Mine vehicle path planning method based on model simulation in mine environment

Abstract

The application provides a mine vehicle path planning method based on model simulation in a mine environment, and relates to the technical field of automatic control, which collects mine environment data and inputs the data into a preset mine environment dynamic simulation system; a three-dimensional slope curvature matrix is constructed based on terrain elevation data, a dynamic center of gravity offset of the mine vehicle in the driving process is calculated according to the three-dimensional slope curvature matrix, the dynamic center of gravity offset is converted into a path generation value by using a dynamic weighting method, and a path safety level diagram is generated; a tire adhesion force distribution diagram is established according to road surface friction coefficient data, an initial path is generated by using an improved RRT algorithm in combination with obstacle distribution data, the curvature of the initial path is adjusted by the vehicle load state of each path point, and a final path planning scheme is obtained. The application can effectively avoid the problems of rollover, skidding or low path execution efficiency of the mine vehicle in the complex mine environment, and realizes efficient and safe path planning.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and more specifically, to a method for path planning of mining vehicles in a mining environment based on model simulation. Background Technology

[0002] With the continuous expansion of mining resource development, mining transportation operations are gradually moving towards automation and intelligence to improve operational efficiency, reduce labor costs, and ensure personnel safety. Mining vehicles, as core equipment in mining transportation, directly impact mine production efficiency and operational safety through their path planning. Traditional mining vehicle path planning mainly relies on fixed routes or path planning algorithms based on static environments. While these methods are practical in simple scenarios, they exhibit significant limitations in complex and dynamic mining environments. In recent years, with the development of topographic mapping technologies (such as LiDAR and RTK positioning systems), dynamic environment modeling, and intelligent algorithms (such as Fast Random Tree Algorithm, RRT), path planning technologies based on 3D terrain modeling, dynamic obstacle monitoring, and path optimization methods have gradually become a research hotspot. In particular, improvements in multi-sensor fusion technology and path planning algorithms have made path planning for mining vehicles under extreme terrain and complex load conditions increasingly possible.

[0003] However, existing path planning technologies still have many shortcomings in practical applications in mining environments. On the one hand, mining environments are dynamic and complex, with features such as undulating terrain, varying road surface friction coefficients, and the random distribution and movement of obstacles. This places higher demands on the accuracy and real-time performance of path planning. Most existing technologies are based on two-dimensional or static environmental models, lacking comprehensive consideration of three-dimensional slope, curvature, and vehicle dynamic characteristics (such as load and center of gravity changes), making it difficult to guarantee the safety of the planned path. On the other hand, existing path planning algorithms (such as the traditional RRT algorithm) typically employ fixed step sizes or simple random sampling strategies, which are prone to low path search efficiency and local search bottlenecks in mining environments. Furthermore, road surface conditions (such as friction coefficients) and vehicle states (such as load changes) are often ignored in path planning, potentially leading to slippage, rollovers, or excessive energy consumption during actual path execution. These shortcomings limit the practical application effectiveness of mining vehicles in complex mining environments. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention is proposed. This invention provides a model simulation-based method for planning the path of mining vehicles in a mining environment. This method can, to some extent, solve the problems of low matching between the path planning model and the actual environment due to the inability to accurately obtain complex terrain information in the mining environment; and the lack of sufficient consideration of the center of gravity shift characteristics of mining vehicles under dynamic load conditions, leading to safety hazards such as rollover and slippage in complex mining environments.

[0005] According to one aspect of the present invention, a method for mine vehicle path planning in a mining environment based on model simulation is provided, comprising:

[0006] Collect topographic elevation data, road surface friction coefficient data, and obstacle distribution data in the mining environment, and input the collected data into a preset dynamic simulation system for the mining environment;

[0007] A three-dimensional slope curvature matrix is ​​constructed based on the terrain elevation data. The dynamic center of gravity offset of the mining vehicle during the driving process is calculated based on the three-dimensional slope curvature matrix. The dynamic center of gravity offset is converted into path cost value using a dynamic weighting method to generate a path safety level map.

[0008] On the path safety level map, a tire adhesion distribution map is established based on the road surface friction coefficient data. Combined with the obstacle distribution data, an improved RRT algorithm is used to generate an initial path. The curvature of the initial path is adjusted by the vehicle load status at each path point to obtain the final path planning scheme.

