Path planning method and system for autonomous operation of underground shovel

Abstract

This invention discloses a path planning method and system for autonomous operation of underground loader. The method includes: acquiring map data of the autonomous operation area of ​​the target underground loader; obtaining the skeleton information of the autonomous operation area using a skeleton extraction algorithm; obtaining the shortest reachable path from a set starting point to a set ending point; extracting the turning points on the shortest reachable path and sequentially connecting the set starting point, each turning point, and the set ending point to obtain an initial polyline path; interpolating and trimming the initial polyline path to obtain a collision-free polyline from the set starting point to the set ending point; determining vehicle U-turns and detecting intersections; constructing V-shaped U-turn paths and replacing corners at intersections to complete the path planning for autonomous operation of the underground loader from the set starting point to the set ending point. This invention not only achieves path planning for autonomous operation of underground loader, but also features continuous curvature, higher reliability, and better feasibility and accuracy.

Description

Technical Field

[0001] This invention belongs to the field of intelligent mining technology, specifically relating to a path planning method and system for autonomous operation of underground loaders. Background Technology

[0002] With the increasing demand for mineral resources, mining production is increasingly extending into deeper and more dangerous underground environments. To address the dual challenges of operational safety and production efficiency, the mining industry is gradually moving towards autonomous mining. Among various underground equipment, loaders are the backbone of material transportation, and their production efficiency directly determines the ore extraction capacity of the entire mining area. Therefore, planning the autonomous operation paths of underground loaders is of great significance.

[0003] The operation of underground loaders typically follows a standard "V"-shaped work cycle consisting of loading, transporting, and unloading. A key characteristic of this cycle is that the bucket must always be facing the work target during the loading and unloading phases. Therefore, during the transport phase, the vehicle must perform a reverse maneuver (U-turn) at a pre-defined decision point (usually an intersection) to reverse its direction of travel. These forced reverse maneuvers, performed within extremely confined spaces and based on intersections, present significant geometric and kinematic challenges to trajectory planning.

[0004] Currently, common path planning schemes for underground loader operators mainly include grid-based search methods and sampling methods. Grid-based search methods, such as Dijkstra's algorithm and A* algorithm, often have high computational costs on large-scale and cluttered mine maps and tend to generate jagged paths with many inflection points, resulting in poor performance. Sampling methods, such as PRM, RRT, and RRT*, often encounter the typical "narrow passage" problem in narrow underground tunnels, exhibiting a significant heavy-tailed running time distribution and large fluctuations in solution quality. If kinematically feasible local connectors are lacking, the generated paths usually require post-processing smoothing, which may reduce the clearance between the operator and obstacles or violate the turning path sharpness constraint. Summary of the Invention

[0005] One of the objectives of this invention is to provide a highly reliable and accurate path planning method for autonomous operation of underground scrapers.

[0006] The second objective of this invention is to provide a system for implementing the path planning method for autonomous operation of underground scrapers.

[0007] The path planning method for autonomous operation of underground loaders provided by this invention includes the following steps:

[0008] S1. Obtain map data information of the autonomous operating area of ​​the target underground loader;

[0009] S2. Based on the width of the target underground loader and the set safety margin, the skeleton extraction algorithm is used to obtain the skeleton information of the autonomous operation area from the map data information obtained in step S1.

[0010] S3. Based on the skeleton information obtained in step S2, a grid-based search method is used to obtain the shortest reachable path from the set starting point to the set ending point;

[0011] S4. Based on the heading angle changes of each point on the path, extract the turning points on the shortest reachable path obtained in step S3, and connect the set starting point, each turning point and the set ending point in sequence to obtain the initial polyline path.

[0012] S5. Based on the geometric positional relationship between each line segment in the initial polyline path and the autonomous operation area, interpolate and trim the initial polyline path to obtain a collision-free polyline from the set starting point to the set ending point within the autonomous operation area.

[0013] S6. Determine vehicle U-turns based on the positional relationship between the direction vectors of the set starting point and the set ending point, and detect intersections based on the positional relationship between the neighborhood of each point on the shortest reachable path and the shortest reachable path.

[0014] S7. Based on the intersection detection results obtained in step S6, and based on the corner information at each intersection, construct the V-shaped U-turn path at the intersection and replace the corners in the route to obtain the final route planning line, thus completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

[0015] Step S1 specifically includes the following steps:

[0016] Obtain map data information of the autonomous operating area of ​​the target underground loader;

[0017] The obtained map data is rasterized, with the grid containing obstacles set to 1 and the grid that the target underground loader can pass through set to 0, thus obtaining a binary occupancy grid map of the target underground loader's autonomous operating area.

[0018] Step S2 specifically includes the following steps:

[0019] Set an obstacle set Obstacle set This includes all grid cells in a binary occupied grid map that have a value of 1;

[0020] Calculate obstacle radius for , The width of the target underground loader. The set safety threshold;

[0021] Based on the obtained obstacle radius Using a binary occupancy grid map, we can obtain areas that can be safely traversed within the autonomous operation area. for ,in For binary occupied grid maps, Point To the obstacle set The distance;

[0022] The skeleton extraction algorithm is used to extract the region. The skeleton, the skeleton information of the autonomous operation area .

[0023] Step S3 specifically includes the following steps:

[0024] Obtain the starting point of the target underground loader in its autonomous operating area. and the end point of the task ;

[0025] In skeleton information Get distance to the starting point of the operation The nearest point is used as the starting point. ;

[0026] In skeleton information Get distance to the end point of the operation The nearest point is used as the endpoint. ;

[0027] A grid-based search method is used to search the skeleton information. Searching on the [top] will yield results starting from the set starting point. To the set destination Shortest reachable path .

[0028] The S4 mentioned above specifically includes the following steps:

[0029] Set sampling span for ,in The average width of the alleyway. Raster resolution;

[0030] For the shortest reachable path The i-th point in ,exist Within the range, set the i-th point. rear heading angle for The i-th point Forward heading angle for Where N is the shortest reachable path. The total number of points It is a two-parameter arctangent function;

[0031] The local heading change was calculated. for ;in, To normalize the angle to operate;

[0032] For local heading changes Make a judgment: If Then the point Marked as a turning point; if Then point It is not a turning point; The threshold value is set for determining the turning point;

[0033] Represent all the marked inflection points as The turning point is on the shortest reachable path. The corresponding index in the table represents what is meant by Then it exists At all inflection points, the index difference should not exceed the sampling span. The turning point is divided into the same curve segment. For each curve segment Take the midpoint as Heart ;

[0034] According to the index, check all the bends. Sort the sequences to obtain the core sequence. for , The total number of apex bends;

[0035] Set the starting point , Bend Core Sequence Each apex and set finish line Connect them sequentially to obtain the initial polyline path. .

