Concrete management system, concrete management method, and concrete management program

Abstract

To reduce labor required for evaluation of concrete at a construction site.SOLUTION: The concrete management system 1 includes the image acquiring unit 111 that acquires a captured image acquired from the imaging device 10A, the recognition unit 112 that determines the start and end of unloading of concrete on the basis of the captured image, and the prediction unit 114 that predicts the workability of concrete, the recognition unit 112 records unloading start information in the storage unit when it is determined that an unloading start condition is satisfied on the basis of the captured image, and the prediction unit 114 predicts the workability of concrete to be unloaded after the start of unloading in the captured image and records a prediction result in the storage unit in association with the unloading start information. When the information on the workability is output and the recognition unit 112 determines that the unloading end condition is satisfied based on the captured image, the prediction result and the unloading end information are recorded in the storage unit in association with each other.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to a concrete management system, a concrete management method, and a concrete management program for managing the quality of concrete at the time of unloading. [Background technology]

[0002] Pouring fresh concrete (hereinafter simply referred to as concrete) of poor quality can cause initial defects such as poor filling. Therefore, when concrete transported by truck is used on site, it is necessary to evaluate the quality of the concrete. This quality is traditionally evaluated using workability, which relates to the concrete's resistance to deformation and fluidity, and its resistance to material segregation.

[0003] Workability can also be described as the ease of work or construction, and is evaluated qualitatively and relatively. Workability is affected by a variety of factors, including the unit water content, the content of coarse and fine aggregate, the maximum size of coarse aggregate, additives such as water reducers and their amounts, and some of these factors have a relative effect on workability. For this reason, evaluating workability is difficult.

[0004] Conventionally, workability has been evaluated by periodic slump tests when concrete is unloaded. Furthermore, workers visually check the workability of the concrete at any time during unloading. However, slump tests are sampling tests, and they cannot grasp the quality fluctuations of the concrete unloaded between tests. Furthermore, visual checks require the worker's experience.

[0005] Therefore, technology for evaluating the flow state of concrete at work sites, etc., based on image information has also been studied (see, for example, Patent Document 1). In the technology described in this document, the flow state of a concrete sample during a slump test at work sites is captured without considering imaging conditions such as the imaging position and imaging angle, and the captured image is processed to analyze changes in the surface contour, thereby evaluating the concrete. In addition to evaluating concrete during slump tests, there is also research being done to evaluate concrete samples using AI-based machine learning (deep learning) using communication terminal devices carried by workers at construction sites, etc. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Publication No. 2019-211247 Summary of the Invention [Problem to be solved by the invention]

[0007] However, in the system described in Patent Document 1, a worker or the like needs to adjust the imaging optical axis of the communication terminal device and take images using the communication terminal device. Furthermore, when using such a system, the worker or the like needs to use the communication terminal device to start and end measurements, and also needs to monitor for deviations from the management standards. In this case, a worker must always be stationed near the unloading position. [Means for solving the problem]

[0008] The present disclosure provides a concrete management system including an image acquisition unit that acquires photographed images acquired from a photographing device, a recognition unit that determines the start and end of unloading of concrete based on the photographed images, and a prediction unit that predicts the workability of the concrete, wherein the recognition unit records unloading start information in a storage unit when it determines that an unloading start condition is met based on the photographed images, the prediction unit predicts the workability of the concrete to be unloaded after the unloading start in the photographed images, and records the prediction result in the storage unit in association with the unloading start information and outputs information related to the workability, and wherein the recognition unit determines that an unloading end condition is met based on the photographed images, and records the prediction result in association with unloading end information in the storage unit.

[0009] The present disclosure provides a concrete management method for managing concrete using a storage unit and a control unit, wherein the control unit acquires photographed images from an imaging device, and when it determines based on the photographed images that an unloading start condition is satisfied, records unloading start information in the storage unit, predicts the workability of the concrete to be unloaded after the unloading start in the photographed images, records the prediction result in the storage unit in association with the unloading start information, and outputs information related to the workability, and when it determines based on the photographed images that an unloading end condition is satisfied, records the prediction result in association with unloading end information in the storage unit.

[0010] The present disclosure provides a concrete management program for managing concrete using a storage unit and a control unit, and causes the control unit to function as a means for acquiring photographed images from an imaging device, and, when it is determined based on the photographed images that an unloading start condition has been met, recording unloading start information in the storage unit, predicting the workability of the concrete to be unloaded from the photographed images, recording the prediction result in the storage unit in association with the unloading start information, and outputting information related to the workability, and, when it is determined based on the photographed images that an unloading end condition has been met, recording the prediction result in the storage unit in association with unloading end information. [Effects of the Invention]

