean the position of the speaker.

  According to the fourth invention, when extracting a region in an environment where a speaker and a listener exist, the region can be divided with a sense close to that of a human.

  The fifth invention is dependent on the first to fourth inventions, and the execution means is the enlargement means and the area generation until the sum of the areas extracted by the area extraction means exceeds a certain of the entire range of the map information. The means and the area extracting means are repeatedly executed.

  In the fifth invention, the execution means repeatedly executes the enlargement means, the area generation means, and the area extraction means until the total sum of the areas extracted by the area extraction means exceeds a certain value in the entire range of the map information. For example, if there is an area where only about 10% of the entire map information area can be extracted, the extraction of the area will be terminated assuming that such an area is not worth extracting. is there.

  According to the fifth aspect of the invention, since the area extraction process is not performed more than necessary, useless processes can be omitted.

  According to a sixth aspect of the present invention, there is provided a processor of a computer having storage means for storing map information in which a certain environment in which an obstacle including at least an object is registered is viewed from directly above, along an arbitrary straight line in the map information. A drawing step for drawing a plurality of circles having a center, an enlargement step for enlarging each of the plurality of circles drawn by the drawing step until just before contacting the obstacle, and a plurality of circles enlarged by the enlargement step A region generating step for generating one or a plurality of regions by combining circles having centers along the straight line, and extracting one region from one or a plurality of regions generated by the region generating step according to a predetermined rule The region extraction step and the map information excluding the region extracted by the region extraction step are subjected to an enlargement step and a region generation step. -Up, and to perform the execution step of executing repeatedly region extraction step, a region dividing program.

A seventh invention is an area division method for a computer comprising storage means for storing map information obtained by looking down from above on a certain environment in which an obstacle including at least an object is registered, and (a) a map In the information, a plurality of circles having a center along an arbitrary straight line are drawn, (b) the circle drawn in step (a) is enlarged until just before contacting the obstacle, (c) enlarged in step (b). Among the plurality of circles formed by combining circles having centers along the same straight line to generate one or a plurality of regions, (d) 1 or 1 generated by step (c) according to a predetermined rule Extracting one region from a plurality of regions, and (e) step (d)
Is a region dividing method in which step (b), step (c), and step (d) are repeatedly executed on the map information excluding the region extracted by (1).

  In the sixth and seventh inventions, as in the first invention, it can be divided into areas that can be transmitted to humans.

  An eighth invention is a communication robot comprising the area dividing device according to any one of claims 1 to 5.

  In the eighth invention, the communication robot (10) that performs the communication loss action by at least one of the body movement and the voice is the area dividing device (24, 28, 58, 60, 64, 66, 68, 72, 74, 76, 78, 80, etc.).

  According to the eighth aspect, when providing a service, it is possible to execute an interaction using a gesture (physical action) and a voice (utterance) referring to an area that can be transmitted to a human.

  A ninth invention is a communication robot which is arranged in a certain environment and executes a communication action including at least one of body movement and voice with a human being, and is a divided region obtained by dividing the environment into a plurality of regions Storage means for storing information, area selection means for selecting one area corresponding to a human question among a plurality of areas indicated by the divided area information stored by the storage means, at least one selected by the area selection means The action determination means for determining the body movement according to the size of the area, and at least the one area selected by the area selection means, and the content of the sound according to the positional relationship between the person and the person The body motion determined by the utterance content determination means and the motion determination means is executed, and the utterance content determined by the utterance content determination means is uttered The Rukoto comprises instruction means for instructing a region selected by the region selecting means to a human, a communication robot.

  In the ninth invention, the communication robot (10) is arranged in a certain environment and executes communication behavior including at least one of body movement and voice with a human. This communication robot includes a storage means (64), an area selection means (60, S9), an action determination means (60, S11, S245), an utterance content determination means (60, S11, S253, S255, S261), an instruction means ( 60, S13). The storage means stores divided area information obtained by dividing the environment into a plurality of areas. The area selecting unit selects one area corresponding to a human question from among a plurality of areas indicated by the divided area information stored by the storage unit. For example, when asked where to put luggage or wait for a person, one area corresponding to each case is selected. The motion determining means determines the physical motion according to at least the size of one area selected by the area selecting means. That is, an instruction operation for instructing a human to the selected one area is determined. Further, the utterance content determination means determines the content of the sound in accordance with at least the positional relationship between the one area selected by the area selection means, itself, and a person. The instructing means performs the body motion determined by the motion determining means and utters the content of the utterance determined by the utterance content determining means, thereby instructing one person the one area selected by the area selecting means. .

  According to the ninth aspect, since an area corresponding to a human question is selected and designated, an interaction similar to that performed by a human can be realized.

  The tenth invention is according to the ninth invention, and when not only one area selected by the area selecting means but also another area is indicated when the body movement determined by the action determining means is executed. In addition, it further includes first current position updating means for bringing its own position closer to the one area.

  In the tenth invention, the communication robot is configured such that the first current position update unit (60, S249, S251, S259) is selected by the region selection unit when the body motion determined by the motion determination unit is executed. When not only the above-mentioned area but also other areas are pointed, the position of itself is brought close to the one area.

  According to the tenth aspect of the invention, for example, when an area is indicated by a pointing operation, it is possible to operate so that only the area is pointed by bringing its own position closer. Therefore, for example, even if an instruction word such as here, there, over there, this side, that side, or that side is uttered, it is possible to reliably indicate the selected one region. In other words, human-like interaction is possible.

  An eleventh invention is according to the ninth invention, and is a relative language information storage means for storing relative language information including an allocation area allocated to a specific object and a language expression for indicating the allocation area. When the physical motion determined by the motion determining means is executed, if not only one area selected by the area selecting means but also other areas are indicated, is one area included in the allocated area? The utterance content determination means further includes a language expression that indicates the allocation area when it is determined by the area inclusion determination means that only one area is included in the allocation area. Is determined as the audio content.

  In the eleventh invention, the communication robot further includes relative language information storage means (64) and area inclusion determination means (60, S257). The relative language information storage means stores relative language information (204g) including an allocation area allocated to a specific object and a language expression for indicating the allocation area. The area inclusion determination means, when executing the body movement determined by the movement determination means, indicates not only one area selected by the area selection means but also other areas (“YES” in S249). It is determined whether or not one area is included in the allocation area. The utterance content determination means determines, when the area inclusion determination means determines that only one area is included in the allocation area, the linguistic expression indicating the allocation area as speech content.

  According to the eleventh aspect of the invention, even when a region other than the one region selected by physical movement is indicated, when only the selected one region is included in the allocation region, Since the utterance is expressed by the linguistic expression indicating the allocation area, the selected one area can be surely indicated.

  The twelfth invention is dependent on the eleventh invention, and when it is determined by the area inclusion determination means that not only one area but also other areas are included, its own position is changed to the one area. Second current position updating means for bringing the second current position closer is further provided.

  In a twelfth aspect, the communication robot further includes second current position updating means (60, S257, S259). The second current position update unit determines that the area inclusion determination unit includes not only one area but also other areas (“YES” in S257), and changes the position of the second current position update unit to the first area. Move closer to.

  In the twelfth invention, similarly to the tenth invention, for example, when an area is indicated by a pointing operation, the user can be operated to point only at the area by bringing his / her position closer.

According to this invention, since the circles that do not contact the obstacles are joined together, a certain range surrounded by the obstacles can be extracted as one region. That is, it can be divided into areas that can be transmitted to humans.

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

FIG. 1 is a front view of the appearance of a communication robot according to an embodiment of the present invention as seen from the front. FIG. 2 is a block diagram showing an electrical configuration of the communication robot shown in FIG. FIG. 3 is an illustrative view showing object map information according to the object map information data stored in the communication robot shown in FIG. FIG. 4 is an illustrative view showing a camera image according to the camera image data stored in the communication robot shown in FIG. FIG. 5 is an illustrative view showing an example of a grid map. FIG. 6 is an illustrative view showing an example of a reference list for pairs of points and line inclinations. FIG. 7 is an illustrative view for explaining a method of determining a frame for drawing a stretchable circle. FIG. 8 is an illustrative view for explaining a drawing method of an expansion / contraction circle in a certain frame in the frame. FIG. 9 is an illustrative view showing an example of drawing an expansion / contraction circle having a center along a certain straight line and a balloon generated thereby. FIG. 10 is an illustrative view showing an example in which an expansion / contraction circle having a center is drawn along a straight line inclined by 30 ° from the example shown in FIG. 9 and a balloon generated thereby. FIG. 11 is an illustrative view showing divided areas. FIG. 12 is an illustrative view showing an instruction word switching model of this embodiment. FIG. 13 is an illustrative view for explaining a method of determining whether or not a communication robot is brought close to an area when executing a communication action referring to the area. FIG. 14 is an illustrative view showing an example of a memory map of the memory shown in FIG. FIG. 15 is an illustrative view showing an example of a relative language table. FIG. 16 is an illustrative view showing a relative language assignment area set corresponding to a fixed object included in the object map information shown in FIG. FIG. 17 is a flowchart showing the overall processing of the CPU shown in FIG. FIG. 18 is a flowchart showing a part of the area division processing of the CPU shown in FIG. FIG. 19 is another part of the CPU area dividing process shown in FIG. 2, and is a flowchart subsequent to FIG. FIG. 20 is a flowchart showing the CPU candidate set acquisition processing shown in FIG. FIG. 21 is a flowchart showing a first part of the balloon acquisition process of the CPU shown in FIG. FIG. 22 is a second part of the balloon acquisition process of the CPU shown in FIG. 2, and is a flowchart subsequent to FIG. FIG. 23 is a third part of the balloon acquisition process of the CPU shown in FIG. 2, and is a flowchart subsequent to FIG. 24 is a fourth part of the balloon acquisition process of the CPU shown in FIG. 2, and is a flowchart subsequent to FIG. FIG. 25 is a flowchart showing a part of the area selection processing of the CPU shown in FIG. FIG. 26 is another part of the CPU area selection process shown in FIG. 2, and is a flowchart subsequent to FIG. FIG. 27 is a flowchart showing a part of the CPU control information determination process shown in FIG. FIG. 28 is another part of the CPU control information determination process shown in FIG. 2, and is a flowchart subsequent to FIG.

  Referring to FIG. 1, a communication robot (hereinafter simply referred to as “robot”) 10 of this embodiment is an interaction-oriented robot, and mainly gestures with a communication target (communication target) such as a human being. The robot 10 has a function of executing communication actions including gestures (physical movements) such as hand gestures and voices, etc. The robot 10 provides various services such as reception and guidance, for example, a certain company or event venue. It is arranged in various places or situations (environments) such as (shopping malls, exhibition halls) etc. The robot 10 can run in the environment.

