KR100779088B1 - Apparatus for controlling robot and method thereof - Google Patents

Abstract

Robot control apparatus and method according to the present invention comprises a state analysis unit for determining whether to belong to a predetermined instability state by evaluating the current situation based on at least one or more perception information; And a target generation unit configured to set a behavior target of the robot based on the result of the current situation and the instability state and compare it with a predetermined value system, and then receive and correct the result of the behavior of the robot as the perceptual information. The robot can actively induce behavior by injecting the robot into a processing procedure and a value system to solve various unstable states that may occur in the situation of the user and the surrounding environment of the robot.

Description

Apparatus for controlling robot and method

1 is a block diagram showing the configuration of a robot control apparatus according to the present invention.

2 is a flowchart illustrating a process of a robot control method according to the present invention.

FIG. 3 is a flowchart illustrating the instability determination step S220 of FIG. 2 in more detail.

FIG. 4 is a step illustrating in more detail the step S230 of instructing a target action to the robot of FIG. 2.

5 is a view schematically showing how the robot control apparatus according to the present invention is actually linked with other devices.

The present invention relates to a device and a method for controlling the robot to operate with its own purpose, and more particularly, to solve the various unstable conditions that may occur in the situation of the user and the environment of the robot surroundings the robot The present invention provides an apparatus and a method capable of actively expressing the same.

Representative of the prior art in connection with the present invention is an active control technology based on desires and emotions employed in robots of the so-called Pet Robot 群, represented by AIBO of Sony, Japan. For example, Aibo has a built-in method of deriving actions by using internal and external stimuli such as hunger and scolding. According to this method, Aibo actively receives the behavior trend by repeatedly applying various arithmetic and matrix operations that can receive a plurality of perceptual evaluations and synchronization states as inputs and calculate correlations between the elements constituting the inputs. Equation 1 below shows an example of a formula for deriving a tendency of appearance from instinctive desire.

According to Equation 1, high fatigue and low curiosity indicate that the tendency of resting behavior will be enhanced. Fatigue, curiosity, and the like are continuously calculated based on the needs and emotion models of the robot and combined with the external stimulus values detected from various sensors arranged in the robot, and are input to the behavior selection arithmetic equation as described above. This allows the robot to actively express a series of actions.

By the way, an active expression method based on desire and emotion may be useful for deriving an instinctive behavior, but there is a problem that it is difficult to use for eliciting higher-level behavior. Here, the high-level acts are as follows.

First, it is an act that considers users. In other words, the robot should not be faithful to its own desires and emotions, but should be able to actively derive its services, that is, services that can enhance user's comfort, comfort, and enjoyment. For example, if the situation analysis indicates that the user is in a living room and the living room is hot and low in humidity, and the user is annoyed, the robot creates and based on the goal to make the room comfortable for the user. Must be able to provide adequate services.

Second, by continuously interpreting changes in the surrounding situation, it should be possible to determine whether an unstable situation occurs and to derive goal-oriented actions that can solve the instability that occurred. For example, if it detects a situation where the temperature in a particular area of the room is constantly rising late at night, the robot determines that an emergency has occurred and alerts or alerts the user to resolve it. You should be able to examine the area in more detail.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems and needs, and proposes an apparatus and method for providing an internal knowledge processing mechanism capable of actively expressing the behavior of the high-level object of the robot. It is.

In order to solve the above technical problem, the robot control apparatus according to the present invention includes: a state analysis unit that determines whether the device belongs to a predetermined instability state by evaluating a current situation based on at least one or more perceptual information input from the outside; And a target generation unit configured to set a behavior target of the robot based on the result of the current situation and the instability state and compare it with a predetermined value system, and then receive and correct the result of the behavior of the robot as the perceptual information. It is characterized by.

In order to solve the above technical problem, the robot control method according to the present invention includes: establishing at least one duty to be performed by a robot; Evaluating a current situation based on at least one piece of perceptual information input from the outside to determine whether the device belongs to a predetermined instability state; Instructing the action after setting the action target of the robot by comparing the current situation and the instability state determination result with the duty; And receiving and correcting the result of performing the action as the perceptual information. The method may further include an active robot control method and system.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of the present invention. For convenience of description and easy understanding, the robot control apparatus and the method according to the present invention will be described together. 1 is a block diagram showing the configuration of a robot control apparatus according to the present invention. 2 is a flowchart illustrating a process of a robot control method according to the present invention. 3 is a flowchart illustrating the instability state determination step S220 of FIG. 2 in more detail, and FIG. 4 is a step illustrating the step S230 of instructing a target action to the robot of FIG. 2 in more detail.

