Unmanned vehicle path planning method in strip mine environment

Abstract

The invention provides an unmanned vehicle path planning method in an open-pit mine environment, and relates to the technical field of automatic driving, and the method comprises the steps: obtaining the three-dimensional data of the open-pit mine environment through a laser radar, and constructing a point cloud map through the vehicle motion state information of an inertial measurement unit, so as to achieve the preliminary positioning of a vehicle; establishing a joint optimization model of the laser radar and the inertial measurement unit, inputting the data of the laser radar and the inertial measurement unit into the joint optimization model, calculating the final pose of the vehicle, and updating the point cloud map in real time to reflect the real-time position information of the vehicle. According to the invention, through the laser radar, the inertial measurement unit and real-time environment information, accurate positioning of the vehicle, dynamic optimization of global and local paths and safety cooperative control are realized, and the driving safety, adaptability and efficiency of the unmanned vehicle in a complex strip mine environment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving, and particularly to a path planning method for driverless vehicles in an open-pit mine environment. Background Art

[0002] Traditional open-pit mine transportation heavily relies on manual driving, which not only increases the human resource cost, but also is prone to operation errors and safety accidents due to the influence of drivers' skills, experience and fatigue status. When facing complex terrains and adverse weather conditions, the driving speed of manually driven transport vehicles is limited, and frequent stops for inspections are required, resulting in low overall transport efficiency and affecting the mining progress of the mine. The open-pit mine environment is complex and changeable, with many potential safety hazards. It is difficult for manually driven vehicles to accurately judge the surrounding environment in real time, especially in the case of poor visibility or complex road conditions, and accidents such as collisions and rollovers are extremely likely to occur.

[0003] Due to the need to hire a large number of drivers and bear their wages, benefits and other expenses, as well as compensation and maintenance costs caused by accidents, the overall operating cost of the traditional open-pit mine transportation method remains high. It is difficult for the transportation method to achieve intelligent dispatching and management of vehicles, and it is impossible to dynamically adjust the transportation route and vehicle configuration according to the actual needs of the mine, resulting in insufficient utilization of resources and low transportation efficiency.

[0004] In addition, in the face of the unique complex terrains and adverse weather conditions in open-pit mines, traditional transport vehicles are often difficult to adapt to, prone to failures or damages, increasing the maintenance cost and downtime. In the transportation method, vehicle operation data is often ignored or not fully utilized. There is rich information contained in these data, such as vehicle status, driving trajectory, environmental conditions, etc., and these data are often wasted or not effectively analyzed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a path planning method for driverless vehicles in an open-pit mine environment, which realizes the functions of autonomous navigation, path planning, obstacle avoidance and stable control of the vehicle.

[0006] To solve the above technical problems, the technical solution of the present invention is as follows: In a first aspect, a path planning method for driverless vehicles in an open-pit mine environment, the method includes: Obtaining three-dimensional data of the open-pit mine environment through a lidar, and constructing a point cloud map through the vehicle motion state information of an inertial measurement unit to achieve the initial positioning of the vehicle; Establishing a joint optimization model of the lidar and the inertial measurement unit, inputting the data of the lidar and the inertial measurement unit into the joint optimization model, calculating the final pose of the vehicle, and updating the point cloud map in real time to reflect the real-time position information of the vehicle; Based on the point cloud map updated in real time, conduct a preliminary global path planning to obtain a globally optimized path; Based on the globally optimized path and real-time environmental information, conduct a preliminary local path planning, and through a reward and punishment function, dynamically optimize the local path to determine the final driving path; Evaluate the environmental factors of the open-pit mine in real time to obtain the evaluation results of environmental factors, and based on the evaluation results of environmental factors, dynamically adjust the final driving path to formulate a driving plan for the driverless vehicle; Based on the driving plan of the driverless vehicle and real-time vehicle-to-vehicle information, plan the vehicle following and intersection safety cooperation process, and conduct real-time tracking of the local path to obtain the tracking results. Based on the tracking results and real-time environmental information, control the vehicle.

[0007] Furthermore, obtain the three-dimensional data of the open-pit mine environment through a lidar, and construct a point cloud map through the vehicle motion state information of an inertial measurement unit to achieve the preliminary positioning of the vehicle, including: Collect the three-dimensional data of the open-pit mine environment using a lidar, and preprocess the three-dimensional data to generate initial point cloud data; Through the vehicle motion state information provided by the inertial measurement unit in real time, correct and optimize the initial point cloud data to construct a point cloud map; Based on the point cloud map, achieve the preliminary positioning of the vehicle.

[0008] Furthermore, establish a joint optimization model of the lidar and the inertial measurement unit, input the data of the lidar and the inertial measurement unit into the joint optimization model, calculate the final pose of the vehicle, and update the point cloud map in real time to reflect the real-time position information of the vehicle, including: Input the point cloud data collected by the lidar and the acceleration and angular velocity motion state data provided by the inertial measurement unit into the joint optimization model; Define an error function to quantify the point cloud matching error, motion state error, and fusion error between the matching error and the motion state error; Iteratively optimize the error function. In each iteration, according to the currently estimated vehicle pose, recalculate the point cloud matching error and the motion state error provided by the inertial measurement unit; Continuously iterate and optimize. When the preset number of iterations is reached, output the final pose of the vehicle, including position, speed, and attitude information; Based on the final pose of the vehicle, update the point cloud map in real time.

[0009] Furthermore, based on the point cloud map updated in real time, conduct a preliminary global path planning to obtain a globally optimized path, including: Determine the starting and ending positions of path planning in the global map, and convert the starting and ending positions into coordinate points in the map coordinate system; Create an open list and a closed list, and add the starting node to the open list; Select the current node from the open list, remove the current node from the open list, add it to the closed list, and mark it as evaluated; Traverse all adjacent nodes of the current node, and perform operations of adding to the open list, setting the parent node, or updating the cost and parent node according to whether the adjacent node is in the closed list and whether it is already in the open list; When expanding adjacent nodes, use the Euclidean distance to estimate the distance from the adjacent node to the end point to guide the search direction, and when the search reaches the end point or cannot continue, construct a globally optimized path from the starting point to the end point by backtracking the parent node pointer.

[0010] Furthermore, based on the globally optimized path and real-time environment information, conduct a preliminary planning of the local path, and dynamically optimize the local path through a reward and punishment function to determine the final driving path, including: Construct a virtual force field environment around the vehicle, where the target position generates gravitational force to guide the vehicle forward, and obstacles generate repulsive force; Under the force field environment, plan a preliminary local path of the vehicle, and dynamically optimize the local path through a reward and punishment function to determine the final driving path.

[0011] Furthermore, evaluate the environmental factors of the open-pit mine in real time to obtain the evaluation result of the environmental factor, and dynamically adjust the final driving path according to the evaluation result of the environmental factor to formulate a driving plan for the driverless vehicle, including: Collect the environmental data of the open-pit mine in real time, including environmental factors such as climate conditions, road conditions, dust or fog concentration; Convert the environmental data of the open-pit mine into the evaluation result of the environmental factor according to the preset rules and fuzzy logic; Dynamically adjust the driving plan of the driverless vehicle according to the evaluation result of the environmental factor, including adjusting parameters such as the acceleration, speed, and steering angle of the vehicle; Formulate a driving plan for the driverless vehicle according to the adjusted driving plan of the driverless vehicle, including information such as the driving route, speed planning, and acceleration planning of the vehicle.

