A path analysis and planning system for a tracking robot based on computational mathematics

Abstract

The application relates to the technical field of path planning, and discloses a path analysis and planning system of a tracing robot based on computational mathematics; an image splicing module is used for aligning and splicing sub-images into regional images; a matrix node mode is used for converting the regional images into matrix nodes based on a grid distinguishing criterion; a path planning module is used for planning a target path in the matrix nodes; a mode selection module is used for selecting a planning control mode according to a tracing state; and a planning control module is used for controlling a terminal to control the tracing robot to maintain current movement, adjust leftward or rightward, and re-plan a path. The application can discover abnormal obstacle phenomena in the first time when the tracing robot is affected or hindered by an obstacle object, and can control the tracing robot to perform targeted obstacle avoidance movement according to the severity of the influence and hindering of the actual obstacle object, so that the dynamic situation of the tracing robot during movement towards a destination can be dynamically detected in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of path planning, more particularly, the present application relates to a tracing robot path analysis planning system based on computational mathematics. BACKGROUND

[0002] The tracing robot is an important branch of mobile robots, which is widely used in warehouse logistics, workshop management and other scenes. Its core is to automatically identify and track the preset or dynamic path identifier, and realize autonomous movement from the starting point to the end point. Especially in the transfer of materials and products in the workshop, due to the random movement of personnel and equipment, the personnel and equipment are prone to interfere with the movement path of the inspection robot, resulting in the tracing robot unable to move according to the preset path. Therefore, it is necessary to analyze and plan the control and processing of the movement path of the tracing robot.

[0003] The patent application with publication number CN110716560A discloses a mobile robot path analysis planning method, which processes the road section, collects relevant data of the mobile robot, performs primary screening of the path, and performs secondary screening to obtain the optimal path. The optimal path can be obtained by analyzing the overall operation parameters and obstacle records of the mobile robot carrying goods, which can avoid the situation that the mobile robot carrying goods cannot run according to the planned route due to the volume of goods exceeding the standard of the planned road, and improve the rationality of route planning.

[0004] The existing path analysis planning system usually relies on the detection data of real-time sensors, so that the tracing robot cannot quickly plan a reliable global path before the task starts, and it is difficult to ensure that the tracing robot can reach the destination with the shortest movement distance. In addition, when the tracing robot encounters obstacles during movement, it cannot accurately analyze the impact of obstacles on the tracing robot in the first time, so it cannot reasonably formulate obstacle avoidance instructions matching the severity of the obstacle impact, resulting in low efficiency of adaptive obstacle avoidance operation of the tracing robot during real-time movement, and the tracing robot is prone to accidental collision with obstacles, which may interrupt the task, thereby reducing the real-time and accuracy of the path analysis control of the tracing robot.

[0005] In view of this, the present application provides a tracing robot path analysis planning system based on computational mathematics to solve the above problems. SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a tracing robot path analysis planning system based on computational mathematics, applied to a control terminal, comprising:

[0007] The image splicing module acquires sub-images of the planning area at the same time, splices the remaining sub-images after removing the overlapping area existing in the sub-images, and generates an area image.

[0008] The matrix node mode converts the area image into a two-dimensional grid map with nodes based on a grid division criterion, identifies the entity attribute of the grid in the planning area, and assigns values to the grid based on the entity attribute to generate a matrix node.

[0009] The path planning module marks the starting grid and the ending grid in the matrix node based on the criterion that one point coordinate corresponds to one grid, and plans the target path of the tracing robot in the matrix node in combination with the matrix optimization algorithm.

[0010] The mode selection module acquires the tracing video of the tracing robot on the target path in real time, analyzes the obstacles in the tracing video, determines the real-time tracing state of the tracing robot, and selects the corresponding planning control mode according to the tracing state.

[0011] The planning control module formulates the control instructions corresponding to the planning control mode to control the terminal to control the tracing robot to maintain the current movement, adjust left or right, and re-plan the path until the destination is reached.

[0012] Further, the splicing method of the area image is:

[0013] At the same time, A cameras are used to shoot images of the planning area from a top-down angle to obtain A sub-images, and an outer boundary line is generated after drawing a line along the outer edge of the sub-image.

[0014] Two diagonal lines of four corner points are drawn in the A sub-images respectively, and the intersection of the two diagonal lines is marked as a center point.

[0015] The length of the diagonal line in the sub-image is measured, 5% of the length of the diagonal line is marked as a calibration length, and four overlapping points are obtained by marking point positions at a distance of one calibration length from the adjacent corner point at both ends of the two diagonal lines.

[0016] The inner boundary line is generated by connecting the four overlapping points in sequence, and the area between the inner boundary line and the outer boundary line in the sub-image is marked as an overlapping area.

[0017] The overlapping area of the A sub-images is removed respectively, and the remaining A sub-images are one-to-one aligned and spliced with the planning area in space to generate an area image.

[0018] Further, the grid division criterion is that half of the projection width of the target road is used as the grid division standard.