Robot control apparatus according to the present invention is largely composed of the state analysis unit 110 and the target generation unit 120. The state analysis unit 110 evaluates the current situation based on the at least one perceptual information input from the outside based on the built duty S210 to determine whether it belongs to a predetermined instability state (S220). The target generation unit 120 sets an action target of the robot by comparing with a predetermined value system based on the current situation and the instability state determination result of the state analysis unit 110, and then directs the action (S230). In response to the feedback of the performance of the robot, the feedback is received as the perceptual information and modified to approach the target value (S240).

First, detailed configuration blocks of the state analysis unit 110 and the target generation unit 120 will be described in detail.

State analysis unit 110 is a attention list unit 111 for storing the symbols that are the criterion that can be interpreted as the unstable state as the main cues, the situation analysis unit for encoding the perceptual information into a symbol corresponding to the one-to-one correspondence ( 113, the contextual unit 115 storing the symbolized perceptual information, and whenever the contents of the contextual unit 115 is changed, the contextual unit 115 is searched to determine whether the critical clue exists, and if inferred, A situation monitor 117 for outputting a driving signal indicating a start, and an instability analysis unit 119 for determining whether or not entering into an unstable state by performing inference based on a predetermined rule according to the drive signal.

On the other hand, the target generation unit 120 receives the determination result of the value system unit 121, the state analysis unit 110 and the current situation to store the obligation to be performed by the robot the value system unit 121 A goal setting unit 123 for setting an action target to be performed by the robot after comparing the duty with the duty, and an action selection unit 125 for outputting an action guide to the robot to perform the set action target. do.

The detailed operation will now be described under the above configuration. First, the attention list 111 stores clue information for discovering that a situation in which the robot should pay attention has occurred (S221). The state analysis unit 110 receives various perception information from the external perception means 100 including various sensors, cameras, and voice recognition devices. Perception information is input to the situation analysis unit 113, and the situation analysis unit 113 generates the situation detailed data by analyzing and encoding each of the input perception information or by combining various perception information (S223). The situation detailed data generated in this way is recorded in the situation unit 115. Some of the situation-specific data recorded in the situation unit 115 is re-entered into the situation analysis unit 113 to be modified or deleted. This is for maintaining consistency of the context-specific data content described in the context section 115. The situation unit 115 continuously receives the situation-specific data generated as a result of the situation analysis, and its contents change dynamically. The situation monitor 117 continuously monitors the change contents of the situation unit 115. The situation monitor 117 transmits the fact that the data stored in the attention list unit 111 appears in the situation unit 115 to the instability state analysis unit 119. The unstable state analysis unit 119 determines whether an unstable state has occurred by combining the contents of the attentional cues transmitted by the situation monitor 117 and the contents of the situation unit 115 (S225 to S227). When it is determined that the instability state has occurred, the instability state analysis unit 119 transmits the generated instability state to the target generation unit 120 (S229). The goal setting unit 123 of the goal generation unit 120 comprehensively determines the instability state and the contents of the situation unit 115 and reflects it on the value system unit 121 to determine which action goal to set for the user. Transfer to action selector 125. The action selector 125 selects an action that can achieve the set goal (S231 to S233). The external action specification and execution means (for example, the robot) that has received the selected action thus determines and actually performs the selected action. The execution result is input to the state analysis unit 110 again as perceptual information, and if the execution result is successful, it is determined that the target set is achieved. If the execution result is not successful, since the set goal is not achieved, a new goal is set or the goal achievement behavior is stopped through the above-described procedure (S235).

In performing the above functions, data transmitted and received between components that perform each function is data described by Knowledge Representation Language, and each component performs a series of logical inference functions based thereon. Perform the necessary judgment function. In addition, the value system unit 121 is a knowledge base that describes obligations, missions, and the like to be adhered to by the robot. The value system unit 121 is a set of data that promotes the safety of the robot behavior and sets the fundamental direction of the robot behavior.