[0011] According to the present disclosure, it is possible to reduce the effort required to evaluate concrete at construction sites. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a schematic diagram illustrating an example of a construction site according to an embodiment. [Figure 2] FIG. 2 is a schematic diagram of a hardware configuration of the embodiment. [Figure 3] FIG. 2 is a diagram illustrating the configuration of the concrete management system according to the embodiment. [Figure 4] FIG. 2 is a conceptual diagram illustrating a work and a time frame according to the embodiment. [Figure 5] An explanatory diagram of the data recorded in the memory unit of the same embodiment, where (a) is a diagram of the first teacher information and the first prediction model, (b) is a diagram of the second teacher information and the second prediction model, (c) is a diagram of the third teacher information and the third prediction model, and (d) is a diagram of the fourth teacher information and the fourth prediction model. [Figure 6] 1 shows an example of an image of concrete taken during unloading in the same embodiment, where (a) shows the surface texture of a concrete with a slump of 17.5, (b) shows the surface texture of a concrete with a slump of 12.5, and (c) shows the surface texture of a concrete with a slump of 6.5. [Figure 7]FIG. 2 is a diagram showing an example of a captured image according to the embodiment. [Figure 8] These are explanatory diagrams of the concrete pile-up conditions during unloading in this embodiment, where (a) is level 1, (b) is level 2, (c) is level 3, (d) is level 4, (e) is level 5, (f) is level 6, (g) is level 7, and (h) is level 8. [Figure 9] 10A, 10B, and 10C are diagrams showing planar distributions of flow velocity of concrete in the same embodiment, where (a) is a diagram showing a slump of 17.5, (b) is a diagram showing a slump of 12, and (c) is a diagram showing a slump of 8.5. [Figure 10] FIG. 2 is a diagram showing measurement positions of the flow velocity of concrete in the same embodiment. [Figure 11] FIG. 2( a ) is a diagram showing the change in flow velocity of concrete over time in the same embodiment, and FIG. 2( b ) is a diagram showing the results of a spectrum analysis of the flow velocity of concrete during unloading in the same embodiment. [Figure 12] 10 is a flowchart showing a processing procedure at the time of unloading in the embodiment. [Figure 13] 10 is a flowchart showing a processing procedure at the time of unloading in the embodiment. [Figure 14] FIG. 10 is a diagram illustrating an example of a screen displayed on the administrator terminal of the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] An embodiment of the concrete control system will be described below with reference to Figures 1 to 14. The concrete control system evaluates the workability of concrete used at a construction site.

[0014] As shown in Figure 1, at a construction site C1, concrete transported by a transport vehicle A1 is supplied from a chute A2 to a hopper P2 of a pump vehicle P1. The concrete is then transported to a pouring site by a pressure-feeding device of the pump vehicle P1.

[0015] The concrete control system 1 comprises a processing terminal 10 and an administrator terminal 30. The processing terminal 10 is installed near the transport vehicle A1 at the construction site C1. The processing terminal 10 comprises a photographing device. The photographing device photographs the concrete flowing (falling) from the chute A2 into the hopper P2. The processing terminal 10 is connected to the administrator terminal 30 via a network NW such as the Internet so that data can be sent and received.

[0016] (Hardware configuration description) 2, the hardware configuration of the information processing device H10 that constitutes the processing terminal 10 and the administrator terminal 30 will be described. The information processing device H10 includes a communication device H11, an input device H12, an output device H13, a storage unit H14, and a processor H15. Note that this hardware configuration is an example, and it can also be realized by other hardware.

[0017] The communication device H11 is an interface that establishes a communication path with other devices and executes data transmission and reception, and is, for example, a network interface card or a wireless interface.

[0018] The input device H12 is a device that accepts input of various information, and is, for example, a touch panel, a mouse, a keyboard, a camera, etc. The output device H13 is a display that displays various information, a speaker, etc.

[0019] The memory unit H14 is a storage device that stores data for executing various functions of the processing terminal 10 and the manager terminal 30, and various programs such as a concrete management program. Examples of the memory unit H14 include a ROM, a RAM, a hard disk, and an SSD.

[0020] The processor H15 controls each process using programs and data stored in the storage unit H14. Examples of the processor H15 include a CPU, an MPU, a GPU, an NPU, etc. The processor H15 loads programs stored in a ROM or the like into a RAM and executes various processes for each process.

[0021] The processor H15 is not limited to a processor that performs all of its processing using software. For example, the processor H15 may include a dedicated hardware circuit (e.g., an application-specific integrated circuit (ASIC)) that performs hardware processing for at least some of the processing it performs. That is, the processor H15 may be configured as a circuit including: (1) one or more processors that operate according to a computer program (software); (2) one or more dedicated hardware circuits that perform at least some of the various processes; or (3) a combination thereof. The processor includes a CPU and memory such as RAM and ROM, and the memory stores program code or instructions configured to cause the CPU to perform processing. The memory, i.e., computer-readable medium, includes any available medium that can be accessed by a general-purpose or dedicated computer.

[0022] (Functions of concrete management system) Next, each function of the concrete management system will be explained using Figure 3. The processing terminal 10 includes a camera 10A and a processing terminal main body 10B. The processing terminal 10 is a computer terminal that uses images captured by the camera 10A to capture images (flowing images) of concrete being unloaded from the chute A2 of the transport vehicle A1 to the hopper P2 (concrete inlet) of the pump vehicle P1. The camera 10A is built into the housing of the processing terminal main body 10B and is controlled by the processing terminal main body 10B. The processing terminal main body 10B can automatically or manually adjust the shooting conditions of the camera 10A. Videos are captured under the following conditions: ISO sensitivity of 100 to 400, shutter speed of 1 / 250 to 1 / 1000 seconds, resolution of 1080P HD, and frame rate of 30 fps. In this embodiment, the processing terminal 10 is installed on a tripod in front of the chute A2. The installation height of the processing terminal 10 is, for example, 1.6 m to 1.7 m. The photographing device 10A may be externally attached to the processing terminal main body 10B.