[0012] Furthermore, according to the driving plan of the driverless vehicle and the real-time vehicle-to-vehicle information, plan the vehicle following and intersection safety cooperation process, and perform real-time tracking on the local path to obtain the tracking result, and control the vehicle according to the tracking result and the real-time environment information, including: Track the local optimized path in real time, and obtain the final control sequence of the vehicle in the next period of time according to the current state of the vehicle, including position, speed, acceleration, and a preset prediction model, including steering, accelerating, and decelerating actions taken by the vehicle; During the real-time tracking process, continuously update the prediction model according to the actual state of the vehicle and real-time environment information, including road conditions and obstacle positions, to obtain a comprehensive evaluation of the tracking results and real-time environment information; Control the vehicle according to the comprehensive evaluation of the tracking results and real-time environment information, including adjusting the steering angle, accelerating, and decelerating of the vehicle, to ensure that the vehicle travels stably and safely along the planned path.

[0013] In a second aspect, a path planning system for an unmanned vehicle in an open-pit mine environment includes: An environment perception module for real-time collecting environmental factor data such as the climate, road conditions, dust or fog concentration of the open-pit mine environment; A vehicle state monitoring module for obtaining the current state of the unmanned vehicle, including position, speed, and acceleration; A path planning module for planning the global final path of the unmanned vehicle in the open-pit mine environment through a preset prediction model according to the data of the environment perception module and the vehicle state monitoring module; An obstacle detection and avoidance module for using lidar and cameras to real-time detect obstacles on the path, including static obstacles and dynamic obstacles, and adjusting the original path according to the position, speed, and moving direction of the obstacles; A real-time tracking and optimization module for real-time tracking the global final path and obtaining the final control sequence of the vehicle in the next period of time according to the current state of the vehicle and real-time environment information, including steering, accelerating, and decelerating actions; A vehicle control module for controlling the unmanned vehicle according to the control sequence of the real-time tracking and optimization module to make the vehicle travel along the planned path.

[0014] The above solution of the present invention at least includes the following beneficial effects: By using lidar to obtain three-dimensional data of the open-pit mine environment and combining the vehicle motion state information of the inertial measurement unit, a high-precision point cloud map is constructed, realizing the initial positioning of the vehicle. Further, a joint optimization model of lidar and inertial measurement unit is established, which can more accurately calculate the final pose of the vehicle and real-time update the point cloud map to ensure the real-time and accuracy of the map. This improves the positioning accuracy and map update efficiency of the unmanned vehicle in the complex open-pit mine environment.

[0015] The initial global path planning is carried out to obtain the globally optimized path, providing the overall driving direction for the vehicle. Then, based on the global path and real-time environmental information, the initial local path planning is carried out, and the local path is dynamically optimized through a reward and punishment function to determine the final driving path that better conforms to the actual driving conditions. This method of combining global and local path planning improves the driving efficiency of the vehicle.

[0016] The environmental factors of the open-pit mine, such as climate, road conditions, dust / fog concentration, etc., are evaluated in real time to obtain the evaluation results of environmental factors, and the final driving path is dynamically adjusted according to these results. This enables the driverless vehicle to better adapt to the complex and changeable environmental conditions of the open-pit mine and improves the driving safety. At the same time, the vehicle following and intersection safety cooperation processes are planned, and the local path is tracked in real time, and the vehicle is controlled according to the tracking results and real-time environmental information, further enhancing the safety and stability of the vehicle during driving.

[0017] The method realizes the intelligent path planning and driving control of the driverless vehicle in the open-pit mine environment, reduces the dependence on traditional manual driving, and reduces the labor cost. At the same time, through real-time environmental assessment and path adjustment, as well as the safety cooperation between vehicles, the intelligent management level of the entire mine transportation system is improved, providing strong support for the efficient and safe mining of the open-pit mine. Brief Description of the Drawings

[0018] Figure 1 It is a schematic flowchart of a method for path planning of a driverless vehicle in an open-pit mine environment provided by an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of a path planning system for a driverless vehicle in an open-pit mine environment provided by an embodiment of the present invention. Detailed Embodiments

[0020] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0021] As Figure 1 shown, an embodiment of the present invention proposes a method for path planning of a driverless vehicle in an open-pit mine environment, and the method includes the following steps: Step 11, obtaining three-dimensional data of the open-pit mine environment through a lidar and vehicle motion state information of an inertial measurement unit, and constructing a point cloud map to achieve the initial positioning of the vehicle; Step 12: Establish a joint optimization model for the lidar and the inertial measurement unit. Input the data of the lidar and the inertial measurement unit into the joint optimization model, calculate the final pose of the vehicle, and update the point cloud map in real time to reflect the real-time position information of the vehicle. Step 13: Based on the point cloud map updated in real time, conduct a preliminary global path planning to obtain a globally optimized path. Step 14: According to the globally optimized path and the real-time environmental information, conduct a preliminary local path planning, and dynamically optimize the local path through a reward and punishment function to determine the final driving path. Step 15: Evaluate the environmental factors of the open-pit mine in real time to obtain the evaluation result of the environmental factors. According to the evaluation result of the environmental factors, dynamically adjust the final driving path to formulate a driving plan for the driverless vehicle. Step 16: According to the driving plan of the driverless vehicle and the real-time vehicle-to-vehicle information, plan the vehicle following and intersection safety cooperation process, and track the local path in real time to obtain the tracking result. Control the vehicle according to the tracking result and the real-time environmental information.

[0022] In the embodiment of the present invention, the lidar can accurately capture the three-dimensional data of the open-pit mine environment, providing a basis for constructing a detailed point cloud map. Combining the vehicle motion state information of the inertial measurement unit can update the vehicle position in real time and realize the preliminary positioning of the vehicle.

[0023] Step 12: The joint optimization model of the lidar and the inertial measurement unit can make full use of the data complementarity of the two, improving the accuracy and stability of vehicle pose calculation. Updating the point cloud map in real time to reflect the real-time position information of the vehicle provides a dynamic and accurate environmental model for path planning and vehicle control.

[0024] Step 13: Based on the point cloud map updated in real time, the optimal path can be globally planned to avoid the trap of local optimal solutions. The globally optimized path provides a clear driving direction for the vehicle, reducing the blindness and uncertainty during driving.

[0025] Step 14: Combining the globally optimized path and the real-time environmental information can flexibly adjust the local path to adapt to environmental changes. The reward and punishment function can guide the vehicle to avoid obstacles, optimize the driving path, and improve the driving efficiency and safety.

[0026] Step 15: Evaluate the environmental factors of the open-pit mine in real time, such as climate, road conditions, etc., providing a basis for the dynamic adjustment of the driving path. Adjusting the driving path according to the evaluation result of the environmental factors can ensure that the vehicle can drive safely and efficiently under various environmental conditions.

[0027] Step 16, plan the vehicle following and intersection safety collaboration process, which can ensure that the vehicle maintains a safe distance from other vehicles during driving and avoid collision accidents. Real-time tracking of the local path and controlling the vehicle according to the tracking results and real-time environmental information can ensure that the vehicle drives safely and stably along the planned path. This improves the intelligence and automation level of the entire driverless system and reduces labor costs and safety risks.

