Determination and control of activities of an emotional system

• US 20040093121A1
• Filed: 03/06/2003
• Published: 05/13/2004
• Est. Priority Date: 11/11/2002
• Status: Active Grant

• Current Assignee: Alfred Schurmann
• Original Assignee: Alfred Schurmann
• Inventors: Schurmann, Alfred
• Granted Patent: US 7,024,277 B2
• Time in Patent Office: Days
• US Class Current: 700/245
• CPC Class Codes: G06N 3/004 Artificial life, i.e. computing arrangements simulating life

Abstract

A method and subsystem, called method DKA, for determining and controlling activites of an emotional system, aS, which belongs to the class of autonomous, motivated agents and robots, is described. The method DKA determines the current motivation of the system aS to execute an activity. This motivation is determined by stimulus patterns in situation models, and intensities of satisfactions and desires, with regard to needs of the system aS. Priorities of activities of aS are determined by motivations. DKA controls sub-activities which, by orders of the system aS, are executed by other agents/robots. The method DKA can judge which objects, situations and activities are good (at present) and which are bad for aS. The used internal representation of the world of aS can contain very abstract situation models.

Emotional systems, steered by the method DKA, are especially suitable: for attending ill people, as artificial servants making housemaid works for routine services for clients.

Claims

1. Claimed is a method and a subsystem, called in the claims below “

method DKA”

, for determination and control of activities of a non-biological emotional system, denoted in the claims below by aS, comprising;

i. the priority of execution of an activity or a command depends on the current motivation of the said emotional system aS to execute this activity or this command, respectively, where this motivation depends on needs, satisfactions and desires of the system aS, and goal situations of the said activity (said command respectively);

ii. the method DKA uses activity models (called also activity schemas) which contain sub-activities, situation models, goal situations and stimulus patterns, where the said situation models belong to the below (in (v)) said graph SeG;

iii. the control of execution of said activity or command is realized by execution of sub-activities and control-activities of the activity, or the command, respectively;

iv. the priority of execution and control of said sub-activity depends on the motivation of the systems aS to execute this sub-activity, where this motivation depends on needs, satisfactions and desires of the system aS, and goal situations of the activity in which the said sub-activity occurs;

v. the method DKA depends on the following components of the system aS;

(a) needs of the systems aS, (b) intensities of satisfactions and desires of the system aS, with regard to the said needs of aS, (c) sub-system WP of the system aS for perception of objects and simple situations of the world of the system aS, (d) a set of said activity schemas which the system aS can execute, (e) a graph, SeG, (called below semantic graph) which contains the models of objects and situations appearing in the world of the system aS;

vi. the object and situation models in the said semantic graph SeG contain stimulus patterns with regard to the said needs of the system aS;

on the basis of these stimulus patterns and the intensities of satisfactions and desires of the system aS, the method DKA determines stimulus intensities of the perceived objects and situations;

on the basis of the said stimulus intensities, the method DKA (thus, the system aS) can judge how good or how bad is an object or a situation for the system aS, at present time;

vii. each said intensity of satisfaction and desire of the system aS concerns a said need of the system aS.

2. Claimed is a method for determining the said, in claim 1, current motivation of the said system aS to execute a said, in claim 1, activity, comprising:

i. this motivation is determined by;

(a) the said, in claim 1, stimulus intensities of the goal and end situations of the said activity, (b) the intensities of satisfactions and desires of the system aS, at the present time and in the future, with regard to the said, in claim 1, needs of the system aS, (c) the said, in claim 1, needs of the system aS;

ii. the said stimulus intensity of a goal or end situation of said activity is determined by;

(a) the said, in claim 1, stimulus patterns in this goal or end situation, (b) the intensities of satisfactions and desires of the system aS, at the present time and in the future, with regard to the said, in claim 1, needs of the system aS;

iii. on the basis of the said motivation to execute an activity, the method DKA (thus also the system aS) can judge how good or how bad is the execution of an activity for the system aS, at the present time.

3. Claimed is a method for determining the said, in claim 1, motivation to execute a command comprising:

this motivation is determined by;

(a) the said, in claim 1 and 2, motivation of the said system aS to execute the activity which the said command concerns, (b) the said, in claim 1, needs of the system aS, especially by the following two needs;

WA(Pa)—

the need to perform and fulfil the commands and wishes of an authorized person, Pa, or an authorized system, Pa, AN—

the need for recognition, acknowledgment and self-esteem, (c) the said, in claim 1, satisfactions and desires of the system aS, especially by the desire concerning the said need WA(Pa) and the desire for getting recognition and acknowledgment from the said Pa.

