JP7141615B2 - Robot, robot control method and program - Google Patents

Description

TECHNICAL FIELD The present invention relates to a robot or the like that performs solution search for a group of state variables, and more particularly to a robot or the like that performs solution search for a group of state variables for controlling the movement of its legs, for example.

In a search for a solution such as a satisfiability problem (SAT) by a general computer, as the number of state variables increases, the amount of calculation increases explosively, and it takes a long time to obtain a correct answer. Therefore, conventionally, an amoeba-type computer has been proposed (see, for example, Patent Document 1). In amoeba-type computers, even if the number of state variables increases, the amount of computation increases, for example, only linearly, and efficient solution search can be expected.

JP 2014-85733 A

The inventors of the present application have devised a robot (hereinafter referred to as an amoeba-type robot) that uses the principle of an amoeba-type computer as a device for controlling the movement of its legs.

However, in conventional amoeboid computers, regarding the state variable group for controlling the movement of the robot's legs, the solutions are not updated in response to feedback from the controller, resulting in a state in which steady state solutions are obtained repeatedly and solution search is inadequate. As a result of falling into a state where it is possible, efficient solution search becomes difficult, and the motion of the robot may become unstable.

Also, robots other than the amoeba type may not be able to perform efficient solution search for the state information group, resulting in unstable motion.

A robot according to a first aspect of the present invention includes a storage unit that stores one or more negative rules that are rules associated with a partial state information group that is a set of one or more state information and that indicate prohibited items. , two or more legs and a state information group that is a set of state information indicating at least the state of each of the two or more legs; a control unit that acquires a state information group that does not match the rules and controls each of the two or more legs based on the state information group, and at least some of the two or more legs are: A robot that does not, at least temporarily, perform an action based on a group of partial state information.

With such a configuration, it is difficult for the robot to fall into a state of repeatedly obtaining a steady-state solution or a state in which the solution search becomes impossible.

In the above configuration, "at least some of the two or more legs do not perform an operation based on the partial state information group at least temporarily" means, for example, two or more legs performed by the control unit. This is the result of different delays in the control of each leg of the . The delay occurring in each of the two or more legs may be, for example, on the order of several milliseconds to several hundred milliseconds. A delay fluctuating within such a range is preferably realized by an analog core circuit, for example, as described in the third invention to be described later, but may be realized by the control unit in terms of software. . In that case, if the unit time in the operation of the control unit is, for example, several milliseconds, the control unit sends a control signal to each leg, for example, about 1,000 to 100,000 times the unit time. By delaying, a fluctuating delay in the range of several milliseconds to several hundred milliseconds in each leg can be realized.

That is, the operation timescale of a robot having a device including a control unit as described in the embodiment differs from the operation timescale inside the device, so that the operation of the robot is prevented from becoming unstable. In order to do so, the controller should generate a delay whose length corresponds to the timescale of the motion of the robot and which is different between the legs.

Alternatively, in the above configuration, "at least some of the two or more legs do not, at least temporarily, perform an action based on the partial state information group" satisfies a predetermined condition. In this case, the control unit may be the result of controlling at least some of the two or more legs so as not to perform an operation based on the partial state information group, at least temporarily. The predetermined condition is usually a condition regarding information other than the partial state information group. Information other than the partial state information group is, for example, information outside the robot (environmental information) such as an image from a camera, information from a temperature sensor or a humidity sensor.

The condition regarding the image may be, for example, that the high-frequency component is greater than or equal to the threshold. Increased chances of avoiding falls. The condition regarding the temperature information may be, for example, that the temperature is lower than or equal to or less than a threshold, so that when the ground is frozen, no action based on the partial state information group is performed, and the robot This increases the possibility of avoiding collisions and falls due to slipping. The condition regarding the humidity information may be, for example, that the humidity is higher than or equal to or greater than the threshold, whereby when the ground is wet, the operation based on the partial state information group is not performed, and the robot is The possibility of avoiding a collision or falling due to a slip increases.

In addition, in the robot of the second invention, in contrast to the first invention, each of the two or more legs has a sensor for detecting whether or not the leg is on the ground, and the state information group is a robot that includes state information indicating detection results by sensors in each of two or more legs.

With such a configuration, an amoeba robot capable of efficient solution search for a group of state variables for controlling the movement of its legs is realized.

The robot of the third invention, in contrast to the first or second invention, further comprises a core circuit having two or more partial circuits each having a diode, a capacitor, and a switch, and the two or more partial circuits comprise two The above circuit for controlling the movement of the leg, the control unit acquires a state information group, which is a set of state information indicating at least the state of each of the two or more partial circuits, and obtains one or more of each of the state information groups. A state information group that does not match the negative rule corresponding to the partial state information group is acquired, and based on the state information group, control is performed to either flow or stop the current in each of the two or more partial circuits. , at least some of the one or more capacitors included in each of the two or more partial circuits have different time constants for charging or discharging, and the different time constants result in at least some of the two or more partial circuits included in the core circuit. is a robot that does not, at least temporarily, operate based on the partial state information group.

With such a configuration, it is difficult for the robot to fall into a state of repeatedly obtaining a steady-state solution or a state in which the solution search becomes impossible.

In addition, in the robot of the fourth invention, in contrast to the third invention, two or more partial circuits are connected via a hub, and a current is supplied to each of the two or more partial circuits from the center of the hub. is.

With such a configuration, the total current in the hub is conserved according to Kirchhoff's law.

Further, in the robot of the fifth invention, in contrast to the third or fourth invention, the control unit acquires a state information group, which is a set of state information indicating at least the state of each of the two or more partial circuits. information group acquisition means; negative rule acquisition means for acquiring, from a storage unit, one or more negative rules corresponding to one or more partial status information groups of the status information group; and one or more negative rules acquired by the negative rule acquisition means. A positive example acquiring means for acquiring a positive case state information group that does not match any of the rules; A robot comprising control means for performing control for movement.

With such a configuration, for example, efficient solution search can be performed with respect to the state variable group for controlling the movement of the legs of the amoeba robot.

Further, in the robot of the sixth invention, in contrast to any one of the third to fifth inventions, the charging time or discharging time of the capacity of the capacitor is controlled according to the result of the state information group acquired by the control unit. It is a robot that takes longer than the time to perform

With such a configuration, the state change may not catch up with the control, and as a result, the operation may not be performed, increasing the possibility of finding the correct answer.

In addition, in the robot of the seventh invention, in contrast to the fifth invention, the positive example acquisition means is a robot that acquires a state information group of positive cases by trial and error.

With such a configuration, it is possible to efficiently acquire a group of status information of positive cases.

In addition, in the robot of the eighth invention, in contrast to any one of the first to seventh inventions, the storage unit has a number of legs indicating a state of landing that is equal to or less than a predetermined number. In addition, it is a robot that stores the first negative rule that prohibits raising the leg.

Such a configuration realizes an amoeba-type robot that can accurately control the movement of its legs.

Further, in the robot of the ninth invention, in contrast to any one of the first to eighth inventions, each of the two or more legs is movable back and forth, and the storage part is provided between the two or more legs. This robot stores a second negative rule that prohibits forward and backward movement of one or more legs out of two or more legs, corresponding to the forward and backward positional relationship of .

Such a configuration realizes an amoeba-type robot that can accurately control the movement of its legs.

In addition, in the robot of the tenth invention, in contrast to any one of the first to tenth inventions, each of the two or more legs can be raised and lowered, and the storage section relates to each of the two or more legs, A third negative rule that prohibits forward movement with the leg down, corresponding to the state of the leg being on the ground, and a leg It is a robot that stores the fourth negative rule that prohibits it from moving backwards with its tail raised.