  As will be described in detail later, in this embodiment, the robot 10 also functions as an area dividing device.

  FIG. 1 is a front view showing the appearance of the robot 10 of this embodiment. The robot 10 includes a carriage 18, and two wheels 20 and one follower wheel 22 for autonomously moving the robot 10 are provided on the lower surface of the carriage 18. The two wheels 20 are independently driven by a wheel motor 24 (see FIG. 3), and the carriage 18, that is, the robot 10 can be moved in any direction, front, back, left, and right. The slave wheel 22 is an auxiliary wheel that assists the wheel 20. Therefore, the robot 10 can move in the arranged space by autonomous control.

  A cylindrical sensor mounting panel 26 is provided on the carriage 18, and a number of distance sensors 28 are mounted on the sensor mounting panel 26. These distance sensors 28 measure the distance from the sensor mounting panel 26, that is, the object (human being, obstacle, etc.) around the robot 10.

  The body 30 is provided on the sensor mounting panel 26 so as to stand upright. Further, as described above, it is further provided at the front center upper portion of the body 30 (a position corresponding to a human chest), and measures the distance mainly to a human in front of the robot 10. Further, the body 30 is provided with a support column 32 extending from substantially the center of the upper end of the side surface, and an omnidirectional camera 34 is provided on the support column 32. The omnidirectional camera 34 captures the surroundings of the robot 10 and is distinguished from an eye camera 58 described later. As this omnidirectional camera 34, for example, a camera using a solid-state imaging device such as a CCD or a CMOS can be adopted. It should be noted that the installation positions of the distance sensor 28 and the omnidirectional camera 34 are not limited to the relevant parts and can be changed as appropriate.

  An upper arm 38R and an upper arm 38L are provided at upper end portions on both sides of the body 30 (a position corresponding to a human shoulder) by a shoulder joint 36R and a shoulder joint 36L, respectively. Although illustration is omitted, each of the shoulder joint 36R and the shoulder joint 36L has three orthogonal degrees of freedom. That is, the shoulder joint 36R can control the angle of the upper arm 38R around each of the three orthogonal axes. A certain axis (yaw axis) of the shoulder joint 36R is an axis parallel to the longitudinal direction (or axis) of the upper arm 38R, and the other two axes (pitch axis and roll axis) are orthogonal to the axis from different directions. It is an axis to do. Similarly, the shoulder joint 36L can control the angle of the upper arm 38L around each of three orthogonal axes. A certain axis (yaw axis) of the shoulder joint 36L is an axis parallel to the longitudinal direction (or axis) of the upper arm 38L, and the other two axes (pitch axis and roll axis) are orthogonal to the axis from different directions. It is an axis to do.

In addition, an elbow joint 40R and an elbow joint 40L are provided at the tips of the upper arm 38R and the upper arm 38L, respectively. Although illustration is omitted, each of the elbow joint 40R and the elbow joint 40L has one degree of freedom, and the angles of the forearm 42R and the forearm 42L can be controlled around this axis (pitch axis).

  A sphere 44R and a sphere 44L corresponding to a human hand are fixedly provided at the tips of the forearm 42R and the forearm 42L, respectively. However, when a finger or palm function is required, a “hand” in the shape of a human hand can be used. Although not shown, the front surface of the carriage 18, the portion corresponding to the shoulder including the shoulder joint 36R and the shoulder joint 36L, the upper arm 38R, the upper arm 38L, the forearm 42R, the forearm 42L, the sphere 44R and the sphere 44L are respectively A contact sensor (shown generically in FIG. 3) 46 is provided. The contact sensor 46 on the front surface of the carriage 18 detects contact of a person or an obstacle with the carriage 18. Therefore, when the robot 10 is in contact with a human being or an obstacle during its movement, the robot 10 can detect the contact and immediately stop driving the wheel 20 to suddenly stop the movement of the robot 10. Further, the other contact sensors 46 detect whether or not the respective parts are touched. In addition, the installation position of the contact sensor 46 is not limited to the said site | part, and may be provided in an appropriate position (position corresponding to a person's chest, abdomen, side, back, and waist).

  A neck joint 48 is provided at the upper center of the body 30 (a position corresponding to a person's neck), and a head 50 is further provided thereon. Although illustration is omitted, the neck joint 48 has a degree of freedom of three axes, and the angle can be controlled around each of the three axes. A certain axis (yaw axis) is an axis directed directly above (vertically upward) of the robot 10, and the other two axes (pitch axis and roll axis) are axes orthogonal to each other in different directions.

  The head 50 is provided with a speaker 52 at a position corresponding to a human mouth. The speaker 52 is used for the robot 10 to communicate with a person around it by voice or sound. A microphone 54R and a microphone 54L are provided at a position corresponding to a human ear. Hereinafter, the right microphone 54R and the left microphone 54L may be collectively referred to as a microphone 54. The microphone 54 captures ambient sounds, particularly a human voice that is an object for performing communication. Furthermore, an eyeball portion 56R and an eyeball portion 56L are provided at positions corresponding to human eyes. The eyeball portion 56R and the eyeball portion 56L include an eye camera 58R and an eye camera 58L, respectively. Hereinafter, the right eyeball portion 56R and the left eyeball portion 56L may be collectively referred to as the eyeball portion 56. Further, the right eye camera 58R and the left eye camera 58L may be collectively referred to as an eye camera 58.

  The eye camera 58 captures a human face approaching the robot 10, other parts or objects, and captures a corresponding video signal. The eye camera 58 can be the same camera as the omnidirectional camera 34 described above. For example, the eye camera 58 is fixed in the eyeball unit 56, and the eyeball unit 56 is attached to a predetermined position in the head 50 via an eyeball support unit (not shown). Although illustration is omitted, the eyeball support portion has two degrees of freedom, and the angle can be controlled around each of these axes. For example, one of the two axes is an axis (yaw axis) in a direction toward the top of the head 50, and the other is orthogonal to the one axis and goes straight in a direction in which the front side (face) of the head 50 faces. It is an axis (pitch axis) in the direction to be performed. By rotating the eyeball support portion around each of the two axes, the tip (front) side of the eyeball portion 56 or the eye camera 58 is displaced, and the camera axis, that is, the line-of-sight direction is moved. It should be noted that the installation positions of the above-described speaker 52, microphone 54, and eye camera 58 are not limited to the portions, and may be provided at appropriate positions.

  As described above, the robot 10 of this embodiment includes independent two-axis driving of the wheel 20, three degrees of freedom of the shoulder joint 36 (6 degrees of freedom on the left and right), and one degree of freedom of the elbow joint 40 (two degrees of freedom on the left and right). It has a total of 17 degrees of freedom: 3 degrees of freedom for the neck joint 48 and 2 degrees of freedom for the eyeball support (4 degrees of freedom on the left and right).

  FIG. 2 is a block diagram showing an electrical configuration of the robot 10. Referring to FIG. 2, robot 10 includes a CPU 60. The CPU 60 is also called a microcomputer or a processor, and is connected to the memory 64, the motor control board 66, the sensor input / output board 68, and the audio input / output board 70 via the bus 62.

  The memory 64 includes a ROM, an HDD, and a RAM (not shown). In the ROM and HDD, a control program for controlling the operation of the robot 10 is stored in advance. For example, a detection program for detecting the output (sensor information) of each sensor and a communication program for transmitting / receiving necessary data and commands (messages) to / from an external computer are recorded. The RAM is used as a work memory or a buffer memory.

  The motor control board 66 is composed of, for example, a DSP, and controls driving of motors of axes such as arms, neck joints, and eyeballs. That is, the motor control board 66 receives control data from the CPU 60 and controls two motors (in FIG. 2, collectively referred to as “right eyeball motor 72”) that control the respective angles of the two axes of the right eyeball portion 56R. Control the rotation angle. Similarly, the motor control board 66 receives control data from the CPU 60 and controls two motors for controlling the respective angles of the two axes of the left eyeball unit 56L (in FIG. 2, collectively referred to as “left eyeball motor 74”). ) To control the rotation angle.

  The motor control board 66 receives control data from the CPU 60, and includes three motors for controlling the angles of the three orthogonal axes of the right shoulder joint 36R and one motor for controlling the angle of the right elbow joint 40R. The rotation angles of a total of four motors (collectively indicated as “right arm motor 76” in FIG. 2) are controlled. Similarly, the motor control board 66 receives control data from the CPU 60, three motors for controlling the angles of the three orthogonal axes of the left shoulder joint 36L, and one motor for controlling the angle of the left elbow joint 40L. The rotation angles of a total of four motors (collectively indicated as “left arm motor 78” in FIG. 2) are controlled.

  Further, the motor control board 66 receives control data from the CPU 60, and controls three motors for controlling the angles of the three orthogonal axes of the neck joint 48 (in FIG. 2, collectively indicated as "head motor 80"). Control the rotation angle. The motor control board 66 receives control data from the CPU 60 and controls the rotation angles of the two motors that drive the wheels 20 (collectively indicated as “wheel motors 24” in FIG. 2). In this embodiment, a motor other than the wheel motor 24 uses a stepping motor (that is, a pulse motor) in order to simplify the control. However, a DC motor may be used similarly to the wheel motor 24. In addition, the actuator that drives the body part of the robot 10 is not limited to a motor that uses a current as a power source, and may be appropriately changed. For example, in another embodiment, an air actuator may be applied.

  Similarly, the sensor input / output board 68 is also constituted by a DSP, and takes in signals from each sensor and gives them to the CPU 60. That is, data relating to the reflection time from each of the distance sensors 28 is input to the CPU 60 through the sensor input / output board 68. As the distance sensor 28, an infrared distance sensor or a laser distance sensor can be used, or both of them may be used. The video signal from the omnidirectional camera 34 is input to the CPU 60 after being subjected to predetermined processing by the sensor input / output board 68 as necessary. Similarly, the video signal from the eye camera 58 is also input to the CPU 60. Further, signals from the plurality of contact sensors described above (collectively indicated as “contact sensors 46” in FIG. 3) are provided to the CPU 60 via the sensor input / output board 68.

Similarly, the voice input / output board 70 is also configured by a DSP, and voice or voice according to voice synthesis data provided from the CPU 60 is output from the speaker 52. Also, voice input from the microphone 54 is given to the CPU 60 via the voice input / output board 70.

  The CPU 60 is connected to the communication LAN board 82 via the bus 62. The communication LAN board 82 is configured by a DSP and provides transmission data given from the CPU 60 to the wireless communication device 84, and sends the transmission data from the wireless communication device 84 to an external computer (such as the storage control device 102) via the network 100. Send. The communication LAN board 82 receives data via the wireless communication device 84 and gives the received data to the CPU 60. That is, the robot 10 can perform wireless communication with the storage control device 102 by the communication LAN board 82 and the wireless communication device 84.

