Method for controlling a mobile object, control device for a mobile object, mobile object and program

Abstract

To suppress deterioration in accuracy of input value calculation.SOLUTION: A mobile body control method includes the steps of: calculating an input value to be inputted to a mobile body based on a target output value of the mobile body; inputting the input value to the mobile body to control the mobile body; acquiring an output value of the mobile body when the input value has been inputted to the mobile body; inputting the input value in an output prediction model which imitates the behavior of the mobile body to calculate an estimated output value of the mobile body when the input value has been inputted; acquiring a search parameter as a coefficient indicating correspondence of a difference between the output value and the estimated output value to an adaptive parameter indicating the behavior of the mobile body; and calculating the adaptive parameter based on the difference between the output value and the estimated output value and the search parameter, in which the input value is calculated based on the target output value and the adaptive parameter in the step of calculating the input value.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] This disclosure relates to a method for controlling a mobile body, a control device for a mobile body, a mobile body, and a program. [Background technology]

[0002] In recent years, mobile objects such as underwater vehicles, ships, and aircraft are increasingly being equipped with control devices that automatically operate them. For example, Patent Document 1 describes adjusting control parameters such as gain values ​​in feedback control based on maneuvering data so that the target vessel satisfies specified response characteristics. Patent Document 2 describes calculating control parameters that exhibit behavior close to a normative model designed to show desired response characteristics, based on a transfer function model identified during a calibration flight of an unmanned aerial vehicle. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-187469 [Patent Document 2] Japanese Patent Publication No. 2015-45963 [Overview of the Initiative] [Problems that the invention aims to solve]

[0004] In Patent Documents 1 and 2, when controlling a moving object, control parameters (input values) are calculated using various parameters that are fixed values ​​set in advance through calibration or other means. However, due to the influence of external disturbances, for example, the pre-set parameters and input values ​​may no longer correspond appropriately, which may reduce the accuracy of the control parameter calculation.

[0005] This disclosure aims to solve the aforementioned problems and to provide a control method for a mobile body, a control device for a mobile body, a mobile body, and a program that can suppress a decrease in the accuracy of input value calculation. [Means for solving the problem]

[0006] A method for controlling a mobile body according to this disclosure includes the steps of: obtaining a target output value of the mobile body; calculating an input value to be input to the mobile body based on the target output value; inputting the input value to the mobile body to control the mobile body; obtaining the output value of the mobile body when the input value is input to the mobile body; inputting the input value to an output prediction model that simulates the behavior of the mobile body to calculate an estimated output value of the mobile body when the input value is input; obtaining a search parameter which is a coefficient that shows the correspondence between the error between the output value and the estimated output value and an adaptive parameter that indicates the behavior of the mobile body; and calculating the adaptive parameter based on the error between the output value and the estimated output value and the search parameter, wherein in the step of calculating the input value, the input value is calculated based on the target output value and the adaptive parameter.

[0007] The control device for a mobile body according to this disclosure includes: a first acquisition unit that acquires a target output value of the mobile body; a first calculation unit that calculates an input value to be input to the mobile body based on the target output value; a movement control unit that inputs the input value to the mobile body and controls the mobile body; a second acquisition unit that acquires the output value of the mobile body when the input value is input to the mobile body; a second calculation unit that inputs the input value to an output prediction model that simulates the behavior of the mobile body and calculates an estimated output value of the mobile body when the input value is input; a third acquisition unit that acquires a search parameter, which is a coefficient indicating the correspondence between the error between the output value and the estimated output value and an adaptive parameter that indicates the behavior of the mobile body; and a third calculation unit that calculates the adaptive parameter based on the error between the output value and the estimated output value and the search parameter, wherein the first calculation unit calculates the input value based on the target output value and the adaptive parameter.

[0008] The mobile body relating to this disclosure has the control device.

[0009] The program according to the present disclosure includes steps of obtaining a target output value of a moving body, calculating an input value to be input to the moving body based on the target output value, inputting the input value to the moving body to control the moving body, obtaining an output value of the moving body when the input value is input to the moving body, inputting the input value to an output prediction model that simulates the behavior of the moving body to calculate an estimated output value of the moving body when the input value is input, obtaining a search parameter that is a coefficient indicating a correspondence relationship between an error between the output value and the estimated output value and an adaptation parameter indicating the behavior of the moving body, and calculating the adaptation parameter based on the error between the output value and the estimated output value and the search parameter, and causing a computer to execute the steps. In the step of calculating the input value, the input value is calculated based on the target output value and the adaptation parameter.

Effect of the Invention

[0010] According to the present disclosure, it is possible to suppress a decrease in the calculation accuracy of the input value.

Brief Description of the Drawings

[0011] [Figure 1] FIG. 1 is a schematic diagram showing an example of a moving body according to the present embodiment. [Figure 2] FIG. 2 is a schematic block diagram of a control device according to the present embodiment. [Figure 3] FIG. 3 is a schematic block diagram for explaining a method of controlling a moving body by a control device. [Figure 4] FIG. 4 is a flowchart for explaining a control flow of a moving body according to the present embodiment. [Figure 5] FIG. 5 is a schematic diagram for explaining the setting of a search parameter. [Figure 6] FIG. 6 is a flowchart for explaining a processing flow of robust processing. [Figure 7] FIG. 7 is a graph showing an example of a simulation result of an output value. [Figure 8]FIG. 8 is a flowchart for explaining the processing flow of the determination process. **Embodiments for Carrying Out the Invention**

[0012] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. Note that the present disclosure is not limited by this embodiment, and when there are multiple embodiments, those configured by combining each embodiment are also included.