[0036] Step S5 specifically includes the following steps:

[0037] Interpolation stage: from the initial polyline path Starting from the first point, proceed sequentially to each adjacent point. , Line segments Make a judgment: if Not all located in the region Inside, then at the point and points Insert a point between , Shortest reachable path Points and indexes in Values ,in For point Shortest reachable path The index value in For point Shortest reachable path The index value in The floor symbol;

[0038] Repeat the above interpolation steps until all line segments formed by adjacent points lie within the region. Within, an initial collision-free polyline is obtained. ;

[0039] Cutting stage:

[0040] Initialize the most recently retained vertex for ;

[0041] From the initial collision-free polygonal line The first point in Begin by testing the initial collision-free polylines sequentially. Each vertex in Perform the detection: If starting from the current most recently retained vertex arrive line segments All located in the region If inside, then delete the vertex. Otherwise, preserve the vertices. and the current most recently retained vertex Update to vertices ;

[0042] Repeat the above cutting steps to finally obtain a collision-free zigzag line. .

[0043] Step S6 specifically includes the following steps:

[0044] Determining if a vehicle is making a U-turn:

[0045] Set starting point The direction vector of the loader at the location is represented as: Set the endpoint The direction vector of the loader at the location is represented as: Collision-free zigzag lines Represented as , direction vector at the starting point of the path for Path endpoint direction vector Represented as The direction in which the loader faces is defined as the direction of the loader's bucket.

[0046] like If so, it is determined that assistance is needed to turn the vehicle around;

[0047] Intersection detection:

[0048] Shortest reachable path Perform intersection detection: Set c as the shortest reachable path. The point on, For the 8-neighborhood of c; when More than The neighborhood is the shortest reachable path If the above is true, then c will be set as a candidate point for the intersection.

[0049] Step S7 includes the following steps:

[0050] Based on the intersection detection results obtained in step S6, the system first searches for U-turn points where the path has already turned at the intersection and constructs the corresponding U-turn structure. If no U-turn point can be found, a U-turn and back maneuver mode centered on the intersection is set, and a V-shaped U-turn path is constructed at the intersection. Finally, all corners in the route are constructed using CAC structures to obtain the final path planning route, thus completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

[0051] Step S7 specifically includes the following situations:

[0052] Scenario 1: The set path has already turned at the intersection:

[0053] Based on the intersection detection results obtained in step S6, select any intersection candidate point. In collision-free fold lines Get the distance to the intersection candidate point nearest point And obtain the local corner triplet. ;

[0054] Set the far endpoint: from the intersection candidate point Start, along the skeleton Shortest reachable path Traverse outwards along the other skeleton branches, enumerating the far endpoints of each branch. ; distal end The definition is: the shortest reachable path on the branch. Connected endpoints to the far endpoint All line segments are located in the region Within, and with the longest length in the branch direction; where, if it is a maneuvering mode of first turning into the branch and then completing the reverse, then the shortest reachable path on the branch. The connected endpoints are If the driving pattern involves first proceeding straight through the intersection into the side lane and then turning back, then the shortest reachable path on the branch is... The connected endpoints are ;

[0055] Scenario 1-1: Setting a maneuver mode where the vehicle first turns and enters a side tunnel before reversing:

[0056] In this maneuver mode, the vehicle first turns into a side branch at the intersection, then proceeds along the side lane, shifts gears and reverses in the side lane, and finally returns to the main lane.

[0057] In skeleton information Several distant endpoints are obtained by searching on the lateral branches other than the shortest reachable path R. For each remote endpoint The intersection point is obtained. for and the intersection As the new corner apex of the intersection;

[0058] sequentially at all remote endpoints The triplet it belongs to Construct the CAC structure and obtain the corresponding tangent point. If the tangent point is located on the line segment... If the inside is selected, then that tangent point is chosen as the turning point. ;

[0059] If a U-turn point that meets the conditions exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... pass Turn to ,exist The driving direction is reversed at this point, and then along the straight segment from Back At this point, replace the corresponding corner with: turning section. and a section at the U-turn point The opposite line segment at the location ;

[0060] If there is no U-turn point that meets the conditions If so, continue with the following steps;

[0061] Scenario 1-2: Setting a driving mode where the vehicle first goes straight through the intersection, enters the side lane, and then turns back in the opposite direction:

[0062] In this maneuvering mode, the vehicle first goes straight through the intersection and enters the side lane, then reverses, and finally turns back to the main lane.

[0063] In skeleton information Several distant endpoints are obtained by searching on the lateral branches other than the shortest reachable path R. For each remote endpoint The intersection point is obtained. for ;

[0064] sequentially at all remote endpoints The triplet it belongs to Construct the CAC structure on the line segment and obtain the corresponding tangent point. If the tangent point is located on the line segment... If the inside is selected, then that tangent point is chosen as the turning point. ;

[0065] If a U-turn point exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... Drive straight to ,exist The driving direction is reversed at this point, and then... Turn and return At this point, replace the corresponding corner with a straight line segment. and a period of time After reversing, pass Turning section ;

[0066] If there is no U-turn point If so, continue with the following steps;

[0067] Scenario 2: Set up a U-turn / reverse maneuver mode centered on the intersection:

[0068] Based on the intersection detection results obtained in step S6, select any intersection candidate point. In collision-free fold lines Get the distance to the intersection candidate point The two nearest neighboring points and ;

[0069] Candidate points at intersections direct as and set the remote endpoint Shortest reachable path The connected endpoints are ;

[0070] Skeletal information corresponding to lateral branches The search yielded several remote endpoints. ;

[0071] For each remote endpoint There are two sets of turning triples. and In the triplet Constructing a CAC structure at the point yields a line segment. Several first tangent points on; then triplet Constructing a CAC structure at the point yields a line segment. Several second tangent points on the line segment; among several first tangent points and several second tangent points, select those where both the first and second tangent points lie on the line segment. Inside, and at a distance The more distant tangent point is used as the turning point. ;

[0072] If a U-turn point exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... pass Turn into the side alley and reach the U-turn point. Afterwards, The driving direction is reversed at this point, and then... After turning back ;

[0073] If there is no U-turn point If so, it is determined that under the current safety constraints, V-shaped turns assisted by intersections are not supported;

[0074] The remaining corners of the route are then constructed using CAC structures to obtain the final path planning route.

[0075] The construction of the CAC structure includes the following steps:

[0076] The transition section is constructed using a symmetrical spiral curve-circular arc-spiral curve profile; curvature constraints are satisfied during the construction process. and sharpness constraints ;in, For curvature, For the set curvature threshold, Sharpness is defined as the rate of change of curvature. The set sharpness threshold;

[0077] When constructing a spiral curve, the following methods are used: Perform curve construction; during the construction process, solve... The parametric triplet of the following cyclotron curve - circular arc - cyclotron curve and tangent point position This allows for the determination of the corresponding spiral curve-circular arc-spiral curve substitution segment; among which, The length of the spiral curve, The length of the arc. The maximum curvature of the cycloid is denoted by . The coordinates of the point of tangency between the straight section entering the curve and the first spiral curve are given. The coordinates are the points of tangency between the straight section exiting the curve and the second spiral curve.