  Further, the wireless tag reader 86 is connected to the CPU 60 via the bus 62. The wireless tag reader 86 receives a radio wave superimposed with identification information transmitted from the wireless tag (RFID tag) via an antenna (not shown). The wireless tag reader 86 amplifies the received radio wave signal, separates the identification signal from the radio wave signal, demodulates (decodes) the identification information, and supplies the identification information to the CPU 60. The wireless tag is attached to a person in a company or event venue, and the wireless tag reader 86 detects a wireless tag within a communicable range. Note that the wireless tag may be an active type or a passive type that is driven in accordance with radio waves transmitted from the wireless tag reader 86.

  As described above, the robot 10 is arranged in a certain environment and provides services such as reception and guidance in the environment. In this embodiment, a case will be described in which communication behavior is executed with a human being as a communication target in a certain environment while referring to a partial area in the provision of such a service.

  For example, when considering communication between people, when asked where to place the luggage, the person (guide) who is asked looks for an area where the luggage can be placed and points to it here (over there, over there) ) Or this side (that side, that side), etc., and tell the listener where to place the luggage (area). However, referring to objects (objects) that exist in the environment such as walls, pillars, desks, chairs, for example, instead of directives such as “next to desks”, “in front of pillars”, There are cases where a language of expression (hereinafter referred to as “relative language”) determined according to a specific object is spoken.

  In this embodiment, the robot 10 is caused to execute the same communication behavior as that of the above guider. That is, in a certain environment (space), an area is extracted (divided), an appropriate area is selected in response to a question (request) from a human, and the selected area is transmitted by communication behavior. At this time, appropriate body movements and utterance contents are also selected.

  Hereinafter, a method for extracting (dividing) a region, a method for selecting a region, and execution of communication behavior with reference to the region will be sequentially described.

  First, when an area is extracted (divided), a grid map as a base is generated. The grid map is generated based on the object map information, the camera image, and the distance information.

The object map information is a two-dimensional map (schematic diagram) as seen from directly above the environment, and an object (fixed object) that is fixedly arranged in the environment is described in its shape (size). . Although omitted in FIG. 3, in the object map information, the arrangement positions of the fixed objects are managed in two-dimensional coordinates. For example, the origin O of the two-dimensional coordinates is set at the lower left vertex of the square frame that defines the environment shown in FIG. However, in the two-dimensional coordinate system, the horizontal direction in FIG. 3 is the X-axis direction, and the vertical direction in FIG. 3 is the Y-axis direction. Also, the right direction in FIG. 3 is the positive direction of the X axis, and the upward direction in FIG. 3 is the positive direction of the Y axis.

  In this embodiment, the fixed object includes a wall, a pillar, a flower bed, a counter, a library, and the like. That is, an object that cannot move or an object that rarely moves corresponds to a fixed object.

  FIG. 4 shows an example of a camera image. For example, the camera image is an image obtained by photographing the environment in which the robot 10 is arranged with the eye camera 58 of the robot 10. The robot 10 manages its own position (current position) in the same coordinate system as the coordinate system set in the object map information in the environment, and when an image is captured by the eye camera 58, the captured image is displayed. The image is stored in the memory 64 in association with the photographed position.

  As can be seen from the camera image shown in FIG. 4, in addition to the fixed objects shown in FIG. 3, movable objects such as tables and chairs (hereinafter referred to as “movable objects”) are also arranged in the environment. Is done. Such a movable object is considered to be moved at a relatively high frequency, so it is not registered in the object map information, and is detected from the camera image and distance information as will be described later. The

  In this embodiment, the movable object includes a desk, a table, a chair, a sofa, and the like.

  When the distance is detected by the distance sensor 28 at a certain position (XY coordinates) as the robot 10 moves in the environment, the distance information is associated with the position, the distance, the direction (angle), and the like. Is information (data) related to the distance including. With such distance information, it is possible to know the existence of an object (object) and its position.

  Here, a method for detecting a movable object from a camera image and distance information will be described. In this embodiment, edge detection of image processing and segmentation are used for detection of the movable object. For edge detection, Canny's edge detection method (J. Canny. A computational approach to edge detection. IEEE Trans. On Pattern Analysis and Machine Intelligence, Vol. 8, No. 6, 11 1986.) is used. For the segmentation, information from the distance sensor 28 that detects the boundary of the object and information on the area of the object extracted as a connected area are used. In this embodiment, the connected area is a lump (area) that is detected to have the same color, and when it is determined that there is some object from the distance information for such an area, the object Recognize that there is a (movable object). In other words, when a planar constant region having the same color is detected from the camera image and the distance is detected at the position of the constant region with reference to the distance information, the constant region has a height. Therefore, it can be determined that the object is present.

  However, since the above-described fixed object is also included in the camera image, naturally, this fixed object is also detected by edge detection and segmentation.

Further, as can be seen from the camera image shown in FIG. 4, a line (floor pattern) is formed on the floor surface by a joint such as a tile. Such a floor pattern is also acquired (detected) based on the camera image. This is because when a person refers to an area, the area may be distinguished by a pattern. In this embodiment, OpenCV (OpenCV1.0. Http://opencv.willowgarage.com/wiki/.) Is used to extract the floor pattern. Further, the above-mentioned Canny edge detection method is used to extract a floor pattern using a Gaussian filter for noise removal.

  In this embodiment, since a certain indoor environment is assumed, the floor pattern is detected. However, when the environment is outdoor, the pattern of the ground (or concrete or road) is used. Is detected. However, even outdoors, in an arcade town or the like, tiles may be laid out in the same manner as the floor surface.

  However, since the movable object and the floor pattern obtained based on the camera image are detected in the image when the three-dimensional real space is viewed obliquely from above with the eye camera 58, the image is the same as the object map information. In addition, the real space is converted into a two-dimensional image (information) as seen from above. However, when the camera image in which the movable object and the floor pattern are detected is converted into a two-dimensional image, a conversion matrix created in the process of calibrating the eye camera 58 is used. Therefore, even when the camera image is distorted (distortion near the edge of the image) due to the angle of the eye camera 58 or the lens center blur at the time of shooting, it is correctly converted into two-dimensional information. .

  Then, for example, the detected movable object and the floor pattern are registered in the object map information, and the grid map is generated by subdividing the pattern into a lattice shape.

  A plurality of areas (area groups) are extracted (divided) using the grid map generated in this way. FIG. 5A shows an example of a grid map, but for simplicity, a fixed object, a movable object, and a floor pattern (hereinafter, these may be collectively referred to as “obstacles”) are omitted. is there.

  In this embodiment, each grid (frame) included in the grid is set so that one side corresponds to a length of 20 cm in the real space. Therefore, the size of each frame is also changed according to the scale of the grid map.

  In addition, as shown in FIG. 5B, identification information (index number) is attached to each frame, and balloons (regions) based on each frame are generated in the order according to the index number. The The method for generating the balloon (region) will be described in detail later.

  In this embodiment, when a grid map is generated, a reference list as shown in FIG. 6 is created corresponding to the grid map. This reference list is a list that is referred to when drawing a circle (hereinafter referred to as an “expandable circle”) C (g) that is a base for generating a balloon. As shown in FIG. 6, in the reference list, a set of the coordinate (point g) of the center of the frame and the slope d of the straight line P is described corresponding to the index number of each frame. As can be seen from FIG. 6, a straight line P having different inclinations from 0 ° to 30 ° is set for one frame (point g). In this embodiment, the inclination d is set in units of 30 ° from 0 ° to 330 °. In this embodiment, a pair of a point g and a slope d is referred to in ascending order of both the index number and the slope angle.

  The slope d may be set more finely, but the calculation becomes enormous. In an environment where objects are arranged in an orderly manner, it is considered that the inclination d may be set more roughly.

For example, as shown in FIG. 7 (A), a straight line P having a slope d of 0 ° at a point g of a certain frame of the grid.
Is set, first, an expansion / contraction circle C (g) centered on the point g is drawn. However, at the beginning of drawing the expansion / contraction circle C (g), the radius is 0, and the radius is increased step by step from that state. In this embodiment, the length corresponding to one stage is half the length of one side of one frame. Accordingly, the stretchable circle C (g) drawn first is a circle that is inscribed in the frame centered on the point g, as shown in FIG. In the case shown in FIG. 8A, when the one-step radius is further increased, the result is shown in FIG. 8B. In this way, the expansion / contraction circle C (g) is enlarged.

  In this embodiment, the radius of the expansion / contraction circle C (g) is increased by one step until it touches (impacts) an obstacle (a fixed object, a movable object, and a floor pattern) in the grid map. When the expansion / contraction circle C (g) collides with the obstacle, the radius of the expansion / contraction circle C (g) is returned to the length immediately before the collision (one step before). Similarly, for each of all the frames (points g) through which the straight line P passes along the straight line P with the inclination d, an expansion / contraction circle C (g) centered on the point g is drawn. In the example shown in FIG. 7A, the expansion / contraction circle C (g) is drawn on the hatched frame.

  In this embodiment, when one stretch circle C (g) is drawn, the stretch circle C (g) is drawn next to the frame (point g) adjacent to the frame (point g) and passing through the straight line P. And enlarged. For example, with reference to a pair of frames (point g) indicated by the reference list, a stretchable circle C (g) is sequentially arranged in a frame (point g) that is basically adjacent to the right direction (plus direction of the X axis) and passes through the straight line P. Draw. When the drawing of the expansion / contraction circle C (g) in the right direction is completed, the drawing is basically adjacent to the left direction (minus direction of the Y axis) with reference to the set of frames (point g) indicated by the reference list. Then, the expansion / contraction circle C (g) is sequentially drawn on the frame (point g) through which the straight line P passes.

  Therefore, in the case shown in FIG. 7A, when the expansion / contraction circle C (g) is drawn for the frame in which the point g is described, the expansion / contraction circle C (g) is drawn for the next frame to the right. repeat. When the drawing of the expansion / contraction circle C (g) in the right direction is finished, the expansion / contraction circle C (g) is drawn for the left adjacent frame of the frame in which the point g is described, and then the expansion / contraction circle C for the left adjacent frame is drawn. Draw (g).

