Patent Application
20190065930
The present invention relates to an emotion identifying system, a system and a computer-readable medium.
A terminal that studies conversations between a user and another person that the user is talking to on the phone and accumulates, in a reply table, replies from the other person on the phone to questions from the user has been known (please see Patent Literature 1, for example). In addition, an emotion generating apparatus including a neural network that receives an input of user information, equipment information and a current emotional state of itself to output the next emotional state has been known (please see Patent Literature 2, for example). In addition, a technique to store spatiotemporal patterns in an associative memory including a plurality of electronic neurons having a layer neural network relation having directional artificial synapse connectivity has been known (please see Patent Literature 3, for example).
[Patent Literature 1] Japanese Patent Application Publication No. 2011-253389
[Patent Literature 2] Japanese Patent Application Publication No. H10-254592
[Patent Literature 3] Japanese Translation of PCT International Patent Application No. 2013-535067
Conventionally, there has been a drawback in a process of identifying an emotion of a target object that the state where it becomes easier for a particular type of emotion to emerge depending on particular situations cannot be handled appropriately.
Various embodiments of the present invention may be described with reference to flowcharts and block diagrams whose blocks may represent (1) steps of processes in which operations are performed or (2) units of apparatuses responsible for performing operations. Certain steps and units may be implemented by dedicated circuitry, programmable circuitry supplied with computer-readable instructions stored on computer-readable media, and/or processors supplied with computer-readable instructions stored on computer-readable media. Dedicated circuitry may include digital and/or analog hardware circuits and may include integrated circuits (IC) and/or discrete circuits. Programmable circuitry may include reconfigurable hardware circuits comprising logical AND, OR, XOR, NAND, NOR, and other logical operations, flip-flops, registers, memory elements, etc., such as field-programmable gate arrays (FPGA), programmable logic arrays (PLA), etc.
Computer-readable media may include any tangible device that can store instructions for execution by a suitable device, such that the computer-readable medium having instructions stored therein comprises an article of manufacture including instructions which can be executed to create means for performing operations specified in the flowcharts or block diagrams. Examples of computer-readable media may include an electronic storage medium, a magnetic storage medium, an optical storage medium, an electromagnetic storage medium, a semiconductor storage medium, etc. More specific examples of computer-readable media may include a floppy (registered trademark) disk, a diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an electrically erasable programmable read-only memory (EEPROM), a static random access memory (SRAM), a compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a BLU-RAY(registered trademark) disc, a memory stick, an integrated circuit card, etc.
Computer-readable instructions may include assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, JAVA (registered trademark), C++, etc., and conventional procedural programming languages, such as the “C” programming language or similar programming languages.
Computer-readable instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, or to programmable circuitry, locally or via a local area network (LAN), wide area network (WAN) such as the Internet, etc., to execute the computer-readable instructions to create means for performing operations specified in the flowcharts or block diagrams. Examples of processors include computer processors, processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc.
Hereinafter, (some) embodiment(s) of the present invention will be described. The embodiment(s) do(es) not limit the invention according to the claims, and all the combinations of the features described in the embodiment(s) are not necessarily essential to means provided by aspects of the invention.
The robot 40a, robot 40b and robot 40c are arranged at home, for example. A user 50a is a user of the robot 40a. The user 50a is a family member of the home at which the robot 40a is arranged or the user 50a is another person visiting the home. The robot 40a interacts with the user 50a through a conversation or the like. Likewise, each of the robot 40b and robot 40c interacts with a corresponding one of a user 50b and a user 50c through a conversation or the like. The robot 40a, robot 40b and robot 40c can be applied to usage forms in which they are arranged at a reception desk of a shop or office or the like, and serves visiting customers, for example. Usage forms of the robot 40a, robot 40b and robot 40c are not limited to these usage forms.
The server 60 is provided remotely from the robot 40a, robot 40b and robot 40c. The server 60 can control the robot 40a, robot 40b and robot 40c through a communication network 90. For example, the server 60 acquires, through the communication network 90, sensor information detected at the robot 40a, decides an emotion of the robot 40a or operation that the robot 40a is caused to perform based on the acquired sensor information, and issues an instruction to the robot 40a by transmitting control information through the communication network 90.
The robot 40b and robot 40c have approximately the same functions as those of the robot 40a. In an explanation of the system 10, the robot 40a, robot 40b and robot 40c are collectively referred to as a robot 40 in some cases. In addition, in the system 10, processes of identifying an emotion of the robot 40 and deciding operation thereof are executed by the server 60. However, those processes may be partially or entirely executed by the robot 40.