[0028] In a preferred embodiment of the present invention, for the above step 11, obtaining three-dimensional data of the open-pit mine environment through a lidar and constructing a point cloud map based on the vehicle motion state information of the inertial measurement unit to achieve the preliminary positioning of the vehicle may include: Step 112, collect three-dimensional data of the open-pit mine environment using a lidar and preprocess the three-dimensional data to generate initial point cloud data; Step 113, perform correction and optimization processing on the initial point cloud data through the vehicle motion state information provided by the inertial measurement unit in real time to construct a point cloud map; Step 114, achieve the preliminary positioning of the vehicle according to the point cloud map.

[0029] In the embodiment of the present invention, according to the environmental characteristics of the open-pit mine and the vehicle driving requirements, select a suitable lidar model and scanning parameters. Start the lidar and scan the open-pit mine environment in all directions and at multiple angles to obtain rich three-dimensional data. Preprocess the collected original three-dimensional data, including denoising, filtering, cropping, etc. Remove invalid data points generated by equipment noise, environmental interference, etc., and retain effective environmental information. At the same time, smooth the data to reduce the impact of data fluctuations on subsequent processing. Organize the preprocessed data according to a certain format and rules to form initial point cloud data. The point cloud data contains the three-dimensional coordinate information of various objects in the environment and is the basis for constructing a point cloud map.

[0030] Step 113, the inertial measurement unit (IMU) can provide real-time vehicle motion state information such as attitude angle, acceleration, and speed. This information is crucial for correcting and optimizing the point cloud data. Use the vehicle motion state information provided by the IMU to correct the initial point cloud data. Through the attitude angle information, the distortion of the point cloud data caused by vehicle attitude changes can be corrected; through the acceleration and speed information, the offset of the point cloud data caused by vehicle motion can be corrected. On the basis of correction, further optimize the point cloud data. For example, splice the point cloud data collected at different times through a registration algorithm to form a complete point cloud map; reduce the number of point cloud data through a downsampling algorithm to improve processing efficiency; optimize the shape of the point cloud data through a smoothing algorithm to make it closer to the real environment.

[0031] Step 114, according to the characteristics of the point cloud map and the requirements of vehicle positioning, select an appropriate positioning algorithm. Using the selected positioning algorithm, search for feature points or regions in the point cloud map that match the current vehicle position. By comparing the coordinate information of the feature points or regions, determine the preliminary position of the vehicle in the point cloud map.

[0032] Verify the preliminary positioning result to ensure its accuracy and reliability. This can be done by comparing it with the results of other positioning means (such as GPS).

[0033] Suppose in an open-pit mine environment, an autonomous vehicle is driving. The lidar continuously scans the surrounding environment to generate initial point cloud data. At the same time, the IMU provides real-time information such as the vehicle's attitude angle and acceleration. By combining this information with the initial point cloud data, the point cloud data is corrected and optimized, and finally a complete point cloud map is constructed. In the point cloud map, the outline of the mine, the position of the road, and the distribution of obstacles can be clearly seen. The vehicle achieves preliminary positioning based on the information in the point cloud map and determines the next driving direction.

[0034] By combining the use of lidar and IMU, it is possible to obtain more accurate environmental information around the vehicle and achieve precise positioning of the vehicle. This helps to improve the driving safety and stability of autonomous vehicles in the open-pit mine environment. The open-pit mine environment is complex and changeable, and traditional positioning methods may be interfered by environmental factors. The point cloud map positioning method based on lidar and IMU can adapt to various complex environments, such as dust and fog, light changes, etc., to ensure that the vehicle can be accurately positioned in any situation. Accurate positioning information provides the vehicle with a clear driving direction and target position. This helps to reduce the blind driving and unnecessary exploration time of the vehicle and improve the driving efficiency. At the same time, the obstacle information in the point cloud map can also help the vehicle plan the driving path in advance to avoid collisions and delays. The application of autonomous vehicles can reduce the dependence on traditional manual driving and lower the labor cost. The point cloud map-based positioning method, as a key part of autonomous driving technology, provides strong support for the realization of this goal.

[0035] In a preferred embodiment of the present invention, for the above step 12, establishing a joint optimization model of the lidar and the inertial measurement unit, inputting the data of the lidar and the inertial measurement unit into the joint optimization model, calculating the final pose of the vehicle, and updating the point cloud map in real time to reflect the real-time position information of the vehicle may include: Step 123, input the point cloud data collected by the lidar and the acceleration and angular velocity motion state data provided by the inertial measurement unit into the joint optimization model; Step 124, define an error function to quantify the point cloud matching error, the motion state error, and the fusion error between the matching error and the motion state error; Step 125, iteratively optimize the error function. In each iteration, according to the currently estimated vehicle pose, recalculate the point cloud matching error and the motion state error provided by the inertial measurement unit. Step 126, continuously iteratively optimize. When the preset number of iterations is reached, output the final pose of the vehicle, including position, speed, and attitude information. Step 127, according to the final pose of the vehicle, update the point cloud map in real time.

[0036] In the embodiments of the present invention, it is ensured that the data of the lidar and the inertial measurement unit are synchronized in time. This is usually achieved through timestamps to ensure that each data point has a corresponding time tag. Preprocess the point cloud data of the lidar, including denoising, filtering, cropping, etc., to extract effective environmental features. At the same time, calibrate and filter the data of the inertial measurement unit to eliminate noise and bias.

[0037] Input the preprocessed point cloud data and the motion state data (acceleration, angular velocity) of the inertial measurement unit into the joint optimization model. These data will serve as the basis for model optimization.

[0038] Step 124, calculate the matching error between the current point cloud data and the point cloud data in the previous point cloud data or the map. This is achieved by comparing the distances of feature points or feature surfaces. Calculate the error between the motion state data provided by the inertial measurement unit and the motion state predicted based on the previous pose. This includes errors in position, speed, and attitude. Combine the point cloud matching error and the motion state error to define a comprehensive error function. This function should be able to reflect the weights and mutual relationships between the two, so as to consider them simultaneously during the optimization process.

[0039] Step 125, provide an initial vehicle pose estimate, which can be based on the pose at the previous moment or on the data of other sensors. Iteratively optimize the error function. In each iteration, according to the currently estimated vehicle pose, recalculate the point cloud matching error and the motion state error, and update the pose estimate to reduce these errors.

[0040] Step 126, set a number of iterations or an error threshold as the termination condition for iteration. When this condition is reached, stop the iteration and output the final pose of the vehicle. Output the final pose including information such as the position of the vehicle (such as longitude, latitude, altitude), speed (such as linear velocity, angular velocity), and attitude (such as roll angle, pitch angle, yaw angle).

[0041] Step 127, combine the final pose of the vehicle with the point cloud data to update the point cloud map. This includes adding new point cloud data to the map and updating the positions of objects in the map according to the movement of the vehicle.

[0042] Maintain the point cloud map, including removing redundant data, optimizing the data structure, and maintaining the real-time and accuracy of the map, etc.

[0043] Suppose an autonomous vehicle is driving in an open-pit mine. The lidar continuously scans the surrounding environment and generates point cloud data, while the inertial measurement unit provides the vehicle's acceleration and angular velocity information. Input these data into the joint optimization model, and calculate the final pose of the vehicle through iterative optimization. In each iteration, the model will consider the matching error of the point cloud data and the motion state error of the inertial measurement unit, and adjust the pose estimation to reduce these errors. Finally, when the iteration reaches the preset number of times, output the final pose of the vehicle, and update the point cloud map in real time according to this pose. In this way, the vehicle can accurately know where it is and the situation of the surrounding environment.

