Off-road capability determination device and off-road capability determination system

Abstract

To quickly determine traveling performance on a driving surface.SOLUTION: A traveling performance determination device includes a data acquisition unit that acquires measurement data measured by a test vehicle that has traveled on a driving surface to be evaluated, a storage unit that performs machine learning on the measurement data acquired by the test vehicle traveling on various driving surfaces and stores a determination program for determining the traveling performance of the driving surface based on the measurement data, and an operation unit that processes the measurement data to be determined based on the determination program stored in the storage unit and determines the traveling performance of the driving surface to be evaluated. The measurement data includes vertical vibration data when the test vehicle has traveled on the driving surface to be evaluated.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a rutting determination device and a rutting determination system for a path on which a construction machine travels.

Background Art

[0002] Construction machines, especially those used in civil engineering work, travel on an unpaved driving surface. As a device for determining the driving surface to be traveled, Patent Document 1 describes a system used when traveling on a set road, but it describes a system that photographs the behavior of other vehicles and detects the state of the road in the traveling direction based on that behavior.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Here, construction machines may work on an unpaved driving surface (ground), such as at a construction site, for example, in rivers or on the coast. Also, in the event of a landslide or other disaster, construction machines may be used for recovery. When a construction machine travels on a soft ground containing moisture, depending on the state of the driving surface, there is a risk that the running part may sink into the driving surface and become immovable, or that the machine may not be able to maintain its posture.

[0005] <000,028>[[ID=e39]]As a method for determining the state of the driving surface, there is a method in which an operator presses a cone penetrometer into the ground and performs a penetration test to measure the strength of the ground, but the work becomes complicated. Also, workers need to work on-site. When multiple vehicles, such as on a road, travel in a fixed direction, the system of Patent Document 1 can be used, but it is difficult to apply when the evaluation target driving surface is not a road.

[0006] This disclosure has been made in view of the above-mentioned problems, and aims to provide a traction determination device and traction determination system that can quickly determine the traction of a running surface. [Means for solving the problem]

[0007] To achieve the above objectives, a drivability determination device according to one aspect of the present disclosure includes: an information acquisition unit that acquires measurement data measured by a test vehicle that travels on a road surface to be evaluated; a storage unit that stores a determination program that uses machine learning to acquire measurement data obtained by the test vehicle traveling on various road surfaces and determines the drivability of the road surface based on the measurement data; and a calculation unit that processes the measurement data to be determined based on the determination program stored in the storage unit and determines the drivability of the road surface to be evaluated, wherein the measurement data includes data on vertical vibration when the test vehicle travels on the road surface to be evaluated.

[0008] The calculation unit performs a Fourier transform on the vibration data, calculates a value for each frequency component, and inputs the values ​​for each frequency component into the judgment program.

[0009] The measurement data includes data on the shape of the evaluation target surface before and after the test vehicle travels over it.

[0010] The determination program makes a determination based on the results of learning the frequency data and the shape data in a single model.

[0011] The determination program learns the frequency data and the shape data using separate models and makes a determination based on the calculated results of each.

[0012] The measurement data includes data on the output of the test vehicle while it is driving on the evaluation target surface.

[0013] The calculation unit determines whether a vehicle larger than the test vehicle can travel on the evaluation target travel surface.

[0014] The determination program performs machine learning using, as teacher data, measurement data obtained by traveling on a driving surface associated with having the wearability and measurement data obtained by traveling on a driving surface associated with not having the wearability.

[0015] To achieve the above object, a wearability determination system according to an aspect of the present disclosure includes the wearability determination device according to any one of the above and a test vehicle that travels on an evaluation target driving surface and supplies measurement data measured during traveling to the wearability determination device.

[0016] The wearability determination device is separate from the test vehicle and receives measurement data through communication with the test vehicle.

Advantages of the Invention

[0017] According to the present disclosure, it is possible to quickly determine the wearability of a driving surface.

