A path planning method and system based on material loading and unloading in mining areas

Abstract

The present application relates to a path planning method and system based on material loading and unloading in mining areas. In the prior art, the main difficulty in intelligent vehicle path planning is how to achieve balance while optimizing numerous tasks, while bypassing collisions and flexibly adapting to the changing environment. This requires determining a method that can effectively navigate complex multi-task optimization and seamlessly integrate real-time environmental information. The present application is based on a path planning method and system for material loading and unloading in mining areas. The intelligent vehicle path planning method that combines a multi-task particle swarm optimization method and a window adaptive method successfully optimizes various tasks and incorporates real-time environmental data into its trajectory planning, thereby improving the flexibility of intelligent vehicles in dynamic environments, and also improving the effectiveness and efficiency of intelligent vehicle path planning in complex and dynamic environments.

Description

Technical Field

[0001] The present application relates to the field of path planning, and in particular to a path planning method and system based on material loading and unloading in a mining area. Background Art

[0002] The mining material handling industry is in the process of transitioning from manual operation to automation. With the rapid development of e-commerce, the efficiency of mining material handling has become key to the logistics industry. Smart vehicles have become key assets in different industries, especially in certain environments (such as mining areas, warehouses, and manufacturing facilities), where their ability to effectively transport goods in a controlled environment is highly valued. Smart vehicles operate by following predefined routes or using navigation algorithms to depict their trajectories. However, in complex and dynamic environments, path planning for smart vehicles becomes a daunting task that requires optimizing multiple tasks simultaneously, especially those including energy conservation, transport duration, and collision avoidance. Path planning for smart vehicles has been the subject of extensive research, such as using various algorithms, including A * Search algorithms, Dijkstra algorithms, and fast exploration random tree algorithms are used to determine the optimal route for smart vehicles. However, these traditional methods may not be able to address the dynamic nature of the environment and issues related to smart vehicle operation. In the prior art, the main difficulty in smart vehicle path planning during material loading and unloading in mining areas is how to achieve a balance while optimizing many tasks, while avoiding collisions and flexibly adapting to the changing environment. This requires determining a method that can effectively navigate complex multi-task optimization and seamlessly integrate real-time environmental information. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a path planning method and system based on material loading and unloading in mining areas, which successfully optimizes various tasks, including energy usage, transportation time and collision avoidance, incorporates real-time environmental data into its trajectory planning, improves the flexibility of smart vehicles in dynamic environments, and also improves the effectiveness and efficiency of smart vehicle path planning in complex and dynamic environments.

[0004] This application first discloses a path planning method based on material loading and unloading in a mining area, and its specific steps are as follows:

[0005] Step 1: Initialize the environment map, obstacles, and set environment parameters; create an environment map of the mining area, including setting the starting point and the target point, marking all known obstacles on the map and defining the population size (the number of particles in the particle swarm); determine the number of variables and weights, which will affect the decision of the path planning algorithm;

[0006] Step 2: Initialize the particle swarm; randomly distribute the particle swarm in the working area of ​​the map, and each particle represents a potential solution for the intelligent vehicle path;

[0007] Step 3: Optimize speed and trajectory selection; Based on the shortest path objective function and the maximum smoothness objective function, use the multi-task particle swarm optimization algorithm to optimize the speed selection within the adaptive window; Apply the window adaptive algorithm to select the highest-scoring trajectory for the smart vehicle based on the optimized speed;

[0008] Step 4: Output the optimal path; check whether the smart vehicle has reached the target point; determine whether the termination condition is met; if the smart vehicle reaches the target point and meets the termination condition, output the current optimal path; if the target point or the termination condition is not reached, the particle swarm is updated and returns to step 3 to continue execution.

[0009] Preferably, the detailed steps of the window adaptive algorithm in step 3 are:

[0010] Step 21: Perception and localization: Get sensor data readings to perceive the environment and estimate the current position and direction of the smart vehicle;

[0011] Step 22: motion prediction: predicting the future motion of the smart vehicle based on the current state of the smart vehicle and the dynamic model; wherein the current state of the smart vehicle includes speed, acceleration, and steering capability;

[0012] Step 23: Generation of an adaptive window: defining an adaptive window to represent feasible speed and steering combinations of the smart vehicle, and setting maximum and minimum limits for linear and angular velocities according to the capabilities of the smart vehicle;