[0078] The construction of the CAC structure specifically includes the following steps:

[0079] Step 1: Set the entry heading angle at the corner of the broken line as... The heading angle at the exit of the curve is The heading change angle was calculated. for ;

[0080] For sharpness constant The curve segment exists ,in Let be the curvature at arc length s. The maximum curvature of the cyclotron curve Values ;

[0081] Heading changes caused by a single-sided spiral slope Represented as Therefore, the heading changes caused by the two symmetrical spiral curve slopes are obtained. for The curvature of the middle arc segment is constant. And the corresponding arc length is The corresponding change in heading Values ; Obtain the total heading change for ;

[0082] set up ;

[0083] Step 2: Use the pure spiral curve mode:

[0084] at this time, ,but The peak curvature required for the pure cyclotron mode is ;

[0085] like satisfy Then it is determined that the change in course can be achieved through two segments with sharpness of [missing information]. The symmetric cyclotron curve is realized; the corresponding solution is:

[0086]

[0087]

[0088]

[0089] In the formula for The symbol;

[0090] like If the course change cannot be achieved through two segments with a sharpness of [value missing], then it is determined that the course change cannot be achieved through two segments with a sharpness of [value missing]. The symmetrical cyclotron curve is achieved, and subsequent steps are continued;

[0091] Step 3: Use the spiral curve-circular arc-spiral curve pattern:

[0092] set up At this point, the slope length is fixed. ;

[0093] The remaining heading changes are completed by constant curvature circular arcs, represented as:

[0094]

[0095]

[0096] In the formula The heading change included in the circular arc segment;

[0097] Step 4: Settings On point The heading angle at the point of entry into the curve is The heading angle at the exit of the curve is Set the unit tangent vector for the curve. for unit tangent vector at the exit of the curve for ; Point of tangency at the bend for The point of tangency at the bend for ;in, for arrive distance, for arrive The distance;

[0098] Interchange the X-axis with Alignment, with the origin at the starting point of the entry curve segment. At this time, a parameter is The spiral curve-circular arc-spiral curve segment, from arrive The generated displacement vector Represented as , This represents the displacement along the x-axis of the local coordinate system. This represents the displacement along the y-axis in the local coordinate system; the calculation yields... and will according to Rotate back to the global coordinate system to obtain the displacement vector in the global coordinate system. for ,in It is a two-dimensional rotation matrix;

[0099] exist ;

[0100] Will Projected to and the corresponding normal vector Above: Order ,Will exist Basis expansion yields the results. and .

[0101] This invention also provides a system for implementing the path planning method for autonomous operation of underground shovel loader, comprising a data acquisition module, a skeleton extraction module, a reachability extraction module, a polyline extraction module, a collision-free extraction module, an intersection detection module, and a path planning module; the data acquisition module, skeleton extraction module, reachability extraction module, polyline extraction module, collision-free extraction module, intersection detection module, and path planning module are connected in series; the data acquisition module is used to acquire map data information of the autonomous operation area of ​​the target underground shovel loader and upload the data information to the skeleton extraction module; the skeleton extraction module is used to obtain skeleton information of the autonomous operation area based on the acquired map data information, the width of the target underground shovel loader, and a set safety margin, using a skeleton extraction algorithm, and upload the data information to the reachability extraction module; the reachability extraction module is used to obtain the shortest reachable path from a set starting point to a set ending point using a grid-based search method based on the received data information and the obtained skeleton information, and upload the data information to the polyline extraction module; the polyline extraction module is used to obtain the shortest reachable path from each point on the path based on the received data information and the obtained skeleton information, using a grid-based search method, and upload the data information to the polyline extraction module; the polyline extraction module is used to obtain the shortest reachable path based on each point on the path, using a grid-based search algorithm, based on the received data information. The system calculates the turning points on the shortest reachable path by changing the heading angle of the points, and then connects the set starting point, each turning point, and the set ending point to obtain an initial polyline path. This data is then uploaded to the collision-free extraction module. The collision-free extraction module, based on the received data and the geometric positional relationship between each line segment in the initial polyline path and the autonomous operating area, interpolates and trims the initial polyline path to obtain a collision-free polyline from the set starting point to the set ending point within the autonomous operating area. This data is then uploaded to the intersection detection module. The intersection detection module, based on the received data and the positional relationship between the direction vectors of the set starting point and the set ending point, determines whether a vehicle needs to turn around. It also detects intersections based on the positional relationship between the neighborhood of each point on the shortest reachable path and the shortest reachable path itself. This data is then uploaded to the path planning module. The path planning module, based on the received data and the corner information at each intersection, constructs a V-shaped U-turn path at the intersection and replaces corners in the route to obtain the final planned route, completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

[0102] The path planning method and system for autonomous operation of underground loaders provided by this invention extracts and processes information on the autonomous operation area of ​​the target underground loader, and performs collision detection, U-turn detection and V-shaped U-turn path construction based on safety threshold settings. This not only realizes path planning for autonomous operation of underground loaders, but also results in smooth path curvature, higher reliability, and better feasibility and accuracy. Attached Figure Description

[0103] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0104] Figure 2 This is a schematic diagram illustrating the construction of the V-shaped U-turn path in the method of the present invention; wherein, Figure 2 (a) is a schematic diagram of the original path corresponding to case 1-1. Figure 2 (b) is a schematic diagram of the V-shaped U-turn path construction corresponding to scenario 1-1. Figure 2 (c) is a schematic diagram of the original path corresponding to cases 1-2. Figure 2 (d) is a schematic diagram of the V-shaped U-turn path construction corresponding to scenarios 1-2. Figure 2 (e) is a schematic diagram of the original path corresponding to case 2. Figure 2 (f) is a schematic diagram of the V-shaped U-turn path construction corresponding to case 2.

[0105] Figure 3 This is a measured map of a mine, representing an embodiment of the method of the present invention; wherein, Figure 3 (a) is a schematic diagram of a single alleyway scene. Figure 3 (b) is a schematic diagram of a multi-branched alleyway scene. Figure 3 (c) is a schematic diagram of a complex network of alleyways.

[0106] Figure 4 This is a schematic diagram showing the comparison results of the total path length and path structure in an embodiment of the method of the present invention.

[0107] Figure 5 This is a schematic diagram showing the comparison results between the distance to the obstacle and the method embodiment of the present invention.

[0108] Figure 6 This is a schematic diagram comparing the running time of embodiments of the method of the present invention; wherein, Figure 6 (a) is a schematic diagram illustrating the running time of the A* algorithm. Figure 6 (b) is a schematic diagram illustrating the running time of the PRT* algorithm. Figure 6 (c) is a schematic diagram illustrating the running time of the HHA algorithm. Figure 6 (d) is a schematic diagram of the running time of the method of the present invention. Figure 6 (e) is a schematic diagram comparing average time and path length.

[0109] Figure 7 This is a schematic diagram of the functional modules of the system of the present invention. Detailed Implementation

[0110] like Figure 1 The diagram shown is a flowchart of the method of the present invention: The path planning method for autonomous operation of underground loader disclosed in this invention includes the following steps:

[0111] S1. Obtain map data of the autonomous operating area of ​​the target underground loader; specifically including the following steps:

[0112] Obtain map data information of the autonomous operating area of ​​the target underground loader;

[0113] The obtained map data is rasterized, with the grid containing the obstacle set to 1 and the grid that the target underground loader can pass through set to 0, thus obtaining a binary occupancy grid map of the target underground loader's autonomous operating area.

