JP7173059B2 - DISTRIBUTED CONTROL SYSTEM, CONTROL METHOD FOR DISTRIBUTED CONTROL SYSTEM, AND COMPUTER PROGRAM - Google Patents

Description

The present invention relates to a distributed control system, a method of controlling a distributed control system, and a computer program.

A system is known in which a control object is transported by a plurality of robots (see Patent Document 1, for example). Patent Literature 1 discloses a cooperative transport robot system in which various models are used to estimate an external force applied to a control object and to calculate a control amount for the control object. Patent Literature 2 discloses a flying object provided with thrust generating means for generating thrust for controlling its position and attitude. Non-Patent Document 1 discloses a system in which a plurality of robots transport objects to be controlled. In this system, the velocity of the controlled object is modeled, and the controlled object is controlled using various estimated parameters in a state in which the characteristics of the controlled object and the positions at which multiple robots grip the controlled object are unknown. .

JP 2015-99524 A JP 2009-248808 A

Preston Culbertson, Mac Schwager, "Decentralized Adaptive Control for Collaborative Manipulation", IEEE International Conference on Robotics and Automation (ICRA), 21-25 May 2018

However, in the technique described in Patent Literature 1, various models are used to estimate the external force and determine the control amount. Therefore, if the parameters of the model used are different from the actual parameters, there is a risk that the control will be greatly affected. In addition, in the technique described in Patent Document 2, since centralized control is performed, the control amount of each thrust generating means is determined after all information such as disturbance is captured. Therefore, various situation determinations and a function of adjusting the control law according to the determination results are indispensable, which increases the calculation load and the number of sensors of the system. In addition, the method of determining the situation and the method of adjusting the control law cannot be set unless it is known in advance what will happen, so there is a problem in terms of robustness against various characteristics and environmental changes. In the technique described in Non-Patent Document 1, in order for the estimated parameters to be close to the actual parameters, the structure of the modeled controlled object and the actual controlled object are very similar, and the unknown parameter fluctuations is small. Therefore, it is difficult to control the controlled object using the technique of Non-Patent Document 1 except for very simple cases and limited conditions. Also, depending on the setting of the gain value of the adjustable parameter, it may take time to converge to an appropriate value or diverge. Therefore, it is difficult to reliably operate the control system under various conditions using the technique described in Non-Patent Document 1.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and it is an object of the present invention to stably control a controlled object even in the absence of detailed information about the controlled object and a control device that controls the controlled object. . A distributed control system for achieving a common goal with respect to an arbitrary controlled object, wherein a plurality of agents respectively installed in a plurality of controlled devices performing actions for achieving the goal in cooperation, A plurality of agents respectively controlling movements of a plurality of controlled devices; a central control unit that uniformly transmits to the plurality of agents first information, which is a target value of a state, or second information, which is information related to each of the observation value and the target value; wherein said agent receives said first information or said second information obtained from said central control unit without exchanging information regarding movement control of said other controlled device with said other agent. to cause the state of the controlled object to follow the target value by controlling the movement of the controlled device. In addition, the present invention can also be implemented as the following modes.

The present invention has been made to solve the above-described problems, and can be implemented as the following modes.

(1) According to one aspect of the present invention, a distributed control system is provided. This distributed control system consists of a plurality of agents respectively installed in a plurality of controlled devices that work together to achieve a common goal, and controls the movements of the plurality of controlled devices. a plurality of agents; and a central control unit that obtains information about observed values for achieving the objective and transmits information about the observed values to the plurality of agents, wherein the agents interact with other agents. without exchanging information about the control of the controlled device, using the information about the observed value obtained from the central control unit, to control the movement of the controlled device so as to perform the operation for achieving the purpose.

According to this configuration, the plurality of agents use the information on the observed values transmitted from the central control unit to control the movements of the controlled devices on which they are mounted. Each agent cooperates to cause the controlled devices to perform actions for achieving a common goal without exchanging information on the movements of the controlled devices controlled by each other. That is, the agent indirectly considers the influence of the controlled device controlled by itself according to the movement of the controlled device controlled by another agent using the observed value, The influence is modeled as a simple output characteristic. As a result, the agent can control the controlled device even if information such as the position of another controlled device cannot be accurately grasped. That is, according to this configuration, the object to be controlled can be stably controlled. Also, a plurality of controlled devices are independently controlled by each installed agent. Therefore, even if the characteristics of a certain controlled device or the characteristics of the central control unit itself change, the common object can be achieved by controlling other controlled devices. As a result, even if one of the plurality of controlled devices is replaced, there is no need to reconfigure the replaced controlled device to match the other controlled devices.

(2) In the distributed control system of the aspect described above, the object includes moving an object, and the central control unit controls the movement of the object when each of the plurality of controlled devices applies force to the object. from the amount of change in the state of the controlled device to calculate an error (or a command value) with respect to the target, and transmit the error to the plurality of agents, and the agent calculates a command value using the error to calculate the command value of the controlled device can control the movement of
According to this configuration, the error (or the command value) is calculated by the central control unit, and the central control unit performs the arithmetic processing that is commonly performed by each agent. The load on the entire distributed control system can be reduced compared to the case of calculating .

(3) In the distributed control system of the above aspect, the central control unit includes sensors for detecting changes in the state of the object when the plurality of controlled devices apply forces to the object, and , an error calculation unit for calculating the command value including the difference resolved into a plurality of coordinate axes from the difference between the amount of change in the state of the object detected by the sensor; and the command calculated by the error calculation unit. a communication unit configured to transmit a value to the agent, the agent using a receiving unit configured to receive the command value; and a positive/negative value of the difference, which is included in the command value and which is resolved along the predetermined axis, and and an agent-side controller for controlling movement of the controlled device without using the magnitude of the resolved difference along the predetermined axis.
According to this configuration, the agent-side control section decomposes the amount of change in the state of the object acquired via the receiving section into differences along one or more axes including the predetermined axis. The agent-side control unit controls the movement of the controlled device by using only the sign of the difference, without using the magnitude of the difference resolved along the predetermined axis. That is, the agent-side control unit determines the output to the controlled device using only positive/negative determination results of the difference resolved along the predetermined axis. This eliminates the need for the agent-side controller to calculate different outputs that change according to the magnitude of the difference. Therefore, the computational load is suppressed, and the controlled device is easily controlled.