[0023] The processing terminal body 10B evaluates the concrete using the captured image. The processing terminal body 10B includes a control unit 11, a prediction result storage unit 12, a teacher information storage unit 13, and a learning result storage unit .

[0024] The control unit 11 performs the processes described below (processes including an image acquisition stage, a learning stage, a prediction stage, a determination stage, etc.) By executing programs for each process, the control unit 11 functions as an image acquisition unit 111, a recognition unit 112, a learning unit 113, a prediction unit 114, a determination unit 115, etc.

[0025] The image acquisition unit 111 acquires images captured at the time of unloading concrete from the image capture device 10A at the construction site C1. In the learning process and prediction process of the first and second prediction models described below, the recognition unit 112 performs preprocessing such as graying a color image and equalizing its histogram.

[0026] The recognition unit 112 also determines whether unloading has started based on the captured image. In this embodiment, the unloading start condition for determining that unloading has started is that the chute A2 has been detected in the preprocessed captured image. The recognition unit 112 performs a recognition process for the chute A2 in the captured image, and if the chute A2 is newly detected, it determines that the unloading start condition has been met and records unloading start information in the prediction result storage unit 12.

[0027] The recognition unit 112 also determines whether unloading has been completed based on the captured image. In this embodiment, the unloading completion condition for determining that unloading has been completed is that the recognized chute A2 has disappeared from the captured image. If the recognized chute A2 has disappeared from the captured image, the recognition unit 112 determines that the unloading completion condition has been met, and associates the prediction result with unloading completion information and records it in the prediction result storage unit 12.

[0028] Furthermore, the recognition unit 112 detects concrete flowing down the chute A2 through image recognition. The prediction unit 114 starts and stops predicting workability based on the recognition results of the recognition unit 112. Specifically, if the recognition unit 112 newly detects concrete in the captured image after the unloading start condition is satisfied, the prediction unit 114 predicts workability. If the recognition unit 112 no longer detects concrete in the captured image during workability prediction, the prediction unit 114 stops predicting workability.

[0029] The learning unit 113 creates a model that predicts workability from an image of concrete being unloaded. In this embodiment, first to fourth prediction models are generated by performing first to fourth learning processes, which will be described later.

[0030] The prediction unit 114 predicts the workability of concrete by applying the acquired image to the first to fourth prediction models. For example, the prediction unit 114 may predict workability by performing first to fourth prediction processes using the first to fourth prediction models, respectively.

[0031] If the predicted workability exceeds the management standard, the determination unit 115 outputs an alarm to the output device H13 or the like and also transmits an alarm to the manager terminal 30 based on the predicted workability.

[0032] The training information storage unit 13 stores training information used to generate a workability prediction model. In this embodiment, first to fourth training information are stored to generate four types of prediction models. As described below, the first to fourth training information are information that associates workability with information obtained from images captured during concrete unloading. In this embodiment, the workability is measured using slump calculated by a slump test. To create the training information, for example, images of several types of concrete with different slumps are taken. For each type of concrete, the concrete unloaded onto the pump truck P1 is pumped into the hopper A3 of the transport vehicle A1 and poured into the chute A2. The concrete is circulated for approximately two hours, and the flow state of the concrete is photographed every 15 minutes for two minutes. At the same time, the slump is measured by a slump test. Furthermore, if the prediction accuracy is low, the images captured for workability prediction are used as images for additional learning.

[0033] Prediction models of concrete workability are recorded in the learning result storage unit 14. These prediction models are recorded when a learning process is performed. In this embodiment, four types of prediction models (first to fourth prediction models) are recorded.

[0034] (Recording of work information) The process of recording workpiece information by the recognition unit 112 will be described with reference to Fig. 4. The workpiece is the period from the start to the end of the work of unloading concrete by the multiple transport vehicles A1.

[0035] To recognize the chute A2, the recognition unit 112 performs preprocessing on the captured image, including graying and histogram equalization. The preprocessing may include edge extraction, which detects areas of the captured image where brightness changes significantly as edges. The recognition unit 112 also performs feature extraction on the preprocessed captured image to recognize the shape of the chute A2. The outlet of the chute A2 has a semicircular shape (curved shape). Therefore, the recognition unit 112 determines that it has recognized the chute A2 when it extracts at least a semicircular edge as a feature from the preprocessed image.

[0036] The recognition unit 112 records the work from the start to the end of the work as one piece of work information 120. The work information 120 includes the work start date and time and the work end date and time. The recognition unit 112 recognizes the start and end of the work based on the user's operation of a work start button and a work end button displayed on the screen of a display, which is, for example, the output device H13.