[0019] The conversion method of the two-dimensional grid map is:

[0020] B roads for the tracing robot to pass through are identified in the area image through computer vision technology, and the projection values of the B roads along the width direction are measured one by one to obtain B projection widths;

[0021] The road corresponding to the minimum value of the projection width is recorded as a target road, and half of the projection width of the target road is taken as the grid length, and the area image is divided into C array distributed grids;

[0022] The center points of the C grids are marked one by one, and the inner circles are drawn in the C grids respectively with the grid length as the radius and the center points of the grids as the centers, and four points are marked on the inner circles at equal angles to obtain four nodes.

[0023] Further, the entity attribute includes an obstacle attribute and a passable attribute;

[0024] The generation method of the matrix node is:

[0025] The point coordinates of the four nodes in the C grids in the electronic map are queried one by one, and the object type at the position corresponding to the point coordinates in the planning area is queried;

[0026] The nodes with the object type as roads are recorded as target nodes, and the number of target nodes in the C grids is counted one by one to obtain a target value;

[0027] When the target value is 4, the entity attribute of the grid is the passable attribute; when the target value is not 4, the entity attribute of the grid is the obstacle attribute;

[0028] The grids with the entity attributes of the passable attribute and the obstacle attribute are respectively assigned values of 0 and 1, and the C grids are sequentially numbered in ascending order according to the way of row by row from top to bottom and left to right in the row, with 1 as the first number, to generate the matrix node with the number.

[0029] Further, the marking method of the starting grid and the ending grid is:

[0030] The real-time position of the tracing robot at the current time is queried through the positioning chip, and the real-time position of the tracing robot and the point coordinates of the destination on the electronic map are marked as the starting coordinates and the ending coordinates respectively;

[0031] The length and width of the electronic map are reduced or expanded in proportion to the standard length and standard width of the matrix node to be consistent with the size of the standard length and standard width;

[0032] The electronic map after reduction or expansion is rotated counterclockwise until the electronic map and the matrix node are aligned in spatial position to generate a target map.

[0033] The target map is overlaid on the matrix node, and the starting coordinate and the ending coordinate are recorded as the starting grid and the ending grid, respectively.

[0034] Further, the target path planning method is:

[0035] The horizontal direction is taken as the row, and the vertical direction is taken as the column, the matrix node is converted into a map matrix with i rows and y columns, the assignment of the C grids is transferred to the C elements of the map matrix, and the element located at the top left corner of the map matrix is recorded as the initial point;

[0036] The row value and the column value of the element corresponding to the starting grid and the ending grid in the map matrix are marked, respectively, and the starting value and the ending value are generated after the row value and the column value of the element are combined;

[0037] The cost of crossing an element is recorded as a unit cost, the actual distance between the initial point and the starting grid is measured, and is recorded as a cost matrix;

[0038] After comparing the row value and the column value of the ending value with the row value and the column value of the starting value, respectively, the estimated distance from the starting grid to the ending grid is calculated, and is recorded as E heuristic matrices;

[0039] The calculation formula of the heuristic matrix is:

[0040] ;

[0041] In the formula, is the heuristic matrix, is the row value of the ending value, is the row value of the starting value, is the column value of the ending value, is the column value of the starting value;

[0042] After adding the cost matrix and the E heuristic matrices one by one, the path span value is calculated, and the minimum value of the path span value is taken as the span amplitude of the target path, and the A* search algorithm is combined to automatically plan the target path in the matrix node.

[0043] Further, the tracing state includes a normal state, an influence state and an avoidance state:

[0044] The determination method of the tracing state is:

[0045] The tracing video of the tracing robot during the tracing period is obtained in real time, the tracing image at the last tracing time in the tracing period is intercepted from the tracing video, and the obstacle object in the tracing image is identified based on computer vision technology;

[0046] When there are no obstacles in the tracking image, the tracking state is determined to be the normal state;

[0047] When there are obstacles in the tracking image, the tracking image is divided into left, middle and right regions in the vertical direction, and the area values ​​of the left and right regions are adjusted until the area values ​​of the left and right regions are both one-sixth of the area value of the tracking image. Then, the obstacle region is generated by drawing a line along the outer edge of the obstacle.

[0048] If the entire obstacle area is located inside the left or right region, the tracking state is determined to be the affected state.

[0049] If the obstacle area is partially or entirely located inside the central area, the tracking state will be determined as an avoidance state.

[0050] Furthermore, the planning and control modes include path maintenance mode, path adjustment mode, and obstacle avoidance planning mode;

[0051] The method for selecting the planning control mode is as follows:

[0052] When the tracking robot is in a normal state, an affected state, or an obstacle avoidance state, select the path maintenance mode, path adjustment mode, or obstacle avoidance planning mode respectively.

[0053] Furthermore, the control commands include commands to maintain the current motion, commands to adjust left or right, and commands to replan the target path;

[0054] The method for formulating control commands is as follows:

[0055] When selecting path maintenance mode, path adjustment mode, or obstacle avoidance planning mode, commands to maintain the current movement, adjust to the left or right, or replan the path are generated respectively.