(4) In the distributed control system of the aspect described above, when the difference resolved along the predetermined axis is positive, the agent-side control unit presets output to the controlled device, and if the difference resolved along the predetermined axis is zero or less, regardless of the magnitude of the difference resolved along the predetermined axis, output to the controlled device Output may be zero.
According to this configuration, the agent-side control unit uniformly sets the output to the controlled device to zero when the difference resolved along the predetermined axis is zero or less. This further simplifies the control of the controlled device.

(5) In the distributed control system of the aspect described above, the agent-side control unit is arranged such that the difference resolved along the predetermined axis is zero or less and the magnitude of the difference resolved along the predetermined axis is less than a predetermined value. In this case, a preset negative constant may be output to the controlled device.
According to this configuration, the agent-side control unit sets the output of the controlled device to a negative constant when the difference resolved along the predetermined axis is a negative value greater than or equal to a predetermined value. Therefore, when the difference resolved along the predetermined axis is zero or less, the agent-side control unit switches the output between two ways according to the magnitude of the absolute value. And the precision and responsiveness of control for objects can be improved.

(6) In the distributed control system of the aspect described above, the agents prioritize the resolved differences in descending order of absolute value, and the sum of the values using the finite number of resolved differences with the highest priority An error integrating section that calculates a certain integrated error, and a device control section that controls the movement of the controlled device using the calculated integrated error may be provided.
According to this configuration, only a finite number of decomposed differences in descending order of priority are used as differences to be output to the controlled device. That is, a difference that has a small effect on the output is not included in the calculation target as the output of the controlled device. As a result, it is possible to suppress the influence of the low-priority difference on the output of the controlled device while reducing the computational load.

(7) In the distributed control system of the aspect described above, the agent multiplies each of the decomposed differences by a weight individually set for each decomposed difference. and a device control unit for controlling the movement of the controlled device using the calculated integrated error.
According to this configuration, each value is calculated by multiplying the decomposed difference by the weight set for each decomposed difference. By changing the weighting, it is possible to adjust the parameters of the controlled device that have a large effect on the objective so that they have a large effect on the integrated error. Thereby, it is possible to improve the control performance for the difference having a high degree of importance.

(8) In the distributed control system of the above aspect, the agent calculates the sum of each of the decomposed differences that is equal to or greater than a threshold value set individually for each decomposed difference. An error integrating section that calculates a certain integrated error, and a device control section that controls the movement of the controlled device using the calculated integrated error.
According to this configuration, when the decomposed difference is equal to or greater than the threshold value set in advance for each difference, it is included in the calculation target of the integrated error. Therefore, by changing the threshold value, it is possible to include differences that have a large effect on the purpose in the integration error, and to exclude differences that have a small impact from the integration error. As a result, it is possible to reduce the computational load on the system of this configuration while suppressing the influence on the output of the controlled device.

(9) In the distributed control system of the aspect described above, the agent includes an error integration unit that calculates integrated errors obtained by weighting and ranking the decomposed differences, and a dynamic and a device control unit for controlling the movement of the controlled device using the dynamic adjustment error whose phase and gain are adjusted by the dynamic error adjustment unit. good too.
Each controlled device is affected by changes in the states of other controlled devices that are indirectly included in the integrated error as a change in objective. According to this configuration, the decomposed difference used to calculate the integrated error is not the value as it is, but the value to which dynamic processing has been added. Therefore, it is possible to dynamically change the effect of error between agents and between agents, and highly adaptive control to changes in the target and environment, such as responding to differences in responsiveness and changing the behavior of each agent. system can be constructed.

(10) In the distributed control system of the above aspect, the object includes moving an object, and the central control unit controls the movement of the object when each of the plurality of controlled devices applies force to the object. is acquired as information about the observed value and transmitted to the plurality of agents, and the agents control the movement of the controlled device using the amount of change in the state of the object. good.
By using the amount of change in the state of the object as information about the observed value, this configuration can be used to move the object to the desired location.

(11) In the distributed control system of the above aspect, the central control unit or the agent determines the amount of change in the state of the object when each of the plurality of controlled devices applies force to the object to achieve the objective. It may be acquired in advance before the control for.
According to this configuration, by using the control of the controlled device and the amount of change in the state of the object that have been obtained and linked in advance, it is possible to improve the accuracy of achieving the objective.

It should be noted that the present invention can be implemented in various aspects, for example, a control device, a distributed control system, devices and systems comprising these, control methods, and computers for executing these systems and methods. It can be implemented in the form of a program, a server device for distributing the computer program, a non-temporary storage medium storing the computer program, or the like.

1 is a schematic perspective view of a distributed control system according to an embodiment of the invention; FIG. 1 is a schematic block diagram of a central controller and one controlled device; FIG. 4 is a flow chart showing a control method of the distributed control system; It is a graph showing time transition of the position change of an object. It is a graph showing the time transition of the attitude|position change of an object. It is the time transition of each output in the period shown in FIG. It is the time transition of each output in the period shown in FIG. FIG. 11 is a schematic block diagram of a central control device and one controlled device in a third modification of the first embodiment; FIG. 11 is a flow chart of an output calculation method in the third modified example of the first embodiment; FIG. FIG. 11 is an explanatory diagram of follow-up performance in a third modified example of the first embodiment; FIG. 5 is an explanatory diagram of follow-up performance in a comparative example; FIG. 11 is a schematic block diagram of a central control device and one controlled device in a fourth modification of the first embodiment; It is a schematic block diagram of the distributed control system of 2nd Embodiment. 8 is a flow chart of a control method of the distributed control system of the second embodiment;

<First embodiment>
FIG. 1 is a schematic perspective view of a distributed control system 100 according to an embodiment of the invention. The distributed control system 100 of the first embodiment is a transport system that flies and transports an object OB to be transported to a destination. As shown in FIG. 1, the distributed control system 100 includes a central controller (central controller) 50 attached to an object OB, and four controlled units controlled using information transmitted from the central controller 50. and devices 10-40. Note that in FIG. 1, the four controlled devices 10 to 40 and the central control device 50 are represented in a simplified manner.