Such a configuration realizes an amoeba-type robot that can accurately control the movement of its legs.

In addition, in the robot of the eleventh invention, in contrast to any one of the first to tenth inventions, the storage unit has two or more legs, and when the legs are positioned forward or backward, , a robot that stores a fifth negative rule that prohibits the movement of other legs that have a predetermined positional relationship with the leg.

Such a configuration realizes an amoeba-type robot that can accurately control the movement of its legs.

In addition, in the robot of the twelfth invention, in contrast to any one of the first to eleventh inventions, the storage unit has two or more legs, and the legs are raised and lowered and grounded. It is a robot that stores a sixth negative rule that permits prohibited movement according to a combination of state and non-state.

Such a configuration realizes an amoeba-type robot that can accurately control the movement of its legs.

In addition, the robot of the thirteenth invention is a robot that can permit two or more types of motions for the same group of state information according to the sixth negative rule in contrast to the twelfth invention.

With such a configuration, the possibility of escaping from a state in which steady-state solutions are repeatedly obtained or a state in which solution search becomes impossible increases, and as a result, an amoeba-type robot that can accurately control the movement of its legs is realized.

According to the present invention, for example, as a result of efficient solution search for the state variable group, a robot whose motion is less likely to become unstable is realized.

Block diagram of the robot in the embodiment Block diagram of the same device Data structure diagram of dominance rules Flowchart explaining the operation of the device Overview of the robot A diagram showing the circuit configuration of the same electronic amoeba together with a single-celled amoeba. A diagram showing the circuit configuration of the pseudopodia unit and error signal waveforms with different characteristics given to the pseudopodia unit. A diagram showing the time evolution of the four variables (a) and the search time fluctuation (b) according to the change in the error period λ for each error probability Diagram showing an example of the internal configuration of the same computer system

Hereinafter, embodiments of the robot will be described with reference to the drawings. It should be noted that, since components denoted by the same reference numerals in the embodiments perform similar operations, repetitive description may be omitted.

FIG. 1 is a block diagram of the robot according to this embodiment. The robot comprises a device 1 , a body 2 , two or more (eg, four, eight, etc.) legs 3 and a drive mechanism 4 .

Device 1 will be described later. The body 2 may be made of, for example, a flexible material, have a hollow portion (not shown), and pressurize the inside of the hollow portion. However, the material and structure of the body 2 do not matter.

The legs 3 are movable up and down and back and forth. The leg portion 3 may also be made of, for example, a flexible material such as rubber or elastomer, have a hollow portion (not shown), and pressurize the interior of the hollow portion. In this case, the movement of the leg 3 up and down and back and forth is realized by its expansion and contraction and bending. Alternatively, the legs 3 may be made of a hard material such as metal or carbon fiber and have joints. Up-down and back-and-forth movements in this case are achieved by bending the joints.

The leg 3 also has a sensor 31 . The sensor 31 detects whether or not the leg 3 is on the ground, and outputs the detection result. The detection result is, for example, information indicating whether or not the leg 3 is on the ground (for example, "on" indicating the state of being on the ground, "off" indicating the state of not being on the ground, etc.). . Alternatively, the detection result may be a signal that is not output when the ground is not detected and is output when the ground is detected, or it is continuously output when the ground is not detected and the ground is detected. A signal to be stopped according to the detection may be used. However, the material and configuration of the leg portion 3 are not limited.

Further, the sensor 31 is, for example, a touch sensor, but may be a distance sensor, a push button, or the like, and any type of element can be used as long as it can detect whether or not the sensor is on the ground. Also, the form of the detection result output by the sensor 31 is not limited.

The drive mechanism 4 is a mechanism (for example, an actuator) for realizing vertical and longitudinal movement of each of the two or more legs 3 . The drive mechanism 4 includes, for example, a blower for injecting and discharging air into and from the hollow portion of each of the two or more legs 3, one or more wires attached to the hollow portion, and one or more wires. It has an electric reel for winding. Alternatively, the drive mechanism 4 may have one or more motors for bending the joints for each of the two or more legs 3 . However, the configuration of the driving mechanism 4 is not limited.

The device 1 drives two or more legs 3 via a drive mechanism 4 . Specifically, the device 1 drives the two or more legs 3 by controlling the current of two or more electric elements such as electric reels and motors that constitute the drive mechanism 4, thereby controlling the movement of the robot. Realize movement.

Note that the robot may be, for example, a commercially available multi-legged robot, and its hardware configuration does not matter. Also, the device 1 may be provided in a device other than a robot. Devices other than robots are, for example, chemical reaction simulators. Chemical reaction simulators, for example, simulate chemical reaction processes of covalently bonded organic molecular compounds and predict reaction products. Here, a binding state that violates one or more physical principles or general chemical reaction rules is defined as a negative rule.

FIG. 2 is a block diagram of the device 1. As shown in FIG. The device 1 comprises a storage section 11 , a core circuit 12 and a control section 13 .

The core circuit 12 comprises two or more partial circuits 121 . The two or more partial circuits 121 are connected via, for example, a hub, and current is supplied to each of the two or more partial circuits 121 from the center of this hub. With such a configuration, the total current in the hub is conserved according to Kirchhoff's law, and as a result, for example, the movement of an amoeba-type robot corresponding to an amoeba whose volume is conserved can be realized.

The control unit 13 includes state information group acquisition means 131 , negative rule acquisition means 132 , positive example acquisition means 133 , and control means 134 .

The storage unit 11 configuring the device 1 can store various types of information. Various types of information are, for example, one or more negative rules. Negative rules are rules regarding prohibited items. Prohibitions are things that cannot be done. A negative rule is, for example, a rule associated with a partial state information group, which is a set of one or more pieces of state information, and is a rule indicating a prohibited matter. A partial state information group is a part of the state information group. A status information group is a set of one or more status information. State information is information indicating a state. Details of the state information, the state information group, and the partial state information group will be described later.

Prohibited items may be, for example, prohibited items related to the movement of two or more legs 3 in order to avoid overturning of the robot. Specifically, the negative rule is, for example, a rule such as "it is prohibited to raise three or more legs at once".

Note that the storage unit 11 can also store, for example, regular rules. A positive rule is a rule that must be followed. However, it is preferable that the storage unit 11 does not store the correct rule. Alternatively, even if there is no positive rule, it can be said that the robot can perform accurate movements based on the negative rule.

The storage unit 11 stores, for example, a first negative rule. The first negative rule is that when the number of legs 3 indicating a state of landing among the two or more legs 3 possessed by the robot is equal to or less than a predetermined number, the leg is lifted. This rule prohibits it. The predetermined number is for example 2, but could also be 1 if the robot has only two legs 3 .

The first negative rule is, for example, "if (the number of legs with the sensor on is two or less), then (the leg is prohibited from being lifted)", but the form of expression does not matter.

The storage unit 11 also stores, for example, a second negative rule. The second negative rule is a rule that prohibits forward and backward movement of one or more legs 3 out of the two or more legs 3, corresponding to the forward and backward positional relationship between the two or more legs 3. .

The second negative rule is, for example, "if (either right or left front leg is behind), then (one leg is prohibited from moving backward)", "if (either right or left rear leg is One leg is in front), then (one leg is prohibited from moving forward)”, etc., but it may be a revocation of permission, and the form of expression does not matter.