[0044] By combining the data of the lidar and the inertial measurement unit, and using the joint optimization model for iterative optimization, the final pose of the vehicle can be calculated more accurately, improving the positioning accuracy. The real-time updated point cloud map provides the vehicle with rich environmental information, including the location of obstacles, the shape of the road, etc., enhancing the vehicle's perception ability of the environment. Accurate positioning and real-time environmental perception enable the vehicle to better plan the driving path, avoid collisions and falling into dangerous areas, and improve driving safety. The joint optimization model can handle the data fusion problem of the lidar and the inertial measurement unit, enabling the vehicle to drive stably in the complex and changeable open-pit mine environment. In the case of poor or unavailable GPS signals, the vehicle can still perform positioning and navigation through the data of the lidar and the inertial measurement unit, reducing the dependence on GPS.

[0045] In a preferred embodiment of the present invention, the calculation formula of the error function is: ; Wherein, represents the error function, represents the number of points in the source point cloud and the target point cloud; represents the index of the point currently being processed; represents the th point's coordinates in three-dimensional space; represents the identity matrix; represents the rotation angle; represents the unit vector of the rotation axis; represents the matrix formed after the outer product of the vector with itself; represents the coordinates of the th point in the source point cloud; represents that after rotation, the source point cloud along , , The distances of translation in the axial directions; Denotes the vehicle acceleration vector; Denotes the preset acceleration vector; Denotes the vehicle angular velocity vector; Denotes the preset vehicle angular velocity vector; ; Wherein, , , Denotes the components of the unit vector of the rotation axis ; ; .

[0046] In an embodiment of the present invention, the source point cloud and the target point cloud are determined, and the number of their points is obtained .

[0047] Initialize the rotation angle , the unit vector of the rotation axis (including the components , , ), the translation vector ([[]] ). Obtain the vehicle acceleration vector and the angular velocity vector from the inertial measurement unit. Set the preset acceleration vector and the angular velocity vector .

[0048] Use the rotation angle and the unit vector of the rotation axis to calculate the rotation matrix. The rotation matrix is composed of the identity matrix , the outer product matrix of the rotation axis and the cross product matrix of the rotation axis (the skew-symmetric matrix constructed by ). Specifically in the calculation, first construct the skew-symmetric matrix, and then use the Rodriguez rotation formula to calculate the rotation matrix. For each point ( ) in the source point cloud, apply the rotation matrix and the translation vector to calculate its predicted position in the target point cloud. Calculate the square of the Euclidean distance between the predicted position and the actual position ( ), and sum over all points to obtain the total sum of the point cloud matching errors. Calculate the square of the Euclidean distance between the vehicle acceleration vector and the preset acceleration vector . Calculate the square of the Euclidean distance between the vehicle angular velocity vector and the preset angular velocity vector The squared Euclidean distance between them. Add the point cloud matching error, acceleration error, and angular velocity error to obtain the total error For the rotation angle θ, the unit vector of the rotation axis , the translation vector ( ) are iteratively optimized to minimize the total error . In each iteration, update these parameters and recalculate the total error until the preset number of iterations or error threshold is reached. When the iterative optimization ends, output the final rotation angle, rotation axis, and translation vector, which constitute the final pose of the vehicle. According to the final pose of the vehicle, add the new point cloud data to the point cloud map and update the object positions in the map

[0049] By comprehensively considering the point cloud matching error and the motion state error, and iteratively optimizing the pose parameters, the position of the vehicle in the three-dimensional space can be determined more accurately, improving the positioning accuracy. This method does not rely on external navigation signals (such as GPS), so it can work stably in environments where GPS signals are poor or unavailable (such as open-pit mines), enhancing the adaptability to complex environments. By updating the point cloud map in real time, the real-time position of the vehicle and environmental changes can be reflected, providing the vehicle with the latest environmental information, which helps to plan the driving path and avoid collisions. Accurate positioning and real-time environmental perception enable the vehicle to better understand the surrounding environment, make decisions in a timely manner, avoid potential dangers, and improve driving safety. This method combines the data of lidar and inertial measurement unit, making full use of their respective advantages, and improving the reliability and accuracy of positioning

[0050] In a preferred embodiment of the present invention, step 13 above, according to the point cloud map updated in real time, perform a preliminary global path planning to obtain a globally optimized path, which may include Step 131, determine the starting and ending positions of the path planning in the global map, and convert the starting and ending positions into coordinate points in the map coordinate system Step 132, create an open list and a closed list, and add the starting node to the open list Step 133, select the current node from the open list, remove the current node from the open list, add it to the closed list, and mark it as evaluated Step 134, traverse all adjacent nodes of the current node, and perform operations of adding to the open list, setting the parent node, or updating the cost and the parent node according to whether the adjacent node is in the closed list and whether it is already in the open list Step 135, when expanding adjacent nodes, use the Euclidean distance to estimate the distance from the adjacent node to the end point to guide the search direction, and when the search reaches the end point or cannot continue, construct a globally optimized path from the starting point to the end point by backtracking the parent node pointer

[0051] In an embodiment of the present invention, according to task requirements or user input, the starting and ending positions of path planning are clearly specified in the global map. The actual position information (such as longitude and latitude, altitude, etc.) of the starting and ending points is converted into coordinate points in the map coordinate system. This involves converting geographical coordinates into points in the two-dimensional or three-dimensional coordinate system of the map.

[0052] Step 132: Create two lists, namely the open list (used to store nodes to be evaluated) and the closed list (used to store evaluated nodes). Add the starting node as the first node to be evaluated to the open list. The cost of the starting node (referring to the cumulative cost from the starting point to this node) is initialized to 0, and the parent node is set to null (because it is the starting point).

[0053] Step 133: According to a certain selection strategy (such as the minimum cost, the closest to the end point, etc.), select a node from the open list as the current node. Remove the current node from the open list and add it to the closed list, marking it as evaluated to avoid repeated evaluation.

[0054] Step 134: Check each adjacent node of the current node one by one. For each adjacent node, determine whether it is in the closed list (evaluated) or already in the open list (to be evaluated). If the adjacent node is not in the open list or the closed list, add it to the open list, set its parent node to the current node, and calculate and set its cost (usually including the cumulative cost from the starting point to this node and the estimated cost from this node to the end point). If the adjacent node is already in the open list, but the cost of reaching this node through the current node is lower, update its cost and parent node.

[0055] Step 135: When expanding adjacent nodes, use the Euclidean distance to estimate the straight-line distance from the adjacent node to the end point as heuristic information to guide the search direction. When the search reaches the end point node, or when the open list is empty and no better path can be found, the search terminates. By backtracking the parent node pointer, starting from the end point node, connect its parent nodes in sequence until backtracking to the starting point node, thereby constructing a globally optimized path from the starting point to the end point.

[0056] Suppose there is a two-dimensional map representing a certain area of a city. Each point on the map can be represented by a two-dimensional coordinate, for example ( , ). Now, there is an autonomous vehicle that needs to travel from the starting point (0, 0) to the end point (10, 10). There are some obstacles on the map, such as buildings, road construction areas, etc., and these obstacles are marked as impassable areas on the map.

[0057] The specified starting point and ending point are: Starting point position: (0, 0), Ending point position: (10, 10).