  FIG. 7B shows an example in which a straight line P having an inclination d = 30 ° is set with respect to the point g shown in FIG. Also in this case, the expansion / contraction circle C (g) is drawn for all the frames through which the straight line P passes. That is, the expansion / contraction circle C is drawn for all the hatched frames. However, in the case shown in FIG. 7B, first, when the expansion / contraction circle C (g) is drawn for the frame in which the point g is described, the expansion / contraction circle C (g) is next displayed in the right adjacent frame. Drawing is performed, and an expansion / contraction circle C (g) is drawn on a frame adjacent to the upper side. As described above, it is basically adjacent in the right direction because it may be adjacent in the upward direction as shown in FIG. 8B or may be adjacent in the downward direction as described later. It is. When the drawing of the expansion / contraction circle C (g) in the right direction is finished, the expansion / contraction circle C (g) is drawn in the frame adjacent to the left of the frame in which the point g is written, and the expansion / contraction frame is displayed in the frame adjacent to the lower side thereof. C (g) is drawn.

  In this embodiment, the expansion / contraction circle C (g) is drawn for all the frames through which the straight line P passes, the point g of the adjacent frame is on the obstacle, or two expansion / contraction circles C drawn on the adjacent frame. When the radius of (g) is greatly different, the drawing of the expansion / contraction circle C (g) in the current set indicated by the reference list is ended. In this way, not only when the expansion / contraction circle C (g) is drawn for all the frames through which the straight line P passes, or when the point g of the adjacent frame is on the obstacle, two expansion / contractions drawn in the adjacent frame Even when the radius of the circle C (g) is greatly different, it is considered that the drawing of the expansion / contraction circle C (g) is terminated in such a case because the region is divided by an obstacle or the like. This is because it is not appropriate to extract as one region.

When the drawing of the expansion / contraction circle C (g) for the current set indicated by the reference list is finished, all the expansion / contraction circles C (g) drawn for the current set are combined. In this embodiment, all the frames included in all the stretchable circles C (g) drawn for the current set are filled, and one region (balloon) P (C ( g) and d) are generated.

  In this way, the drawing of the expansion / contraction circle C (g) is completed in the set of the frame point g (indicated by the index number) and the slope d of the straight line P, and the balloon P (C (g), d) When generated, drawing of the expansion / contraction circle C (g) for the next set is started. That is, the slope d of the straight line P is rotated by 30 °. However, if the slope d of the straight line P of the immediately preceding set is 330 °, the balloon P (C for the first set (the straight line P with d = 0 °) for the frame point g indicated by the next index number. The generation processes (g) and d) are executed.

  However, although omitted in FIG. 6, when the generation process of the balloon P (C (g), d) for one set is finished, a check is added to the reference list column corresponding to the set. That is, it is registered that the generation process of the balloon P (C (g), d) is finished for the group.

  FIG. 9 and FIG. 10 are diagrams illustrating examples of the result of drawing and enlarging the expansion / contraction circle C (g) focusing on a certain frame (not shown) when there are four obstacles in the environment. FIG. 9A and 9B show the case where the slope d of the straight line P is 0 °, and FIGS. 10A and 10B show the case where the slope d of the straight line P is 30 °. Show. However, in FIG. 9 and FIG. 10, for the sake of easy understanding, only a representative expansion / contraction circle C (g) is shown, but actually, a larger number of expansion / contraction circles C (g) are drawn.

  In FIG. 9A, the slope d of the straight line P is 0 °, and the two obstacles arranged above and below the straight line P are slanted. ) Is smaller. Accordingly, as shown in FIG. 9B, unlike the shape of the space (region) surrounded by the four obstacles, a balloon P (C (g), d) having a shape close to a rhombus is generated.

  On the other hand, in FIG. 10 (A), the inclination d of the straight line P is 30 °, and two obstacles are arranged above and below the straight line P so as to be parallel to this, so that these two obstacles are arranged. An expansion / contraction circle C (g) having the same or substantially the same diameter as the interval to be drawn is drawn along the straight line P. Therefore, as shown in FIG. 10B, the balloon P (C (g), d) is generated so as to coincide with or substantially coincide with the area surrounded by the four obstacles.

  Such balloons P (C (g), d) are extracted for each set of reference lists. From all the extracted balloons P (C (g), d), 1 Balloon P (C (g), d) is selected and registered (added) in a set (candidate set) Regions-part as a candidate for a region to be divided. Here, a method for selecting one balloon P (C (g), d) will be described.

  In general, it can be said that a region near a speaker is recognized more accurately than a region far away. In order to express this knowledge, the concept of energy will be used. Simply put, a small area in the vicinity has the same energy as a large area in the distance. Assuming that the distance between the current position of the robot 10 and the frame (point g) is d (g), the energy E (g) of the frame can be expressed as Equation 1.

[Equation 1]
E (g) = 1 / d (g) ²
For each extracted balloon P (C (g), d), energy E (g) is calculated for all the frames included in the balloon P (C (g), d), and the sum is also obtained. Then, one balloon P (C (g), d) having the maximum sum of energy E (g) is selected, and a candidate set R
It is registered (added) to the “egions-part”.

  When one balloon P (C (g), d) is registered in the candidate set Region-part, the remaining frames excluding the frame included in the balloon P (C (g), d) are shown in FIG. The reference list as shown in FIG. 6 is created, and the drawing process of the stretchable circle C (g) (the process of creating the balloon P (C (g), d)) is executed according to the newly created reference list. Then, for each balloon P (C (g), d), the sum of the energy E (g) is obtained, and the balloon P (C (g), d) having the maximum sum becomes the candidate set Regions-part. Registered (added). Such processing is repeatedly executed. When the balloon P (C (g), d) as a region candidate cannot be generated, the generation process of the candidate set Regions-part ends. In this embodiment, if a frame exceeding the number corresponding to 90% of the total number of frames existing in the grid map is excluded from the target for extracting (generating) the balloon P (C (g), d), the candidate The generation of the set Regions-part is completed. This is because it is considered that almost no frame that can be extracted as a region remains. As a result, execution of useless processing can be avoided.

  Such a generation process of the candidate set Regions-part is executed every time the parameter ES for edge detection from the camera image is variably set between 0 and 1. However, in this embodiment, the parameter ES is sequentially set from 0 to 0.3, 0.6, 0.9, and 1.0. When the parameter ES is set to 0, a grid map when there is no floor pattern is generated. Also, as the parameter ES increases, the floor pattern appears clearly.

  In addition, a candidate set Regions-part generated when the parameter ES is 0 is set as a parent hierarchy, and when the parameter ES is increased, an area included in the generated candidate set Regions-part (balloon P ( Of C (g), d)), a region (balloon P (C (g), d)) included in the parent layer candidate set Regions-part is registered as a child layer region. However, if the child hierarchy area included in the parent hierarchy area occupies 80% or more of the parent hierarchy area, it is determined that the area is substantially the same and is removed from the child hierarchy area. . In this way, the areas arranged in the tree structure Tree are extracted as a set Regions of divided areas. FIG. 11 shows an example of divided regions R1-R7 in the environment where the camera image shown in FIG. 4 is taken. As shown in FIG. 11, the region R1 includes R2-R4. This indicates that the region R1 is a parent hierarchy and the region R2-R4 is a child hierarchy thereof. Thus, the child hierarchy R2-R4 appears in the area R1 of the parent hierarchy because the pattern of the floor appears by increasing the parameter ES, thereby narrowing the range for drawing the expansion / contraction circle C (g). Because it becomes. However, as shown in FIG. 11, the position of the robot 10 in the environment is slightly below the regions R1 and R2.

  One region R corresponding to a question (request) from a human is selected from the set Regions of the divided regions. For example, when a person asks where to place a package, a region having an area equal to or larger than the bottom area of the package is extracted as a candidate region R_Cand from a plurality of regions included in the set Regions of divided regions. Then, one region R satisfying a predetermined condition is selected from the extracted candidates. In this embodiment, the predetermined condition is that the distance from the robot 10 (specific position) is the shortest distance. However, the predetermined condition may be that the distance to the questioned person is the shortest distance.

  When the region R is selected, the gesture (physical motion) of the robot 10 when transmitting the region R to the requesting person, the content and position of the utterance (the position of the robot 10 when performing the gesture or utterance) are determined. The That is, the control information of the robot 10 that performs the communication action with reference to the region R is determined.

  For example, in order to cause the robot 10 to execute a communication action referring to the region R, when a human refers to the region R, what gesture is used, what instruction is used, and how the region R is recognized. It was derived by experiment.

  Ten subjects (four men in their twenties and thirties and six women) participated in the experiment. In the experiment, the subject was instructed to stand 1 meter away from the experimenter at five locations in a building. At that time, the test subject and the experimenter faced the same direction. In each place, six tasks (corresponding to requests for the robot 10) that are considered to have different sizes of necessary areas were prepared. The task T1 is “put a bag”. Task T2 is “stand for people and wait”. Task T3 is “talking”. Task T4 is “put about 20 moving packages”. Task T5 is “Guiding kindergarten children that come one after another and waiting”. Task T6 is “run a radio controlled car”. First, each subject was instructed to think of an area where each task could be performed, and then to convey the area to the experimenter.

  However, in this embodiment, in order to construct an instruction word model for referring to a region, the utterance contents are “here”, “there”, “over there”, “this side”, “that side”, “that side”. It was restricted to select from six. Each subject was also instructed to use gestures.

  As a result of such an experiment, when referring to the region, the arm and the index finger are extended toward the region, and the body operation that temporarily maintains the state (referred to as “first operation” for convenience of explanation), the arm, The body motion (referred to as “second motion” for convenience of explanation) that draws a circle with the index finger to indicate the region after extending the index finger toward the region occupied the majority. Further, when the ratio between the number of times of the first operation and the number of times of the second operation was examined, the number of cases where the area was referred to by the second operation exceeded 80%. Therefore, in this embodiment, when referring to the region, the robot 10 is caused to execute the second operation. However, since the robot 10 does not have a portion corresponding to a finger, the region is indicated by the whole hand formed in a spherical shape.

  In addition, when referring to the area by the second motion, it was also derived based on the result of the above experiment that the size of the circle drawn with the index finger differs depending on the size of the area. Specifically, in the experiment, the subject referred to in the second motion, and the number of pixels in the region drawn on the photograph was counted. As a result, when the number of pixels is greater than 0 and less than or equal to 50, that is, when the size of the area viewed from the subject is small, a small circle is drawn. When the number of pixels is more than 50 and less than or equal to 400, that is, when the size of the area viewed from the subject is medium, a medium-sized circle is drawn. Furthermore, when the number of pixels is larger than 400, that is, when the size of the area seen from the subject is large, a large circle is drawn.

As for the directive, when the area of the region is larger than 1.7 m ² , the test subject frequently uses the directive with “side” (“this side”, “that side”, “that side”). high. In addition, when the area of the region is 1.7 m ² or less, the test subject frequently uses an instruction word without “side” (“here”, “there”, “over there”). Therefore, when the robot 10 is made to speak an instruction word, it can be determined whether or not to add “side” according to the size of the area to be referred to.