[0013] Step 24: trajectory evaluation; generate a set of candidate trajectories by sampling the velocity within the adaptive window; evaluate each trajectory according to the corresponding criteria in proximity to obstacles, proximity to targets, and motion smoothness; assign a score or cost to each trajectory according to the evaluation criteria;

[0014] Step 25: trajectory selection; select the trajectory with the lowest cost or the highest score as the ideal trajectory of the smart vehicle; ensure that the selected trajectory is collision-free and satisfies the dynamic constraints of the smart vehicle;

[0015] Step 26: motion execution; converting the ideal trajectory into a control command of the smart vehicle actuator; executing the control command to make the smart vehicle move along the desired trajectory;

[0016] Step 27: Loop; repeat steps 21 to 26 to update the motion plan of the smart vehicle according to changes in the environment and the state of the smart vehicle; adjust the adaptive window and re-evaluate the trajectory to adapt to the corresponding dynamic changes.

[0017] Preferably, the specific steps of the multi-task particle swarm optimization algorithm in step 3 are as follows:

[0018] Step 31: Initialization; randomly generate the position and velocity of a particle group in the search space; set the personal best position and fitness of each particle in the initialization position to the initial value; the particle determines the global best position with its best fitness;

[0019] Step 32: Update the particle's velocity and position;

[0020] Step 33: Evaluation: Calculate the fitness of the objective function of each particle's new position by combining the multi-task function;

[0021] Step 34: Update personal and global best; This is done by comparing the fitness value with the personal best fitness of each particle target; if the current fitness is greater than the personal best fitness value, then update the personal best position and fitness value of the objective function; the global best fitness of each target is calculated by comparing the fitness value of the particle with the current global best fitness value to determine whether the current fitness is superior to any objective function, and if so, update the global best position and fitness value of the target accordingly;

[0022] Step 35: Pareto dominance; each particle in the particle swarm is compared based on its performance on the multi-task function to determine the dominance relationship; a dominance level is assigned to each particle based on the Pareto dominance relationship, and the weight is dynamically adjusted based on the performance of the particle and the distribution of solutions on the Pareto frontier;

[0023] Step 36: Non-dominated sorting: The particles are divided into different non-dominated fronts according to their dominance levels, expressed as follows:

[0024]

[0025] Among them, L i is the crowding distance in the non-dominated front to which each particle belongs; and They are the fitness values ​​of the particles before and after the current particle on a certain objective function; and is the highest and lowest fitness value of the objective function among all non-dominated solutions; N is the number of objective functions; k is the total number of particles in the non-dominated frontier;

[0026] Step 37: Selection and reproduction; Select particles from the non-dominated front according to their dominance level and crowding distance; Reproduce new particles by combining the selected particles through crossover and mutation operations;

[0027] Step 38: Termination condition; by checking whether the termination condition is met; if the termination condition is met, stop the algorithm; otherwise, return to step 32.

[0028] Preferably, the shortest path objective function and the maximum smoothness objective function in step 3 are specifically:

[0029] The shortest path objective function is expressed as follows:

[0030]

[0031] Among them, F1 is the objective function of the shortest path; P i is the position of the i-th point of the smart vehicle in the working area space; n(ta) is the total steering angle of the mobile smart vehicle; x i and i are the axes of the horizontal and vertical velocity components of the smart vehicle at the i-th point; this function aims to minimize the total distance when the smart vehicle moves on the planned path, thereby finding the shortest path from the starting point to the end point.

[0032] The maximum smoothness objective function is expressed as follows:

[0033]

[0034] Among them, F2 is the objective function for smoothness, x i and i The coordinate axes representing the horizontal and vertical velocity components of the i-th point of the smart vehicle in the working area. These components together determine the motion state of the smart vehicle at that point. The arctan function is used to calculate the steering angle between consecutive points on the path. The arctan function accepts the horizontal and vertical distances between two points as input and returns the angle from the x-axis to the line connecting the two points. The purpose of this function is to increase the smoothness of the path by minimizing the sum of the steering angles on the path, which helps to reduce sharp turns and speed changes of the smart vehicle during movement, thereby improving the smoothness and efficiency of driving. i and P i+1 Represent the current point and the next point on the path respectively.

