Driving information planning method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a driving information planning method and device, electronic equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: acquiring an initial position and a target position of each first mine car in a mining area; for each first tramcar, constructing a reference line of the tramcar based on the initial position of the tramcar and the target position of the tramcar; for each first mine car, first driving information of the mine car is obtained through a kinematic model of the mine car and a reference line of the mine car; and for each first tramcar, obtaining second driving information of the first tramcar based on the first driving information of each second tramcar and the reference line of the tramcar. Each mine car can combine the driving information obtained through prediction construction of other mine cars with the reference line of the mine car, so that actual driving information which does not collide with other mine cars is planned, and an intelligent center body for mine car scheduling does not need to be independently arranged. Therefore, the reasoning complexity is not increased in a super-linear manner along with the increase of mine cars, and the calculation cost is saved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically, to a driving information planning method, apparatus, electronic device, and storage medium. Background Technology

[0002] In real-world applications such as mines, using mining trucks for mining and transportation is a common practice. To ensure the quality and efficiency of these operations, it is crucial to prevent collisions between the trucks. Current solutions primarily involve establishing a "central command point" within the mining area. This central command point controls each mining truck sequentially, enabling free scheduling of each truck and thus preventing collisions.

[0003] However, in real-world applications, the quality of communication links within mining areas is unstable and prone to disconnections. Some mining trucks that are far from the "central command point" may not be able to receive the dispatch instructions from the "central command point," which may lead to mining truck collisions. Therefore, this scheme that relies on central dispatch is difficult to maintain continuous service. Moreover, in terms of real-time performance, the reasoning complexity of this centralized dispatch method increases linearly or even superlinearly with the number of mining trucks and the scale of joint output. The generation of joint trajectories for multiple vehicles is difficult to meet strict real-time constraints. Summary of the Invention

[0004] The purpose of this application is to at least solve one of the aforementioned technical defects. The technical solution provided by the embodiments of this application is as follows: In a first aspect, embodiments of this application provide a driving information planning method, including: Obtain the initial and target positions of each mining vehicle within the mining area; For each first mining car, a reference line is constructed based on the initial position and the target position of the first mining car; the reference line is used to characterize the path of the first mining car from the initial position to the target position in the mining area; the first mining car can be any mining car in the mining area; For each first mining car, the first driving information of the first mining car is obtained through its own kinematic model and its own reference line; For each first mining car, the first driving information of the first mining car is updated based on the first driving information of each second mining car and the reference line of the mining car. The updated first driving information is used as the second driving information of the first mining car. Both the first driving information and the second driving information include the reference line of the corresponding mining car and the motion state at multiple points on the reference line. The second driving information of each mining car does not have the same point at the same time. The second mining car is any other mining car in the mining area other than the first mining car.

[0005] Secondly, embodiments of this application provide a driving information planning device, comprising: The location acquisition module is used to obtain the initial and target locations of each mining truck within the mining area; The reference line construction module is used to construct a reference line for each first mining car based on its initial position and target position. The reference line is used to represent the path of the first mining car from its initial position to its target position in the mining area. The first mining car can be any mining car in the mining area. The first driving information determination module is used to obtain the first driving information of each first mining car by using its own kinematic model and its own reference line. The second driving information determination module is used to update the first driving information of each first mine car based on the first driving information of each second mine car and the reference line of the mine car, and use the updated first driving information as the second driving information of the first mine car; both the first driving information and the second driving information include the reference line of the corresponding mine car and the motion state at multiple points on the reference line; the second driving information of each mine car does not have the same point at the same time; the second mine car is any other mine car in the mining area other than the first mine car.

[0006] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory; The processor executes a computer program to implement the method provided in the first aspect embodiment or any alternative embodiment of the first aspect.

[0007] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method provided in the first aspect embodiment or any optional embodiment of the first aspect.

[0008] The beneficial effects of the technical solutions provided in this application are: First, each first mining car can determine its own first driving information based on its own kinematic model and position information, and send this first driving information to each second mining car, so that each first mining car can know the driving status of each second mining car and make subsequent adjustments to its own driving situation.

[0009] Secondly, each first mining truck can plan its own reference line based on its current position and target position, and then combine its own reference line with the first driving information of each second mining truck. In this way, it can avoid collisions with each other when planning its own second driving information.

[0010] The solution provided in this application embodiment allows each first mining truck to obtain driving information predicted by surrounding second mining trucks, and then combine the driving information predicted by each second mining truck with its own reference line to plan actual driving information that does not collide with other mining trucks. This eliminates the need to set up a separate intelligent center for mining truck scheduling in the mining area. Furthermore, since each first mining truck plans its driving information with itself as the center, the reasoning complexity does not increase superlinearly with the increase in the number of first mining trucks, thus saving computational costs. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.