4. Claimed are the said, in claim 1, activity schemas (or activity models, or in short:

activities) comprising;

i. an activity schema contains a graph, GA;

the nodes of this graph are the said, in claim 1, situation models, and the arcs of this graph are the said, in claim 1, sub-activities, where these situation models belong to the said, in claim 1, semantic graph SeG, and sub-activity can contain a control-activity if this sub-activity should be executed by an other object (machine, agent, robot, computer or an other system);

ii. said, in (i), control-activity of a said sub-activity is executed by the system aS in order to control the execution of the said sub-activity (by an other object);

iii. some nodes of the said graph GA are goal situations of the said activity schema;

iv. when the method DKA decides to execute an activity schema, then the said goal situations of the activity schema have positive stimulus intensities, for the system aS;

v. the method DKA determines the priority of an activity by the said, in claim 1 and 2, motivation to execute this activity;

vi. an activity scheme contains phrases—

texts in a natural language—

the meaning of which is the activity schema;

on the basis of these phrases, the method DKA can name and express activities by words and simple sentences in a natural language.

5. Claimed is a method for determining the said, in claim 1, motivation to execute a said, in claim 1 and 4, sub-activity which occurs in a said, in claim 1 and 4, activity, comprising:

i. the said motivation to execute a sub-activity is determined by;

(a) the said, in claim 1 and 2, stimulus intensities of the goal and end situations of the said activity, (b) the said, in claim 1 and 2, needs of the system aS, (c) the said, in claim 1 and 2, intensities of satisfactions and desires of the system aS, at the present time and in the future, with regard to the said needs of the system aS;

ii. on the basis of the said motivation to execute a sub-activity, the method DKA (thus also the system aS) can judge how good or how bad is the execution of a said sub-activity for the system aS, at present time.

6. Claimed is a method and a subsystem for said, in claim 1 and 4, control of execution of a said, in claim 1, 4 and 5, sub-activity by an other object (machine, agent, robot, computer or an other system), comprising:

i. when a said sub-activity is not personally executed by the system aS, then this sub-activity can contain a said, in claim 4, control-activity—

in this case, the method DKA (thus also the system aS) controls the execution of the sub-activity by an other object (e.g. agent, machine, robot);

ii. the said, in (i), control-activity of a said sub-activity contains;

(a) situation models for possible situations which can appear when the said sub-activity is being executed—

as results of this execution, (b) elementary activities (e.g. motion activities, observation activities) which the system aS executes in order to recognize the said, in (a), situations and, eventually, to correct the execution of the said sub-activity by the said object;

iii. the current priority of execution of the said control-activity is determined by the duration of the execution of this sub-activity after the last control, and by the said, in claim 1 and 5, motivation to execute the sub-activity in which the said control-activity occurs;

iv. immediate after execution of the said control-activity, the said (in (iii)) priority equals zero ;

this priority increases after that and can exceed the value of the motivation to execute the said, in (iii), sub-activity.

7. Claimed is a subsystem of an emotional system aS, called below subsystem WP, for perception of objects and simple situations, comprising:

i. the sensors of said subsystem WP are directed to these objects and situations, OS, in the surrounding of the subsystem WP (thus also of the system aS), for which the system aS, at present time, has high intensities of desires for attention (des(aS,AUw(OS),t) in the patent description) and curiosity/knowledge (des(aS,NEw(OS),t) in the patent description) with regard to the said objects and situations OS;

ii. the said subsystem WP builds, by the internal representations of objects which the subsystem WP actually have perceived, a tree, TOS, such that;

(a) each node, KtOS(O), of the tree represents an object, O, which has been recognized by the subsystem WP, (b) when the subsystem WP has recognized that an object, O, is in object Q, then node KTOS(O) is a successor of the node KTOS(Q) (which represents the object Q ), (c) when it is perceived that a situation is in object Q then this fact is registered in the node KTOS(Q);

iii. the subsystem WP uses a graph, SeG, which represents the structure and relations of the object and situation models which represent the surrounding (the world) of the emotional system aS;

an object or situation model, OSM, in graph SeG, contains phrases such that (a) each phrase in the model OSM is a word, a simple sentence or a pattern of a simple sentence in a natural language, (b) the meaning of the said phrases in model OSM is given (described) by the model OSM;

this enables;

the subsystem WP to perceive objects and situations only by the said phrases, the system aS to communicate by simple sentences in a natural language.