Brief Description of the Drawings

[0018] [Figure 1] FIG. 1 is a schematic diagram showing a schematic configuration of the wearability determination system of the present embodiment. [Figure 2] FIG. 2 is a block diagram of the wearability determination system of the present embodiment. [Figure 3] FIG. 3 is a flowchart showing an example of a learning process of the process of the wearability determination system. [Figure 4] FIG. 4 is an explanatory diagram for explaining an example of a method for obtaining measurement data for learning. [Figure 5] FIG. 5 is a schematic diagram showing an example of measurement data. [Figure 6] FIG. 6 is a schematic diagram showing an example of processed measurement data. [Figure 7] FIG. 7 is a schematic diagram showing an example of a model of machine learning. [Figure 8] FIG. 8 is a flowchart showing an example of a determination process of the wearability determination system. [Figure 9]FIG. 9 is a diagram showing an example of processing measurement data determined to have no running performance. [Figure 10] FIG. 10 is a diagram showing an example of processing measurement data determined to have running performance. [Figure 11] FIG. 11 is a flowchart showing an example of the determination process of the running performance determination system. [Figure 12] FIG. 12 is a flowchart showing an example of the determination process of the running performance determination system.

Embodiments for Carrying Out the Invention

[0019] Hereinafter, embodiments of the running performance determination system according to the present invention will be described in detail based on the drawings. Note that the present invention is not limited to the description of the following embodiments. Also, the components in the following embodiments include those that can be replaced by those skilled in the art, those that are easy to replace, or those that are substantially the same. Furthermore, the components in the following described embodiments can be subjected to various omissions, substitutions, or changes in configuration without departing from the gist of the present invention. In the following embodiments, one embodiment of the running performance determination system according to the present invention is exemplified, and the necessary components will be described while omitting other components

[0020] Figure 1 is a schematic diagram showing the general configuration of the off-road capability determination system of this embodiment. Figure 2 is a block diagram of the off-road capability determination system of this embodiment. The off-road capability determination system 10 shown in Figures 1 and 2 includes a test vehicle 12, a remote control device 13, and an off-road capability determination device 14. The off-road capability determination system 10 determines the off-road capability by having the test vehicle 12 travel on the target travel surface (ground, travel road surface). The criteria for off-road capability include the calculation of the cone index. By calculating the cone index, it is possible to determine the off-road capability of construction machinery, for example, whether it is possible to travel on the same surface with construction machinery other than the test vehicle 12. Note that the criteria for off-road capability are not limited to the cone index, and various criteria that can determine the off-road capability of construction machinery can be used. In addition, although the test vehicle 12 and the off-road capability determination device 14 are separate components in the off-road capability determination system 10 of this embodiment, they may be integrated. Furthermore, the off-road capability determination device 14 includes a machine learning function and a function to perform off-road capability determination processing on measurement data acquired by the test vehicle 12 using the results of machine learning, but each of these may be performed on a separate computing device.

[0021] The test vehicle 12 is a small construction machine capable of traveling on rough terrain, such as a skid steer loader. The test vehicle 12 comprises a main body 22 and a drive unit 24. The main body 22 is the frame of the test vehicle 12 and houses the drive source, control functions, etc. The test vehicle 12 in this embodiment is a remotely operated device and does not have a driver's seat, but it may have a driver's seat and an operating unit. Alternatively, the test vehicle 12 may be a device in which an operator stands on the running surface at the rear of the test vehicle 12 and operates the test vehicle 12. The drive unit 24 is the part that contacts the running surface and moves the test vehicle 12. In this embodiment, the drive unit 24 is a track. The drive unit 24 may also be a tire.

[0022] The test vehicle 12 has a main body 22 equipped with a behavior detection unit 26, a front shape detection unit 28, a rear shape detection unit 30, and position detection units 32, 34, and 36, which serve as sensors for evaluating the condition of the road surface under evaluation while driving. The test vehicle 12 also includes a control unit 40 and a communication unit 42.