[0009] Furthermore, the dynamic simulation system for the mining environment is constructed using a three-dimensional scene engine;

[0010] The rendering unit of the 3D scene engine includes a terrain rendering module, a road surface attribute rendering module, and an obstacle rendering module.

[0011] The terrain rendering module uses a quadtree hierarchical segmentation method to establish terrain mesh models of different precision.

[0012] The road surface attribute rendering module displays the road surface friction coefficient data using a color gradient method;

[0013] The obstacle rendering module selects different 3D models for rendering based on the type of obstacle. If it is a fixed obstacle, a static bounding box model is used. If it encounters a moving obstacle, a dynamic bounding box model is used with added velocity vector labels.

[0014] Furthermore, the formula for calculating the center of gravity offset is expressed as follows:

[0015]

[0016] in, This refers to the vehicle's wheelbase. For the height of the center of mass, This is the load distribution influence coefficient. Location point The equivalent pavement inclination at that location, This refers to the stiffness coefficient of the vehicle's suspension system. The relative height difference between the wheels For vehicle load capacity, The wheelbase of the vehicle. For longitudinal curvature value and This represents the lateral curvature value.

[0017] Furthermore, the formulas for calculating the longitudinal curvature value and the transverse curvature value are expressed as follows:

[0018]

[0019]

[0020] in, Location point The elevation value.

[0021] Furthermore, the improved RRT algorithm is achieved by introducing a comprehensive load compensation factor. The formula is expressed as:

[0022]

[0023] in, , , These are the weighting coefficients. The rated load threshold, To allow the maximum center of gravity offset, The comprehensive curvature value of the road surface includes the longitudinal curvature value. and lateral curvature value .

[0024] Furthermore, the weighting coefficients Based on load sensitivity analysis, it can be expressed as:

[0025]

[0026] Among them, The vehicle's baseline mass (unloaded mass), The rollover stability coefficient under no-load conditions. The minimum allowable stability coefficient;

[0027] The weighting coefficient The dynamic response characteristics based on the center of gravity shift are expressed as:

[0028]

[0029] in, To achieve the designed maximum speed, Minimum turning radius, For roll stiffness, The wheelbase of the vehicle. The current height of the center of mass;

[0030] The weighting coefficient Considering the coupling between road surface curvature and vehicle steering characteristics, it can be expressed as:

[0031]

[0032] in, To achieve the designed maximum speed, Minimum turning radius, This refers to the tilt stiffness.

[0033] Furthermore, the improved RRT algorithm employs an adaptive step size for path search, expressed by the formula:

[0034]

[0035]

[0036] in, As the reference step size, For safety reasons, For path risk level, , , This is the risk weighting coefficient. For the maximum permissible speed, This is a model for predicting the road surface friction coefficient.

[0037] Furthermore, the road surface friction coefficient prediction model is expressed by the following formula:

[0038]

[0039] in, As the reference friction coefficient, ( , () represents the coordinates of the reference point. This is the spatial correlation coefficient. The periodic variation range of the road surface The distance from the sampling point to the reference point. For road surface characteristic period, For speed influence coefficient, The speed is the vehicle speed.

[0040] Furthermore, during the path smoothing optimization process of the improved RRT algorithm, the path curvature must satisfy the following constraints:

[0041]

[0042] in, For path curvature; , The first derivative of the path, , This is the path second derivative.

[0043] Furthermore, integrity verification is performed based on the smoothly converged path, and the comprehensive evaluation index of the entire path is calculated. The value, expressed by the formula, is:

[0044]

[0045] in, This is the total path length. , , These are the weighting coefficients. is the rate of change of curvature.

[0046] Compared with existing technologies, this invention constructs a three-dimensional slope curvature matrix by collecting terrain elevation data, road surface friction coefficient data, and obstacle distribution data in the mining environment. It then combines this matrix with the center of gravity shift characteristics of mining vehicles under dynamic load conditions to convert it into a path cost value to generate a path safety level map, thereby improving the matching degree between the path planning model and the actual environment. Furthermore, the improved RRT algorithm, combined with the dynamic load status of path points, adjusts the path curvature, significantly improving the efficiency and safety of path planning. In addition, this invention comprehensively considers the road surface friction coefficient, dynamic obstacle distribution, and vehicle operating parameters, making the planned path more robust and adaptable. It effectively avoids problems such as rollover, skidding, or low path execution efficiency of mining vehicles in complex mining environments, thus achieving efficient and safe path planning. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0048] Figure 1 This is a flowchart of a model simulation-based path planning method for mining vehicles in a mining environment, according to an embodiment of the present invention. Detailed Implementation

