Lane marking construction vehicle, construction method and construction management program

Abstract

To provide a lane marking construction vehicle capable of forming lane markings on road surfaces with sufficient accuracy while responding to speeding-up of lane marking construction.SOLUTION: In a lane marking construction vehicle according to the present invention, a control part 21 constituting a control device 17 executes processing for extracting existing lane markings by using camera images captured by a forward cameras 15 and a downward camera 16 to adjust a paint injecting position based on lane marking extraction results, and processing for instructing injection start / stop timing to a coating device 11, a bead spraying device 13, and a release agent spraying device 14. In the processing for extracting the existing lane markings, an AI image generation model is used to generate an AI image that extracts only a white line portion of the lane markings from a composite image that includes unnecessary information other than the lane markings, and based on the AI image, unnecessary information other than the lane markings is deleted from the composite image that includes unnecessary information other than the lane markings.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a marking construction vehicle, construction method, and construction management program for constructing new markings over existing markings. [Background technology]

[0002] Conventionally, markings on road surfaces have been carried out using vehicles (trucks, etc.) equipped with an applicator for applying marking paint. Specifically, markings are applied by moving the vehicle while spraying molten paint heated to approximately 200°C from an applicator with a spraying mechanism onto the road surface.

[0003] In this case, in the conventional construction of markings, the vehicle travel speed is about 10 km / h at the fastest, and the performance of the coating device is 1 m when traveling at 10 km / h. 2 With a spray capacity of about 3 kg per unit area, it is possible to form markings on the road surface with sufficient precision (uniform and sufficient coating thickness, sufficient clarity of the start and end positions of the markings, etc.).

[0004] For example, Patent Document 1 listed below discloses a road marking construction vehicle that paints marking lines while adjusting the position of a spray gun by processing camera images, as an example of a vehicle that can form marking lines on the road surface with sufficient accuracy.

[0005] Specifically, the road marking construction vehicle described in Patent Document 1 below is equipped with a painting device (equivalent to the above-mentioned application device) having a spray gun that dispenses paint, a first camera that images the area in front of the vehicle, a second camera that images the road surface in front of the spray gun, a third camera that images the road surface behind the spray gun, and a control unit.

[0006] In the road marking construction vehicle configured as described above, the control unit combines the images captured by each camera, generates a virtual line based on the target line in the combined image, and outputs a correction signal based on the relative position of the virtual line and the vehicle body in the vehicle width direction. Then, in this road marking construction vehicle, the position of the spray gun (painting device) is adjusted based on the correction signal, and paint is sprayed from the spray gun.

[0007] Furthermore, the control unit detects the start and end positions of the existing markings based on the images captured by the first and second cameras, and the painting device paints the new markings over the existing markings while switching between starting and stopping paint discharge based on the start and end positions detected by the control unit. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] JP 2019-56237 A (Patent 6976511) Summary of the Invention [Problem to be solved by the invention]

[0009] As described above, conventional lane marking construction uses a vehicle, but the vehicle is designed to travel at a slow speed of approximately 10 km / h at most. For example, Patent Document 1 discloses that an ideal course and a predicted course are superimposed on a forward image captured by a first camera and displayed on a display device, allowing the driver to recognize any deviation of the predicted course from the ideal course while driving, and then correct the course while viewing the display device to bring the predicted course closer to the ideal course. In other words, since the driver drives while visually checking the display device, it can be seen that the traveling speed of the road marking construction vehicle described in Patent Document 1 is designed to be slow enough to allow safe driving even while viewing the display device.

[0010] On the other hand, in recent years, there has been a demand for expressways and other roads to shorten construction times, quickly reopen construction sites to traffic, and return road conditions to their original state as quickly as possible. To achieve this, efforts are underway to develop high-speed marking construction technologies, for example, using high-speed vehicles. Specifically, there is a demand for marking construction vehicles that can construct markings (such as lane marks) with sufficient precision (uniform and sufficient paint thickness, sufficient clarity of the start and end positions of the markings, etc.) even when traveling at a speed of 50 km / h.

[0011] However, the road marking construction vehicle described in Patent Document 1 above depends on the driving technique of the driver and is not designed to construct road markings when traveling at high speeds, for example, when traveling at 50 km / h, lane markings.

[0012] Furthermore, in the construction using the road marking construction vehicle described in Patent Document 1, image processing is performed on the images captured by the camera, i.e., processing of the captured images is performed, but there is a possibility that misrecognition of the marking lines may occur, such as when something that is not actually a marking line is processed as a marking line due to the influence of dirt, scratches, unevenness, light reflection, shadows, etc. on the road surface that is captured in the image captured by the camera. Such misrecognition may result in the marking lines not being drawn correctly, and the accuracy of the completed markings may deteriorate.

[0013] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a marking line construction vehicle, construction method, and construction management program that can form marking lines on the road surface with sufficient accuracy while responding to the increased speed of marking line construction. [Means for solving the problem]

[0014] The marking construction vehicle of the present invention is equipped with an application device that sprays paint for marking construction onto the road surface, and a sliding mechanism that slides the application device in the vehicle width direction to adjust the paint spray position, and is a marking construction vehicle that travels along existing marking lines and overlays new marking lines on top of the existing marking lines, and is equipped with a camera that periodically captures images of the road surface within a predetermined imaging range, and a control device that has a control unit that executes control processing related to marking construction and a memory unit for storing various information obtained during the control processing.

[0015] The control unit also comprises an image processing unit that performs image processing to individually create a bird's-eye view of camera images captured periodically and stores the bird's-eye view road surface image in a memory unit; a construction control unit that extracts a first center line, which is the center line of information recognized as a white line portion, from the road surface image periodically generated by the image processing unit and stores the first center line in the memory unit; an AI image generation unit that generates an AI image in which only the white line portion of the lane marking line is extracted by inputting AI execution target images extracted at regular intervals from the road surface image generated by the image processing unit into a trained AI image generation model that has undergone machine learning using, as an explanatory variable, lane marking line images containing unnecessary information generated by performing the above-mentioned image processing on camera images capturing lane marking lines captured at various locations and as an objective variable, a teacher image that has annotated the lane marking line images to remove unnecessary information and extract the white line portion of the lane marking line; and an AI control unit that extracts a second center line, which is the center line in the width direction of the white line portion of the lane marking line, from the AI ​​image and stores the second center line in the memory unit.