The four controlled devices 10-40 have the same configuration and shape. Each controlled device 10 to 40 is fixed at a position where it is possible to move the object OB to a target, and its position is unknown. In FIG. 1, absolute coordinates AC composed of X-, Y-, and Z-axes, and relative coordinates RC set in the central controller 50 composed of x-, y-, and z-axes. It is shown. The Z-axis in the absolute coordinates AC and the z-axis in the relative coordinates RC are axes parallel to the direction of gravity. In this embodiment, the angle around the x-axis is defined as φ, the angle around the y-axis is defined as θ, and the angle around the z-axis is defined as ψ.

FIG. 2 is a schematic block diagram of central controller 50 and one controlled device 10 . As shown in FIG. 2 , the central controller 50 includes a sensor 51 that acquires information about observed values, a first controller 52 that controls the central controller 50 , a battery 57 for the first controller 52 and the sensor 51 . and Sensor 51 is an acceleration sensor and a gyro sensor. Since the central controller 50 is fixed to the object OB, the sensor 51 detects the position (coordinate values) and orientation (angles φ, θ, ψ) of the object OB in the relative coordinates RC. Note that the detected position and orientation of the object OB are hereinafter also referred to as state quantities of the object OB.

As shown in FIG. 2, the first control unit 52 includes a CPU (Central Processing Unit) 53, a ROM (Read Only Memory) 55, a RAM (Random Access Memory) 56, and a communication unit 54. . The communication unit 54 transmits various types of information to each of the controlled devices 10 to 40 by performing wireless communication. Also, the communication unit 54 acquires the coordinate values and the orientation of the destination, which is the destination of the object OB, in the absolute coordinates AC, which are input from an input unit (for example, a keyboard or a mouse) (not shown). A destination in this embodiment corresponds to a common goal that the four controlled devices 10 to 40 cooperate to achieve. The coordinate values and attitude of the destination in the absolute coordinates AC are also simply referred to as target values (targets). The communication unit 54 broadcasts the state quantity of the object OB from the central control unit 50 every time the clock period dT(s) elapses. Note that the communication unit 54 may unicast transmit the state quantity of the object OB from the central control device 50 to the four controlled devices 10 to 40 .

The CPU 53 is connected to the ROM 55 and the RAM 56 , and functions as a state prediction section 531 and an error calculation section 532 by developing a computer program stored in the ROM 55 in the RAM 56 and executing it. The state prediction unit 531 uses the detection values of the sensor 51 to calculate state quantities including the acceleration and angular velocity of the object OB. The state prediction unit 531 calculates the predicted position (x(t+τ), y(t+τ), z(t+τ)) and the predicted yaw angle (Ψ( Calculate the predicted value of t+τ)).

The error calculation unit 532 uses the predicted value after τ seconds calculated by the state prediction unit 531 and the target value of the object OB acquired via the communication unit 54 to calculate the state quantity after τ seconds and the target value. The prediction errors (e _x , e _y , e _z , e _Ψ ) of are calculated by the following equation (2). Note that the prediction errors (e _x , e _y , e _z , e _Ψ ) in the first embodiment correspond to the target error and the command value. Also, each component (eg, " _ex ") included in the prediction error (ex, _ey , _ez , e[ _psi ]) corresponds to an error resolved along a predetermined _axis .

Note that the prediction errors (e _x , e _y , e _z , e _Ψ ) represented by Equation (2) are errors in the relative coordinates RC. Therefore, the error calculator 532 converts the target values (X _m , Y _m ) in the absolute coordinates AC to the target values (x _m , y _m ) in the relative coordinates RC by performing the coordinate conversion shown in the following equation (3). Convert to The prediction errors (e _x , e _y , e _z , e _Ψ ) calculated using equations (2) and (3) are transmitted to each of the controlled devices 10-40 via the communication unit 54. FIG.

Controlled device 10 shown in FIG. A battery 11 is provided. The propeller 13 is rotated by driving the motor 12, and the controlled device 10 flies. In FIG. 2, since the controlled devices 10 to 40 have the same configuration, only the controlled device 10 will be described. The same control as that for the controlled device 10 is performed.

The second control unit 14 includes a CPU 15, a ROM 17, a RAM 18, a receiving unit 16, and a storage unit 19 including a hard disk drive (HDD). The receiving unit 16 receives prediction errors ( _ex , _ey , _ez , e[ _psi ]) transmitted from the communication unit 54 of the central control unit 50 by performing wireless communication. The CPU 15, ROM 17, and RAM 18 correspond to the agent-side controller.

The CPU 15 is connected to the ROM 17 and the RAM 18 , and functions as an error integrating section 151 and a motor control section (apparatus control section) 152 by developing a computer program stored in the ROM 17 in the RAM 18 and executing it. The error integrator 151 calculates an integrated error E _i that takes into consideration the effect of each input on each output, represented by the following equation (4).

In the equation (4), in this embodiment, the error integration unit 151 calculates the action _{coefficient q} The integrated error E _i is calculated by setting positive (q _ij =1) for _{ij and negative (q ij =} −1) for the increasing action coefficient q _ij . That is, the error integration unit 151 assigns a preset constant action coefficient q _ij using the positive/negative of each element such as “e _x ” with respect to u _i irrespective of the size of each element. Also, the error integration unit 151 uses the weight g _i to rank the targets to be controlled. That is, the error integration unit 151 calculates the integrated error E _i which is the sum of the values obtained by multiplying each element such as “e _x ” by a weight g _i individually set for each element. .

The error integrating section 151 calculates the output u _i using the calculated integrated error E _i and the following equation (5). The motor control unit 152 controls the position and orientation of the object OB by outputting a voltage corresponding to the calculated output _ui to the motor 12 .

FIG. 3 is a flow chart showing the control method of the distributed control system 100. As shown in FIG. The flowchart in FIG. 3 shows that the position (Xi, Yi, _Zi ) and yaw angle _{( Ψi} ) of each controlled device 10 follow the target values _{( Xmi} , _Ymi , _Zmi , _Ψmi ). It shows the control flow when it is controlled as follows. As shown in FIG. 3, first, the first control unit 52 acquires a target value at the absolute coordinates AC as the transport destination of the object OB via the communication unit 54 (step S1). The error calculator 532 converts the target value in the absolute coordinates AC obtained using Equation (3) into a target value in the relative coordinates RC (step S2).