In addition, regarding the proper use (indicator switching) of “ko”, “so”, and “a”, the results of the above-mentioned experiment and Japanese Patent Application Laid-Open No. 2006-231497 (Japanese Patent Application Laid-Open No. 2006-231497, previously filed by the applicant of this application) It is determined according to the indicator determination model and the indicator determination method disclosed in (Reference). The instruction word switching model of this embodiment is shown in FIG. However, in this embodiment, in order to refer to the area, a point closest to the robot 10 (speaker) in the area to be referred to (hereinafter referred to as “the closest point of the area”) is replaced with the object in the above reference. I think.

Specifically, when the closest point of the region is closer to the speaker (robot 10) than to the listener (requesting person), “ko (here, this side)”, “so (there, that side) ) ”And“ Ah (that, that side) ”are determined. When the parameter _dSO is equal to or _smaller than the parameter _fKS ( _{dSL, θSO, θLO} ), it is determined as “here” or “this side”. When the parameter _dSO is equal to or less than the parameter f _SA (d _SL, θ _SO, θ _LO ), it is determined as “there” or “the side”. When the parameter _dSO is greater than or equal to the parameter f _SA (d _SL, θ _SO, θ _LO ), it is determined as “over there” or “that side”.

If the nearest point of the region is closer to the listener than the speaker, “So” and “A” are determined as follows. When the parameter d _LO is equal to or less than the parameter f _SA (d _SL, θ _SO, θ _LO ), it is determined as “there” or “the side”. When the parameter d _LO is equal to or _larger than the parameter f _SA (d _SL, θ _SO, θ _LO ), it is determined as “over there” or “that side”.

The parameter _dSO is the distance from the speaker to the nearest point of the area. The parameter d _LO is the distance from the listener to the closest point of the region. The parameter f _KS is a boundary curve between “ko” and “so”. The parameter f _SA is a boundary curve between “So” and “A”. The parameter d _SL is the distance between the speaker and the listener. The parameter θ _SO is an angle formed by a line segment connecting the speaker and the listener and a line segment connecting the speaker and the nearest point. The parameter θ _LO is an angle formed by a line segment connecting the speaker and the listener and a line segment connecting the listener and the nearest point.

  Therefore, it is determined whether or not the instruction word has “side” according to the size of the area to be referred to, and any one of “ko”, “so”, and “a” is determined according to the instruction word switching model as shown in FIG. Is determined.

  In FIG. 12, only the upper half of the “ko”, “so”, and “a” regions is shown, but the lower half of these regions is represented symmetrically with the upper half.

Next, the position of the robot 10 when referring to the region R will be described. As described above, when the robot 10 refers to the region R, the robot 10 performs the second operation. Therefore, as illustrated in FIG. 13, the robot R refers to the region R (R11 in FIG. 13) referenced by the circle drawn by the second operation. ) May be instructed to other areas (R12 in FIG. 13). In such a case, it is determined whether or not the region R to be referred to is included in the region R _Er assigned to the relative language (referred to as “relative language assigned region” for convenience of description). Here, when the area R is included in the relative language assignment area R _Er , the area R is referred to by utterance using the relative language instead of the instruction word. However, if the relative language assignment area R _Er includes the other adjacent area (R12 in FIG. 13), the area R cannot be correctly indicated. In addition, when the area R is not included in the relative allocation area R _Er , the area R cannot be indicated by utterance using a relative language.

In such a case, the position of the robot 10 is brought close to the region R so that only the region R referred to by the circle drawn by the second operation is indicated. For example, the robot 10 is brought closer to the reference region R by a predetermined distance D (in this embodiment, 20 cm), and each time whether or not only the region R to which the circle drawn by the second action refers is indicated. To be judged. However, as shown in FIG. 13, a circle or an ellipse (instruction area A _P ) obtained by projecting a circle drawn on the longitudinal axis of the arm of the robot 10 and tilted by an angle θ _P onto a two-dimensional plane. It is determined whether or not to designate only the region R or other regions. However, the angle theta _P, as described above, are varied according to the size of the region R.

  FIG. 14 shows an example of the memory map 200 of the memory 64 (RAM) shown in FIG. As shown in FIG. 14, the memory 64 includes a program storage area 202 and a data storage area 204. The program storage area 202 stores a main processing program 202a, an area division program 202b, an area selection program 202c, a control program 202d, and the like.

  The main processing program 202a is a program for processing a main routine for overall control of the robot 12. The area dividing program 202b is a program for extracting (dividing) a tree-structured area group from the grid map. The area selection program 202c is a program for selecting one area from a tree-structured area group in response to a request from a human. The control program 202d is a program for executing at least one of body movement and speech according to a behavior module of communication behavior.

  Although not shown, in the program storage area, as described above, the detection program and the communication program are stored, and the shooting program and the voice recognition program are also stored. The photographing program is a program for executing photographing processing using the omnidirectional camera 34 and the eye camera 56 and storing the photographed image (camera image) in the memory 64. The speech recognition program is a program for detecting speech uttered by a human through the microphone 54 and recognizing the detected speech by using DP matching or HMM method using dictionary data (not shown).

  The data storage area 204 is provided with a request buffer 204a. The request buffer 204a is a storage area for storing the contents of questions (requests) from humans in chronological order. Therefore, after processing a human request, the request is deleted from the request buffer 204a.

  The data storage area 204 stores map data 204b, camera image data 204c, distance data 204d, area group data 204e, instruction word switching data 204f, and relative language table data 204g.

  The map data is data about object map information in which the fixed object shown in FIG. 3 is described. The camera image data 204c is data on a camera image when the inside of the environment is imaged by the eye camera 58 of the robot 10. The distance data 204d is data on the direction (angle) and distance stored in correspondence with the position (XY coordinates).

The area group data 204e is data regarding a tree-structured area group acquired according to the area division program 202b. The instruction word switching data 204f is data on parameters (d _SO , d _LO , f _KS , f _SA ) for switching the instruction words corresponding to the instruction word switching model shown in FIG. The relative language table data 204g is data about an area set in advance for the fixed object and the utterance content corresponding to the area. This is an utterance content and a corresponding region that are used to easily distinguish the region when it is difficult to refer to the region in the instruction word. FIG. 15 shows a specific example of the relative language table. In the example shown in FIG. 15, the relative language table describes the coordinates of the four vertices and the relative language expression Er that define the relative language allocation area R _Er corresponding to the ID. For example, the relative language assignment area REr is set in advance corresponding to the fixed object in the object map information shown in FIG. 3, as indicated by the dotted frame in FIG.

Returning to FIG. 14, although not shown, the data storage area 204 stores other data necessary for the control of the robot 10 and is provided with a counter (timer) and a flag.

  FIG. 17 is a flowchart showing the entire robot control process of the CPU 60 shown in FIG. For example, when the robot 10 encounters or is touched by a human, the CPU 60 starts the entire process. As shown in FIG. 17, when starting the entire process, the CPU 60 acquires object map information in step S1. That is, the CPU 60 reads the map data 204b from a nonvolatile memory or an external computer. In the next step 3, a camera image is acquired. That is, the CPU 60 acquires an image (camera image) captured by the eye camera 58 and stores the camera image data 204c corresponding to the acquired camera image in the data storage area 204.

  The robot 10 is initially placed at a predetermined position in the environment where the XY coordinates are known, and then updates its current position based on the movement direction and the movement distance when the robot 10 is running. However, an environmental sensor may be provided so that the current position can be corrected as appropriate.

  In the next step S5, distance information is acquired. That is, the CPU 60 moves the robot 10 in advance and corresponds to the position of the robot 10 when the distance is detected by the distance sensor 28, and the distance data 204d about the direction (angle) of the robot 10 at that time and the detected distance. Is obtained from a non-volatile memory or the like and stored in the data storage area 204. This is mainly used to add information about movable objects to the grid map.

  Subsequently, in step S7, an area division process (see FIGS. 18 and 19) described later is executed. In the next step S9, an area selection process (see FIGS. 25 and 26) described later is executed. In step S11, control information determination processing (see FIGS. 27 and 28), which will be described later, is executed. In step S13, the control information (body motion G, utterance content E, position Pos) determined in step S11 is executed. That is, a communication action referring to the region R is executed.

  Subsequently, in step S15, it is determined whether there is another question (request). That is, the CPU 60 determines whether all requests according to the request data stored in the request buffer 204a have been processed. If “YES” in the step S15, that is, if there is another question, the process returns to the step S9. On the other hand, if “YES” in the step S15, the entire process is ended as it is.

  18 and 19 are flowcharts of the area dividing process in step S7 shown in FIG. As shown in FIG. 18, when starting the region group extraction process, the CPU 60 initializes the tree structure of the region in step S31. That is, the CPU 60 resets the area group data 204e. In the subsequent step S33, an edge detection parameter ES is initialized (ES = 0). However, the parameter ES is changed in the range of 0 to 1, and in this embodiment, it is changed by 0.3 as will be described later. The smaller the value of this parameter ES, the harder it is to detect the boundary of the region.

  In the subsequent step S35, edge detection is executed from the camera image with the parameter ES. In the next step S37, a floor pattern is detected. However, when the parameter ES is 0, the floor pattern is not detected. In the next step S39, the movable object is detected from the camera image and the distance information. However, the fixed object is also detected by executing the process of step S39. In step S41, a grid map is generated. That is, the information on the movable object and the floor pattern is added to the object map information, and is subdivided into a lattice shape.

  Next, a candidate set acquisition process (see FIG. 20) described later is executed in step S43, and one region Rc is extracted from the candidate set Regions-part in step S45 shown in FIG. In the next step S47, a region Rp that is an upper layer of the region Rc is searched from the tree structure Tree. That is, the CPU 60 searches for a region Rp including the region Rc.

  In step S49, it is determined whether or not the region Rc overlaps the region Rp by a threshold value or more. Here, the threshold value is 80%, and when the region Rc is in a range of 80% or more of the region Rp, it is determined that they are the same or substantially the same region. Therefore, if “YES” in the step S49, that is, if the region Rc overlaps the region Rp by a threshold value or more, the region Rc is discarded in a step S51, and the process proceeds to a step S55. On the other hand, if “NO” in the step S49, that is, if the region Rc does not overlap the region Rp by a threshold value or more, the region Rc is registered as a child hierarchy of the region Rp, and the process proceeds to a step S55.

  In step S55, it is determined whether an area remains in the candidate set Regions-part. If “YES” in the step S55, that is, if an area remains in the candidate set Regions-part, the process returns to the step S45 to extract the next area Rc. On the other hand, if “NO” in the step S55, that is, if there is no region remaining in the candidate set Regions-part, it is determined whether or not the parameter ES is 1 in a step S57.