[0016] Then, with the camera calibrated in advance so that the lane markings are captured in the center of the camera image when an ideal route is traveled along the lane markings, the construction control unit further reads second center lines stored in the memory unit, calculates an expected center line by connecting the second center lines, and defines a white line detection range by giving the expected center line a predetermined tolerance. Next, the construction control unit reads first center lines from the memory unit, extracts center lines within the white line detection range from the first center line, calculates the amount of deviation between the extracted center line and the center of the road surface image, calculates the amount of adjustment for the spray position based on this deviation, and further identifies the start and end positions of the lane markings based on the center lines within the white line detection range extracted above, and calculates the paint spray start and spray stop times based on the identified positions. Thereafter, a spray position adjustment signal indicating the adjustment amount of the spray position and a paint spray start trigger signal that triggers the start of paint spraying are generated, and these signals are sent to the slide mechanism and the application device in time for the paint spray start time, and a paint spray stop trigger signal that triggers the paint spray stop is generated and this signal is sent to the application device in time for the paint spray stop time.

[0017] That is, in the lane marking construction vehicle according to the present invention, the control unit extracts existing lane markings using camera images captured by the forward camera and the downward camera, adjusts the paint spraying position based on the lane marking extraction results, and instructs the applicator, bead sprayer, and stripper sprayer on the spray start / stop timing. In the process of extracting existing lane markings, an AI image generation model is used to generate an AI image that extracts only the white line portion of the lane markings from a composite image that includes unnecessary information other than the lane markings, and then, based on this AI image, removes unnecessary information other than the lane markings from the composite image that includes unnecessary information other than the lane markings.

[0018] This allows the process of adjusting the paint spray position and instructing the application device, bead spray device and stripping agent spray device on the spray start / stop timing to be performed based on information on the white line portion of the demarcation line, with unnecessary information other than the demarcation line removed.This means that demarcation lines can be formed on the road surface with sufficient precision (uniform and sufficient paint film thickness, sufficient clarity of the start and end positions of the demarcation line, etc.) while accommodating faster construction of demarcation lines.

[0019] In addition, the marking construction vehicle of the present invention further comprises a bead spraying device that sprays beads onto the paint sprayed onto the road surface by the applicator, and a stripping agent spraying device that sprays a specified amount of stripping agent toward the end position of the marking line, and the construction control unit further calculates the start time and stop time of bead and stripping agent spraying based on the start and end positions of the marking line identified above, generates a bead spray start trigger signal that triggers the start of bead spraying, and sends this signal to the bead spraying device in time for the bead spray start time, generates a bead spray stop trigger signal that triggers the stop of bead spraying, and sends this signal to the bead spraying device in time for the bead spray stop time, and generates a stripping agent spray start trigger signal that triggers the start of stripping agent spraying, and sends this signal to the stripping agent spraying device in time for the stripping agent spray start time.

[0020] In addition, in the marking construction vehicle of the present invention, the bead spraying device and the stripping agent spraying device are held slidably in conjunction with the application device by a slide mechanism, and it is desirable that they be arranged in the following order from the front of the vehicle: stripping agent spraying device, application device, bead spraying device.

[0021] In addition, in the lane marking construction vehicle according to the present invention, the camera is configured with at least one camera, and if multiple cameras are configured, they are installed so that their imaging areas partially overlap, and all the cameras capture images simultaneously to capture the imaging range, and the image processing unit further performs image processing to combine the images captured by each camera simultaneously.The imaging range is preferably set appropriately based on the grace time required from the image of the road surface until the start or stop of paint spraying.

[0022] In addition, the construction method of the present invention is a construction method using a marking construction vehicle that includes an applicator that sprays paint for marking construction onto the road surface, a slide mechanism that slides the applicator in the vehicle width direction to adjust the paint spray position, a camera that periodically captures images of the road surface within a predetermined imaging range, and a control device that has a control unit that executes control processing related to marking construction and a memory unit for storing various information obtained during the control processing, and that travels along existing marking lines while overlaying the existing markings with new markings.

[0023] Then, in a state where the camera has been calibrated in advance so that the lane markings are captured in the center of the camera image when an ideal route is traveled along the lane markings, the control unit performs image processing to individually capture bird's-eye views of the camera images periodically taken and stores the bird's-eye view road surface image in a storage unit; a first center line extraction step to extract a first center line that is the center line of information recognized as a white line portion from the road surface image periodically generated in the image processing step and store the first center line in a storage unit; an AI image generation step to generate an AI image in which only the white line portion of the lane markings is extracted by inputting target images for AI execution extracted at regular intervals from the road surface image generated in the image processing step into a trained AI image generation model that has undergone machine learning using, as explanatory variables, lane marking images containing unnecessary information generated by performing the image processing on camera images capturing lane markings captured at various locations and training images in which unnecessary information has been deleted by annotating the lane marking images and the white line portion of the lane markings has been extracted; and a second center line extraction step of extracting a second center line that is a center line and storing the second center line in a storage unit; a detection range definition step of reading out the second center lines stored in the storage unit, determining an expected center line by connecting the second center lines, and defining a white line detection range that has a predetermined tolerance for the expected center line; an adjustment amount calculation step of reading out the first center line from the storage unit, extracting center lines within the white line detection range from the first center line, calculating the amount of deviation between the extracted center line and the center of the road surface image, and calculating the amount of adjustment of the injection position based on the amount of deviation; a time calculation step of identifying the start and end positions of the demarcation line based on the center line, and calculating the paint spray start and spray stop times based on the identified positions; a signal processing step of generating a spray position adjustment signal indicating the amount of adjustment of the spray position and a paint spray start trigger signal that triggers the start of paint spraying, and transmitting these signals to the slide mechanism and the coating device in time for the paint spray start time, and also generating a paint spray stop trigger signal that triggers the stop of paint spraying, and transmitting this signal to the coating device in time for the paint spray stop time;The method is characterized by carrying out the following steps.

[0024] In addition, in the construction method of the present invention, the marking construction vehicle further comprises a bead spraying device that sprays beads onto the paint sprayed onto the road surface by the applicator, and a stripping agent spraying device that sprays a specified amount of stripping agent toward the end position of the marking line, and in the time calculation step, the start time and stop time of bead and stripping agent spraying are calculated based on the identified start and end positions of the marking line, and in the signal processing step, a bead spray start trigger signal that triggers the start of bead spraying is generated and sent to the bead spraying device in time for the bead spray start time, a bead spray stop trigger signal that triggers the stop of bead spraying is generated and sent to the bead spraying device in time for the bead spray stop time, and a stripping agent spray start trigger signal that triggers the start of stripping agent spraying is generated and sent to the stripping agent spraying device in time for the stripping agent spray start time.