The sensor 51 acquires sensor values of the acceleration sensor and the gyro sensor (step S3). The state prediction unit 531 calculates the state quantity of the object OB (the position and orientation of the object OB) using the detection value of the sensor 51 (step S4). The error calculation unit 532 calculates the predicted values (x(t+τ), y(t+τ), z(t+τ)), Ψ (t+τ)) is used to calculate prediction errors (e _x , e _y , e _z , e _Ψ ) represented by Equation (2) (step S5).

Each of the controlled devices 10 to 40 acquires prediction errors (e _x , e _y , e _z , e _Ψ ) through their respective receivers (step S6). Although each of the controlled devices 10 to 40 performs the control flow described below, since the control flow in each of the controlled devices 10 to 40 is the same, the control flow performed by the controlled device 10 will be described. A description of the control flow performed by the control devices 20 to 40 is omitted.

The error integration unit 151 of the controlled device 10 uses the obtained prediction errors (e _x , e _y , e _z , e _Ψ ) to calculate the integrated error E _i shown in Equation (4) (step S7). . The error integrating unit 151 uses the calculated integrated error E _i to calculate the output u _i shown in Equation (5), and the motor control unit 152 uses the output u _i to control the motor 12 (step S8 ).

The error integration unit 151 updates the time t ₁ to the time obtained by adding the clock cycle dT (step S9). The error integration unit 151 determines whether or not the time t ₁ has passed the calculation cycle Tc(s) (step S10). The calculation cycle Tc is the time required to calculate various parameters. When it is determined that the time t ₁ has not passed the calculation cycle Tc (step S10: NO), the error integration unit 151 continues to wait for the calculation cycle Tc of the time t ₁ to elapse. When it is determined that the time t ₁ has passed the calculation period Tc (step S10: YES), the error integration unit 151 resets the time t ₁ to zero (step S12). After that, the receiving unit 16 of the controlled device 10 acquires the prediction errors ( _ex , _ey , _ez , e[ _psi ]) (step S6), and the processes after step S7 are performed. These control flows are performed in each of the controlled devices 10 to 40 until the object OB reaches its destination.

FIG. 4 is a graph showing the temporal transition of the position change of the object OB. FIG. 4 shows the transition on each coordinate axis until the position of the object OB reaches the target value ( _Xm , _Ym , _Zm ). As shown in FIG. 4, the object OB approaches the target value over time.

FIG. 5 is a graph showing the temporal transition of the attitude change of the object OB. FIG. 5 shows the transition until the appearance of the object OB changes to the target posture (Ψ _m ) as the position of the object OB approaches the target value. As shown in FIG. 5, the object OB approaches the target posture over time.

FIG. 6 shows temporal transitions of the outputs u ₁ to u ₄ during the period T1 shown in FIG. FIG. 7 shows time transitions of the outputs u ₁ to u ₄ during the period T2 shown in FIG. FIG. 6 shows the control amount of the motor in a transient state before the object OB reaches the target value. On the other hand, FIG. 7 shows the control variables of the motor in steady state after the object OB reaches the target value. As shown in FIGS. 6 and 7, the output u ₁ -u ₄ of each controlled device 10-40 determined by the integrated error E _i switches between the predetermined value U _b shown in equation (5) and 0. is happening frequently.

As described above, the distributed control system 100 of the first embodiment includes a plurality of second controllers 14 mounted on the controlled devices 10 to 40 and the central controller 50 . Controlled devices 10-40 cooperate to achieve a common goal of transporting object OB to its destination. The central control unit 50 detects the state quantity of the object OB from the sensor 51 and transmits the detected state quantity to each second control unit. For example, the second control unit 14 of the controlled device 10 controls the controlled device 10 without exchanging control information with the other controlled devices 20-40. Therefore, in the first embodiment, the second control unit 14 causes the controlled device 10 controlled by the second control unit 14 to use observation values of the effects of control of the other controlled devices 20 to 40 on the object OB. It is only necessary to indirectly consider the effect of the control itself on the target as a simple output characteristic model. As a result, the second control unit 14 can control the controlled device 10 even if information such as the positions of the other controlled devices 20 to 40 cannot be accurately grasped. That is, according to the distributed control system 100, the object OB can be stably controlled. In addition, each of the controlled devices 10-40 is independently controlled by the second controller 14 mounted on each of the controlled devices 10-40. Therefore, even if the characteristics of a certain controlled device or the characteristics of the central control device 50 change, the object OB is transported to the destination under the control of another controlled device. As a result, even if one of the plurality of controlled devices is replaced, it is not necessary to reconfigure the replaced controlled device to match the other controlled devices.

In addition, the central control unit 50 of the first embodiment calculates the error (or command value) with respect to the target, transmits the error to each of the controlled devices 10 to 40, and transmits the command value within each of the controlled devices 10 to 40. Calculate As a result, the central control unit 50 performs the arithmetic processing commonly performed by the controlled devices 10 to 40 . Therefore, the load on the distributed control system 100 is reduced compared to the case where each of the controlled devices 10 to 40 performs everything from error calculation to command value calculation.

Also, the error integration unit 151 of the first embodiment determines the output u _i depending on whether the integrated error E _i is positive or negative, as shown in Equation (5). Therefore, the error integration unit 151 can calculate the integrated error _Ei without calculating the magnitude of each component of the prediction error, using only the positive/negative determination of the integrated error _Ei to determine the output _ui . As a result, the calculation load on the error integration unit 151 is suppressed, and the controlled device 10 is easily controlled.

Also, the error integration unit 151 of the first embodiment sets the output u _i to zero when the integration error E _i is negative, as shown in Equation (5). Therefore, the computational load of the error integration unit 151 is further suppressed, and the controlled device 10 is controlled more easily.

Also, the error integrating unit 151 of the first embodiment calculates integrated error E _i using a value obtained by multiplying each difference by weight g _i as shown in Equation (4). Thus, by changing the weight g _i , it is possible to increase the influence of the integrated error E _i of the errors of high importance during the transportation of the object OB. This improves the accuracy of the amount of change in the state of the object OB.

<Modified example of embodiment>
The present invention is not limited to the above-described embodiments, and can be implemented in various aspects without departing from the scope of the invention. For example, the following modifications are possible.