[0058] Create an Open List to store the nodes to be evaluated. Create a Closed List to store the evaluated nodes. Add the starting node (0, 0) to the Open List, with its cost initialized to 0 and the parent node set to null. Assume that the minimum cost is used as the selection strategy. Select the node with the minimum cost from the Open List as the current node, which is the starting node (0, 0) initially. Traverse all adjacent nodes of the current node (in a two-dimensional map, it is the four directions of up, down, left, and right). For each adjacent node, check if it is in the Closed List (already evaluated) or if it is already in the Open List (to be evaluated). If it is not in the Open List either, add it to the Open List, set its parent node to the current node, and calculate its cost (including the cumulative cost from the starting point to this node and the estimated cost from this node to the ending point. Here, the estimated cost can use the Euclidean distance). If it is already in the Open List, but the cost to reach this node through the current node is lower, update its cost and parent node.

[0059] When expanding adjacent nodes, use the Euclidean distance to estimate the straight-line distance from the adjacent node to the ending point (10, 10) as heuristic information to guide the search direction. When the search reaches the ending node (10, 10), or when the Open List is empty and no better path can be found, the search terminates. By backtracking the parent node pointer, starting from the ending node, connect its parent nodes in sequence until backtracking to the starting node, thus constructing a globally optimized path from the starting point to the ending point.

[0060] Assume that during the search process, nodes such as (1, 0), (0, 1), (1, 1), etc. are evaluated in sequence, and the Open List is updated according to the cost and heuristic information. Finally, a globally optimized path from the starting point (0, 0) to the ending point (10, 10) is found. This path may be similar to: (0, 0) -> (1, 0) -> (2, 0) ->... -> (9, 10) -> (10, 10). This path avoids all obstacles and is the final path obtained by comprehensively evaluating the cost and heuristic information.

[0061] By using an open list and a closed list to manage the nodes to be evaluated and the evaluated nodes, and using heuristic information (such as Euclidean distance) to guide the search direction, the efficiency of path planning can be improved and the calculation time can be reduced. By continuously comparing and updating the cost and parent node of the nodes, it is ensured that the finally constructed path is a globally optimized path from the starting point to the ending point, considering all possible path choices. It can handle complex map structures and obstacle distributions, providing a safe and feasible driving route for the vehicle, especially suitable for complex environments such as open-pit mines. By planning a globally optimized path, the vehicle can more accurately predict and avoid potential collision risks, improving the driving safety.

[0062] In a preferred embodiment of the present invention, step 14, according to the globally optimized path and real-time environmental information, perform a preliminary planning of the local path, and dynamically optimize the local path through a reward and punishment function to determine the final driving path, which may include: Step 141, construct a virtual force field environment around the vehicle, wherein, the target position generates gravitational force to guide the vehicle forward, and the obstacle generates repulsive force; Step 142, in the force field environment, plan a preliminary local path of the vehicle, and dynamically optimize the local path through a reward and punishment function to determine the final driving path.

[0063] In an embodiment of the present invention, a three-dimensional virtual force field environment is constructed around the vehicle, and this environment is composed of a gravitational field and a repulsive field. The gravitational field is generated by the target position (i.e., the next key point or the ending point in the globally optimized path), and its magnitude and direction are related to the distance and direction between the vehicle and the target position. The farther the distance, the greater the gravitational force, and the direction points to the target position. The repulsive field is generated by the obstacle, and its magnitude and direction are related to the distance and direction between the vehicle and the obstacle. The closer the distance, the greater the repulsive force, and the direction points to the direction away from the obstacle.

[0064] For the gravitational force, a function inversely proportional to the distance is used for calculation, such as , where is the distance between the vehicle and the target position, is a constant coefficient. For the repulsive force, a function inversely proportional to the square of the distance is used for calculation, and within a certain distance (such as the safety distance of the obstacle), the repulsive force reaches the maximum value, and then rapidly decreases as the distance increases, such as ( ), where is the distance between the vehicle and the obstacle, and are constant coefficients. Superimpose the gravitational field and the repulsive field at the position of the vehicle to obtain the total force received by the vehicle at this position. The magnitude and direction of the total force are determined by the vector sum of the gravitational field and the repulsive field.

[0065] Step 142, in the force field environment, according to the current position and target position of the vehicle, as well as the total force received, plan a preliminary local path. The preliminary path can be obtained by simulating the movement trajectory of the vehicle in the force field, such as using numerical integration methods (such as Euler's method) to iteratively calculate the future position of the vehicle. The reward and punishment function is used to evaluate the quality of the local path and dynamically optimize the path according to the evaluation results.

[0066] The reward part includes: rewards for approaching the target position, rewards for maintaining a safe distance, rewards for following traffic rules (such as speed limits, turn signals, etc.), etc.

[0067] The punishment part includes: punishments for colliding with obstacles, punishments for violating traffic rules, punishments for sudden acceleration or sudden braking, etc. On the basis of the preliminary local path, use the reward and punishment function to iteratively optimize the path. For each possible local path, calculate its reward and punishment value, and select the path with the optimal reward and punishment value as the final driving path. The optimization process can be achieved by simulating the driving situation of the vehicle on different paths and comparing their reward and punishment values.

[0068] Suppose the vehicle is driving in a complex urban environment, surrounded by high-rise buildings, pedestrians, other vehicles, and road facilities, etc. The goal of the vehicle is to drive to the next key point according to the globally optimized path, and this key point is located across a crossroads. During the vehicle's journey to the key point, it needs to avoid obstacles such as pedestrians, other vehicles, and road construction areas. The target position (i.e., the next key point in the globally optimized path) is about 100 meters in front of the vehicle. The magnitude of the gravitational force is inversely proportional to the distance between the vehicle and the target position, such as , where is the distance between the vehicle and the target position, is a constant coefficient, assumed to be 1000. The obstacles include pedestrians in front, a parked car on the left, and a road construction area on the right. The magnitude of the repulsive force is inversely proportional to the square of the distance between the vehicle and the obstacle, and within a certain distance (such as the safe distance of the obstacle, 5 meters), the repulsive force reaches the maximum value, and then decreases rapidly as the distance increases, such as ( ), where is the distance between the vehicle and the obstacle, is 500, is 1000. Superimpose the gravitational field and the repulsive field at the position of the vehicle to obtain the total force received by the vehicle at this position. The magnitude and direction of the total force are determined by the vector sum of the gravitational field and the repulsive field.

[0069] Use Euler's method to simulate the movement trajectory of the vehicle in the force field and iteratively calculate the future position of the vehicle. Suppose the current position of the vehicle is (0, 0, 0) (three-dimensional coordinates, The axis represents height. Here, it is assumed that the vehicle is traveling on the same plane, so the axis value is 0), and the target position is (100, 0, 0). According to the magnitude and direction of the total force, calculate the vehicle's next position, then update the vehicle's position, and continue to calculate the next step until the target position is reached or an obstacle is encountered.

[0070] As the vehicle approaches the target position, the reward value gradually increases, such as = 100 - , where is the distance between the vehicle and the target position. If the vehicle maintains a safe distance from the obstacle (such as 5 meters), it will receive a certain reward, such as = 50. If the vehicle follows traffic rules such as speed limits and uses turn signals, it will receive a certain reward, such as = 30. If the vehicle collides with an obstacle, it will receive a severe penalty, such as = -1000.

