Manipulated variable compensation system

Abstract

The present invention provides a control quantity correction system that can compensate for fluctuations in the amount of control related to driving operations caused by road surface conditions. [Solution] The control input correction system S comprises a road surface condition detection unit 34 for detecting road surface conditions, a first trained model 35, and a correction unit 36. The first trained model 35 is pre-trained using training data that takes control inputs related to driving operations and road surface conditions as input data and control inputs related to driving operations as output data. When control inputs related to driving operations and road surface conditions are input, the first trained model 35 outputs a correction value for the control inputs related to driving operations. The correction unit 36 ​​corrects the control inputs related to driving operations based on the correction value output by the first trained model 35.

Description

Technical Field

[0001] The present invention relates to a system for correcting an operation amount related to a driving operation of a vehicle.

Background Art

[0002] As shown in FIG. 6A, when a forklift travels on a so-called rough road with unevenness on the ground, there is a risk of load collapse or tipping over. Therefore, for example, as disclosed in Patent Document 1, there is a travel control device that suppresses vibration of a load during travel on a rough road.

[0003] This control device detects the load of the load loaded on the fork by a pressure sensor. Next, the control device calculates the frequency wθ of the load fluctuation of the load by the CPU from the detection signal of the pressure sensor, and detects a rough road based on the calculated frequency. Then, when the control device detects a rough road, the driving motor is driven so as to reach a predetermined low speed regardless of the depression amount of the accelerator pedal. Thereby, this control device prevents load collapse during travel on a rough road.

[0004] Further, as shown in FIG. 6B, when the forklift travels on an inclined road surface, even if the steering wheel is not operated (even if the operation amount of the steering wheel is intended to be 0), it may slide to the inclined side. In this case, in order to travel straight, it is necessary to travel while finely adjusting the steering wheel.

[0005] Moreover, what is a concern during travel on a rough road having unevenness is not only prevention of load collapse. For example, when traveling on the unevenness of the road surface, the strength of stepping on the accelerator pedal may vary due to the vibration caused by the unevenness. In this case, depending on the specifications of the forklift, the constant speed operation may be canceled. Also, when the strength of stepping on the brake pedal or the steering angle changes due to the vibration caused by the unevenness, it may lead to load collapse or an accident.

Prior Art Documents

Patent Documents

[0006] [Patent Document 1] Japanese Patent Application Publication No. 11-217200 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] Therefore, the problem that the present invention aims to solve is to provide a system that can correct the amount of operation related to driving operations according to the condition of the road surface. [Means for solving the problem]

[0008] To solve the above problems, the manipulated variable correction system according to the present invention is A road surface condition detection unit that detects the road surface condition, A first pre-trained model takes the input data as the amount of control related to driving operations and the road surface condition, and the output data as the amount of control related to driving operations when passing over the road surface. When the amount of control related to driving operations and the road surface condition are input, the first pre-trained model outputs a correction value for the amount of control related to driving operations. The system includes a correction unit that corrects the control inputs related to driving operations based on the correction values ​​output by the first trained model.

[0009] The above-mentioned manipulated variable correction system is preferably, Road surface conditions include unevenness and / or slope of the road surface.

[0010] The above-mentioned manipulated variable correction system is preferably, The amount of input related to driving operations includes the amount of input from the accelerator pedal.

[0011] The above-mentioned manipulated variable correction system is preferably, The amount of input required for driving includes the amount of input required for braking.

[0012] The above-mentioned manipulated variable correction system is preferably, The amount of input required for driving includes the amount of steering wheel input.

[0013] The above-mentioned manipulated variable correction system is preferably, The input data for the first trained model's training data further includes information about the direction of movement, whether it is moving forward or backward. The first trained model then receives information about the direction of travel and outputs correction values ​​for the control variables related to the driving operation based on the information about the direction of travel.

[0014] The above-mentioned manipulated variable correction system is preferably, It further includes a weight detection unit for detecting the weight of the load, The input data for the first trained model's training data further includes the weight of the load. The first trained model then takes the load weight as input and outputs correction values ​​for the control variables related to the driving operation based on the load weight.

