Improved underground carry-scraper automatic driving reference path development method

Abstract

The invention discloses an improved underground carry-scraper automatic driving reference path development method. According to the method, a geometric model of the carry-scraper, performance parameters of the carry-scraper and a kinematic model of the carry-scraper are combined, the hinging angle and the hinging angular speed of the carry-scraper when the carry-scraper runs on a reference path are estimated through a nonlinear optimization method, and then the minimum distance between a carry-scraper body and a roadway wall is estimated. According to the method, the number of reference paths needing real vehicle verification is effectively reduced by excluding the reference paths with the hinge angle or the hinge angular speed exceeding the performance of the carry-scraper and the distance between the vehicle body and the roadway wall smaller than the safety distance, and the development efficiency of the reference paths is greatly improved. Furthermore, a visual path analysis method with an interaction function is adopted in the method, the method is simple and visual, and the requirements for the technology and experience of operators are reduced.

Description

An Improved Method for Developing Reference Paths for Automatic Driving of Downhole Loaders Technical Field

[0001] This invention relates to the field of automatic driving technology for underground loader operators, and more specifically to an improved method for developing reference paths for automatic driving of underground loader operators. Background Technology

[0002] Automatic driving technology for underground loaders plays a crucial role in the intelligent transformation of mines. Its core value lies in several aspects: enhanced safety, optimized production efficiency, green and low-carbon transformation, and structural restructuring of human resources. The full realization of automatic driving for underground loaders is an inevitable trend with broad application prospects.

[0003] Foreign manufacturers, such as Sandvik and Caterpillar, started earlier in the field of automated driving for underground loaders, and have already commercialized their products and successfully applied them in numerous mines worldwide. While domestic research in this area started later, it has caught up very rapidly. Compared to foreign manufacturers, domestic manufacturers have a better understanding of the actual conditions in domestic mines, and their products are more tailored to the needs of domestic users, helping to break the price monopoly of foreign manufacturers. Therefore, the development of automated driving technology for underground loaders in China is of great significance. Currently, domestic enterprises have achieved independent control over core technologies in the field of automated driving for loaders, and have made significant achievements in core patent technology systems, domestic equipment manufacturing, and the construction of an industrial ecosystem.

[0004] The automatic driving technology of underground loader operators mainly achieves automatic driving by tracking pre-planned reference paths. These reference paths require automatic driving verification (real-vehicle verification) under the supervision of on-board personnel before they can be applied to actual production. Real-vehicle verification is time-consuming and imposes a significant workload on the monitoring personnel. In short, the development of reference paths using existing methods is quite difficult. Therefore, this invention proposes an improved method for developing automatic driving reference paths for underground loader operators. This method effectively reduces the number of reference paths requiring real-vehicle verification, greatly improving the efficiency of reference path development. Furthermore, the method is simple and intuitive, significantly reducing the technical and experience requirements for operators and lowering the difficulty of reference path development. Summary of the Invention

[0005] To address the issues of lengthy development time, high workload for monitoring personnel, and high skill and experience requirements for operators in the development of reference paths for automated driving of underground loader operators, this invention proposes an improved method for developing reference paths for automated driving of underground loader operators.

[0006] This invention is achieved through the following technical solution: an improved method for developing reference paths for automatic driving of underground loader operators, wherein the reference path has a total of There are 1 waypoints, namely , ,…, ,in It is the starting point. It is the destination; each waypoint It has several parameters, including but not limited to the X coordinates of waypoints. Y coordinate The distance from the starting point to the waypoint Curvature at waypoints and the speed at that waypoint ; the distance and curvature The coordinates of waypoints are calculated numerically, a method known in the field of autonomous driving. The road segment starting from is denoted as ,in The path analysis process, as shown in Figure 4, consists of five steps: Step 1: Estimate vehicle acceleration. Assume the vehicle is on the road segment... The vehicle accelerates uniformly, and its acceleration is... Using the law of conservation of energy, we obtain the following equation: in The mass of the loader can be solved from the above equation. Therefore, the time required for the loader to pass through this section of road is... The second step is to estimate the articulation angle and its angular velocity. The kinematic model of the loader is well known in the field of loader autopilot and will not be elaborated here.