[0056] Furthermore, when an instruction to maintain the current motion is given, the tracking robot continues to move along the original target path;

[0057] When a command to adjust to the left or right is given, and the obstacle is located in the left area, the control terminal controls the line-following robot to turn to the right.

[0058] When a command to adjust to the left or right is given, and the obstacle is located in the right area, the control terminal controls the line-following robot to turn to the left.

[0059] When a route replanning instruction is issued, a new target route is replanned.

[0060] The technical advantages of this invention, a path analysis and planning system for a line-following robot based on computational mathematics, are as follows:

[0061] (1): This invention converts the image of the planning area into matrix nodes with obstacle markers, performs spatial transformation on the physical location of entities, and combines matrix optimization algorithm to automatically plan the target path from the starting position to the destination from the matrix nodes. This provides reasonable and accurate route support for the movement path of the line-following robot in the planning area, ensuring that the line-following robot can reach the destination with the shortest movement distance. This avoids the uncontrollable problems that exist in path planning when using pure vision or radar detection, thereby improving the planning efficiency of the target path.

[0062] (2): By collecting tracking videos on the target path and performing obstacle analysis on the tracking videos, the present invention can detect abnormal obstacle phenomena as soon as the tracking robot is affected or obstructed by obstacles. After analyzing the specific impact type of the obstacles, corresponding instruction measures are formulated to control the tracking robot to avoid the obstruction caused by the obstacles. This allows the tracking robot to perform targeted obstacle avoidance movements according to the severity of the actual obstacle. In this way, the dynamic situation of the tracking robot during its movement toward the destination can be detected in real time, and adaptive obstacle avoidance effect can be achieved during the movement of the tracking robot. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of a path analysis and planning system for a line-following robot based on computational mathematics, provided in Embodiment 1 of the present invention.

[0064] Figure 2 This is a flowchart illustrating a path analysis and planning method for a line-following robot based on computational mathematics, provided in Embodiment 2 of the present invention. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Example 1: Please refer to Figure 1 As shown in this embodiment, a path analysis and planning system for a line-following robot based on computational mathematics is applied to a control terminal and includes:

[0067] The image stitching module acquires sub-images of the planned area at the same time, removes overlapping areas in the sub-images, and then aligns and stitches the remaining sub-images to generate a regional image.

[0068] The planning area refers to the area that the line-following robot will pass through and cover when transferring materials and products in the workshop, and serves as the limit of the maximum location range for the line-following robot to perform path analysis and planning in the workshop.

[0069] Sub-images refer to partial images of the planned area captured by cameras installed on the top of the workshop at different locations. These images serve as the basis for subsequent aggregation and stitching of the overall image of the planned area. Since the equipment, personnel, and materials in the planned area will change dynamically at different times, it is necessary to ensure that all sub-images are captured at the same time to achieve the display of the planned area at the same moment.

[0070] After capturing sub-images, these sub-images can only represent the location of a portion of the planned area captured by each camera, and cannot represent the entire planned area. Therefore, it is necessary to align and stitch all the sub-images according to the actual distribution of the planned area, and record the stitched image as the region image.

[0071] When stitching sub-images into a region image, it is necessary to first identify overlapping regions in the sub-images that are repeated with adjacent sub-images, and then remove the overlapping regions to obtain sub-images that can be directly stitched together.

[0072] Specifically, the method for stitching together regional images is as follows:

[0073] At the same time, A cameras are used to capture images of the planned area from an overhead angle, resulting in A sub-images. After drawing lines along the outer edges of the sub-images, the outer boundary line is generated.

[0074] Draw the diagonals of the four corner points in each of the A sub-images to form two diagonals, and record the intersection of the two diagonals as the center point;

[0075] Measure the length of the diagonal in the sub-image, and record 5% of the diagonal length as the calibration length. Mark the points at the two ends of the two diagonals and at a distance of one calibration length from the adjacent corner point to obtain four overlapping points.

[0076] After connecting the four overlapping points in sequence, the inner boundary line is generated, and the area in the sub-image located between the inner boundary line and the outer boundary line is recorded as the overlapping area.

[0077] The overlapping areas in each of the A sub-images are removed, and the remaining A sub-images are spatially aligned and stitched together with the planned area to generate the region image.

[0078] It should be noted that the spatial alignment operation refers to aligning the positions of objects contained in the sub-image with the actual positions of objects contained within the planned area in terms of spatial orientation. This ensures that when the A sub-images are subsequently stitched together to form a regional image, the image displayed in the regional image will be consistent with the image corresponding to the actual planned area, laying the foundation for subsequent path analysis and planning.

[0079] The matrix node mode, based on the raster differentiation criterion, converts the regional image into a two-dimensional raster map with nodes, identifies the entity attributes of the raster in the planned area, and assigns values ​​to the raster based on the entity attributes to generate matrix nodes.