[0071] If the vehicle violates traffic rules, such as speeding or not using turn signals, it will receive a certain penalty, such as = -50. If the vehicle accelerates or brakes suddenly, it will receive a certain penalty, such as = -20. For the initially planned local path, calculate its reward and penalty value. Try different driving strategies (such as slightly adjusting the direction, speed, etc.), plan multiple possible local paths, and calculate the reward and penalty value of each path to obtain the final driving path.

[0072] By constructing a force field environment and dynamically optimizing the path, the vehicle can avoid obstacles in real time, maintain a safe distance, and reduce the risk of collision. The reward and penalty function can guide the vehicle to choose the optimal path, reduce unnecessary turns and stops, and increase the driving speed. The virtual force field environment can be dynamically adjusted according to actual environmental information (such as the position and shape of obstacles), enabling the vehicle to adapt to different driving environments. A smooth and safe driving path can improve the comfort of passengers and reduce the discomfort caused by bumps and sudden brakes. Local path planning is one of the key links in autonomous driving technology, and through continuous optimization and innovation, it can promote the further development and application of autonomous driving technology.

[0073] In a preferred embodiment of the present invention, the calculation formula of the reward and penalty function is: ; where represents the total value of the reward and penalty function; represents the reward coefficient of the path length; represents the planned path length; represents the weighted coefficient of safety; represents the actual distance between the vehicle and the nearest obstacle; Represents the threshold of the safety distance; Represents the penalty coefficient when the distance between the vehicle and the obstacle is less than the safety distance; Represents the weighting coefficient of the speed; Represents the preset vehicle speed; Represents the current speed of the vehicle; Represents the weighting coefficient of the acceleration; Represents the current acceleration of the vehicle.

[0074] In the embodiments of the present invention, the reward coefficient for determining the path length , which determines the influence degree of the path length in the reward and punishment function. The weighting coefficient for determining the safety , which reflects the importance of safety in path planning. Set the threshold of the safety distance , which is the minimum safety distance that should be maintained between the vehicle and the obstacle. Determine the penalty coefficient when the distance between the vehicle and the obstacle is less than the safety distance , which is used to increase the penalty for unsafe behaviors. Set the weighting coefficient of the speed , which determines the weight of the difference between the vehicle speed and the target speed in the reward and punishment function. Set the preset vehicle speed , which is the ideal speed that the vehicle should reach. Determine the weighting coefficient of the acceleration , which is used to evaluate the influence of the vehicle acceleration on the reward and punishment function.

[0075] Use measurement or calculation tools to determine the planned path length . According to the reward coefficient of the path length , calculate the reward part of the path length . The negative sign indicates that the shorter the path, the greater the reward. Measure the actual distance between the vehicle and the nearest obstacle in real time . Compare the actual distance with the safety distance threshold .

[0076] If ≥ , then the vehicle maintains a safe distance from the obstacle and calculates the safety reward .

[0077] If < , then the distance between the vehicle and the obstacle is too close, and calculate the safety penalty: - ( ).

[0078] Measure the current speed of the vehicle in real time . Calculate the difference between the current speed of the vehicle and the preset speed According to the weighted coefficient of speed , calculate the speed reward and punishment part . A positive difference indicates that the vehicle speed is lower than the target speed, and a reward is obtained; a negative difference indicates that the vehicle speed is higher than the target speed, and a punishment is received. The current acceleration of the vehicle is measured in real time . According to the weighted coefficient of acceleration , calculate the acceleration punishment part . The negative sign indicates that the greater the acceleration (whether it is acceleration or deceleration), the greater the punishment

[0079] Add the above-mentioned reward and punishment values of each part to obtain the total value of the reward and punishment function .

[0080] By setting the safety distance threshold and the punishment coefficient , when the distance between the vehicle and the obstacle is too close, the reward and punishment function will give a greater punishment, so as to prompt the vehicle to maintain a safe distance from the obstacle and reduce the collision risk. The reward coefficient of the path length encourages the vehicle to choose a shorter path, which can not only save driving time, but also reduce energy consumption and emissions. The weighted coefficient of speed and the preset speed are set so that the vehicle can maintain a reasonable speed range during driving, neither too fast nor too slow, which helps to improve driving efficiency and safety. The weighted coefficient of acceleration punishes the vehicle acceleration, encourages the vehicle to accelerate and decelerate smoothly, reduces the situations of sudden acceleration and sudden braking, and improves the riding comfort and safety. The reward and punishment function comprehensively considers multiple factors such as path length, safety, speed and acceleration, and can realize the comprehensive optimization of the vehicle driving path, and improve the overall driving efficiency and safety. By adjusting each weighted coefficient and threshold, the reward and punishment function can adapt to different driving environments and driving needs, such as urban congestion sections, highways, etc., and has strong flexibility and adaptability

[0081] In a preferred embodiment of the present invention, in the above step 15, the environmental factors of the open-pit mine are evaluated in real time to obtain the environmental factor evaluation result, and according to the environmental factor evaluation result, the final driving path is dynamically adjusted to formulate a driving plan for the driverless vehicle, which may include: Step 151, collect the environmental data of the open-pit mine in real time, including environmental factors such as climate conditions, road conditions, dust or fog concentration Step 152, convert the environmental data of the open-pit mine into an environmental factor evaluation result according to the preset rules and fuzzy logic Step 153, dynamically adjust the driving plan of the driverless vehicle according to the environmental factor evaluation result, including adjusting parameters such as the acceleration, speed, and steering angle of the vehicle Step 154: Develop a driving plan for the driverless vehicle according to the adjusted driving plan of the driverless vehicle, including the driving route, speed planning, and acceleration planning information of the vehicle.

[0082] In the embodiment of the present invention, a variety of sensors are installed on the driverless vehicle, such as weather sensors (for detecting climate conditions such as temperature, humidity, wind speed, etc.), road surface sensors (for detecting road conditions such as road flatness, wetness, etc.), and dust or fog concentration sensors. The environmental data of the open-pit mine is collected in real time through the sensors, including but not limited to climate conditions (such as temperature, humidity, wind speed, wind direction), road conditions (such as road surface flatness, wetness, pothole distribution), and dust or fog concentration. Pretreatment operations such as cleaning, denoising, and calibration are performed on the collected raw data to ensure the accuracy and reliability of the data.

[0083] Step 152: Establish an environmental factor evaluation model based on historical data. The model should be able to comprehensively consider various environmental factors and output a comprehensive evaluation result. Clear rules and fuzzy logic are set in the evaluation model. For example, when the temperature exceeds a certain threshold, the environment is considered overheated; when the wetness of the road surface reaches a certain level, the road condition is considered poor; when the dust or fog concentration exceeds the safety standard, the visibility is considered low. The real-time collected environmental data is input into the evaluation model and transformed according to the preset rules and fuzzy logic to obtain the environmental factor evaluation result. This result is a grade indicating the quality of the current environment.

[0084] Step 153: Dynamically adjust the driving parameters of the driverless vehicle according to the evaluation result. For example, at overheated or too low temperatures, reduce the vehicle speed to reduce mechanical wear; on a wet road surface, increase the steering angle to improve vehicle stability; when the dust or fog concentration is high, reduce the speed and increase the vehicle distance to ensure safety. Combine the real-time environmental data and the adjusted driving parameters to optimize the driving strategy of the driverless vehicle. For example, select a safer driving route in bad weather, or increase the number of parking inspections when the visibility is low.