[0012] Figure 1 A flowchart illustrating a driving information planning method provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the overall process of a driving information generation method in one example of an embodiment of this application. Figure 3 This is an example of a mine car control simulation diagram in one embodiment of this application; Figure 4 A structural block diagram of a driving information planning device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0013] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0014] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.”

[0015] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0016] The technical solutions of this application and their effects are described below through several exemplary embodiments. It should be noted that the following embodiments can be referenced, borrowed from, or combined with each other. Identical terms, similar features, and similar implementation steps in different embodiments will not be repeated.

[0017] Figure 1 This application provides a flowchart illustrating a driving information planning method. The execution subject of this method can be a terminal (e.g., a computer, mobile phone, etc.) or an intelligent agent, such as... Figure 1 As shown, the method may include: Step S101: Obtain the initial and target positions of each mining vehicle in the mining area.

[0018] In embodiments of this application, a mining area can be an area containing exploitable mineral resources and where mining activities are carried out. A mining car can be a vehicle or rail vehicle used in the mining area to transport ore, waste rock, equipment, or personnel. The initial location can be the geographical location of the mining car in the mining area before it begins its journey, while the target location can be the geographical location the mining car needs to reach within the mining area.

[0019] Specifically, in order to plan the driving information of mining trucks, it is necessary to first determine the approximate driving route of each mining truck. Determining the driving route requires obtaining the current geographical location information of each mining truck and the geographical location information of the destination it needs to reach. The geographical location information can be represented by coordinates.

[0020] Step S102: For each first mining car, construct a reference line for the mining car based on its initial position and target position; the reference line is used to characterize the path of the first mining car from its initial position to its target position in the mining area; the first mining car is any mining car in the mining area.

[0021] In the embodiments of this application, the reference line can be the route taken by the first mine car from the initial position to the target position. This route is not a straight line from the initial position to the target position, but a smooth curve formed by connecting the actual terrain of the mining area.

[0022] Specifically, after obtaining the initial and target positions of each first mining truck, the driving path of the mining truck can be planned in combination with the terrain of the mining area itself. Optionally, a map representing the terrain of the mining area can be pre-entered into the agent of each first mining truck. Then, the agent combines the coordinates of the current position and the target position of the corresponding first mining truck with the map representing the terrain of the mining area to obtain the path from the initial position to the target position.

[0023] Step S103: For each first mine car, obtain the first driving information of the first mine car through its own kinematic model and its own reference line.

[0024] In the embodiments of this application, the kinematic model can be a mathematical model that includes the initial state of each first mining truck (such as position, speed, orientation angle, etc.) and the control quantities of the first mining truck (such as acceleration, steering wheel angle, etc.), and is used to deduce the position of the first mining truck at the next moment. The first driving information can be the possible driving route planning result of each first mining truck simulated by the kinematic model. It is understood that since the first driving information is only obtained through its own kinematic model planning, it does not take into account whether there will be a collision with other mining trucks.

[0025] Specifically, in order to facilitate the actual driving route planning of other mining trucks, each first mining truck also needs to provide other mining trucks with its own predicted driving route planning information. Specifically, each first mining truck, knowing its own reference line, combined with a motion model that includes its own motion state, predicts its own position on its own reference line at various times. This can be understood as setting multiple points on its own reference line, and each point represents its own geographical location at the corresponding time.

[0026] For the kinematic model of the agent, the mining truck used in the mining scenario is modeled as a nonholonomic kinematically constrained model:

[0027] in, The state vector represents the set of all state variables of the mining truck (or vehicle), i.e. [x,y, ,v] T .

[0028] : The vehicle's X-axis position (m) in the current coordinate system.

[0029] : The vehicle's Y-axis position (m) in the current coordinate system.

[0030] : The vehicle's heading angle. That is, the angle (rad) between the direction of the vehicle's front and the positive direction of the X-axis.

[0031] : The vehicle's forward speed (m / s).

[0032] : The velocity component of the vehicle in the X-axis direction (m / s).

[0033] : The velocity component of the vehicle in the Y-axis direction (m / s).

[0034] : The yaw rate of a vehicle, i.e., the rate of change of its heading angle (rad / s).

[0035] Vehicle acceleration (m / s²) 2 ).

[0036] : Control input vector. u contains two variables: .

[0037] : The steering angle of the vehicle's front wheels (rad).

[0038] Vehicle acceleration (m / s²) 2 ).

[0039] Wheelbase (m): The distance between the front and rear axles of a vehicle.