[0080] Two-dimensional raster maps are used to convert a holistic regional image into a matrix image of multiple independently distributed small grids, so that each grid in the two-dimensional raster map corresponds to a different physical meaning. The physical meaning includes, but is not limited to, impassable, freely passable, etc., which enables the two-dimensional raster map to perform orderly and accurate partitioning and segmentation of the holistic regional image.

[0081] When converting a regional image into a two-dimensional raster map, in order to avoid conversion errors and ensure the accuracy of the two-dimensional raster map conversion result, it is necessary to perform the conversion under the constraints of raster differentiation criteria.

[0082] Specifically, the grid division criterion is: half of the projected width of the target road is used as the grid division standard; this ensures that the line-following robot can move freely within each grid and also ensures the consistency of the size of each grid.

[0083] A node is a point in a two-dimensional raster map used to represent the actual object type at different locations in each raster, and serves as a location constraint to determine whether each raster is impassable or freely passable;

[0084] Specifically, the conversion method for two-dimensional raster maps is as follows:

[0085] Computer vision technology is used to identify B roads in the region image that are accessible to the tracking robot, and the projection values ​​of each of the B roads along the width direction are measured to obtain the B projection widths. The projection value refers to the span of the region image occupied by the road from the top view angle, which can be used to numerically represent the size of the road in the region image.

[0086] The road corresponding to the minimum projection width is recorded as the target road, and half of the projection width of the target road is used as the grid side length. The region image is divided into C array-distributed grids.

[0087] The center points of C grid cells are marked one by one. Using one-third of the grid cell's side length as the radius and the grid's center point as the center, inner circles are drawn in each of the C grid cells. Four points are then marked at equal angles on each inner circle, resulting in four nodes. This ensures that the four marked nodes maintain a ring-like structure with equal angles, guaranteeing the independence of each node's position and ensuring the independence of subsequent data collection and analysis results at each node.

[0088] After converting to a two-dimensional raster map with nodes, it is necessary to represent the actual object type corresponding to each node in the raster, and represent the entity attributes of the raster based on the actual object corresponding to the node, so as to determine the passage result represented by the raster in the two-dimensional raster map.

[0089] Specifically, entity attributes include obstacle attributes and access attributes; the obstacle attribute means that the actual object corresponding to the grid in the two-dimensional grid map is an obstacle, and the grid is impassable; the access attribute means that the actual object corresponding to the grid in the two-dimensional grid map is a road, and the grid is freely passable.

[0090] After identifying the entity attributes of the grid, each grid needs to be assigned a value so that the assignment can directly represent the actual passage status of the grid. The matrix node is a matrix structure containing node identifiers obtained after assigning values ​​to the grids in the two-dimensional grid map, and enables the matrix node to accurately represent the location of the line-following robot, the target position, the production equipment and the road.

[0091] Specifically, the method for generating matrix nodes is as follows:

[0092] The coordinates of the four nodes in the C grids on the electronic map are retrieved one by one, and the object type at the corresponding location of the coordinates is retrieved in the planning area. The object type is used to specifically represent the object or equipment that actually corresponds to the coordinates in the workshop planning area. The object type includes, but is not limited to, roads, equipment, etc.

[0093] Nodes whose object type is road are recorded as target nodes, and the number of target nodes in C grid cells is counted one by one to obtain the target value;

[0094] When the target value is 4, it means that the objects corresponding to the four nodes in the grid are all roads, so the entity attribute of the node is recorded as the passage attribute.

[0095] When the target value is not 4, it means that the objects actually corresponding to the four nodes in the grid are not all roads, so the entity attribute of the node is recorded as the obstacle attribute.

[0096] Assign values ​​of 0 and 1 to the grid cells with the attributes of passage and obstacle, respectively. Then, number the C grid cells in ascending order, starting with 1, from top to bottom and from left to right within each row, to generate numbered matrix nodes.

[0097] It should be noted that numbering the matrix nodes ensures that each grid cell in the matrix node has a unique and accurate identification number, thereby providing precise instructions for the path analysis and planning of the subsequent line-following robot.

[0098] The path planning module, based on the spatial transformation criterion, marks the starting and ending grids in the matrix nodes, and combines the matrix optimization algorithm to plan the target path of the line-following robot in the matrix nodes.

[0099] The starting grid refers to the grid corresponding to the current position of the line-following robot within the planning area, which can be used as the starting position for path analysis and planning of the line-following robot;

[0100] The termination grid refers to the grid corresponding to the final position reached by the line-following robot within the planned area, which can be used as the termination position for path analysis and planning of the line-following robot;

[0101] To ensure that an accurate and reasonable path can be provided for the line-following robot to reach its destination, the starting and ending grids need to be accurately marked in the matrix nodes, thus serving as the starting and ending points of the path for the line-following robot to reach its destination.

[0102] Since the starting grid and the ending grid are positional representations on matrix nodes, while the current position of the tracking robot and the final destination position are spatial positional representations on the planning area, the two positions cannot be directly substituted. Under the constraints of spatial transformation criteria, the position in physical space needs to be transformed into the two-dimensional space on the matrix nodes, thereby providing positional support for the subsequent analysis and planning of the target path.