[0035] Preferably, in step 31, the specific steps of initialization are:

[0036]

[0037] Among them, S and V are initialized to the position and velocity of the particle respectively; Fitness is the fitness value of the particle position based on the cost function or objective function; Ub, Lb are the problem areas at the boundaries of the search space; unifrnd is a random generation function; d is the dimension; the particle with the best fitness is determined as the global optimal position and fitness.

[0038] Preferably, in step 32, the specific steps of updating the speed and position of the particle are:

[0039] The velocity and position of each particle in the entire particle group are updated;

[0040]

[0041] Among them, λ1 and λ2 are random numbers generated in the arrangement [0, 1]; ω represents the inertia weight; d is the dimension; and The variable vectors represent the individual best position and the global best position of the particle respectively; α and β represent the acceleration coefficients;

[0042] Update the particle position S using the following formula:

[0043]

[0044] in, and Indicates the position and velocity of the particle numbered id at point t; Indicates the new position of the particle numbered id at time t+1.

[0045] On the other hand, the present application also provides a path planning system based on material loading and unloading in a mining area, characterized in that the system comprises:

[0046] at least one processor; and a memory in communication with the at least one processor; wherein,

[0047] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute any of the above-mentioned path planning system methods based on mine material loading and unloading.

[0048] This scheme proposes a path planning method and system based on material loading and unloading in mining areas. It is a smart vehicle path planning method that combines the multi-task particle swarm optimization method and the window adaptive method. The multi-task particle swarm optimization algorithm promotes the optimization of multiple tasks, while the window adaptive technology has real-time adaptability to changing environmental conditions, successfully optimizing various tasks, including energy usage, transportation time, and collision avoidance, and incorporating real-time environmental data into its trajectory planning, thereby improving the flexibility of smart vehicles in the dynamic environment of material loading and unloading in mining areas, and also improving the effectiveness and efficiency of smart vehicle path planning in the complex and dynamic environment when operating in mining areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Schematic diagram of the arc path of the smart vehicle

[0050] Figure 2Schematic diagram of selecting the optimal path for combining multi-task particle swarm optimization and window adaptive method DETAILED DESCRIPTION

[0051] The detailed description set forth below in conjunction with the accompanying drawings is intended to be a description of various exemplary embodiments of the present invention, and is not intended to represent the only embodiment that can practice the present invention. For the purpose of providing a thorough understanding of the present invention, the detailed description includes specific details. However, it is apparent to those skilled in the art that the present invention can be practiced without these specific details. In some instances, well-known structures and components are shown in block diagram form to avoid blurring the concept of the present invention.

[0052] Smart vehicles are autonomous vehicles used to transport goods within mining areas. In recent years, the technology of smart vehicles has made significant progress, revolutionizing the way goods are transported in multiple industries. Research focuses on improving the efficiency, safety, and adaptability of smart vehicles to meet the evolving needs of material handling, logistics, and transportation in mining areas. For example, cutting-edge sensing technologies such as LiDAR and computer vision are integrated, and these technological advances enable smart vehicles to perceive and navigate complex environments more accurately. Previous research has paved the way for the development of more effective path planning algorithms that optimize the trajectory of smart vehicles while considering factors such as energy consumption, transportation time, and collision avoidance. Smart vehicles are equipped with sensors such as laser scanners and vision systems to perceive their environment and navigate safely. Smart vehicles are popular for their ability to improve the efficiency of material operations, reduce labor costs, and enhance safety. Smart vehicle path planning techniques have been proposed and are roughly divided into traditional and intelligent path planning techniques. Standard algorithms include A * Search, Dijkstra's algorithm and fast exploration random tree algorithm, which rely on predefined maps and graph-based planning methods. Intelligent optimization algorithms use optimization techniques to find the best path according to different tasks.

[0053] Previous work has opened the door to more efficient path planning algorithms that optimize smart vehicle trajectories while taking into account various factors, including energy consumption, travel time, and collision avoidance. Smart vehicles are equipped with sensors such as laser scanners and vision systems to perceive their environment and navigate safely. Smart vehicles have been welcomed for their ability to increase efficiency, reduce labor costs, and improve safety in material handling operations.

[0054] In this application, the proposed method combines the multi-task particle swarm optimization algorithm with the window adaptive algorithm to provide efficient and effective path planning for smart vehicles. The multi-task particle swarm optimization method and the window adaptive algorithm are combined in the path planning of smart vehicles. The multi-task particle swarm optimization method is used to optimize the selection of speed and trajectory within the framework of the window adaptive algorithm, allowing efficient and appropriate decision-making in real-time situations, which is related to the importance of smart vehicle path planning and various techniques for optimizing path selection. Integrating the multi-task particle swarm optimization algorithm and the window adaptive algorithm presents a promising approach to meet the challenges of smart vehicle path planning in complex and dynamic environments.