Hereinafter, an embodiment of the present invention will be described with more specific examples. This embodiment applies the present invention to a robot that actively provides a comfortable indoor environment to a user through an indoor situation analysis.

First, the configuration of the robot device will be briefly described. The configuration of the robot and the robot operating environment shown in this embodiment is as shown in FIG. 5 is a view schematically showing how the robot control apparatus according to the present invention is actually linked with other devices.

The robot device 540 includes a sensor 545, a camera 547, an alarm 543, and a communication module 541 for measuring temperature and humidity together with a driving unit (not shown). The communication module 541 performs a wireless communication function with the home gateway 520. The home gateway 520 serves as a control center for controlling the air conditioner 530 and the electric window 510 having a remote control function, and the robot controls the devices by transmitting a function request to the home gateway 520. .

Next, an active robot control apparatus and a method of operating the method according to the present invention will be described. The temperature and humidity sensor 545 mounted on the robot device 540 continuously detects the temperature and humidity data of the space where the robot is located and transmits the data to the situation analyzer 113. The situation analysis unit 113 signs the situation using a predetermined rule. For example, if the temperature is 25 degrees or more, a symbol corresponding to 'hot' is generated. If the temperature is 10 degrees or less, a symbol corresponding to 'cool' is generated. In addition, the camera 547 of the robot continuously photographs the image and transmits the image to the situation analyzer 113. At the same time, although not shown in the drawing, recognition results obtained by various external perception means are also transmitted to the situation analysis unit 113. For example, when a specific user is recognized through image recognition, the ID of the recognized user is transmitted to the situation analysis unit 113. The situation analysis unit 113 writes to the situation unit 115 a symbol indicating which user is encountered using the received ID. Table 1 below describes the contents of the data described in the situation section 115 at any point in time.

LocatedAt (Me, 101) I (robot) is located at place 101. LocatedAt (Cheolsu, 101) The user with the ID Cheolsu is at 101. Confrented (Cheolsu, 2005-1023T10: 20: 30) Cheolsu encountered a user with an ID. Temperature (101,28) The temperature of 101 is 28 degrees. Humidity (101,80) The humidity of 101 is 80%. Hot (101) 101 is hot Very Humidity (101) 101 is very humid.

The robot issuer or the user registers a symbol such as Hot or VeryHumid in the attention list section 111, which is a clue about the instability of the environment. It also registers symbols, such as Confronted, as it may be necessary to provide services to users when they encounter them. The situation monitor 117 checks the contents of the situation unit 115 whenever the contents of the situation unit 115 changes, and determines whether there is a focused attention clue. In the case as described above, the situation monitor 117 drives the unstable state analysis unit 119 with Hot and VeryHumid as cues. The instability analysis unit 119 comprehensively analyzes the contents of the current situation unit 115 to determine whether an instability state has occurred. The unstable state analysis unit 119 determines whether an unstable state occurs by performing inference based on the following rule.

R10: Hot (? Loc) and VeryHumid (? Loc) and LocatedAt (? Someone,? Loc) and ~ CaughtCold (? Someone)

Instability (HotAndHumid, Uncomfortable,? Someone,? Loc)

R11: TooNoisy (? Loc) and LocatedAt (? Someone,? Loc) and ~ PlayingMedia (? Someone,? Loc)

Instability (Noisy, Uncomfortable, someone, loc)

Rule R10 determines that "if a place is hot and very humid and there is someone there, it is unstable because it is hot and humid." According to this rule, it is not judged that an unstable condition has occurred if no one is in the hot and humid place. In addition, R10 includes a condition (~ CaughtCold (? Someone)) that if a user in the place has a cold, he will not see it as unstable. This is because the user may catch a cold and intentionally increase the boiler temperature and turn on the humidifier. Rule R11 means that if a place is very noisy and the user is there, the user will see the state as unstable unless the user has burned media such as a movie or music.

The instability state analysis unit 119 transmits the sensed instability state to the target generation unit 120 so that an action target of the robot can be generated. The goal setting unit 123 generates a goal by combining the generated instability state, the data of the preset value system unit 121, and other situations of the situation unit 1152. First, the value system unit 121 includes the following knowledge.