[0085] Step 154: Re-plan the driving route of the driverless vehicle based on the adjusted driving parameters and strategies. The route should be able to avoid dangerous areas and ensure the vehicle arrives at the destination safely. According to the environmental factor evaluation result and the vehicle performance parameters, formulate a reasonable speed plan. The plan should be able to ensure that the vehicle maintains a stable driving speed under different road conditions. Combine the vehicle dynamics characteristics and driving requirements to formulate an acceleration plan. The plan should be able to ensure that the vehicle remains stable during acceleration and deceleration and avoid potential safety hazards caused by sudden acceleration and sudden braking.

[0086] Suppose in an open-pit mine, an autonomous vehicle is performing a transportation task. During the driving process, the sensors on the vehicle detect that the temperature suddenly rises above 40 degrees Celsius, and at the same time, the road surface sensors find that there are many potholes in the front section of the road. At this time, the environmental factor evaluation model evaluates the current environment as "high temperature and poor road conditions" according to the preset rules and fuzzy logic. Based on this evaluation result, the autonomous vehicle automatically adjusts the driving plan. First, the vehicle reduces the driving speed to reduce the wear of the engine and tires caused by high temperature; at the same time, the vehicle increases the steering angle to better adapt to the potholed road surface and ensure driving stability. Next, the vehicle re-plans the driving route according to the adjusted speed and steering angle to avoid areas with more potholes. Finally, the vehicle formulates a new speed plan and acceleration plan to ensure that it can reach the destination safely and smoothly under high temperature and potholed road conditions.

[0087] By evaluating the environmental factors of the open-pit mine in real time and dynamically adjusting the driving plan, the autonomous vehicle can better adapt to complex and changeable environmental conditions and reduce the risk of accidents. The autonomous vehicle can automatically adjust the driving parameters and strategies according to the evaluation results of environmental factors, improving the vehicle's adaptability and flexibility to different environmental conditions. By dynamically planning the driving route and speed plan, the autonomous vehicle can avoid dangerous areas and congested sections, improve driving efficiency, and reduce transportation time. In harsh environments such as high temperature and slippery roads, the autonomous vehicle reduces mechanical wear and damage by adjusting the driving parameters and strategies, thus extending the service life of the vehicle. Real-time evaluation of environmental factors and dynamic adjustment of the driving plan are important manifestations of the intelligence of autonomous vehicles, which helps to promote the intelligent and automated process of open-pit mine mining.

[0088] In a preferred embodiment of the present invention, step 16 above, according to the driving plan of the autonomous vehicle and the real-time vehicle-to-vehicle information, plans the vehicle following and intersection safety cooperation process, and performs real-time tracking on the local path to obtain the tracking result. According to the tracking result and the real-time environmental information, controlling the vehicle may include: Step 161, perform real-time tracking on the locally optimized path. According to the current state of the vehicle, including position, speed, acceleration, and a preset prediction model, obtain the final control sequence of the vehicle within a certain period of time in the future, including the steering, acceleration, and deceleration actions taken by the vehicle; Step 162, during the real-time tracking process, continuously update the prediction model according to the actual state of the vehicle and the real-time environmental information, including road conditions and obstacle positions, to obtain a comprehensive evaluation of the tracking result and the real-time environmental information; Step 163, according to the comprehensive evaluation of the tracking result and the real-time environmental information, control the vehicle, including adjusting the steering angle, acceleration, deceleration, etc. of the vehicle to ensure that the vehicle travels stably and safely along the planned path.

[0089] In the embodiments of the present invention, based on the locally optimized path in the driving scheme of the driverless vehicle, the specific tracking targets are clarified, including the starting point, the ending point, and the key points along the path. Through the sensors and the positioning system on the vehicle, the current state information of the vehicle is obtained in real time, including the position, speed, acceleration, etc.

[0090] According to the current state of the vehicle and a preset prediction model (such as a Kalman filter), the motion trajectory of the vehicle in the next period of time is predicted. Based on the predicted motion trajectory and the locally optimized path, a control sequence that the vehicle needs to take in the next period of time is generated, including the specific instructions and time points for actions such as steering, accelerating, and decelerating.

[0091] Step 162: Continuously obtain the actual state information of the vehicle through the sensors and the positioning system, and compare it with the predicted values in the prediction model. Use sensors such as radar and cameras on the vehicle to monitor the environmental information such as the road conditions and the positions of obstacles in real time. According to the deviation between the actual state of the vehicle and the predicted values and the changes in the real-time environmental information, the parameters of the prediction model are dynamically adjusted to improve the accuracy of the prediction. Combining the results of the prediction model and the real-time environmental information, the driving state of the vehicle is comprehensively evaluated to determine whether the vehicle deviates from the planned path and whether there are potential safety risks.

[0092] Step 163: Analyze the comprehensive evaluation results to determine the current control requirements of the vehicle, such as whether it is necessary to adjust the steering angle, accelerate or decelerate, etc. Based on the evaluation results and the analysis, a specific control strategy is formulated. For example, if the vehicle deviates from the planned path, it is corrected by adjusting the steering angle; if an obstacle appears ahead, deceleration or avoidance is used to ensure safety. The formulated control strategy is converted into specific control instructions, and the vehicle is controlled through the actuators of the vehicle (such as the steering system and the power system). After executing the control instructions, the actual state and environmental information of the vehicle are obtained again to perform a new round of prediction, evaluation, and control, forming a closed-loop control process.

[0093] Suppose a driverless vehicle is driving along a locally optimized path, and suddenly a falling rock appears on the road ahead. At this time, the sensors of the vehicle detect this obstacle in real time and immediately update the real-time environmental information. The prediction model predicts that if the vehicle continues to drive along the original path, it will collide with the falling rock according to this new information.

[0094] Based on this prediction result, the comprehensive evaluation system determines that the vehicle needs to take evasive measures immediately. Therefore, the control strategy is formulated to turn to the left and decelerate appropriately to avoid the falling rock. This control instruction is immediately sent to the actuators of the vehicle, and the vehicle quickly responds, successfully avoiding the obstacle and continuing to drive stably along the adjusted path.

[0095] By tracking the locally optimized path in real time and adjusting according to real-time environmental information, it is possible to timely detect and avoid obstacles, reduce the collision risk, and improve driving safety. The vehicle can dynamically adjust its driving strategy based on real-time environmental information and its own state, enhancing the vehicle's adaptability to different road conditions and environments. Through real-time tracking and prediction, the vehicle can drive more accurately along the planned path, reducing unnecessary deviations and corrections and improving driving efficiency. The control of the vehicle is more precise and timely, bringing a smoother and more comfortable driving experience to passengers. The real-time tracking and control system is an important part of driverless technology. By continuously optimizing and improving this system, the overall progress and development of driverless technology can be promoted.

[0096] As Figure 2 shown, an embodiment of the present invention also provides a path planning system for a driverless vehicle in an open-pit mine environment, including: An environment perception module for real-time collecting environmental factor data such as the climate, road conditions, dust or fog concentration of the open-pit mine environment; A vehicle state monitoring module for obtaining the current state of the driverless vehicle, including position, speed, and acceleration; A path planning module for planning the global final path of the driverless vehicle in the open-pit mine environment according to the data of the environment perception module and the vehicle state monitoring module through a preset prediction model; An obstacle detection and avoidance module for using lidar and cameras to detect obstacles on the path in real time, including static obstacles and dynamic obstacles, and adjusting the original path according to the position, speed, and moving direction of the obstacles; A real-time tracking and optimization module for real-time tracking of the global final path and obtaining the final control sequence of the vehicle in a future period of time according to the current state of the vehicle and real-time environmental information, including steering, accelerating, and decelerating actions; A vehicle control module for controlling the driverless vehicle according to the control sequence of the real-time tracking and optimization module to make the vehicle drive along the planned path.