[0013] (Mobile object) FIG. 1 is a schematic diagram showing an example of a mobile object according to this embodiment. The mobile object 10 according to this embodiment is a device that moves autonomously. The mobile object 10 may be an underwater vehicle that moves in water W (underwater), or may be a ship that moves on the water surface of water W. Further, the mobile object 10 is not limited to an underwater vehicle or a ship, and may be an aircraft that flies, or may be a vehicle that travels on a road surface.

[0014] As shown in FIG. 1, the mobile object 10 includes a control target 12, an output detection unit 14, and a control device 16. The control target 12 is a device controlled by the control device 16, and for example, is a drive mechanism that drives the mobile object 10. The output detection unit 14 is a device that detects an output value from the mobile object 10. The output value from the mobile object 10 refers to the state of the mobile object 10 when the control target 12 is controlled by the control device 16, and in other words, refers to the control output from the mobile object 10 in response to the control of the control target 12 by the control device 16. For example, when performing depth control of the mobile object 10 that is an underwater vehicle, the output value may refer to the depth of the mobile object 10. In this case, the output detection unit 14 may be a sensor that detects the depth of the mobile object 10. Also, for example, when performing navigation control of the mobile object 10 that is a ship, the output value may refer to the position (coordinates) of the mobile object 10. In this case, the output detection unit 14 may be a sensor that detects the position of the mobile object 10. <​​The control device 16 is a device that controls the movement of the mobile body 10. The control device 16 controls the movement of the mobile body 10 by inputting input values ​​to the controlled object 12. That is, when the control device 16 inputs input values ​​to the controlled object 12, the controlled object 12 is driven according to those input values ​​to move the mobile body 10. The input values ​​here refer to the control inputs (control parameters) input to the controlled object 12. For example, when controlling the depth of a mobile body 10 that is an underwater navigating object, the input values ​​may be manipulated variables for changing the depth of the mobile body 10 (e.g., steering angle or payload amount). Also, for example, when controlling the navigation of a mobile body 10 that is a ship, the input values ​​may be manipulated variables for changing the position of the mobile body 10 (e.g., steering angle or applied torque).

[0016] Figure 2 is a schematic block diagram of the control device according to this embodiment. The control device 16 is, for example, a computer and, as shown in Figure 2, has a communication unit 20, a storage unit 22, and a control unit 24. The communication unit 20 is a communication module that communicates with an external device, such as an antenna. The communication method used by the communication unit 20 may be arbitrary and may be wireless communication or wired communication. The storage unit 22 is a memory that stores the calculation contents and program information of the control unit 24, and includes at least one of the following: RAM (Random Access Memory), main memory such as ROM (Read Only Memory), and external storage device such as HDD (Hard Disk Drive).

[0017] The control unit 24 is an arithmetic unit and includes arithmetic circuits such as a CPU (Central Processing Unit). The control unit 24 includes a first acquisition unit 30, a first calculation unit 32, a movement control unit 34, a second acquisition unit 36, a second calculation unit 38, a third acquisition unit 40, and a third calculation unit 42. The control unit 24 reads a program (software) from the storage unit 22 and executes it to realize the first acquisition unit 30, the first calculation unit 32, the movement control unit 34, the second acquisition unit 36, the second calculation unit 38, the third acquisition unit 40, and the third calculation unit 42, and executes their processing. The control unit 24 may execute processing with a single CPU, or it may have multiple CPUs and execute processing with those multiple CPUs. Furthermore, at least a portion of the first acquisition unit 30, the first calculation unit 32, the movement control unit 34, the second acquisition unit 36, the second calculation unit 38, the third acquisition unit 40, and the third calculation unit 42 may be implemented using hardware circuits. Also, the program for the control unit 24 stored in the storage unit 22 may be stored on a recording medium that the control device 16 can read.

[0018] The control method of the mobile body 10 by the control unit 24 will be described below. Figure 3 is a schematic block diagram illustrating the control method of the mobile body by the control device.

[0019] (Obtaining the target output value) As shown in Figure 3, the first acquisition unit 30 (target output value acquisition unit) acquires the target output value r. The target output value r is the target value of the output value of the mobile body 10. For example, when performing depth control of a mobile body 10 that is an underwater navigation body, the target output value r may be the target depth of the mobile body 10. Also, for example, when performing navigation control of a mobile body 10 that is a ship, the target output value r may be the target position of the mobile body 10 (the target destination position). The first acquisition unit 30 may set the target output value r as appropriate according to the purpose of the mobile body 10. Note that the target output value r is not limited to the target value of the output value of the mobile body 10 itself, but may also be an manipulated variable (control input) used to set the output value of the mobile body 10 to the target value. In other words, the target output value r may be a value corresponding to the target value of the output value of the mobile body 10.

[0020] (Calculation of input values) The first calculation unit 32 (input value calculation unit) calculates an input value u to be input to the mobile body 10 (control target 12) based on the target output value r. The first calculation unit 32 calculates the input value u based on the target output value r and the adaptive parameters ω, θ, and σ calculated immediately beforehand. The adaptive parameters are parameters that indicate the behavior of the mobile body 10. The mobile body 10 may behave differently from the ideal due to environmental factors such as disturbances, and the adaptive parameters can be said to correspond to the amount of deviation of the actual behavior of the mobile body 10 from the ideal behavior, and they fluctuate in real time. Therefore, in this embodiment, the adaptive parameters are calculated in real time by the third calculation unit 42, which will be described later, and the first calculation unit 32 calculates the input value u so as to cancel out the parameters that indicate disturbances among the adaptive parameters. The calculation of the input value u that cancels out the parameters that indicate disturbances will be described later. In this embodiment, there are three types of adaptive parameters: ω, θ, and σ, but the number and types are not limited to these and may be arbitrary. The method for calculating the adaptive parameters will also be described later.