[0114] S2. Based on the width of the target underground loader and the set safety margin, the skeleton extraction algorithm is used to obtain the skeleton information of the autonomous operation area using the map data information obtained in step S1; specifically, it includes the following steps:

[0115] Set an obstacle set Obstacle set This includes all grid cells in a binary occupied grid map that have a value of 1;

[0116] Calculate obstacle radius for , The width of the target underground loader. The set safety threshold;

[0117] Based on the obtained obstacle radius Using a binary occupancy grid map, we can obtain areas that can be safely traversed within the autonomous operation area. for ,in For binary occupied grid maps, Point To the obstacle set The distance;

[0118] The skeleton extraction algorithm (such as the Zhang–Suen algorithm) is used to extract the region. The skeleton, the skeleton information of the autonomous operation area ;

[0119] S3. Based on the skeleton information obtained in step S2, a grid-based search method is used to obtain the shortest reachable path from the set starting point to the set ending point; specifically, the following steps are included:

[0120] Obtain the starting point of the target underground loader in its autonomous operating area. and the end point of the task ;

[0121] In skeleton information Get distance to the starting point of the operation The nearest point is used as the starting point. ;

[0122] In skeleton information Get distance to the end point of the operation The nearest point is used as the endpoint. ;

[0123] A grid-based search method (such as the A* algorithm) is used to search the skeleton information. Searching on the [top] will yield results starting from the set starting point. To the set destination Shortest reachable path ;

[0124] S4. Based on the heading angle changes at each point on the path, extract the turning points on the shortest reachable path obtained in step S3, and connect the set starting point, each turning point, and the set ending point sequentially to obtain the initial polyline path; specifically, this includes the following steps:

[0125] Set sampling span for ,in The average width of the alleyway. This is the grid resolution; this value strikes a balance between noise suppression and sensitivity in detecting small-radius curves.

[0126] For the shortest reachable path The i-th point in ,exist Within the range, set the i-th point. rear heading angle for The i-th point Forward heading angle for Where N is the shortest reachable path. The total number of points It is a two-parameter arctangent function;

[0127] The local heading change was calculated. for ;in, To normalize the angle to operate;

[0128] For local heading changes Make a judgment: If Then the point Marked as a turning point; if Then point It is not a turning point; The threshold value is set for determining the turning point;

[0129] Represent all the marked inflection points as The turning point is on the shortest reachable path. The corresponding index in the table represents what is meant by Then it exists At all inflection points, the index difference should not exceed the sampling span. The turning point is divided into the same curve segment. For each curve segment Take the midpoint as Heart ;

[0130] According to the index, check all the bends. Sort the sequences to obtain the core sequence. for , The total number of apex bends;

[0131] Set the starting point , Bend Core Sequence Each apex and set finish line Connect them sequentially to obtain the initial polyline path. ;

[0132] However, the initial polyline path obtained at this time The following issues exist: Due to the limitation of the BCD threshold, some small curves below the threshold are not explicitly captured, which may lead to collisions with obstacles; moreover, due to the influence of irregular alley geometry and intersection structure, the path may contain redundant vertices.

[0133] S5. Based on the geometric positional relationship between each line segment in the initial polyline path and the autonomous operation area, interpolate and trim the initial polyline path to obtain a collision-free polyline from a set starting point to a set ending point within the autonomous operation area; specifically including the following steps:

[0134] Interpolation stage: from the initial polyline path Starting from the first point, proceed sequentially to each adjacent point. , Line segments Make a judgment: if Not all located in the region Inside, then at the point and points Insert a point between , Shortest reachable path Points and indexes in Values ,in For point Shortest reachable path The index value in For point Shortest reachable path The index value in The floor symbol;

[0135] Repeat the above interpolation steps until all line segments formed by adjacent points lie within the region. Within, an initial collision-free polyline is obtained. This step ensures that the path is collision-free within the safe domain.

[0136] Cutting stage:

[0137] Initialize the most recently retained vertex for ;

[0138] From the initial collision-free polygonal line The first point in Begin by testing the initial collision-free polylines sequentially. Each vertex in Perform the detection: If starting from the current most recently retained vertex arrive line segments All located in the region If inside, then delete the vertex. Otherwise, preserve the vertices. and the current most recently retained vertex Update to vertices ;

[0139] Repeat the above cutting steps to finally obtain a collision-free zigzag line. This step can reduce redundancy.

[0140] S6. Determine if a vehicle is making a U-turn based on the positional relationship between the direction vectors of the set starting point and the set ending point; detect intersections based on the positional relationship between the neighborhood of each point on the shortest reachable path and the shortest reachable path itself; specifically including the following steps:

[0141] Isolated tunnels are often too narrow for unmanned loaders to make a V-shaped U-turn in place. When the difference in attitude between the starting and ending points necessitates a reversal of the travel direction, this solution requires performing the maneuver in the form of a "V-shaped U-turn" at a nearby intersection. This process requires expanding the already optimized polygonal line by adding an intersection vertex and a dedicated U-turn point; specifically as follows... Figure 2 As shown;

[0142] Determining if a vehicle is making a U-turn:

[0143] Set starting point The direction vector of the loader at the location is represented as: Set the endpoint The direction vector of the loader at the location is represented as: Collision-free zigzag lines Represented as , direction vector at the starting point of the path for Path endpoint direction vector Represented as The direction in which the loader faces is defined as the direction of the loader's bucket.

[0144] like If so, it is determined that assistance is needed to turn the vehicle around;

[0145] Intersection detection:

[0146] Shortest reachable path Perform intersection detection: Set c as the shortest reachable path. The point on, For the 8-neighborhood of c; when More than The neighborhood is the shortest reachable path If the above is true, then c will be set as a candidate point for the intersection;

[0147] S7. Based on the intersection detection results obtained in step S6, and according to the corner information at each intersection, construct the V-shaped U-turn path at the intersection and replace the corners in the route to obtain the final route planning, thus completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point; including the following steps:

[0148] Based on the intersection detection results obtained in step S6, the system first searches for U-turn points where the path has already turned at the intersection and constructs the corresponding U-turn structure. If no U-turn point can be found, a U-turn and back maneuver mode centered on the intersection is set, and a V-shaped U-turn path is constructed at the intersection. Finally, all corners in the route are constructed using CAC structures to obtain the final route planning, thus completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

[0149] In practice, the following situations are included:

[0150] Scenario 1: The set path has already turned at the intersection:

[0151] Based on the intersection detection results obtained in step S6, select any intersection candidate point. In collision-free fold lines Get the distance to the intersection candidate point nearest point And obtain the local corner triplet. ;

[0152] In this case, there is no collision-free broken line. Since there is already a corner at the intersection, this method tends to "embed" the V-shaped U-turn into this existing turn in order to avoid adding extra macro-turns;

[0153] Set the far endpoint: from the intersection candidate point Start, along the skeleton Shortest reachable path Traverse outwards along the other skeleton branches, enumerating the far endpoints of each branch. ; distal end The definition is: the shortest reachable path on the branch. Connected endpoints to the far endpoint All line segments are located in the region Within, and with the longest length in the branch direction; where, if it is a maneuvering mode of first turning into the branch and then completing the reverse, then the shortest reachable path on the branch. The connected endpoints are If the driving pattern involves first proceeding straight through the intersection into the side lane and then turning back, then the shortest reachable path on the branch is... The connected endpoints are ;

[0154] Scenario 1-1: Setting a maneuver mode where the vehicle first turns and enters a side tunnel before reversing:

[0155] In this maneuver mode, the vehicle first turns into a side branch at the intersection, then proceeds along the side lane, shifts gears and reverses in the side lane, and finally returns to the main lane.