[0049] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0050] As mentioned in the background section, existing technologies have two main problems: First, due to the inability to accurately obtain three-dimensional elevation data, road surface friction coefficients, and real-time distribution of obstacles in complex terrain in mining environments, existing path planning methods are usually based on static or low-precision environmental models, resulting in a mismatch between the path planning results and the actual environment, which can easily lead to safety hazards such as vehicle rollover or skidding. Second, existing path planning algorithms do not fully consider the combined effects of the center of gravity shift characteristics of mining vehicles under dynamic load conditions, path curvature constraints, and road surface friction coefficients, resulting in a lack of safety and robustness of the planned path in complex mining environments.

[0051] Figure 1 This is a system block diagram of a model simulation-based path planning method for mining vehicles in a mining environment, according to an embodiment of the present invention. Figure 1 As shown, a method for mine vehicle path planning in a model-simulation-based mining environment includes:

[0052] S1: Collect topographic elevation data, road surface friction coefficient data, and obstacle distribution data in the mining environment, and input the collected data into the preset dynamic simulation system of the mining environment.

[0053] The data collected included topographic elevation data, road surface friction coefficient data, and obstacle distribution data in the mining environment. The topographic elevation data included three-dimensional topographic information of the main transportation road and mining operation area. When using a LiDAR scanner to acquire three-dimensional point cloud data of the mining area, the scanner's sampling accuracy was set to 0.02m, and the scanning range covered the entire mining operation area. Noise points were removed from the acquired three-dimensional point cloud data using the least squares method, and the Kriging interpolation algorithm was used to perform spatial interpolation to complete the data.

[0054] Road surface friction coefficient data are collected by ground sensor arrays deployed along the main transportation roads in the mining area. Each set of ground sensor arrays is deployed every 50m along the road. Each set of sensor arrays consists of 4 pressure sensors and 2 friction force sensors. The sampling frequency is 100Hz. When a change in road surface condition is detected, the sensor array automatically increases the sampling frequency to 200Hz.

[0055] In the obstacle distribution data, the spatial coordinates and geometric dimensions of fixed obstacles are obtained through an RTK positioning system with a positioning accuracy better than 0.05m. For moving obstacles, they are tracked in real time by a 77GHz millimeter-wave radar installed on the roadside. The millimeter-wave radar has a detection range of 200m, an angular resolution of 0.1 degrees, and records the obstacle's movement trajectory data in the last 10 seconds.

[0056] The collected terrain elevation data was discretized into 0.5m × 0.5m grids. Simultaneously, road surface friction coefficient data and obstacle distribution data were mapped to their corresponding locations on the discretized grid map using spatial coordinates. Specifically, a unified spatial coordinate system was established, with the southwest corner of the mining area selected as the origin to create a rectangular coordinate system, where the x-axis points east, the y-axis points north, and the z-axis points to the zenith. When mapping the road surface friction coefficient data, the spatial coordinates (x, y, z) of the friction coefficient data collected by the ground sensor array were determined based on the sensor's installation location. A bilinear interpolation algorithm was used to calculate the friction coefficient in the sensor interval area. If a grid cell spans the sampling areas of multiple sensors, then... The interpolation result at the center point of the grid cell is taken as the friction coefficient value of the grid. For the mapping of obstacle distribution data, when a fixed obstacle is encountered, the coordinates of the obstacle boundary points obtained by the RTK system are transformed to a unified coordinate system. The set of grid cells covered by the obstacle is determined by the polygon filling algorithm. If a moving obstacle is encountered, the position coordinates of the moving obstacle at the current moment are calculated based on the real-time position data obtained by the millimeter-wave radar. The cell occupied by the obstacle in the grid map is determined by combining the geometric size of the obstacle. At the same time, the movement direction and speed information of the obstacle are marked for each occupied grid cell. After the mapping is completed, each grid cell contains the friction coefficient attribute value of that position and the obstacle occupancy status information.

[0057] After data mapping is completed, the processed terrain elevation data, road surface friction coefficient data, and obstacle distribution data are input into the preset dynamic simulation system of the mine environment. The dynamic simulation system of the mine environment is constructed using a three-dimensional scene engine.