The robot 40 acquires various types of information detected by sensors such as information about sounds and images of a user 50 and external forces received by the robot 40, and transmits the information to the server 60. The server 60 uses a neural network (NN) to generate control information for controlling the robot 40 based on the information acquired from the robot 40.
As influence information to influence identification of an emotion, the server 60 uses information corresponding to an amount of an endocrine substance in a living body to identify an emotion from the generated input information. For example, the server 60 uses a neural network that uses, as parameters, secretion amounts of endocrine substances in a brain such as an amount of dopamine, an amount of noradrenalin, an amount of vasopressin or an amount of serotonin to identify an emotion. In the neural network, an increase in an amount of dopamine may influence generation of an emotion classified as “pleased”. In addition, an increase in an amount of noradrenalin may influence generation of an emotion classified as “anger”. In addition, an increase in an amount of serotonin may have an influence in a direction to suppress the intensity of an emotion such as “pleased” or “anger”. A neural network determines what kind of emotion becomes easier to emerge, based on the total sum of the secretion amounts of these endocrine substances.
A process of identifying an emotion is explained specifically. The server 60 generates information to be input to the neural network based on sensor information acquired from the robot 40. In addition, the server 60 decides secretion amounts of endocrine substances based on the sensor information. For example, if the remaining battery amount is small, the amount of noradrenalin is increased. The server 60 uses the neural network in which artificial neurons are linked at coupling coefficients influenced by endocrine substances, to identify an emotion of the robot 40.
Here, if for example the continuous operation time of the robot 40 becomes long, and the robot 40 is in a situation as if it lacks rest, the server 60 controls a process related to the neural network such that it becomes easier for an emotion corresponding to the paranoid-schizoid position in psychology to emerge. For example, the server 60 increases the coupling coefficient of artificial synapses to influence artificial neurons corresponding to an emotion belonging to the paranoid-schizoid position. In addition, the server 60 expands the range occupied by an emotion corresponding to the paranoid-schizoid position on an emotion map on which a plurality of emotions are mapped. An emotion belonging to the paranoid-schizoid position refers to an emotion representing a psychological state that easily leads to a behavior of a schizoid mechanism, for example.
The server 60 transmits, to the robot 40, control information including information indicating an emotion of the robot 40 identified using the neural network and information indicating an emotion to be expanded on the emotion map. The robot 40 displays the emotion map based on the control information received from the server 60, and highlight-displays the emotion of the robot 40 identified by the server 60 on the emotion map. In addition, if an emotion belonging to the paranoid-schizoid position is identified, the server 60 may include, in the control information, an instruction to cause the robot 40 to show a behavior corresponding to the paranoid-schizoid position like a schizoid mechanism. Thereby, the robot 40 can be caused to be in human-like psychological states. In this manner, the system 10 allows appropriate handling of the fact that it becomes easier for a particular emotion to be generated depending on situations in which the robot 40 is put, in the process of identifying an emotion of the robot 40.
The control target 160 includes a display unit mentioned below. The control target 160 includes a speaker. The control target 160 also includes motors to drive movable parts such as limbs or a head part of the robot 40, or the like.
The sensor unit 120 has various types of sensor such as a microphone, a gyro sensor, a motor sensor, a camera, a battery remaining capacity sensor or an infrared sensor. The microphone of the sensor unit 120 acquires ambient sounds. For example, the microphone of the sensor unit 120 acquires sounds of the user 50. The camera of the sensor unit 120 captures images using visible light and generates image information about moving images or still images. The infrared sensor of the sensor unit 120 detects objects around it using infrared rays. The gyro sensor of the sensor unit 120 detects the angular velocities of the entire robot 40 and each unit of the robot 40. The motor sensor of the sensor unit 120 detects the rotation angles of drive axes of motors to drive movable parts of the robot 40. The battery remaining capacity sensor of the sensor unit 120 detects the remaining capacity of a battery provided to the robot 40.
The sensor unit 120 outputs, to the information processing unit 130, various types of sensor data such as sound data acquired by the microphone, images captured by the camera, angular velocities detected by the gyro sensor, rotation angles detected by the motor sensor, a remaining capacity detected by the battery remaining capacity sensor or object information detected by the infrared sensor. The information processing unit 130 supplies the communicating unit 102 with an acquired sensor signal, and causes it to transmit the sensor signal to the server 60. In addition, based on control information acquired from the server 60, the information processing unit 130 causes the robot 40 to utter through the speaker or causes limbs of the robot 40 to operate.