[0023] The behavior detection unit 26 is, for example, located in the center of the main body 22. The behavior detection unit 26 is equipped with an acceleration sensor and detects vertical vibrations of the test vehicle 12. The behavior detection unit 26 may also detect the speed, attitude, etc., of the test vehicle 12. The behavior detection unit 26 is, for example, an IMU (Inertial Measurement Unit). The behavior detection unit 26 may also acquire information from the control unit 40 (input values ​​to the drive unit 24), i.e., operation information, as behavior information.

[0024] The front shape detection unit 28 is positioned at the front end of the main body 22 (the front side when the drive unit 24 of the test vehicle 12 is moved forward). The front shape detection unit 28 detects the shape of the evaluation target running surface on the front side of the vehicle. In other words, the front shape detection unit 28 detects the shape of the evaluation target running surface before the test vehicle 12 passes over it. The rear shape detection unit 30 is positioned at the rear end of the main body 22 (the rear side when the drive unit 24 of the test vehicle 12 is moved forward). The rear shape detection unit 30 detects the shape of the evaluation target running surface on the rear side of the vehicle. In other words, the rear shape detection unit 30 detects the shape of the evaluation target running surface after the test vehicle 12 has passed over it. The front shape detection unit 28 and the rear shape detection unit 30 can use various devices that can detect the shape of the running surface in the target area, such as a camera, LiDAR (Light detection and ranging), and an infrared sensor (a sensor that emits infrared light as detection light and detects the distance to the running surface by detecting its reflection).

[0025] The position detection units 32, 34, and 36 detect the position relative to the road surface under evaluation. The position detection units 32, 34, and 36 can use GSNN (Global Navigation Satellite System) sensors or beacons that communicate with a base station whose position relative to the area in which the test vehicle 12 travels is set. The position detection unit 32 is positioned next to the behavior detection unit 26. The position detection unit 34 is positioned next to the front shape detection unit 28. The position detection unit 36 ​​is positioned next to the rear shape detection unit 30. By acquiring their respective position information, the position detection units 32, 34, and 36 detect the position of the test vehicle 12 along with the corresponding positions of the behavior detection unit 26, the front shape detection unit 28, and the rear shape detection unit 30. This allows the positions of the behavior detection unit 26, the front shape detection unit 28, and the rear shape detection unit 30 to be identified, and their respective detection information to be associated with each other.

[0026] The control unit 40 controls the movement of the test vehicle 12 and outputs information detected by sensors. The control unit 40 includes arithmetic functions executed by a processor such as a CPU (Central Processing Unit) microprocessor, microcomputer, DSP (Digital Signal Processor), and system LSI (Large Scale Integration), and storage functions executed by non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Registered Trademark) (Electrically Erasable Programmable Read Only Memory), magnetic disks, flexible disks, optical disks, compact disks, minidiscs, or DVDs, and controls the behavior of the test vehicle 12. The control unit 40 also outputs information acquired by various sensors to the off-road capability determination device 14 via the communication unit 42.

[0027] The communication unit 42 is connected to a communication network in a state where it can communicate data with each other, and sends and receives data. The communication network may be constructed including public communication lines and dedicated communication lines. The communication unit 42 communicates with the operation unit that receives remote control input and outputs the input operation information to the control unit 40. The communication unit 42 also outputs information detected by the sensors to the off-road capability determination device 14.

[0028] The remote control device 13 remotely operates the test vehicle 12. The remote control device 13 may be an integrated device with the off-road capability determination device 40. The remote control device 13 communicates with the test vehicle 12 and controls the movement of the test vehicle 12. The method of remote operation of the remote control device 13 is not particularly limited; for example, a mechanism similar to the control unit of the test vehicle 12 may be provided to detect input operations, or operations input via a keyboard, mouse, etc., may be detected.

[0029] The off-road capability determination device 40 uses measurement data received from the test vehicle 12 to determine the off-road capability of the test vehicle 12. Furthermore, the off-road capability determination device 40 uses the measurement data received from the test vehicle 12 and the corresponding input information as training data to perform machine learning and create a determination program for determining off-road capability. In other words, it performs training on a set learning model.