[0058] The rendering unit of the 3D scene engine includes a terrain rendering module, a road surface attribute rendering module, and an obstacle rendering module.

[0059] When the terrain rendering module renders terrain elevation data, it uses a quadtree hierarchical segmentation method to establish terrain grid models with different precisions. A high-resolution grid with a precision of 0.5m is used within 50m of the observation point, while a low-resolution grid with a precision of 2m is used in areas beyond 50m.

[0060] The road surface attribute rendering module displays the road surface friction coefficient data using a color gradient method, where areas with a friction coefficient greater than 0.8 are displayed in green, areas with a friction coefficient between 0.5 and 0.8 are displayed in yellow, and areas with a friction coefficient less than 0.5 are displayed in red;

[0061] The obstacle rendering module selects different 3D models for rendering based on the type of obstacle. For fixed obstacles, a static bounding box model is used. When encountering moving obstacles, a dynamic bounding box model is used with added velocity vector labels.

[0062] The 3D scene engine updates the scene at a frequency of 50Hz. When rendering dynamically, it prioritizes rendering scene elements within a 100m radius around the vehicle and updates the terrain, road surface attributes, and obstacle status within that radius in real time.

[0063] S2: Construct a three-dimensional slope curvature matrix based on the terrain elevation data, calculate the dynamic center of gravity offset of the mining vehicle during the driving process according to the three-dimensional slope curvature matrix, and use a dynamic weighting method to convert the dynamic center of gravity offset into path cost value to generate a path safety level map.

[0064] A three-dimensional slope and curvature matrix is ​​constructed based on terrain elevation data. First, a bicubic spline interpolation algorithm is used to calculate the first and second derivatives of each grid point, thereby obtaining the slope and curvature values ​​for that point. For each location point... Its slope value The calculation formula is:

[0065]

[0066] in, This is the elevation value of that point.

[0067] Curvature values ​​include longitudinal curvature values. and lateral curvature value The formulas are as follows:

[0068]

[0069]

[0070] A three-dimensional slope-curvature matrix is ​​constructed based on slope and curvature values, with the location of each grid point... corresponding matrix elements It consists of a triplet of data, that is .

[0071] The dynamic center of gravity offset of the mining vehicle during operation is calculated based on the three-dimensional slope curvature matrix, taking into account the vehicle's wheelbase. Wheelbase Suspension stiffness coefficient and load capacity When the vehicle is at the location point At that time, taking a 5m×5m local area centered on this location, first calculate the equivalent pavement inclination angle within this area. :

[0072]

[0073] in, This is the influence coefficient of road surface curvature.

[0074] Further calculations are needed to determine the relative height difference between the four wheels of the vehicle. And introduce the load distribution influence coefficient. :

[0075]

[0076] in, For the vehicle's own weight. This represents the vehicle's current actual load capacity.

[0077] Furthermore, a vehicle stability equation considering suspension characteristics is constructed. First, the vehicle rollover stability equation is established, expressed as:

[0078]

[0079] in, For the first Vertical support force of each wheel The wheelbase of the vehicle. For the first The distance from each wheel to the vehicle's centerline To account for the height of the center of gravity after loading, The vehicle's lateral tilt angle (rad) This is lateral acceleration.

[0080] Considering the influence of suspension characteristics, the vertical support force of the wheel is calculated as follows:

[0081]

[0082] in, For the first The suspension compression of each wheel This refers to the stiffness coefficient of the vehicle's suspension system. This is the suspension damping coefficient. The height of the vehicle's center of gravity when unloaded. This is the equivalent road surface inclination angle.

[0083] Introducing the load distribution influence coefficient Center of mass height Expressed as:

[0084]

[0085] Therefore, the vehicle stability equation is obtained as follows:

[0086]

[0087]

[0088] in, For the vehicle stability moment, when This indicates that the vehicle is stable. This indicates that the vehicle is at risk of overturning.

[0089] Solving this equation yields the dynamic center of gravity offset of the vehicle at that location. :

[0090]

[0091] in, This refers to the vehicle's wheelbase. Location point The equivalent pavement inclination at that location.