Next, the functional block configuration of the server 60 is explained. The server 60 has a processing unit 270, a communicating unit 202 and a storage unit 280. The processing unit 270 includes a secretion information generating unit 200, an input information generating unit 210, an adjusting unit 220, an emotion identifying unit 260, a control unit 250 and a determining unit 252. The emotion identifying unit 260 has an NN operation unit 230 and a judging unit 240. The communicating unit 202 has an information acquiring unit 204.
The communicating unit 202 is responsible for communication with the robot 40. The communicating unit 202 may be a communication device such as a network IF. The storage unit 280 has a storage medium such as a hard disk drive or a flash memory. In addition, the storage unit 280 has a volatile storage device such as a RAM. The storage unit 280 stores data required for execution of processes by the processing unit 270 or the like, beside program codes to be read out by the processing unit 270 at the time of execution and various types of transient data.
The information acquiring unit 204 acquires information for deciding an emotion of the robot 40. For example, the information acquiring unit 202 acquires, through the network 90, information detected at the sensor unit 120 of the robot 40. The judging unit 240 judges whether or not judgement factors associated respectively with a plurality of emotions are each met, based on the information acquired by the information acquiring unit 202.
If the information acquired by the information acquiring unit 202 matches a predetermined condition about a predetermined type of emotion, the adjusting unit 220 expands the range occupied by the predetermined type of emotion on a multi-emotion map. For example, the determining unit 252 determines whether or not the information acquired by the information acquiring unit 202 matches a predetermined condition about a predetermined type of emotion. If the determining unit 252 determines that the information acquired by the information acquiring unit 202 matches a predetermined condition, the adjusting unit 220 expands the range occupied by the predetermined type of emotion on the multi-emotion map.
The emotion map is one example of a space in which emotions are mapped. In addition, the predetermined type of emotion is an emotion corresponding to the paranoid-schizoid position, for example. In addition, the predetermined condition is a predetermined condition for determining that the robot 40 is in a predetermined situation corresponding to the paranoid-schizoid position. The predetermined type of emotion may be an emotion indicating a psychological situation to make transitions between the paranoid-schizoid position and the depression position.
The emotion identifying unit 260 identifies an emotion which is among a plurality of emotions and for which the judging unit 240 judges that an associated judgement factor is met. The control unit 250 outputs information indicating the emotion identified by the emotion identifying unit 260 and the ranges occupied respectively by the plurality of emotions on the emotion map. For example, the control unit 250 causes information indicating the emotion and the ranges to be transmitted from the communicating unit 202 to the robot 40.
Based on the information acquired by the information acquiring unit 204, the adjusting unit 220 further adjusts operation parameters to influence judgement about one or more judgement factors which are among a plurality of judgement factors and associated with emotions the ranges of which are to be expanded on the emotion map, such that it becomes easier for the one or more judgement factors to be met. Thereby, if it is determined based on the information acquired by the information acquiring unit 204 that the robot 40 is in a situation corresponding to the paranoid-schizoid position, it is possible to make it easier for an emotion corresponding to the paranoid-schizoid position to be identified as an emotion of the robot 40.
The plurality of judgement factors may be a plurality of artificial neurons included in the neural network to which input information based on the information acquired by the information acquiring unit 204 is input. Any emotion among a plurality of emotions may be associated with each among the plurality of artificial neurons. The emotion identifying unit 260 may identify an emotion associated with a firing artificial neuron among the plurality of artificial neurons among the plurality of emotions. The operation parameters may be coupling coefficients of artificial synapses included in the neural network. The adjusting unit 220 increases the coupling coefficients of strong coupling artificial synapses connected to one or more artificial neurons associated with an emotion having a range which is to be expanded on the emotion map.
The secretion information generating unit 200 generates secretion information indicating the secretion amounts of one or more endocrine substances based on the information acquired by the information acquiring unit 204. Then, the adjusting unit 220 adjusts the coupling coefficients based on the secretion amounts of the endocrine substances indicated by the secretion information, and further adjusts the coupling coefficients connected to one or more judgement factors associated with the predetermined type of emotion if the range occupied by the predetermined type of emotion on the emotion map is to be expanded. Thereby, in addition to adjustment of synapse coupling according to internal secretion, adjustment can be made such that it becomes easier for a particular type of emotion to emerge depending on the situation of the robot 40.
The control unit 250 highlight-displays the range on the emotion map occupied by an emotion identified by the emotion identifying unit 260 on the emotion map. Thereby, the user 50 can easily recognize an emotion of the robot 40. In addition, if it is determined that the robot 40 is in a situation corresponding to the paranoid-schizoid position, the range occupied by the emotion corresponding to the paranoid-schizoid position on the emotion map is displayed as an expanded range. Thereby, the feeling of the robot 40 can be recognized easily.