[0015] The above-mentioned manipulated variable correction system is preferably, It further includes a fork height detection unit that detects the height of the forks, The input data for the first trained model's training data further includes the fork height, and the first trained model, upon receiving the fork height as input, outputs a correction value for the control variable related to the driving operation based on the fork height.

[0016] The above-mentioned manipulated variable correction system is preferably, The input data for the first trained model's training data also includes the vehicle speed and the wheel position. The first trained model then takes the vehicle speed and wheel position as input and outputs correction values ​​for the control variables related to the vehicle operation based on the vehicle speed and wheel position.

[0017] The above-mentioned manipulated variable correction system is preferably, The input data for the first trained model's training data also includes the weight of the attachment. The first learned model further receives the weight of the attachment as input and outputs a correction value for the operation amount related to the traveling operation based on the weight of the attachment.

[0018] The operation amount correction system preferably further includes a road surface camera that captures the road surface in the traveling direction to generate a road surface image, The road surface state detection unit is pre-trained with teacher data that uses the road surface image as input data and outputs the road surface state as output data, and has a second learned model that outputs the road surface state when a road surface image is input.

[0019] The operation amount correction system preferably further includes a load camera that captures the load to be loaded to generate a load image, a third learned model that is pre-trained with teacher data that uses the load image as input data and outputs the center of gravity position of the load as output data, and outputs the center of gravity position of the load when a load image is input, The input data of the teacher data of the first learned model further includes the center of gravity position of the load, The first learned model further receives the center of gravity position of the load as input and outputs a correction value for the operation amount based on the center of gravity position of the load.

[0020] To solve the above problems, a forklift according to the present invention includes the operation amount correction system.

[0021] To solve the above problems, an operation amount correction program according to the present invention is an operation amount correction program used for a vehicle having a computer, The computer is caused to a road surface state detection unit that detects the road surface state, and a first learned model that is pre-trained with teacher data that uses the operation amount related to the traveling operation and the road surface state as input data and outputs the operation amount related to the traveling operation when passing through the road surface as output data, and outputs a correction value for the operation amount related to the traveling operation when the operation amount related to the traveling operation and the road surface state are input, A correction unit is used to correct the control variables related to driving operations based on the correction values ​​output by the first trained model. [Effects of the Invention]

[0022] The control input correction system according to the present invention can correct the control input related to driving operations according to the road surface conditions. [Brief explanation of the drawing]

[0023] [Figure 1] This is a schematic side view of a forklift with an operation amount correction system according to one embodiment of the present invention. [Figure 2] Figure 1 is a schematic perspective view showing the movement of a forklift, where A represents an uneven road surface and B represents an inclined road surface. [Figure 3] This is a functional block diagram of the manipulated variable correction system. [Figure 4] This figure shows the operation of the first trained model and the road surface trained model. [Figure 5] This is a flowchart illustrating the operation flow of the manipulated variable correction system. [Figure 6] This is a schematic perspective view illustrating the driving motion of a conventional forklift, where A represents an uneven road surface and B represents an inclined road surface. [Modes for carrying out the invention]

[0024] Hereinafter, an embodiment of the operation amount correction system of the present invention will be described with reference to the attached figures. The vehicle in this embodiment is a counterbalanced forklift, but this is merely an example, and the vehicle according to the present invention is not limited to this. For example, the vehicle according to the present invention may be a general vehicle, an industrial vehicle, or another type of forklift.

[0025] <Forklift Configuration> First, the configuration of the forklift 1 will be described. Figure 1 is a schematic side view of a forklift 1 equipped with an operation amount correction system S according to one embodiment of the present invention. Figure 2 is a schematic perspective view showing the forklift 1 shown in Figure 1 in motion, where Figure 2A shows an uneven road surface RS and Figure 2B shows an inclined road surface RS. As shown in Figure 1, the forklift 1 includes front wheels 10, rear wheels 11, a body 13, a driver's seat 14, an operating unit 16, a head guard 17, a road surface camera 19, a mast 20, left and right forks 21, and a lifting device (not shown).