[0092] Dynamic center of gravity offset Decomposed into horizontal components and longitudinal components The dynamic weighted method is used to convert the dynamic centroid offset into path cost. The formula is expressed as:

[0093]

[0094] in, This is the horizontal stability weighting coefficient, with a value range of [0.6, 0.8]. This is the longitudinal stability weighting coefficient, with a value range of [0.2, 0.4].

[0095] When the lateral center of gravity shifts Greater than the vehicle's wheelbase width When the value is 20%, the path cost at the corresponding location increases by a factor of 1.5. If simultaneously the longitudinal center of gravity shifts... Greater than the vehicle wheelbase length If the path cost is increased by 15%, the path cost increases by a factor of 2. When both the lateral and longitudinal center-of-gravity offsets are less than 50% of their corresponding thresholds, the path cost remains unchanged. A path safety level map is generated based on the calculated path cost. The path safety level map uses a five-level classification method. The value range [0,1] is evenly divided into five intervals, which correspond to five levels in ascending order: safe, relatively safe, moderate, relatively dangerous and dangerous.

[0096] S3: On the path safety level map, establish a tire adhesion distribution map based on the road surface friction coefficient data, combine the obstacle distribution data, use the improved RRT algorithm to generate an initial path, and adjust the curvature of the initial path according to the vehicle load status at each path point to obtain the final path planning scheme.

[0097] Based on the established path safety level map, the road surface friction coefficient is made continuous using the bilinear interpolation method, and a road surface friction coefficient prediction model is constructed. The formula is as follows:

[0098]

[0099] in, As the reference friction coefficient, ( , () represents the coordinates of the reference point. This is the spatial correlation coefficient. The periodic variation range of the road surface The distance from the sampling point to the reference point. For road surface characteristic period, For speed influence coefficient, The speed is the vehicle speed.

[0100] Based on the friction coefficient distribution, calculate the adhesion force ellipse equation under the tire-ground dynamics model:

[0101]

[0102] in, For longitudinal force, It is a lateral force. For vertical loads, This refers to the tire slip angle.

[0103] It is important to note that vertical loads The impact of dynamic transfer needs to be considered:

[0104]

[0105] in, For longitudinal acceleration, For lateral acceleration, For the height of the center of mass, This is the load-bearing influence coefficient.

[0106] Vertical load In the calculation formula, the first term represents the static uniform load, the second term describes the load transfer between the front and rear wheels caused by longitudinal acceleration and deceleration, and the third term characterizes the lateral load transfer during the steering process.

[0107] Furthermore, an improved RRT algorithm is constructed, incorporating a comprehensive load compensation factor. :

[0108]

[0109] in, , , These are the weighting coefficients. The rated load threshold, To allow the maximum center of gravity offset, This refers to the overall curvature of the road surface.

[0110] Furthermore, The determination is based on load sensitivity analysis:

[0111]

[0112] Among them, The vehicle's baseline mass (unloaded mass), The rollover stability coefficient under no-load conditions. This is the minimum allowable stability coefficient.

[0113] The determination is based on the dynamic response characteristics of the center of gravity shift:

[0114]

[0115] in, The height of the unloaded center of gravity. To allow the maximum center of gravity offset, For suspension stiffness, This refers to the wheelbase.

[0116] The determination needs to consider the coupling between road surface curvature and vehicle steering characteristics:

[0117]

[0118] in, To achieve the designed maximum speed, Minimum turning radius, For roll stiffness, The wheelbase is the distance between the wheels. This is the current height of the center of mass.

[0119] The three weighting coefficients must satisfy the following constraints:

[0120]

[0121]

[0122]

[0123] When the RRT algorithm begins path search, it first calculates the load compensation factor based on the current vehicle status. During each search for a new node, after selecting the nearest node from the random tree, the algorithm no longer extends to the random sampling point using a fixed step size. Instead, it dynamically calculates the step size based on an adaptive step size, as expressed in the formula:

[0124]

[0125]

[0126] in, As the reference step size, For safety reasons, For path risk level, , , This is the risk weighting coefficient. This is the maximum permissible speed.

[0127] Specifically: when the load Exceeding the rated load When it is 80%, or the center of gravity offset Exceeding the maximum allowed offset When it is 60%, The value will drop below 0.6, at which point the search step size will automatically decrease to the baseline step size. Below 60%. Meanwhile, when the risk level of the search area... A value greater than 0.7 indicates a center of gravity shift exceeding 70% of the permissible value, a vehicle speed exceeding the speed limit by 70%, or a road surface friction coefficient falling below 70% of the reference value. This will further reduce the step size to about 50% of the original.