[0021] In this embodiment, the first calculation unit 32 calculates the input value u using feedback control. That is, the first calculation unit 32 calculates the input value u based on the target output value r, the adaptive parameters calculated immediately before, and the feedback difference value, which is the difference between the target output value r acquired immediately before and the output value y detected immediately before. In other words, the first calculation unit 32 calculates the input value u in such a way that it cancels out the parameters that indicate disturbances among the adaptive parameters, and that the output value y follows the target output value r through feedback control. However, feedback control is not essential.

[0022] In the example shown in Figure 3, the first calculation unit 32 calculates an adaptive input value ua, which is an input value that cancels out the parameter indicating disturbance among the adaptive parameters, based on the target output value r and the adaptive parameters. In this case, the first calculation unit 32 may remove high-frequency components that constitute noise from the adaptive input value ua using a low-pass filter A1. The high-frequency components to be removed may be components above a predetermined frequency, and the predetermined frequency here may be set as appropriate. Then, the first calculation unit 32 uses a feedback controller FB to calculate a difference input value ub, which is a correction input value (correction gain) that makes the feedback difference value less than or equal to a predetermined value (e.g., zero), from the feedback difference value detected immediately before. The feedback controller FB may be any algorithm capable of calculating the difference input value ub from the feedback difference value. Based on the adaptive input value ua and the difference input value ub, the first calculation unit 32 calculates an input value u, which cancels out the parameter indicating disturbance among the adaptive parameters, and which follows the target output value r through feedback control. Note that the method for calculating the input value u shown in Figure 3 is just one example and is not limited thereto. For example, the first calculation unit 32 may calculate an input value based on the target output value r and the difference input value ub so that the output value y follows the target output value r through feedback control, and then calculate an input value u that cancels out the parameter indicating disturbance based on that input value and the adaptive input value ua.

[0023] (Control of mobile objects) The movement control unit 34 controls the moving body 10 (control target 12) by inputting an input value u to it. When the input value u is input, the moving body 10 is driven according to the value of the input value u (control input) and enters a predetermined state (indicating a predetermined output value y).

[0024] (Retrieve output value) The second acquisition unit 36 ​​(output value acquisition unit) acquires the output value y of the mobile body 10. That is, the second acquisition unit 36 ​​acquires the output value y from the mobile body 10, which has entered a predetermined state due to the input value u. In this embodiment, the second acquisition unit 36 ​​controls the output detection unit 14 to cause the output detection unit 14 to detect the output value y of the mobile body 10, and acquires the detection result as the output value y. The output value y can be said to be the actual control output of the mobile body 10.

[0025] The control device 16 inputs the input value u to the mobile body 10 to control it, and at the same time inputs the input value u to the output prediction model A2, which is a mathematical model (algorithm) that simulates the behavior of the mobile body 10. The control device 16 then calculates the amount of deviation (adaptive parameter) between the actual behavior of the mobile body 10 and the ideal behavior from the error between the output value from the output prediction model A2 and the actual output value y, and calculates an input value u that cancels out this deviation. This makes it possible to set the input value u to cancel out the effects of external disturbances, for example, and suppresses a decrease in the accuracy of the input value u calculation. The calculation process of the adaptive parameter will be explained in detail below.

[0026] (Calculation of estimated output value) The second calculation unit 38 (estimated output value calculation unit) inputs the input value u into the output prediction model A2 and calculates the estimated output value ya, which is the estimated output value of the moving body 10 when the input value u is input. That is, the second calculation unit 38 substitutes the input value u into the mathematical model shown in the output prediction model A2 and calculates the estimated output value ya, which is the output value in the output prediction model A2. In this embodiment, the second calculation unit 38 calculates the estimated output value ya using the following equation (1) as the output prediction model A2.

[0027]

number

[0028] In equation (1), t is time, Am, Bm, Cm, and α are predetermined coefficients, and x v and x gω, θ, and σ are adaptive parameters, and u z α is the input value u. The adaptive parameter ω is the so-called input gain, a quantity that represents the change in the degree to which the input value u affects the behavior of the moving body 10, for example, the change in the magnitude of the moment (rudder effect) generated when the rudder angle is changed. The adaptive parameter θ is a parameter that represents the change in the dynamics of the moving body, and corresponds to state-dependent changes. The adaptive parameter σ is a parameter that represents the change in the dynamics of the moving body, and corresponds to changes independent of the state. The second calculation unit 38 uses the adaptive parameters calculated by the third calculation unit 42 immediately before as the adaptive parameters in equation A2. α is one of the search parameters described later. Furthermore, α is the stabilization term coefficient (a coefficient multiplied by the feedback difference value) when calculating the estimated output value ya, and can be said to be a coefficient for adjusting the trade-off between adaptive speed and responsiveness (vibration).

[0029] (Calculation of adaptive parameters) The third calculation unit 42 (adaptive parameter calculation unit) calculates the error e between the output value y from the mobile body 10 and the estimated output value ya from the output prediction model A2. y Based on the search parameters obtained by the third acquisition unit 40 described later, the adaptive parameters ω, θ, and σ are calculated. y This refers to the difference between the output value y and the estimated output value ya. The search parameter is the error e. y This parameter shows the correspondence between the adaptive parameter and the error e. y These are coefficients in the mathematical model that calculates adaptive parameters based on [the given formula]. The search parameters are pre-set fixed values ​​and do not change in real time. The method for setting the search parameters will be described later.

[0030] In this embodiment, the third calculation unit 42 calculates the adaptive parameter ω, adaptive parameter θ, and adaptive parameter σ using the following equations (2) to (4). However, the adaptive parameters are not limited to being calculated using equations (2) to (4), and the error ey It may be calculated by any method using the search parameter and the adaptation parameter.

[0031]

Number

Number

Number

[0032] In equations (2) to (4), Γ ω , Γ θ , Γ σ are search parameters and can be said to be parameters related to the speed of adjustment by the adaptation parameter. Also, Proj(θ, y) is a projection operator and has the value shown in the following equation (5). The same applies to Proj(ω, y) and Proj(σ, y).

