A digital processing system and method for road construction

Abstract

The present invention discloses a digital processing system and method for road construction, which belongs to the field of artificial intelligence technology. The present invention obtains real-time video data of a construction site, uses a convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data; generates a movement plan for movable obstacles, records the position of immovable obstacles, and updates the database; sends instructions to operators at the construction site according to the position and movement plan of the movable obstacles; if the movable obstacle cannot move on its own, uses Canny edge detection to identify the edge of the obstacle, extracts the coordinates of the edge points, and controls the automated equipment to remove the movable obstacle when the movable obstacle is within the operating range of the automated equipment; performs path planning, designs the best road construction route, and avoids immovable obstacles; uses a genetic algorithm to formulate a road construction plan, and dynamically adjusts the construction plan during the construction process.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a digital processing system and method applied to road construction. Background Art

[0002] Road construction is a key link in the process of modern urban construction and infrastructure improvement. However, traditional road construction methods often face multiple challenges, especially in terms of obstacle management, construction efficiency and safety at the construction site.

[0003] Traditional methods rely on manual identification and management of obstacles at construction sites, which is time-consuming and prone to errors. It is difficult for construction personnel to obtain on-site conditions in real time, resulting in delayed decision-making and affecting construction progress. Existing technologies are slow to respond to the dynamic changes of movable obstacles and fail to effectively achieve real-time tracking and management, making it difficult to adjust construction plans and increasing risks during construction. Traditional path planning methods fail to fully consider the distribution of immovable obstacles, and construction route designs are often poorly optimized, which may lead to unnecessary detours and affect construction efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide a digital processing system and method for road construction to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A digital processing method for road construction, the method comprising the following steps:

[0007] S100, obtaining real-time video data of the construction site through the camera at the road construction site, using a convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data; marking movable obstacles and tracking them in real time, using the Dijkstra algorithm to generate a movement plan, and for immovable obstacles, recording the location and updating the database;

[0008] S200, according to the position and movement plan of the movable obstacle, send instructions to the operator at the construction site to guide him to remove the movable obstacle; determine whether the movable obstacle cannot move on its own, if it cannot move on its own, use Canny edge detection to identify the edge of the obstacle, extract the coordinates of the edge point, and when the movable obstacle is within the operating range of the automation equipment, control the automation equipment to remove the movable obstacle;

[0009] S300, after the movable obstacles are removed, the Dijkstra algorithm is used to perform path planning based on the distribution of the immovable obstacles to design the best road construction route to avoid the immovable obstacles;

[0010] S400. According to the properties of the immovable obstacles, a genetic algorithm is used to formulate a road construction plan based on the optimal road construction route, including a timetable and resource allocation for road construction. During the construction process, the changes of obstacles and the construction progress are monitored in real time, and the construction plan is dynamically adjusted.

[0011] According to step S100, the real-time video data is converted into image frames, the frame rate is set to capture moving obstacles, and each frame of the image is preprocessed, including denoising, contrast enhancement and normalization processing; image samples containing various movable and immovable obstacles are collected, and the image samples cover different angles, lighting and environmental conditions, and the samples are labeled to identify movable obstacles and immovable obstacles;

[0012] A convolutional neural network architecture for image recognition is designed, including a convolutional layer, a pooling layer, and a fully connected layer, to extract image features; the convolutional neural network is trained using a labeled data set, and network parameters are adjusted to improve the model's accuracy in identifying obstacles; the trained convolutional neural network is applied to image frames to identify obstacles in the image, and the output results include the type, position, and size of the obstacle, and the types of obstacles include movable obstacles and immovable obstacles.

[0013] According to step S100, a unique identifier ID is generated for the movable obstacle mark, and real-time tracking is performed. By identifying the same ID in consecutive image frames, the dynamic position update of the movable obstacle is realized; the position of the movable obstacle is predicted and updated using a Kalman filter, and the state of each movable obstacle is represented by a vector: , where x is the current horizontal coordinate of the movable obstacle, indicating its horizontal position at the construction site, and y is the current vertical coordinate of the movable obstacle, indicating its vertical position at the construction site. is the horizontal velocity of the movable obstacle, indicating its moving rate in the x direction, is the vertical velocity of the movable obstacle, indicating its moving rate in the y direction;

[0014] The construction area is assumed to be a graph G=(V,E), where V is a set of vertices representing construction points, and E is a set of edges representing accessible paths. Weight represents the distance cost from vertex i to vertex j, Indicates that the edge connecting vertex i and vertex j constitutes part of the path; Dijkstra's algorithm is used to find the shortest path from the starting point s to each other point i, and the formula is as follows:

[0015] ;

[0016] in, is the currently known shortest distance from the starting point s to the vertex i, is the currently known shortest distance from starting point s to vertex j.