<First Modification of First Embodiment>
In the first embodiment, the error integration unit 151 uses the weight g _ij to calculate the integrated error E _i as shown in Equation (4). Prioritize from the m controlled devices as shown in (6). Based on the result of the priority order, the error integration unit 151 calculates a finite number of prediction errors (e _ij (i=1, 2, . . . , 4), (j=x, y , z, ψ) ) and the action coefficient q _ij , the integrated error E _i is calculated. The same control as in the first embodiment is performed using the calculated integrated error E _i of the first modified example.

In the first modified example, the error integration unit 151 prioritizes the product of the absolute value of the element included in each prediction error and the weight g _ij in descending order, as shown in Equation (6). The error integration unit 151 uses only the differences of a finite number of elements (one element in Equation (7)) in descending order of priority to calculate the integration error E _i . As a result, by changing the number of elements used to calculate the integrated error _Ei , the parameters of the controlled devices 10 to 40 that greatly affect the object OB can have a greater influence on the integrated error Ei. This improves the accuracy of the amount of change in the state of the object OB.

<Second Modification of First Embodiment>
In the second modification, unlike the first embodiment, the threshold C _i for determining each element included in each prediction error (e _j ) of m controlled devices is stored in the storage unit 19 in advance. ing. The error integration unit 151 determines whether or not the absolute value of each element is equal to or greater than the corresponding threshold C _i , and uses only the elements equal to or greater than the threshold C _i to calculate the integrated error E _i . For example, when there is a relationship as shown in Equation (8) below, the error integration unit ₁₅₁ calculates integrated error Ei shown in Equation (9) below. The error integration unit 151 calculates an integrated error E _i that maximizes the product of the absolute value of each error and the weight g _i as shown in the following equation (9) from the result of the priority order. The same control as in the first embodiment is performed using the calculated integrated error E _i of the second modification.

In the second modified example, the error integration unit 151 uses only differences whose absolute values are equal to or greater than a threshold value C _i set individually in advance, as shown in Equation (8), and uses Equation (9) as The integrated error E _i is calculated as shown in . Accordingly, by changing each threshold value C _i used for calculating the integrated error E _i , the influence of the parameters of the controlled devices 10 to 40 that have a large effect on the object OB or are important on the integrated error E _i can be increased. can. This improves the accuracy of the amount of change in the state of the object OB.

<Third Modification of First Embodiment>
FIG. 8 is a schematic block diagram of a central control device 50 and one controlled device 10a in the third modification of the first embodiment. In the third modification, compared with the first embodiment, the CPU 15a of the second control unit 14a of the controlled device 10a performs dynamic processing on the integrated error. , the output calculation of the error integration unit 151a, and the control of the motor control unit 152a are different, and the rest of the configuration and control are the same as those of the first embodiment. Therefore, in the third modified example, the configuration and control different from those of the first embodiment will be described, and the description of the same configuration and control will be omitted. In the third modified example, other controlled devices 20a to 40a also have the same configuration as the controlled device 10a.

The error adjustment unit 153 dynamically processes each element included in the prediction errors (e _x , e _y , e _z , e _Ψ ) received from the central control unit 50 via the reception unit 16, A prediction error (corresponding to a phase described later) is calculated. In the third modified example, the error integrating unit 151 calculates the integrated error _{Ei using the prediction error after dynamic processing instead of the prediction error (ex, ey , ez} , _e [ _psi ]).

Specifically, the error adjuster 153 uses the prediction error to calculate the dynamically processed phase α _i represented by Equation (10). Note that the numerator in Equation (10) corresponds to the prediction error. The maximum output value U _max in the denominator is the output value when the maximum voltage is applied to the motor 12 . The motor control unit 152a of the third modified example controls the motion of the controlled device 10a using the prediction error whose phase and gain are adjusted by the error adjustment unit 153. FIG. Note that the prediction error in the third modified example corresponds to the dynamic adjustment error in which the phase and gain are adjusted.

In the first embodiment, as shown in equation (5), the output u _i is U _b for ON and zero for OFF. The change switches the output u _i between ON and OFF. Therefore, the error integration unit 151a sets the phase α _i as shown in the following equation (10). Note that K _i in Equation (10) is a function that determines the tracking performance, and its parameters are appropriately set.

The error integrator 151a calculates the output _ui represented by the following equation (11) using the calculated phase of equation (10).

The error integration unit 151a smoothes the calculated output u _i using the time constant τ and the output value T _i as shown in the following equation (12).

FIG. 9 is a flowchart of a method for calculating the output u _i in the third modified example of the first embodiment. The calculation flow shown in FIG. 9 corresponds to the subflow of step S7 in the control flow of the first embodiment shown in FIG. Therefore, when the prediction error is acquired by the controlled device 10a in step S6 of FIG. 5, the error adjustment unit 153 calculates the phase α _i using the prediction error and equation (10) (step S71 ).

The error integration unit 151a calculates the phase of the oscillator by calculating Equation (10). The error integration unit 151a uses the calculated phase of the oscillator to calculate the output u _i represented by Equation (11) (step S72). Further, the error integration unit 151a calculates a value smoothed by the time constant τ and the output value T _i by substituting the calculated output u _i into Equation (12) (step S73). After that, the smoothed values are used to perform the processing from step S8 onward in FIG.

FIG. 10 is an explanatory diagram of follow-up performance in the third modification of the first embodiment. FIG. 11 is an explanatory diagram of follow-up performance in a comparative example. FIGS. 10 and 11 show time transitions LN1 and LN2 of the output u _i with respect to the time transition LN _tr of the target value indicated by the dashed line. FIG. 10 shows the results when the following relational expression (13) is used in the formula (10), and FIG. 11 shows the results when the following relational formula (14) is used in the formula (10).

The time transition LN1 of the output _ui of the third modified example shown in FIG. 10 has a smaller amplitude than the time transition LN2 of the comparative example shown in FIG. Also, the sum of the differences between the time transition LN1 and the target value is smaller than the sum of the differences between the time transition LN2 and the target value. That is, the follow-up performance of the time transition LN1 of the third modified example is superior to the follow-up performance of the time transition LN2 of the comparative example.

When the error adjustment unit 153 of the third modification performs the processing of equation (11), even if ej is not sufficiently small, Ki(ej, αi)≈0 due to the structure of Ki, and even if e is large, α _i can reduce the vibration caused by the delay in the update of

<Fourth Modification of First Embodiment>
FIG. 12 is a schematic block diagram of a central control device 50b and one controlled device 10b in the fourth modification of the first embodiment. In the fourth modification, part of the functions performed by the first control unit 52 of the central control unit 50 in the first embodiment are largely different in that they are performed by the second control unit 14b of each controlled device. Therefore, in the fourth modification, differences from the first embodiment will be described, and descriptions of the same configuration and control as in the first embodiment will be omitted.