[0030] The off-road capability determination device 14 includes a communication unit 50, an information acquisition unit 52, an input unit 54, an output unit 56, a calculation unit 58, and a storage unit 60.

[0031] The communication unit 50 is connected to a communication network in a state where mutual data communication is possible, and it sends and receives data. The information acquisition unit 52 acquires information acquired by the communication unit 50 through communication with the test vehicle 12, specifically information acquired by various sensors when the vehicle is driving on the road surface to be evaluated.

[0032] The input unit 54 is used by the operator operating the off-road capability determination device 14 to input operations. The input unit 54 can be a keyboard, mouse, touch panel, etc. The output unit 56 displays various information to the operator operating the off-road capability determination device 14. The output unit 56 can display the determination results generated by the calculation unit 58. The output unit 56 may include a display device such as a liquid crystal display (LCD), organic electro-luminescence display (OELD), or inorganic electro-luminescence display (IELD), and an audio output unit such as a speaker. The output unit 56 may include an input device such as a touchscreen.

[0033] The arithmetic unit 58 performs various processes on the information detected by the test vehicle 12 based on the programs and data stored in the storage unit 60. In particular, in this embodiment, the arithmetic unit 58 performs a process to determine the drivability of the evaluation target road surface that the test vehicle 12 has traveled on, and performs machine learning of the model used for the determination. The arithmetic unit 58 may be implemented including a processor such as a CPU (Central Processing Unit) microprocessor, a microcomputer, a DSP (Digital Signal Processor), or a system LSI (Large Scale Integration). The arithmetic unit 58 reads the programs stored in the storage unit 60 and expands them into working memory such as RAM, and causes the CPU or other processor to execute the instructions contained in the programs expanded into working memory. As a result, the arithmetic unit 58 can perform various processes based on each function. The arithmetic unit 58 has a machine learning unit 70 and a determination processing unit 72. Each part of the arithmetic unit 58 will be described later.

[0034] The storage unit 60 stores programs and data for realizing various processes executed by the arithmetic unit 58. The functions provided by the programs stored in the storage unit 60 include a function to determine the drivability of the surface traveled by the test vehicle 12, and a function to perform machine learning on the model used in the determination program. The data stored in the storage unit 60 includes measurement data 86 received from the test vehicle 12. The measurement data 86 may also be a processed dataset that serves as training data for machine learning. The storage unit 60 may be implemented using non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Registered Trademark) (Electrically Erasable Programmable Read Only Memory), magnetic disks, flexible disks, optical disks, compact disks, minidiscs, or DVDs.

[0035] The storage unit 60 includes a data processing program 80, a judgment program 82, and a learning program 84. The data processing program 80 processes the measurement data measured by the test vehicle 12 into data that can be processed by the judgment program 82 and the learning program 84. For example, the data processing program 80 performs a Fourier transform on vibration data from the measurement data for a specified period and outputs the frequency components (hereinafter described as frequencies) for that period as output data. The data processing program 80 also performs a process to associate multiple acquired data. Specifically, the data processing program 80 associates the processed data with position information and performs a process to associate the Fourier transformed data and the shape processing result data in the same domain.

[0036] The judgment program 82 processes the measurement data obtained when the test vehicle 12 travels on the evaluation target surface, which has been processed by the data processing program 80, using a pre-trained program, and outputs a judgment result for off-road capability based on the processing results.

[0037] The learning program 84 uses multiple measurement data and data associated with the evaluation value of the vehicle's ability to travel for each measurement data as training data to perform machine learning on a learning model for vibration judgment (hereinafter referred to as the learning model). The learning model is not particularly limited, and various models for processing data can be used. Furthermore, the learning program may train separate models for each parameter of the measurement data, or it may train a single model by associating multiple parameters of the measurement data.