[0055] The specific steps are as follows:

[0056] Step 1: Initialize the environment map, obstacles, and set environment parameters; create an environment map of the mining area, including setting the starting point and the target point, marking all known obstacles on the map and defining the population size (the number of particles in the particle swarm); determine the number of variables and weights, which will affect the decision of the path planning algorithm;

[0057] Step 2: Initialize the particle swarm; randomly distribute the particle swarm in the working area of ​​the map, and each particle represents a potential solution for the intelligent vehicle path;

[0058] Step 3: Optimize speed and trajectory selection; Based on the shortest path objective function and the maximum smoothness objective function, use the multi-task particle swarm optimization algorithm to optimize the speed selection within the adaptive window; Apply the window adaptive algorithm to select the highest-scoring trajectory for the smart vehicle based on the optimized speed;

[0059] Step 4: Output the optimal path; check whether the smart vehicle has reached the target point; determine whether the termination condition is met; if the smart vehicle reaches the target point and meets the termination condition, output the current optimal path; if the target point or the termination condition is not reached, the particle swarm is updated and returns to step 3 to continue execution.

[0060] Depend on Figure 1It can be seen that the smart vehicle is responsible for transporting materials within a given mining area, moving from the starting point to the desired destination. In order to ensure efficient and obstacle-free movement, the smart vehicle must be able to bypass static obstacles while maintaining the shortest path and minimizing the steering angle. This requirement is necessary to meet multiple goals at the same time. The working area of ​​the smart vehicle is defined as a physical area, represented by R×R, where obstacles are mapped to parts of the working area to ensure that the movement between the smart vehicle and obstacles is collision-free. Assume that the smart vehicle can obtain environmental information about its working area. Let α and θ be the angles of the vehicle body with the x-axis, respectively, assuming that the two wheels are parallel and satisfy the Ackermann steering geometry, and L is the wheelbase. The distance of the smart vehicle moving at an adjusted speed in a static obstacle represents the shortest path between points, following the given formula:

[0061]

[0062] Where α and θ are the vehicle body, x-axis angle and steering wheel angle respectively, the horizontal x and vertical y are the velocity components of the axis, re-expressed as; dist is the distance path between points, and n is the total number of steering angles.

[0063] The direction of motion of the smart vehicle body is affected by various physical factors such as gravity acceleration, running speed, and vehicle mass. In addition, in a dynamic environment with obstacles in the workspace, the speed of the smart vehicle is limited by factors such as curvature, turning radius, maximum steering speed, static friction coefficient, and minimum radius. The smart vehicle steering system is constrained by the following program with motion constraints: Therefore, the running speed of the smart vehicle is affected by the following acceleration formula with a limiting minimum speed radius:

[0064]

[0065] In the formula, v i and v max are the running speed and the maximum running speed respectively; u and g are the static friction coefficient and the gravity acceleration parameter respectively; v i and m are the running speed and vehicle mass variables respectively; R and k are the curvature turning radius and maximum steering speed variables respectively.

[0066] Optimization systems based on particle swarm optimization are inspired by the cooperative behavior of schools of fish and flocks of birds. In particle swarm optimization, a group of particles represent potential solutions, and these particles search the space in search of the best answer: the personal most prominent solution of a particle and the overall most prominent solution of the swarm influence how it moves. There are several stages in the process of running particle swarm optimization - initializing the swarm, updating the positions and velocities of particles, evaluating the objective function of the new positions of particles, updating personal and global best, and checking the termination condition are the basic steps of particle swarm optimization.

[0067] The initialization phase is to initialize the particle swarm with random positions or positions and velocities in the search space. In this phase, the individual optimal positions are also set, as well as the fitness of each particle's objective function with respect to its initial position.

[0068]

[0069] Among them, S and V are initialized to the position and velocity of the particle respectively; Fitness is the fitness value of the particle position based on the cost function or objective function; Ub, Lb are the problem areas at the boundaries of the search space; unifrnd is a random generation function; d is the dimension; the particle with the best fitness is determined as the global optimal position and fitness.