[0007] The curvature of the trajectory of the origin of the front vehicle coordinate system when the loader is moving is derived from the kinematic model. Satisfy the following equation: in It refers to the driving speed. It is the hinge angle. It is the hinge angular velocity. This refers to the wheelbase; scrapers have a special characteristic where the distances from the front and rear axles to the hinge point are the same, hence the term "wheelbase" for the scraper. (Waypoint) The hinge angle at the point is The hinge angular velocity is Combining the curvature equations described above, the problem of estimating the hinge angle and hinge angular velocity at each point on the reference path is transformed into an optimization problem (finding the minimum value) with the following non-negative objective function: Where the nonnegative constant The weights are used for the objective function. The physical meaning of the objective function is quite intuitive and will not be elaborated here. In one specific implementation of this invention, Satisfactory results can be obtained when the value is 1000. The optimization problem can be set with various initial conditions and constraints. Furthermore, the optimization problem can be solved using commercial or open-source nonlinear optimization tools. This invention uses the open-source CasaADi / Ipopt library, which yields satisfactory results.

[0008] Step 3: Estimate the poses of the front and rear vehicle bodies in the map coordinate system at any road point on the reference path. The origin of the front vehicle body reference system is located at this road point, the X-axis direction is the tangent direction of the reference path at this road point, and the Y-axis direction is 90 degrees counterclockwise from the X-axis direction. Therefore, the pose of the front vehicle body reference system in the map coordinate system is known. Furthermore, using the hinge angle of the loader at this road point and the loader wheelbase, the pose of the rear vehicle body coordinate system in the map coordinate system can be obtained through coordinate transformation. The calculation method is well known in the field of autonomous driving and will not be elaborated here. In particular, this invention uses the open-source Eigen3 library for the relevant calculations of coordinate transformation.

[0009] Step 4: Estimating the minimum distance between the vehicle body and the tunnel wall. The following describes the method of estimating the minimum distance between the vehicle body and the tunnel wall using the front vehicle body as an example.

[0010] First, this invention models the vehicle body using polygons in the front and rear vehicle coordinate systems, respectively. Let the vertex coordinates of the polygons used in this model be... common Points, such as the polygonal model of the front vehicle body indicated by 130 in Figure 1; further, each side of the polygon is divided into equal parts with a length not exceeding Multiple line segments, The typical value is 0.1 meters; the coordinates of the vertices of the equally divided line segment are denoted as . common The third step has already obtained the poses of the front and rear vehicle coordinate systems in the map coordinate system. Therefore, the coordinates of the above vertices are projected into the map coordinate system through coordinate transformation to become... Furthermore, through calculation The minimum distance between the vehicle body and the tunnel wall is estimated by the minimum distance between the vehicle body and the tunnel wall.

[0011] This invention employs different calculation methods for raster maps and vector maps. For raster maps, a K-dimensional tree is first constructed using the center points of the alleyway wall raster, followed by the K-nearest neighbor algorithm for calculation. The minimum distance to the midpoint of the K-dimensional tree, this minimum distance is the point. The minimum distance to the tunnel wall is used as an estimate of the minimum distance from the vehicle body to the tunnel wall. Specifically, this invention uses the open-source FLANN library to implement the aforementioned K-dimensional tree construction and K-nearest neighbor search. For vector maps, the tunnel walls are represented by several line segments, denoted as... ;calculate With the above line segment set The minimum distance is used as an estimate of the minimum distance from the vehicle body to the tunnel wall; the distance from the point to the line segment is calculated using a mature algorithm in computational geometry, and the required calculation is performed using the open-source computational geometry algorithm library CGAL.