[0017] According to step S200, the content of the instruction includes: a target location, operation steps, time requirements and safety tips; wherein the target location specifies the current location of the movable obstacle and the safe area to which it needs to be removed, the operation steps describe the specific route of the movement plan in real time and display a specific route map, the time requirement includes the time limit for completing the movement plan, and the safety tips include the location of the immovable obstacle to remind the operator to avoid it; the instruction is sent to the relevant operator in real time through the mobile application;

[0018] Operators are required to confirm and provide feedback after receiving instructions and adjust the movement plan according to the on-site conditions; during operation, the positions of movable obstacles are monitored, and when new obstacles appear or the path is blocked, a new movement plan is generated and instructions are resent.

[0019] Use sensors to monitor the movement status of movable obstacles, including speed and position. If the speed is zero and the position remains unchanged for a certain period of time, it is preliminarily determined to be in a stationary state. Monitor the surrounding environment of movable obstacles at road construction sites. If there are no construction workers or machines around, it is determined that the obstacles cannot move on their own.

[0020] Determine the operating range of the automated equipment and express it in coordinates to form an operable area; compare the position of the movable obstacle that cannot move on its own with the operating range of the automated equipment. If the position of the obstacle is within the operating range, the condition is met;

[0021] Confirm that the obstacle cannot move on its own and is within the operating range, and generate a removal instruction, which includes the target location and required operating steps; send instructions to the automation equipment through the embedded controller, instructing it to perform specific actions, and the automation equipment executes the predetermined action, safely removes the obstacle along the specified path, and places it in a safe area.

[0022] The operating range of the automation equipment is defined as a polygonal area or a circular area. The specific shape depends on the working characteristics of the equipment. If a polygon is used, the operating range of the polygonal area is within the vertex coordinates. If a circle is used, the operating range CA of the circular area is defined by the center point C and the radius r. The coordinates of C are ;

[0023] Use Canny edge detection to identify the edge of the obstacle, extract the coordinates of the edge points, record all the identified edge points as a set, and form the outline of the obstacle; use the Jarvis algorithm to construct the minimum convex hull according to the coordinates of the edge points to obtain the polygonal representation of the obstacle; when the operating range is defined as a polygonal area, use the ray method to determine whether the point is inside the polygon, send a horizontal ray from the point to be detected, and count the number of intersections with the polygon edge; if the number of intersections is an odd number, the point is inside the polygon; if it is an even number, it is outside the polygon; when the operating range is defined as a circular area, determine whether each edge point is within the circular range, as defined below: ,in, are the edge point coordinates.

[0024] Before edge detection, Gaussian filtering is applied to the input image to reduce noise. The gradient amplitude and direction of each pixel are calculated to identify areas in the image with significant brightness changes. After determining the potential edge locations, the non-maximum suppression algorithm suppresses non-edge points along the gradient direction. Two thresholds are used to classify edges into strong edges, weak edges, and non-edges. Strong edges are retained, while weak edges are retained when connected to strong edges. The parts of weak edges connected to strong edges are determined as valid edges, thus forming an edge map. After completing Canny edge detection, the identified edge points are extracted and recorded as a set to form the outline of the obstacle.

[0025] Select the leftmost point from the edge point set as the starting point. Starting from the current point, check all other edge points and find the point that forms the smallest angle with the current point as the next edge point. Add the next point found to the convex hull and update the current point to the newly found point. Repeat the above search process until returning to the starting point to form a closed polygon. During the construction process, it is necessary to ensure that each new point is an edge point. After being processed by the Jarvis algorithm, the final generated polygon represents the outer contour of the obstacle.

[0026] According to step S300, a map of the construction site is constructed based on the collected obstacle information, the map includes the coordinates of all immovable obstacles and their boundaries, and the passable areas are marked, and the map is represented in a grid form; the Dijkstra algorithm is initialized, and a starting point and an ending point are selected, wherein the starting point is the construction starting position and the ending point is the construction target position; the distance values ​​of all nodes are initialized to infinity, the distance of the starting point is set to zero, and a priority queue is created to store unprocessed nodes; starting from the starting point, all adjacent nodes are visited, the distance from the starting point to each adjacent node is calculated, and its distance value is updated; for each adjacent node, if the node is an area of ​​an immovable obstacle, the node is skipped; after visiting each node, it is marked as processed to avoid repeated calculations until the ending point is found or all passable nodes are processed;

[0027] After finding the end point, use the saved path information to trace back from the end point to the starting point to construct the optimal path, which avoids all immovable obstacles; generate the final construction roadmap based on the traced path to clarify the direction and distance of each step.