[0037] When the recognition unit 112 newly detects the chute A2, it determines that unloading has begun, and includes unloading start information in the workpiece information 120. Furthermore, when the recognition unit 112 newly detects concrete in the captured image, the prediction unit 114 predicts workability. Furthermore, when the recognition unit 112 is no longer able to detect concrete in the captured image during workability prediction, the prediction unit 114 stops predicting workability. This is because if the prediction unit 114 predicts workability based on the captured image when the recognition unit 112 does not detect concrete, an abnormal value will be output as the prediction result. For example, even if the flow of concrete temporarily stops for some reason while the transport vehicle A1 is parked at the unloading position, this can prevent an abnormal value from being output.

[0038] Furthermore, when the state where the chute A2 was detected changes to a state where the chute A2 is not detected, the recognition unit 112 measures the elapsed time from the point where the chute A2 is no longer detected. Possible cases where the chute A2 cannot be detected include when the transport vehicle A1 finishes unloading and leaves the unloading position, and when an obstacle such as a worker enters between the processing terminal 10 and the chute A2.

[0039] The recognition unit 112 determines that unloading is complete when the elapsed time exceeds a reference time. In other words, even if an obstruction enters between the processing terminal 10 and the chute A2, if the time during which the photographing of the chute A2 is obstructed is equal to or shorter than the reference time, the recognition unit 112 does not determine that unloading is complete. When the recognition unit 112 determines that unloading is complete, it includes unloading completion information in the work information 120. For example, the recognition unit 112 registers the period from the start of unloading to the end of unloading as one time frame for the first transport vehicle A1. In other words, the time frame is the period from the start to the end of unloading by one transport vehicle A1.

[0040] Furthermore, when the recognition unit 112 detects a new chute A2, it determines that unloading by the second transporter A1 has begun, and starts a time frame. The work information 120 generated in this manner includes time frames equal to the number of transporters A1. Each piece of time frame information is stored in association with a prediction result, which will be described later.

[0041] (Workability learning and prediction processing) Next, the workability learning process and prediction process will be described with reference to FIG. First, the first learning process and the first prediction process will be described.

[0042] 5(a), the first teaching information 221 is information that associates workability with an image that focuses on feature information corresponding to the surface condition of the concrete at the time of unloading. Here, slump is associated as workability.

[0043] Figure 6(a) shows unloading images of "slump 17.5", Figure 6(b) shows "slump 12.5", and Figure 6(c) shows "slump 6.5". The larger the slump, the higher the fluidity. FIG. 7 is an example of a screen 10C displayed on the display, which is the output device H13 of the processing terminal 10. When the recognition unit 112 determines that the chute A2 has been detected, it sets a chute detection frame F1 so that the recognized outlet of the chute A2 fits within the chute detection frame F1. The recognition unit 112 then sets a concrete detection frame F2 within the chute detection frame F1. The recognition unit 112 determines whether concrete has been recognized based on the image within the concrete detection frame F2. For example, the control unit 11 determines whether concrete has flowed down based on whether or not unevenness on the concrete surface has been recognized. When the recognition unit 112 determines that concrete has flowed down, the learning unit 113 learns workability based on the image within the concrete detection frame F2. The prediction unit 114 performs a first learning process and a first prediction process based on the image within the concrete detection frame F2. As illustrated in FIG. 7, the image of the concrete detection frame F2 and the image of the chute detection frame F1 may be displayed within frames F4 and F5.

[0044] In addition, a detection frame for detecting the concrete pile condition may be further provided below the chute detection frame F1 or the concrete detection frame F2. As the slump decreases, the unevenness of the surface of the flowing concrete increases. In addition, the pile shape of the concrete that falls into the hopper of the pump truck P1 also increases.

[0045] In the first learning process, the learning unit 113 uses the first teacher information 221 to generate a first prediction model 231 by a deep learning convolutional neural network (CNN), and records the first prediction model 231 in the learning result storage unit 14.

[0046] In the first prediction process, the prediction unit 114 performs image processing within the concrete detection frame F2, and predicts and outputs the slump using the first prediction model 231. The second learning process and the second prediction process will be described.

[0047] As shown in FIG. 5(b), the second teaching information 222 is information that associates workability with an image that focuses on the pile shape of concrete during unloading. In this embodiment, the area below the concrete detection frame F2, where the concrete has flowed down the chute A2 and accumulated in the hopper P2, is cut out and used.

[0048] As shown in Figure 8, the pile shape is classified into eight levels. Levels 1 to 3 shown in Figure 8(a), (b), and (c) indicate a state where the proportion of the area occupied by concrete is small to large, respectively, and where no concrete is piled up. Levels 4 to 7 shown in Figure 8(d), (e), (f), and (g) indicate a state where the proportion of the area occupied by concrete is small, medium, large, and extra large, respectively, and where the concrete is piled up. Level 8 in Figure 8(h) indicates a state where the piled concrete is falling apart. Each level is also associated with a workability (slump).

[0049] Then, in the second learning process, the learning unit 113 uses the second teacher information 222 to generate a second prediction model 232 by a deep learning CNN (Convolutional Neural Network), and records the second prediction model 232 in the learning result storage unit 14.