The storage unit 11 also stores two negative rules, the third and the fourth. The third negative rule is a rule that prohibits moving forward with the leg 3 in a lowered state, corresponding to the state in which the leg 3 is on the ground. . The fourth negative rule is a rule that similarly relates to two or more legs 3 and prohibits the leg 3 from moving backward in a raised state in response to the state in which the leg 3 does not land. be.

The third negative rule is, for example, "if (sensor on), then (moving forward + prohibiting lowered state, and allowing backward movement + lowered state)", and the fourth negative rule is, for example, "if (sensor off), then (prohibit backward movement + raised state, and allow forward movement + raised state), but the expression format is not limited.

The storage unit 11 also stores, for example, a fifth negative rule. The fifth negative rule is that for each of two or more legs 3, if the position of the leg 3 is forward or backward, the movement of the other leg 3 that has a predetermined positional relationship with the leg 3 is a rule that prohibits

The fifth negative rule is, for example, "if (the leg is in front), then (the leg in the point-symmetrical position is prohibited from moving other than forward)", "if (the leg is in the back), then (It is prohibited for the leg in a point-symmetrical position to move in any direction other than backward)”, but the form of expression does not matter.

The storage unit 11 also stores, for example, a sixth negative rule. The sixth negative rule is a rule that permits prohibited movements for each of two or more legs 3, depending on the combination of the state of raising and lowering the leg 3 and whether or not it is in contact with the ground. be. Combinations are, for example, regarding whether or not the leg is raised and lowered, and whether or not the leg is on the ground. There are four states: a state in which the legs are lowered and grounded (sensor on), and a state in which the legs are raised and grounded (sensors on).

The sixth negative rule is, for example, the rule "if (with leg down and sensor off), then (change direction with leg down") that permits the prohibited movement "change direction with leg down". permission to change)”.

Alternatively, the sixth negative rule is a rule that permits the prohibited movement "lowering the leg in that direction", "if (with the leg raised and the sensor off), then (lowering the leg in that direction"). allowed)”.

Or, the sixth negative rule is the rule "if (leg up and sensor off), then (change direction and drop leg") that permits the prohibited movement "change direction and drop leg". allowed)”.

Alternatively, the sixth negative rule is the rule "if (with the leg lowered and the sensor on)" that permits the prohibited movement "raising the leg in that direction", then (raising the leg in that direction allowed).

Or, the sixth negative rule is the rule "if (leg down and sensor on), then (change direction and lift leg") that allows the prohibited movement "turn leg up". allowed)”.

Alternatively, the sixth negative rule is a rule that permits the prohibited movement “lowering the leg in that direction”, “if (with the leg raised and the sensor on), then (lowering the leg in that direction”). is prohibited)” is acceptable.

However, the expression form of the sixth negative rule does not matter. Also, the sixth negative rule may include, for example, rule groups corresponding to two or more of the above six examples.

It should be noted that the sixth negative rule, for example, the second and third, fourth and fifth of the above six examples, for the same state information group, two or more movement. As a result, for example, in a specific example to be described later, the possibility of escaping from a state in which steady-state solutions are repeatedly obtained or a state in which solution search is impossible increases, and as a result, the movement of the leg 3 of the amoeba-type robot can be accurately controlled. You can control it. The state in which the solution search becomes impossible is a state in which all the variables try to take the same value at the same time, causing a conflict. As a result of the system being locked in this state, the robot 1 stops moving or falls over in an unstable place.

Each of the two or more partial circuits 121 forming the core circuit 12 has a diode, a capacitor and a switch. A diode allows current to flow only in one direction (forward) and not in the opposite direction. The capacitor stores electricity flowing through the diode and releases the stored electricity through the switch.

It is preferable that the two or more partial circuits 121 have different capacitor charging or discharging time constants (hereinafter sometimes referred to as charging/discharging time constants or simply time constants). The charge/discharge time constant is basically determined by the capacitance of the capacitor, but also varies depending on other circuit elements (eg, diodes, resistors, switches, etc.) connected to the capacitor. Therefore, even if the capacitance of the capacitors is the same between the two or more partial circuits 121, for example, the values of the resistors and the characteristics of the diodes are different, and the thresholds of the NOT gates that realize the switches are different. If so, the charging/discharging time constant will be different.

The switch switches between a state in which the capacitor can be charged and a state in which the capacitor can be discharged. That is, the capacitor can be charged when the switch is OFF and discharged when the switch is ON.

Subcircuit 121 typically also has a resistor. A resistor, for example, increases the time constant for charging and discharging a capacitor. A time constant is the time required to charge and discharge a capacitor. The timing at which the charging of the capacitor is completed is delayed by a time corresponding to the time constant with respect to the timing at which the switch is turned on. Similarly, with respect to the timing when the switch is turned off, the timing when discharging from the charged capacitor is also delayed by a time corresponding to the time constant.

Note that the switch is composed of, for example, a NOT gate, but an inverter may be used, and any means for realizing it may be used. It goes without saying that the partial circuit 121 may also have other elements. The other element is, for example, a field effect transistor (FET), which will be described in a specific example, but the type of element does not matter.

The control unit 13 controls movements of the two or more legs 3 via the two or more partial circuits 121 forming the core circuit 12 . Specifically, the control unit 13 acquires a state information group that is a set of state information indicating at least the state of each of the two or more partial circuits 121 that constitute the core circuit 12, A state information group that does not match the negative rule corresponding to the state information group is obtained, and based on the state information group, current is supplied to each of the two or more partial circuits 121, or current is stopped. conduct.

State information is information indicating a state. The state is, for example, the state of the partial circuit 121 . The information indicating the state of the partial circuit 121 is, for example, the amount of electricity stored in the capacitor, or information indicating whether the capacitor is being charged or discharged.

Alternatively, the state may be the state of the sensor 31 . The state information indicating the state of the sensor 31 is, for example, "on" or "off" as described above, but the form of expression is not limited.

Alternatively, the state may be the state of the leg 3 . The state information indicating the state of the leg portion 3 includes, for example, "leg raised state", "leg lowered state", "backward moving state", forward moving state, "grounded state", It is "ungrounded state" and the like.

The “grounded state” and “non-grounded state” are equivalent to “on” and “off” of the sensor 31 . Other information such as “leg raised state” may be acquired using state information indicating the state of the partial circuit 121, for example. However, the state information may be any information that indicates the state of the components of the robot, such as information that indicates the state of the drive mechanism 4 (for example, the current value of the electric reel, the angle of the joint, etc.). Also, information indicating the state of a component of a device other than a robot may be used.

A state information group is a set of various types of state information as described above. The state information group includes at least state information of each of the two or more partial circuits 121 forming the core circuit 12 . In the case of the robot shown in FIG. 1, the state information group also normally includes state information of the sensor 31.

The state information group may be, for example, a state variable group {X1, X2, X3, X4, . . . } described in a specific example. Each of the state variables "X1", "X2", etc. forming the state variable group takes a value of either true (eg "1") or false (eg "0").

In this case, the storage unit 11 may store, for example, two or more negative rules as shown in FIG. FIG. 3 is a data structure diagram of a negative rule. A negative rule has a state information group and prohibited matters. The state information group is composed of one or more state variables X1 to X4.

The state variable X1 is, for example, a variable indicating whether or not the leg 3 is raised, and is "1" corresponding to the true "leg raised state" or false "leg lowered state". It takes any value of "0". Note that the correspondence between true/false and "0" and "1" may be reversed (same below).