[0103] Specifically, the spatial transformation criterion is: one point coordinate corresponds to one grid; this ensures that the current position and final destination of the line-following robot can correspond one-to-one with a unique grid.

[0104] The marking method for the starting and ending grids is as follows:

[0105] The positioning chip is used to query the real-time location of the line-following robot at the current moment, and the real-time location and destination coordinates of the line-following robot are marked on the electronic map, which are recorded as the start coordinates and the end coordinates, respectively.

[0106] Using the length and width of the matrix nodes as the standard length and width, the length and width of the electronic map are proportionally reduced or expanded to match the standard length and width. When the length and width of the electronic map are greater than the standard length and width, the length and width of the electronic map need to be reduced, and vice versa, the length and width of the electronic map are expanded, thereby achieving the applicability reduction or expansion effect of the electronic map.

[0107] Rotate the reduced or expanded electronic map counterclockwise until the electronic map and matrix nodes are aligned in space to generate the target map;

[0108] The target map is overlaid on the matrix nodes, and the grids containing the starting and ending coordinates are designated as the starting grid and the ending grid, respectively.

[0109] After marking the starting and ending grids, these grids can be used as the starting and ending points of the target path. This allows for the automatic analysis and planning of the target path for the tracking robot from the starting grid to the ending grid, using a path search algorithm based on matrix operations within the matrix nodes.

[0110] In this embodiment, the target path is a path with the starting grid and the ending grid as the starting point and the ending grid, ensuring that the tracking robot can reach the destination position with the shortest movement distance.

[0111] Specifically, the method for planning the target path is as follows:

[0112] Using the horizontal direction as rows and the vertical direction as columns, transform the matrix nodes into a map matrix with i rows and y columns;

[0113] Transfer the values ​​of the C grid cells to the C elements of the map matrix, and denote the element located in the top left corner of the map matrix as the initial point;

[0114] Mark the row and column values ​​of the corresponding elements of the starting and ending grids in the map matrix respectively, and combine the row and column values ​​of the elements to generate the starting and ending values;

[0115] The cost of crossing an element is denoted as the unit cost, and the actual distance between the initial point and the starting grid is measured and denoted as the cost matrix.

[0116] After comparing the row and column values ​​of the termination value with the row and column values ​​of the starting value, the estimated distance from the starting grid to the termination grid is calculated and denoted as E heuristic matrices.

[0117] The formula for calculating the heuristic matrix is:

[0118] ;

[0119] In the formula, For heuristic matrices, The row value is the terminating value. The row value is the starting value. The column value is the terminating value. The column values ​​are the initial values;

[0120] The path span value is calculated by adding the cost matrix to each of the E heuristic matrices one by one.

[0121] The formula for calculating the path span value is:

[0122] ;

[0123] In the formula, The path span value. The cost matrix;

[0124] The minimum value of the path span is used as the span of the target path. Combined with the A* search algorithm, the target path is automatically planned in the matrix nodes.

[0125] It should be noted that the A* search algorithm is an algorithm that finds the lowest passage cost on the graphical plane. As an existing technology in this field, it expands the search space of the traditional A algorithm by taking into account the current position and movement mode of the line-following robot, and generates a target path that conforms to the characteristics of the line-following robot.

[0126] For example, when planning a target path in an earthquake matrix, if the initial value is... When the starting grid moves one grid to the right, it will reach... At this point, the inspection robot has essentially moved one unit at a cost. If it moves down one grid, it will reach... At this point, the inspection robot has also moved by one unit. If the value of the grid after moving down is 1, it means that the grid is impassable. In this case, it should move up, right, or left by one grid to avoid the impact of the obstacle and thus achieve the effect of automatic search for the target path.

[0127] The mode selection module collects real-time video of the line-following robot on the target path, performs obstacle analysis on the line-following video, determines the real-time line-following status of the line-following robot, and selects the corresponding planning and control mode based on the line-following status; the planning and control modes include path maintenance mode, path adjustment mode, and obstacle avoidance planning mode.

[0128] Once the target path is generated, the control terminal can send the route information corresponding to the target path to the tracking robot and control the tracking robot to move towards the destination according to the target path, thereby realizing the effective transfer of materials and products in the workshop.

[0129] Since the people and objects in the planned area of ​​the workshop are not stationary, when people and objects move within the planned area and enter the target path, they may interfere with the movement trajectory of the line-following robot, causing the line-following robot to be unable to move according to the original target path. Therefore, it is necessary to monitor and analyze the real-time movement of the inspection robot on the target path.

[0130] Line-following video is a video image used to represent the dynamic changes of a line-following robot's real-time movement on a target path, and serves as the basis for subsequent judgments on whether the line-following robot will collide with or come into contact with obstacles during its movement; the line-following video is obtained by real-time shooting by a camera installed on the line-following robot.