[0128] For example, if the reference step size Set to 1.0m, when the vehicle load reaches 85% of its rated load, The coefficient of friction of the road surface in this area drops to 0.55, while the coefficient of friction of the road surface in this area is only 65% ​​of the baseline value. If the value is approximately 0.75, then the final search step size will be reduced to approximately 0.25m.

[0129] When the load is less than 50% of the rated load and the center of gravity offset is less than 30% of the maximum allowable value, If the risk level R(x,y) is less than 0.3, the search step size can be maintained at more than 80% of the baseline step size, thus improving search efficiency.

[0130] Furthermore, in each iteration of the RRT algorithm, uniform random sampling is no longer used; instead, a probability density function based on the Boltzmann distribution is employed for sampling, as expressed by the formula:

[0131]

[0132]

[0133] in, For sampling probability density, As the normalization factor, For temperature control parameters, Let be the state energy function. , , The weighting coefficient for the energy term. To the shortest distance to the obstacle, For safe distance threshold, The path tangent angle at the current position. The target path direction angle.

[0134] When the algorithm begins its search, it first divides the entire planning space into grid cells and calculates the state energy for each grid cell. Specifically, the state energy of a grid cell will increase accordingly when the following conditions occur: path risk. When the energy value is greater than 0.7, the energy value increases by 200%; when the distance to the obstacle is less than 80% of the safe distance, the energy value increases by 150%; when the deviation of the azimuth angle from the target direction exceeds 45 degrees, the energy value increases by 100%. For example, if the baseline energy value is 1.0, when the path risk of this unit is 0.8 and the distance to the obstacle is 75% of the safe distance, its final energy value will increase to 3.5.

[0135] After completing the path search based on the calculated state energy distribution, an initial path is obtained, which is formed by connecting discrete sampling points.

[0136] First, the initial path undergoes node screening preprocessing. Specifically, when the distance between adjacent nodes is less than 0.5 meters, these nodes are merged into a midpoint. When the turning angle of a node is greater than 60 degrees, transition nodes are added 0.8 meters before and after that midpoint. For the screened path node set, a modified cubic spline curve is used for piecewise smoothing, with every three consecutive nodes forming a smoothing unit. Within each smoothing unit, four control points are calculated based on the positions and orientation angles of the first and last nodes: the starting point P0, the ending point P3, and P1 and P2 located at 1.2 times the node distance along the tangent vector direction, respectively. The curve shape is then iteratively optimized by adjusting the positions of the control points. The curvature of the smoothed curve segment must satisfy the following constraints:

[0137]

[0138] in, For path curvature; , The first derivative of the path, , This is the path second derivative.

[0139] When the curvature of the smoothed curve segment exceeds the above constraints, the internal control points P1 and P2 of the segment are simultaneously moved outward by 10% of their original distance each time, until the curvature meets the constraints or the maximum number of adjustments (5 times) is reached. If the rate of change of curvature exceeds 0.2 / m², an auxiliary control point Pm is added at the midpoint of the segment, initially set as the midpoint between P1 and P2, and then the optimal position is searched in the direction perpendicular to P1P2 with a step size of 0.1 meters. At the junction of curve segments, if the curvature difference between adjacent segments is greater than 0.1 / m, the endpoint tangent vectors of the two segments are adjusted simultaneously: the angle between the tangent vectors is reduced by 5 degrees each time, and the control point positions are updated accordingly, until the curvature difference is less than the threshold or the maximum number of adjustments is reached.

[0140] When the minimum distance between the smoothed path and the obstacle is less than the safe distance, the gradient descent method is used for path adjustment: the closest point between the current path and the obstacle is calculated, and all control points within a 3-meter radius of this point are adjusted simultaneously. The adjustment direction is the normal direction away from the obstacle, with a step size starting from 0.2 meters. When the distance increase is less than 1 centimeter, the step size is halved until the safe distance requirement is met or the maximum number of adjustments is reached. The entire smoothing process is completed within a maximum of 50 iterations. If all constraints are still not met, the RRT algorithm is returned to re-search the path segment. In addition, when the vehicle load increases by more than 30% of the rated load, or the center of gravity shifts by more than 40% of the maximum allowable value, or the local road surface friction coefficient decreases by more than 35%, if any of the above conditions are met, the path segment 2.5 times the vehicle length backward from the current position is marked as a replanning area. The starting and ending points and direction angles of this segment remain unchanged, and the RRT search and smoothing process is re-executed. If a path that meets all constraints cannot be obtained after 3 replanning iterations, the vehicle speed is reduced by 20% and the process is repeated.