[0025] In addition, in the construction method of the present invention, it is desirable that the bead spraying device and the stripping agent spraying device be held slidably in conjunction with the applicator device by a slide mechanism, and that they be arranged in the following order from the front of the vehicle: stripping agent spraying device, applicator device, bead spraying device.

[0026] In the construction method according to the present invention, the camera is configured with at least one camera, and if multiple cameras are configured, they are installed so that their imaging areas partially overlap, and all the cameras simultaneously capture images of the imaging range, and the image processing step further performs image processing to combine the images captured by the cameras simultaneously.The imaging range is preferably set appropriately based on the grace time required from the image capture of the road surface until the start or stop of paint spraying.

[0027] The construction management program of the present invention is a construction management program for a marking construction vehicle that includes an applicator that sprays marking paint onto a road surface, a slide mechanism that slides the applicator in the vehicle width direction to adjust the paint spray position, a camera that periodically captures images of the road surface within a preset imaging range, and a control device that has a control unit that executes control processing related to marking construction and a memory unit for storing various information obtained during the control processing, and that overpaints new markings over existing markings while traveling along the existing markings.The program is characterized by causing a computer operating as the control unit to execute a series of processes including each of the above-mentioned steps. [Effects of the Invention]

[0028] According to the present invention, it is possible to provide a marking construction vehicle, construction method, and construction management program that can form marking lines on a road surface with sufficient accuracy while responding to the increased speed of marking construction. [Brief explanation of the drawings]

[0029] [Figure 1] FIG. 1 is a schematic diagram showing the general configuration of a lane marking construction vehicle according to the present invention. [Figure 2] FIG. 2 is a diagram illustrating an example of the hardware configuration of a computer that operates as a control device. [Figure 3] FIG. 3 is a functional block diagram showing the functions of the control unit that executes the construction management program. [Figure 4] FIG. 4 is a diagram illustrating an example of the hardware configuration of a computer that operates as a learning model generation system. [Figure 5] FIG. 5 is a diagram showing an example of a composite image. [Figure 6] FIG. 6 is a diagram showing an example of a teacher image. [Figure 7] FIG. 7 is a diagram showing the evaluation (learning curve) of a model based on machine learning. [Figure 8] FIG. 8 is a flowchart showing an example of the construction control process. [Figure 9]FIG. 9 is a diagram illustrating an example of a composite image. [Figure 10] FIG. 10 is a diagram showing an example of an input (synthetic image) to an AI image generation model and an output (AI image) from the AI ​​image generation model. DETAILED DESCRIPTION OF THE INVENTION

[0030] Hereinafter, embodiments of a lane marking construction vehicle, a construction method, and a construction management program according to the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to these embodiments.

[0031] <Configuration> 1 is a schematic diagram showing the overall configuration of a lane marking construction vehicle according to this embodiment, with (a) being a schematic diagram of the lane marking construction vehicle as seen from above, and (b) being a schematic diagram of the lane marking construction vehicle as seen from the right side. The lane marking construction vehicle 1 used in this embodiment is a lane marking construction vehicle that travels along existing lane marks and paints new lane marks over the existing lane marks, and FIG. 1 is a schematic diagram showing the components involved in lane marking construction.

[0032] In Figure 1, the coating device 11 mounted on the marking construction vehicle 1 is a device that sprays paint for marking construction onto the road surface, and is equipped with a spraying mechanism (not shown) that sprays the supplied paint toward the road surface.

[0033] The applicator 11 has a structure in which an opening 11a for supplying paint is provided at the top and a paint nozzle 11b at the bottom, the opening having dimensions approximately equal to the width of the demarcation line (for example, about 150 mm), and is held by a slide mechanism 12 installed on the loading platform so that the paint nozzle 11b is on the outside of the loading platform (so that it protrudes outward from the right side of the demarcation line construction vehicle 1) (see Figure 1(a)). This slide mechanism 12 is arranged so that it can move back and forth in the horizontal direction (vehicle width direction) perpendicular to the traveling direction of the demarcation line construction vehicle 1, and is used to slide the applicator 11 it holds in the vehicle width direction as needed to fine-tune the paint spray position.

[0034] Furthermore, in order to supply the paint to be sprayed onto the road surface to the applicator 11, the loading platform of the marking construction vehicle 1 is equipped with, for example, a melting pot, which is a tank for melting and storing the dissolved paint, and a paint supply device having a flow path for supplying the paint from the melting pot to the applicator 11 (see FIG. 1(a)), and furthermore, a distance measuring sensor 11c is installed on top of the applicator 11 to measure the distance to the liquid surface of the paint supplied to the applicator 11 (see FIG. 1(b)). The amount of paint supplied from the melting pot to the applicator 11 (supply amount) is controlled based on the distance measured by the distance measuring sensor 11c.

[0035] The melting pot is fixed in a mounted state on the loading platform of the marking construction vehicle 1, and has a paint outlet on its side. The paint supply device has a structure in which one side of the paint flow path is connected to the paint outlet of the melting pot, and the other side (tip side) of the flow path is located directly above the opening 11a provided in the coating device 11, and the paint flowing through the flow path in the paint supply device along the slope flows down from the tip toward the opening 11a of the coating device 11.

[0036] In addition, the loading platform of the marking construction vehicle 1 is loaded with, for example, a generator that serves as the power source for each of the installed devices, a compressor that supplies compressed air, an LPG cylinder that serves as a heat source for the melting furnace, etc. (see Figure 1).

[0037] The marking construction vehicle 1 is also equipped with a bead sprayer 13 that sprays (sprays) beads supplied from a bead tank (not shown) onto the paint sprayed onto the road surface by the applicator 11, and a spray gun-type stripper sprayer 14 that momentarily sprays (sprays) a specified amount of stripper toward the end of the marking. In this embodiment, the stripper sprayer 14 is used to clean up any dripping paint after spraying by the applicator 11 stops (beyond the end of the marking). The positions of the bead sprayer 13 and the stripper sprayer 14 are adjusted in conjunction with the applicator 11 by the sliding action of the slide mechanism 12. As an example, the slide mechanism 12 is arranged with the stripper sprayer 14, applicator 11, and bead sprayer 13 in this order from the front of the vehicle.