[0033]

Number

[0034] The adaptation parameter calculated by the third calculation unit 42 is used as the adaptation parameter of the output prediction model A2 when calculating the estimated output value ya next time. That is, the second calculation unit 38 updates the output prediction model A2 using the adaptation parameter calculated immediately before by the third calculation unit 42. As a result, it becomes possible to calculate the estimated output value ya using the latest adaptation parameter, and the input value u can be calculated while appropriately reflecting the disturbance.

[0035] Also, the adaptation parameter calculated by the third calculation unit 42 is used as the adaptation parameter when calculating the input value u next time. That is, the first calculation unit 32 calculates the input value u based on the target output value r and the adaptation parameter calculated immediately before. This will be specifically described below.

[0036] Here, the output values ​​when the mobile body 10 exhibits ideal behavior can be treated as those shown in the following equations (6) and (7).

[0037]

number

number

[0038] In contrast, the actual behavior of the moving body 10 may differ from the ideal behavior due to environmental factors such as disturbances, as described above. The output value, which also takes into account the deviation from the ideal behavior (adaptive parameter), is expressed by equation (8). As shown in equation (9), if we let θ(t)||x(t)||+σ(t) in equation (8) be f(x, t), then this f(x, t) becomes a parameter indicating the disturbance. Therefore, the first calculation unit 32 calculates u(t) as the input value u (adaptive input value ua in this embodiment) that cancels out f(x, t) in equation (8). This makes it possible to set an input value u that cancels out the effect of disturbances, and even if disturbances are present, it becomes possible to appropriately track the target output value r.

[0039]

number

number

[0040] The control device 16 controls the movement of the mobile body 10 by executing the processes described above at predetermined intervals.

[0041] (Control flow) The control flow of the mobile body 10 by the control device 16 described above will now be explained. Figure 4 is a flowchart illustrating the control flow of the mobile body according to this embodiment. As shown in Figure 4, the control device 16 acquires the target output value r, the adaptive parameter calculated immediately before, and the feedback difference value calculated immediately before (step S10), and calculates the input value u based on these (step S12). Then, the control device 16 inputs the input value u to the control target 12 to control the mobile body 10 (step S14) and acquires the output value y, which is the control output from the mobile body 10 (step S16). The control device 16 also inputs the input value u to the output prediction model A2 to calculate the estimated output value ya of the mobile body 10 (step S18). After acquiring the output value y and calculating the estimated output value ya, the control device 16 calculates the error e between the output value y and the estimated output value ya. y Based on the search parameters pre-set by the third acquisition unit 40, adaptive parameters are calculated (step S20). If the process is to be terminated (step S22; Yes), this process is terminated. If the process is not to be terminated (step S22; No), the output prediction model A2 is updated with the calculated adaptive parameters, and the process returns to step S10 to continue.

[0042] (Setting the search parameters) Next, the method for setting the search parameters by the third acquisition unit 40 will be explained. As described above, the search parameters are coefficients used to calculate the adaptive parameters. While the adaptive parameters are calculated sequentially during the control of the mobile body 10, the search parameters are fixed values ​​that are set in advance.

[0043] Figure 5 is a schematic diagram illustrating the setting of the search parameters. The third acquisition unit 40 sets the search parameters using the detailed prediction model B1. The detailed prediction model B1 is a simulator (algorithm) that, upon receiving input values ​​based on the target output value, the search parameters, and variable parameters which are variable parameters related to the movement of the mobile body 10, performs calculations to simulate the behavior of the mobile body 10 and outputs an estimated output value of the mobile body 10. The input values ​​based on the target output value refer to input values ​​that can achieve the set target output value and may be calculated by any method. Variable parameters are parameters that change due to unknown environmental changes when the mobile body 10 moves, such as buoyancy and fluid coefficients. The detailed prediction model B1 calculates an estimated output value of the mobile body 10 in the same way as the output prediction model A2, but since it uses variable parameters that were not used in the output prediction model A2 as input values, it can calculate the estimated output value with higher accuracy than the output prediction model A2. The output prediction model A2 may have any configuration that can calculate an estimated output value based on the input values ​​to the mobile body 10, the search parameters, and the variation parameters.

[0044] The third acquisition unit 40 selects input values ​​based on the target output value, the values ​​of the search parameters, and the values ​​of the variation parameters, and inputs the selected input values, search parameters, and variation parameters into the detailed prediction model B1 to obtain estimated output values ​​as calculation results. The third acquisition unit 40 obtains multiple estimated output values ​​by changing at least one of the input values, search parameters, and variation parameters and performing this process multiple times. Based on each estimated output value output from the detailed prediction model B1, the third acquisition unit 40 sets search parameters for calculating adaptive parameters. In other words, the third acquisition unit 40 sets preferred search parameters based on multiple calculation results from the detailed prediction model B1.

[0045] In this embodiment, the third acquisition unit 40 sets the search parameters using robust processing, which searches for search parameters that reduce the degree of deviation between the estimated output value and the target output value, even when the variable parameters change. The robust processing will be described below.

[0046] (Robust processing) In robust processing, the third acquisition unit 40 sets the search parameters using a surrogate optimization method. Specifically, the third acquisition unit 40 inputs the selected input values, search parameters, and variation parameters into the detailed prediction model B1 and obtains the estimated output value as a calculation result. Then, based on the estimated output value, the third acquisition unit 40 calculates the degree of deviation of the estimated output value from the target output value on which the input values ​​were based. Here, the degree of deviation refers to the degree of deviation of the estimated output value from the target output value, and can be said to be an evaluation value of how bad the result (risk) will be depending on the variation parameters and search parameters. For example, when controlling the depth of an underwater vehicle, the time it takes for the depth to fall within a predetermined range of the target depth may be used as the degree of deviation.