[0012] Step 5: Check if the constraints are met. For all waypoints on the reference path, compare the data obtained in the above analysis process with the vehicle constraints and safety constraints to check whether all waypoints on the reference path meet the constraint conditions.

[0013] Working Principle and Process: This invention improves upon the traditional two-step approach of developing reference paths for automated driving of underground loaders, which involves path planning and actual vehicle verification. A path analysis step is added before actual vehicle verification. Path analysis uses mathematical and physical methods to analyze the reference path, estimating the loader's speed, acceleration, articulation angle, articulation angular velocity, and distances between the front and rear vehicle bodies and the tunnel walls at any waypoint along the reference path. These data are compared with vehicle constraints and safety constraints. Only reference paths that meet the constraints are verified in actual vehicle testing. Reference paths that do not meet the constraints require adjustment through path planning, followed by re-analysis, until the constraints are met before actual vehicle verification can proceed.

[0014] The beneficial effects of this invention are as follows: Compared with existing reference path development methods, the improved reference path development method for automatic driving of underground loaders proposed in this invention adds a path analysis stage before actual vehicle verification. This path analysis stage eliminates most paths that loaders cannot automatically track, improving the efficiency of reference path development and reducing the workload of monitoring personnel. Furthermore, the path analysis adopts a visual and interactive working method, making operation simple and intuitive, and reducing the requirements for operators. Attached Figure Description

[0015] Figure 1 is a schematic diagram of an automatic driving scenario for an underground loader; where: 100 Reference path start point; 110 Reference path end point; 120 Waypoints on the reference path, whose information includes X and Y coordinates and driving speed, etc.; 130 Front body of the loader; 140 Rear body of the loader; 150 Hinge point between the front and rear bodies of the loader, the front and rear bodies rotate around the hinge point to achieve vehicle steering; 160 Coordinate system of the front body of the loader; 170 Coordinate system of the rear body of the loader; 180 Underground tunnel wall; Figure 2 is a schematic diagram of the existing reference path development process; Figure 3 is a schematic diagram of the improved reference path development process of this invention; Figure 4 is a schematic diagram of the reference path analysis process of this invention; Figure 5 is a reference path analysis screen of this invention; where: 500 Distance analysis sub-screen from front and rear bodies to tunnel wall; 501 Minimum distance curve from front body to tunnel wall; 502 Minimum distance curve from rear body to tunnel wall; 510 Hinge angle and angular velocity analysis sub-screen; 511 Hinge Angle Curve 512 Hinge Angular Velocity Curve 520 Acceleration Analysis Sub-screen 521 Velocity Curve 522 Acceleration Curve. Detailed Implementation

[0016] The present invention will be further described below with reference to the embodiments and accompanying drawings.

[0017] The automatic driving of underground loader is mainly achieved by tracking a pre-planned reference path. To facilitate the explanation of the specific content of this invention, Figure 1 provides a schematic diagram of an automatic driving scenario for an underground loader. In Figure 1, the loader, consisting of a front body (130) and a rear body (140), operates in a tunnel enclosed by tunnel walls (180). The automatic driving of the underground loader can use a grid map or a vector map, with the tunnel walls represented by grids or line segments in the map coordinate system. The front and rear bodies are hinged at a hinge point (150), and the loader steers by controlling the hinge angle. The reference coordinate system (160) logically attached to the front vehicle body is called the front vehicle reference system. Its origin is at the center point of the front axle of the vehicle, with its X-axis pointing forward, Y-axis pointing left, and Z-axis pointing upward. The reference system (170) logically attached to the rear vehicle body is called the rear vehicle reference system. Its origin is at the center point of the rear axle of the vehicle, with its X-axis pointing forward, Y-axis pointing left, and Z-axis pointing upward. The reference path consists of several discrete waypoints (120), including the starting point (100) and ending point (110) of the path. Each waypoint is accompanied by several parameters, including but not limited to its X-coordinate, Y-coordinate, curvature, and driving speed in the map coordinate system.