[0038] Next, the functions of each part of the calculation unit 58 will be explained. The machine learning unit 70 is executed by processing the learning program 84. The machine learning unit 70 performs machine learning on the learning model using multiple measurement data and data to which the evaluation value of off-road capability is associated for each measurement data as training data. The judgment processing unit 72 is executed by processing the judgment program 82. The judgment processing unit 72 processes the measurement data acquired when the test vehicle 12 travels on the evaluation target road surface and determines the off-road capability.

[0039] Next, we will explain the processing of the off-road capability determination system 10 using Figures 3 to 12. First, we will explain the process of creating the determination criteria for the determination process using machine learning, using Figures 3 to 7. Figure 3 is a flowchart showing an example of the learning process of the off-road capability determination system. Figure 4 is an explanatory diagram to show an example of how to acquire measurement data for learning. Figure 5 is a schematic diagram showing an example of measurement data. Figure 6 is a schematic diagram showing an example of processed measurement data. Figure 7 is a schematic diagram showing an example of a machine learning model.

[0040] The off-road capability determination system 10 drives the test vehicle 12 over the target road surface and acquires measurement data (step S12). The target road surface is one whose off-road capability is known. The evaluation of the target road surface may be performed before or after driving. The off-road capability criterion is one that can evaluate whether a construction machine larger and heavier than the test vehicle 12 can traverse the road surface. In this embodiment, the case of determining whether or not off-road capability exists will be described.

[0041] For example, in the test area 100 shown in Figure 4, a travel route 102 is set. The travel route 102 includes areas 110 and 114. Area 110 is a travel surface with soft ground and no traversability. Area 110 is, for example, filled with cohesive soil to a predetermined depth, for example, 0.15 m, while area 114 is a travel surface with firm ground and traversability. The test vehicle 10 acquires measurement data for the travel surface with no traversability by acquiring data when traveling through area 110. The test vehicle 10 acquires measurement data for the travel surface with traversability by acquiring data when traveling through area 114. The test vehicle 10 also travels through areas 110 and 114 at a predetermined speed. Specifically, vibration data shown in Figure 5 is acquired. In this embodiment, the vibration to be measured is the vertical component of the measurement data. In the measurement results shown in Figure 5, period 120 represents the vibration measurement results when traveling through region 110, and period 122 represents the vibration measurement results when traveling through region 114. The correspondence between the vibration measurement data and the travel position is determined based on the information from the position detection unit 28. In this embodiment, vertical vibration is described, but the travel speed in other directions and the shape of the travel surface before and after travel (change in shape) can also be obtained as measurement data.

[0042] The off-road capability determination system 10 processes the acquired measurement data into data that can be used for machine learning (step S14). The off-road capability determination system 10 processes the measurement data with the data processing program 80. In this embodiment, the vibration data from the measurement data is Fourier transformed. As a result, the results of frequency analysis of the vibration are calculated, as shown in Figure 6. As a result of processing, the off-road capability determination system 10 calculates output values ​​for each frequency within a set range of frequency bands. An example of a frequency band is the range from 1 Hz to n Hz. The measurement data is, for example, data measured with a period of 0.6 seconds.

[0043] The off-road capability determination system 10 associates the off-road capability determination result with the measurement data (step S16). The off-road capability determination system 10 associates data from travel through area 110 with "no off-road capability" and data from travel through area 114 with "off-road capability."

[0044] The off-road capability determination system 10 determines whether the acquisition of measurement data is complete (step S18). If the off-road capability determination system 10 determines that the acquisition of measurement data is not complete (No in step S18), it returns to step S12 and acquires more measurement data. The off-road capability determination system 10 repeats the process from step S12 to step S16 to accumulate measurement data to be used for learning.

[0045] When the off-road capability determination system 10 determines that the acquisition of measurement data is complete (Yes in step S18), it performs machine learning using processed measurement data as input and the off-road capability determination result as output (step S20). Here, machine learning is performed by inputting training data to the learning model 130, as shown in Figure 7. The learning model 130 is a model in which multiple intermediate layers 142 are set between the input layer 140 and the output layer 144. In machine learning, the input values ​​of the measurement data are input to the input layer 140, the off-road capability determination result is input to the output layer 144, and training is performed to calculate the coefficients of the intermediate layers 142, thereby making the learning model 130 a trained model. For example, in this embodiment, the output values ​​of the angular frequency at 1 Hz, calculated by Fourier transforming the vibration, are input to the input layer 140.