[0028] According to step S400, a construction plan is formulated using a genetic algorithm based on the optimal road construction route and in combination with immovable obstacles. The construction plan includes the division of construction stages, the main tasks of each stage, resource requirements and time arrangements; the priorities and interdependencies of the construction tasks are analyzed and a timetable is formulated; the construction process is divided into several stages, and the start and end times of each stage are determined; according to the construction plan, human, material and equipment resources are allocated to determine the type and quantity of resources required for each stage; during the construction process, the progress and obstacle status of the construction site are monitored in real time; based on real-time monitoring data, the deviation of the construction progress from the plan is evaluated, and when obstacles change or construction is delayed, the causes are analyzed and adjustment plans are proposed.

[0029] Using a genetic algorithm, a set of initial populations of construction plans are randomly generated. Each plan includes the division of construction phases, main tasks, resource requirements and time arrangements. A fitness function is designed to evaluate the pros and cons of each plan based on factors such as the efficiency of the plan, resource utilization and construction time. Plans with excellent performance are selected for reproduction based on the fitness value to ensure the transmission of excellent genes. The selected plans are crossed to generate new plans that combine the advantages of different plans. Some parameters of the newly generated plans are randomly adjusted to increase diversity and avoid local optimality. Fitness evaluation, selection, crossover and mutation are repeated until the preset termination condition is reached.

[0030] Break down the entire construction project into several specific tasks, and clarify the content and objectives of each task. By analyzing the logical relationship between the tasks, determine which tasks must be completed first and which tasks can be carried out in parallel. Determine the priority based on the dependency and impact of the tasks to ensure that key tasks are handled first.

[0031] According to the nature of the task and construction requirements, the construction process is divided into several stages, each with a clear start and end time. Based on past experience or expert opinions, the time required for each task is estimated and allocated to the corresponding stage. The time schedule of all stages and their corresponding tasks is organized into a timetable to ensure the overall construction process proceeds in an orderly manner.

[0032] A digital processing system applied to road construction, comprising:

[0033] Obstacle recognition module: including: video data acquisition unit, obstacle recognition unit and obstacle tracking unit; wherein, the video data acquisition unit obtains real-time video data of the construction site through the camera at the road construction site, the obstacle recognition unit uses convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data, the obstacle tracking unit marks movable obstacles and tracks them in real time, uses Dijkstra algorithm to generate movement plans, and records the positions of immovable obstacles and updates the database;

[0034] Obstacle management module: including: an instruction sending unit, a motion capability judgment unit and an automation equipment control unit; wherein the instruction sending unit sends instructions to the operators at the construction site according to the position and movement plan of the movable obstacle, instructing them to remove the movable obstacle; the motion capability judgment unit judges whether the movable obstacle is unable to move on its own; if it is unable to move on its own, the automation equipment control unit uses Canny edge detection to identify the edge of the obstacle, extracts the coordinates of the edge points, and controls the automation equipment to remove the movable obstacle when the movable obstacle is within the operating range of the automation equipment;

[0035] Construction route design module: including: path planning unit; after the movable obstacles are removed, the path planning unit uses the Dijkstra algorithm to plan the path based on the distribution of immovable obstacles, designs the best road construction route, and avoids immovable obstacles;

[0036] Construction plan formulation module: includes: a construction plan formulation unit and a construction plan dynamic adjustment unit; wherein, the construction plan formulation unit uses an artificial intelligence algorithm to formulate a road construction plan based on the nature of immovable obstacles and the route of the optimal road construction route, including a road construction schedule and resource allocation; during the construction process, the construction plan dynamic adjustment unit monitors the changes in obstacles and the construction progress in real time and dynamically adjusts the construction plan.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The present invention can monitor the status of movable obstacles in real time and respond quickly to achieve control of automated equipment, ensuring flexibility and safety of the construction site.

[0039] 2. The present invention uses the Dijkstra algorithm for path planning and designs the optimal construction route according to the distribution of immovable obstacles, thereby avoiding unnecessary detours and improving construction efficiency.

[0040] 3. The present invention formulates a detailed construction plan through artificial intelligence algorithms, monitors the construction progress and obstacle changes in real time, and can dynamically adjust the construction plan to ensure efficient allocation and utilization of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the steps of a digital processing method applied to road construction of the present invention;

[0042] Figure 2 The present invention is a system structure diagram of a digital processing system applied to road construction. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution.