[0018] For ease of explanation, the path segment between two adjacent waypoints is referred to as a road segment. Figure 1 uses a raster map; however, this invention is equally applicable to vector maps. For clarity, it is assumed below that the loader is traveling in the forward direction. Due to the symmetrical design of the loader, the reverse travel scenario can be handled similarly. Path tracking means that, ideally, by controlling the loader's articulation angle in real time, the origin of the front vehicle reference frame is always kept on the reference path, and its X-axis is tangential to the reference path.

[0019] The automated driving system for underground loaders typically employs a path development process as shown in Figure 2. As illustrated, this process first involves a path planning step to obtain alternative reference paths, followed by a real-vehicle verification step to confirm the traversability of the reference paths. If the real-vehicle verification fails, the system returns to the path planning step to adjust the path. The path planning step involves creating and editing reference paths using path planning tools, or obtaining reference paths by recording data from manual driving. It also allows for modification of parameters such as the location or speed of existing reference path waypoints. The real-vehicle verification step involves assigning a driver as a supervisor on the loader, with the automated driving system controlling the loader along the reference path to verify its traversability. During real-vehicle verification, if any sections become impassable, the supervisor will apply the emergency brakes to prevent accidents. For safety reasons, the initially planned reference path typically operates at a relatively low speed. After successful real-vehicle verification, the speed is increased using path planning tools, and then the increased speed is verified again. Several rounds of path adjustment and verification cycles are usually required to obtain a reference path that simultaneously meets both production efficiency and safety requirements. The curvature of different sections of the reference path varies, therefore the speed settings for different sections typically differ, with higher speeds on straight sections and lower speeds around curves. Clearly, there are numerous ways to adjust waypoint positions and speeds. If each adjustment requires verification on a real vehicle, path development efficiency becomes extremely low. Furthermore, it's not easy to intuitively visualize how the loader's operation will change after waypoint and speed adjustments, placing high demands on the operator's skills and experience. In short, the current path development methods commonly used for automatic underground loader operation suffer from the dual problems of low efficiency and high demands on operator skills and experience, representing pain points that urgently need improvement.

[0020] To address the problems of current methods, this example proposes an improved method for developing reference paths for automated driving of underground loaders, as shown in Figure 3. Comparing Figures 2 and 3, it can be seen that this invention adds a path analysis step between the path planning step and the actual vehicle verification step. The path analysis step analyzes the reference path using mathematical and physical methods. Specifically, for all waypoints on the reference path, the articulation angle, articulation angular velocity, pose of the front and rear vehicle bodies in the map coordinate system, and vehicle acceleration when the loader reaches that point along the reference path are estimated. Then, the minimum distance between the loader body and the tunnel wall is estimated using the pose of the front and rear vehicle bodies in the map coordinate system. Different brands and models of loaders have different allowable ranges for articulation angle, articulation angular velocity, velocity, and acceleration; these ranges serve as vehicle constraints for path analysis. For safety reasons, the loader must maintain a sufficient safety distance from the tunnel wall during operation; this safety distance constitutes the safety constraint used in path analysis. The constraints used in path analysis include the aforementioned vehicle constraints and safety constraints. For waypoints on the reference path, the data obtained during path analysis is compared with the two types of constraints mentioned above to determine whether all waypoints meet the constraints. If there are waypoints on the path that do not meet the constraints, the reference path is considered to be unconstrained. Reference paths that do not meet the constraints do not need to be verified on actual vehicles. Furthermore, for reference paths that do not meet the constraints, operators can focus on modifying those waypoints that do not meet the constraints, making targeted modifications and improving path development efficiency.