[0046] The off-road capability determination system 10 stores the learned program in the memory unit (step S22).

[0047] Next, using Figures 8 to 10, we will explain the process of determining off-road capability by driving on the target driving surface. Figure 8 is a flowchart showing an example of the determination process of the off-road capability determination system. Figure 9 is a diagram showing an example of measurement data in the range of 1 Hz to 15 Hz where off-road capability is determined to be absent. Figure 10 is a diagram showing an example of measurement data in the range of 1 Hz to 15 Hz where off-road capability is determined to be present. The process shown in Figure 8 is the case when processing is performed based on vibration data detected by the behavior detection unit 26 from the measurement data.

[0048] The off-road capability determination system 10 acquires measurement data (step S32). Specifically, the off-road capability determination system 10 obtains data detected by sensors during driving by the test vehicle 12 on the evaluation surface via communication to the off-road capability determination device 14.

[0049] The off-road capability determination system 10 performs a Fourier transform on the vibration data and calculates values ​​for each frequency (step S34). The off-road capability determination device 14 performs a Fourier transform on the measurement data to calculate the waveforms shown in Figures 9 and 10, and calculates the output (amplitude) for each frequency in the target frequency range. Here, Figure 9 is the result of performing a Fourier transform on measurement data when driving on a surface with no off-road capability. Figure 10 is the result of performing a Fourier transform on measurement data when driving on a surface with off-road capability. As shown in Figures 9 and 10, the output changes depending on whether or not the surface has off-road capability.

[0050] The off-road capability determination system 10 inputs the processed data into the trained program (step S36). The off-road capability determination device 14 inputs the output information for each frequency calculated in step S34 into the trained program.

[0051] The off-road capability determination system 10 outputs a determination result (step S38). In this embodiment, it outputs a determination result indicating whether or not the vehicle has off-road capability.

[0052] As described above, the off-road capability evaluation system 10 can evaluate off-road capability with high accuracy by having the test vehicle 12 drive on the target driving surface, acquiring data detected during the drive, particularly vertical vibration data, and evaluating off-road capability based on the acquired vibration data. Furthermore, since the data necessary for evaluation can be acquired by actually driving the test vehicle 12, data acquisition becomes easier. Specifically, evaluations such as using a cone penetrometer on the target driving surface to evaluate it become unnecessary. In addition, off-road capability can be evaluated for the area that the test vehicle 12 has driven on. This allows the target driving surface to be wide-ranging, and evaluations can be performed for each area.

[0053] The off-road capability evaluation system 10 evaluates off-road capability based on whether a construction machine larger than the test vehicle 12 can travel the same route. By running the test vehicle 12, which is capable of traveling on rough terrain, the system can evaluate whether the construction machine to be used for the work can travel the target route.

[0054] Furthermore, by using data obtained by Fourier transforming vibration data, the ground conditions of the target travel surface can be evaluated more appropriately.

[0055] In the process shown in Figure 8, off-road capability is evaluated based on vertical vibration, but in addition to vibration, off-road capability may also be evaluated based on the shape of the running surface. Figure 11 is a flowchart showing an example of the determination process of the off-road capability determination system. The process shown in Figure 11 is when processing is performed based on vibration data and shape data detected by the behavior detection unit 26 from the measurement data.

[0056] The off-road capability determination system 10 acquires measurement data (step S32). Specifically, the off-road capability determination system 10 obtains data detected by sensors during driving by the test vehicle 12 on the evaluation surface via communication to the off-road capability determination device 14.