[0070] The velocity and position of each particle in the entire particle group are updated;

[0071]

[0072] Among them, λ1 and λ2 are random numbers generated in the arrangement [0, 1]; ω represents the inertia weight; d is the dimension; and The variable vectors are used to represent the personal best position and global best position of the particle respectively; α and β represent the acceleration coefficients. The particle position S is updated using the following formula:

[0073]

[0074] in, and Indicates the position and velocity of the particle numbered id at point t; Indicates the new position of the particle numbered id at time t+1.

[0075] Evaluation phase: Using a fitness function specific to the problem being solved, the evaluation phase proceeds by evaluating the fitness of each particle's new position. After this, the personal and global best states can be updated, for example, by comparing each particle's fitness to its personal best fitness and updating the personal best position if the current fitness is higher. If necessary, the particle's fitness is compared to the global best fitness and the global best position and fitness are updated. The termination condition phase determines whether the termination condition has been met, such as reaching a desired fitness level or a maximum number of iterations. If the termination condition is met, the algorithm stops; otherwise, it returns to the update phase.

[0076] The window adaptive algorithm is a reactive navigation method in the proposed path planning method, which allows the intelligent vehicle to perform reactive navigation in real time according to the current state of the environment and the capabilities of the intelligent vehicle. The following are the detailed steps of the window adaptive algorithm:

[0077] Step 21: Perception and localization: Get sensor data readings to perceive the environment and estimate the current position and direction of the smart vehicle;

[0078] Step 22: motion prediction: predicting the future motion of the smart vehicle based on the current state of the smart vehicle and the dynamic model; wherein the current state of the smart vehicle includes speed, acceleration, and steering capability;

[0079] Step 23: Generation of an adaptive window: defining an adaptive window to represent feasible speed and steering combinations of the smart vehicle, and setting maximum and minimum limits for linear and angular velocities according to the capabilities of the smart vehicle;

[0080] Step 24: trajectory evaluation; generate a set of candidate trajectories by sampling the velocity within the adaptive window; evaluate each trajectory according to the corresponding criteria in proximity to obstacles, proximity to targets, and motion smoothness; assign a score or cost to each trajectory according to the evaluation criteria;

[0081] Step 25: trajectory selection; select the trajectory with the lowest cost or the highest score as the ideal trajectory of the smart vehicle; ensure that the selected trajectory is collision-free and satisfies the dynamic constraints of the smart vehicle;

[0082] Step 26: motion execution; converting the ideal trajectory into a control command of the smart vehicle actuator; executing the control command to make the smart vehicle move along the desired trajectory;

[0083] Step 27: Loop; repeat steps 21 to 26 to update the motion plan of the smart vehicle according to changes in the environment and the state of the smart vehicle; adjust the adaptive window and re-evaluate the trajectory to adapt to the corresponding dynamic changes.

[0084] This algorithm enables smart vehicles to navigate safely and efficiently in dynamic environments by iteratively generating and evaluating trajectories within an adaptive window. It takes into account the dynamic constraints of the smart vehicle, obstacle avoidance, and target reaching tasks to generate an appropriate motion plan.

[0085] In this application, the optimal intelligent vehicle path planning is achieved through multi-task particle swarm optimization in window adaptive technology. When managing multiple tasks of a particle swarm, a particle swarm with different positions and speeds represents a potential solution. Particles search for the best answer in the search space to optimize the specified task. The multi-task particle swarm optimization algorithm includes the following stages: randomly initialize the position and speed of the particles, evaluate the goals set for the intelligent vehicle path planning, evaluate the fitness of each particle, update the personal best and global best according to the personal best position of each particle, update the global best position using the individual and collective best solutions, update the speed and position of the particles, and whether the termination conditions are met.

[0086] The multi-task includes two objective functions: the shortest path objective function and the maximum smoothness objective function. The shortest path objective function is modeled by calculating the distance along the identified points to guide the intelligent vehicle to move forward or backward, turn or move as path planning. The first objective function of the shortest path is expressed as follows:

[0087]

[0088] Among them, F1 is the objective function of the shortest path; P i is the position of the i-th point of the smart vehicle in the working area space; n(ta) is the total steering angle of the mobile smart vehicle; x i and i are the axes of the horizontal and vertical velocity components of the smart vehicle at the i-th point; this function aims to minimize the total distance when the smart vehicle moves on the planned path, thereby finding the shortest path from the starting point to the end point.