[0045] According to one embodiment of the present invention, Figure 1 A step diagram of a digital processing method applied to road construction is shown in the figure. A digital processing method applied to road construction comprises the following steps:

[0046] S100, obtaining real-time video data of the construction site through the camera at the road construction site, using a convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data; marking movable obstacles and tracking them in real time, using the Dijkstra algorithm to generate a movement plan, and for immovable obstacles, recording the location and updating the database;

[0047] S200, according to the position and movement plan of the movable obstacle, send instructions to the operator at the construction site to guide him to remove the movable obstacle; determine whether the movable obstacle cannot move on its own, if it cannot move on its own, use Canny edge detection to identify the edge of the obstacle, extract the coordinates of the edge point, and when the movable obstacle is within the operating range of the automation equipment, control the automation equipment to remove the movable obstacle;

[0048] S300, after the movable obstacles are removed, the Dijkstra algorithm is used to perform path planning based on the distribution of the immovable obstacles to design the best road construction route to avoid the immovable obstacles;

[0049] S400. According to the properties of the immovable obstacles, a genetic algorithm is used to formulate a road construction plan based on the optimal road construction route, including a timetable and resource allocation for road construction. During the construction process, the changes of obstacles and the construction progress are monitored in real time, and the construction plan is dynamically adjusted.

[0050] According to step S100, the real-time video data is converted into image frames, the frame rate is set to capture moving obstacles, and each frame of the image is preprocessed, including denoising, contrast enhancement and normalization processing; image samples containing various movable and immovable obstacles are collected, and the image samples cover different angles, lighting and environmental conditions, and the samples are labeled to identify movable obstacles and immovable obstacles;

[0051] A convolutional neural network architecture for image recognition is designed, including a convolutional layer, a pooling layer, and a fully connected layer, to extract image features; the convolutional neural network is trained using a labeled data set, and network parameters are adjusted to improve the model's accuracy in identifying obstacles; the trained convolutional neural network is applied to image frames to identify obstacles in the image, and the output results include the type, position, and size of the obstacle, and the types of obstacles include movable obstacles and immovable obstacles.

[0052] In this embodiment, real-time video data is obtained from the camera at the road construction site, and the video stream is converted into image frames. The frame rate is set to 30 frames per second to ensure that the moving obstacles at the construction site are captured. Diverse image samples are collected from different angles, lighting conditions and weather conditions at the construction site, covering different scenes such as day, night, sunny and rainy days. The samples include various types of movable obstacles, such as vehicles, pedestrians, and immovable obstacles, such as street lights and guardrails. The bounding box and type of the obstacles are manually annotated using annotation tools, and movable obstacles and immovable obstacles are classified. The annotation results include the category, location and size of the obstacle.

[0053] The labeled image samples are divided into a training set and a validation set, 80% for training and 20% for validation, with a total of 10,000 samples. The training process lasts for 50 epochs, and each epoch iteration completes the learning of the entire training set. The recognition accuracy of the model reaches 92% and 88% on the training set and validation set, respectively.

[0054] In the simulated experimental environment, the video duration is 10 hours. The total number of images is 1,080,000 frames, specifically 30 frames per second, for 10 hours. Movable obstacles: vehicles, workers, mechanical equipment, etc., a total of 500 different samples. Immovable obstacles: street lights, guardrails, temporary signs, etc., a total of 300 different samples. The recognition accuracy of movable obstacles is 93%, of which the recognition accuracy of vehicles is 95% and the recognition accuracy of workers is 90%. The recognition accuracy of immovable obstacles is 90%, of which the recognition accuracy of fixed facilities is higher.

[0055] According to step S100, a unique identifier ID is generated for the movable obstacle mark, and real-time tracking is performed. By identifying the same ID in consecutive image frames, the dynamic position update of the movable obstacle is realized; the position of the movable obstacle is predicted and updated using a Kalman filter, and the state of each movable obstacle is represented by a vector: , where x is the current horizontal coordinate of the movable obstacle, indicating its horizontal position at the construction site, and y is the current vertical coordinate of the movable obstacle, indicating its vertical position at the construction site. is the horizontal velocity of the movable obstacle, indicating its moving rate in the x direction, is the vertical velocity of the movable obstacle, indicating its moving rate in the y direction;

[0056] The construction area is assumed to be a graph G=(V,E), where V is a set of vertices representing construction points, and E is a set of edges representing accessible paths. Weight represents the distance cost from vertex i to vertex j, Indicates that the edge connecting vertex i and vertex j constitutes part of the path; Dijkstra's algorithm is used to find the shortest path from the starting point s to each other point i, and the formula is as follows:

[0057] ;

[0058] in, is the currently known shortest distance from the starting point s to the vertex i, is the currently known shortest distance from starting point s to vertex j.

[0059] In this example, a construction area is simulated, which contains 10 vertices and 15 edges, and some of the areas have immovable obstacles. The system identifies 5 movable obstacles, including 3 construction vehicles and 2 pedestrians, and each obstacle appears at different locations in the construction area at different times.

[0060] 5 movable obstacles were tracked continuously for a total of 5,000 frames. Tracking accuracy: 97%, with an error margin of less than 10 pixels. The Kalman filter was able to successfully predict the position and resume tracking when some obstacles were occluded or moving quickly. The shortest path planned by the system successfully avoided the immovable obstacles and found the best solution among 15 navigable paths. The path length was the shortest, and the construction vehicle successfully reached the target point according to the planned route, avoiding collisions and delays.