[0156] In skeleton information Several distant endpoints are obtained by searching on the lateral branches other than the shortest reachable path R. For each remote endpoint The intersection point is obtained. for and the intersection As the new corner apex of the intersection;

[0157] sequentially at all remote endpoints The triplet it belongs to Construct the CAC structure and obtain the corresponding tangent point. If the tangent point is located on the line segment... If the inside is selected, then that tangent point is chosen as the turning point. ;

[0158] If a U-turn point that meets the conditions exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... pass Turn to ,exist The driving direction is reversed at this point, and then along the straight segment from Back At this point, replace the corresponding corner with: turning section. and a section at the U-turn point The opposite line segment at the location ;

[0159] If there is no U-turn point that meets the conditions If so, continue with the subsequent steps;

[0160] Scenario 1-2: Setting a driving mode where the vehicle first goes straight through the intersection, enters the side lane, and then turns back in the opposite direction:

[0161] In this maneuvering mode, the vehicle first goes straight through the intersection and enters the side lane, then reverses, and finally turns back to the main lane.

[0162] In skeleton information Several distant endpoints are obtained by searching on the lateral branches other than the shortest reachable path R. For each remote endpoint The intersection point is obtained. for ;

[0163] sequentially at all remote endpoints The triplet it belongs to Construct the CAC structure on the line segment and obtain the corresponding tangent point. If the tangent point is located on the line segment... If the inside is selected, then that tangent point is chosen as the turning point. ;

[0164] If a U-turn point exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... Drive straight to ,exist The driving direction is reversed at this point, and then... Turn and return At this point, replace the corresponding corner with a straight line segment. and a period of time After reversing, pass Turning section ;

[0165] If there is no U-turn point If so, continue with the following steps;

[0166] Scenario 2: Set up a U-turn / reverse maneuver mode centered on the intersection:

[0167] Based on the intersection detection results obtained in step S6, select any intersection candidate point. In collision-free fold lines Get the distance to the intersection candidate point The two nearest neighboring points and ;

[0168] Candidate points at intersections direct as and set the remote endpoint Shortest reachable path The connected endpoints are ;

[0169] Skeletal information corresponding to lateral branches The search yielded several remote endpoints. ;

[0170] For each remote endpoint There are two sets of turning triples. and In the triplet Constructing a CAC structure at the point yields a line segment. Several first tangent points on; then triplet Constructing a CAC structure at the point yields a line segment. Several second tangent points on the line segment; among several first tangent points and several second tangent points, select those where both the first and second tangent points lie on the line segment. Inside, and at a distance The more distant tangent point is used as the turning point. ;

[0171] If a U-turn point exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... pass Turn into the side alley and reach the U-turn point. Afterwards, The driving direction is reversed at this point, and then... After turning back ;

[0172] If there is no U-turn point If so, it is determined that under the current safety constraints, V-shaped turns assisted by intersections are not supported;

[0173] At this point, it can be ensured that the curvature and sharpness of the entire V-shaped turn section remain within the kinematic limits of the underground loader, and the entire turn maneuver is always within the safe range.

[0174] The remaining corners of the route are then constructed using CAC structures to obtain the final path planning route.

[0175] The construction of the CAC structure includes the following steps:

[0176] The transition section is constructed using a symmetrical spiral curve-circular arc-spiral curve profile; curvature constraints are satisfied during the construction process. and sharpness constraints ;in, For curvature, For the set curvature threshold, Sharpness is defined as the rate of change of curvature. The set sharpness threshold;

[0177] When constructing a spiral curve, the following methods are used: Perform curve construction; during the construction process, solve... The parametric triplet of the following cyclotron curve - circular arc - cyclotron curve and tangent point position This allows for the determination of the corresponding spiral curve-circular arc-spiral curve substitution segment; among which, The length of the spiral curve, The length of the arc. The maximum curvature of the cycloid is denoted by . The coordinates of the point of tangency between the straight section entering the curve and the first spiral curve are given. The coordinates are the points of tangency between the straight section exiting the curve and the second spiral curve.

[0178] In practice, the following steps can be taken:

[0179] Step 1: Set the entry heading angle at the corner of the broken line as... The heading angle at the exit of the curve is The heading change angle was calculated. for ;

[0180] For sharpness constant The curve segment exists ,in Let be the curvature at arc length s. The maximum curvature of the cyclotron curve Values ;

[0181] Heading changes caused by a single-sided spiral slope Represented as Therefore, the heading changes caused by the two symmetrical spiral curve slopes are obtained. for The curvature of the middle arc segment is constant. And the corresponding arc length is The corresponding change in heading Values ; Obtain the total heading change for ;

[0182] set up ;

[0183] Step 2: Use the pure spiral curve mode:

[0184] First, try using the pure spiral curve mode. If that fails, then use the spiral curve-circular arc-spiral curve mode.

[0185] at this time, ,but The peak curvature required for the pure cyclotron mode is ;

[0186] like satisfy Then it is determined that the change in course can be achieved through two segments with sharpness of [missing information]. The symmetric cyclotron curve is realized; the corresponding solution is:

[0187]

[0188]

[0189]

[0190] In the formula for The symbol;

[0191] like If the course change cannot be achieved through two segments with a sharpness of [value missing], then it is determined that the course change cannot be achieved through two segments with a sharpness of [value missing]. The symmetrical cyclotron curve is achieved, and subsequent steps are continued;

[0192] Step 3: Use the spiral curve-circular arc-spiral curve pattern:

[0193] set up At this point, the slope length is fixed. ;

[0194] The remaining heading changes are completed by constant curvature circular arcs, represented as:

[0195]

[0196]

[0197] In the formula The heading change included in the circular arc segment;

[0198] Step 4: Settings On point The heading angle at the point of entry into the curve is The heading angle at the exit of the curve is Set the unit tangent vector for the curve. for unit tangent vector at the exit of the curve for ; Point of tangency at the bend for The point of tangency at the bend for ;in, for arrive distance, for arrive The distance;

[0199] Interchange the X-axis with Alignment, with the origin at the starting point of the entry curve segment. At this time, a parameter is The spiral curve-circular arc-spiral curve segment, from arrive The generated displacement vector Represented as , This represents the displacement along the x-axis of the local coordinate system. This represents the displacement along the y-axis in the local coordinate system; the calculation yields... and will according to Rotate back to the global coordinate system to obtain the displacement vector in the global coordinate system. for ,in A two-dimensional rotation matrix; a displacement vector. It can be calculated using the standard Fresnel integral of the cyclotron curve and the geometric formula of the circular arc;

[0200] exist ;

[0201] Will Projected to and the corresponding normal vector Above: Order ,Will exist Basis expansion yields the results. and .

[0202] In practical implementation, the CAC tangent construction is applied to... All corners, thus obtaining a global [line / path]. The path is continuous, and each turning maneuver is as compact as possible, achieving the minimum space required by the vehicle's kinematic limits.