The state variable X2 is, for example, a variable indicating whether or not the leg 3 is moved forward, and is "1" corresponding to true "state of forward movement" or false "state of forward movement". It takes any value of "0" corresponding to the "non-existent state". Similarly, the state variable X3 is, for example, a variable indicating whether or not the leg 3 has moved backward, and is "1" corresponding to true "backward movement state" or false "backward movement state". It takes any value of "0" corresponding to the state "not

Further, the state variable X4 is, for example, a variable indicating whether or not the leg 3 is in contact with the ground. Takes one of the corresponding "0" values.

Two or more stored negative rules are associated with IDs (for example, "1", "2", etc.). For example, the negative rule corresponding to ID "1" (hereinafter sometimes referred to as negative rule 1) is positioned symmetrically with the state information group "0, 1, 0, 1..." It is forbidden to move the legs other than forward.” In addition, the negative rule (negative rule 2) corresponding to the ID "2" is the state information group "0, 0, 1, 1..." "Prohibit" and have.

Similarly, the negative rule 3 includes the state information group "Null, 0, 0, 1..." and the prohibition "prohibit forward movement + lowered state, and allow backward movement + lowered state". Negative rule 4 has a state information group "Null, 0, 0, 0 ..." and a prohibition "prohibit backward movement + raised state, and forward movement + raised state" Permission”. Note that "Null" indicates that the value of the state variable may be either "0" or "1".

However, the data structure of the negative rule does not matter. A negative rule may be a bounceback rule in a specific example described later.

The state information group acquisition unit 131 acquires, for example, for each of two or more partial circuits 121, a state information group including one or more pieces of state information of the partial circuits 121 concerned.

Alternatively, the state information group acquisition means 131 obtains, for each of two or more legs 3, one or more state information of the partial circuit 121 corresponding to the leg 3 and state information of the sensor 31 corresponding to the leg 3. You may acquire the state information group containing.

Negative rule obtaining means 132 obtains, for each of two or more status information groups obtained by status information group obtaining means 131, one or two or more negative rules corresponding to one or two or more partial status information groups possessed by the status information group. A rule is acquired from the storage unit 11 .

It can be said that the partial state information group is a subset of the state information group, which is a set of various types of state information. For example, in the case of the state variable group {X1, X2, X3, X4}, the one or more partial state information groups are {X1}, {X2}, {X1, X2}, {X2, X3}. . . , {X1, X2, X3}, {X1, X2, X4} . . . , {X1, X2, X3, X4}.

For example, the negative rule acquisition unit 132 searches the storage unit 11 using the state information group as a key for each of one or more partial state variable groups as described above, and acquires a negative rule corresponding to the state variable group. Therefore, one or more partial state information groups can be obtained for one state information group, and one or more negative rules can be obtained for each of them.

The positive case obtaining means 133 obtains a state information group of positive cases. A positive example state information group is a state information group that does not match any of the one or more negative rules acquired by the negative rule acquisition means 132 .

The positive case acquisition unit 133 acquires a state information group of positive cases, for example, by trial and error. Trial and error refers to the process of repeatedly trying and failing to finally solve an unknown problem when there is no known solution. Trial and error may be, for example, a solution search for a group of state variables as explained in a specific example.

That is, the positive example acquiring means 133, for example, obtains a state variable group that does not match any of the one or more negative rules (this is {X1=1, X2=0, X3=0, X4=1} ), and using the inverted state variable group (that is, {X1 = 0, X2 = 1, X3 = 1, X4 = 0}), the state after a unit time Δt The process of calculating the variable group is repeated until the previous calculation result and the current calculation result match.

Then, the positive example acquisition unit 133 acquires a state variable group in which the previous calculation result and the current calculation result match as a positive example state information group. Note that the positive case acquisition unit 133 may store the state information group of positive cases thus acquired in the storage unit 11 .

The control means 134 controls the operation of either supplying or stopping the current to each of the two or more partial circuits 121 based on the positive case state information group acquired by the positive case acquiring means 133 . Also good. Specifically, for example, of the conditions (if...) and permission items (then...) that constitute the sixth negative rule, the former (for example, "the state in which the leg is raised and the sensor is off") etc.) corresponds to the positive example of the state information group, and the latter (for example, "permit to change direction while putting the foot down") corresponds to the control based on the positive example of the state information group.

In the device 1 configured as described above, it is preferable that at least some of the one or more capacitors included in each of the two or more partial circuits 121 constituting the core circuit 12 have different capacitances. For example, when each of two or more partial circuits 121 has only one capacitor, the difference in capacitance means that the capacitance varies between the two or more capacitors. Alternatively, if two or more partial circuits 121 each have two or more capacitors, the difference in capacitance is that, for example, the two or more capacitors included in one partial circuit 121 have different capacitances. It may be that the two or more partial circuits 121 have different degrees of variation in capacitance. Due to the difference in capacitance, at least some partial circuits 121 out of the two or more partial circuits 121 of the core circuit 12 do not operate based on the partial state information group, at least temporarily.

It should be noted that not operating includes, for example, a case where the partial circuit 121 changes slowly in response to the control of the control unit 134 and, as a result, the partial circuit 121 does not operate.

Alternatively, the charge/discharge time constants of the capacitors may be different between the two or more partial circuits 121 forming the core circuit 12 . Note that the charge/discharge time constant varies, for example, depending on the capacitance of the capacitor or a combination of the capacitance of the capacitor and the value of the resistance.

Due to the difference in the charging and discharging time constants of the capacitors, in at least some of the partial circuits 121, the charging time or discharging time of the capacitance of the capacitors is controlled according to the result of the acquisition of the positive state information group by the control unit 13. As a result, there are cases where the part of the partial circuit 121 does not operate.

In a specific example described later, even if the charge/discharge time constant of the capacitor is longer than the feedback time from the controller to the core circuit 12, at least part of the partial circuit 121 may not operate. .

In this way, at least a portion of the partial circuits 121 are not operated, so that a state of repeatedly obtaining a steady state solution and a solution search for the state variable group for controlling the motion of each of the two or more legs 3 of the robot are not performed. As a result of being less likely to fall into an impossible state, efficient solution search can be performed.

The storage unit 11 is preferably a non-volatile recording medium such as a flash memory, but can also be realized by a volatile recording medium such as a RAM.

It does not matter how the information is stored in the storage unit 11 or the like. For example, information may be stored in the storage unit 11 or the like via a recording medium such as a memory card, or information transmitted via a network, a communication line, or the like may be stored in the storage unit 11 or the like. Alternatively, information input via an input device may be stored in the storage unit 11 or the like. Any input device such as a touch panel, a keyboard, or the like may be used.

The control unit 13, the state information group acquisition means 131, the negative rule acquisition means 132, the positive example acquisition means 133, and the control means 134 can usually be implemented by an MPU, memory, or the like. The processing procedure of the control unit 13 and the like is normally realized by software, and the software is recorded in a recording medium such as a ROM. However, the processing procedure may be realized by hardware (dedicated circuit).

Next, the operation of the device 1 will be explained using the flowchart of FIG. The processing of this flow chart is executed in software by a computer such as a microcomputer, but may be executed in hardware by a digital circuit using logic gates or an analog circuit that performs logical operations.

(Step S401) The control unit 13 sets the initial value 1 to the variable i. The variable i is a variable for sequentially selecting unselected partial circuits 121 out of the two or more partial circuits 121 forming the core circuit 12 . In the case of an analog circuit, since it operates in continuous time, the variable i is convenient for understanding.

(Step S402) The control unit 13 determines whether or not the i-th partial circuit 121 exists. If the i-th partial circuit 121 exists, the process proceeds to step S403, and if the i-th partial circuit 121 does not exist, the process returns to step S401.

(Step S403) The state information group acquisition unit 131 acquires a state information group corresponding to the i-th partial circuit 121. FIG.