[0131] After capturing the tracking video, it is necessary to perform obstacle analysis on the information contained in the tracking video to determine whether there are any obstacles in the tracking video that may hinder the movement of the tracking robot. Based on the results of the obstacle analysis, the tracking status of the tracking robot at the current moment can be determined.

[0132] The tracking state is used to indicate whether the tracking robot encounters obstacles that affect its movement along the target path at the current moment, and serves as the basis for selecting the real-time motion mode of the tracking robot in the future.

[0133] Specifically, the tracking states include normal state, affected state, and avoidance state. The normal state means that the tracking robot does not encounter any obstacles at the current moment. The affected state means that the tracking robot encounters minor obstacles at the current moment. The avoidance state means that the tracking robot encounters severe obstacles at the current moment.

[0134] The method for determining the tracking state is as follows:

[0135] The system acquires real-time tracking videos of the tracking robot during the tracking period, extracts the tracking image at the last tracking moment in the tracking period from the tracking video, and identifies obstacles in the tracking image based on computer vision technology. Obstacles are used to represent the specific objects contained in the tracking image, including but not limited to equipment, personnel, materials, etc.

[0136] When there are no obstacles in the tracking image, the tracking robot has not encountered any obstacles at the current moment, and the actual motion state of the tracking robot will not change. The tracking state is then determined as the normal state.

[0137] When there are obstacles in the tracking image, the tracking image is divided into left, middle and right regions along the vertical direction, and the area values ​​of the left and right regions are adjusted until the area values ​​of the left and right regions are both one-sixth of the area value of the tracking image, and then the adjustment is stopped.

[0138] After drawing lines along the outer edge of the obstacle, an obstacle region is generated, and the positional relationship between the obstacle region and the left and right regions is analyzed.

[0139] If the obstacle area is entirely located inside the left or right area, the obstacle has encroached on the trajectory path of the line-following robot, and the actual motion state of the line-following robot will change slightly. In this case, the line-following state is determined to be the affected state.

[0140] If the obstacle area is partially or entirely located inside the central area, the obstacle intrudes into the trajectory path of the line-following robot, and the actual motion state of the line-following robot will change drastically. In this case, the line-following state is determined to be an avoidance state.

[0141] Once the tracking state of the tracking robot is determined, the corresponding planning and control mode can be selected according to the different tracking states. This allows the planning and control mode to be used to analyze and control the movement mode of the tracking robot after the current moment, thereby driving the tracking robot to dynamically adjust and process its working state according to the real-time situation at the current moment.

[0142] Specifically, the planning and control modes include path maintenance mode, path adjustment mode, and obstacle avoidance planning mode. Path maintenance mode refers to the line-following robot being in a normal, unchanged mode. Path adjustment mode refers to the line-following robot being in a mode where the path is slightly changed and adjusted. Obstacle avoidance planning mode refers to the line-following robot being in a mode where the path is significantly changed and adjusted.

[0143] The method for selecting the planning control mode is as follows:

[0144] When the tracking robot is in the normal tracking state, there is no need to change or adjust the robot's movement mode, so the path maintenance mode is selected.

[0145] When the tracking robot is in an affected state, and it is necessary to make a small change to the robot's movement mode, then select the path adjustment mode.

[0146] When the tracking robot is in an obstacle avoidance state, and a significant change in the robot's movement is required, then the obstacle avoidance planning mode should be selected.

[0147] The planning and control module formulates control instructions corresponding to the planning and control mode. The control terminal sends the control instructions to the line-following robot and controls the line-following robot to execute the control instructions. The control instructions include instructions to maintain the current motion, instructions to adjust left or right, and instructions to replan the path.

[0148] Once the corresponding planning and control mode is selected for the line-following robot, different specific methods can be used to plan and control the robot in different modes to ensure that it can move quickly and accurately toward its final destination, while avoiding collisions with obstacles during its movement.

[0149] To ensure that the line-following robot can move normally on the target path, it is necessary to formulate corresponding control instructions under different planning and control modes, so that the control instructions can serve as the direct basis for the control terminal to control the line-following robot to make changes and adjustments at the current moment.

[0150] Specifically, control commands include commands to maintain the current motion, commands to adjust left or right, and commands to replan the target path.

[0151] The method for formulating control commands is as follows:

[0152] When the path-maintaining mode is selected, the tracking robot can move along the original target path without any changes or adjustments, and the current motion is maintained by formulating the command.

[0153] When the path adjustment mode is selected, the line-following robot can move along the original target path, but if a small adjustment is needed, then a left or right adjustment command is given.

[0154] When obstacle avoidance planning mode is selected, the line-following robot cannot move along the original target path and needs to make significant changes and adjustments, so a path replanning instruction is generated.

[0155] After the corresponding control commands are formulated, the control terminal can send the control commands to the line-following robot and plan and control the movement of the line-following robot according to the different actual control commands.

[0156] Specifically, once the instruction to maintain the current motion is given, there is no need to adjust the movement mode of the line-following robot; the line-following robot can simply continue moving along the original target path.