[0047] The third acquisition unit 40 calculates a correspondence between the degree of deviation, the search parameters, and the variable parameters, based on the calculated deviation and the search parameters and variable parameters used in the calculation. The third acquisition unit 40 selects the input values, search parameters, and variable parameters to be used in the next calculation from this correspondence, inputs the selected input values, search parameters, and variable parameters into the detailed prediction model B1, and obtains the estimated output value as the calculation result. The method of selecting the input values, search parameters, and variable parameters to be used in the next calculation based on the correspondence may be arbitrary, but for example, values ​​in which the search parameters and variable parameters have deviated by a predetermined value or more from the previous values ​​may be selected. The third acquisition unit 40 also calculates the degree of deviation for the newly calculated estimated output value and updates the correspondence using the newly calculated degree of deviation and the search parameters and variable parameters used in the new calculation. The third acquisition unit 40 can learn the correspondence by repeating this process for a predetermined number of times and continuously updating the correspondence, thereby constructing a correspondence that accurately reflects the correspondence between the degree of deviation, the search parameters, and the variable parameters.

[0048] Figure 5 shows an example of a calculated correspondence. In Figure 5, for each pair P of search parameters and variable parameters, the detailed prediction model B1 calculates an estimated output value, and the degree of deviation is calculated. Based on these degree of deviations for each pair P, a correspondence SM between the degree of deviation, the search parameters, and the variable parameters is set. In other words, the third acquisition unit 40 calculates the correspondence SM as a function with the degree of deviation as the solution, using the search parameters and variable parameters as variables, based on the calculation results for multiple pairs P.

[0049] The third acquisition unit 40 sets the search parameters used to calculate the adaptive parameters based on this correspondence. Specifically, based on the correspondence, the third acquisition unit 40 calculates the degree of deviation when the variable parameter is changed while the search parameter is fixed, as shown at point Q in Figure 5. Based on these calculated degrees of deviation, the third acquisition unit 40 sets the search parameters for calculating the adaptive parameters. For example, the third acquisition unit 40 may set the search parameters for calculating the adaptive parameters such that the maximum value (worst-case value with the largest degree of deviation) among the degrees of deviation when the variable parameter is changed while the search parameter is fixed is less than or equal to a predetermined value. Alternatively, the third acquisition unit 40 may set the search parameters for calculating the adaptive parameters such that the average value of the degrees of deviation when the variable parameter is changed while the search parameter is fixed is less than or equal to a predetermined value. By setting the search parameters in this way, it becomes possible to select search parameters that reduce the degree of deviation of the estimated output value from the target output value even when the variable parameter changes, and the input value u can be calculated with high accuracy, for example, even in the case of an unknown disturbance.

[0050] The method for setting the search parameters is not limited to using the robust processing described above. For example, the calculation of the estimated output values ​​by the detailed prediction model B1 may be performed multiple times, and the search parameters may be set using any method based on those estimated output values. Alternatively, for example, the search parameters may be set using a metaheuristic method.

[0051] The robust processing flow described above will now be explained. Figure 6 is a flowchart illustrating the robust processing flow. As shown in Figure 6, the control device 16 selects the values ​​of the search parameters, the values ​​of the variable parameters, and the input values ​​based on the target output value (step S30), inputs the selected search parameters, variable parameters, and input values ​​into the detailed prediction model B1, and calculates the estimated output value (step S32). Based on the calculated estimated output value, the control device 16 calculates the degree of deviation of the estimated output value from the target output value (step S34), and calculates the correspondence between the search parameters, variable parameters, and the degree of deviation (step S36). Then, it determines whether the number of calculated estimated output values ​​(number of calculations) is greater than or equal to a predetermined value (step S38). If it is less than the predetermined value (step S38; No), it returns to step S30, selects the search parameters, variable parameters, and input values ​​to be used in the next calculation based on the calculated correspondence, and continues processing to learn the correspondence. Note that the search parameters, variable parameters, and input values ​​used in the first calculation may be selected arbitrarily. On the other hand, if the number of calculated estimated output values ​​is greater than or equal to a predetermined value (Step S38; Yes), the correspondence calculation process is terminated, and search parameters for calculating adaptive parameters are set based on the calculated correspondence (Step S39).

[0052] In this embodiment, the control device 16 set the search parameters, but the entity that sets the search parameters is not limited to the control device 16. For example, a device other than the control device 16 may set the search parameters, and the control device 16 may acquire the search parameters set by that device.

[0053] (Output value when using this control) Next, we will describe the simulation results for the output values ​​when using the control according to this embodiment described above. Figure 7 is a graph showing an example of the simulation results for the output values. In this simulation, we simulated the case of controlling the depth of a mobile body 10, which is an underwater navigation body, and used the depth of the mobile body 10 as the output value. In Figure 7, the horizontal axis is time, and the vertical axis is depth (output value). In this simulation, we simulated the fluctuation of the output value (depth) when the payload (weight) fluctuates while the target output (target depth) is set to D1. In Figure 7, line L1 shows the fluctuation of depth when the input value u is calculated using the target output value r and the feedback difference value (difference input value ub) without using adaptive parameters. Lines L2 and L3 show the fluctuation of depth when the input value u is calculated using the target output value r and adaptive parameters, as in this embodiment. Note that line L2 shows the result when the search parameters used to calculate the adaptive parameters are arbitrarily set without using robust processing, while line L3 shows an example when the search parameters used to calculate the adaptive parameters are set using robust processing.

[0054] As shown by lines L1, L2, and L3, it can be seen that by using adaptive parameters, the time it takes to fall within the predetermined range of the target depth is shortened, and the input values ​​can be calculated with high accuracy. Furthermore, as shown by lines L2 and L3, it can be seen that by using robust processing, the time it takes to fall within the predetermined range of the target depth is further shortened, and the input values ​​can be calculated with even higher accuracy.