[0026] The vehicle has one front wheel 10 and one rear wheel 11 on each side, and the vehicle body 13 is positioned above the front wheel 10 and the rear wheel 11. The size of the front wheel 10 and the size of the rear wheel 11 are different. In this embodiment, the front wheel 10 is the drive wheel and the rear wheel 11 is the driven wheel.

[0027] The driver's seat 14 is located on the vehicle body 13, and the control unit 16 is located in the driver's seat 14. The head guard 17 is located above the driver's seat 14. In this embodiment, the road surface camera 19 has cameras provided one at the front and one at the rear on the head guard 17. The road surface camera 19 is configured to capture images of the road surface RS in front and the road surface RS behind to generate a road surface image RP. The number of cameras in the road surface camera 19 is not particularly limited.

[0028] The mast 20 is located at the front of the vehicle body 13, and the forks 21 are configured to lift and lower along the mast 20 by a lifting device that scoops up the load W. The "height of the forks 21," which will be explained later, refers to the height of the forks 21 as they are lifted by the lifting device (i.e., the height of the load W).

[0029] Figure 3 is a functional block diagram of the control input correction system S. As shown in Figure 3, the control unit 16 includes a steering wheel 160, an accelerator control unit 161, a brake control unit 162, and a forward / reverse switching unit 163.

[0030] The handle 160 is configured to control the turning angle of the forklift 1 by the amount of its operation. In this embodiment, the handle 160 is configured as a steering wheel, but is not limited to this.

[0031] The accelerator control unit 161 is configured to control the travel speed of the forklift 1 by the amount of its operation. In this embodiment, the accelerator control unit 161 is configured as an accelerator pedal, but is not limited to this.

[0032] The brake operating unit 162 is configured to control the braking force of the forklift 1's movement by the amount of its operation. In this embodiment, the brake operating unit 162 is composed of a brake pedal 25, but is not limited to this.

[0033] The forward / reverse switching unit 163 is configured to switch the direction of travel of the forklift 1 forward or backward. In this embodiment, the forward / reverse switching unit 163 is configured as a lever, but is not limited to this.

[0034] The forklift 1 further includes a travel device 23, a brake 25, a travel speed sensor 26, and a control unit 30.

[0035] The traveling device 23 includes a motor or engine for driving the wheels, a swivel device for rotating the wheels, and the like. The traveling device 23 is configured to rotate and accelerate / decelerate the forklift 1 according to commands from the power control unit 37, which will be described later. The traveling device 23 also adjusts the swivel angle and travel speed according to commands from the power control unit 37.

[0036] The brake 25 is configured to decelerate the forklift 1 in accordance with the commands of the power control unit 37. The brake 25 also adjusts the braking force in accordance with the commands of the power control unit 37.

[0037] The travel speed sensor 26 detects the travel speed of the forklift 1. The travel speed sensor 26 may, for example, estimate the travel speed by detecting the rotational speed of the drive wheels (front wheels 10), and its configuration is not particularly limited.

[0038] The control unit 30 is composed of a computer and includes a storage device, an arithmetic unit, and memory. The storage device stores a manipulated variable correction program that causes the computer to operate as a road surface condition detection unit 34, a first learned model 35, a road surface learned model 340, and a correction unit 36, which will be described later.

[0039] <Functional Configuration of the Manipulated Variable Correction System> Next, the functional configuration of the control amount correction system S will be described. The control unit 30 includes a storage unit 31, a weight detection unit 32, a fork height detection unit 33, a road surface condition detection unit 34, a first learned model 35, a correction unit 36, and a power control unit 37.

[0040] The memory unit 31 stores the weight of the forklift 1 and the weight of any attachments fitted to the forklift 1. The attachments are not particularly limited and may include, for example, a roll clamp or a load stabilizer. If the forklift 1 does not have any attachments fitted, the memory unit 31 stores the weight of the attachments as 0 kg. The memory unit 31 also stores the positions of the front wheels 10 and rear wheels 11 of the forklift 1.