[0089] The second objective function is the maximum smoothness function objective, which is expressed as follows:

[0090]

[0091] Among them, F2 is the objective function for smoothness, x i and i The coordinate axes representing the horizontal and vertical velocity components of the i-th point of the smart vehicle in the working area. Here, "velocity components" refer to the velocity components of the smart vehicle in the horizontal (x-axis) and vertical (y-axis) directions. These components together determine the motion state of the smart vehicle at that point. It uses the inverse tangent (arctan) function to calculate the steering angle between consecutive points on the path. The inverse tangent function accepts the horizontal and vertical distances between two points as input and returns the angle from the x-axis to the line connecting the two points. The purpose of this function is to increase the smoothness of the path by minimizing the sum of the steering angles on the path, which helps to reduce sharp turns and speed changes of the smart vehicle during movement, thereby improving the smoothness and efficiency of driving; P i and P i+1 Represent the current point and the next point on the path respectively.

[0092] The maximum smoothness function focuses on the continuity and smoothness of the path, which is particularly important for the navigation of smart vehicles in real environments because it can reduce mechanical wear, improve ride comfort, and potentially improve energy efficiency. In this way, the multi-task particle swarm optimization algorithm is able to consider both the length and smoothness of the path and provide a comprehensive optimized path for the smart vehicle.

[0093] The multi-task particle swarm optimization algorithm is able to consider multiple objectives simultaneously and find a set of solutions that balance different objectives and provide the best trade-off between multiple objectives, rather than searching for a single optimal solution.

[0094] The specific steps of the multi-task particle swarm optimization algorithm are as follows:

[0095] Step 31: Initialization; randomly generate the position and velocity of a particle group in the search space; set the personal best position and fitness of each particle in the initialization position to the initial value; the particle determines the global best position with its best fitness;

[0096] Step 32: Update the velocity and position of the particle; use formula (4) to update the velocity V of each particle, and use formula (5) to update the position S of the particle;

[0097] Step 33: Evaluation: Calculate the fitness of the objective function of each particle's new position by combining multiple task functions, where the task functions are the shortest path objective function and the maximum smoothness objective function;

[0098] Step 34: Update personal and global best; This is done by comparing the fitness value with the personal best fitness of each particle target; if the current fitness is greater than the personal best fitness value, then update the personal best position and fitness value of the objective function; the global best fitness of each target is calculated by comparing the fitness value of the particle with the current global best fitness value to determine whether the current fitness is superior to any objective function, and if so, update the global best position and fitness value of the target accordingly;

[0099] Step 35: Pareto dominance; each particle in the particle swarm is compared based on its performance on the multi-task function to determine the dominance relationship; a dominance level is assigned to each particle based on the Pareto dominance relationship, and the weight is dynamically adjusted based on the performance of the particle and the distribution of solutions on the Pareto frontier;

[0100] Step 36: Non-dominated sorting: The particles are divided into different non-dominated fronts according to their dominance levels, expressed as follows:

[0101]

[0102] Among them, L i is the crowding distance in the non-dominated front to which each particle belongs; and They are the fitness values ​​of the particles before and after the current particle on a certain objective function; and is the highest and lowest fitness value of the objective function among all non-dominated solutions; N is the number of objective functions; k is the total number of particles in the non-dominated frontier;

[0103] Step 37: Selection and reproduction; Select particles from the non-dominated front according to their dominance level and crowding distance; Reproduce new particles by combining the selected particles through crossover and mutation operations;

[0104] Step 38: Termination condition; by checking whether the termination condition is met, the termination condition is specifically whether the set maximum number of iterations is exceeded; if the termination condition is met, stop the algorithm; otherwise, return to step 32.

[0105] Among them, the Pareto advantage relationship is that if a particle is no worse than another particle in all tasks and is better in at least one task, then this particle is considered to dominate the other particle in the Pareto sense; according to the Pareto advantage relationship, each particle is assigned a dominance level. Generally, particles with lower levels are closer to the Pareto frontier, which means they are better solutions; during the optimization process, the weight of each task is adjusted accordingly according to the performance of the particle on the Pareto frontier; this dynamic adjustment helps the algorithm to explore the solution space more effectively and find better solutions. The adjustment of weights also takes into account the characteristics of the optimization problem, such as the trade-offs and conflict levels between different tasks; the algorithm takes into account the distribution of solutions on the Pareto frontier to ensure that the solutions found are both dispersed and close to the ideal solution; in this way, the algorithm is able to find a set of solutions that balance different tasks in multi-task problems, which are non-inferior in the Pareto sense and provide decision makers with a series of feasible options.