[0021] The path analysis method proposed in this invention will be described in detail below, and the process is shown in Figure 4. Again, for ease of explanation, it is assumed that the loader is traveling in the forward direction. The loader's structure is symmetrical front and rear; therefore, the reverse travel scenario can be handled similarly. Let there be a total of [number missing] on the reference path. There are 1 waypoints, namely , ,…, ,in It is the starting point. It is the destination; each waypoint It has several parameters, including but not limited to the X coordinates of waypoints. Y coordinate The distance from the starting point to the waypoint Curvature at waypoints and the speed at that waypoint ; the distance and curvature The coordinates of waypoints are calculated numerically, a method known in the field of autonomous driving. The road segment starting from is denoted as ,in The path analysis process, as shown in Figure 4, consists of five steps: Step 1: Estimate vehicle acceleration. Assume the vehicle is on the road segment... The vehicle accelerates uniformly, and its acceleration is... Using the law of conservation of energy, we obtain the following equation: in The mass of the loader can be solved from the above equation. Therefore, the time required for the loader to pass through this section of road is... .

[0022] As an example, the acceleration analysis sub-screen (520) in Figure 5 shows the results of performing acceleration estimation and data visualization on the reference path in Figure 1. This screen includes a velocity curve (521) and an acceleration curve (522). The acceleration curve (522) shows that the acceleration is greater than zero at the beginning of the reference path, approximately zero in the middle of the reference path, and less than zero at the end of the reference path. The results of the acceleration analysis are consistent with the velocity planning represented by the velocity curve (521).

[0023] The second step: estimating the articulation angle and its angular velocity. The kinematic model of the loader is well known in the field of loader autopilot, and will not be elaborated here.

[0024] The curvature of the trajectory of the origin of the front vehicle coordinate system when the loader is moving is derived from the kinematic model. Satisfy the following equation: in It refers to the driving speed. It is the hinge angle. It is the hinge angular velocity. This refers to the wheelbase; scrapers have a special characteristic where the distances from the front and rear axles to the hinge point are the same, hence the term "wheelbase" for the scraper. (Waypoint) The hinge angle at the point is The hinge angular velocity is Combining the curvature equations described above, the problem of estimating the hinge angle and hinge angular velocity at each point on the reference path is transformed into an optimization problem (finding the minimum value) with the following non-negative objective function: Where the nonnegative constant The weights are used for the objective function. The physical meaning of the objective function is quite intuitive and will not be elaborated here. In one specific implementation of this invention, Satisfactory results can be obtained when the value is 1000. The optimization problem can be set with various initial conditions and constraints. Furthermore, the optimization problem can be solved using commercial or open-source nonlinear optimization tools. This invention uses the open-source CasaADi / Ipopt library, which yields satisfactory results.

[0025] As an example, the articulation angle and angular velocity analysis sub-screen (510) in Figure 5 shows the results of articulation angle and angular velocity estimation and data visualization for the reference path in Figure 1. This screen includes the articulation angle curve (511) and the articulation angular velocity curve (512). In the specific implementation of the present invention, the articulation angle is defined as positive when the vehicle turns left, negative when turning right, and zero when going straight. The articulation angle curve (511) shows that the articulation angles at both ends of the reference path are close to zero, the articulation angles in the middle turning section are greater than zero, and the articulation angle is the largest at about 16 meters. This is consistent with the characteristics of the reference path shown in Figure 1. The reference path is relatively straight at both ends, and makes a large left turn in the middle, with the sharpest turn at about 16 meters. The articulation angular velocity curve (512) shows that the articulation angular velocity is approximately zero in the straight sections at both ends of the reference path, and greater than zero in the middle turning section of the reference path, which is consistent with the characteristics of the reference path shown in Figure 1. Comparing the hinge angle curve (511) and the hinge angular velocity curve (512), it can be seen that the hinge angular velocity is greater than zero where the hinge angle gradually increases, less than zero where the hinge angle gradually decreases, and approximately zero where the hinge angle remains basically unchanged. This indicates that the estimates of the hinge angle and the hinge angular velocity are consistent.