  If “NO” in the step S57, that is, if the parameter ES is not 1, the parameter ES is increased by a predetermined value in a step S59, and the process returns to the step S35 shown in FIG. For example, the predetermined value is set to 0.3. However, if the parameter ES exceeds 1 (1.0), the value is set to 1.0. Therefore, the parameter ES is set to 0.3, 0.6, 0.9, and 1.0. However, this is an example, and the parameter ES may be increased by 0.1 or 0.2. On the other hand, if “YES” in the step S57, that is, if the parameter ES is 1, an area in the tree structure Tree is extracted as a set Regions of divided areas in a step S61, and the process returns to the entire processing. That is, an area that can be transmitted to a human is extracted (divided). Although illustration is omitted, the data of the set Regions of the divided areas is also stored in the work area of the data storage area 204.

  FIG. 20 is a flowchart of the candidate set acquisition process shown in step S43 of FIG. As shown in FIG. 20, when starting the candidate set acquisition process, the CPU 60 creates a reference list as shown in FIG. 6 corresponding to the grid map in step S81. In the next step S83, a balloon acquisition process (see FIGS. 21 to 24), which will be described later, is executed. In step S85, the energy E (P (C (g)) of all the balloons P (C (g), d). , D)). In step S87, the balloon P (C (g_max), d_max) having the maximum energy is extracted. That is, a larger balloon P (C (g), d) closer to the robot 10 is extracted.

  In the next step S89, the extracted balloon P (C (g_max), d_max) is registered in the candidate set Regions-part. Although not shown, the candidate set Regions-part is stored in the work area of the memory 64. In a subsequent step S91, the balloon P (C (g_max), d_max) is deleted from the grid map. That is, all the frames included in the balloon P (C (g_max), d_max) are excluded from the region extraction range in the grid map.

In a succeeding step S93, it is determined whether or not a frame that can extract an area remains in the grid map. In this embodiment, the CPU 60 determines whether or not the number of frames that is 10% or more of the total number of frames remains. If “YES” in the step S93, that is, if there remains a frame that can extract an area in the grid map, the process returns to the step S81. That is, index numbers are assigned to the remaining frames, and the reference list as shown in FIG. 6 is recreated. On the other hand, if “NO” in the step S93, that is, if there remains no frame that can extract an area in the grid map, the process returns to the entire process as it is.

  21 to 24 are flowcharts of the balloon acquisition process in step S83 shown in FIG. As shown in FIG. 21, when the balloon acquisition process is started, the CPU 60 sets the first set of points g and straight line P (d = 0 °) in the reference list in step S111. In the next step S113, the radius of the expansion / contraction circle C (g) centered on the point g is expanded by one step. That is, the CPU 60 extends the radius of the expansion / contraction circle C (g) by half the length of one side of one frame (cell). In a succeeding step S115, it is determined whether or not the expansion / contraction circle C (g) hits an obstacle. That is, the CPU 60 determines whether or not the expansion / contraction circle C (g) has touched at least the frame on which the obstacle is drawn.

  If “NO” in the step S115, that is, if the expansion / contraction circle C (g) does not hit the obstacle, the process returns to the step S113. On the other hand, if “YES” in the step S115, that is, if the expansion / contraction circle C (g) hits an obstacle, in step S117, the radius of the expansion / contraction circle C (g) one step before is set, and the step S119 is performed. The point g is set to the variable g-pre, and the radius of the expansion / contraction circle C (g) is set to the variable r-pre. However, variable g-pre shows the coordinate about the present point g, and variable r-pre shows the radius of the expansion-contraction circle C (g) in the point g set to the variable g-pre.

  In a succeeding step S121, it is determined whether or not the inclination d is 90 ° or 270 °. If “NO” in the step S121, that is, if the inclination d is not 90 ° or 270 °, in the step S123, the positive direction of the X axis or the Y axis direction ( A point g-cur adjacent to the plus direction or minus direction is acquired, and the process proceeds to step S127 shown in FIG. That is, in step S123, the point g-cur of the frame that is next to the frame where the point g-pre exists (next to the right side, next to the upper side, or next to the lower side) and through which the straight line P passes is acquired. . On the other hand, if “YES” in the step S121, that is, if the inclination d is 90 ° or 270 °, in the step S125, the point on the straight line P having the inclination d is a point adjacent to the positive direction of the Y axis of the point g-pre. g-cur is acquired, and the process proceeds to step S127.

  As shown in FIG. 22, in step S127, it is determined whether or not there is a point g-cur. That is, whether or not the expansion / contraction circle C (g) is drawn on all the frames adjacent to the positive direction of the X axis of the frame indicated by the current reference list on the straight line P (may be adjacent to the Y axis direction). to decide. If “NO” in the step S127, that is, if there is no point g-cur, it is determined that the drawing of the expansion / contraction circle C (g) in the plus direction of the X axis is finished, and the process proceeds to the step S143 shown in FIG. . On the other hand, if “YES” in the step S127, that is, if there is a point g-cur, it is determined whether or not the point g-cur is on an obstacle in a step S129.

  If “YES” in the step S129, that is, if the point g-cur is on the obstacle, the drawing of the expansion / contraction circle C (g) in the positive direction of the X axis is ended, and the process proceeds to the step S143. This is because it is considered that the area is divided (disconnected) by the obstacle. On the other hand, if “NO” in the step S129, that is, if the point g-cur is not on the obstacle, the radius of the expansion / contraction circle C (g-cur) centering on the point p-cur is set by one step in a step S131. Expanding. In a succeeding step S133, it is determined whether or not the expansion / contraction circle C (g-cur) hits an obstacle.

  If “NO” in the step S133, that is, if the expansion / contraction circle C (g-cur) does not hit the obstacle, the process returns to the step S131 as it is. On the other hand, if “YES” in the step S133, that is, if the expanding / contracting circle C (g-cur) hits the obstacle, the radius of the expanding / contracting circle C (g-cur) is set to the previous radius in a step S135. In step S137, the radius of the expansion / contraction circle C (g-cur) is set to the variable r-cur.

  In step S139, it is determined whether r-pre / r-cur is greater than or equal to a first threshold value (eg, 1.3). If “YES” in the step S139, that is, if r-pre / r-cur is equal to or larger than the first threshold value, it is determined that the region should be divided, and the X-axis expansion / contraction circle C (g ) Is ended, and the process proceeds to step S143. On the other hand, if “NO” in the step S139, that is, if r-pre / r-cur is less than the first threshold value, the r-pre / r-cur is set to a second threshold value (for example, 0.7) in the step S141. ) Determine if: If “NO” in the step S141, that is, if r-pre / r-cur exceeds the second threshold value, the process returns to the step S121 shown in FIG. Continue drawing g). On the other hand, if “YES” in the step S141, that is, if r-pre / r-cur is equal to or smaller than a second threshold value (for example, 0.7), it is determined that the region should be divided, and the X axis The drawing of the expansion / contraction circle C (g) in the plus direction is terminated, and the process proceeds to step S143.

  In steps S139 and S141, it is determined whether or not the radius (size) of the expansion / contraction circle C (g) drawn in the adjacent frame has changed abruptly. It is judged that it should be done.

  As shown in FIG. 23, in step S143, the frame is returned to the point g of the frame of interest indicated by the reference list. This is because the drawing of the expansion / contraction circle C (g) in the minus direction of the X axis is started following the plus direction of the X axis. The drawing process of the X-axis minus direction expansion / contraction circle C (g) is the same as the above-described drawing process of the X axis plus direction, so the overlapping contents will be described briefly. .

  In the next step S45, it is determined whether or not the inclination d is 90 ° or 270 °. If “NO” in the step S145, in a step S147, the point g-cur adjacent to the point g-pre on the straight line P having the inclination d in the minus direction of the X axis or the Y axis direction (plus direction or minus direction) is obtained. Acquire and go to step S151. On the other hand, if “YES” in the step S145, that is, the point g-cur adjacent to the point g-pre on the straight line P with the inclination d in the negative direction on the Y axis is acquired, and the process proceeds to the step S151.

  In step S151, it is determined whether or not there is a point g-cur. If “NO” in the step S151, the process proceeds to a step S167 shown in FIG. On the other hand, if “YES” in the step S151, it is determined whether or not the point g-cur is on an obstacle in a step S153. If “YES” in the step S153, the process proceeds to a step S167 as it is. On the other hand, if “NO” in the step S153, the radius of the expansion / contraction circle C (g-cur) centered on the point g-cur is expanded by one step in a step S155. In step S157, it is determined whether the expansion / contraction circle C (g-cur) hits an obstacle.

If “NO” in the step S157, the process returns to the step S155 as it is. On the other hand, if “YES” in the step S157, the radius of the expansion / contraction circle C (g-cur) is set to a previous radius in the step S159 shown in FIG. 24, and the expansion / contraction circle C (g -Cur) radius is set to the variable r-cur. In a succeeding step S163, it is determined whether or not r-pre / r-cur is not less than the first threshold value. If “YES” in the step S163, the process proceeds to a step S167 as it is. On the other hand, if “NO” in the step S163, it is determined whether or not r-pre / r-cur is equal to or smaller than a second threshold value in a step S165. If “NO” in the step S165, the process returns to the step S145 shown in FIG. 23, and the drawing of the expansion / contraction circle C (g) in the minus direction of the X axis is continued. On the other hand, if “YES” in the step S165, the process proceeds to a step S167.

  In step S167, all the expansion / contraction circles C (g) are combined to generate a balloon P (C (g), d). That is, all the frames included in the expansion / contraction circle C (g) drawn for the current set indicated by the reference list are extracted as one lump, and a balloon P (C (g), d) is generated.

  In the next step S169, it is determined whether or not the slope d of the straight line P indicated by the reference list is 330 °. If “NO” in the step S169, that is, if the slope d of the straight line P indicated by the reference list is not 330 °, the straight line P having the slope d indicated by the next group is set in the current frame point in a step S171. Then, the process returns to step S113 in FIG. That is, in step S171, a straight line P with the inclination d changed by 30 ° is set at a point g in the same frame as the immediately preceding frame.

  On the other hand, if “YES” in the step S169, that is, if the inclination d of the straight line P indicated by the reference list is 330 °, the drawing process of the expansion / contraction circle C (g) is performed on all the pairs of the reference list in a step S173. Determine if it has been executed. If “NO” in the step S173, that is, if there remains a pair that has not been subjected to the drawing process of the expansion / contraction circle C (g) in the reference list, in step S175, the frame point g and the next pair indicated by A straight line P with an inclination d is set, and the process returns to step S113. That is, in step S175, a straight line P having an inclination d = 0 ° is set at the point g of the frame indicated by the index number next to the immediately preceding set of frames. On the other hand, if “YES” in the step S175, that is, if the drawing process of the expansion / contraction circle C (g) is executed for all the pairs in the reference list, the process returns to the candidate set acquisition process.