[0038] The lane marking construction vehicle 1 also includes a front camera 15 that captures images of the area in front of the vehicle body, a lower camera 16 that captures images of the lower right side of the vehicle body, and a control device 17 that processes (images) the camera images captured simultaneously and periodically by the front camera 15 and the lower camera 16. The front camera 15 is fixed at a height above the driver's seat with its lens facing diagonally downward so that it can capture images of lane markings painted on the road surface ahead. The lower camera 16 is installed above the front end of the loading platform and is fixed with its lens facing downward so that it can capture images of lane markings painted on the road surface ahead of the paint spraying position. Note that the camera images captured by the front camera 15 and the lower camera 16 contain information (unnecessary information other than lane markings) that may cause the images to be mistaken for lane markings, such as dirt, scratches, unevenness, light reflections, and shadows on the road surface. The locations at which each camera is installed are not limited to those described above; they may be installed anywhere on the lane marking construction vehicle 1 as long as they can capture the desired image area.

[0039] The forward camera 15 and the downward camera 16 are installed so that their imaging areas partially overlap (see FIG. 1(b)). The imaging ranges of the forward camera 15 and the downward camera 16 are appropriately set, taking into consideration the grace period (e.g., 0.6 seconds) required from capturing an image of the road surface (existing markings) until paint spraying begins / stops. That is, the imaging range is set based on the sum of the processing time (e.g., 0.4 seconds) required from capturing an image to transmitting a control signal (e.g., a paint spray start / stop trigger signal) and the delay time (e.g., 0.2 seconds) due to equipment, etc., from transmitting the control signal until paint spraying begins / stops. In this embodiment, as an example, the imaging range is set to approximately 9 meters forward from the applicator 11, taking into consideration the travel distance (8.34 meters) during the grace period when traveling at 50 km / h (=13.9 meters / second). Note that in this embodiment, as an example, two cameras are used to simultaneously and periodically capture images of the road surface on which existing markings are painted. However, this is not limited to this. For example, it is sufficient that the set imaging range can be imaged, and imaging may be performed periodically with one camera, or imaging may be performed simultaneously and periodically with three or more cameras.

[0040] The control device 17 controls the construction of markings by the marking construction vehicle 1 of this embodiment. Specifically, it periodically receives camera images captured simultaneously by the forward camera 15 and the downward camera 16, and performs image processing to combine these camera images each time they are received, extraction processing to extract existing markings from the combined image obtained by image processing, adjustment processing to fine-tune the paint spray position based on the extraction results, and signal processing to generate and transmit control signals related to marking construction (such as a spray position adjustment signal and a spray start / stop trigger signal, which will be described later) based on the extraction results.

[0041] That is, in the marking construction vehicle 1 of this embodiment, the control device 17 extracts existing markings using camera images captured by the forward camera 15 and the downward camera 16, adjusts the paint spray position based on the marking line extraction results (adjustment in the vehicle width direction using the slide mechanism 12), and further instructs the applicator 11, bead sprayer 13, and stripper sprayer 14 on the spray start / stop timing, thereby performing marking construction by painting new markings over the existing markings.

[0042] It should be noted that, for the sake of convenience of explanation, the various components (melting tank, etc.) mounted on the above-mentioned marking construction vehicle 1 are listed as appropriate to the configurations related to the characteristic operations of this embodiment, and do not describe all the configurations and functions required to carry out marking construction.

[0043] <Control device configuration> Next, the configuration of the control device 17 in the lane marking construction vehicle 1 of this embodiment will be specifically described. Figure 2 is a diagram showing an example of the hardware configuration of a computer that operates as the control device 17. In the lane marking construction vehicle 1 of this embodiment, the control device 17 operates as a host computer that performs various control processes related to lane marking construction (hereinafter referred to as construction control processes).

[0044] In FIG. 2, the control device 17 includes a control unit 21 configured with a CPU (Central Processing Unit) and an FPGA (Field Programmable Gate Array), a storage unit 22 including various memories such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and a receiving unit 23 and a transmitting unit 24 that operate as an I / F with an external device.

[0045] 2, the control unit 21 executes a construction management program for carrying out marking construction, for example, painting new markings over existing markings, in order to realize the construction control process by the control device 17 in the marking construction vehicle 1 of this embodiment. The memory unit 22 stores a program (construction management program) related to the construction control process of this embodiment, various information (information such as white line center lines and white line detection ranges, which will be described later), and various data obtained during the processing (image data such as camera images captured by each camera, composite images after image processing, and AI images, which will be described later). The control unit 21 reads out and executes the construction management program stored in the memory unit 22 to carry out the construction control process of this embodiment.

[0046] The storage unit 22 is not limited to an internal memory, but may be an external storage medium such as a DVD (Digital Versatile Disc) or SD memory, or may be configured with both an internal memory and an external storage medium (such as a DVD or SD memory). For ease of explanation, the hardware configuration of the control device 17 shown in FIG. 2 lists the configuration related to the construction control processing of this embodiment, and does not represent all the functions of the computer that constitutes the control device 17.

[0047] Furthermore, the control device 17 is assumed to be a dedicated computer as shown in FIG. 2, but is not limited to this and may be, for example, a general-purpose PC such as a desktop computer or a laptop computer, or may be a portable terminal such as a smartphone or a tablet terminal.

[0048] 3 is a functional block diagram showing the functions of the control unit 21 that executes the construction management program. In FIG. 3, the control unit 21 includes an image processing unit 31 that generates a composite image (a composite road surface image) by performing image processing to synthesize periodically received camera images (images simultaneously captured by the forward camera 15 and the downward camera 16) into a bird's-eye view; an AI image generation unit 32 that operates as an AI image generation model (performing AI image prediction processing) to generate an AI image that extracts only the white line portion of the lane marking from the composite image (an image that includes information that may cause the lane marking to be erroneously recognized, such as road surface dirt, scratches, unevenness, light reflection, and shadows); an AI control unit 33 that performs pre-processing and post-processing of the AI ​​image prediction processing by the AI ​​image generation model; and a construction control unit 34 that performs signal processing to adjust the paint spray position and control the start / stop of paint spraying. The detailed operation of each unit will be described later.

[0049] <Learning model generation process> Next, before explaining the construction control process by the control device 17 of the marking construction vehicle 1 of this embodiment, we will explain in detail the process of generating the learning model that is the premise for this process, i.e., the process of generating the above-mentioned AI image generation model.

[0050] 4 is a diagram showing an example of the hardware configuration of a computer that operates as the learning model generation system 41. This learning model generation system 41 operates as a host computer that performs a process (hereinafter referred to as a learning model generation process) that generates a learning model (AI image generation model) that performs the above-mentioned AI image prediction process.