[0061] According to step S200, the content of the instruction includes: a target location, operation steps, time requirements and safety tips; wherein the target location specifies the current location of the movable obstacle and the safe area to which it needs to be removed, the operation steps describe the specific route of the movement plan in real time and display a specific route map, the time requirement includes the time limit for completing the movement plan, and the safety tips include the location of the immovable obstacle to remind the operator to avoid it; the instruction is sent to the relevant operator in real time through the mobile application;

[0062] Operators are required to confirm and provide feedback after receiving instructions and adjust the movement plan according to the on-site conditions; during operation, the positions of movable obstacles are monitored, and when new obstacles appear or the path is blocked, a new movement plan is generated and instructions are resent.

[0063] Use sensors to monitor the movement status of movable obstacles, including speed and position. If the speed is zero and the position remains unchanged for a certain period of time, it is preliminarily determined to be in a stationary state. Monitor the surrounding environment of movable obstacles at road construction sites. If there are no construction workers or machines around, it is determined that the obstacles cannot move on their own.

[0064] Determine the operating range of the automated equipment and express it in coordinates to form an operable area; compare the position of the movable obstacle that cannot move on its own with the operating range of the automated equipment. If the position of the obstacle is within the operating range, the condition is met;

[0065] Confirm that the obstacle cannot move on its own and is within the operating range, and generate a removal instruction, which includes the target location and required operating steps; send instructions to the automation equipment through the embedded controller, instructing it to perform specific actions, and the automation equipment executes the predetermined action, safely removes the obstacle along the specified path, and places it in a safe area.

[0066] The operating range of the automation equipment is defined as a polygonal area or a circular area. The specific shape depends on the working characteristics of the equipment. If a polygon is used, the operating range of the polygonal area is within the vertex coordinates. If a circle is used, the operating range CA of the circular area is defined by the center point C and the radius r. The coordinates of C are ; In this embodiment, the operating range of the device is a pentagon, which is composed of vertex coordinates, and the vertex set is: [(2,3), (5,1), (8,2), (7,6), (3,7)].

[0067] Use Canny edge detection to identify the edge of the obstacle, extract the coordinates of the edge points, record all the identified edge points as a set, and form the outline of the obstacle; use the Jarvis algorithm to construct the minimum convex hull according to the coordinates of the edge points to obtain the polygonal representation of the obstacle; when the operating range is defined as a polygonal area, use the ray method to determine whether the point is inside the polygon, send a horizontal ray from the point to be detected, and count the number of intersections with the polygon edge; if the number of intersections is an odd number, the point is inside the polygon; if it is an even number, it is outside the polygon; when the operating range is defined as a circular area, determine whether each edge point is within the circular range, as defined below: ,in, are the edge point coordinates.

[0068] The extracted edge point set is: [(1,1), (2,2), (4,1), (6,3), (5,5), (3,6)]. After being processed by the Jarvis algorithm, the minimum convex hull constructed is: [(1,1), (4,1), (6,3), (5,5), (3,6)], forming a closed polygon to represent the outline of the obstacle.

[0069] Before edge detection, a Gaussian filter is applied to the input image with a kernel size of 5x5 and a standard deviation of 1.4 to reduce noise. The gradient amplitude and direction of each pixel are calculated to identify areas with significant brightness changes in the image. After determining the potential edge locations, the non-maximum suppression algorithm suppresses non-edge points along the gradient direction. Two thresholds are used, with the high threshold set to 100 and the low threshold set to 50, to classify edges into strong edges, weak edges, and non-edges. Strong edges will be retained, while weak edges will be retained when connected to strong edges. The parts of weak edges connected to strong edges are determined as valid edges, thus forming an edge map. After completing Canny edge detection, the identified edge points will be extracted and recorded as a set to form the outline of the obstacle.

[0070] Select the leftmost point from the edge point set as the starting point. Starting from the current point, check all other edge points and find the point that forms the smallest angle with the current point as the next edge point. Add the next point found to the convex hull and update the current point to the newly found point. Repeat the above search process until returning to the starting point to form a closed polygon. During the construction process, it is necessary to ensure that each new point is an edge point. After being processed by the Jarvis algorithm, the final generated polygon represents the outer contour of the obstacle.