[0055] (Decision on whether to use robust processing) The third acquisition unit 40 may decide whether or not to set the search parameters using robust processing based on the type of expected environmental change (environmental change to be adapted) of the mobile body 10. The third acquisition unit 40 decides to set the search parameters using robust processing if the expected environmental change satisfies predetermined conditions, and decides not to use robust processing if the expected environmental change does not satisfy predetermined conditions. The predetermined conditions can be set arbitrarily; for example, the user may input information on whether or not the predetermined conditions are met, and the third acquisition unit 40 may acquire the information input by the user on whether or not the predetermined conditions are met. If the information satisfies the predetermined conditions, the third acquisition unit 40 decides to set the search parameters using robust processing, assuming that the expected environmental change satisfies the predetermined conditions. On the other hand, if the information does not satisfy the predetermined conditions, the third acquisition unit 40 decides to set the search parameters without using robust processing, assuming that the expected environmental change does not satisfy the predetermined conditions.

[0056] The predetermined conditions can be set arbitrarily, but they may be considered satisfied when it is assumed that the degree of change in the environment cannot be predicted in advance. For example, the predetermined conditions may be considered satisfied when it is assumed that the rudder effect of the underwater navigating body (the degree of change in depth in response to the control input that changes the direction of the moving body 10) will fluctuate. The rudder effect fluctuates when the fluid coefficient changes due to, for example, rain, and the degree of such fluctuation is difficult to predict. By using robust processing, it is possible to set search parameters that are robust to environmental changes, making it particularly effective when the degree of change in the environment cannot be predicted in advance.

[0057] On the other hand, if, for example, the payload (weight) of an underwater vehicle is expected to fluctuate, it may be determined that the predetermined conditions are not met. Since the payload is expected to be predictable in advance, the degree of fluctuation is relatively easy to predict. When the degree of fluctuation is easily predictable, search parameters that are robust to environmental changes become unnecessary, and the computational load can be reduced by not using robust processing that requires computational load for setting search parameters.

[0058] (Determining the search parameters to set) The third acquisition unit 40 may determine the types of search parameters to set based on the types of expected environmental changes of the moving body 10. That is, the third acquisition unit 40 may decide which search parameters to automatically set according to the types of expected environmental changes. For example, the third acquisition unit 40 may set fewer types of search parameters when the expected environmental changes satisfy predetermined conditions than it does when the expected environmental changes do not satisfy predetermined conditions. For example, if predetermined conditions are met (for example, if it is indicated that the rudder effect of the underwater vehicle is expected to fluctuate), Γ ω , Γ θ , Γ σ , and α may be automatically set as search parameters. Also, for example, if the predetermined conditions are not met (for example, if it is indicated that the payload of the underwater vehicle is expected to fluctuate), Γ ω , Γ θ , Γ σ , α, ω max , θ max , and σ max This can be used as an automatically set search parameter. ω max , θ max , and σ max This represents the upper limit of the adaptive parameters. If the predetermined conditions are not met and robust processing is not used, the computational load can be reduced, thus preventing the computational load from becoming excessive even if the number of automatically set search parameters is increased.

[0059] When performing this process, search parameters other than those determined to be automatically set may be set by the user, or pre-set values ​​may be used.

[0060] An example of the processing flow of the decision process described above will now be explained. Figure 8 is a flowchart illustrating the processing flow of the decision process. As shown in Figure 8, the control device 16 determines the search parameters to be automatically set based on the type of expected environmental change (step S40), and determines the method for setting the search parameters based on the type of expected environmental change (step S42). That is, the control device 16 decides whether to set the search parameters by robust processing or by a method other than robust processing. The method other than robust processing may be arbitrary, but for example, it may be a method in which the calculation processing of estimated output values ​​by the detailed prediction model B1 is performed multiple times, and the search parameters are set by an arbitrary method from those estimated output values, and a method with a lower computational load than robust processing is preferred. Once the control device 16 has determined the search parameters to be automatically set and the method for setting them, it sets the values ​​of the search parameters that have been set to be automatically set using the determined setting method (step S44).

[0061] Furthermore, the order in which the types of search parameters to be automatically set are determined and the method of setting them is not limited to that shown in Figure 8 and may be arbitrary. Also, the processes of determining the types of search parameters to be automatically set and determining the method of setting the search parameters are not mandatory; these processes may be omitted, or at least one of these processes may be performed.

[0062] (effect) A control method for a mobile body 10 according to a first aspect of this disclosure includes the steps of: obtaining a target output value r of the mobile body 10; calculating an input value u to be input to the mobile body 10 based on the target output value r; controlling the mobile body 10 by inputting the input value u to the mobile body 10; obtaining the output value y of the mobile body 10 when the input value u is input to the mobile body 10; inputting the input value u to an output prediction model A2 that simulates the behavior of the mobile body 10 to calculate an estimated output value ya of the mobile body 10 when the input value u is input; and the error e between the output value y and the estimated output value ya. yThe steps include obtaining search parameters, which are coefficients that show the correspondence with adaptive parameters that indicate the behavior of the moving object 10, and the error e between the output value y and the estimated output value ya. y The process includes the steps of calculating adaptive parameters based on the search parameters and the input value u, where the input value u is calculated based on the target output value r and the adaptive parameters.

[0063] According to this disclosure, error e y Adaptive parameters are calculated using the search parameters, and the input value u is calculated based on the adaptive parameters. According to this disclosure, by sequentially calculating adaptive parameters based on the search parameters, it becomes possible to calculate an input value u that can achieve the target output value r with high accuracy, even when there are unknown disturbances.