(Step S404) The negative rule acquisition unit 132 acquires from the storage unit 11 one or more negative rules corresponding to the state information group acquired in step S403.

(Step S405) The positive example obtaining unit 133 obtains a group of positive state information that does not match any of the one or more negative rules obtained in step S404.

(Step S406) The control means 134 uses the state information group of the positive example acquired in step S405 to control the operation of either supplying or stopping the current to the i-th partial circuit 121. .

(Step S407) The control unit 13 increments the variable i. After that, the process returns to step S402.

In the flowchart of FIG. 4, the process ends when the power of the device 1 is turned off or a process end interrupt occurs.

A specific operation example of the robot according to the present embodiment will be described below. The robot of this example is an amoeba type robot as shown in FIG. Although this robot has four legs, it may have any number of legs, such as eight or six, as long as it has two or more legs. Each leg (pseudopod unit) of this robot corresponds to each leg 3 of the robot shown in FIG. Each leg has joints and is movable in up and down and forward and backward directions. A touch sensor corresponding to the sensor 31 is provided at the tip of each leg.

This robot then has an Electronic amoeba that corresponds to device 1 . Electronic amoebas are inspired by the prey behavior of single-celled amoeboid organisms (hereafter single-celled amoeba) and are designed to search for solutions to combinatorial optimization problems. Single-celled amoebas seek to maximize food intake while meeting given constraints.

Specifically, first, single-celled amoeba extend pseudopodia in one direction with some degree of fluctuation. Electronic amoeba exhibit such pseudopodic movements by means of unidirectional currents flowing through diodes within the pseudopodia unit, corresponding to state variables. Deformation fluctuations and errors in single-celled amoeba are realized electronically in electronic amoeba using a random bit generator. Alternatively, similar fluctuations and errors can be obtained by introducing noise into the power supply or the like.

Second, unicellular amoebae are constrained in their deformation by volume conservation. Constraints on deformation by volume conservation give rise to spatiotemporal correlations between pseudopodia. In the electronic amoeba, volume conservation is expressed by Kirchhoff's law at the center of the star-shaped network of pseudopodia units.

Third, unicellular amoebae contract their pseudopodia when exposed to light. This light avoidance is used to intentionally deform the amoeba shape. The property of light avoidance is represented by current control by field effect transistors with capacitors in each pseudopod.

FIG. 6 shows the circuitry of an electronic amoeba as described above, together with a unicellular amoeba. The electronic amoeba has an amoeba core corresponding to the core circuit 12 , a controller corresponding to the control section 13 , and a memory (not shown) corresponding to the storage section 11 .

Stored in memory is a bounceback rule that corresponds to the negative rule. The bounceback rule can be said to be a rule corresponding to the "given constraints" of the single-celled amoeba. Bounceback rules may also be built into the program. Furthermore, a logic function written in a program may be directly implemented by a circuit of logic gates.

The amoeba core has two or more (eight in FIG. 6) pseudopod units corresponding to partial circuits 121 . These two or more pseudopodia units are connected via a star-topology network, which represents a hub. Current is supplied to the center (hub) of the star network, and the supplied current is distributed to each of the two or more pseudopodia units. A current Ii is input to each pseudopodia unit, and the output Xi of each pseudopodia unit is provided to the controller. The controller generates a feedback signal yi based on the bounceback rule for the output Xi from each pseudopodia unit and provides it to each pseudopodia unit. A feedback signal is a signal for updating a state variable and is also called a bounceback signal.

Each pseudopod is also provided with an error signal ei generated by a random bit generator (not shown). It can be said that the error signal is a signal that realizes the deformation error or fluctuation of the single-celled amoeba.

FIG. 7 shows the circuit configuration of each pseudopodia unit and the waveforms of three error signals with different characteristics given to each pseudopodia unit. First, regarding the circuit configuration, the input terminal of the pseudopod unit is connected to the anode side of the diode via the resistor R, and the cathode side of the diode is connected to the output terminal via the NOT gate. The cathode side of the diode is also grounded through a capacitor C, and a first FFT is connected between the cathode side of the diode and capacitor C, and in series with the first FET between capacitor C and ground. A second FET is connected respectively. A feedback signal yi is applied to the first FET, and an error signal ei is applied to the second FET.

For example, the capacitance of the capacitor C of each pseudopod unit is 10 nF and the resistance R is 100Ω. However, it is preferable that one or more of the capacitance and resistance values vary among the pseudopod units. For example, it is desirable that the relative variation in capacitance, etc. between pseudopodia units be less than 100%. The relative variability can be said to be a value that relatively indicates the degree of variability in two or more groups with different units, averages, or the like, for example. The relative variation is, for example, the relative standard deviation (also referred to as coefficient of variation) obtained by dividing the standard deviation by the arithmetic mean, but any value can be used as long as it can relatively express the degree of variation. This is because if the variation is too large, the device 1 may not operate.

The current injected into the hub is 10 μA. The number of variables N is 12 and the number of redundant variables is 2N=24. The clock frequency of the controller is 84 MHz, which is sufficiently faster than the charging/discharging speed of the capacitor C.

A random bit string generated by the Xorshift method is used as the error signal that causes the fluctuation. The error signal produces a "0" regardless of the bounceback signal.

Error signals with different characteristics such as e1 to e3 shown in the lower part of FIG. 7 affect the solution search. The error probability p means how many pseudopodia units received an error signal at one time. The error period λ is the sum of the error bit width and the average time interval between adjacent bits. Increasing λ increases both the bit width and the time interval of the error. The state variable Xi of the electronic amoeba is obtained from the output of the pseudopod unit (ie the output voltage of the NOT gate).

FIG. 8 shows the time evolution of the four variables (a) and the search time variation (b) according to the change in the error period λ for each error probability.

In FIG. 8(a), the error probability p is 0.65 and the number of state variables is 24. Starting the search process from the initial state Xi_j=1 for all i and j, each of the state variables changes and sometimes shows an oscillatory behavior. When an error signal is randomly imposed on the pseudopodia unit, the electronic amoeba goes into another state. After applying the error signal several times, the system reaches a steady state and stabilizes. This confirms that the steady-state state variables always satisfy the constraints, that is, the amoeba core can find a solution.

On the other hand, the oscillatory state suggests that the amoeba core cannot resolve discrepancies between variables. The system becomes localized and it becomes necessary to provide an error signal to the system in order to escape from the localized state. In other words, the amoeba core maintains its dynamics until a final solution is found. This is because the bounceback rule continues to flip at least one of the state variables until all state variables satisfy the constraint.

Also, from FIG. 8(b), it can be seen that the error probability is related to the solution retrieval efficiency. Each curve has a minimum value, which indicates the existence of an optimal error period that minimizes the solution search time. This data shows that the optimal error period depends on the error probability and is of the order of the charging time constant of the capacitance, thereby minimizing the optimal error probability p It can be seen that there exists a value of

Thus, the electronic amoeba of the present example finds the proper state of walking straight, depending on the sensor 31 signal and the previous state. If the electronic amoeba is constrained to avoid falling or getting lost, it doesn't need to be programmed how to walk.