[0203] The planned path obtained by the method of this invention is mainly straight with fewer turns, which is more conducive to subsequent trajectory tracking and control. Moreover, the path strictly follows the limits of maximum curvature and maximum sharpness, making the path more feasible. At the same time, the solution of this invention solves the problem of the loader needing to turn around, maintains a large safety clearance from obstacles and a short path length, and has a short planning time.

[0204] The effects of the method of the present invention will be further illustrated below with reference to an embodiment:

[0205] In a measured map of a certain mine (such as...) Figure 3 As shown in the figure, using 12 trajectory planning tasks with different start and end points, the A* method, RRT* method, and the recently developed advanced method HHA (described in application document No. 202511299246.8), commonly used in trajectory planning, were compared. Each task was repeated 20 times. The A* method is the scheme proposed by Hart in his 1968 paper "A Formal Basis for the Heuristic Determination of Minimum Cost Paths"; the RRT* method is the scheme proposed by Karaman in his 2010 paper "Optimalkinodynamic motion planning using incremental sampling-based methods".

[0206] The metrics include: path length L and normalized length. (where L is the path length, (Euclidean distance between start and end points); based on threshold , The components of the segment (straight line / circular arc / curvature variation segment); the distribution of curvature and sharpness and the constraint satisfaction rate (falling on...) , The proportion within the range); net distance statistics and security satisfaction rate based on Euclidean distance transformation (net distance greater than...). The percentage of the total path length; and the end-to-end runtime (excluding I / O) from planner call to output of the final path. Unless otherwise specified, all statistics are aggregated on 12 start-end pairs by default.

[0207] Path length and segment composition: results are as follows Figure 4 As shown, in all 12 non-V-turn start-end point tasks, the proposed method, while maintaining an overall path length close to that of the grid baseline method, generates trajectories that are significantly dominated by straight line segments. The average normalized path length of the proposed method is... The median value is 1.076 (1.033), slightly better than A* and HHA (both around 1.09), and significantly smaller than RRT* (around 1.26). In terms of composition, the straight segments in this invention's scheme account for an average of 87.3%, while curvature variation segments account for only 12.1%; whereas baseline methods typically allocate only 20–50% of the path length to straight segments, with the remainder consumed by turns. The results show that in multi-branched tunnels, this invention's scheme often eliminates a large number of small turns while achieving shorter paths than A*; in grid-like tunnels, this invention occasionally accepts a length "premium" of about 2–3% to retain longer straight segments, which is more beneficial for subsequent trajectory tracking.

[0208] Kinematic feasibility (curvature and sharpness): The fit to the turning limit constraints is very tight. For all non-U-turn tasks, curvature and sharpness remain within the preset boundaries almost throughout the entire path, with an average constraint satisfaction rate of 100%. In contrast, A*, RRT*, and HHA exceed the limits for a significant portion of the arc length. The average curvature amplitude of the proposed method is approximately 0.013, and the average sharpness amplitude is approximately 0.0048, both significantly lower than the baseline methods; peak values ​​are also strongly suppressed: the average maximum... Approximately 0.12, average maximum The peak value is approximately 0.038, while the baseline method typically produces a peak value an order of magnitude higher. As shown in Table 1, the combination of vertex optimization and curvature-continuous C–A–C transition effectively eliminates steering rate shocks, thereby generating a kinematically “clean” trajectory.

[0209]

[0210] Net distance and safety: Results as follows Figure 5 As shown, while merging turns and shortening the path, the solution of this invention still maintains sufficient obstacle clearance. The average clearance is 2.77m, and the safety satisfaction rate is close to 100%, which is comparable to or better than A* and RRT* in most scenarios. Figure 5 In the study, the few samples close to the threshold lasted for a very short time, which can be attributed to errors caused by discretization and resampling, and has no substantial impact on the overall security conclusion.

[0211] Runtime: Under the same hardware and software environment, the proposed solution is a fast and stable deterministic planner. Its average runtime is approximately 17.4 ms (median 16.9 ms), while A* is approximately 20.2 ms, HHA is approximately 13.7 ms, and the proposed solution is approximately 16.6 ms. The time distribution of RRT* exhibits a significant heavy-tailed characteristic: the average is approximately 8 ms, but the maximum value can reach 84 ms. Therefore, its average value needs to be understood in conjunction with the success rate and worst-case latency. Detailed runtime statistics are shown in Table 2.

[0212]

[0213] Figure 6 Box plots of 12 sets of start-endpoint pairs for each method are shown, along with the scatter plot relationship between average running time and path length. The scalability of the proposed solution is good as the complexity of the scene increases. In single-lane, multi-branch, and grid layouts, the average running time of the proposed solution remains within 20ms, which is approximately 5%, 7%, and 28% faster than A*, respectively.

[0214] like Figure 7 The diagram shows the functional modules of the system of the present invention: The system disclosed in this invention, which implements the path planning method for autonomous operation of underground shovel loader, includes a data acquisition module, a skeleton extraction module, a reachability extraction module, a polyline extraction module, a collision-free extraction module, an intersection detection module, and a path planning module; these modules are connected in series. The data acquisition module acquires map data information of the autonomous operation area of ​​the target underground shovel loader and uploads the data information to the skeleton extraction module. The skeleton extraction module, based on the received data information and the acquired map data information, uses a skeleton extraction algorithm to obtain the skeleton information of the autonomous operation area, based on the width of the target underground shovel loader and a set safety margin, and uploads the data information to the reachability extraction module. The reachability extraction module, based on the received data information and the obtained skeleton information, uses a grid-based search method to obtain the shortest reachable path from a set starting point to a set ending point, and uploads the data information to the polyline extraction module. The polyline extraction module, based on the received data information and the obtained skeleton information, uses a grid-based search method to obtain the shortest reachable path from a set starting point to a set ending point, and uploads the data information to the polyline extraction module. Based on the heading angle changes of each point along the path, the system extracts the bend points on the shortest reachable path and connects the set starting point, each bend point, and the set ending point sequentially to obtain an initial polyline path. This data is then uploaded to the collision-free extraction module. The collision-free extraction module, based on the received data and the geometric positional relationship between each line segment in the initial polyline path and the autonomous operating area, interpolates and trims the initial polyline path to obtain a collision-free polyline from the set starting point to the set ending point within the autonomous operating area. This data is then uploaded to the intersection detection module. The intersection detection module, based on the received data and the positional relationship between the direction vectors of the set starting point and the set ending point, determines whether a vehicle needs to turn around. It also detects intersections based on the positional relationship between the neighborhood of each point on the shortest reachable path and the shortest reachable path itself, and uploads this data to the path planning module. The path planning module, based on the received data and the corner information at each intersection, constructs a V-shaped U-turn path at the intersection and replaces corners in the route to obtain the final planned route, completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