[0026] Step 3: Estimate the poses of the front and rear vehicle bodies in the map coordinate system at any road point on the reference path. The origin of the front vehicle body's reference system is located at this road point, the X-axis direction is the tangent direction of the reference path at this road point, and the Y-axis direction is 90 degrees counterclockwise from the X-axis direction. Therefore, the pose of the front vehicle body's reference system in the map coordinate system is known. Further, using the hinge angle and wheelbase of the loader at this road point, the pose of the rear vehicle body's coordinate system in the map coordinate system can be obtained through coordinate transformation. Following common practice in robotics and autonomous driving, the poses of the front and rear vehicle bodies in the map coordinate system can be represented using a homogeneous transformation matrix. The homogeneous transformation matrix representing the pose of the front vehicle body is: in The coordinates of the origin of the front vehicle's coordinate system in the map coordinate system are the coordinates of the current road point. Let be the rotation angle of the front vehicle's coordinate system relative to the map coordinate system, which is also the direction angle of the tangent of the reference path at the current waypoint. Based on the structural characteristics of the loader and the basic principles of rigid body kinematics, the homogeneous transformation matrix representing the pose of the rear vehicle can be obtained as follows: In the formula It is the wheelbase, which is a structural constant of the vehicle; It is the hinge angle, estimated through the second step mentioned above. The homogeneous transformation matrix and pose can be converted to each other based on the basic knowledge of rigid body kinematics, which will not be elaborated further in this paper.

[0027] As an example, the poses of the front vehicle coordinate system (160) and the rear vehicle coordinate system (170) in Figure 1 are calculated and visualized according to the above method. Figure 1 shows the correctness of the above calculation method.

[0028] Step 4: Estimating the minimum distance between the vehicle body and the tunnel wall. The following describes the method of estimating the minimum distance between the vehicle body and the tunnel wall using the front vehicle body as an example.

[0029] First, this invention models the vehicle body using polygons in the front and rear vehicle coordinate systems, respectively. Let the vertex coordinates of the polygons used in this model be... common Points, such as the polygonal model of the front vehicle body indicated by 130 in Figure 1; further, each side of the polygon is divided into equal parts with a length not exceeding Multiple line segments, The typical value is 0.1 meters; the coordinates of the vertices of the equally divided line segment are denoted as . common The third step has already obtained the poses of the front and rear vehicle coordinate systems in the map coordinate system. Therefore, the coordinates of the above vertices are projected into the map coordinate system through coordinate transformation to become... Specifically, first, the coordinates Convert to homogeneous coordinates Then, the above homogeneous coordinates are sequentially compared with the homogeneous transformation matrix. (When processing the front vehicle body) or homogeneous transformation matrix (When processing the rear of the vehicle) Multiply to obtain homogeneous coordinates Finally, the obtained homogeneous coordinates are converted into ordinary coordinates. For methods of converting between homogeneous coordinates and ordinary coordinates, please refer to the basic knowledge of rigid body kinematics; furthermore, through calculation... The minimum distance between the vehicle body and the tunnel wall is estimated by the minimum distance between the vehicle body and the tunnel wall.

[0030] This invention employs different calculation methods for raster maps and vector maps. For raster maps, a K-dimensional tree is first constructed using the center points of the alleyway wall raster, followed by the K-nearest neighbor algorithm for calculation. The minimum distance to the midpoint of the K-dimensional tree, this minimum distance is the point. The minimum distance to the tunnel wall is used as an estimate of the minimum distance from the vehicle body to the tunnel wall. Specifically, this invention uses the open-source FLANN library to implement the aforementioned K-dimensional tree construction and K-nearest neighbor search. For vector maps, the tunnel walls are represented by several line segments, denoted as... ;calculate With the above line segment set The minimum distance is used as an estimate of the minimum distance from the vehicle body to the tunnel wall; the distance from the point to the line segment is calculated using a mature algorithm in computational geometry, and the required calculation is performed using the open-source computational geometry algorithm library CGAL.