[0041] The weight detection unit 32 is configured to detect the weight of the load W. For example, the forklift 1 may further include sensors for detecting the weight of the load W, and the weight detection unit 32 may detect the weight of the load W by reading the measured values ​​of these sensors. The method by which the weight detection unit 32 detects the weight of the load W is not particularly limited.

[0042] The fork height detection unit 33 is configured to detect the height of the forks 21 (the position of the load W being loaded). For example, the forklift 1 may further include sensors for detecting the height of the forks 21, and the fork height detection unit 33 may detect the height of the forks 21 by reading the measured values ​​of these sensors. The method by which the fork height detection unit 33 detects the height of the forks 21 is not particularly limited.

[0043] The road surface condition detection unit 34 detects the road surface condition RC based on the road surface image RP. The road surface condition detection unit 34 detects the road surface condition RC in front of or behind the forklift 1 in response to the operation of the forward / reverse switching unit 163.

[0044] The road surface condition RC includes at least the degree of unevenness in the area where the wheels are expected to pass. The road surface condition detection unit 34 may indicate the degree of unevenness, for example, with a score. In this case, the score may be 0 if the unevenness is within a predetermined range, a positive value corresponding to the degree of the convexity if there is a convexity beyond the predetermined range, and a negative value corresponding to the degree of the concaveness if there is a concaveity beyond the predetermined range.

[0045] The road surface condition detection unit 34 may also detect the location of irregularities. The location of irregularities on the road surface RS may include, for example, the distance CD between the irregularity and the wheel, as shown in Figure 1. This allows the maneuvering amount correction system S to determine the timing for correcting the maneuvering amount, which will be explained later, based on the distance CD.

[0046] The method by which the road surface condition detection unit 34 detects the road surface condition RC is not particularly limited. In this embodiment, the road surface condition detection unit 34 has a road surface learned model 340. The road surface learned model 340 corresponds to the second learned model of the present invention.

[0047] The road surface-trained model 340 is pre-trained using training data that takes road surface image RP as input data and outputs the location and degree of unevenness of the road surface RS as output data. As shown in Figure 4, when road surface image RP is input, it is configured to output the road surface condition RC, i.e., the location and degree of unevenness of the road surface RS.

[0048] The road surface condition detection unit 34 may be further configured to detect the inclination of the road surface RS. In this case, for example, the forklift 1 may further include a sensor for detecting the inclination of the road surface RS, and the road surface condition detection unit 34 may detect the inclination of the road surface RS by reading the output of this sensor. Alternatively, the output data of the training data of the road surface trained model 340 may include the slope of the road surface RS, and the road surface trained model 340 may be configured to output the inclination of the road surface RS when a road surface image RP is input. The inclination of the road surface RS may be indicated, for example, by a score corresponding to the degree of the inclination. The configuration in which the road surface condition detection unit 34 detects the inclination of the road surface RS is not particularly limited.

[0049] The first pre-trained model 35 is pre-trained using training data that takes the input data of the driving operation controls and the road surface condition RC as input data, and the output data of the driving operation controls when passing over road surface RS. As a result, the first pre-trained model 35 has pre-learned the correlation between the road surface condition RC and the driving operation controls that change due to the influence of the road surface condition RC. The driving operation controls include the controls of the accelerator control unit 161, the steering wheel control unit 160, and the brake control unit 162. As shown in Figure 4, the first pre-trained model 35 is configured to output a correction value for the driving operation controls when it receives the driving operation controls and the road surface condition RC as input.

[0050] In the first trained model 35 of this embodiment, information about the direction of travel, such as whether it is moving forward or backward, is also included in the input data of the training data. Because the forklift 1 is carrying a load W and the positions and sizes of the front wheels 10 and rear wheels 11 are different, the vibrations when passing over uneven surfaces RS differ between moving forward and backward. Therefore, the first trained model 35 can output correction values ​​for the operation amounts related to driving operations that are suitable for moving forward and backward by learning for both forward and reverse.