[0064] A control method for a mobile body 10 according to a second aspect of this disclosure is a control method for a mobile body 10 according to a first aspect, wherein in the step of acquiring search parameters, the input value u to the mobile body 10, the search parameters, and variable parameters which are variable parameters related to the movement of the mobile body 10 are input to a detailed prediction model B1 that simulates the behavior of the mobile body 10, and the process of acquiring an estimated output value output from the detailed prediction model B1 is performed multiple times while changing at least one of the search parameters and variable parameters. Then, based on each estimated output value output from the detailed prediction model B1, search parameters for calculating adaptive parameters are set. According to this disclosure, since search parameters are set based on multiple calculation results from the detailed prediction model B1, the accuracy of calculating adaptive parameters based on search parameters is improved, and it becomes possible to calculate input values ​​u that can achieve a target output value r with high accuracy.

[0065] A control method for a mobile body 10 according to a third aspect of this disclosure is a control method for a mobile body 10 according to the first or second aspect, wherein in the step of acquiring search parameters, a robust process is performed in which the degree of deviation of the estimated output value from the target output value, the search parameters, and the fluctuation parameters are calculated based on the respective estimated output values ​​output from the detailed prediction model B1, and the search parameters are set based on this correspondence. According to this disclosure, since the search parameters are set based on the degree of deviation and the correspondence between the search parameters and the fluctuation parameters, the accuracy of calculating adaptive parameters can be improved without making the computational load when setting the search parameters excessive.

[0066] A control method for a mobile body 10 according to a fourth aspect of this disclosure is a control method for a mobile body 10 according to a third aspect, wherein in the step of acquiring search parameters, as a robust process, the degree of deviation is calculated when the variable parameters are changed while the search parameters are fixed, based on the correspondence relationship, and search parameters for calculating adaptive parameters are set based on these degree of deviation. According to this disclosure, since the search parameters are set from the degree of deviation when the variable parameters are changed while the search parameters are fixed, it becomes possible to select search parameters that are robust to disturbances (environmental changes), for example, and it becomes possible to calculate the input value u with high accuracy even when there is an unknown disturbance.

[0067] A control method for a mobile body 10 according to a fifth aspect of this disclosure is a control method for a mobile body 10 according to a third or fourth aspect, wherein in the step of acquiring search parameters, it is determined whether or not to set the search parameters using robust processing based on the type of expected environmental change of the mobile body 10. According to this disclosure, since it is determined whether or not to use robust processing according to the expected environmental change, it becomes possible to appropriately calculate the input value u according to the environmental change.

[0068] A control method for a mobile body 10 according to the sixth aspect of this disclosure is a control method for a mobile body 10 according to any of the first to fifth aspects, wherein in the step of acquiring search parameters, the type of search parameter to be set is determined based on the type of expected environmental change of the mobile body 10. According to this disclosure, since the search parameter to be set is determined according to the expected environmental change, it becomes possible to appropriately calculate the input value u according to the environmental change.

[0069] A control method for a mobile body 10 according to the seventh aspect of this disclosure is a control method for a mobile body 10 according to any of the first to sixth aspects, wherein in the step of calculating an estimated output value, the output prediction model A2 is updated using the adaptive parameters calculated immediately beforehand. According to this disclosure, since the output prediction model A2 is updated using the adaptive parameters calculated immediately beforehand, it becomes possible to adjust the output prediction model A2 in accordance with environmental changes, and it becomes possible to appropriately calculate the input value u in accordance with environmental changes.

[0070] The control method for a mobile body 10 according to the eighth aspect of this disclosure is a control method for a mobile body 10 according to any of the first to seventh aspects, wherein in the step of calculating the input value u, the input value u is calculated in such a way that it cancels out the parameter that indicates a disturbance among the adaptive parameters. According to this disclosure, since the input value u is calculated in such a way that it cancels out the parameter that indicates a disturbance, it becomes possible to appropriately calculate the input value u in accordance with environmental changes.

[0071] A control method for a mobile body 10 according to the ninth aspect of this disclosure is a control method for a mobile body 10 according to any of the first to eighth aspects, wherein in the step of calculating the input value u, the input value u is calculated based on the difference (feedback difference value) between the target output value r and the output value y. According to this disclosure, since the input value u is calculated using feedback control and adaptive parameters, it becomes possible to appropriately calculate an input value u that can achieve the target output value r while responding to environmental changes.

[0072] A control method for a mobile body 10 according to the tenth aspect of this disclosure is a control method for a mobile body 10 according to any of the first to ninth aspects, which controls the depth of the underwater navigation body, which is the mobile body 10. According to this disclosure, the depth of the underwater navigation body can be appropriately controlled.

[0073] A control method for a mobile body 10 according to the 11th aspect of this disclosure is a control method for a mobile body 10 according to any of the 1st to 9th aspects, which controls the navigation of a ship that is a mobile body 10. According to this disclosure, the navigation of a ship can be appropriately controlled.

[0074] Although embodiments of the present disclosure have been described above, the embodiments are not limited to those described herein. Furthermore, the aforementioned components include those that can be easily conceived by those skilled in the art, those that are substantially the same, and those that fall within the so-called equivalent range. Moreover, the aforementioned components can be combined as appropriate. Furthermore, various omissions, substitutions, or modifications of the components can be made without departing from the gist of the embodiments described above. [Explanation of Symbols]

[0075] 10 Mobile Units 12. Controlled object 16 Control device 30 First acquisition part 32 First Calculation Unit 34 Movement Control Unit 36 Second acquisition part 38 Second Calculation Section 40 Third acquisition part 42 Third Calculation Unit r Target output value u Input value y output value