[0141] The smoothing process is considered complete when the maximum curvature of all curve segments does not exceed 95% of the dynamic constraints, the rate of curvature change is less than 0.18 / m², the curvature difference between adjacent segments is less than 0.08 / m, and the minimum distance to the obstacle is greater than 1.6 times the vehicle width.

[0142] The converged path is validated for completeness, and the overall evaluation index of the entire path is calculated. The value, expressed by the formula, is:

[0143]

[0144] in, This is the total path length. , , These are the weighting coefficients. is the rate of change of curvature.

[0145] like If the value is less than a preset threshold, the path is output to the control system as the final executable path. Specifically, this includes converting the smoothed path into a standard path data structure, sampling a path point every 0.2 meters, and calculating and recording the following information for each path point:

[0146] Global coordinates Accurate to centimeters, tangential angle Accuracy to 0.1 degrees, curvature Accuracy to 0.01 / meter, rate of curvature change Accurate to 0.01 m², the cumulative distance s from the starting point of the path is accurate to centimeters.

[0147] At the same time, each path point is associated with its corresponding vehicle status parameters: current load (accurate to 10kg), center of gravity offset (accurate to 1cm), motion state compensation factor α(m,D) (accurate to 0.01); and environmental parameters: local road surface adhesion coefficient (accurate to 0.01), environmental risk level R(x,y) (accurate to 0.01).

[0148] Next, the path point sequence is segmented. Whenever the curvature change exceeds 0.05 / meter or the road surface parameter change exceeds 10%, the path is marked as a new segment. A recommended speed is calculated for each segment: the baseline speed is 40 km / h; when the maximum curvature within the segment exceeds 0.1 / meter, the speed decreases by 20%; when the load exceeds 75%, the speed decreases by 15%; when the road surface adhesion coefficient is below 0.6, the speed decreases by 25%; when multiple factors occur simultaneously, the deceleration ratio is calculated by combining these factors.

[0149] This generates a sequence of path execution instructions, with each waypoint containing: target position, target heading, recommended speed, and steering angle constraints. For the initial segment from the path start point to the first waypoint, a special starting strategy is generated: if the initial heading deviation is greater than 5 degrees, the maximum speed is set to 15 km / h until the heading deviation is reduced to within 3 degrees. For the stopping segment from the last waypoint to the endpoint, a deceleration curve is planned: uniform deceleration begins at a position 2.5 times the vehicle length from the endpoint to ensure a smooth stop.

[0150] The complete path data and execution instructions are packaged into a standard format: the file header contains the total path length, total number of segments, and total number of path points; the path point data area contains complete status information for each point; the segment information area contains the start and end indexes, motion constraints, and recommended parameters for each segment; and the control instruction area contains the execution strategies for each key point. Each field in the data packet has fixed precision and format requirements to ensure the control system can accurately parse and execute it. For example: coordinate values ​​are uniformly stored as floating-point numbers in meters, accurate to three decimal places; angle values ​​are uniformly converted to radians and stored accurately to four decimal places; and vehicle speed values ​​are uniformly stored in meters per second, accurate to two decimal places.

[0151] In summary, a model simulation-based path planning method for mining vehicles in a mining environment, based on embodiments of the present invention, is elucidated. This method constructs a three-dimensional slope curvature matrix by collecting terrain elevation data, road surface friction coefficient data, and obstacle distribution data from the mining environment. Combined with the center-of-gravity shift characteristics of mining vehicles under dynamic load conditions, this matrix is ​​transformed into a path cost value to generate a path safety level map, thereby improving the matching degree between the path planning model and the actual environment. The improved RRT algorithm, combined with the dynamic load state of path points, adjusts the path curvature, significantly improving the efficiency and safety of path planning. Furthermore, the present invention comprehensively considers road surface friction coefficient, dynamic obstacle distribution, and vehicle operating parameters, making the planned path more robust and adaptable. This effectively avoids problems such as rollover, slippage, or low path execution efficiency for mining vehicles in complex mining environments, thus achieving efficient and safe path planning.