Claims

1. A path planning method for autonomous operation of underground loader, characterized in that... Includes the following steps: S1. Obtain map data information of the autonomous operating area of ​​the target underground loader; S2. Based on the width of the target underground loader and the set safety margin, the skeleton extraction algorithm is used to obtain the skeleton information of the autonomous operation area from the map data information obtained in step S1. S3. Based on the skeleton information obtained in step S2, a grid-based search method is used to obtain the shortest reachable path from the set starting point to the set ending point; S4. Based on the heading angle changes at each point on the path, extract the turning points on the shortest reachable path obtained in step S3, and connect the set starting point, each turning point, and the set ending point sequentially to obtain the initial polyline path; specifically, this includes the following steps: Set sampling span for ,in The average width of the alleyway. Raster resolution; For the shortest reachable path The i-th point in ,exist Within the range, set the i-th point. rear heading angle for The i-th point Forward heading angle for Where N is the shortest reachable path. The total number of points It is a two-parameter arctangent function; The local heading change was calculated. for ;in, To normalize the angle to operate; For local heading changes Make a judgment: If Then the point Marked as a turning point; if Then point It is not a turning point; The threshold value is set for determining the turning point; Represent all the marked inflection points as The turning point is on the shortest reachable path. The corresponding index in the table represents what is meant by Then it exists At all inflection points, the index difference should not exceed the sampling span. The turning point is divided into the same curve segment. For each curve segment Take the midpoint as Heart ; According to the index, check all the bends. Sort the sequences to obtain the core sequence. for , The total number of apex bends; Set the starting point , Bend Core Sequence Each apex and set finish line Connect them sequentially to obtain the initial polyline path. ; S5. Based on the geometric positional relationship between each line segment in the initial polyline path and the autonomous operation area, interpolate and trim the initial polyline path to obtain a collision-free polyline from a set starting point to a set ending point within the autonomous operation area; specifically including the following steps: Interpolation stage: from the initial polyline path Starting from the first point, proceed sequentially to each adjacent point. , Line segments Make a judgment: if Not all located in the region Inside, then at the point and points Insert a point between , Shortest reachable path Points and indexes in Values ,in For point Shortest reachable path The index value in For point Shortest reachable path The index value in The floor symbol; Repeat the above interpolation steps until all line segments formed by adjacent points lie within the region. Within, an initial collision-free polyline is obtained. ; Cutting stage: Initialize the most recently retained vertex for ; From the initial collision-free polygonal line The first point in Begin by testing the initial collision-free polylines sequentially. Each vertex in Perform the detection: If starting from the current most recently retained vertex arrive line segments All located in the region If inside, then delete the vertex. Otherwise, preserve the vertices. and the current most recently retained vertex Update to vertices ; Repeat the above cutting steps to finally obtain a collision-free zigzag line. ; S6. Determine if a vehicle is making a U-turn based on the positional relationship between the direction vectors of the set starting point and the set ending point; detect intersections based on the positional relationship between the neighborhood of each point on the shortest reachable path and the shortest reachable path itself; specifically including the following steps: Determining if a vehicle is making a U-turn: Set starting point The direction vector of the loader at the location is represented as: Set the endpoint The direction vector of the loader at the location is represented as: Collision-free zigzag lines Represented as , direction vector at the starting point of the path for Path endpoint direction vector Represented as The direction in which the loader faces is defined as the direction of the loader's bucket. like If so, it is determined that assistance is needed to turn the vehicle around; Intersection detection: Shortest reachable path Perform intersection detection: Set c as the shortest reachable path. The point on, For the 8-neighborhood of c; when More than The neighborhood is the shortest reachable path If the above is true, then c will be set as a candidate point for the intersection; S7. Based on the intersection detection results obtained in step S6, and based on the corner information at each intersection, construct the V-shaped U-turn path at the intersection and replace the corners in the route to obtain the final route planning line, thus completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

2. The path planning method for autonomous operation of underground loaders according to claim 1, characterized in that... Step S1 specifically includes the following steps: Obtain map data information of the autonomous operating area of ​​the target underground loader; The obtained map data is rasterized, with the grid containing the obstacle set to 1 and the grid that the target underground loader can pass through set to 0, thus obtaining a binary occupancy grid map of the target underground loader's autonomous operating area. Step S2 specifically includes the following steps: Set an obstacle set Obstacle set This includes all grid cells in a binary occupied grid map that have a value of 1; Calculate obstacle radius for , The width of the target underground loader. The set safety threshold; Based on the obtained obstacle radius Using a binary occupancy grid map, we can obtain areas that can be safely traversed within the autonomous operation area. for ,in For binary occupied grid maps, Point To the obstacle set The distance; The skeleton extraction algorithm is used to extract the region. The skeleton, the skeleton information of the autonomous operation area ; Step S3 specifically includes the following steps: Obtain the starting point of the target underground loader in its autonomous operating area. and the end point of the task ; In skeleton information Get distance to the starting point of the operation The nearest point is used as the starting point. ; In skeleton information Get distance to the end point of the operation The nearest point is used as the endpoint. ; A grid-based search method is used to search the skeleton information. Searching on the [top] will yield results starting from the set starting point. To the set destination Shortest reachable path .

3. The path planning method for autonomous operation of underground loader as described in claim 2, characterized in that... Step S7 includes the following steps: Based on the intersection detection results obtained in step S6, the system first searches for U-turn points where the path has already turned at the intersection and constructs the corresponding U-turn structure. If no U-turn point can be found, a U-turn and back maneuver mode centered on the intersection is set, and a V-shaped U-turn path is constructed at the intersection. Finally, all corners in the route are constructed using CAC structures to obtain the final path planning route, thus completing the path planning for the underground loader's autonomous operation from the set starting point to the set ending point.