In addition, the control unit 250 controls the robot 40 according to an emotion identified by the emotion identifying unit 260. For example, if the emotion identifying unit 260 decides an emotion “pleased” as an emotion of the robot 40, the control unit 250 generates control information for causing the robot 40 to show a behavior expressing pleasure and causes the control information to be transmitted to the robot 40. In addition, the control unit 250 may generate control information for causing the speaker of the robot 40 to output cheerfully sounding sounds and cause the robot 40 to engage in a conversation with a user using the sounds. On the other hand, if an emotion corresponding to the paranoid-schizoid position is identified as an emotion of the robot 40, the control unit 250 causes control information for causing the robot to show a behavior corresponding to the paranoid-schizoid position like a schizoid mechanism to be transmitted from the communicating unit 202 to the robot 40. Thereby, the feeling of the robot 40 can be appropriately expressed through behaviors of the robot 40 or the like.
The functions of each unit of the server 60 may be realized by a computer. For example, the processing unit 270 may be realized by a processor such as an MPU or the like, and the storage unit 280 may be realized by a recording medium such as a non-volatile memory. The storage unit 280 may store a program to be executed by a processor. By the processor executing the program, the secretion information generating unit 200, input information generating unit 210, adjusting unit 220, NN operation unit 230, judging unit 240, emotion identifying unit 260, determining unit 252 and control unit 250 may be implemented, and control of the storage unit 280 may be realized. The program may be read out of a recording medium 290 such as an optical disc by the processor and stored in the storage unit 280, or may be provided to the server 60 through a network and stored in the storage unit 280. The storage unit 280 and recording medium 290 may be computer-readable non-transitory recording media.
The neural network which is a target of operation of the NN operation unit 230 includes artificial neurons which are assigned to the respective emotions shown on the emotion map 300. The neural network also has a plurality of artificial neurons for input assigned to, respectively, a first input and a second input positioned at the innermost part of the concentric circles on the emotion map 300. The respective artificial neurons for input assigned to the first input and the second input receive inputs of input information generated by the input information generating unit 210 based on the information acquired by the information acquiring unit 204. Then, artificial neurons are connected by artificial synapses generally from inner parts toward outer parts to form the neural network. It may be determined by a design whether input information is input to artificial neurons for input assigned to the first input, input to artificial neurons for input assigned to the second input, or input to both the artificial neurons for input assigned to the first input and the artificial neurons for input assigned to the second input. The artificial neurons for input assigned to the first input are generally linked to artificial neurons corresponding to emotions positioned on the left side of the emotion map 300. Because of this, if input information is input to the artificial neurons for input assigned to the first input, it becomes easier for emotions positioned on the left side of the emotion map 300 to be generated. In addition, the artificial neurons for input assigned to the second input are linked to artificial neurons corresponding to emotions positioned on the right side of the emotion map 300. Then, if input information is input to the artificial neurons for input assigned to the second input, it becomes easier for emotions positioned on the right side of the emotion map 300 to be generated.
Based on the input information, the NN operation unit 230 performs operation of the neural network repeatedly, and the judging unit 240 judges the firing state of each artificial neuron. The emotion identifying unit 260 judges an emotion of the robot 40 based on the firing state of each artificial neuron. For example, the emotion identifying unit 260 judges an emotion to which a firing artificial neuron is assigned as one emotion felt by the robot 40.
On the emotion map 300, emotions generally representing comfortable feeling are arranged on the top side of the concentric circles. In addition, emotions representing generally uncomfortable feeling are arranged on the bottom side of the concentric circles. Here, the emotion map 300 includes emotions belonging to the paranoid-schizoid position. Emotions belonging to the paranoid-schizoid position are generally emotions arranged in regions extending from the center of the concentric circles along the horizontal axis of the concentric circles. On the emotion map 300, emotions belonging to the paranoid-schizoid position are “hesitant”, “confused/disturbed”, “awful”, “scared (awfully)” and the like. The regions extending from the center of the concentric circles along the horizontal axis of the concentric circles are regions where comfortable feeling arranged on the top side above the center of the concentric circles and uncomfortable feeling on the bottom side below the center of the concentric circles are mixed. That is, the paranoid-schizoid position can be said to represent a psychological state which is chaotic and formless. Emotions at the paranoid-schizoid position shown in
If the robot 40 encounters a situation to fall into the paranoid-schizoid position, the server 60 performs control such that it becomes easier for an emotion corresponding to the paranoid-schizoid position to emerge. In addition, if the robot 40 encounters such a situation, the feeling of the robot 40 is presented in an easier-to-understand manner by causing the robot 40 to display the emotion map on which the ranges occupied by emotions such as “hesitant” corresponding to the paranoid-schizoid position are expanded. Specific details of these processes are mentioned below.