Claims

1. The steps include obtaining the target output value of the mobile object, The steps include: calculating the input value to be input to the mobile body based on the target output value; The steps include: inputting the input value to the moving body to control the moving body, The steps include: obtaining an output value which is the actual control output of the moving body when the input value is input to the moving body; The steps include inputting the input value into an output prediction model that simulates the behavior of the moving object, and calculating the estimated output value of the moving object when the input value is input, The steps include obtaining a search parameter, which is a coefficient indicating the correspondence between the error between the output value and the estimated output value and the adaptive parameter representing the behavior of the moving object, The step includes calculating the adaptive parameter based on the error between the output value and the estimated output value and the search parameter, In the step of calculating the input value, the input value is calculated based on the target output value and the adaptive parameter. The adaptive parameter is a parameter that indicates the amount of deviation of the actual behavior of the moving body from its ideal behavior, and is a parameter used in the output prediction model to calculate the estimated output value. In the step of obtaining the aforementioned search parameters, The process of inputting the input values ​​to the moving body, the search parameters, and the variable parameters, which are variable parameters related to the movement of the moving body, into a detailed prediction model that simulates the behavior of the moving body, and obtaining estimated output values ​​output from the detailed prediction model, is performed multiple times while changing at least one of the search parameters and the variable parameters. Based on the respective estimated output values ​​output from the detailed prediction model, the search parameters for calculating the adaptive parameters are set. A method for controlling a moving object.

2. In the step of obtaining the aforementioned search parameters, A method for controlling a moving body according to claim 1, comprising: calculating the degree of deviation of the estimated output value from the target output value based on the respective estimated output values ​​output from the detailed prediction model, the correspondence between the search parameter and the variation parameter, and performing a robust process to set the search parameter based on the correspondence.

3. The method for controlling a moving body according to claim 2, wherein in the step of obtaining the search parameters, as a robust process, the degree of deviation is calculated when the variable parameters are changed while the search parameters are fixed based on the correspondence relationship, and the search parameters for calculating the adaptive parameters are set based on these degrees of deviation.

4. The method for controlling a mobile body according to claim 2 or 3, wherein in the step of acquiring the search parameters, it is determined whether or not to set the search parameters using the robust processing based on the type of expected environmental change of the mobile body.

5. The method for controlling a mobile body according to any one of claims 1 to 3, wherein in the step of acquiring the search parameters, the type of search parameters to be set is determined based on the type of expected environmental change of the mobile body.

6. A method for controlling a moving body according to any one of claims 1 to 3, wherein in the step of calculating the estimated output value, the output prediction model is updated using the adaptive parameter calculated immediately beforehand.

7. The method for controlling a moving body according to any one of claims 1 to 3, wherein in the step of calculating the input value, the input value is calculated so as to cancel out the parameter that indicates a disturbance among the adaptive parameters.

8. The method for controlling a moving body according to any one of claims 1 to 3, wherein in the step of calculating the input value, the input value is calculated based on the difference between the target output value and the output value.

9. A method for controlling a moving body, which is an underwater navigation body, according to any one of claims 1 to 3, for controlling the depth of the moving body.

10. A method for controlling a moving body, according to any one of claims 1 to 3, for controlling the navigation of the aforementioned moving body, a vessel.

11. A first acquisition unit that acquires the target output value of the mobile object, A first calculation unit calculates an input value to be input to the mobile body based on the target output value, A movement control unit that inputs the input value to the moving body and controls the moving body, A second acquisition unit acquires an output value which is the actual control output of the moving body when the input value is input to the moving body, A second calculation unit inputs the input value into an output prediction model that simulates the behavior of the moving object, and calculates the estimated output value of the moving object when the input value is input. A third acquisition unit acquires a search parameter which is a coefficient indicating the correspondence between the error between the output value and the estimated output value and the adaptive parameter that indicates the behavior of the moving object. Includes a third calculation unit that calculates the adaptive parameter based on the error between the output value and the estimated output value and the search parameter, The first calculation unit calculates the input value based on the target output value and the adaptive parameter, The adaptive parameter is a parameter that indicates the amount of deviation of the actual behavior of the moving body from its ideal behavior, and is a parameter used in the output prediction model to calculate the estimated output value. The aforementioned acquisition unit is: The process of inputting the input values ​​to the moving body, the search parameters, and the variable parameters, which are variable parameters related to the movement of the moving body, into a detailed prediction model that simulates the behavior of the moving body, and obtaining estimated output values ​​output from the detailed prediction model, is performed multiple times while changing at least one of the search parameters and the variable parameters. Based on the respective estimated output values ​​output from the detailed prediction model, the search parameters for calculating the adaptive parameters are set. A control device for mobile vehicles.

12. A mobile body having the control device described in claim 11.

13. The steps include obtaining the target output value of the mobile object, The steps include: calculating the input value to be input to the mobile body based on the target output value; The steps include: inputting the input value to the moving body to control the moving body, The steps include: obtaining an output value which is the actual control output of the moving body when the input value is input to the moving body; The steps include inputting the input value into an output prediction model that simulates the behavior of the moving object, and calculating the estimated output value of the moving object when the input value is input, The steps include obtaining a search parameter, which is a coefficient indicating the correspondence between the error between the output value and the estimated output value and the adaptive parameter representing the behavior of the moving object, A program that causes a computer to perform the steps of calculating the adaptive parameter based on the error between the output value and the estimated output value and the search parameter, In the step of calculating the input value, the input value is calculated based on the target output value and the adaptive parameter. The adaptive parameter is a parameter that indicates the amount of deviation of the actual behavior of the moving body from its ideal behavior, and is a parameter used in the output prediction model to calculate the estimated output value. In the step of obtaining the aforementioned search parameters, The process of inputting the input values ​​to the moving body, the search parameters, and the variable parameters, which are variable parameters related to the movement of the moving body, into a detailed prediction model that simulates the behavior of the moving body, and obtaining estimated output values ​​output from the detailed prediction model, is performed multiple times while changing at least one of the search parameters and the variable parameters. Based on the respective estimated output values ​​output from the detailed prediction model, the search parameters for calculating the adaptive parameters are set. program.

Images