Claims

1. A method for mine vehicle path planning in a mining environment based on model simulation, characterized in that, include: Collect topographic elevation data, road surface friction coefficient data, and obstacle distribution data in the mining environment, and input the collected data into a preset dynamic simulation system for the mining environment; A three-dimensional slope curvature matrix is ​​constructed based on the terrain elevation data. The dynamic center of gravity offset of the mining vehicle during the driving process is calculated based on the three-dimensional slope curvature matrix. The dynamic center of gravity offset is converted into path cost value using a dynamic weighting method to generate a path safety level map. On the path safety level map, a tire adhesion distribution map is established based on the road surface friction coefficient data. Combined with the obstacle distribution data, an improved RRT algorithm is used to generate an initial path. The curvature of the initial path is adjusted by the vehicle load status at each path point to obtain the final path planning scheme.

2. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 1, characterized in that, The dynamic simulation system for the mining environment is built using a three-dimensional scene engine. The rendering unit of the 3D scene engine includes a terrain rendering module, a road surface attribute rendering module, and an obstacle rendering module. The terrain rendering module uses a quadtree hierarchical segmentation method to establish terrain mesh models of different precision. The road surface attribute rendering module displays the road surface friction coefficient data using a color gradient method; The obstacle rendering module selects different 3D models for rendering based on the type of obstacle. If it is a fixed obstacle, a static bounding box model is used. If it encounters a moving obstacle, a dynamic bounding box model is used with added velocity vector labels.

3. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 2, characterized in that, The formula for calculating the center of gravity offset is as follows: in, This refers to the vehicle's wheelbase. For the height of the center of mass, This is the load distribution influence coefficient. Location point The equivalent pavement inclination at that location, This refers to the stiffness coefficient of the vehicle's suspension system. The relative height difference between the wheels For vehicle load capacity, The wheelbase of the vehicle. For longitudinal curvature value and This represents the lateral curvature value.

4. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 3, characterized in that, The formulas for calculating the longitudinal curvature value and the transverse curvature value are expressed as follows: in, Location point The elevation value.

5. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 4, characterized in that, The improved RRT algorithm is achieved by introducing a comprehensive load compensation factor. The formula is expressed as: in, , , These are the weighting coefficients. The rated load threshold, To allow the maximum center of gravity offset, The comprehensive curvature value of the road surface includes the longitudinal curvature value. and lateral curvature value .

6. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 5, characterized in that, The weighting coefficient Based on load sensitivity analysis, it can be expressed as: Among them, The vehicle's baseline mass (unloaded mass), The rollover stability coefficient under no-load conditions. The minimum allowable stability coefficient; The weighting coefficient The dynamic response characteristics based on the center of gravity shift are expressed as: in, To achieve the designed maximum speed, Minimum turning radius, For roll stiffness, The wheelbase of the vehicle. The current height of the center of mass; The weighting coefficient Considering the coupling between road surface curvature and vehicle steering characteristics, it can be expressed as: in, To achieve the designed maximum speed, Minimum turning radius, This refers to the tilt stiffness.

7. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 6, characterized in that, The improved RRT algorithm uses an adaptive step size for path search, expressed by the formula: in, As the reference step size, For safety reasons, For path risk level, , , This is the risk weighting coefficient. For the maximum permissible speed, This is a model for predicting the road surface friction coefficient.

8. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 7, characterized in that, The road surface friction coefficient prediction model is expressed by the following formula: in, As the reference friction coefficient, ( , () represents the coordinates of the reference point. This is the spatial correlation coefficient. The periodic variation range of the road surface The distance from the sampling point to the reference point. For road surface characteristic period, For speed influence coefficient, The speed is the vehicle speed.

9. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 8, characterized in that, During the path smoothing optimization process of the improved RRT algorithm, the path curvature must meet the following constraints: in, For path curvature; , The first derivative of the path, , This is the path second derivative.

10. The method for mine vehicle path planning in a mining environment based on model simulation according to claim 8, characterized in that, Integrity verification is performed based on a smoothly converged path, and a comprehensive evaluation index for the entire path is calculated. The value, expressed by the formula, is: in, This is the total path length. , , These are the weighting coefficients. is the rate of change of curvature.

Images