E1 indicates input information based on a detection signal. The artificial neuron N1 is an artificial neuron for input. The artificial neuron N1 receives inputs of n input information E11 to input information En1 respectively generated based on detection signals from sensors.
The artificial synapse S12 is an artificial synapse connecting the artificial neuron N1 and the artificial neuron N2. In particular, the artificial synapse S12 is an artificial synapse for inputting an output from the artificial neuron N1 to the artificial neuron N2. The artificial synapse S14 is an artificial synapse connecting the artificial neuron N1 and the artificial neuron N4. In particular, the artificial synapse S14 is an artificial synapse for inputting an output from the artificial neuron N1 to the artificial neuron N4. Assuming that j and k are integers, an artificial synapse for inputting an output from an artificial neuron N to an artificial neuron Nk is denoted as an artificial synapse Sjk.
Here, assuming that i is an integer, each artificial neuron is denoted with Ni. Ni has, as parameters, Si representing the status of Ni, Vim representing the internal state of an artificial neuron represented by Ni, and Ti representing a threshold of firing of Ni. In addition, an artificial synapse Sjk has a coupling coefficient BSjk as a parameter. In the present embodiment, artificial neurons are collectively referred to as an artificial neuron N in some cases, omitting their suffixes. In addition, artificial synapses are collectively referred to as an artificial synapse S in some cases, omitting their suffixes. Likewise, parameters of artificial neurons are also collectively referred to as internal information Vm, a threshold T or a status S in some cases, omitting their suffixes.
The status S, internal state Vm and threshold T of the artificial neuron N are parameters that can be updated over time. The status S is information related to the firing state of a neuron, and at least indicates whether the artificial neuron N is in a firing state or non-firing state. The internal state Vm is information related to the membrane potential of a neuron, and is one example of parameters representing the internal state or output of the artificial neuron N.
In addition, a coupling coefficient BS which is a parameter of the artificial synapse S is a parameter that can be updated over time. The coupling coefficient BS is information related to the plasticity of a synapse, and indicates the strength of coupling between artificial neurons N coupled by the artificial synapse S.
The NN operation unit 230 updates the above-mentioned parameters in the neural network according to the input information, and calculates the internal state Vm of each artificial neuron N. In the present embodiment, the artificial neuron N gets to have the status S of being in a “firing” state if the internal state Vm exceeds the threshold T. If in the firing state, a predetermined signal is output from the artificial neuron N for a predetermined length of time. After the passage of a predetermined length of time, the status S of N returns to “non-firing”.
Here, contents of operation performed by the NN operation unit 230 are explained more specifically taking N2 as an example. The NN operation unit 230 calculates an input I2 to N2 according to BS12×Vm1×f(S1)+BS42×Vm4×f(S4). Here, f(S) is a function that gives 0 if S is a value representing non-firing, and gives 1 if S is a value indicating a rising phase or falling phase. f(s) corresponds to a model in which a synapse transmits an action potential only if a neuron fire. f(s) may be 1. This corresponds to a model in which a membrane potential is transmitted regardless of the firing state of a neuron. A function corresponding to another transmission model of membrane potentials may be applied as f(s).
Generally speaking, the NN operation unit 230 calculates an input Ii to Ni according to ΣjBSji×Vmj×f(Sj)+ΣjΣji. The NN operation unit 230 uses BSji, Vmj, Sj and Ej at the current clock time to calculate inputs Ii, Si and the like Ni at the next clock time. The NN operation unit 230 repeats this temporally repetitively to decide the status S of each artificial neuron N in real-time. Then, the judging unit 240 judges the status S of each artificial neuron N, and the emotion identifying unit 260 judges an emotion of the robot 40 based on a result of judgement about the judging unit 240. For example, if an artificial neuron to which an emotion “pleased” in
Here, the adjusting unit 220 adjusts the coupling coefficient BS of the artificial synapse S based on the information acquired from the robot 40. For example, if it is detected that the remaining capacity of a storage battery that the robot 40 has which indicates its remaining amount is 50% or lower, the secretion information generating unit 200 increases the secretion amount of “noradrenalin” as an internal variable. Then, the adjusting unit 220 adjusts the coupling coefficient BS of the artificial synapse S associated with “noradrenalin” based on the secretion amount of “noradrenalin”. As mentioned below, generation of “noradrenalin” is set to increase the coupling coefficient BS of the artificial synapse S on the path of firing of artificial emotion neurons corresponding to emotions such as “anxiety” or “anger”, for example. Thereby, “noradrenalin” has an influence in the direction to make it easier for emotions such as “anxiety” or “anger” to be generated.