[0031] As an example, Figure 1 shows the working scenario under the grid map condition. The sub-screen (500) of Figure 5, which analyzes the distance between the front and rear vehicle bodies and the tunnel wall, shows the results after performing grid map-based minimum distance estimation between the front and rear vehicle bodies and the tunnel wall for the reference path shown in Figure 1 and visualizing the data. This screen includes the minimum distance curve between the front vehicle body and the tunnel wall (501) and the minimum distance curve between the rear vehicle body and the tunnel wall (502). The minimum distance curve between the front vehicle body and the tunnel wall (501) shows that the minimum distance between the front vehicle body and the tunnel wall is about 0.45 meters, located at a distance of about 9.5 meters. The minimum distance curve between the rear vehicle body and the tunnel wall (502) shows that the minimum distance between the rear vehicle body and the tunnel wall is about 0.35 meters, located at a distance of about 20 meters. Outside the distance range of 18.5 meters to 21.3 meters, the minimum distance between the rear vehicle body and the tunnel wall is greater than 0.4 meters.

[0032] Step 5: Check if the constraints are met. For all waypoints on the reference path, compare the data obtained in the above analysis process with the vehicle constraints and safety constraints to check whether all waypoints on the reference path meet the constraint conditions.

[0033] As an example, the safety constraint is set as follows: the minimum distance between the front and rear vehicle bodies and the tunnel wall is not less than 0.4 meters; the vehicle constraint is set as follows: the articulation angle does not exceed 0.7 radians, the articulation angular velocity does not exceed 0.7 radians / second, and the acceleration does not exceed 1.0 m / s². Figure 5 shows the analysis sub-screen (500) of the distance analysis between the front and rear vehicle bodies and the tunnel wall, indicating that the minimum distance between the rear vehicle body and the tunnel wall is approximately 0.35 meters around 20 meters, which does not meet the safety constraint. Therefore, the positions of the waypoints around 20 meters on the reference path in Figure 1 need to be optimized. As can be seen from the articulation angle and angular velocity analysis sub-screen (510) and the acceleration analysis sub-screen (520) in Figure 5, the vehicle articulation angle, articulation angular velocity, and acceleration at all waypoints on the reference path in Figure 1 meet the vehicle constraint.

[0034] Figure 5 is one of the analysis screens provided by the path analysis in this invention. It helps operators understand the actual operation of the loader after path adjustment. The first sub-figure in Figure 5 analyzes the minimum distance between the front and rear loader bodies and the tunnel wall when the vehicle passes through different positions on the reference path; the second sub-figure analyzes the articulation angle and articulation angular velocity when the vehicle passes through different positions on the reference path; the third sub-figure analyzes the speed and acceleration of the vehicle when passing through different positions on the reference path. It should be noted that Figure 1 is actually one of the path analysis visualization screens of this invention. Through this screen, operators can move the vehicle along the reference path and intuitively understand the articulation angle of the loader, the configuration of the front and rear loader bodies, and the positional relationship between the front and rear loader bodies and the tunnel wall at each waypoint.

[0035] By checking whether the reference path meets the constraints and through intuitive analysis, this invention can effectively solve the problems of low efficiency and high requirements for operator skills and experience in the current path development methods used for automatic driving of underground loaders.

Claims

1. An improved method for developing reference paths for automatic driving of downhole loaders, characterized in that... The process includes the following steps: Step 1: Estimate the vehicle acceleration; Step 2: Estimate the hinge angle and its angular velocity; Step 3: Estimate the pose of the front and rear vehicle bodies in the map coordinate system; Step 4: Estimate the minimum distance from the vehicle body to the tunnel wall; Step 5: Check if the constraints are met.