4. The path planning method for autonomous operation of underground loader as described in claim 3, characterized in that... Step S7 specifically includes the following situations: Scenario 1: The set path has already turned at the intersection: Based on the intersection detection results obtained in step S6, select any intersection candidate point. In collision-free fold lines Get the distance to the intersection candidate point nearest point And obtain the local corner triplet. ; Set the far endpoint: from the intersection candidate point Start, along the skeleton Shortest reachable path Traverse outwards along the other skeleton branches, enumerating the far endpoints of each branch. ; Distal point The definition is: the shortest reachable path on the branch. Connected endpoints to the far endpoint All line segments are located in the region Within, and with the longest length in the branch direction; where, if it is a maneuvering mode of first turning into the branch and then completing the reverse, then the shortest reachable path on the branch. The connected endpoints are If the driving pattern involves first proceeding straight through the intersection into the side lane and then turning back, then the shortest reachable path on the branch is... The connected endpoints are ; Scenario 1-1: Setting a maneuver mode where the vehicle first turns and enters a side tunnel before reversing: In this maneuver mode, the vehicle first turns into a side branch at the intersection, then proceeds along the side lane, shifts gears and reverses in the side lane, and finally returns to the main lane. In skeleton information Several distant endpoints are obtained by searching on the lateral branches other than the shortest reachable path R. For each remote endpoint The intersection point is obtained. for and the intersection As the new corner apex of the intersection; sequentially at all remote endpoints The triplet it belongs to Construct the CAC structure and obtain the corresponding tangent point. If the tangent point is located on the line segment... If the inside is selected, then that tangent point is chosen as the turning point. ; If a U-turn point that meets the conditions exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... pass Turn to ,exist The driving direction is reversed at this point, and then along the straight segment from Back At this point, replace the corresponding corner with: turning section. and a section at the U-turn point The opposite line segment at the location ; If there is no U-turn point that meets the conditions If so, continue with the subsequent steps; Scenario 1-2: Setting a driving mode where the vehicle first goes straight through the intersection, enters the side lane, and then turns back in the opposite direction: In this maneuvering mode, the vehicle first goes straight through the intersection and enters the side lane, then reverses, and finally turns back to the main lane. In skeleton information Several distant endpoints are obtained by searching on the lateral branches other than the shortest reachable path R. For each remote endpoint The intersection point is obtained. for ; sequentially at all remote endpoints The triplet it belongs to Construct the CAC structure on the line segment and obtain the corresponding tangent point. If the tangent point is located on the line segment... If the inside is selected, then that tangent point is chosen as the turning point. ; If a U-turn point exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... Drive straight to ,exist The driving direction is reversed at this point, and then... Turn and return At this point, replace the corresponding corner with a straight line segment. and a period of time After reversing, pass Turning section ; If there is no U-turn point If so, continue with the subsequent steps; Scenario 2: Set up a U-turn / reverse maneuver mode centered on the intersection: Based on the intersection detection results obtained in step S6, select any intersection candidate point. In collision-free fold lines Get the distance to the intersection candidate point The two nearest neighboring points and ; Candidate points at intersections direct as and set the remote endpoint Shortest reachable path The connected endpoints are ; Skeletal information corresponding to lateral branches The search yielded several remote endpoints. ; For each remote endpoint There are two sets of turning triples. and In the triplet Constructing a CAC structure at the point yields a line segment. Several first tangent points on; then triplet Constructing a CAC structure at the point yields a line segment. Several second tangent points on the line segment; among several first tangent points and several second tangent points, select those where both the first and second tangent points lie on the line segment. Inside, and at a distance The more distant tangent point is used as the turning point. ; If a U-turn point exists Then the V-shaped U-turn path is completed; at this time, the vehicle departs from... pass Turn into the side alley and reach the U-turn point. Afterwards, The driving direction is reversed at this point, and then... After turning back ; If there is no U-turn point If so, it is determined that under the current safety constraints, V-shaped turns assisted by intersections are not supported; The remaining corners of the route are then constructed using CAC structures to obtain the final path planning route.

5. The path planning method for autonomous operation of underground loader according to claim 4, characterized in that... The construction of the CAC structure includes the following steps: The transition section is constructed using a symmetrical spiral curve-circular arc-spiral curve profile; curvature constraints are satisfied during the construction process. and sharpness constraints ;in, For curvature, For the set curvature threshold, Sharpness is defined as the rate of change of curvature. The set sharpness threshold; When constructing a spiral curve, the following methods are used: Perform curve construction; during the construction process, solve... The parametric triplet of the following cyclotron curve - circular arc - cyclotron curve and tangent point position This allows for the determination of the corresponding spiral curve-circular arc-spiral curve substitution segment; among which, The length of the spiral curve, The length of the arc. The maximum curvature of the cycloid is denoted by . The coordinates of the point of tangency between the straight section entering the curve and the first spiral curve are given. The coordinates are the points of tangency between the straight section exiting the curve and the second spiral curve.

6. The path planning method for autonomous operation of underground loader according to claim 5, characterized in that... The construction of the CAC structure specifically includes the following steps: Step 1: Set the entry heading angle at the corner of the broken line as... The heading angle at the exit of the curve is The heading change angle was calculated. for ; For sharpness constant The curve segment exists ,in Let be the curvature at arc length s. The maximum curvature of the cyclotron curve Values ; Heading changes caused by a single-sided spiral slope Represented as Therefore, the heading changes caused by the two symmetrical spiral curve slopes are obtained. for The curvature of the middle arc segment is constant. And the corresponding arc length is The corresponding change in heading Values ; Obtain the total heading change for ; set up ; Step 2: Use the pure spiral curve mode: at this time, ,but The peak curvature required for the pure cyclotron mode is ; like satisfy Then it is determined that the change in course can be achieved through two segments with sharpness of [missing information]. The symmetric cyclotron curve is realized; the corresponding solution is: In the formula for The symbol; like If the course change cannot be achieved through two segments with a sharpness of [value missing], then it is determined that the course change cannot be achieved through two segments with a sharpness of [value missing]. The symmetrical cyclotron curve is achieved, and subsequent steps are continued; Step 3: Use the spiral curve-circular arc-spiral curve pattern: set up At this point, the slope length is fixed. ; The remaining heading changes are completed by constant curvature circular arcs, represented as: In the formula The heading change included in the circular arc segment; Step 4: Settings On point The heading angle at the point of entry into the curve is The heading angle at the exit of the curve is ; Set the unit tangent vector for the curve. for unit tangent vector at the exit of the curve for ; Point of tangency at the bend for The point of tangency at the bend for ;in, for arrive distance, for arrive The distance; Interchange the X-axis with Alignment, with the origin at the starting point of the entry curve segment. At this time, a parameter is The spiral curve-circular arc-spiral curve segment, from arrive The generated displacement vector Represented as , This represents the displacement along the x-axis of the local coordinate system. This represents the displacement along the y-axis in the local coordinate system; the calculation yields... and will according to Rotate back to the global coordinate system to obtain the displacement vector in the global coordinate system. for ,in It is a two-dimensional rotation matrix; exist ; Will Projected to and the corresponding normal vector Above: Order ,Will exist Basis expansion yields the results. and .

7. A system for implementing the path planning method for autonomous operation of underground loaders as described in any one of claims 1 to 6, characterized in that... It includes a data acquisition module, a skeleton extraction module, a reachability extraction module, a polyline extraction module, a collision-free extraction module, an intersection detection module, and a path planning module; these modules are connected in series; the data acquisition module is used to acquire map data information of the autonomous operating area of ​​the target underground loader and upload the data information to the skeleton extraction module. The skeleton extraction module is used to extract the skeleton information of the autonomous operation area based on the received data information and the acquired map data information, based on the width of the target underground loader and the set safety margin, and upload the data information to the skeleton extraction module. The reachability extraction module is used to obtain the shortest reachable path from a set starting point to a set ending point based on the received data and the obtained skeleton information, using a grid-based search method, and then uploads the data to the polyline extraction module. The polyline extraction module is used to extract the turning points on the shortest reachable path based on the changes in heading angles of each point on the path, according to the received data information. It then connects the set starting point, each turning point, and the set ending point in sequence to obtain the initial polyline path and uploads the data information to the collision-free extraction module. The collision-free extraction module is used to interpolate and trim the initial polyline path based on the received data information and the geometric positional relationship between each line segment in the initial polyline path and the autonomous operation area to obtain a collision-free polyline from the set starting point to the set ending point within the autonomous operation area and uploads the data information to the intersection detection module. The intersection detection module is used to determine vehicle U-turns based on the received data and the positional relationship between the direction vectors of the set starting point and the set ending point. It also performs intersection detection based on the positional relationship between the neighborhood of each point on the shortest reachable path and the shortest reachable path, and uploads the data to the path planning module. The path planning module is used to construct V-shaped U-turn paths at intersections and replace corners in the route based on the received data and corner information at each intersection, so as to obtain the final path planning route and complete the path planning for the underground loader to autonomously operate from the set starting point to the set ending point.

Images