The secretion amounts of internal secreted substances are associated with the coupling coefficients BS of particular artificial synapses S. Thereby, according to information acquired at the robot 40, it is possible to change, through the secretion amounts of internal secreted substances, how easily signals are transmitted at artificial synapses S at respective locations in the neural network. Because of this, it becomes possible to generate variety of emotions from information acquired at the robot 40.
As illustrated in the figure, values associated with the increase factor of BS14 and the increase factor of the BS45 increase as the amount of noradrenalin increases. On the other hand, values associated with the increase factor of BS43 decrease as the amount of noradrenalin increases. Thereby, for example, in the neural network shown in
Here, in the example explained, the coupling coefficient BS of the artificial synapse S is adjusted in a direction to make it easier for an artificial neuron N at the output destination to fire. However, increase factors may be set such that the coupling coefficient BS of the artificial synapse S can be adjusted in a direction to make it harder for the artificial neuron N at the output destination to fire. For example, if an artificial synapse S is a strong coupling one, it is possible to make it harder for an artificial neuron N at the output destination to fire by making an increase factor smaller. On the other hand, if an artificial synapse S is connecting artificial neurons N with inhibitory coupling, it can be made harder for an artificial neuron N at the output destination to fire by making the increase factor larger, and it can be made easier for the artificial neuron N at the output destination to fire by making the increase factor smaller.
The adjusting unit 220 refers to the coupling coefficient correspondence information to adjust a corresponding coupling coefficient BS by an amount corresponding to the total secretion amount of each internal secreted substance. Thereby, the adjusted amount of a coupling coefficient BS can be adjusted in a complicated manner based on information acquired at the robot 40, and eventually, emotion artificial neurons can be caused to fire in a variety of combinations. Moreover, by associating the relationship between information acquired at the robot 40 and endocrine substances with the relationship between each endocrine substance and a coupling coefficient BS while giving meaning to them as if the subject is a human, natural emotions which are usually felt by humans can be generated.
With reference to
The storage unit 280 stores information in which identification information and an amount of increase of each of BS12, B23, B36, B14, BS45 and BS56 are associated, in association with the condition “amount of serotonin>threshold”. The storage unit 280 stores information in which identification information and an amount of increase of each of BS12, B23, B36, B14, BS45 and BS56 are associated, in association with the condition “duration of lack of rest>threshold”. In addition, likewise, the storage unit 280 stores information in which identification information and an amount of increase of each of BS12, B23, B36, B14, BS45 and BS56 are associated, in association with the condition “duration of absence-of-family-member state>threshold”.
The determining unit 252 acquires the endocrine amount of serotonin from the secretion information generating unit 200, refers to the correspondence information shown in
The determining unit 252 measures the duration of lack of rest based on information acquired by the information acquiring unit 204. Specifically, the determining unit 252 determines whether or not the robot 40 is in operation based on information acquired by the information acquiring unit 204, and if the continuous operation time of the robot 40 exceeds a predetermined length of time, starts measuring the duration of lack of rest. Then, the determining unit 252 refers to the correspondence information shown in
In addition, the determining unit 252 measures the duration of absence-of-family-member state based on information acquired by the information acquiring unit 204. Specifically, the determining unit 252 determines whether or not people recognized based on information acquired by the information acquiring unit 204 include a person that the robot 40 learned as being a family member. Then, if a state occurs where it is determined that a person learned as being a family member is not included, the determining unit 252 starts measuring the duration of absence-of-family-member state. Then, the determining unit 252 refers to the correspondence information shown in
For example, the adjusting unit 220 increases BS56 by 0.5.