[0051] 4, the learning model generation system 41 includes a control unit 51 configured with a CPU, FPGA, etc., a storage unit 52 including various memories such as ROM and RAM, an input unit 53 including a user interface such as a keyboard and mouse, an interface (I / F) unit 54 performing input / output processing such as printing and scanning, a display unit 55 as a display, and a communication unit 56 that communicates with the outside via a predetermined network. Note that, although FIG. 4 shows the learning model generation system 41 including the input unit 53 including a user interface such as a keyboard and mouse, the learning model generation system 41 used in this embodiment is not limited to this, and may be configured without the input unit 53 or in combination with the input unit 53 by providing the display unit 55 with a touch panel function.

[0052] The control unit 51 executes, for example, a learning model generation program for generating a learning model that performs the above-mentioned AI image prediction processing in order to realize the learning model generation processing by the learning model generation system 41. The storage unit 52 stores the learning model generation program and various information (such as a data set for machine learning) related to the learning model generation processing, as well as various data obtained during the processing. The control unit 51 performs the learning model generation processing by reading and executing the learning model generation program stored in the storage unit 52.

[0053] The storage unit 52 is not limited to an internal memory, but may be an external storage medium such as a DVD or SD memory, or may be configured with both an internal memory and an external storage medium (such as a DVD or SD memory). For ease of explanation, the hardware configuration of the learning model generation system 41 of this embodiment lists the configuration related to the learning model generation process, and does not represent all the functions of the computer that constitutes the learning model generation system 41.

[0054] Furthermore, the learning model generation system 41 of this embodiment is intended for general-purpose PCs such as desktop personal computers and laptop computers, but is not limited to these and may also be, for example, mobile terminals such as smartphones and tablet terminals, and may also be realized in the control device 17 described above.

[0055] 4, the control unit 51 generates, as an example of a learning model, an AI image generation model that predicts (generates) an AI image in which only the white line portion of a lane marking is extracted from a composite image (a composite road surface image) that includes information that may cause the lane marking to be erroneously recognized, such as dirt on the road surface, scratches, unevenness, light reflection, shadows, etc. (unnecessary information other than the lane marking.) This AI image generation model is generated using a known machine learning algorithm, such as a neural network.

[0056] In this embodiment, in order to perform machine learning, the vehicle 1 is driven on a highway with the front camera 15 and the lower camera 16 attached to it, and images of lane marks at, for example, 2,000 locations (camera images at 2,000 locations taken by the front camera 15 and camera images at 2,000 locations taken by the lower camera 16) are captured, and these 2,000 lane mark images are stored in the memory unit 52 of the learning model generation system 41 via the communication unit 56. Note that lane marks are continuous diagonal boundary lines with a white line portion of 8 m and a spacing of 12 m, drawn to separate driving lanes on a highway, and are one type of lane marking.

[0057] Furthermore, for each location, the control unit 51 reads out the lane mark image captured by the front camera 15 and the lane mark image captured by the lower camera 16 from the storage unit 52, combines them, and stores the combined image (combined lane mark image) generated for each location in the storage unit 52. During this combination, the control unit 51 uses known image processing to generate a combined image that is viewed from above as if it were captured from the sky.

[0058] Figure 5 shows an example of a composite image, and these 2,000 composite images serve as input data for machine learning. These composite images contain a lot of information that could cause the images to be misrecognized as lane markings, such as unnecessary information other than lane markings. They also contain lane markings that have faded due to wear.

[0059] The control unit 51 also performed annotation on the composite images of the 2,000 locations to generate training images (training data) for machine learning. Specifically, unnecessary information other than the lane markings was deleted from the composite images of the 2,000 locations, and the white line portions of the lane markings were extracted. FIG. 6 is a diagram showing an example of a training image. The control unit 51 then generated a dataset for machine learning by linking each composite image to a training image corresponding to the composite image. That is, the dataset used in the machine learning algorithm uses the composite images of the 2,000 locations as explanatory variables (input data) and training images corresponding to the composite images as objective variables (training data). The control unit 51 then stored the previously generated dataset in the storage unit 52.

[0060] Thereafter, the control unit 51 reads out from the storage unit 52 data sets for 1600 of the above-mentioned 2000 locations, and performs supervised learning using the data sets using a neural network, which is one of the machine learning algorithms, to generate an AI image generation model that predicts (performs AI image prediction processing) an AI image in which only the white line portions of lane marks are extracted from a composite image that includes information that may cause the lane marks to be erroneously recognized (unnecessary information other than lane marks, such as dirt on the road surface, scratches, unevenness, light reflection, shadows, etc.). That is, in this embodiment, a machine learning algorithm is used to learn the correlation between the composite images for each location and the teacher images that individually correspond to these composite images, thereby generating an AI image generation model that performs the above-mentioned AI image prediction processing.

[0061] In this embodiment, as described above, 2,000 datasets were prepared for machine learning. Of these, 1,600 were used as training data for generating an AI image generation model, 200 were used as evaluation data, and the remaining 200 were used as test data. Here, the training data is a dataset used to train the AI ​​image generation model, the evaluation data is a dataset used to verify the performance of the model, and the test data is a dataset used for final evaluation of the model by comparison with training images. Furthermore, the number of training data prepared is sufficient to achieve a desired prediction accuracy (e.g., an IOU (Intersection over Union) of 80% or more) without causing overfitting or underfitting.

[0062] FIG. 7 is a diagram showing the evaluation (learning curve) of a machine learning model, showing, for example, the transition of loss (loss) versus the degree of learning (epochs). As shown in FIG. 7, the AI ​​image generation model in this embodiment shows that the loss decreases as learning progresses, confirming that learning is progressing smoothly. Furthermore, since the loss becomes sufficiently small from about 20 epochs onwards and there is little change thereafter, it can be said that an epoch number of about 20 is suitable for learning an AI image generation model. Furthermore, when 200 items of test data were used to perform AI image prediction processing using the generated AI image generation model, a comparison was made between the AI ​​images obtained and the teacher images. As a result, an IOU of 80% or more was achieved, and good prediction results were obtained.

[0063] <An example of a machine learning algorithm> As described above, the machine learning algorithm used in this embodiment is a known neural network. Therefore, in supervised learning using a data set, the control unit 51 calculates the IOU between the prediction result (AI image) that is the output of the output layer of the neural network and the teacher image, and performs repeated learning for a predetermined number of epochs so as to minimize the loss.