[0071] According to step S300, a map of the construction site is constructed based on the collected obstacle information, the map includes the coordinates of all immovable obstacles and their boundaries, and the passable areas are marked, and the map is represented in a grid form; the Dijkstra algorithm is initialized, and a starting point and an ending point are selected, wherein the starting point is the construction starting position and the ending point is the construction target position; the distance values ​​of all nodes are initialized to infinity, the distance of the starting point is set to zero, and a priority queue is created to store unprocessed nodes; starting from the starting point, all adjacent nodes are visited, the distance from the starting point to each adjacent node is calculated, and its distance value is updated; for each adjacent node, if the node is an area of ​​an immovable obstacle, the node is skipped; after visiting each node, it is marked as processed to avoid repeated calculations until the ending point is found or all passable nodes are processed;

[0072] After finding the end point, use the saved path information to trace back from the end point to the starting point to construct the optimal path, which avoids all immovable obstacles; generate the final construction roadmap based on the traced path to clarify the direction and distance of each step.

[0073] According to step S400, a construction plan is formulated using a genetic algorithm based on the optimal road construction route and in combination with immovable obstacles. The construction plan includes the division of construction stages, the main tasks of each stage, resource requirements and time arrangements; the priorities and interdependencies of the construction tasks are analyzed and a timetable is formulated; the construction process is divided into several stages, and the start and end times of each stage are determined; according to the construction plan, human, material and equipment resources are allocated to determine the type and quantity of resources required for each stage; during the construction process, the progress and obstacle status of the construction site are monitored in real time; based on real-time monitoring data, the deviation of the construction progress from the plan is evaluated, and when obstacles change or construction is delayed, the causes are analyzed and adjustment plans are proposed.

[0074] According to another embodiment of the present invention, Figure 2 As shown in the system structure diagram of a digital processing system applied to road construction, a digital processing system applied to road construction includes:

[0075] Obstacle recognition module: including: video data acquisition unit, obstacle recognition unit and obstacle tracking unit; wherein, the video data acquisition unit obtains real-time video data of the construction site through the camera at the road construction site, the obstacle recognition unit uses convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data, the obstacle tracking unit marks movable obstacles and tracks them in real time, uses Dijkstra algorithm to generate movement plans, and records the positions of immovable obstacles and updates the database;

[0076] Obstacle management module: including: an instruction sending unit, a motion capability judgment unit and an automation equipment control unit; wherein the instruction sending unit sends instructions to the operators at the construction site according to the position and movement plan of the movable obstacle, instructing them to remove the movable obstacle; the motion capability judgment unit judges whether the movable obstacle is unable to move on its own; if it is unable to move on its own, the automation equipment control unit uses Canny edge detection to identify the edge of the obstacle, extracts the coordinates of the edge points, and controls the automation equipment to remove the movable obstacle when the movable obstacle is within the operating range of the automation equipment;

[0077] Construction route design module: including: path planning unit; after the movable obstacles are removed, the path planning unit uses the Dijkstra algorithm to plan the path based on the distribution of immovable obstacles, designs the best road construction route, and avoids immovable obstacles;

[0078] Construction plan formulation module: includes: a construction plan formulation unit and a construction plan dynamic adjustment unit; wherein, the construction plan formulation unit uses an artificial intelligence algorithm to formulate a road construction plan based on the nature of immovable obstacles and the route of the optimal road construction route, including a road construction schedule and resource allocation; during the construction process, the construction plan dynamic adjustment unit monitors the changes in obstacles and the construction progress in real time and dynamically adjusts the construction plan.

[0079] In this embodiment, a construction schedule and resource allocation are formulated based on the construction route. Schedule: 3 days of construction on section AB and 2 days of construction on section BD. Resource allocation: Allocate corresponding equipment and manpower to each section. For example, section AB requires 3 excavators and 10 workers, and section BD requires 2 excavators and 8 workers. The progress and obstacles of the construction site are monitored in real time through sensors and cameras. For example, a new obstacle is detected at (6,6), which will affect the construction route. Use AI algorithms to dynamically adjust the construction route, avoid new obstacles, and rearrange resources and schedules.

[0080] The new construction route is: A->B->C->D. The adjusted construction schedule is: 2 days for section AB, 2 days for section BC, and 2 days for section CD. The new resource allocation plan is: 2 excavators are needed for section AB, 3 excavators are needed for section BC, and 2 excavators are needed for section CD.