R20: User (? Someone) → OughtToMake (? Someone, Comfortable)

R21: User (? Someone) → OughtToBeTo (? Someone, Polite)

R22: Uncomfortable = ~ Comfortable

R23: Impolite = ~ Polite

Rule R20 can be interpreted as a task to make the user comfortable, and Rule R21 can be interpreted as a task to behave politely to the user. The expressions R22 and R23 each represent a conflicting concept. This value system provides clue information that verifies whether the instability that occurs prevents robots from complying with their missions and explains what the robots intend to do with their intentions. The goal setting unit 123 determines that a discrepancy has occurred in the robot's value system through the following rule and what goal should be set to resolve the discrepancy.

R30: OughtToMake (? Someone,? X) and Instability (? Cause,? Y, someone,? Loc) and? X = ~? Y

-> Resolve (? Cause,? Someone,? Loc)

R31: Resolve (? Cause,? Someone,? Loc) and Resolvent (? Cause,? Goal)-> Goal (? Goal,? Someone,? Loc)

R32: Resolvent (HotAndHumid, CoolAndDry)

Rule R30 means that when a task is given through the value system and the instability makes it impossible to complete the task (as described by? X = ~? Y), the causal factor is to be resolved. Rule R31 is a rule that finds a resolving clue for a cause factor and sets it as a target. And, R32 is a fact that states that the solution to HotAndHumid is CoolAndDry. Therefore, in view of the data flows thus far, the target setting unit 123 sets the target as follows.

Goal (CoolAndDry, Cheolsu, 101)

This means "make the place l01 cool and dry, for a user with the ID Cheolsu".

Finally, the action selector 125 sets a means for achieving CoolAndDry set as a target. The action selector 125 utilizes the following knowledge.

R40: Action (OpenWindow, WindowClosed, ClearWether, CoolAndDry)

R41: Action (TurnOnAirconditioner, WindowClosed, DoorClosed, CoolAndDry)

R42: Goal (? Resolvent,? Someone,? Loc) and Action (? Action,? Precondition1,? Precondition2,? Resolvent) and? Precondition1 (? Loc) and

? precondition (? loc)-> Perform (? action,? loc)

Rules R40 and R41 dictate the means by which a certain goal can be achieved. First, the R40 suggests opening the window, indicating that the window must be closed and the weather should be pleasant. And the effect of the action is CoolAndDry. The next R41 suggests the act of turning on the air conditioner, specifying that windows and doors must be closed as a precondition.

Rule R42 determines the actions to be performed by combining goals, means of achievement, and current status. If the windows and doors are closed and the weather is cloudy, the action of choice will be to turn on the air conditioner. However, if the doors and windows are closed and the weather is clear, a conflict arises because both R40 and R41 are satisfied. In this case, the conflict selection can be resolved by extending the behavior selection policy by setting the priority among the actions as follows.

R40: Action (OpenWindow, WindowClosed, ClearWether, CoolAndDry)

R41: Action (TurnOnAirconditioner, WindowClosed, DoorClosed, CoolAndDry)

R42: Superior (OpenWindow, TurnOnAirconditioner)

R43: Goal (? Resolvent,? Someone,? Loc) and Action (? Action,? Precondition1,? Precondition2,? Resolvent) and? Precondition (? Loc) and? Precondion2 (? Loc)-> TemporaryPerform (? Action,? loc)

R44: TemporaryPerform (? Action1,? Loc) and TemporaryPerform (? Action2,? Loc) and Superior (? Action1,? Action2)-> Perform (? Action1,? Loc)

Knowledge R42 declares that the opening of windows is an act that takes precedence over turning on the air conditioner. In addition, R43 makes an intermediate decision instead of determining a final action, and when a plurality of actions are selected, R43 selects a high priority action to determine a final result.

The action thus determined is transferred to the action specification and execution means (for example, a robot) and expressed as actual physical service performance. For example, in FIG. 5, if the user is in the room and the act of opening the window due to the hot and humid reason is derived, the robot device 540 opens the electric window 510 of the room by the home gateway 520 through the communication module. The electric window 510 is opened according to the command.

One thing to note is that the robot device 540 can present the intention of performing the action to the user by storing the knowledge processing content in the process described above.

In other words, in the above example, the robot said, "We opened the window to make Cheolsu comfortable. It was because of Cheolsu's discomfort that the conflicting part of the value system. He also asked why Cheolsu felt uncomfortable. You might answer, "The room is hot and humid."