[0157] When a left or right adjustment command is given, it is necessary to determine whether the obstacle is located in the left or right region. If the obstacle is in the left region, and the line-following robot encounters an obstacle on its left side, the control terminal sends a right turn command to the line-following robot's motion mechanism to avoid the obstacle. If the obstacle is in the right region, and the line-following robot encounters an obstacle on its right side, the control terminal sends a left turn command to the line-following robot's motion mechanism to avoid the obstacle.

[0158] When a path replanning instruction is given, the tracking robot can no longer continue moving along the original target path. The original target path needs to be discarded and a new target path needs to be replanned. At this time, the starting grid needs to be redefined in the matrix node, and the target path replanning process needs to be performed. Then, it is necessary to return to the path planning module to plan the new target path and perform subsequent related analysis and control processing.

[0159] It should be noted that before the line-following robot reaches its final destination, the target path analysis and planning operation will be performed in the manner described above. This process will continue until the line-following robot reaches its final destination. This achieves the diverse effects of dynamic monitoring, real-time analysis, and obstacle avoidance planning of the line-following robot's movement path when transferring materials and products within the planned area, thus realizing the intelligent, efficient, and dynamic planning and control objectives of the line-following robot.

[0160] Example 2: Please refer to Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A path analysis and planning method for a line-following robot based on computational mathematics is provided, applied to a control terminal, and implemented based on a path analysis and planning system for a line-following robot based on computational mathematics, including:

[0161] S01: At the same time, acquire sub-images of the planned area, remove overlapping areas in the sub-images, align and stitch the remaining sub-images to generate the area image;

[0162] S02: Based on the grid differentiation criteria, the regional image is converted into a two-dimensional grid map with nodes, the entity attributes of the grid in the planning area are identified, and the grid is assigned values ​​based on the entity attributes to generate matrix nodes;

[0163] S03: Based on the principle that one point coordinate corresponds to one grid, the starting grid and the ending grid are marked in the matrix nodes, and the target path of the line-following robot is planned in the matrix nodes by combining the matrix optimization algorithm.

[0164] S04: Real-time acquisition of tracking video of the tracking robot on the target path, obstacle analysis of the tracking video, determination of the real-time tracking status of the tracking robot, and selection of the corresponding planning and control mode based on the tracking status.

[0165] S05: Formulate control instructions corresponding to the planning control mode. The control terminal controls the line-following robot to maintain the current movement, adjust to the left or right, and replan the path until the destination is reached.

[0166] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A path analysis and planning system for a line-following robot based on computational mathematics, applied to a control terminal, characterized in that, include: The image stitching module acquires sub-images of the planned area at the same time. After removing overlapping areas in the sub-images, it aligns and stitches the remaining sub-images to generate a regional image. The matrix node mode, based on the raster differentiation criterion, converts the regional image into a two-dimensional raster map with nodes, identifies the entity attributes of the raster in the planned area, and assigns values ​​to the raster based on the entity attributes to generate matrix nodes. The grid division criterion is: half of the projected width of the target road is used as the grid division standard; The conversion method for two-dimensional raster maps is as follows: Using computer vision technology, B roads in the area image are identified for the pathfinder robot to travel on, and the projection values ​​of the B roads along the width direction are measured one by one to obtain the B projection widths. The road corresponding to the minimum projection width is recorded as the target road, and half of the projection width of the target road is used as the grid side length. The region image is divided into C array-distributed grids. Mark the center point of each of the C grid cells one by one. Using one-third of the grid cell side length as the radius and the center point of the grid cell as the center, draw inner circles in each of the C grid cells. Mark four points at equal angles on the inner circles to obtain four nodes. Entity attributes include obstacle attributes and passability attributes; The method for generating matrix nodes is as follows: Query the coordinates of the four nodes in each of the C grids on the electronic map, and then query the object type at the corresponding location of the coordinates in the planning area. Nodes whose object type is road are recorded as target nodes, and the number of target nodes in C grid cells is counted one by one to obtain the target value; When the target value is 4, the entity attribute of the grille is the passage attribute; when the target value is not 4, the entity attribute of the grille is the obstacle attribute. Assign values ​​of 0 and 1 to the grid cells with the attributes of passage and obstacle respectively, and number the C grid cells in ascending order, starting with 1, in a row-by-row manner and from left to right within each row, to generate numbered matrix nodes. The path planning module, based on the principle that one point coordinate corresponds to one grid, marks the starting grid and the ending grid in the matrix nodes, and combines the matrix optimization algorithm to plan the target path of the line-following robot in the matrix nodes. The method for planning the target path is as follows: Using the horizontal direction as rows and the vertical direction as columns, the matrix nodes are converted into a map matrix with i rows and y columns. The values ​​of C grid cells are transferred to C elements of the map matrix, and the element located in the top left corner of the map matrix is ​​recorded as the initial point. Mark the row and column values ​​of the corresponding elements of the starting and ending grids in the map matrix respectively, and combine the row and column values ​​of the elements to generate the starting and ending values; The cost of crossing an element is denoted as the unit cost, and the actual distance between the initial point and the starting grid is measured and denoted as the cost matrix. After comparing the row and column values ​​of the termination value with the row and column values ​​of the starting value, the estimated distance from the starting grid to the termination grid is calculated and denoted as E heuristic matrices. The formula for calculating the heuristic matrix is: ； In the formula, For heuristic matrices, The row value is the terminating value. The row value is the starting value. The column value is the terminating value. The column values ​​are the initial values; After adding the cost matrix to each of the E heuristic matrices, the path span value is calculated. The minimum path span value is used as the span of the target path. Combined with the A* search algorithm, the target path is automatically planned in the matrix nodes. The mode selection module collects real-time video of the line-following robot on the target path, performs obstacle analysis on the line-following video, determines the real-time line-following status of the line-following robot, and selects the corresponding planning and control mode based on the line-following status. The planning and control module generates control commands corresponding to the planning and control mode. The control terminal controls the line-following robot to maintain its current movement, adjust to the left or right, and replan the path until it reaches its destination.