Thereby, coupling coefficients BS of artificial synapses S on the path from the artificial neuron N1 to the artificial neuron N3 are increased, and it becomes easier for the artificial neuron N3 to fire if information is input to the second input. Because of this, it becomes easier to identify, as an emotion of the robot 40, an emotion “hesitant” belonging to the paranoid-schizoid position on the emotion map 300 in
In
At Step 806, the input information generating unit 210 and secretion information generating unit 200 acquire information received by the information acquiring unit 204 from the robot 40. At Step 808, the input information generating unit 210 generates information to be input to artificial neurons for input assigned to the first input and information to be input to artificial neurons for input assigned to the second input. At Step 810, the secretion information generating unit 200 calculates secretion amounts of endocrine substances for example based on the correspondence information explained with reference to
At Step 811, the determining unit 252 determines whether or not paranoid-schizoid conditions are met, based on the information acquired at Step 806. If it is determined that the paranoid-schizoid conditions are met (Y), the process proceeds to Step 830, and if it is determined that the paranoid-schizoid conditions are not met (N), the process proceeds to Step 812. If the paranoid-schizoid conditions are met, at Step 830, the adjusting unit 220 decides amounts of increase of the coupling coefficients BS. For example, the adjusting unit 220 decides the amounts of increase corresponding to the paranoid-schizoid conditions that are met, as explained with reference to
Subsequently, at Step 832, the adjusting unit 220 decides an expansion amount of each emotion belonging to the paranoid-schizoid position. For example, the longer the duration of lack of rest is, the larger the expansion amount of the range occupied by the emotion corresponding to the paranoid-schizoid position decided by the adjusting unit 220 is. In addition, the longer the duration of absence-of-family-member state is, the expansion amount of the range occupied by the emotion corresponding to the paranoid-schizoid position decided by the adjusting unit 220 may be. The adjusting unit 220 reduces the ranges occupied by other emotions according to the expansion amount of the range occupied by the emotion corresponding to the paranoid-schizoid position. After the process at Step 832, the process proceeds to Step 812.
At Step 812, the adjusting unit 220 calculates the coupling coefficient BS of each artificial synapse S. For example, if NO in the determination at Step 811, the adjusting unit 220 decides each coupling coefficient BS based on the endocrine amount decided at Step 810. On the other hand, if YES in the determination at Step 811, the adjusting unit 220 decides each coupling coefficient BS according to the sum of the coupling coefficient BS decided based on the endocrine amount decided at Step 810 and the increase/decrease amount decided at Step 830.
Subsequently, at Step 814, the NN operation unit 230 calculates an input I to each artificial neuron N according to the formula explained with reference to
Subsequently, at Step 820, the judging unit 240 judges a firing artificial neuron N based on the internal state Vm and threshold T of each artificial neuron N. At Step 822, the emotion identifying unit 260 judges an emotion of the robot 40 based on the firing artificial neuron. Thereby, the emotion identifying unit 260 assigns an emotion corresponding to the firing artificial neuron as an emotion of the robot 40. The emotion identifying unit 260 may determine that the robot 40 is more intensely feeling an emotion corresponding to an artificial neuron N whose internal state Vm is lager among firing artificial neurons. Subsequently, at Step 824, the control unit 250 control each unit of the robot 40 based on the emotion judged at Step 822.
At S830, the server 60 determines whether or not to end the loop. For example, if it is instructed to end the emotion generating process, the server 60 determines to end the loop. If the loop is not to be ended, the process returns to S804, and calculation for the still next temporal step is performed. If the loop is to be ended, this flow is ended.
The control unit 250 of the server 60 causes the display unit 162 of the robot 40 to display an object 900 corresponding to the emotion map 300. Specifically, the control unit 250 causes information indicating the range occupied by each emotion in each concentric circle and the emotion name to be transmitted in the control information to the robot 40. In addition, the control unit 250 causes information identifying an emotion corresponding to a firing artificial neuron N to be transmitted in control information.
The information processing unit 130 of the robot 40 causes the display unit 162 to display the object 900 based on the received control information. If artificial neurons corresponding to “relieved”, “at peace” and “encouraged” fire, as shown in the object 900, the information processing unit 130 highlights the ranges occupied by “relieved”, “at peace” and “encouraged” more than the ranges occupied by other emotions, and makes them displayed on the display unit 162. For example, the information processing unit 130 may display, on the display unit 162, the ranges occupied by emotions “relieved”, “at peace” and “encouraged” in colors different from those of the ranges occupied by the other emotions.
The control unit 250 of the server 60 decides the range occupied by each emotion in each concentric circle based on an expansion amount of the range occupied by the emotion received from the adjusting unit 220. Then, the control unit 250 causes information indicating the decided range occupied by each emotion and the emotion name to be transmitted in the control information to the robot 40. In addition, the control unit 250 causes information identifying an emotion corresponding to a firing artificial neuron N to be transmitted in control information.