As described above, the active robot control method proposed by the present invention can actively control the robot by injecting and inducing the robot to intentionally keep the user and the environment safe and comfortable and act accordingly. You can make it possible to explain the value system of why you did it for the actions you performed.

The robot control method according to the present invention can also be embodied as computer readable codes on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD_ROM, magnetic tape, floppy disks, and optical data storage, and may also include those implemented in the form of carrier waves (e.g., transmission over the Internet). . The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

As described above, optimal embodiments have been disclosed in the drawings and the specification. Although specific terms have been used herein, they are used only for the purpose of describing the present invention and are not intended to limit the scope of the present invention as defined in the claims or the claims. Therefore, those skilled in the art will understand that various modifications and equivalent other embodiments are possible from this. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

As described above, the robot control apparatus and method thereof according to the present invention inject a robot into a robot by injecting a processing procedure and a value system to solve various unstable states that may occur in a situation in which the user is present and a situation of the robot's surrounding environment. This can actively induce action.

As a result, the robot can actively preserve the user's safety and solve the instability of the environment, and the robot maker or the user can easily give the robot duties, duties, and manners through the value system.

It can also be made to explain how and why the robot performed the action.

In addition, it is possible to improve the user-friendliness and stability of the robot by making it more human-like in deriving the active and intellectual behavior of the robot, which can be one of the driving forces to expand the personal and home robot industry.

Claims (9)

A state analysis unit for determining whether the current state is included in a predetermined instability state based on at least one piece of perceptual information input from the outside; And Based on the present situation and the result of the unstable state determination, a goal generation is set by comparing the result of the performance of the robot with the perception information after setting the behavior goal of the robot by comparing with the value system that defines the performance obligation of the robot. Including; And transmitting and receiving data described in a Knowledge Representative Language between the state analyzer and the target generator. delete The method of claim 1, wherein the state analysis unit Attention list unit for storing the symbols that are the criterion that can be interpreted as the unstable state as the main cues; A situation analysis unit for encoding the perceptual information into a symbol corresponding to one-to-one and outputting the symbol; A situation unit which stores the encoded perceptual information; A situation monitor for searching for the situation unit and checking whether there is a key point of interest whenever the contents of the situation unit are changed, and outputting a driving signal instructing to start inference if present; And And an unstable state analysis unit for determining whether the vehicle enters an unstable state by performing inference based on a predetermined rule according to the driving signal. The method of claim 1, wherein the target generation unit A value system unit for storing duties to be performed by the robot; A goal setting unit which receives the determination result of the state analysis unit and the current situation, accesses the value system unit, compares the duty system, and sets an action target to be performed by the robot; And And a behavior selection unit for outputting a behavior instruction to the robot to perform the set behavior goal. (a) establishing at least one or more duties the robot should perform; (b) evaluating the current situation based on at least one of the perceptual information input from the outside to determine whether it belongs to a predetermined instability state; (c) instructing the action after setting the action target of the robot by comparing the current state and instability state determination result with the duty; And (d) receiving and correcting the result of performing the action as the feedback information; The robot control method, characterized in that the obligation, the current situation and the instability state determination results are described in Knowledge Representative Language. The method of claim 5, wherein step (b) (b1) storing the symbols that serve as the criterion for determining the unstable state as key cues; (b2) encoding the perceptual information into a one-to-one corresponding symbol and storing the perceptual information as situation data; (b3) starting to infer if the situation data includes the critical clue in comparison with the critical clues whenever the situation data is changed; And (b4) performing the inference based on a predetermined rule to determine whether or not entering into an unstable state; robot control method comprising a. The method of claim 5, wherein step (c) (c1) storing the duty that the robot should perform; (c2) setting an action target to be performed by the robot after comparing the result of the unstable state determination with the current situation; And (c3) giving a behavior instruction to the robot to perform the set behavior goal. The method of claim 7, wherein step (c1) Robot control method, characterized in that for storing the obligation in the form of knowledge. (a) storing at least one or more duties for the robot to perform; (b) evaluating the current situation based on at least one of the perceptual information input from the outside to determine whether it belongs to a predetermined instability state; (c) instructing the action after setting the action target of the robot by comparing the current state and instability state determination result with the duty; And and (d) receiving and modifying the result of performing the action as the perceptual information. A computer-readable recording medium having recorded thereon a program that can be executed in a computer, the robot control method comprising: a;

Images