  Although illustration is omitted, if “YES” is determined in the step S169, it is indicated in the reference list that the drawing process has been completed for the pair that has just performed the drawing process of the expansion / contraction circle C (g). Checked.

25 and 26 are flowcharts of the area selection process in step S9 shown in FIG. As shown in FIG. 25, when the area selection process is started, the CPU 60 listens to information about a package from a human in step S191. Here, the CPU 60 outputs a synthesized voice for inquiring about the size of the luggage from the speaker 52. In the next step S193, the bottom area Size of the luggage is acquired. For example, the CPU 60 outputs a voice corresponding to a human response,
By detecting through the microphone 54 and recognizing the sound, the bottom area Size is acquired from the size of the load. Although illustration is omitted, data about a table in which the bottom area corresponding to the size of the package is stored is stored, and the CPU 60 refers to the table to obtain the bottom area Size according to the size of the package. To do.

In this embodiment, the bottom area Size of the load is acquired by the robot 10 interacting with a human. However, the size of the load is estimated from the image captured by the eye camera 58, and the bottom area Size is obtained. May be obtained. Alternatively, the bottom area Size may be estimated (acquired) directly from the image captured by the eye camera 58.

Returning to FIG. 25, in the next step S195, the variable i is initialized (i = 1). This variable i is set to individually identify a plurality of regions R included in the set Regions Region of the divided regions. In the subsequent step S197, the area Ri is read from the set Regions of the divided areas. In the next step S199, it is determined whether or not the area Ri_Size of the region Ri is larger than the bottom area Size. However, the region Ri_Size is obtained by multiplying the number of frames included in the region Ri_Size by an area in real space for one frame (in this embodiment, 20 cm × 20 cm).

If “NO” in the step S199, that is, if the area Ri_Size of the region Ri is equal to or smaller than the bottom area Size, it is determined that the baggage does not have a size for placing a load, and the process proceeds to a step S203 as it is. On the other hand, if “YES” in the step S199, that is, if the area Ri_Size of the region Ri exceeds the bottom area Size, it is determined that the bag has a size for placing the luggage, and the region Ri is determined in step S201. Registration is made to the selection candidate set R_cands, and the process proceeds to step S203. That is, in step S201, the area Ri is selected as a candidate for the area R to be referred (instructed to a person).

  In step S203, the variable i is incremented by 1 (i = i + 1). In step S205, it is determined whether or not the variable i exceeds the maximum value. That is, in step S205, it is determined whether or not the determination process of step S199 has been executed for all the regions Ri registered in the divided region set Regions. If “NO” in the step S205, that is, if the variable i is equal to or less than the maximum value, the process returns to the step S197 as it is. That is, the process of step S199 is executed for the next region Ri. On the other hand, if “YES” in the step S205, that is, if the variable i exceeds the maximum value, the variable j is initialized in a step S27 shown in FIG. 26 (j = 1). This variable j is set to identify an area included in the selection candidate set R_cands.

In the next step S209, an initial value is set to the variable dmin (dmin = ∞). This variable dmin is used to detect the shortest distance between the region Rj and a person. In the subsequent step S211, the region Rj is read from the selection candidate set R_cands. In step S213, it is determined whether or not the distance dj (corresponding to the parameter d _LO described above) between the region Rj and the person is smaller than the variable dmin. However, the distance dj between the region Rj and the person is the distance between the point closest to the person among the points included in the area Rj and the current position of the person.

  If “NO” in the step S213, that is, if the distance dj between the region Rj and the person is equal to or larger than the variable dmin, it is determined that the distance dj is not the shortest distance, and the process directly proceeds to a step S219. On the other hand, if “YES” in the step S213, that is, if the distance dj between the region Rj and the person is smaller than the variable dmin, it is determined that the distance dj is the shortest distance at the present time, and the variable dmin is set in a step S215. Substituting the distance dj, in step S217, the region Rj is set (selected) as the region R that refers to, and the process proceeds to step S219.

  In step S219, the variable j is incremented by 1 (j = j + 1). In a succeeding step S221, it is determined whether or not the variable j exceeds the maximum value. That is, in step S221, it is determined whether or not the processing in step S213 has been executed for all regions Rj registered in the selection candidate set R_cands. If “NO” in the step S221, that is, if the variable j is equal to or less than the maximum value, the process returns to the step S211 and the process of the step S213 is executed for the next region Rj. On the other hand, if “YES” in the step S221, that is, if the variable j exceeds the maximum value, the process returns to the entire processing.

27 and 28 are flowcharts of the control information determination process in step S11 shown in FIG. As shown in FIG. 27, when starting the control information determination process, the CPU 60 executes an initialization process in step S241. Here, the region R is set, and the position Pos of the robot 10 when referring to the region R, the gesture G to be executed, and the utterance content E are cleared. In the next step S243, its own current position is set to Pos ′. Subsequently, in step S245, the gesture G to be executed when referring to the region R is determined. In this embodiment, the arm R is rotated so as to draw a circle, and the region R is designated. Therefore, the size of the circle drawn by hand according to the size of the region R (the arm makes a rotation axis). When the angle θ _P ) is determined and executed, motor control information for holding the arm diagonally and motor control information for rotating the arm to draw a circle are determined.

In the next step S247, a set of regions R_long that can be mistaken for the gesture G is extracted from the set Regions of divided regions. That is, as shown in FIG. 13, assuming that the gesture G has been executed, if other region (R12 in FIG. 13) other than the region R to be referred to (R11 in FIG. 13) is included, The other area (R12) is registered in the set R_wlong as an area that can be mistaken. That is, as shown in FIG. 13, indicated by the instruction area A _P, reference is to extract the region other than the (indicated) region R.

As shown in FIG. 28, in the next step S249, it is determined whether or not the region R to be referred to is included in the region R _Er in the relative language table. If “YES” in the step S249, that is, if the area R to be referred to is included in the area R _Er in the relative table, the process proceeds to a step S257 described later. On the other hand, if “NO” in the step S249, that is, if the area R _Er in the relative language table does not include the area R to be referred to, whether or not the number of elements of the set R_wlong is 0 in the step S251. Judging. If “NO” in the step S251, that is, if the number of elements of the set R_long is 1 or more, the process proceeds to a step S259.

On the other hand, if it is step S251, that is, if the number of elements of the region R_long is 0, the instruction word Ei is determined in step S253. Here, as described above, it is determined whether or not to add a “side” according to the size of the region R, and the distance d _SO between the robot 10 itself and the region R or the distance d _LO between the human and the region R is determined. In response, that is, according to the positional relationship between the closest point of the region R, the current position of the robot 10, and the current position of the human, “ko”, “so”, and “a” are determined. In step S255, the instruction word Ei determined in step S253 is set in the utterance content E, the current position Pos' is set in the position Pos, and the process returns to the entire process.

In step S257, it is determined whether or not there is a region included in the region R _Er among the regions registered in the set R_wlong. If “YES” in the step S257, that is, if there is an area included in the area R _Er among the areas registered in the set R_long, the distance to the area R that refers to the current position Pos ′ in the step S259. The process returns to step S247 shown in FIG. However, in this embodiment, the distance D is 20 cm, and in step S259, the robot 10 is not actually moved, but only the current position Pos ′ is updated, and the gesture G is updated at the current position Pos ′ after the movement. assuming executed, thereby indicated by a circle or an ellipse, that the designated area a _P obtained by projecting the circle drawn, is to extract the regions other than the region R reference.

On the other hand, if “NO” in the step S257, that is, if there is no area included in the area R _Er among the areas registered in the set R_long, the area including the area R in the utterance content E in the step S261. The relative language expression Er described in the relative language table is set corresponding to R _Er , the current position Pos ′ is set at the position Po, and the process returns to the overall processing.

  According to this embodiment, since a grid map in which obstacles including floor patterns and objects are registered is used to extract a region where a plurality of stretchable circles drawn so as not to touch the obstacles are extracted. The area can be divided in the same way as when dividing a space.

Also, according to this embodiment, when detecting edges, the parameters are changed stepwise from a level at which a floor pattern is not detected to a detectable level, so that the entire relatively large space such as a corridor is It can be extracted as a region, or it can be extracted as two or more regions divided into a plurality according to the floor pattern. That is, the area to be referred to can be appropriately changed according to the size of the area to be referred to.

  Furthermore, according to this embodiment, a plurality of regions divided according to a human request by speaking a relative language expression associated with an instruction word or a specific object and performing a pointing operation Since one area selected from the above is referred to, communication behavior similar to that of a human can be executed.

  In this embodiment, the robot generates the grid map, but the grid map may be generated by a computer other than the robot, and the robot may acquire the grid map. In such a case, a distance sensor may be provided in the environment, and another computer may acquire distance information from the distance sensor, or the distance information may be collected by the robot and transmitted to the other computer. May be.

  Further, in this embodiment, the robot divides the area, but the processing until the area is divided (steps S1 to S7 in FIG. 17) is executed by another computer, and the result (divided) The robot may execute the subsequent processing (steps S9 to S15 in FIG. 17) in response to the set of regions (Regions).

  Furthermore, in this embodiment, when the robot performs a pointing operation, the robot is operated to draw a circle (second operation). However, depending on the content of the utterance, the robot may hold the arm extended. You may make it perform insertion operation | movement (1st operation | movement). For example, the second operation is executed only when an instruction word with “side” is uttered, and the first operation is executed when uttering with an instruction word without “side” or a relative language expression. Good. Alternatively, when the utterance is made by an instruction word with “side” or relative language expression, the second operation is executed, and only when the instruction word without “side” is uttered, the first operation is executed. Good.

  Furthermore, in this embodiment, the distance information is acquired in advance by the robot using its own sensor. However, a distance sensor such as a laser range finder is set in the environment, and the distance sensor installed in the environment. You may make it acquire distance information from.

  Further, in this embodiment, when drawing an expansion / contraction circle, straight lines having inclinations of 90 ° and 270 ° are set for each frame, but these two straight lines are substantially the same straight line. Any one of them may not be set.

DESCRIPTION OF SYMBOLS 10 ... Communication robot 24, 72, 74, 76, 78, 80 ... Motor 28 ... Infrared distance sensor 34 ... Omnidirectional camera 52 ... Speaker 54 ... Microphone 58 ... Eye camera 60 ... CPU
64 ... Memory 66 ... Motor control board 68 ... Sensor input / output board 70 ... Voice input / output board 82 ... Communication LAN board 16c ... Acceleration sensor 84 ... Wireless communication device 86 ... Wireless tag reader

The present invention relates to a communication robot , a robot control program, and a robot control method , and more particularly to a communication robot , a robot control program, and a robot control method that execute a communication action with reference to a region, for example.