[0081] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A digital processing method applied to road construction, characterized in that: The method comprises the following steps: S100, obtaining real-time video data of the construction site through the camera at the road construction site, using a convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data; marking movable obstacles and tracking them in real time, using the Dijkstra algorithm to generate a movement plan, and for immovable obstacles, recording the location and updating the database; According to step S100, a unique identifier ID is generated for the movable obstacle mark, and real-time tracking is performed. By identifying the same ID in consecutive image frames, the dynamic position update of the movable obstacle is realized; the position of the movable obstacle is predicted and updated using a Kalman filter, and the state of each movable obstacle is represented by a vector: , where x is the current horizontal coordinate of the movable obstacle, indicating its horizontal position at the construction site, and y is the current vertical coordinate of the movable obstacle, indicating its vertical position at the construction site. is the horizontal velocity of the movable obstacle, indicating its moving rate in the x direction, is the vertical velocity of the movable obstacle, indicating its moving rate in the y direction; The construction area is assumed to be a graph G=(V,E), where V is a set of vertices representing construction points, and E is a set of edges representing accessible paths. Weight represents the distance cost from vertex i to vertex j, Indicates that the edge connecting vertex i and vertex j constitutes part of the path; Dijkstra's algorithm is used to find the shortest path from the starting point s to each other point i, and the formula is as follows: ； in, is the currently known shortest distance from the starting point s to the vertex i, is the currently known shortest distance from the starting point s to the vertex j; S200, according to the position and movement plan of the movable obstacle, send instructions to the operator at the construction site to guide him to remove the movable obstacle; determine whether the movable obstacle cannot move on its own, if it cannot move on its own, use Canny edge detection to identify the edge of the obstacle, extract the coordinates of the edge point, and when the movable obstacle is within the operating range of the automation equipment, control the automation equipment to remove the movable obstacle; According to step S200, the content of the instruction includes: a target location, operation steps, time requirements and safety tips; wherein the target location specifies the current location of the movable obstacle and the safe area to which it needs to be removed, the operation steps describe the specific route of the movement plan in real time and display a specific route map, the time requirement includes the time limit for completing the movement plan, and the safety tips include the location of the immovable obstacle to remind the operator to avoid it; the instruction is sent to the relevant operator in real time through the mobile application; Operators are required to confirm and provide feedback after receiving instructions, and adjust the movement plan according to the on-site situation; during the operation, the position of movable obstacles is monitored, and when new obstacles appear or the path is blocked, a new movement plan is generated and instructions are resent; Use sensors to monitor the movement status of movable obstacles, including speed and position. If the speed is zero and the position remains unchanged, it is preliminarily determined to be stationary. Monitor the surrounding environment of movable obstacles at road construction sites. If there are no construction workers or machines around, it is determined that the obstacles cannot move on their own. Determine the operating range of the automated equipment and express it in coordinates to form an operable area; compare the position of the movable obstacle that cannot move on its own with the operating range of the automated equipment. If the position of the obstacle is within the operating range, the condition is met; Confirm that the obstacle cannot move on its own and is within the operating range, and generate a removal instruction, which includes the target location and the required operating steps; send instructions to the automation equipment through the embedded controller, instructing it to perform specific actions, and the automation equipment performs the predetermined actions, safely removes the obstacle along the specified path, and places it in a safe area; S300, after the movable obstacles are removed, the Dijkstra algorithm is used to perform path planning based on the distribution of the immovable obstacles to design the best road construction route to avoid the immovable obstacles; S400. According to the properties of the immovable obstacles, a genetic algorithm is used to formulate a road construction plan based on the optimal road construction route, including a timetable and resource allocation for road construction. During the construction process, the changes of obstacles and the construction progress are monitored in real time, and the construction plan is dynamically adjusted.

2. A digital processing method for road construction according to claim 1, characterized in that: According to step S100, the real-time video data is converted into image frames, the frame rate is set to capture moving obstacles, and each frame of the image is preprocessed, including denoising, contrast enhancement and normalization; Collect image samples containing various movable and immovable obstacles, covering different angles, lighting and environmental conditions, annotate the samples, and identify movable and immovable obstacles; Design a convolutional neural network architecture for image recognition, including convolutional layers, pooling layers, and fully connected layers to extract image features; The convolutional neural network is trained using a labeled data set, and the network parameters are adjusted to improve the model's accuracy in identifying obstacles. The trained convolutional neural network is applied to image frames to identify obstacles in the image, and the output results include the type, position and size of the obstacle, and the types of obstacles include movable obstacles and immovable obstacles.

3. A digital processing method for road construction according to claim 1, characterized in that: The operating range of the automation equipment is defined as a polygonal area or a circular area. The specific shape depends on the working characteristics of the equipment. If a polygon is used, the operating range of the polygonal area is within the vertex coordinates. If a circle is used, the operating range CA of the circular area is defined by the center point C and the radius r. The coordinates of C are ; Use Canny edge detection to identify the edge of the obstacle, extract the coordinates of the edge points, record all the identified edge points as a set, and form the outline of the obstacle; use the Jarvis algorithm to construct the minimum convex hull based on the coordinates of the edge points to obtain the polygonal representation of the obstacle; when the operating range is defined as a polygonal area, use the ray method to determine whether the point is within the polygon, send a horizontal ray from the point to be detected, and count the number of intersections with the polygon edge; If the number of intersection points is odd, the point is inside the polygon; if it is even, it is outside the polygon; when the operation range is defined as a circular area, determine whether each edge point is within the circular range, as defined below: ,in, are the edge point coordinates.