[0057] The off-road capability determination system 10 performs a Fourier transform on the vibration data and calculates values ​​for each frequency (step S52). In parallel with processing the vibration data, the off-road capability determination system 10 processes the shape data before and after driving (step S54). For example, the off-road capability determination system 10 detects the difference between the shape of the driving surface before and after driving in the same range. Note that the criteria for evaluating the shape are not limited to this.

[0058] The off-road capability determination system 10 inputs the processed data into the trained program (step S36). The off-road capability determination device 14 inputs the output information for each frequency calculated in step S52 and the difference data of the shape calculated in step S54 into the trained program. The trained program in this embodiment is a program created by learning data in which the calculated output information for each frequency and the calculated difference data of the shape are inputs, and the off-road capability determination result of the running surface is output.

[0059] The off-road capability determination system 10 outputs a determination result (step S38). In this embodiment, it outputs a determination result indicating whether or not the vehicle has off-road capability.

[0060] In this way, by adding data on the shape of the running surface before and after driving to the measurement data, the off-road capability can be evaluated with higher accuracy. The off-road capability evaluation system 10 may also include data on the shape of the drive unit 24 of the test vehicle 12 as an evaluation criterion. This makes it possible to evaluate the off-road capability while taking into account the influence of the shape of the part that is in contact with the running surface.

[0061] Here, it is preferable that the number of data points provided to the input layer is greater than that in the 1Hz to 15Hz case of this embodiment. For example, outputs with frequencies higher than 15Hz may also be acquired as data. Alternatively, data may be acquired in units smaller than 1Hz. Furthermore, it is preferable that the data provided to the input layer includes frequencies higher than 15Hz. It is also preferable that the frequency band provided to the input data can be selected as appropriate. By selecting the frequency band, depending on the driving surface, it is possible to analyze data from frequency bands that exclude the frequency bands where outputs occur when obstacles (stones, waste materials, etc.) are included in the measurement data.

[0062] Figure 12 is a flowchart showing an example of the determination process of a vehicle drivability determination system. In the process shown in Figure 11, vibration data and shape data are input into a single trained program, but the system is not limited to this. The vibration data and shape data may be processed by separate trained programs (a trained program for vibration determination and a trained program for shape determination), and the vehicle drivability may be determined based on the results.

[0063] The off-road capability determination system 10 acquires measurement data (step S32). Specifically, the off-road capability determination system 10 obtains data detected by sensors during driving by the test vehicle 12 on the evaluation surface via communication to the off-road capability determination device 14.

[0064] The off-road capability determination system 10 performs a Fourier transform on the vibration data and calculates values ​​for each frequency (step S52). The off-road capability determination system 10 inputs the processed data into a pre-trained program for vibration determination (step S65). The off-road capability determination device 14 inputs the output information for each frequency calculated in step S52 into the pre-trained program for vibration determination. The pre-trained program for vibration determination in this embodiment is a program created by learning data in which the output information for each calculated frequency is input and the off-road capability determination result of the running surface is output.

[0065] The off-road capability determination system 10 processes shape data before and after driving in parallel with processing vibration data (step S54). The off-road capability determination system 10 inputs the processed data into a trained program for shape determination (step S64). The off-road capability determination device 14 inputs the shape difference data calculated in step S54 into the trained program for shape determination. The trained program for shape determination is a program created by learning with training data that takes the calculated shape difference data as input and outputs the off-road capability determination result of the driving surface.

[0066] The off-road capability determination system 10 performs a determination process on the road surface (driving surface) to be evaluated based on the processing results (step S66). The off-road capability determination system 10 makes a determination on off-road capability based on both the processing results based on the vibration data calculated in step S65 and the processing results based on the shape data calculated in step S64. The criteria for determining off-road capability are not particularly limited, and the result is determined by weighting the two evaluation results. If both are determined to be insufficient off-road capability, the result is determined to be insufficient off-road capability, and if at least one is determined to be insufficient off-road capability, the result is determined to be insufficient off-road capability.

[0067] The off-road capability determination system 10 outputs a determination result (step S38). In this embodiment, it outputs a determination result indicating whether or not the vehicle has off-road capability.