As shown in FIG. 12, the central controller 50b includes a sensor 51, a battery 57, and a communication section 54b. The communication unit 54b transmits the detection value of the sensor 51 to the controlled device 10b. The second control unit 14b of the controlled device 10b includes a CPU 15b, a ROM 17, a RAM 18, a receiving unit 16b, and a storage unit 19. The receiving unit 16b receives the detection value of the sensor 51 transmitted from the communication unit 54b. Further, the receiving unit 16b receives the target value, which is the transport destination of the object OB.

The CPU 15 b functions as an error integration section 151 , a motor control section 152 , a state prediction section 154 and an error calculation section 155 . The state prediction section 154 of the fourth modification performs the same function as the state prediction section 531 of the first control section 52 in the first embodiment. Similarly, the error calculator 155 of the fourth modified example performs the same function as the error calculator 532 of the first controller 52 in the first embodiment. In this way, part of the functions performed by the central control device 50 of the first embodiment may be performed by each controlled device (for example, the controlled device 10b).

In the fourth modification, part of the functions performed by the central control device 50 of the first embodiment are performed by the controlled device 10b. Thus, even if some functions of the central control device 50 are performed by the controlled device 10b, transportation of the object OB to the destination is completed as a common purpose.

<Other Modifications of First Embodiment>
Although an example of the distributed control system 100 has been described in the above-described first embodiment and each modified example, various modifications can be made to each configuration and each control of the distributed control system 100 . For example, the controlled device 10 may not include the storage unit 19, and may include other components (for example, an input unit capable of voice input). The controlled devices 10 to 40 control the controlled devices 10 to 40 on which they are mounted without exchanging information with other controlled devices using the detection values of the sensors 51 transmitted from the central control device 50. Various modifications are possible within the range. The controlled devices 10 to 40 do not have the same configuration, and may have different shapes and different output modes.

In the above-described first embodiment and each modified example, an example of each parameter calculated until the controlled devices 10 to 40 are controlled has been described, but each parameter can be applied in various ways within the range of well-known techniques. For example, angles θ and φ may be used in equation (2). The error integration unit 151 of the first embodiment uses the weights g _i shown in equation (4) in order to calculate the integrated error E _i . , (7)) and the threshold determination of the second modified example (formulas (8) and (9)) may be used together. In the fourth modification, at least the dynamic processing represented by Equation (10) may be applied as the dynamic processing of the error, and the parameters represented by Equations (11) and (12) may be omitted. .

Also, the control flow shown in FIG. 3 can be modified in various ways. For example, within the range in which the prediction errors (e _x , e _y , e _z , e _Ψ ) in step S5 are calculated, the steps before and after each step can be interchanged. For example, the process of obtaining the sensor values in step S3 may be performed before the process of obtaining the target values in step S1.

Further, in the first embodiment, as shown in equation (5), the output u _i is determined in two ways by determining whether the integrated error E _i is positive or negative, but it may be determined in three or more ways. For example, when the integrated error E _i is less than zero and the absolute value of the integrated error E _i is greater than or equal to a predetermined value, the error integration unit 151 sets −U _m (U _m >0) as the output u _i good too. As described above, the output differs depending on the magnitude of the absolute value, thereby improving the accuracy of control of the controlled devices 10 to 40 and the object OB.

In addition, the central control device 50 or the second control unit (for example, the second control unit 14) of the controlled devices 10 to 40 causes each of the controlled devices 10 to 40 to apply force to the object OB before conveying the object OB. A detection value of the sensor 51 at the time of addition may be acquired. Each second control unit may use the acquired detection values to actually control the transportation of the object OB. Note that the detected value of the sensor 51 corresponds to information about the observed value.

<Second embodiment>
FIG. 13 is a schematic block diagram of the distributed control system 200 of the second embodiment. A distributed control system 200 of the second embodiment is a system that controls power supplied from a plurality of power sources PW1 to PWn (n=1, 2, . . . n) to power supply destinations FC. As shown in FIG. 13, the distributed control system 200 includes a power supply destination FC, a power detection device S50 that detects the power supplied to the power supply destination FC, and the power sources PW1 to PWn. A plurality of power converters CH1 to CHn for conversion, agents C1 to Cn for controlling each of the power converters CH1 to CHn, and a plurality of detectors for detecting power supplied from each of the power converters CH1 to CHn to a power supply destination FC. of power detection devices S1 to Sn. Since the agents C1 to Cn have the same configuration, only the block diagram of the agent C1 is shown in FIG. 1, and other block diagrams are omitted. In the following, one agent C1 out of a plurality of agents C1-Cn will be explained, and explanations of the other agents C2-Cn will be omitted.

The central control unit 150 includes a CPU 53C, a communication section 154, a ROM 155, and a RAM 156. The communication unit 154 acquires the current power P, which is the total power supplied from the power detection device S50 to the power supply destination FC. Further, the communication unit 154 acquires the required power Pm, which is the supply target to the power supply destination FC, input by another input device. The CPU 53C functions as an error calculator 539 that calculates an error e (=Pm-P) obtained by subtracting the current power P from the required power Pm. The error calculator 539 transmits the calculated error e to each of the agents C1 to Cn via the communication unit 154. FIG.

Each of the agents C1 to Cn uses the error e sent from the error calculator 539 to control the power supplied to the power supply destination FC. One agent C 1 includes a CPU 115 , a receiver 116 , a ROM 117 , a RAM 118 and a memory 119 . The receiving unit 116 receives the error e periodically transmitted from the error calculating unit 539 and the power detected by the power detecting device S1. Storage unit 119 stores a reference value for determining the magnitude of power supplied from power electronics device CH1 to power supply destination FC. Details of the reference value will be described later.

CPU 115 functions as error integration unit 158 and device control unit 159 that controls power conversion device CH1. The error integration unit 158 uses the received error e to define a switching variable β _i represented by Equation (15) below.