[0051] In this embodiment, the first trained model 35 further includes the positions of the wheels (front wheels 10 and rear wheels 11), the weight of the load W, the height of the forks 21 (position of the loaded load W), the travel speed, and the weight of the attachment as input data for the training data. As a result, in addition to the control quantities related to the travel operation and the road surface condition RC, the first trained model 35, upon receiving the wheel positions, direction of travel, weight of the load W, height of the forks 21, travel speed, and weight of the attachment as inputs, outputs corrected values ​​for the control quantities related to the travel operation based on these inputs.

[0052] The vibrations experienced by the forklift 1 as it passes over uneven surfaces vary depending on the wheel position, direction of travel, load weight W, fork height 21, travel speed, and attachment weight. Therefore, the first trained model 35 can output more appropriate correction values ​​for uneven surfaces RS by learning these factors as training data.

[0053] Furthermore, the first trained model 35 can output a correction value for the steering wheel 160 operation amount, taking into account unintended lateral sliding of the forklift 1 when traveling on an inclined road surface RS. Unintended lateral sliding of the forklift 1 when traveling on an inclined road surface RS varies depending on the wheel position, the weight of the load W, the weight of the attachment, the travel speed, and the height of the forks 21. Therefore, by learning these as training data, the first trained model 35 can output a more appropriate correction value for the inclination of the road surface RS. This correction value is a correction value that makes the steering wheel 160 operation amount such that it is possible to travel in a straight line without steering even if the road surface RS is inclined.

[0054] The correction unit 36 ​​corrects the control amount related to driving operation based on the correction value of the control amount related to driving operation output by the first trained model 35. As a result, the control amount related to driving operation is corrected to the control amount intended by the driver H, even when the forklift 1 shakes when passing over unevenness in the road surface RS. The timing of the correction unit 36 ​​correcting the control amount related to driving operation may be based on the position of the unevenness. In addition, the correction unit 36 ​​will not correct the unevenness even if it is in the direction of travel of the forklift 1, unless it coincides with the position of the wheels in the left-right direction.

[0055] Furthermore, if the road surface RS is inclined, the correction unit 36 ​​corrects the amount of steering wheel 160 operation according to the output correction value. As a result, the amount of steering wheel 160 operation is corrected to an amount that allows the vehicle to drive straight without steering, even if the road surface RS is inclined.

[0056] The power control unit 37 controls the running gear 23 and brake 25 based on the operation amount related to the running operation corrected by the correction unit 36, causing the forklift 1 to perform the running operation intended by the driver H.

[0057] <Operation of the manipulated variable correction system> Next, referring to the flowchart in Figure 5, we will explain the operation of the manipulated variable correction system S again.

[0058] (1) The maneuvering amount correction system S has pre-stored information about the forklift 1, such as whether or not the forklift 1 is equipped with an attachment, the weight of the attachment if it is equipped, the position of the wheels, and the height of the forks 21 (see S1 in Figure 5).

[0059] (2) Next, the manipulator correction system S detects the weight of the load W using the weight detection unit 32 (see S2 in Figure 5).

[0060] (3) Next, the control amount correction system S captures the road surface RS in the direction of travel using the road surface camera 19 and generates a road surface image RP (see S3 in Figure 5).

[0061] (4) Next, the manipulator correction system S detects the road surface condition RC in the direction of travel using the road surface condition detection unit 34 (see S4 in Figure 5).

[0062] (5) Next, the control input correction system S outputs a correction value for the control input related to the driving operation based on the first learned model 35, the road surface condition RC, the wheel position, the direction of travel, the weight of the load W, the height of the fork 21, the driving speed, and the weight of the attachment (see S5 in Figure 5).

[0063] (6) Next, the control input correction system S corrects the control input related to the driving operation based on the correction value of the control input related to the driving operation using the correction unit 36 ​​(see S6 in Figure 5).