Another object of the present invention is to provide a novel, co Myunikeshon robot, a robot control program and the robot control method.

Another object of this invention is to select the region corresponding to the human question, since direct it, it is possible to realize the same interaction as the person is performing, co Myunikeshon robot, the robot control It is to provide a program and a robot control method .

1st invention is a communication robot which is arranged in a certain environment, and performs communication action including at least one of a body motion and a voice between humans, and is the divided area which divided the environment into a plurality of areas Storage means for storing information, area selection means for selecting one area corresponding to a human question among a plurality of areas indicated by the divided area information stored by the storage means, at least one selected by the area selection means The action determination means for determining the body movement according to the size of the area, and at least the one area selected by the area selection means, and the content of the sound according to the positional relationship between the person and the person The body motion determined by the utterance content determination means and the motion determination means is executed, and the utterance content determined by the utterance content determination means is uttered The Rukoto comprises instruction means for instructing a region selected by the region selecting means to a human, a communication robot.

In the first invention, the communication robot (10) is arranged in a certain environment and executes communication behavior including at least one of body movement and voice with a human. This communication robot includes a storage means (64), an area selection means (60, S9), an action determination means (60, S11, S245), an utterance content determination means (60, S11, S253, S255, S261), an instruction means ( 60, S13). The storage means stores divided area information obtained by dividing the environment into a plurality of areas. The area selecting unit selects one area corresponding to a human question from among a plurality of areas indicated by the divided area information stored by the storage unit. For example, when asked where to put luggage or wait for a person, one area corresponding to each case is selected. The motion determining means determines the physical motion according to at least the size of one area selected by the area selecting means. That is, an instruction operation for instructing a human to the selected one area is determined. Further, the utterance content determination means determines the content of the sound in accordance with at least the positional relationship between the one area selected by the area selection means, itself, and a person. The instructing means performs the body motion determined by the motion determining means and utters the content of the utterance determined by the utterance content determining means, thereby instructing one person the one area selected by the area selecting means. .

According to the first invention, since an area corresponding to a human question is selected and designated, an interaction similar to that performed by a human can be realized.

The second invention is dependent on the first invention, and indicates not only the one area selected by the area selecting means but also other areas when the body movement determined by the action determining means is executed. In addition, it further includes first current position updating means for bringing its own position closer to the one area.

In the second invention, in the communication robot, the first current position update means (60, S249, S251, S259) is selected by the area selection means when the body motion determined by the motion determination means is executed. When not only the above-mentioned area but also other areas are pointed, the position of itself is brought close to the one area.

According to the second aspect of the invention, for example, when an area is indicated by a pointing operation, it is possible to operate so that only the area is pointed by bringing its own position closer. Therefore, for example, even if an instruction word such as here, there, over there, this side, that side, or that side is uttered, it is possible to reliably indicate the selected one region. In other words, human-like interaction is possible.

A third invention is a relative language information storage unit that is dependent on the first invention and stores relative language information including an allocation area allocated to a specific object and a language expression for indicating the allocation area. When the physical motion determined by the motion determining means is executed, if not only one area selected by the area selecting means but also other areas are indicated, is one area included in the allocated area? The utterance content determination means further includes a language expression that indicates the allocation area when it is determined by the area inclusion determination means that only one area is included in the allocation area. Is determined as the audio content.

In the third invention, the communication robot further includes relative language information storage means (64) and area inclusion determination means (60, S257). The relative language information storage means stores relative language information (204g) including an allocation area allocated to a specific object and a language expression for indicating the allocation area. The area inclusion determination means, when executing the body movement determined by the movement determination means, indicates not only one area selected by the area selection means but also other areas (“YES” in S249). It is determined whether or not one area is included in the allocation area. The utterance content determination means determines, when the area inclusion determination means determines that only one area is included in the allocation area, the linguistic expression indicating the allocation area as speech content.

According to the third aspect of the invention, even when a region other than the one region selected by physical movement is indicated, when only the selected one region is included in the allocation region, Since the utterance is expressed by the linguistic expression indicating the allocation area, the selected one area can be surely indicated.

The fourth invention is dependent on the third invention, and when it is determined by the area inclusion determination means that not only one area but also other areas are included, its own position is changed to the one area. Second current position updating means for bringing the second current position closer is further provided.

In the fourth invention, the communication robot further includes second current position updating means (60, S257, S259). The second current position update unit determines that the area inclusion determination unit includes not only one area but also other areas (“YES” in S257), and changes the position of the second current position update unit to the first area. Move closer to.

In the fourth invention as well, as in the second invention, for example, when an area is indicated by a pointing operation, it is possible to operate so that only the area is pointed by approaching its own position.
5th invention is provided with the memory | storage means to memorize | store the division | segmentation area | region information which divided | segmented a certain environment into the several area | region, The communication action containing at least one of a body movement and a voice is performed between humans in the said environment A robot control program for a communication robot, an area selection step for selecting one area corresponding to a human question from a plurality of areas indicated by the divided area information stored in the storage means in the computer of the communication robot, , According to the size of one area selected in the area selection step, the action determination step for determining the body movement, at least according to the positional relationship between the one area selected in the area selection step, the communication robot, and the human Utterance content determination step for determining the audio content, and An instruction step for instructing a human to the one region selected in the region selection step by causing the communication robot to execute the body movement determined in the determination step and causing the communication robot to utter the content of the utterance determined in the utterance content determination step. This is a robot control program that executes
6th invention is provided with the memory | storage means which memorize | stores the division | segmentation area | region information which divided | segmented a certain environment into the some area | region, Communication which performs the communication action containing at least one of a body motion and a sound between humans in an environment A robot control method for a robot, wherein a computer of a communication robot selects (a) one area corresponding to a human question from among a plurality of areas indicated by divided area information stored in a storage unit; ) At least the body motion is determined according to the size of the one area selected in step (a), and (c) at least the one area selected in step (a), the communication robot, and the human The content of the voice is determined according to the positional relationship, and (d) the body movement determined in step (b) is communique. Causes run Deployment robot, by speech contents of spoken determined in step (c) the communication robot, instructs one of the area selected in step (a) in humans, a robot control method.

Claims (12)

Storage means for storing map information in which a certain environment in which an obstacle including at least an object is registered is viewed from above.
A drawing means for drawing a plurality of circles having a center along an arbitrary straight line in the map information;
An enlarging means for enlarging each of the plurality of circles drawn by the drawing means until immediately before contacting the obstacle;
A region generating unit that generates one or a plurality of regions by combining circles having centers along the same straight line among the plurality of circles expanded by the expanding unit;
In accordance with a predetermined rule, an area extracting means for extracting one area from one or a plurality of areas generated by the area generating means, and the expansion for the map information excluding the area extracted by the area extracting means An area dividing device comprising means, means for repeatedly executing the area generating means, and the area extracting means.   The area dividing device according to claim 1, wherein the obstacle includes a floor surface or a ground pattern in the environment. Further comprising a dividing means for dividing the map information into a grid pattern;
The area dividing device according to claim 1 or 2, wherein the drawing means draws a plurality of circles having a center along a straight line with a different angle for each frame of the map information divided in a lattice shape by the dividing means. .   The area dividing device according to claim 1, wherein the predetermined rule includes that the distance to a specific point is the shortest.   The execution means repeatedly executes the enlargement means, the area generation means, and the area extraction means until the total sum of the areas extracted by the area extraction means exceeds a certain of the entire range of the map information. The area dividing device according to claim 1. In a processor of a computer comprising storage means for storing map information obtained by looking down from above on a certain environment in which an obstacle including at least an object is registered,
A drawing step of drawing a plurality of circles having a center along an arbitrary straight line in the map information;
An enlargement step of enlarging each of the plurality of circles drawn by the drawing step until just before contacting the obstacle;
A region generating step of generating one or a plurality of regions by combining circles having centers along the same straight line among the plurality of circles expanded by the expanding step;
In accordance with a predetermined rule, an area extraction step for extracting one area from one or a plurality of areas generated by the area generation step, and the expansion for the map information excluding the area extracted by the area extraction step A region dividing program for executing an execution step of repeatedly executing a step, the region generation step, and the region extraction step. An area dividing method of a computer comprising storage means for storing map information obtained by looking down from above on a certain environment in which an obstacle including at least an object is registered,
(A) Draw a plurality of circles having a center along an arbitrary straight line in the map information;
(B) Magnify the circle drawn in step (a) until just before contacting the obstacle;
(C) Among the plurality of circles enlarged in the step (b), circles having centers along the same straight line are combined to generate one or a plurality of regions,
(D) extracting one area from one or more areas generated by the step (c) according to a predetermined rule; and (e) adding to the map information excluding the area extracted by the step (d). On the other hand, a region dividing method in which the step (b), the step (c), and the step (d) are repeatedly executed.   A communication robot comprising the area dividing device according to claim 1. A communication robot that is arranged in an environment and performs communication behavior with human beings including at least one of body movement and voice,
Storage means for storing divided area information obtained by dividing the environment into a plurality of areas;
An area selecting means for selecting one area corresponding to the human question among a plurality of areas indicated by the divided area information stored by the storage means;
Motion determining means for determining the body motion according to at least the size of one area selected by the area selecting means;
At least one region selected by the region selecting unit, and the speech content determining unit that determines the content of the voice according to the positional relationship between itself and the person, and the body determined by the motion determining unit A communication robot, comprising: an instruction unit that performs an operation and instructs the human to one region selected by the region selection unit by speaking the content of the utterance determined by the utterance content determination unit.   When the body motion determined by the motion determining means is executed, when not only the one area selected by the area selecting means but also other areas are indicated, the position of itself is brought close to the one area. The communication robot according to claim 9, further comprising first current position updating means. Relative language information storage means for storing relative language information including an assignment area assigned to a specific object and a linguistic expression for indicating the assignment area, and a physical action determined by the action determination means In this case, when including not only one area selected by the area selecting means but also other areas, area inclusion determining means for determining whether or not the one area is included in the allocated area. In addition,
The utterance content determination means determines, when the area inclusion determination means determines that only the one area is included in the allocation area, a language expression indicating the allocation area as the content of the speech. The communication robot according to claim 9.   When the area inclusion determining means determines that not only the first area but also other areas are included, the apparatus further comprises second current position updating means for bringing the position of the area closer to the first area. Item 12. The communication robot according to Item 11.

Images