The error integration unit 158 determines whether the value calculated using Equation (15) is positive or negative. Specifically, when e(t)>0, the error integration unit 158 sets K _i (e) as shown in the following equation (16), and when e(t)≦0, K _i (e ) is set as shown in the following equation (17). The error integrator 158 uses β _i (t) calculated by Equation (15) to switch ON/OFF of the power supply switch of power conversion device CH1 as shown in Relational Equation (18) below.

The device control unit 159 controls the power supplied from the power source PW1 to the power supply destination FC according to the ON/OFF state of the switch shown in Equation (18). The power detection device S1 detects the power supplied from the power conversion device CH1 to the power supply destination FC. If the difference between the power detected by the power detection device S1 and the estimated power supplied by the power conversion device CH1 by control is within the reference value stored in the storage unit 119, the device control unit 159 adjusts the power supply. On the other hand, if the difference is greater than the reference value, device control section 159 stops power supply.

FIG. 14 is a flow chart of the control method of the distributed control system 200 of the second embodiment. In the control flow shown in FIG. 14, first, the communication unit 154 of the central control unit 150 acquires the required power Pm required for the power supply destination FC (step S21). The communication unit 154 acquires the current power P supplied from the power detection device S50 to the power supply destination FC (step S22). The error calculator 539 calculates an error e by subtracting the current power P from the required power Pm (step S23). The error calculation unit 539 transmits the calculated error e to each of the agents C1 to Cn via the communication unit 154. FIG. Transmission of the error e is performed periodically.

The receiver (for example, the receiver 116) of each agent C1 to Cn receives the error e transmitted from the central controller 150 (step S25). Note that the control of one agent C1 will be described below. Since the other agents C2-Cn also perform the same control as the agent C1, the explanation of the control of the other agents C2-Cn is omitted. The error integration unit 158 calculates the switching variable β _i by substituting the received error e into Equation (15) (step S26). The error integration unit 158 determines whether the switching variable β _i is positive or negative using the equation (15) (step S27). The error integration unit 158 uses β _i (t) calculated by the equation (15) to control the converted power of the power converter CH1 shown in the equation (15) (step S28). Device control unit 159 controls power conversion device CH1 to convert the power determined by error integration unit 158 .

The power detection device S1 detects the power supplied to the power supply destination FC converted from the power conversion device CH1 (step S29). Device control unit 159 determines whether or not the difference between the detected power supply and the estimated power converted by power conversion device CH1 under control is within a reference value (step S30). If it is determined that the difference is within the reference value (step S30: YES), the device control unit 159 continues the power supply from the power conversion device CH1 to the power supply destination FC (step S31). The processing after S25 is repeated. On the other hand, when it is determined that the difference is not within the reference value (step S30: NO), the device control unit 159 stops the power supply from the power conversion device CH1 to the power supply destination FC (step S32 ), and the processing after step S25 is repeated.

As described above, even if the distributed control system 200 of the second embodiment is different from the distributed control system 100 of the first embodiment, the system can be modeled by simple output characteristics.

<Modification of Second Embodiment>
Although an example of the distributed control system 200 has been described in the above-described second embodiment, each configuration and each control of the distributed control system 200 can be modified in various ways. For example, the agents C1 to Cn use the detection value of the power detection device S50 transmitted from the central control device 150 to detect the power conversion devices CH1 to CHn, which are the control targets of themselves, without exchanging information with other agents. Various modifications are possible within the control range. Agents C1-Cn, power detectors S1-Sn, power converters CH1-CHn, and power sources PW1-PWn do not have the same configuration, and may have different shapes and different output scales.

The present aspect has been described above based on the embodiments and modifications, but the above-described embodiments are intended to facilitate understanding of the present aspect, and do not limit the present aspect. This aspect may be modified and modified without departing from the spirit and scope of the claims, and this aspect includes equivalents thereof. Also, if the technical features are not described as essential in this specification, they can be deleted as appropriate.

10 to 40, 10a, 10b, 20a... controlled device 11, 57... battery 12... motor 13... propeller 14, 14a, 14b... second control unit (agent)
15, 15a, 15b...CPU
16, 16b, 116... Reception unit 19, 119... Storage unit 50, 50b, 150... Central control unit (central control unit)
51... Sensor 52... First control unit 53, 53C, 115... CPU
54, 54b, 154... Communication unit 100, 200... Distributed control system 151, 158, 151a... Error integration unit 152, 152a... Motor control unit (apparatus control unit)
153... Error adjustment unit (dynamic error adjustment unit)
154, 531 state prediction unit 155, 532, 539 error calculation unit 159 device control unit AC absolute coordinates C1 to Cn agent CH1 to CHn power converter C _i threshold E _i integrated error FC power supply Destination LN1, LN2, LN _tr ... Time transition OB ... Object P ... Current power PW1 to PWn ... Power source Pm ... Requested power RC ... Relative coordinates S1 to Sn, S50 ... Power detection device T ... Target value Tc ... Calculation period T _i ... Output value _Ub ... Predetermined value _Umax ... Maximum output value dT ... Clock cycle e ... Errors e _x , e _y , e _z , e _Ψ , e _ij ... Prediction errors (command values)
α _i ... phase β _i ... switching variable g _i , g _ij ... weight q _ij ... effect coefficient t ₁ ... time u ₁ to u ₄ , u _i ... output

Claims (16)