As described above, according to the present embodiment, the robot stores one or more negative rules that are rules associated with a group of partial state information, which is a set of one or more state information, and that indicate prohibited items. a core circuit 12 having two or more partial circuits 121 each having a diode, a capacitor, and a switch; Then, a state information group that does not match the negative rule corresponding to each of the one or more partial state information groups included in the state information group is acquired, and based on the state information group, current is caused to flow through each of the two or more partial circuits 121. and a control unit 13 that performs control for either operation of stopping. At least some of the one or more capacitors included in each of the two or more partial circuits 121 have different charging or discharging time constants. Partial circuit 121 does not operate based on the partial state information group, at least temporarily. According to such a robot, it is difficult for the robot to fall into a state in which a steady state solution is repeatedly obtained with respect to the group of state variables for controlling the movement of the leg 3 or a state in which the solution search becomes impossible, and efficient solution search can be performed. , the operation becomes unstable and difficult.

In the above robot, the two or more partial circuits 121 are connected via a hub, and current is supplied to each of the two or more partial circuits 121 from the center of the hub. As a result of being conserved according to the law, for example, a motion of a robot corresponding to amoeba whose volume is conserved can be realized.

In the above robot, the control unit 13 includes a state information group acquisition unit 131 for acquiring a state information group, which is a set of state information indicating at least the state of each of the two or more partial circuits 121, and one or more Negative rule acquisition means 132 for acquiring from the storage unit 11 one or more negative rules corresponding to each partial state information group of , and positive examples that do not match any of the one or more negative rules acquired by the negative rule acquisition means 132 A positive case obtaining means 133 for obtaining a state information group, and a control means for performing control for either supplying or stopping the current to each of the two or more partial circuits 121 based on the positive case state information group. 134, efficient solution search can be performed with respect to the state variable group for controlling the motion.

In the above robot, the charge time or discharge time of the capacity of the capacitor is longer than the time until the control unit 13 performs control according to the result of acquiring the state information group. As a result, there are cases where no action is performed, and the possibility of discovering the correct answer increases.

In the above robot, the positive example obtaining means 133 can obtain the state information group of positive cases efficiently by obtaining the state information group of positive cases through trial and error.

In addition, the robot further comprises two or more legs 3, the two or more partial circuits 121 are circuits for controlling the movements of the two or more legs 3, and each of the two or more legs 3 is a leg It has a sensor 31 for detecting whether or not the leg 3 is in a landed state, and the state information group also includes state information indicating the detection result of the sensor 31 of each leg 3 of two or more. Efficient solution search can be performed with respect to the state variable group for controlling the movement of the leg 3 of .

Further, in the above robot, the storage unit 11 stores a first negative rule that prohibits raising the legs when the number of the legs 3 indicating the state of landing is equal to or less than a predetermined number. Therefore, the movement of the leg 3 can be accurately controlled.

Further, in the above robot, each of the two or more legs 3 is movable back and forth, and the storage unit 11 accommodates the two or more legs 3 in correspondence with the front and back positional relationship between the two or more legs 3 . By storing the second negative rule that prohibits the forward and backward movement of one or more legs 3, the movement of the legs 3 can be accurately controlled.

Further, in the above robot, each of the two or more legs 3 can be raised and lowered, and the storage unit 11 is arranged to correspond to the state in which each of the two or more legs 3 is on the ground. A third negative rule that prohibits forward movement with leg 3 down, and a backward movement with leg 3 up corresponding to the state in which leg 3 is not on the ground. By storing the fourth negative rule, the movement of the leg 3 can be accurately controlled.

Further, in the above robot, the storage unit 11 is provided with respect to each of the two or more legs 3, and when the position of the leg 3 is in the front or the rear, another leg having a predetermined positional relationship with the leg 3 is stored. By storing the fifth negative rule that prohibits the movement of leg 3, the movement of leg 3 can be accurately controlled.

In the above robot, the storage unit 11 permits the prohibited movement of each of the two or more legs 3 according to the combination of whether the legs 3 are raised or lowered and whether they are in contact with the ground. By storing the sixth negative rule, the movement of the leg 3 can be accurately controlled.

In the above robot, the sixth negative rule is that two or more types of motions can be permitted for the same group of state information. The possibility of escaping is increased, and as a result, the movement of the leg 3 can be accurately controlled.

In the present embodiment, the two or more partial circuits 121 that constitute the core circuit 12 are analog circuits. As a result, the partial circuit 121 did not operate based on the partial state information group at least temporarily. By, for example, adding different delays to their outputs or stochastically delaying their outputs between the circuits 121, at least some of the partial circuits 121 are at least temporarily dependent on the set of partial state information. It can result in no action taken based on it.

Therefore, according to the present embodiment and the above modification (modification 1), regardless of whether the two or more partial circuits 121 are analog circuits or digital circuits, at least some of the partial circuits 121 are: At least temporarily, it is possible to obtain a result in which no operation is performed based on the partial state information group, and efficient solution search can be performed for the state information group.

Further, even in a modification (modification 2) in which the device 1 does not have the core circuit 12 configured by two or more partial circuits 121 as described above, for example, the control unit 13 may At least some of the two or more legs 3 are at least temporarily , may result in no action based on the partial state information group.

The processing that causes different delays between the legs means, for example, that the timing of changing the control signal to each leg 3 between 0 and 1 is based on the operation clock of the MPU (hereinafter referred to as unit time). However, as long as the processing results in different delays between the legs, the procedure does not matter. Here, the unit time is, for example, several microseconds. However, the unit time may be on the order of less than microseconds, or may be 10 microseconds or more, and its value does not matter.

On the other hand, the delay generated in each of the two or more partial circuits 121 constituting the analog core circuit 12 described in the embodiment fluctuates, for example, within the range of several milliseconds to several hundreds of milliseconds. However, the delay occurring in each partial circuit 121 may be, for example, a time on the order of less than milliseconds or a time of one second or more, and its value and fluctuation range are not limited.

Therefore, in order to realize a delay equivalent to the above-described delay generated in the core circuit 12 by software, the control signal to each leg 3 should be set within a range of, for example, 1,000 to 100,000 times the unit time. , may be delayed randomly or according to a predetermined algorithm.

That is, since the timescale of the motion of the robot is different from the timescale of the motion inside the device 1, in order to avoid the unstable motion of the robot, the control unit 13 should be adjusted to the timescale of the motion of the robot. It is only necessary to produce a delay that is of a length that is appropriate and that fluctuates.

A delay of a predetermined length (eg, 100,000 units of time) occurs only in one or more legs 3 out of two or more legs 3, No delay may occur in the other one or more legs 3 .

As a process therefor, the control unit 13 generates a random number for each of two or more legs 3, for example, and sets the random number (for example, 5390, 1253, 5326, etc.) to a predetermined number (for example, 10). If the remainder (e.g. 0, 3, 6, etc.) is a predetermined value (e.g. 0), then the control signal to the leg 3 is a predetermined time (e.g. 10 10,000 unit time), and if the value is different from a predetermined value (eg, 3, 6, etc.), the control signal to the leg 3 is not delayed. As a result, each of the two or more legs 3 will have its movement delayed by a predetermined time (eg, 100,000 units of time) with a certain probability (eg, 1/10 probability).

Note that the control unit 13 may change the probability by changing the predetermined number when performing the division (for example, if the predetermined number is 100, the movement of each leg 3 delay is 1 in 100).

Alternatively, the occurrence of different delays between the legs is within a predetermined range (for example, 10,000 unit time to 90,000 unit time) in at least some of the two or more legs 3. , which may result in a delay of random length.

As a process therefor, the control unit 13 generates a random number for each of two or more legs 3, for example, and sets the random number (for example, 5390, 1253, 5326, etc.) to a predetermined number (for example, 10). division, and delaying the control signal to the leg 3 by a length (e.g., 0, 30,000 unit time, 60,000 unit time) according to the remainder (e.g., 0, 3, 6, etc.) process. This results in each of the two or more legs 3 delaying its movement by a random length, for example in the range of 0 to 90,000 units of time.