[0064] In this embodiment, a known neural network is used as a machine learning algorithm to generate the AI ​​image generation model described above, but the machine learning algorithm for generating the AI ​​image generation model is not limited to this. For example, other machine learning algorithms such as random forests and boosted decision trees can also be used.

[0065] <Construction method> Next, the construction control process (construction method) by the control device 17 of the lane marking construction vehicle 1 of this embodiment will be described in detail. In this embodiment, as an example, it is assumed that the traveling speed of the lane marking construction vehicle 1 is 50 km / h and that lane markings are constructed on an expressway at this speed. Note that in this embodiment, it is assumed that the cameras have been pre-aligned (calibrated) so that the lane markings (white line portions) are captured in the center of the camera images when the lane marking construction vehicle 1 travels along an ideal route parallel to the lane marks.

[0066] 8 is a flowchart showing an example of the construction control process. In this embodiment, the control unit 21 constituting the control device 17 executes the construction management program to extract existing lane marks using camera images captured by the front camera 15 and the lower camera 16, adjust the paint spray position based on the lane mark extraction results (adjustment in the vehicle width direction by the slide mechanism 12), and instruct the applicator 11, the bead sprayer 13, and the stripper sprayer 14 on the spray start / stop timing.

[0067] Specifically, when the instruction to start construction is received (Step S1, No) while waiting for an instruction to start construction (Step S1, Yes), the control unit 21 first starts acquiring camera images captured by the forward camera 15 and the downward camera 16 via the receiving unit 23 (Step S2). These camera images are captured simultaneously and periodically (e.g., every 20 ms) by the forward camera 15 and the downward camera 16 mounted on the traveling lane marking construction vehicle 1, and are continuously captured until an instruction to stop construction is received (Step S3, Yes) or until an instruction to stop construction is received (Step S3, No). The image processing unit 31 of the control unit 21 then generates a composite image (composite road surface image) by combining and stitching together the camera images captured simultaneously by the forward camera 15 and the downward camera 16 (Step S4), and stores this composite image (road surface image) in the storage unit 22. During this composition, the image processing unit 31 uses known image processing to generate a bird's-eye view composite image, as if it were captured from above. FIG. 9 is a diagram showing an example of a composite image that is generated periodically (for example, every 20 ms).

[0068] Next, the construction control unit 34 of the control unit 21 reads the composite image from the memory unit 22 and extracts from the composite image the coordinate data of the start and end points of the white line portion of the lane mark on the right side in the width direction, the coordinate data of the start and end points of the white line portion of the lane mark on the left side in the width direction, and the lines on both sides (see FIG. 9 ). Note that the extracted information includes unnecessary information (such as road surface dirt) other than the lane marks that were erroneously recognized as lane marks. Then, from all the information extracted as the white line portion, the construction control unit 34 determines the white line center line, which is the center line of the lines on both sides (the center line of the lane mark in the width direction), as white line position information (step S5). This white line center line is stored (added) in the memory unit 22 as a first white line position history based on the composite image (step S6). Note that if the white line portion of the lane mark (including unnecessary information other than the lane mark) is not captured in the read composite image, such as when the area with a 12-meter white line interval is within the imaging range, the white line center line is not added.

[0069] Meanwhile, after generating the composite image by the processing of step S4, the AI ​​control unit 33 extracts composite images to be AI execution target images from the storage unit 22 as preprocessing for the AI ​​image prediction processing (step S7). In this embodiment, as an example, composite images generated periodically (for example, every 20 ms) are counted, and composite images to be AI execution target images are extracted at predetermined regular intervals (for example, count values: 1, 5, 9, 13, ...) (step S7, Yes). In other words, composite images corresponding to count values ​​other than the regular interval count values ​​are not selected as AI execution target images (step S7, No).

[0070] Then, the AI ​​control unit 33 inputs the composite image extracted as the image to be subjected to AI execution to the AI ​​image generation unit 32, which operates as an AI image generation model. The AI ​​image generation unit 32 accepts the input composite image, generates (predicts) an AI image using the AI ​​image generation model (step S8), and stores the generated AI image in the storage unit 22. That is, in this embodiment, the AI ​​image generation model is used to generate an AI image in which only the white line portions of the lane marks are extracted from a composite image that includes information that may cause the image to be erroneously recognized as a lane mark (such as dirt on the road surface, scratches, unevenness, light reflection, and shadows) (AI image prediction processing is performed). Figure 10 is a diagram showing an example of an input (synthetic image) to the AI ​​image generation model and an output (AI image) from the AI ​​image generation model.

[0071] Then, as post-processing of the AI ​​image prediction process, the AI ​​control unit 33 reads from the storage unit 22 the AI ​​image generated by the AI ​​image generation model, and extracts from this AI image the coordinate data of the start and end points of the right and left widthwise sides of the white line portion of the lane mark, as well as the lines on both sides. From this information, the AI ​​control unit 33 determines the white line center line, which is the center line of the lines on both sides (the center line of the lane mark in the widthwise direction), as white line position information (step S9), and stores this white line center line in the storage unit 22 as a second white line position history based on the AI ​​image (step S10). In this embodiment, between the extraction of one AI execution target image and the extraction of the next AI execution target image, the construction control unit 34 determines the white line center lines for four composite images, for example, every 20 ms (step S5), and stores them in the storage unit 22 as a first white line position history (step S6).

[0072] Next, the construction control unit 34 reads the second white line position history from the memory unit 22, calculates a predicted center line (approximate curve) by connecting the white line center lines calculated based on the AI ​​image, and then specifies a white line detection range that includes a margin (for example, an allowable width of about ± several centimeters) around the predicted center line (step S11). That is, the white line detection range is specified in order to remove white line center lines that have been erroneously calculated based on unnecessary information from the first white line position history, which is a white line center line calculated from a composite image that includes information that could cause the lane mark to be erroneously recognized (unnecessary information other than lane marks, such as dirt on the road surface, scratches, unevenness, light reflection, shadows, etc.).

[0073] Next, the construction control unit 34 reads out the first white line position history from the storage unit 22 and extracts the white line center line within the white line detection range (step S12). Then, the construction control unit 34 calculates the amount of deviation between the extracted white line center line and the center of the composite image, and calculates the amount of adjustment of the injection position, i.e., the amount of adjustment of the slide mechanism 12 (slide width in the vehicle width direction) based on this amount of deviation (step S13).