[0050] In the second prediction process, the prediction unit 114 inputs a pile image of the concrete at the time of unloading and predicts and outputs the pile level using the second prediction model 232. The prediction unit 114 predicts a second slump prediction value according to the pile level.

[0051] The third learning process and the third prediction process will be described. As shown in FIG. 5(c), the third teaching information 223 is information that associates workability with the planar distribution of the flow velocity of concrete during unloading.

[0052] Figure 9 shows a schematic representation of the contours of the planar flow velocity distribution in chute A2. Region SP1 is the high-speed region, region SP2 is the medium-speed region, and region SP3 is the low-speed region. Here, we focus on the velocity distribution rather than the velocity itself. Figure 9(a) shows a slump of 17.5, (b) shows a slump of 12, and (c) shows a slump of 8.5. As shown in Figure 9(a), when the slump is large, the difference in flow velocity between the vertical direction is small, but there is a difference in flow velocity between the center and both ends. As shown in Figure 9(b), when the slump is moderate, the difference in flow velocity between the vertical direction becomes large. As shown in Figure 9(c), when the slump becomes smaller, the difference in flow velocity between the center and both ends becomes smaller.

[0053] In this embodiment, as shown in Fig. 10, the velocity is determined by tracking characteristic points in areas a31 to a39 of the concrete on chute A2. Note that for image analysis of flow velocity, the velocity is determined by tracking characteristic points. In the third training information, attention is focused on the characteristics of such planar flow velocity distribution, and the relative flow velocity to the central flow velocity and the magnitude of fluctuation in flow velocity over one second (coefficient of variation) are used as explanatory variables, and workability (slump) is used as a response variable.

[0054] Then, in the third learning process, the learning unit 113 uses the third teacher information 223 to generate a third prediction model 233 consisting of a linear multiple regression equation with the slump as the objective variable and the average value of the relative flow velocity over one second and the magnitude of flow velocity fluctuation (coefficient of variation) as explanatory variables, and records it in the learning result memory unit 14.

[0055] In the third prediction process, the prediction unit 114 inputs an image of concrete flowing down during unloading and predicts and outputs the slump using the third prediction model 233. In this case, the prediction unit 114 obtains the velocity by tracking feature points in areas a31 to a39 in the captured image. Next, the prediction unit 114 calculates the relative flow velocity with respect to the central flow velocity and the magnitude of fluctuation in the flow velocity over one second (coefficient of variation). The prediction unit 114 then inputs the relative flow velocity and the coefficient of variation into a multiple regression equation to obtain a third slump prediction value, thereby predicting workability (slump).

[0056] The fourth learning process and the fourth prediction process will be described. As shown in FIG. 5(d), the fourth teaching information 224 is information that associates workability with the flow velocity fluctuation state of concrete during unloading.

[0057] As shown in Figure 11(a), at a slump of 14.5, the flow velocity exhibits waveform G1, which has large periodic fluctuations, while at a slump of 4.5, it exhibits waveform G2, which has small, irregular fluctuations. When the slump is large, waveform G1 is thought to be dependent on the periodic discharge caused by the rotation of the drum of the transport vehicle A1. When the slump is small, waveform G2 is less affected by the drum rotation and exhibits smaller periodic fluctuations. On the other hand, the friction between chute A2 and the concrete increases, causing the solidified concrete to slip down chute A2, which is thought to result in random fluctuations. Therefore, when each waveform is Fourier transformed, as shown in Figure 11(b), waveform G1 has a high peak in the low frequency range.

[0058] In the fourth learning process, the learning unit 113 calculates intensity average values ​​(first intensity average value, second intensity average value) in two predetermined frequency ranges from the frequency spectrum obtained by Fourier transforming the change in flow velocity over time using the fourth teacher information 224, and calculates the ratio between the first and second intensity average values ​​(intensity ratio).Then, a fourth prediction model that associates the intensity ratio with the slump is generated, and recorded in the learning result storage unit 14.

[0059] In the fourth prediction process, the prediction unit 114 calculates the intensity ratio of the first and second intensity average values ​​calculated from the frequency spectrum obtained by Fourier transforming the change in flow velocity over time, inputs this into the fourth prediction model, and predicts the workability (slump) by obtaining a fourth slump prediction value.

[0060] (Concrete monitoring method) Next, a concrete management method will be described with reference to Figures 12 and 13. The concrete management method includes a prediction model adjustment process and a concrete monitoring process.

[0061] As described above, the first to fourth learning processes are performed in advance using the teacher information storage unit 13 to generate the first to fourth prediction models, which are then recorded in the learning result storage unit . The processing terminal 10 is placed at the construction site C1 in a position where the field of view covers the area from the chute A2 of the transport vehicle A1 to the hopper P2 of the pump vehicle P1. The image capturing device 10A may be capable of automatically changing the magnification (zooming), panning, and tilting.

[0062] 12 shows the procedure for the prediction model adjustment process. The control unit 11 executes a process for acquiring slump test results (step S10). Specifically, a worker acquires concrete transported by the transport vehicle A1 and performs a slump test to measure the slump (measured slump) of the concrete transported to the construction site. The worker then inputs the measured slump from the slump test into the processing terminal 10. In this case, the control unit 11 stores the measured slump in a memory or the like.