Therefore, according to the modified example 2 as well, as a result of efficient solution search for the state information group, a robot whose motion is less likely to become unstable is realized.

According to Modification 2, even when the device 1 is used in a device other than a robot, the control unit 1 causes a delay of a length corresponding to the time scale of the operation of the device, and a fluctuating delay. The operation of the device can be stabilized. Further, inside the device 1, a delay of a length corresponding to the time scale of the operation of the device (for example, about several times the unit time) and a fluctuating delay is generated, so that the operation of the device 1 itself can also be stabilized.

However, if the device 1 has a core circuit 12, the charge/discharge time constant is a charge/discharge time constant corresponding to the time scale of the operation of the device 1 or a device such as a robot having it, and the fluctuating charge/discharge time constant is By using a group of capacitors having such a structure, the control unit 13 does not need to perform the processing described above, and the load on the MPU or the like that implements the control unit 13 is reduced.

Alternatively, when a predetermined condition is satisfied, the control unit 13 causes at least some of the two or more legs 3 to perform an operation based on the partial state information group, at least temporarily. You can control it so that it does not. The predetermined condition is usually a condition regarding information other than the partial state information group. Information other than the partial state information group is, for example, information outside the robot (environmental information) such as an image from a camera, information from a humidity sensor or a temperature sensor.

Specifically, for example, a camera that captures the range in which the robot moves is provided on the robot itself or on the outside thereof. The condition for the image from the camera is that "high frequency components are greater than or equal to the threshold". As a result, for example, when the ground is rough, the motion based on the partial state information group is not performed, increasing the possibility of avoiding the overturning of the robot.

Also, for example, a temperature sensor is provided on the robot itself or in the range in which the robot moves. The condition regarding the temperature information from the temperature sensor is that "the temperature is lower than or equal to the threshold". As a result, for example, when the ground is frozen, the motion based on the partial state information group is not performed, increasing the possibility of avoiding a collision or overturn due to the robot slipping.

Furthermore, humidity sensors are provided, for example, on the robot itself or in the area in which it moves. The condition for humidity information from the humidity sensor is "the humidity is higher than or equal to the threshold". As a result, for example, when the ground is wet due to rain, the motion based on the partial state information group is not performed, thereby increasing the possibility of avoiding a collision or falling due to a slip of the robot.

Furthermore, the processing in this embodiment may be realized by software. Then, this software may be distributed by software download or the like. Also, this software may be recorded on a recording medium such as a CD-ROM and distributed.

The software that implements the robot in this embodiment is, for example, the following program. In other words, the computer of the robot having two or more legs 3 is a rule associated with a partial state information group, which is a set of one or more state information, and one or more negative rules indicating prohibited items. The storage unit 11 can be accessed, and the program causes the computer to acquire a state information group, which is a set of state information indicating at least the state of each of the two or more legs 3, and obtain the state information group. Functions as a control unit 13 that acquires a state information group that does not match the negative rule corresponding to each of the one or more partial state information groups possessed by the group, and controls the two or more legs 3 based on the state information group. , and different delays occur in the control of the two or more legs 3 performed by the control unit 13, so that at least some of the two or more legs 3 are at least temporarily The program is characterized by not performing an operation based on the partial state information group.

FIG. 9 is a diagram showing an example of the internal configuration of a computer system 900 that implements a robot by executing the program according to the embodiment. In FIG. 9, a computer system 900 includes an MPU 911 that is a computer that executes programs, a ROM 912 that stores programs such as a boot-up program, and a ROM 912 that is connected to the MPU 911 to temporarily store instructions of application programs. A RAM 913 that provides a temporary storage space, a storage 914 that stores application programs, system programs, and data, a bus 915 that interconnects the MPU 911, ROM 912, etc., and provides connections to networks such as external networks and internal networks. , a memory card slot 917 and a core circuit 12 . The storage 914 is, for example, flash memory or SSD. Note that the entire computer system 900 may be called a computer.

A program that causes computer system 900 to perform robot functions may be stored, for example, in memory card 920 , inserted into memory card slot 917 , and transferred to storage 914 . Alternatively, the program may be transmitted over a network to computer system 900 and stored in storage 914 . Programs are loaded into RAM 913 during execution. Note that the program may be loaded directly from the memory card 920 or from the network.

The programs do not necessarily include an operating system (OS), third party programs, etc. that cause the computer system 900 to perform the functions of the robot. A program may contain only those portions of instructions that call the appropriate functions or modules in a controlled manner to produce the desired result. How the computer system 900 operates is well known and will not be described in detail. However, the above is just an example, and the hardware configuration of the computer that implements the robot does not matter.

The number of computers that execute the above programs may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed. (However, performing distributed processing on multiple computers is preferable because, for example, different delays can be easily added between computers, or the occurrence of different delays can be expected stochastically. Even when performing centralized processing (for example, time-division processing), it is possible to add different delays or perform control in which different delays are expected stochastically.

In the above embodiments, each process (each function) may be implemented by centralized processing by a single device (system), or may be implemented by distributed processing by a plurality of devices. may However, distributed processing by a plurality of devices is preferable in that, for example, it is possible to easily add different delays between the devices or to predict the occurrence of different delays stochastically. Also, even when centralized processing is performed by one device, it is possible to add different delays or perform control in which the occurrence of different delays can be expected stochastically.

It goes without saying that the present invention is not limited to the above-described embodiments, and that various modifications are possible and are also included within the scope of the present invention.

INDUSTRIAL APPLICABILITY As described above, the robot according to the present invention is useful as a robot or the like because it can efficiently search for a solution for the state information group, and as a result, it has the effect that the movement is less likely to become unstable.

REFERENCE SIGNS LIST 1 device 2 body 3 leg 4 drive mechanism 11 storage unit 12 core circuit 13 control unit 31 sensor 121 partial circuit 131 status information group acquisition means 132 negative rule acquisition means 133 positive example acquisition means 134 control means

Claims (12)