[0064] (7) Next, the control input correction system S controls the running gear 23 and brake 25 by the power control unit 37 according to the control input related to the running operation corrected by the correction unit 36, causing the forklift 1 to perform the running operation intended by the driver H (see S7 in Figure 5).

[0065] The control amount correction system S, with the above configuration, can predict fluctuations in the control amount due to vibrations when passing over uneven surfaces and sliding of the forklift 1 when traveling on slopes, and can correct the control amount to suit the road surface condition RC. As a result, the forklift 1 can travel stably even if its operation is disrupted by the road surface condition RC, and even on an inclined road surface RS.

[0066] Furthermore, for example, if the forklift 1 has a constant speed driving function and the on / off switching of this constant speed driving function is related to the accelerator operation amount, the operation amount correction system S can prevent the constant speed driving function from being unintentionally switched on or off due to the road surface condition RC.

[0067] Although one embodiment of the manipulated variable correction system of the present invention has been described above, the manipulated variable correction system according to the present invention is not limited to the above embodiment. For example, the manipulated variable correction system according to the present invention may be implemented by the following modifications, or by appropriately combining the following modifications.

[0068] <Variation> The control quantities related to driving operations may include, for example, at least one of the control quantities for the accelerator control unit 161, the brake control unit 162, and the steering wheel 160. The correction values ​​for the control quantities related to driving operations output by the first trained model 35 may include at least one correction value for the control quantities for the accelerator control unit 161, the brake control unit 162, and the steering wheel 160. In this case, the control quantity correction system S corrects fluctuations in the control quantities related to driving caused by the road surface condition RC by correcting only at least one of the control quantities for the accelerator control unit 161, the brake control unit 162, and the steering wheel 160.

[0069] The first trained model 35 may be trained to correct for only one of the control variables, either forward or reverse. In this case, the control variable correction system S corrects the control variable related to the driving operation only when moving forward or in reverse.

[0070] The first trained model 35 does not need to have all of the following information included in the training data input: wheel position, direction of travel, load W weight, attachment weight, travel speed, and fork height 21. In this case, the first trained model 35, upon receiving information included in the training data input in addition to the control quantities related to the driving operation and the road surface condition RC, outputs a correction value for the control quantities related to the driving operation based on that input.

[0071] The control input correction system S may further include a load camera that photographs the load W loaded by the forklift 1 and generates a load image, and a third pre-trained model. The load camera may be replaced by a road surface camera 19. The third pre-trained model is pre-trained using training data that takes the load image as input data and outputs the center of gravity position of the load W when it is loaded onto the forklift 1. When a load image is input, it outputs the center of gravity position of the load W. Furthermore, the input data of the training data for the first pre-trained model 35 may include the center of gravity position of the load W, and the first pre-trained model 35 may further receive the center of gravity position of the load W as input and output a correction value for the control input based on the center of gravity position of the load W. Since the vibration and sliding of the forklift 1 based on the road surface condition RC also fluctuate depending on the center of gravity position of the load W, the first pre-trained model 35 can output a more appropriate correction value for the road surface condition RC by learning the center of gravity position of the load W as training data. Furthermore, since the combination of the weight of the load W, the center of gravity of the load W, and the height of the load W (height of the forks 21) affects the vibration and sliding of the forklift 1, it is preferable that the input data for the training data of the first trained model 35 includes the weight of the load W, the center of gravity of the load W, and the height of the forks 21. [Explanation of Symbols]

[0072] S-type control adjustment system H Driver Distance between the CD's uneven surface and the wheel RS road surface RP road surface image RC road surface conditions W load 1 Forklift 10 Front Wheel 11 Rear wheels 13 Car bodies 14. Driver's seat 16 Control section 160 Handle 161 Accelerator control unit 162 Brake operating section 163 Forward / Forward Switching Section 17 Headguard 19 Roadside cameras 20 Mast 21 Forks 23. Running gear 25 Brake 26. Driving speed sensor 30 Control Unit 31 Storage section 32 Weight detection unit 33 Fork height detection unit 34 Road surface condition detection unit 340 Road surface trained model (2nd trained model) 35. First pre-trained model 36 Correction Unit 37 Power Control Unit 100 Conventional forklifts