[0063] The image acquisition unit 111 executes a process for acquiring a pre-captured image (step S11). Specifically, the image acquisition unit 111 uses the photographing device 10A to acquire a photographed image of concrete flowing down from the chute A2 into the hopper P2. At this time, the recognition unit 112 detects the chute A2 using the captured image and sets a chute detection frame F1. The recognition unit 112 also sets a concrete detection frame F2 within the chute detection frame F1 and detects the concrete within the concrete detection frame F2.

[0064] Next, the prediction unit 114 performs a process of adjusting the prediction model (step S12). Specifically, the prediction unit 114 applies the first to fourth prediction models to the images within the concrete detection frame F2 among the acquired captured images, and predicts the first to fourth predicted slump values.

[0065] Next, the prediction unit 114 compares the actually measured slump with the first to fourth predicted slump values, and may select a prediction model that calculates a slump that is within a predetermined allowable range for the actually measured slump.

[0066] FIG. 13 shows the procedure of the concrete monitoring process. First, when the recognition unit 112 receives an operation to start work, it generates new work information 120. The image acquisition unit 111 acquires captured images from the start of work to the end of work. The transport vehicle A1 at the construction site C1 moves to the unloading position. As a result, the chute A2 enters the imaging range of the processing terminal 10.

[0067] Next, the recognition unit 112 performs a recognition process for the shot A2 on the captured image (step S21). The control unit 11 determines whether or not the shot A2 has been detected based on the result of the recognition process for the shot A2 (step S22). If the recognition unit 112 determines that the shot A2 has not been detected (step S22: NO), the process returns to step S21 and the recognition process for the shot A2 is repeated.

[0068] When the control unit 11 determines that the shot A2 has been detected (step S22: YES), it starts the time frame (step S23). Specifically, the recognition unit 112 records the start time of the time frame in the work information 120. In addition, the recognition unit 112 sets a shot detection frame F1 in the captured image.

[0069] Here, when concrete to be used for construction work is poured from the transport vehicle A1 into the pump vehicle P1, the image acquisition unit 111 acquires an unloading image of the concrete flowing down into the hopper P2. Specifically, the recognition unit 112 sets a concrete detection frame F2 within the chute detection frame F1 in the captured image and performs recognition processing of the concrete in the concrete detection frame F2.

[0070] When the recognition unit 112 detects concrete within the concrete detection frame F2, the prediction unit 114 executes a workability prediction process (step S24). The prediction unit 114 applies the acquired unloading image to the first to fourth prediction models to predict the first to fourth predicted slump values. The prediction unit 114 records the first to fourth predicted slump values ​​in association with the time frame started in step S23. Note that if a prediction model is selected in the prediction model adjustment process shown in FIG. 12, the selected prediction model may be applied to the unloading image to predict the predicted slump value.

[0071] Next, the determination unit 115 executes a determination process as to whether a warning is necessary (step S25). Specifically, the determination unit 115 performs a comprehensive evaluation of workability using the first to fourth predicted slump values. Here, the predicted slump value calculated by the prediction model selected in step S24 may be used. For example, if the slump calculated by the selected prediction model deviates from the management standard, it is determined that a warning is necessary. The management standard may be, for example, a single standard value or a standard range. Alternatively, the management standard may be, for example, a plurality of standard values.

[0072] If the determining unit 115 determines that a warning is not required (step S25: NO), the process returns to the workability prediction process using the unloading image (step S24). On the other hand, if the determination unit 115 determines that a warning is necessary (step S25: YES), it executes an alarm output process (step S26). Specifically, the determination unit 115 transmits an alarm message indicating that the management criteria have been exceeded to the administrator terminal 30. The determination unit 115 of the control unit 11 also outputs an alarm to at least one of the output devices H13, which are a display, a speaker, and a lighting device such as a patrol lamp. If multiple reference values ​​are set as the management criteria, the alarm may be changed for each reference value. For example, the color of the image, sound, or lighting light may be changed for each level of the reference value, or the number of devices for which the alarm is output may be increased.

[0073] The recognition unit 112 determines whether the chute A2 has disappeared (step S27). If the recognition unit 112 determines that the chute A2 continues to be recognized in the unloading image and that the chute A2 has not disappeared from the unloading image (step S27: NO), the process returns to step S24.

[0074] When the recognition unit 112 determines that the chute A2 has disappeared from the unloading image (step S27: YES), it determines that the time frame has ended (step S28). The recognition unit 112 records the end time of the time frame in the work information 120.