[0068] Thus, the measurement data can be processed separately to evaluate the off-road capability. This allows, for example, the measurement results of the shape data to be used complementaryly, and if the shape data cannot be measured, the evaluation can be performed using only the vibration data.

[0069] Furthermore, the determination of off-road capability is not limited to outputting whether or not the machine is capable of off-roading. For example, during training, the cone index can be associated with the measurement data to calculate the cone index of the target surface as the off-road capability determination result. Alternatively, off-road capability may be classified into three or more categories according to the range of the cone index. For example, the system may determine whether any construction machine is unable to traverse the surface, only some construction machines are unable to traverse the surface under severe conditions, or all construction machines are able to traverse the surface.

[0070] Furthermore, the test vehicle is not limited to one; multiple test vehicles or test vehicles of different types may be used to accumulate learning data and acquire measurement data on the driving surface to be evaluated. In this case, by inputting information about the test vehicle as input data during learning, evaluations can be performed that correspond to the characteristics of the test vehicle.

[0071] The off-road capability evaluation system 10 may include the driving conditions of the test vehicle and behavioral data other than vibration as parameters of the evaluation criteria, i.e., input values. This allows for absorption of fluctuations caused by the conditions during the test, and enables more accurate determination of off-road capability. [Explanation of Symbols]

[0072] 10. Off-road capability assessment system 12 Test Vehicles 13 Remote control device 14. Off-road capability determination device 22 car bodies 24 Drive unit 26 Behavior detection unit 28 Forward shape detection unit 30 Rear shape detection unit 32, 34, 36 Position detection unit 40 Control Unit 42, 50 Communications Department 52 Information Acquisition Department 54 Input section 56 Output section 58 Arithmetic section 60 Storage section 70 Machine Learning Department 72. Determination Processing Unit 80 Data Processing Programs 82 Judgment Program 84 Learning Programs 86 Measurement Data

Claims

1. An information acquisition unit that acquires measurement data measured by a test vehicle that traveled on an evaluation target driving surface which is an unpaved driving surface and a rough road, A storage unit stores a judgment program that uses machine learning to process measurement data obtained by driving a test vehicle on various surfaces, and that determines the drivability of the surface based on the measurement data. The system includes a calculation unit that processes the measurement data to be judged based on a judgment program stored in the storage unit to determine the drivability of the target running surface, The measurement data includes data on vertical vibration when the test vehicle travels on the evaluation target surface. The calculation unit determines whether a construction machine heavier than the test vehicle can travel on the evaluation target travel surface. Off-road capability assessment device.

2. The off-road capability determination device according to claim 1, wherein the calculation unit performs a Fourier transform on the vibration data, calculates a value for each frequency component, and inputs the value for each frequency component into the determination program.

3. The off-road capability determination device according to claim 1 or 2, wherein the measurement data includes data on the shape of the road surface to be evaluated before and after the test vehicle travels over it.

4. The off-road capability determination device according to claim 3, wherein the determination program determines the off-road capability based on the results of learning the vibration data and the shape data in a single model.

5. The off-road capability determination device according to claim 3, wherein the determination program learns the vibration data and the shape data using separate models and makes a determination based on the respective results calculated.

6. The off-road capability determination device according to any one of claims 1 to 5, wherein the measurement data includes data on the speed of the test vehicle when it travels on the evaluation target surface.

7. The off-road capability determination device according to any one of claims 1 to 6, wherein the determination program is machine-trained using measurement data obtained by driving on a driving surface associated with off-road capability and measurement data obtained by driving on a driving surface associated with inability to drive as training data.

8. A device for determining off-road capability according to any one of claims 1 to 7, A drivability determination system comprising: a test vehicle that travels on a target driving surface and supplies measurement data measured during the journey to the drivability determination device.

9. The off-road capability determination system according to claim 8, wherein the off-road capability determination device is separate from the test vehicle and receives measurement data via communication with the test vehicle.

Images