Claims

1. A road surface condition detection unit that detects the condition of the road surface, A first trained model is pre-trained using training data that takes the amount of operation related to driving and the road surface condition as input data, and the amount of operation related to driving when passing over the road surface as output data, and when the amount of operation related to driving and the road surface condition are input, it outputs a correction value for the amount of operation related to driving. An operation quantity correction system comprising: a correction unit that corrects the operation quantity related to driving operations based on the correction value output by the first trained model.

2. The operation amount correction system according to claim 1, wherein the road surface condition includes the unevenness of the road surface and the inclination of the road surface or either one thereof.

3. The control amount correction system according to claim 1 or 2, wherein the control amount related to driving operations includes the amount of operation of the accelerator.

4. The operation amount correction system according to claim 1 or 2, wherein the operation amount related to driving operations includes the brake operation amount.

5. The operation amount correction system according to claim 1 or 2, wherein the operation amount related to driving operations includes the operation amount of the steering wheel.

6. The input data for the training data of the first trained model further includes information about the direction of movement, whether it is moving forward or backward. The control quantity correction system according to claim 1 or 2, wherein the first trained model is further input with respect to the direction of travel information and outputs a correction value for the control quantity related to the driving operation based on the direction of travel information.

7. It further includes a weight detection unit for detecting the weight of the load, The input data for the training data of the first trained model further includes the weight of the load. The operation quantity correction system according to claim 1, wherein the first trained model is further input to the weight of the load and outputs a correction value for the operation quantity related to the driving operation based on the weight of the load.

8. It further includes a fork height detection unit that detects the height of the forks, The input data for the training data of the first trained model further includes the height of the fork, The operation amount correction system according to claim 1 or 7, wherein the first learned model is further input to the height of the fork and outputs a correction value for the operation amount relating to the driving operation based on the height of the fork.

9. The input data for the training data of the first trained model further includes the driving speed and the position of the wheels. The control quantity correction system according to claim 1 or 2, wherein the first trained model is further input to the travel speed and the position of the wheels, and outputs a correction value for the control quantity relating to the travel operation based on the travel speed and the position of the wheels.

10. The input data for the training data of the first trained model further includes the weight of the attachment. The control amount correction system according to claim 1 or 7, wherein the first learned model is further configured to receive the weight of the attachment and output a correction value for the control amount relating to the driving operation based on the weight of the attachment.

11. The system further includes a road camera that photographs the road surface in the direction of travel and generates a road surface image, The manipulated variable correction system according to claim 1 or 2, wherein the road surface condition detection unit has a second trained model that is pre-trained using training data that takes the road surface image as input data and the road surface condition as output data, and outputs the road surface condition when the road surface image is input.

12. A cargo camera that photographs the cargo being loaded and generates a cargo image, The system further comprises a third pre-trained model that is pre-trained using training data that takes the aforementioned load image as input data and the position of the load's center of gravity as output data, and outputs the position of the load's center of gravity when the load image is input. The input data for the training data of the first trained model further includes the position of the center of gravity of the load. The manipulated variable correction system according to claim 1 or 7, wherein the first trained model is further input to the center of gravity position of the load and outputs a correction value for the manipulated variable based on the center of gravity position of the load.

13. A forklift comprising the maneuvering amount correction system according to claim 1 or 2.

14. A control input correction program used in a vehicle equipped with a computer, To the aforementioned computer, A road surface condition detection unit that detects the condition of the road surface, A first trained model is pre-trained using training data that takes the amount of operation related to driving and the road surface condition as input data, and the amount of operation related to driving when passing over the road surface as output data, and when the amount of operation related to driving and the road surface condition are input, it outputs a correction value for the amount of operation related to driving. A correction unit that corrects the control amount related to driving operations based on the correction value output by the first trained model, and a control amount correction program that is executed by the correction unit.

Images