2. The improved method for developing automatic driving reference paths for downhole loaders according to claim 1, characterized in that: The specific steps are as follows: Assume there are a total of [number] paths on the reference path. There are 1 waypoints, namely: ， ,…, ,in It is the starting point. It is the destination; each waypoint It has several parameters, including but not limited to the X coordinates of waypoints. Y coordinate The distance from the starting point to the waypoint Curvature at waypoints and the speed at that waypoint ; the distance and curvature The coordinates of waypoints are calculated numerically, using waypoints as the basis. The road segment starting from is denoted as ,in The process consists of five steps: Step 1: Estimate vehicle acceleration. Assume the vehicle is on the road segment. The vehicle accelerates uniformly, and its acceleration is... Using the law of conservation of energy, we obtain the following equation: in The mass of the loader can be solved from the above equation. Therefore, the time required for the loader to pass through this section of road is... Step 2: Estimate the hinge angle and its angular velocity. Based on the kinematic model, derive the curvature of the trajectory of the origin of the front vehicle coordinate system when the loader is moving. Satisfy the following equation: in It refers to the driving speed. It is the hinge angle. It is the hinge angular velocity. It is the wheelbase; let the waypoints be set. The hinge angle at the point is The hinge angular velocity is Combining the curvature equations described above, the problem of estimating the hinge angle and hinge angular velocity at each point on the reference path is transformed into an optimization problem with the following non-negative objective function: Where the nonnegative constant The weights are used for the following steps: Step 3: Estimate the pose of the front and rear vehicle bodies in the map coordinate system at any road point on the reference path. The origin of the front vehicle body reference system is located at that road point, the X-axis direction is the tangent direction of the reference path at that road point, and the Y-axis direction is the X-axis direction rotated 90 degrees counterclockwise. Therefore, the pose of the front vehicle body reference system in the map coordinate system is known. Furthermore, using the hinge angle of the loader at that road point and the wheelbase of the loader, the pose of the rear vehicle body coordinate system in the map coordinate system can be obtained through coordinate transformation. Step 4: Estimate the minimum distance from the vehicle body to the tunnel wall. First, model the vehicle body using polygons in the front and rear vehicle coordinate systems respectively. Let the vertex coordinates of the polygons used in this model be... common Furthermore, each side of the polygon is divided into equal parts with a length not exceeding [number of points]. Multiple line segments, The typical value is 0.1 meters; the coordinates of the vertices of the equally divided line segment are denoted as . common The third step has already obtained the poses of the front and rear vehicle coordinate systems in the map coordinate system. Therefore, the coordinates of the above vertices are projected into the map coordinate system through coordinate transformation to become... Furthermore, through calculation The minimum distance between the vehicle body and the tunnel wall is estimated by the minimum distance between the vehicle body and the tunnel wall; the fifth step: check whether the constraints are satisfied. For all waypoints on the reference path, the data obtained in the above analysis process is compared with the vehicle constraints and safety constraints to check whether all waypoints on the reference path meet the constraint conditions.

3. The improved method for developing reference paths for automatic driving of downhole loaders according to claim 2, characterized in that: In the fourth step, calculations are performed. The minimum distance between the vehicle body and the tunnel wall is estimated by finding the minimum distance between the vehicle body and the tunnel wall. For a grid map, a K-dimensional tree is first constructed using the center points of the tunnel wall grid, and then the K-nearest neighbor algorithm is used to calculate the minimum distance between the vehicle body and the tunnel wall. The minimum distance to the midpoint of the K-dimensional tree, this minimum distance is the point. The minimum distance from the tunnel wall is used as an estimate of the minimum distance from the vehicle body to the tunnel wall.

4. The improved method for developing automatic driving reference paths for downhole loaders according to claim 2, characterized in that: In the fourth step, calculations are performed. The minimum distance between the vehicle body and the tunnel wall is estimated by the minimum distance between the vehicle body and the tunnel wall. For vector maps, the tunnel wall is represented by several line segments, which are denoted as... ;calculate With the above line segment set The minimum distance is used as an estimate of the minimum distance from the vehicle body to the tunnel wall.

Images