[0074] Meanwhile, the construction control unit 34 identifies the start and end positions of the lane mark based on the extracted white line center lines (white line center lines extracted every 20 ms), and calculates the spraying start times (paint spraying start time, bead spraying start time, and stripper spraying start time) and spraying stop times (paint spraying stop time, bead spraying stop time) for overpainting the new lane mark on top of the existing lane mark (step S14). In this embodiment, since the bead spraying device 13 is disposed behind the applicator 11 (see FIG. 1), beads are sprayed onto the paint after a time delay corresponding to the distance between the devices after the start of paint spraying. In this embodiment, since the applicator 11 is disposed behind the stripper spraying device 14 (see FIG. 1), paint spraying stops at the end position of the lane mark after a time delay corresponding to the distance between the devices after the stripper is sprayed at the end position of the lane mark.

[0075] Then, the construction control unit 34 generates a spray position adjustment signal indicating the adjustment amount of the slide mechanism 12, a first spray start trigger signal that triggers the application device 11 to start spraying paint, and a second spray start trigger signal that triggers the bead spraying device 13 to start spraying beads, and transmits each of the generated signals via the transmission unit 24 at a timing that is in time for the paint and bead spray start time (timing that takes into account the delay time due to equipment, etc. from the transmission of the control signal to the start / stop of spraying (for example, 0.2 seconds)) (step S15).

[0076] In addition, the construction control unit 34 generates a first spray stop trigger signal that triggers the coating device 11 to stop spraying paint, a second spray stop trigger signal that triggers the bead spraying device 13 to stop spraying beads, and a third spray start trigger signal that triggers the stripping agent spraying device 14 to start spraying stripping agent, and transmits each of the generated signals via the transmitting unit 24 at a timing that is in time for the spray stop time (paint and beads) / spray start time (stripping agent) (timing that takes into account the delay time due to equipment, etc. from the transmission of the control signal to the start / stop of spraying (for example, 0.2 seconds)) (step S15).

[0077] Then, the control unit 21 of the control device 17 continues to perform the above construction control process (construction method) until an instruction to stop construction is received (step S3, Yes).

[0078] In this embodiment, as an example, the traveling speed of the marking construction vehicle 1 is set to 50 km / h, and lane marking construction is performed on an expressway at this speed. However, this is not limited to this, and the construction control process can be similarly applied to other marking construction on expressways, public roads, etc., as long as imaging is performed within an appropriate imaging range according to the planned traveling speed and the above calibration is performed.

[0079] <Effects> As described above, in the marking construction vehicle 1 of this embodiment, the control unit 21 constituting the control device 17 extracts existing markings (such as lane marks) using camera images captured by the forward camera 15 and the downward camera 16, and then performs a process of adjusting the paint spray position (adjusting the vehicle width direction using the slide mechanism 12) based on the results of the marking line extraction, and further performs a process of instructing the applicator 11, the bead sprayer 13, and the stripper sprayer 14 on the spray start / stop timing. In the process of extracting existing markings, an AI image generation model is used to generate an AI image that extracts only the white line portions of the marking lines from a composite image that includes information that could cause the image to be erroneously recognized as a marking line (unnecessary information other than the marking lines, such as road surface dirt, scratches, unevenness, light reflection, and shadows), and then, based on this AI image, unnecessary information other than the marking lines is removed from the composite image (composite road surface image) that includes unnecessary information other than the marking lines.

[0080] This allows the process of adjusting the paint spray position and instructing the application device 11, bead spraying device 13 and stripping agent spraying device 14 on the spray start / stop timing to be performed based on information on the white line portion of the demarcation line, with unnecessary information other than the demarcation line deleted.This means that demarcation lines can be formed on the road surface with sufficient precision (uniform and sufficient paint film thickness, sufficient clarity of the start and end positions of the demarcation line, etc.) while accommodating faster construction of demarcation lines. [Explanation of symbols]

[0081] 1. Lane marking construction vehicle 11 Coating equipment 11a opening 11b spout 11c Distance sensor 12 Slide mechanism 13 Beads scattering device 14 Stripping agent spraying device 15. Front camera 16 Downward Camera 17 Control device 21 Control section 22 Memory section 23 Receiving unit 24 Transmitter 31 Image processing section 32 AI image generation unit 33 AI control section 34 Construction Control Department 41 Learning Model Generation System 51 Control section 52 Storage section 53 Input section 54 Interface (I / F) section 55 Display section 56 Communications Department

Claims

1. A marking construction vehicle is provided with an applicator that sprays marking paint onto a road surface, and a slide mechanism that slides the applicator in the vehicle width direction to adjust the paint spray position, and the vehicle travels along existing markings while overpainting new markings on top of the existing markings. a camera that periodically captures images of the road surface within a preset imaging range; A control device having a control unit that executes control processing related to demarcation line construction and a storage unit for storing various information obtained in the process of the control processing; Equipped with The control unit an image processing unit that performs image processing to individually obtain an overhead view of each camera image captured periodically, and stores the overhead view of the road surface image in the storage unit; a construction control unit that extracts a first center line, which is a center line of information recognized as a white line portion, from the road surface image periodically generated by the image processing unit and stores the first center line in the storage unit; an AI image generation unit that generates an AI image in which only the white line portion of the lane marking is extracted by inputting AI execution target images extracted at regular intervals from the road surface image generated by the image processing unit into a trained AI image generation model that has been machine-learned using, as explanatory variables, lane marking images containing unnecessary information generated by performing the image processing on camera images showing lane markings taken at various locations, and as objective variables, teacher images in which unnecessary information has been deleted by annotating the lane marking images and the white line portion of the lane marking is extracted; an AI control unit that extracts a second center line, which is a center line in a width direction of a white line portion of a lane marking, from the AI ​​image and stores the second center line in the storage unit; Equipped with In a state where the camera has been calibrated in advance so that the lane markings are captured at the center of the camera image when the vehicle is traveling along an ideal route along the lane markings, The construction control unit further reading out the second center lines stored in the storage unit, determining an expected center line by connecting the second center lines, and defining a white line detection range by providing a predetermined allowable width for the expected center line; a first center line is read from the storage unit, a center line within a white line detection range is extracted from the first center line, an amount of deviation between the extracted center line and the center of the road surface image is calculated, an amount of adjustment of the spray position is calculated based on this amount of deviation, and further, a start position and an end position of the marking line are identified based on the extracted center line within the white line detection range, and a paint spray start time and a paint spray stop time are calculated based on the identified positions; a spray position adjustment signal indicating the adjustment amount of the spray position and a paint spray start trigger signal serving as a trigger for starting paint spraying are generated, and these signals are transmitted to the slide mechanism and the coating device in time for the paint spray start time; and a paint spray stop trigger signal serving as a trigger for stopping paint spraying is generated, and this signal is transmitted to the coating device in time for the paint spray stop time. A lane marking construction vehicle characterized by:

2. Furthermore, a bead spraying device that sprays beads onto the paint sprayed onto the road surface by the coating device; a stripping agent spraying device that sprays a specified amount of stripping agent toward the end position of the demarcation line; Equipped with The construction control unit further Calculating the spray start time and spray stop time of the beads and the release agent based on the start and end positions of the specified division line; generating a bead spray start trigger signal that triggers the start of bead spraying, and transmitting this signal to the bead scattering device in time for the bead spray start time; Furthermore, a bead spraying stop trigger signal is generated as a trigger for stopping bead spraying, and this signal is transmitted to the bead scattering device in time for the bead spraying stop time. Furthermore, a stripper injection start trigger signal is generated as a trigger for starting the stripper injection, and this signal is transmitted to the stripper spraying device in time for the stripper injection start time.

2. The lane marking construction vehicle according to claim 1.

3. the bead spraying device and the release agent spraying device are held by the slide mechanism so as to be slidable in conjunction with the applicator device; The release agent spraying device, the application device, and the bead spraying device are arranged in this order from the front of the vehicle.

3. The lane marking construction vehicle according to claim 2.

4. The camera is configured with at least one In the case where the camera is configured with a plurality of cameras, the cameras are installed so that their imaging areas partially overlap, and all the cameras simultaneously capture images of the imaging range, and the image processing unit further performs image processing to combine the images captured by the cameras simultaneously, The imaging range is appropriately set based on the grace time required from when the road surface is imaged until the paint spraying starts or stops.

4. The lane marking construction vehicle according to claim 3.

5. A construction method using a demarcation line construction vehicle that includes an applicator that sprays paint for use in demarcation line construction onto a road surface, a slide mechanism that slides the applicator in the vehicle width direction to adjust the paint spray position, a camera that periodically captures images of the road surface within a preset imaging range, and a control device that has a control unit that executes control processing related to demarcation line construction and a memory unit for storing various information obtained in the process of the control processing, and that travels along existing demarcation lines while overpainting new demarcation lines over the existing demarcation lines, In a state where the camera has been calibrated in advance so that the lane markings are captured at the center of the camera image when the vehicle is traveling along an ideal route along the lane markings, The control unit an image processing step of performing image processing to individually obtain an overhead view of each camera image captured periodically, and storing the overhead view of the road surface image in the storage unit; a first center line extraction step of extracting a first center line, which is a center line of information recognized as a white line portion, from the road surface image periodically generated in the image processing step, and storing the first center line in the storage unit; an AI image generation step in which an AI execution target image extracted at regular intervals from the road surface image generated in the image processing step is input into a trained AI image generation model that has been machine-learned using, as explanatory variables, lane line images containing unnecessary information generated by performing the image processing on camera images showing lane lines captured at various locations, and as objective variables, teacher images in which unnecessary information has been deleted by annotating the lane line images and the white line portions of the lane lines have been extracted; and a second centerline extraction step of extracting a second centerline, which is a centerline in a width direction of a white line portion of a lane marking, from the AI ​​image and storing the second centerline in the storage unit; a detection range defining step of reading out the second center lines stored in the storage unit, calculating an estimated center line by connecting the second center lines, and defining a white line detection range that has a predetermined allowable width around the estimated center line; an adjustment amount calculation step of reading a first center line from the storage unit, extracting a center line within a white line detection range from the first center line, calculating a deviation amount between the extracted center line and the center of the road surface image, and calculating an adjustment amount of the injection position based on the deviation amount; a time calculation step of identifying the start and end positions of the marking line based on the center line within the extracted white line detection range, and calculating the paint spray start time and paint spray stop time based on the identified positions; a signal processing step of generating a spray position adjustment signal indicating an adjustment amount of the spray position and a paint spray start trigger signal which triggers the start of paint spraying, and transmitting these signals to the slide mechanism and the coating device in time for the paint spray start time, and also generating a paint spray stop trigger signal which triggers the paint spray stop time, and transmitting this signal to the coating device in time for the paint spray stop time; To execute A construction method characterized by:

6. The marking construction vehicle further includes a bead spraying device that sprays beads onto the paint sprayed onto the road surface by the coating device, and a stripping agent spraying device that sprays a specified amount of stripping agent toward the end position of the marking line, The time calculation step further comprises: Calculating the spray start time and spray stop time of the beads and the release agent based on the start and end positions of the specified division line; The signal processing step further comprises: generating a bead spray start trigger signal that triggers the start of bead spraying, and transmitting this signal to the bead scattering device in time for the bead spray start time; Furthermore, a bead spraying stop trigger signal is generated as a trigger for stopping bead spraying, and this signal is transmitted to the bead scattering device in time for the bead spraying stop time. Furthermore, a stripper injection start trigger signal is generated as a trigger for starting the stripper injection, and this signal is transmitted to the stripper spraying device in time for the stripper injection start time.

6. The construction method according to claim 5.

7. the bead spraying device and the release agent spraying device are held by the slide mechanism so as to be slidable in conjunction with the applicator device; The release agent spraying device, the application device, and the bead spraying device are arranged in this order from the front of the vehicle.

7. The construction method according to claim 6.

8. The camera is configured with at least one In the case where the camera is configured with a plurality of cameras, the cameras are installed so that their imaging areas partially overlap, and all the cameras simultaneously capture images of the imaging range, and the image processing step further performs image processing to combine the images captured by the cameras simultaneously, The imaging range is appropriately set based on the grace time required from when the road surface is imaged until the paint spraying starts or stops.

8. The construction method according to claim 7.

9. A construction management program for a demarcation line construction vehicle equipped with an applicator that sprays paint for use in demarcation line construction onto a road surface, a slide mechanism that slides the applicator in the vehicle width direction to adjust the paint spray position, a camera that periodically captures images of the road surface within a preset imaging range, and a control device that has a control unit that executes control processing related to demarcation line construction and a memory unit for storing various information obtained in the process of the control processing, and that paints new demarcation lines over existing demarcation lines while traveling along the existing demarcation lines, A computer operating as the control unit A construction management program for executing a series of processes including each step according to any one of claims 5 to 8.

Images