2. The path analysis and planning system for a line-following robot based on computational mathematics according to claim 1, characterized in that, The method for stitching together regional images is as follows: At the same time, A cameras are used to capture images of the planned area from an overhead angle, resulting in A sub-images. After drawing lines along the outer edges of the sub-images, the outer boundary line is generated. Draw two diagonals for each of the four corner points within each of the A sub-images, and mark the intersection of the two diagonals as the center point; Measure the length of the diagonal in the sub-image, and record 5% of the diagonal length as the calibration length. Mark the points at the two ends of the two diagonals and at a distance of one calibration length from the adjacent corner point to obtain four overlapping points. After connecting the four overlapping points in sequence, the inner boundary line is generated, and the area in the sub-image located between the inner boundary line and the outer boundary line is recorded as the overlapping area. Remove the overlapping areas of the A sub-images, and then align and stitch the remaining A sub-images with the planned area to generate the region image.

3. The path analysis and planning system for a line-following robot based on computational mathematics according to claim 2, characterized in that, The marking method for the starting and ending grids is as follows: The positioning chip is used to query the real-time location of the line-following robot at the current moment, and the real-time location and destination coordinates of the line-following robot are marked on the electronic map, which are recorded as the start coordinates and the end coordinates, respectively. Using the length and width of the matrix nodes as the standard length and width, the length and width of the electronic map are proportionally reduced or expanded to match the standard length and width. Rotate the reduced or expanded electronic map counterclockwise until the electronic map and matrix nodes are aligned in space to generate the target map; The target map is overlaid on the matrix nodes, and the grids containing the starting and ending coordinates are designated as the starting grid and the ending grid, respectively.

4. The path analysis and planning system for a line-following robot based on computational mathematics according to claim 3, characterized in that, Tracking states include normal states, affected states, and avoidance states: The method for determining the tracking state is as follows: The system acquires real-time tracking videos of the tracking robot during the tracking period, extracts the tracking image at the last tracking moment in the tracking period from the tracking video, and identifies obstacles in the tracking image based on computer vision technology. When there are no obstacles in the tracking image, the tracking state is determined to be the normal state; When there are obstacles in the tracking image, the tracking image is divided into left, middle and right regions in the vertical direction, and the area values ​​of the left and right regions are adjusted until the area values ​​of the left and right regions are both one-sixth of the area value of the tracking image. Then, the obstacle region is generated by drawing a line along the outer edge of the obstacle. If the entire obstacle area is located inside the left or right region, the tracking state is determined to be the affected state. If the obstacle area is partially or entirely located inside the central area, the tracking state will be determined as an avoidance state.

5. The path analysis and planning system for a line-following robot based on computational mathematics according to claim 4, characterized in that, Planning and control modes include path maintenance mode, path adjustment mode, and obstacle avoidance planning mode; The method for selecting the planning control mode is as follows: When the tracking robot is in a normal state, an affected state, or an obstacle avoidance state, select the path maintenance mode, path adjustment mode, or obstacle avoidance planning mode respectively.

6. The path analysis and planning system for a line-following robot based on computational mathematics according to claim 5, characterized in that, Control commands include commands to maintain current motion, commands to adjust left or right, and commands to replan the target path; The method for formulating control commands is as follows: When selecting path maintenance mode, path adjustment mode, or obstacle avoidance planning mode, commands to maintain the current movement, adjust to the left or right, or replan the path are generated respectively.

7. The path analysis and planning system for a line-following robot based on computational mathematics according to claim 6, characterized in that, When given the instruction to maintain the current motion, the tracking robot continues to move along the original target path; When a command to adjust to the left or right is given, and the obstacle is located in the left area, the control terminal controls the line-following robot to turn to the right. When a command to adjust to the left or right is given, and the obstacle is located in the right area, the control terminal controls the line-following robot to turn to the left. When a route replanning instruction is issued, a new target route is replanned.

Images