The information processing unit 130 of the robot 40 causes the display unit 162 to display an object 1000 based on the received control information. Because the paranoid-schizoid conditions are met in the example shown in
In addition, because the paranoid-schizoid conditions are met, the ranges occupied by emotions “hesitant” and “confused/disturbed”, and “awful” and “scared (awfully)” belonging to the paranoid-schizoid position are expanded. On the other hand, the ranges occupied by emotions “feeling affection”, “happy”, “encouraged”, “scared”, “sad”, “unforgivable”, “sorrow”, “horrible”, “longing” and “pleased” that are mapped on a concentric circle together with emotions “hesitant” and “awful” are displayed as shrunken regions according to expansion of “hesitant” and “awful”. In addition, the ranges occupied by emotions “proud”, “kind”, “reproachful”, “regret”, “hate”, “unendurable”, “guilty” and “wonderful” mapped on a concentric circle together with emotions “confused/disturbed” and “scared (awfully)” are displayed as shrunken regions. Because of this, the user 50 can easily recognize a situation or feeling of the robot 40.
As has been explained above, with the system 10, the robot 40 encounters a situation to fall into the paranoid-schizoid position, it becomes easier for an emotion corresponding to the paranoid-schizoid position to emerge. In addition, because on the display unit 162 of the robot 40, the emotion map on which the ranges occupied by emotions corresponding to the paranoid-schizoid position are expanded is displayed, it is possible to present the feeling of the robot 40 in an easier-to-understand manner. In this manner, the system 10 allows appropriate handling of the state where it becomes easier for a particular type of emotion to emerge depending on situations, in a process of identifying an emotion of the robot 40.
In the system 10 according to the above-mentioned embodiment, if the paranoid-schizoid conditions are met, control is performed such that not only the ranges occupied by emotions corresponding to the paranoid-schizoid position are expanded, but also it is made easier by the adjusting unit 220 for emotions corresponding to the paranoid-schizoid position to emerge. However, even if the paranoid-schizoid conditions are met, it may be not made easier by the adjusting unit 220 for emotions corresponding to the paranoid-schizoid position to emerge.
In addition, in the system 10 according to the above-mentioned embodiment, emotions are assigned to respective artificial neurons N. However, emotions may not be assigned to artificial neurons N. In this case, if the paranoid-schizoid conditions are met, more artificial neurons N are associated to emotions of the paranoid-schizoid position by expanding the ranges occupied by the emotions corresponding to the paranoid-schizoid position. Thereby, control can also be performed such that it becomes easier for emotions corresponding to the paranoid-schizoid position to be generated.
In addition, in the system 10 according to the above-mentioned embodiment, if the paranoid-schizoid conditions are met, the ranges occupied by emotions corresponding to the paranoid-schizoid position are expanded. However, instead of expanding the ranges occupied by emotions corresponding to the paranoid-schizoid position, the emotion map 300 may be rotated about the center of the concentric circles on the emotion map 300.
The functions of the server 60 that are explained above may be implemented by one or more computers. At least some functions of the server 60 may be implemented a virtual machine. In addition, at least some of the functions of the server 60 may be implemented by cloud computing. In addition, the robot 40 is one example of a target object. Various forms other than robots may be adopted as the form of target objects. For example, target objects may be vehicles such as automobile or motorcycle or home appliances, or the like.
While the embodiments of the present invention have been described, the technical scope of the invention is not limited to the above described embodiments. It is apparent to persons skilled in the art that various alterations and improvements can be added to the above-described embodiments. It is also apparent from the scope of the claims that the embodiments added with such alterations or improvements can be included in the technical scope of the invention.
The operations, procedures, steps, and stages of each process performed by an apparatus, system, program, and method shown in the claims, embodiments, or diagrams can be performed in any order as long as the order is not indicated by “prior to,” “before,” or the like and as long as the output from a previous process is not used in a later process. Even if the process flow is described using phrases such as “first” or “next” in the claims, embodiments, or diagrams, it does not necessarily mean that the process must be performed in this order.
10: system; 40: robot; 50: user; 90: network; 102: communicating unit; 120: sensor unit; 130: information processing unit; 160: control target; 162: display unit; 120: sensor unit; 200: secretion information generating unit; 210: input information generating unit; 220: adjusting unit; 230: NN operation unit; 240: judging unit; 250: control unit; 252: determining unit; 260: emotion identifying unit; 270: processing unit; 280: storage unit; 290: recording medium; 300: emotion map; 900: object; 1000: object
| Number | Date | Country | Kind |
|---|---|---|---|
| 2016-091998 | Apr 2016 | JP | national |
The contents of the following patent applications are incorporated herein by reference: Japanese Patent Application No. 2016-091998 filed on Apr. 28, 2016 and International Patent Application No. PCT/JP2017/014876 filed on Apr. 11, 2017.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/JP2017/014876 | Apr 2017 | US |
| Child | 16170045 | US |