[0106] Among them, the purpose of non-dominated sorting is to determine the position of each particle in the non-dominated front; maintain the diversity in the front to avoid all particles concentrating in a certain part of the Pareto front; through the calculation of the crowding distance, give priority to those particles that are sparsely distributed on the Pareto front to promote the diversity of solutions; help the particle swarm optimization algorithm explore different areas of the Pareto front, so as to find a set of solutions that strike a balance between multiple objectives.

[0107] Particles are selected based on their dominance level and crowding distance on the Pareto front. Particles with lower dominance levels are more likely to be selected because they are closer to the Pareto front. Particles with larger crowding distances are also preferred to maintain the diversity of solutions. New particles generated by crossover and mutation operations are added to the population to replace some old particles. This ensures that there are always fresh solutions in the particle swarm, which helps the algorithm continue to explore unknown areas of the Pareto front.

[0108] From the above, it can be seen that the present application also provides a path planning system based on material loading and unloading in a mining area, characterized in that the system includes:

[0109] at least one processor; and a memory in communication with the at least one processor; wherein,

[0110] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a path planning system method based on mine material loading and unloading as described in any one of claims 1-6.

[0111] In summary, this solution is based on the path planning method and system for material loading and unloading in mining areas. It is an intelligent vehicle path planning method that combines the multi-task particle swarm optimization method and the window adaptive method. It successfully optimizes various tasks and incorporates real-time environmental data into its trajectory planning. It improves the flexibility of intelligent vehicles in dynamic environments, and also improves the effectiveness and efficiency of intelligent vehicle path planning in complex and dynamic environments.

[0112] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A path planning method based on material loading and unloading in a mining area, the specific steps of which are as follows: Step 1: Initialize the environment map, obstacles, and set environment parameters; create an environment map of the mining area, including setting the starting point and the target point, marking all known obstacles on the map and defining the population size; determine the number and weight of variables, which will affect the decision of the path planning algorithm; Step 2: Initialize the particle swarm; randomly distribute the particle swarm in the working area of ​​the map, and each particle represents a potential solution for the intelligent vehicle path; Step 3: Optimize speed and trajectory selection; Based on the shortest path objective function and the maximum smoothness objective function, use the multi-task particle swarm optimization algorithm to optimize the speed selection within the adaptive window; Apply the window adaptive algorithm to select the highest-scoring trajectory for the smart vehicle based on the optimized speed; in, The specific steps of the multi-task particle swarm optimization algorithm are as follows: Step 31: Initialization; randomly generate the position and velocity of a particle group in the search space; set the personal best position and fitness of each particle in the initialization position to the initial value; the particle determines the global best position with its best fitness; Step 32: Update the particle's velocity and position; Step 33: Evaluation: Calculate the fitness of the objective function of each particle's new position by combining the multi-task function; Step 34: Update personal and global best; This is done by comparing the fitness value with the personal best fitness of each particle target; if the current fitness is greater than the personal best fitness value, then update the personal best position and fitness value of the objective function; the global best fitness of each target is calculated by comparing the fitness value of the particle with the current global best fitness value to determine whether the current fitness is superior to any objective function, and if so, update the global best position and fitness value of the target accordingly; Step 35: Pareto dominance; each particle in the particle swarm is compared based on its performance on the multi-task function to determine the dominance relationship; a dominance level is assigned to each particle based on the Pareto dominance relationship, and the weight is dynamically adjusted based on the performance of the particle and the distribution of solutions on the Pareto frontier; Step 36: Non-dominated sorting: The particles are divided into different non-dominated fronts according to their dominance levels, expressed as follows: Among them, L i is the crowding distance in the non-dominated front to which each particle belongs; and They are the fitness values ​​of the particles before and after the current particle on a certain objective function; and is the highest and lowest fitness value of the objective function among all non-dominated solutions; N is the number of objective functions; k is the total number of particles in the non-dominated frontier; Step 37: selection and reproduction; selecting particles from the non-dominated frontier according to their dominance level and crowding distance, selecting particles with small dominance level and large crowding distance as selection targets; breeding new particles by combining the selected particles through crossover and mutation operations; Step 38: Termination condition; by checking whether the termination condition is met; if the termination condition is met, stop the algorithm; otherwise, return to step 32; The detailed steps of the window adaptive algorithm are: Step 21: Perception and localization: Get sensor data readings to perceive the environment and estimate the current position and direction of the smart vehicle; Step 22: motion prediction: predicting the future motion of the smart vehicle based on the current state of the smart vehicle and the dynamic model; wherein the current state of the smart vehicle includes speed, acceleration, and steering capability; Step 23: Generation of an adaptive window: defining an adaptive window to represent feasible speed and steering combinations of the smart vehicle, and setting maximum and minimum limits for linear and angular velocities according to the capabilities of the smart vehicle; Step 24: trajectory evaluation; generate a set of candidate trajectories by sampling the velocity within the adaptive window; evaluate each trajectory according to the corresponding criteria in proximity to obstacles, proximity to targets, and motion smoothness; assign a score or cost to each trajectory according to the evaluation criteria; Step 25: trajectory selection; select the trajectory with the lowest cost or the highest score as the ideal trajectory of the smart vehicle; ensure that the selected trajectory is collision-free and satisfies the dynamic constraints of the smart vehicle; Step 26: motion execution; converting the ideal trajectory into a control command of the smart vehicle actuator; executing the control command to make the smart vehicle move along the desired trajectory; Step 27: loop; continuously repeat steps 21 to 26 according to changes in the environment and the state of the smart vehicle to update the motion plan of the smart vehicle; adjust the adaptive window and re-evaluate the trajectory to adapt to the corresponding dynamic changes; Step 4: Output the optimal path; check whether the smart vehicle has reached the target point; determine whether the termination condition is met; if the smart vehicle reaches the target point and meets the termination condition, output the current optimal path; if the target point or the termination condition is not reached, the particle swarm is updated and returns to step 3 to continue execution.