[0075] Thereafter, the recognition unit 112 returns to step S21 and performs the recognition process for a new shot A2. (Administrator terminal screen) Referring to FIG. 14, the management screen 31 output to the output device H13 of the manager terminal 30 will be described. The manager terminal 30 can be used at a location remote from the construction site C1. The management screen 31 includes a continuous display area 31A, display areas 31B and 31C, and a most recent value display area D. The continuous display area 31A displays the 5-second average predicted slump value of the workpiece. The display area 31B displays the average predicted slump value of one transport vehicle A1. The transport vehicle A1 to be displayed may be the most recent transport vehicle A1, and may be selectable on the management screen 31. The display area 31C displays the average predicted slump values ​​of three transport vehicles A1. The average predicted slump values ​​of the three transport vehicles A1 may be distinguished by different display modes such as color or shape. This allows the manager to grasp changes in the predicted slump value for each transport vehicle A1 and to perform detailed process management. The most recent value display area 31D displays the most recent predicted slump value.

[0076] (Effects of this embodiment) According to the above embodiment, the following effects can be obtained. (1) The processing terminal 10 determines the start and end of unloading based on the captured images. This allows workability evaluation to be performed even if the worker is not always at the unloading position. This reduces the labor required to evaluate concrete at the construction site C1. Furthermore, by determining the start and end of unloading, workability can be evaluated for each transport vehicle A1, allowing for strict process management.

[0077] (2) The processing terminal 10 determines that unloading has started when it detects the chute A2, which is a predetermined part, and determines that unloading has ended when it determines that the chute A2 has disappeared. This makes it possible to detect the approach and exit of the transport vehicle A1 to and from the unloading position.

[0078] (3) The processing terminal 10 is configured to automatically determine whether or not concrete is present on the chute A2. This prevents an abnormal value from being output when there is no concrete on the chute A2.

[0079] (4) When the workability exceeds the management standard, the processing terminal 10 sends an alarm, so that the manager can be instructed to make improvements. (5) If the processing terminal 10 cannot recognize the chute for a predetermined time or longer, it determines that the transport vehicle A1 has left the unloading position. In other words, if the time during which the chute A2 disappears is less than a predetermined time, for example, due to a temporary obstruction to the chute A2, it determines that unloading should continue. Therefore, it is possible to automatically determine the end of unloading even if there is no worker present to determine the on-site situation.

[0080] The above embodiment can be modified as follows: The above embodiment and the following modifications can be combined with each other within the scope of technical compatibility. (Workability) In the above embodiment, slump is used as the workability for evaluating the quality of concrete. However, workability is not limited to slump, and material segregation may also be used as the workability. Furthermore, slump flow may be used as an index of fluidity. In this case, a prediction model is generated by a learning process using, as training information, photographed images associated with the workability evaluation values ​​calculated from each test.

[0081] (Unloading start conditions and unloading end conditions) In the above embodiment, the unloading start condition is that the recognition unit 112 detects the chute A2. Alternatively or additionally, the unloading start condition may be that the recognition unit 112 detects a specific movement of a part of the transport vehicle A1 (e.g., the chute A2) or that the recognition unit 112 detects a specific part of the transport vehicle A1 (e.g., a license plate, an edge of the chute A2 other than the discharge port, etc.).

[0082] In addition, the condition for completing unloading may be that the recognition unit 112 detects a specific movement of a part such as the chute A2, or that the recognition unit 112 determines that a specific part of the transport vehicle A1 that it recognized (for example, the license plate) has disappeared from the captured image.

[0083] (Image Processing) In the above embodiment, the recognition unit 112 uses image processing to recognize the chute A2 as a specific part of the transport vehicle A1 for determining the start and end of unloading. Alternatively or in addition to this, the recognition unit 112 may recognize other parts of the transport vehicle A1 as long as the image capturing range of the image capturing device 10A can include flowing concrete. For example, the recognition unit 112 may recognize a drum, scoop, hopper, license plate, or a marking attached for image recognition.

[0084] In the above embodiment, an image of concrete being unloaded (a flow-down image) is captured as it flows from the chute A2 of the transport vehicle A1 into the hopper P2 of the pump vehicle P1. As long as the image is of concrete being unloaded from the chute A2 of the transport vehicle A1, the unloading destination is not limited to the hopper P2 of the pump vehicle P1.

[0085] (Prediction model) In the above embodiment, an image of the concrete detection frame F2 was used in the first prediction process, but images of other regions may also be used. Here, the prediction model to be used may be specified depending on the situation at the time of unloading. For example, different regions may be used depending on the time of day at the time of unloading. The way shadows caused by convex parts of the concrete surface texture appear differs between morning and evening and daytime. Therefore, a region that can emphasize the concaves and convexes may be selected depending on the time of day.

[0086] In the above embodiment, the first to fourth prediction models are generated by the first to fourth learning processes, and the first to fourth prediction processes are performed. The methods for predicting concrete workability are not limited to four types.

[0087] In the above embodiment, one of the first to fourth prediction models may be selected. However, if the model to be used is determined in advance, the prediction model adjustment process (S10 to S12) may be omitted.

[0088] In the above embodiment, in the learning process and prediction process of the first and second prediction models, images of the concrete flow status are processed by graying color images and equalizing histograms. The image processing method is not limited to graying and equalizing histograms, and any image processing method that can emphasize the surface condition of the concrete is sufficient.