4. A digital processing method for road construction according to claim 3, characterized in that: Before edge detection, Gaussian filtering is applied to the input image to reduce noise. The gradient amplitude and direction of each pixel are calculated to identify areas in the image with significant brightness changes. After determining the potential edge locations, the non-maximum suppression algorithm suppresses non-edge points along the gradient direction. Two thresholds are used to classify edges into strong edges, weak edges, and non-edges. Strong edges are retained, while weak edges are retained when connected to strong edges. The parts of weak edges connected to strong edges are determined as valid edges, thus forming an edge map. After completing Canny edge detection, the identified edge points are extracted and recorded as a set to form the outline of the obstacle. Select the leftmost point from the edge point set as the starting point. Starting from the current point, check all other edge points and find the point that forms the smallest angle with the current point as the next edge point. Add the next point found to the convex hull and update the current point to the newly found point. Repeat the above search process until returning to the starting point to form a closed polygon. During the construction process, it is necessary to ensure that each new point is an edge point. After being processed by the Jarvis algorithm, the final generated polygon represents the outer contour of the obstacle.

5. The digital processing method for road construction according to claim 1, characterized in that: According to step S300, a map of the construction site is constructed based on the collected obstacle information, the map includes the coordinates of all immovable obstacles and their boundaries, and the passable areas are marked, and the map is represented in a grid form; the Dijkstra algorithm is initialized, and a starting point and an ending point are selected, wherein the starting point is the construction starting position and the ending point is the construction target position; the distance values ​​of all nodes are initialized to infinity, the distance of the starting point is set to zero, and a priority queue is created to store unprocessed nodes; starting from the starting point, all adjacent nodes are visited, the distance from the starting point to each adjacent node is calculated, and its distance value is updated; for each adjacent node, if the node is an area of ​​an immovable obstacle, the node is skipped; after visiting each node, it is marked as processed to avoid repeated calculations until the ending point is found or all passable nodes are processed; After finding the end point, use the saved path information to trace back from the end point to the starting point to construct the optimal path, which avoids all immovable obstacles; generate the final construction roadmap based on the traced path to clarify the direction and distance of each step.

6. A digital processing method for road construction according to claim 1, characterized in that: According to step S400, a construction plan is formulated using a genetic algorithm based on the optimal road construction route and in combination with immovable obstacles. The construction plan includes the division of construction stages, the main tasks of each stage, resource requirements and time arrangements; the priorities and interdependencies of the construction tasks are analyzed and a timetable is formulated; the construction process is divided into several stages, and the start and end time of each stage is determined; according to the construction plan, human, material and equipment resources are allocated to determine the type and quantity of resources required for each stage; during the construction process, the progress and obstacle status of the construction site are monitored in real time; Based on real-time monitoring data, the deviation between construction progress and plan is evaluated. When obstacles change or construction is delayed, the causes are analyzed and adjustment plans are proposed.

7. A digital processing system for road construction, using a digital processing method for road construction according to any one of claims 1 to 6, characterized in that: include: Obstacle recognition module: including: video data acquisition unit, obstacle recognition unit and obstacle tracking unit; wherein, the video data acquisition unit obtains real-time video data of the construction site through the camera at the road construction site, the obstacle recognition unit uses convolutional neural network to identify movable obstacles and immovable obstacles in the real-time video data, the obstacle tracking unit marks movable obstacles and tracks them in real time, uses Dijkstra algorithm to generate movement plans, and records the positions of immovable obstacles and updates the database; Obstacle management module: including: an instruction sending unit, a motion capability judgment unit and an automation equipment control unit; wherein the instruction sending unit sends instructions to the operators at the construction site according to the position and movement plan of the movable obstacle, instructing them to remove the movable obstacle; the motion capability judgment unit judges whether the movable obstacle is unable to move on its own; if it is unable to move on its own, the automation equipment control unit uses Canny edge detection to identify the edge of the obstacle, extracts the coordinates of the edge points, and controls the automation equipment to remove the movable obstacle when the movable obstacle is within the operating range of the automation equipment; Construction route design module: including: path planning unit; after the movable obstacles are removed, the path planning unit uses the Dijkstra algorithm to plan the path based on the distribution of immovable obstacles, designs the best road construction route, and avoids immovable obstacles; Construction plan formulation module: includes: a construction plan formulation unit and a construction plan dynamic adjustment unit; wherein, the construction plan formulation unit uses an artificial intelligence algorithm to formulate a road construction plan based on the nature of immovable obstacles and the route of the optimal road construction route, including a road construction schedule and resource allocation; during the construction process, the construction plan dynamic adjustment unit monitors the changes in obstacles and the construction progress in real time and dynamically adjusts the construction plan.

Images