2. A path planning method based on mining material loading and unloading according to claim 1, wherein: The shortest path objective function and the maximum smoothness objective function in step 3 are as follows: The shortest path objective function is expressed as follows: Among them, F1 is the objective function of the shortest path; P i is the position of the i-th point of the smart vehicle in the working area space; n(ta) is the total steering angle of the mobile smart vehicle; x i and i are the axes of the horizontal and vertical velocity components of the smart vehicle at the i-th point, respectively; this function aims to minimize the total distance when the smart vehicle moves on the planned path, thereby finding the shortest path from the starting point to the end point; The maximum smoothness objective function is expressed as follows: Among them, F2 is the objective function for smoothness, x i and i The coordinate axes representing the horizontal and vertical velocity components of the i-th point of the smart vehicle in the working area. These components together determine the motion state of the smart vehicle at that point. The arctan function is used to calculate the steering angle between consecutive points on the path. The arctan function accepts the horizontal and vertical distances between two points as input and returns the angle from the x-axis to the line connecting the two points. The purpose of this function is to increase the smoothness of the path by minimizing the sum of the steering angles on the path, which helps to reduce sharp turns and speed changes of the smart vehicle during movement, thereby improving the smoothness and efficiency of driving. i and P i+1 Represent the current point and the next point on the path respectively.

3. A path planning method based on mining material loading and unloading according to claim 1, wherein: Step 31, the specific steps of initialization are: Among them, S and V are initialized to the position and velocity of the particle respectively; Fitness is the fitness value of the particle position based on the cost function or objective function; Ub, Lb are the problem areas at the boundaries of the search space; unifrnd is a random generation function; d is the dimension; the particle with the best fitness is determined as the global optimal position and fitness.

4. A path planning method based on mining material loading and unloading according to claim 1, wherein: Step 32, the specific steps of updating the speed and position of the particle are: The velocity and position of each particle in the entire particle group are updated; Among them, λ1 and λ2 are random numbers generated in the arrangement [0, 1]; ω represents the inertia weight; d is the dimension; and The variable vectors represent the individual best position and the global best position of the particle respectively; α and β represent the acceleration coefficients; Update the particle position S using the following formula: in, and Indicates the position and velocity of the particle numbered id at point t; Indicates the new position of the particle numbered id at time t+1.

5. A path planning system based on material loading and unloading in a mining area, characterized in that: The system comprises: at least one processor; and a memory in communication with the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a path planning system method based on mine material loading and unloading as described in any one of claims 1-4.

Images