[0089] In the above embodiment, in the process of adjusting the prediction model (step S12), a prediction model to be used for managing the concrete to be unloaded is selected. Here, the priority of the prediction models to be used may be determined. In this case, the determination unit 115 ranks the prediction models in order of proximity to the measured slump, from the first slump to the fourth slump, and assigns higher priority to them. Then, the determination unit 115 sets weighting values ​​according to the priority. In this case, higher weighting values ​​are assigned to models closer to the measured slump.

[0090] In the process of determining whether a warning is necessary (step S25), the determination unit 115 may calculate a comprehensive evaluation value of workability by multiplying the first to fourth slumps by weighting values. The determination unit 115 then compares the comprehensive evaluation value with a reference range. If the comprehensive evaluation value is outside the reference range, the determination unit 115 executes a warning process (step S26).

[0091] In the above embodiment, if at least one of the predicted slump values ​​calculated by the prediction model selected in step S24 falls outside the reference range, it is determined that a warning is necessary. The conditions for warning are not limited to this, and can be set appropriately by the administrator, etc. For example, it may be determined that a warning is necessary if all of the first to fourth predicted slump values ​​fall outside the reference range.

[0092] (Judgment Department) The judgment unit 115 may change the judgment method depending on the time of day. In this case, a judgment method table in which predicted slump values ​​to be used for judgment depending on the time of day are predetermined is stored in the learning result storage unit 14 or the like, and one or more predicted slump values ​​are used depending on the current time. This makes it possible to use a judgment method that makes it easy to predict workability depending on, for example, the solar radiation situation.

[0093] In the above embodiment, the determination unit 115 outputs a warning, but the prediction unit 114 may output a warning. In the above embodiment, the processing terminal 10 functions as the recognition unit 112, the learning unit 113, and the prediction unit 114. Alternatively, a support server connected to the processing terminal 10 and the administrator terminal 30 may function as at least one of the recognition unit 112, the learning unit 113, and the prediction unit 114. The support server, like the processing terminal 10, includes hardware such as a communication device H11, an input device H12, an output device H13, a storage unit H14, and a processor H15. For example, the support server may include a prediction result storage unit 12, a teacher information storage unit 13, and a learning result storage unit 14, function as the recognition unit 112, the learning unit 113, and the prediction unit 114, and execute each process of the processing terminal 10 described above. [Explanation of symbols]

[0094] A1...transport vehicle, A2...chute, P1...pump vehicle, P2...hopper, 10...on-site terminal, 11...control unit, 111...image acquisition unit, 112...recognition unit, 112...learning unit, 114...prediction unit, 115...determination unit, 12...prediction result storage unit, 13...teaching information storage unit, 14...learning result storage unit

Claims

1. an image acquisition unit that acquires a captured image acquired from an imaging device; a recognition unit that determines the start and end of unloading of concrete based on the captured image; A prediction unit that predicts the workability of the concrete, When the recognition unit determines that an unloading start condition is satisfied based on the captured image, the recognition unit records unloading start information in a storage unit; The prediction unit predicts the workability of the concrete to be unloaded after the start of unloading in the photographed image, records the prediction result in the storage unit in association with the unloading start information, and outputs information related to the workability, A concrete management system in which, when the recognition unit determines that the unloading completion condition has been met based on the captured image, the prediction result and unloading completion information are associated and recorded in the memory unit.

2. The recognition unit determines that the unloading start condition is satisfied when a predetermined part of the transport vehicle transporting the concrete is newly detected in the captured image, The concrete management system according to claim 1 , wherein the recognition unit determines that the unloading termination condition is satisfied when the detected part disappears from the captured image.

3. The concrete management system of claim 1, wherein the recognition unit predicts the workability when the concrete is newly detected in the captured image after the unloading start condition is met, and stops predicting the workability when the concrete is no longer detected in the captured image during the workability prediction.

4. 2. The concrete control system according to claim 1, wherein when the predicted workability exceeds a control standard, an alarm is sent to an administrator terminal.

5. 3. The concrete management system according to claim 2, wherein the recognition unit determines that the transport vehicle has moved from the unloading location when the recognized part cannot be recognized for a predetermined period of time or longer.

6. A method for managing concrete using a memory unit and a control unit, The control unit Acquire the captured image acquired from the imaging device; When it is determined that the unloading start condition is satisfied based on the photographed image, unloading start information is recorded in the storage unit; In the photographed image, the workability of the concrete to be unloaded after the start of unloading is predicted, and the prediction result is recorded in the storage unit in association with the unloading start information, and information regarding the workability is output; A concrete management method in which, when it is determined that the unloading completion condition is met based on the captured image, the prediction result and unloading completion information are associated and recorded in the memory unit.

7. A program for managing concrete using a memory unit and a control unit, The control unit Acquire the captured image acquired from the imaging device; When it is determined that the unloading start condition is satisfied based on the photographed image, unloading start information is recorded in the storage unit; In the photographed image, the workability of the concrete to be unloaded after the start of unloading is predicted, and the prediction result is recorded in the storage unit in association with the unloading start information, and information regarding the workability is output; A concrete management program that functions as a means for associating the prediction result with unloading completion information and recording it in the memory unit when it is determined that the unloading completion condition has been met based on the captured image.

Images