a storage unit that stores one or more negative rules that are rules associated with a partial state information group that is a set of one or two or more pieces of state information and that indicate prohibited items;
two or more legs;
a core circuit comprising two or more sub-circuits comprising diodes, capacitors and switches, the sub-circuits controlling movement of the two or more legs;
Obtaining a state information group that is a set of state information indicating at least the state of each of the two or more legs, and obtaining a state information group that does not match a negative rule corresponding to each of one or more partial state information groups included in the state information group and a control unit that controls each of the two or more legs based on the state information group,
a robot in which at least some of the two or more legs do not perform movements based on the partial state information group, at least temporarily ;
The control unit
Acquiring a group of status information that also indicates the status of each of the two or more partial circuits, acquiring a group of status information that does not match a negative rule corresponding to each of the one or more groups of partial status information included in the group of status information, and obtaining a group of status information that does not match the negative rule controlling the operation of either passing current through or stopping the current in each of the two or more partial circuits based on the information group;
At least some of the one or more capacitors included in each of the two or more partial circuits have different time constants for charging or discharging,
The robot, wherein at least some partial circuits among the two or more partial circuits included in the core circuit do not operate based on the partial state information group at least temporarily due to the different time constants . each of the two or more legs has a sensor for detecting whether or not the leg is on the ground;
2. The robot according to claim 1, wherein the state information group includes state information indicating detection results by the sensors of the two or more legs. the two or more partial circuits are connected via a hub;
3. The robot according to claim 1 , wherein current is supplied from the center of said hub to each of said two or more partial circuits. The control unit
state information group acquisition means for acquiring a state information group, which is a set of state information indicating at least the state of each of the two or more partial circuits;
negative rule acquisition means for acquiring from the storage unit one or more negative rules corresponding to one or more partial state information groups included in the state information group;
positive example acquisition means for acquiring a positive example state information group that does not match any of the one or more negative rules acquired by the negative rule acquisition means;
4. Control means for controlling an operation to either pass or stop the current to each of the two or more partial circuits based on the state information group of the positive example. or the robot according to item 1 . 5. The robot according to any one of claims 1 to 4 , wherein the charging time or discharging time of the capacity of the capacitor is longer than the time until the control unit performs the control according to the result of acquiring the state information group. . The positive example acquisition means is
5. The robot according to claim 4 , wherein the state information group of positive examples is obtained by trial and error. a storage unit that stores one or more negative rules that are rules associated with a partial state information group that is a set of one or two or more pieces of state information and that indicate prohibited items;
two or more legs;
Obtaining a state information group that is a set of state information indicating at least the state of each of the two or more legs, and obtaining a state information group that does not match a negative rule corresponding to each of one or more partial state information groups included in the state information group and a control unit that controls each of the two or more legs based on the state information group,
a robot in which at least some of the two or more legs do not perform movements based on the partial state information group, at least temporarily;
The one or more negative rules are
includes a first negative rule that prohibits lifting a leg if the number of legs showing a landing condition is less than or equal to a predetermined number, or
In two or more legs that can move back and forth, one or more legs out of the two or more legs are prohibited from moving back and forth in accordance with the front and back positional relationship between the two or more legs. contains a second negative rule, or
In two or more legs that can be raised and lowered, it is prohibited to move each of the two or more legs forward in a lowered state corresponding to the state in which the leg is on the ground. Include a third negative rule and a fourth negative rule that prohibits moving the leg up and backwards in response to the leg not landing on the ground, or
Regarding each of the two or more legs, a fifth negative rule that prohibits the movement of other legs that have a predetermined positional relationship with the leg when the position of the leg is forward or backward. contains or
The robot includes a sixth negative rule that permits prohibited movements for each of the two or more legs according to a combination of a state of raising and lowering the leg and whether or not the leg is on the ground . 8. The robot according to claim 7 , wherein said sixth negative rule can permit two or more types of motion for the same state information group. A robot control method for controlling a robot having two or more legs,
It is a rule associated with a partial state information group that is a set of one or more state information, and is realized by a storage unit that stores one or more negative rules indicating prohibited items, and a control unit,
The control unit acquires a state information group that is a set of state information indicating at least the state of each of the two or more legs, and determines a negative rule corresponding to each of one or more partial state information groups included in the state information group. A control step of acquiring a group of state information that does not match and controlling each of the two or more legs based on the group of state information,
A robot control method, wherein at least some of the two or more legs do not, at least temporarily, perform an action based on the partial state information group ,
In the control step,
A state information group indicating the state of each of the two or more partial circuits in the core circuit, which is two or more partial circuits having a diode, a capacitor, and a switch, and has a partial circuit that controls the movement of the two or more legs. obtaining a group of state information that does not match the negative rule corresponding to each of the one or more partial state information groups possessed by the group of state information, and applying current to each of the two or more partial circuits based on the group of state information or stop the current,
At least some of the one or more capacitors included in each of the two or more partial circuits have different time constants for charging or discharging,
A robot control method according to claim 1, wherein at least some of the two or more partial circuits included in the core circuit do not operate based on the partial state information group at least temporarily due to the different time constants . A robot control method for controlling a robot having two or more legs,
It is a rule associated with a partial state information group that is a set of one or more state information, and is realized by a storage unit that stores one or more negative rules indicating prohibited items, and a control unit,
The control unit acquires a state information group that is a set of state information indicating at least the state of each of the two or more legs, and determines a negative rule corresponding to each of one or more partial state information groups included in the state information group. A control step of acquiring a group of state information that does not match and controlling each of the two or more legs based on the group of state information,
At least some of the two or more legs do not perform an action based on the partial state information group, at least temporarily,
The one or more negative rules are
includes a first negative rule that prohibits lifting a leg if the number of legs showing a landing condition is less than or equal to a predetermined number, or
In two or more legs that can move back and forth, one or more legs out of the two or more legs are prohibited from moving back and forth in accordance with the front and back positional relationship between the two or more legs. contains a second negative rule, or
In two or more legs that can be raised and lowered, it is prohibited to move each of the two or more legs forward in a lowered state corresponding to the state in which the leg is on the ground. Include a third negative rule and a fourth negative rule that prohibits moving the leg up and backwards in response to the leg not landing on the ground, or
Regarding each of the two or more legs, a fifth negative rule that prohibits the movement of other legs that have a predetermined positional relationship with the leg when the position of the leg is forward or backward. contains or
A robot control method including a sixth negative rule that permits prohibited movement of each of the two or more legs according to a combination of a state of raising and lowering the leg and whether or not the leg is on the ground. . A computer of a robot having two or more legs stores one or more negative rules that are rules associated with a partial state information group that is a set of one or two or more pieces of state information and that indicate prohibited matters. is accessible to the containment
said computer,
Obtaining a state information group that is a set of state information indicating at least the state of each of the two or more legs, and obtaining a state information group that does not match a negative rule corresponding to each of one or more partial state information groups included in the state information group and functions as a control unit that controls each of the two or more legs based on the state information group,
A program characterized in that at least some of the two or more legs do not, at least temporarily, perform an action based on the partial state information group ,
The control unit
A state information group indicating the state of each of the two or more partial circuits in the core circuit, which is two or more partial circuits having a diode, a capacitor, and a switch, and has a partial circuit that controls the movement of the two or more legs. obtaining a group of state information that does not match the negative rule corresponding to each of the one or more partial state information groups possessed by the group of state information, and applying current to each of the two or more partial circuits based on the group of state information or stop the current,
At least some of the one or more capacitors included in each of the two or more partial circuits have different time constants for charging or discharging,
Since the time constants are different, at least some of the two or more partial circuits included in the core circuit do not operate based on the partial state information group at least temporarily. A program that makes a computer work . A computer of a robot having two or more legs stores one or more negative rules that are rules associated with a partial state information group that is a set of one or two or more pieces of state information and that indicate prohibited matters. is accessible to the containment
said computer,
Obtaining a state information group that is a set of state information indicating at least the state of each of the two or more legs, and obtaining a state information group that does not match a negative rule corresponding to each of one or more partial state information groups included in the state information group and functions as a control unit that controls each of the two or more legs based on the state information group,
A program characterized in that at least some of the two or more legs do not, at least temporarily, perform an action based on the partial state information group,
The one or more negative rules are
includes a first negative rule that prohibits lifting a leg if the number of legs showing a landing condition is less than or equal to a predetermined number, or
In two or more legs that can move back and forth, one or more legs out of the two or more legs are prohibited from moving back and forth in accordance with the front and back positional relationship between the two or more legs. contains a second negative rule, or
In two or more legs that can be raised and lowered, it is prohibited to move each of the two or more legs forward in a lowered state corresponding to the state in which the leg is on the ground. Include a third negative rule and a fourth negative rule that prohibits moving the leg up and backwards in response to the leg not landing on the ground, or
Regarding each of the two or more legs, a fifth negative rule that prohibits the movement of other legs that have a predetermined positional relationship with the leg when the position of the leg is forward or backward. contains or
A program comprising a sixth negative rule that permits prohibited movements for each of the two or more legs according to a combination of a state of raising and lowering the leg and whether or not the leg is on the ground.

Images