[0097] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for unmanned vehicle path planning in an open-pit mine environment, characterized in that: The method comprises: The three-dimensional data of the open-pit mine environment is obtained through the LiDAR, and the vehicle motion status information of the inertial measurement unit is used to build a point cloud map to achieve the initial positioning of the vehicle; Establish a joint optimization model of LiDAR and IMU, input the data of LiDAR and IMU into the joint optimization model, calculate the final position of the vehicle, and update the point cloud map in real time to reflect the real-time position information of the vehicle; According to the point cloud map updated in real time, preliminary planning of the global path is carried out to obtain the global optimized path; According to the global optimization path and real-time environmental information, the local path is initially planned, and the local path is dynamically optimized through the reward and punishment function to determine the final driving path; Real-time evaluation of environmental factors in open-pit mines, obtaining environmental factor evaluation results, and dynamically adjusting the final driving path based on the environmental factor evaluation results to formulate a driving plan for unmanned vehicles; According to the driving plan of the unmanned vehicle and the real-time vehicle-to-vehicle information, the vehicle following and intersection safety coordination processes are planned, and the local path is tracked in real time to obtain the tracking results. The vehicle is controlled based on the tracking results and real-time environmental information.

2. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 1, characterized in that: The three-dimensional data of the open-pit mine environment is obtained through the LiDAR, and the vehicle motion status information of the inertial measurement unit is used to build a point cloud map to achieve the initial positioning of the vehicle, including: Use laser radar to collect 3D data of the open-pit mine environment, and pre-process the 3D data to generate initial point cloud data; The initial point cloud data is corrected and optimized through the vehicle motion status information provided in real time by the inertial measurement unit to construct a point cloud map; Based on the point cloud map, the vehicle’s initial positioning is achieved.

3. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 2, characterized in that: Establish a joint optimization model of LiDAR and IMU, input the data of LiDAR and IMU into the joint optimization model, calculate the final position of the vehicle, and update the point cloud map in real time to reflect the real-time position information of the vehicle, including: The point cloud data collected by the laser radar and the acceleration and angular velocity motion state data provided by the inertial measurement unit are input into the joint optimization model; An error function is defined to quantify the point cloud matching error, the motion state error, and the fusion error between the matching error and the motion state error.

4. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 3, characterized in that: Establish a joint optimization model of LiDAR and IMU, input the data of LiDAR and IMU into the joint optimization model, calculate the final position of the vehicle, and update the point cloud map in real time to reflect the real-time position information of the vehicle, including: The error function is iteratively optimized. In each iteration, the point cloud matching error and the motion state error provided by the inertial measurement unit are recalculated according to the current estimated vehicle pose. Continuously iterate and optimize, and when the preset number of iterations is reached, the final position and posture of the vehicle is output, including position, speed and posture information; The point cloud map is updated in real time based on the final position of the vehicle.

5. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 4, characterized in that: According to the point cloud map updated in real time, preliminary global path planning is carried out to obtain the global optimized path, including: Determine the starting and ending points of the path planning in the global map, and convert the starting and ending points into coordinate points in the map coordinate system; Create an open list and a closed list, and add the starting node to the open list; Select the current node from the open list, remove the current node from the open list, add it to the closed list, and mark it as evaluated.

6. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 5, characterized in that: According to the point cloud map updated in real time, preliminary global path planning is carried out to obtain the global optimized path, including: Traverse all the adjacent nodes of the current node, and add the adjacent nodes to the open list, set the parent node, or update the cost and parent node according to whether the adjacent nodes are in the closed list or in the open list; When expanding adjacent nodes, the Euclidean distance is used to estimate the distance from the adjacent node to the end point to guide the search direction. When the search reaches the end point or cannot proceed, a global optimization path from the start point to the end point is constructed by backtracking the parent node pointer.

7. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 6, characterized in that: According to the global optimization path and real-time environmental information, the local path is initially planned, and the local path is dynamically optimized through the reward and penalty function to determine the final driving path, including: Construct a virtual force field environment around the vehicle, where the target position generates gravity to guide the vehicle forward, and obstacles generate repulsion; In the force field environment, a preliminary local path of the vehicle is planned, and the local path is dynamically optimized through the reward and penalty function to determine the final driving path.

8. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 7, characterized in that: Real-time assessment of the environmental factors of the open-pit mine to obtain the environmental factor assessment results, and dynamically adjust the final driving path based on the environmental factor assessment results to formulate the unmanned vehicle driving plan, including: Real-time collection of environmental data of open-pit mines, including climate conditions, road conditions, and dust or fog concentration environmental factors; According to preset rules and fuzzy logic, the environmental data of open-pit mines are converted into environmental factor assessment results; Dynamically adjust the driving plan of the unmanned vehicle based on the environmental factor assessment results, including adjusting the vehicle's acceleration, speed, and steering angle parameters; According to the adjusted driving plan of the unmanned vehicle, a driving plan for the unmanned vehicle is formulated, including the vehicle's driving route, speed planning, and acceleration planning information.

9. The method for unmanned vehicle path planning in an open-pit mine environment according to claim 8, characterized in that: According to the unmanned vehicle driving plan and real-time vehicle-to-vehicle information, the vehicle following and cross-traffic safety coordination process is planned, and the local path is tracked in real time to obtain the tracking results. According to the tracking results and real-time environmental information, the vehicle is controlled, including: Track the local optimization path in real time, and obtain the final control sequence of the vehicle in the future, including the steering, acceleration, and deceleration actions taken by the vehicle, based on the vehicle's current state, including position, speed, acceleration, and preset prediction models; During real-time tracking, the prediction model is continuously updated based on the actual status of the vehicle and real-time environmental information, including road conditions and obstacle locations, to obtain a comprehensive evaluation of the tracking results and real-time environmental information; Based on the comprehensive evaluation of the tracking results and real-time environmental information, the vehicle is controlled, including adjusting the vehicle's steering angle, acceleration, deceleration, etc., to ensure that the vehicle travels stably and safely along the planned path.

10. A path planning system for an unmanned vehicle in an open-pit mine environment, the system implementing the method according to any one of claims 1 to 7, characterized in that: include: Environmental perception module, used to collect real-time data on environmental factors such as climate, road conditions, and dust or fog concentration in the open-pit mine environment; The vehicle status monitoring module is used to obtain the current status of the unmanned vehicle, including position, speed, and acceleration; The path planning module is used to plan the global final path of the unmanned vehicle in the open-pit mine environment based on the data of the environment perception module and the vehicle status monitoring module through the preset prediction model; The obstacle detection and avoidance module is used to detect obstacles on the path in real time using laser radar and cameras, including static obstacles and dynamic obstacles, and adjust the original path according to the position, speed and movement direction of the obstacles; The real-time tracking and optimization module is used to track the global final path in real time and obtain the final control sequence of the vehicle in the future, including steering, acceleration, and deceleration actions, based on the current state of the vehicle and real-time environmental information; The vehicle control module is used to control the unmanned vehicle according to the control sequence of the real-time tracking and optimization module, so that the vehicle travels along the planned path.

Images