A distributed control system that aims to achieve a common goal for arbitrary controlled objects ,
a plurality of agents respectively installed in a plurality of controlled devices that cooperate to perform actions to achieve the object, the agents respectively controlling movements of the plurality of controlled devices;
A central control unit that acquires information on the controlled object ,
First information that is information about the observed value of the controlled object and a target value of the state of the controlled object for achieving the purpose, or information related to each of the information about the observed value and the target value 2 a central control unit that uniformly transmits information to the plurality of agents;
with
The agent uses the first information or the second information obtained from the central control unit without exchanging information related to movement control of the other controlled device with the other agents. 7. A distributed control system for causing the state of the controlled object to follow the target value by controlling the movement of the controlled device. A distributed control system according to claim 1, wherein
The central control unit or the agent has, as the second information, an error calculation unit that calculates an error between the observed value and the target value,
The distributed control system, wherein each of the plurality of agents calculates the current output of the controlled device using the error as the second information. A distributed control system according to claim 2, wherein
The central control unit or the agent has a state prediction unit that predicts the state of the controlled object after a predetermined time from the current time using information about the observed value,
The error calculation unit calculates, as the second information, a prediction error between the state of the controlled object after the lapse of the predetermined time predicted by the state prediction unit and the target value. A distributed control system according to claim 3 , wherein
The object is to move the object as the object to be controlled to a destination ,
the target value is the position and orientation of the controlled object at the destination;
The distributed control system, wherein the observed value is the current position and attitude of the controlled object. A distributed control system according to claim 4,
The central control unit
a sensor that detects the amount of change in the position and orientation of the object that changes according to the outputs of the plurality of controlled devices;
The error for calculating the observed value of the position and orientation of the controlled object from the detected amount of change, and calculating the error as the second information, which is a difference between the calculated observed value and the target value. a calculation unit;
a communication unit that transmits the error to the plurality of agents;
has
The agent is
a receiver that receives the error;
an agent-side control unit that calculates the current output of the controlled device using the received error;
a distributed control system. A distributed control system according to claim 5 , wherein
The error calculation unit calculates components of the error resolved along each of the predetermined axes in a coordinate axis composed of a plurality of predetermined axes ,
The communication unit transmits each of the components calculated by the error calculation unit to the agent ,
The agent-side control unit is
an error integration unit that calculates an integrated error that is the sum of values obtained by multiplying an action coefficient determined to be positive or negative according to each component, each component, and a weight set according to each component; ,
a device control unit that controls movement of the controlled device using the calculated integrated error ;
If the component multiplied when calculating the integrated error decreases when the output of the controlled device is increased, the action coefficient to be multiplied by the component is set positive, and the output of the controlled device A distributed control system that sets the coefficient of action multiplied by the component negative if it increases when increasing . A distributed control system according to claim 5 , wherein
The error calculation unit calculates components of the error resolved along each of the predetermined axes in a coordinate axis composed of a plurality of predetermined axes,
The communication unit transmits each of the components calculated by the error calculation unit to the agent,
The agent -side control unit is
Prioritization is given in descending order of the product of the absolute value of each of the components resolved along the predetermined axis and the weight set for each of the components, and the finite number of the components with the highest priority; an error integrating unit that calculates an integrated error that is the sum of values obtained by multiplying the coefficients of action set for each of the components ;
a device control unit that controls movement of the controlled device using the calculated integrated error ;
a distributed control system. A distributed control system according to claim 5 , wherein
The error calculation unit calculates components of the error resolved along each of the predetermined axes in a coordinate axis composed of a plurality of predetermined axes,
The communication unit transmits each of the components calculated by the error calculation unit to the agent,
The agent -side control unit is
error integration for calculating an integrated error that is the sum of each of the components resolved along the predetermined axis and calculated using the components that are equal to or greater than a threshold value set individually for each of the components Department and
a device control unit that controls movement of the controlled device using the calculated integrated error ;
a distributed control system. A distributed control system according to any one of claims 6 to 8 ,
The agent-side control unit is
when the integrated error is positive, outputting a preset constant to the controlled device regardless of the magnitude of the integrated error ;
A distributed control system, wherein if the integrated error is less than or equal to zero, the output to the controlled device is set to zero regardless of the magnitude of the integrated error . A distributed control system according to any one of claims 6 to 8 ,
The agent-side control unit is
when the integrated error is positive, outputting a preset constant to the controlled device regardless of the magnitude of the integrated error;
setting the output to the controlled device to zero regardless of the magnitude of the integrated error when the integrated error is less than or equal to zero and is greater than or equal to a predetermined value smaller than zero;
A distributed control system, wherein a preset negative constant is output to the controlled device when the integrated error is less than the predetermined value. A distributed control system according to any one of claims 6 to 10 ,
The agent -side control unit is
a dynamic error adjustment unit that dynamically processes the integrated error obtained by the error integration unit;
a device control unit that controls the output of the controlled device using the dynamic adjustment error whose phase and gain are adjusted by the dynamic error adjustment unit;
a distributed control system. A distributed control system according to claim 5 , wherein
the central control unit acquires and transmits to the plurality of agents changes in the position and orientation of the object that change according to the outputs of the plurality of controlled devices;
The distributed control system, wherein the agent calculates the error and uses the calculated error to calculate the current output of the controlled device. A distributed control system according to any one of claims 5 to 12 ,
The central control unit or the agent obtains in advance the amount of change in the position and orientation of the object, which changes according to the outputs of the plurality of controlled devices, before controlling the controlled object to move to the destination . distributed control system. A distributed control system according to claim 2, wherein
The object is to supply power as the controlled object from a power source to a power supply destination,
The target value is a required power that is a supply target to be supplied to the power supply destination,
The observed value is the power currently being supplied to the power supply destination,
The error calculation unit calculates, as the error as the second information, a power error between the requested power and the power currently being supplied to the power supply destination,
Each of the plurality of agents adjusts power supplied from the power source to the power supply destination according to the power error calculated as the movement of the power conversion device, which is the controlled device to be controlled. , a distributed control system to achieve said power requirements. A control method for a distributed control system that aims to achieve a common goal for arbitrary controlled objects ,
an information acquisition step of acquiring information of the controlled object for achieving the purpose;
information on observed values of the controlled object to a plurality of agents respectively installed in a plurality of controlled devices that cooperate to perform operations for achieving the object and controlling the movement of the plurality of controlled devices respectively ; and uniformly transmit first information that is a target value of the state of the controlled object for achieving the purpose, or second information that is information related to each of the information on the observed value and the target value. an information transmission step;
The movement of the controlled device is controlled using the transmitted first information or the second information without exchanging information on controlling the movement of the other controlled devices between each of the agents . and a control step of causing the state of the controlled object to follow the target value . A computer program that aims to achieve a common goal for any controlled object ,
a function of acquiring information of the controlled object for achieving the purpose;
information on observed values of the controlled object to a plurality of agents respectively installed in a plurality of controlled devices that cooperate to perform operations for achieving the object and controlling the movement of the plurality of controlled devices respectively ; and uniformly transmit first information that is a target value of the state of the controlled object for achieving the purpose, or second information that is information related to each of the information on the observed value and the target value. function and
Controlling the motion of the controlled device using the transmitted first information or the second information without exchanging information on controlling the motion of the other controlled devices between the agents . A function of causing the state of the controlled object to follow the target value by
A computer program that causes a computer to execute

Images