[0040] Step S104: For each first mine car, update the first driving information of the first mine car based on the first driving information of each second mine car and the reference line of the mine car, and use the updated first driving information as the second driving information of the first mine car; both the first driving information and the second driving information include the reference line of the corresponding mine car and the motion state at multiple points on the reference line; the second driving information of each mine car does not have the same point at the same time; the second mine car is any other mine car in the mining area other than the first mine car.

[0041] In the embodiments of this application, the second driving information may be the actual driving route planning result of the mining truck after combining the first driving information of other mining trucks.

[0042] Specifically, after obtaining the first driving information of each second mine car, each first mine car can combine the first driving information of each second mine car with its own reference line to deduce the possible times when each second mine car may appear on its own reference line. Then, based on these times, it plans its own driving information to avoid appearing at the same point on the reference line at the same time as each second mine car.

[0043] In this planning problem, the optimization problem of planning the driving information of a single first mining truck can be represented as: considering a total of K second mining trucks, where the strategy of a single first mining truck is π. i Given a prediction time domain of N, the objective function of the optimization strategy for the i-th mining truck can be expressed as:

[0044] in, : represents the maximum distance traveled by the first mining car i on reference line p;, which depends on the positions of all N mining cars. That is, vehicle i attempts to maximize its relative travel distance with each of the second mining cars.

[0045] : indicates the influence weight of each second mine car j.

[0046] Let : represent the maximum forward distance of each second minecart j on reference line p;. Therefore, the objective is to maximize the maximum forward distance of the first minecart i on the reference line, while also considering the impact of the forward distance of each second minecart on the reference line. This objective function indicates that each minecart wants to move forward as far as possible, while trying to avoid collisions with any of the second minecarts.

[0047] In addition, some constraints need to be added during the above planning process to prevent the optimized driving information from failing to be realized. The specific constraints are as follows:

[0048] Among them, 1. Constraint 1: This constraint represents the state change model of the mining truck, where u is the control input. This equation describes how the mining truck dynamically changes based on its current state and the control input.

[0049] 2. Constraint 2: This constraint ensures that the minimum distance between mine car i and mine car j is d. min This is to avoid collisions. It is a key constraint in multi-agent systems, ensuring that all mining trucks do not collide during operation.

[0050] 3. Constraint 3: This constraint ensures that the minecart's state (such as position and speed) remains within certain upper and lower bounds. For example, the speed cannot exceed a set maximum value. It cannot be lower than the minimum value This ensures that the movement of the mine cars remains within a safe range.

[0051] 4. John 4:

[0052] This constraint ensures that the control inputs of the mine car (such as acceleration and direction control) are also within a certain range, preventing excessively fast or slow acceleration, or control inputs exceeding physical limits.

[0053] 5. Constraint 5:

[0054] This constraint sets the initial state of the minecart. The current state is This applies to all mining trucks. This means that at the start of optimization, the initial state of all mining trucks is known and represents the current state of the system. The above process uses a finite state machine-based method to approximate the Nash equilibrium point of the optimization problem, thus satisfying the trajectory distribution of the dataset.

[0055] This non-cooperative problem can be modeled as a zero-sum Boyce problem. N i This represents the position of the i-th vehicle at the N-th step in the prediction time domain. d represents the distance the mine car has traveled along the reference line. min Let J represent the minimum safe distance between mining trucks. For the i-th truck, the optimal value of the optimization problem is denoted as J. i This is used to characterize its advantage in forward efficiency compared to other mining trucks. As the strategy converges, the multi-agent system as a whole reaches a zero-sum state. The trajectories of each agent gradually approach the Nash equilibrium.

[0056] It is understandable that as long as two minecarts do not appear at the same location at the same time, it can be said that the two minecarts will not collide. Therefore, through the above method, the free scheduling between each minecart can be achieved without the occurrence of minecart collisions.

[0057] Optionally, in the above constraints on the upper and lower limits of the state variables, the upper and lower limits can each be taken as 1.2 times their respective ranges to avoid excessive deviation of the state variables during the solution process.

[0058] The solution provided in this application embodiment is that, firstly, each first mining car can determine its own first driving information based on its own kinematic model and position information, and send this first driving information to each second mining car, so that each first mining car can know the driving status of each second mining car so as to adjust its own driving status in the future.

[0059] Secondly, each first mining truck can plan its own reference line based on its current position and target position, and then combine its own reference line with the first driving information of each second mining truck. In this way, it can avoid collisions with each other when planning its own second driving information.

[0060] The solution provided in this application embodiment allows each first mining truck to obtain driving information predicted by surrounding second mining trucks, and then combine the driving information predicted by each second mining truck with its own reference line to plan actual driving information that does not collide with other mining trucks. This eliminates the need to set up a separate intelligent center for mining truck scheduling in the mining area. Furthermore, since each first mining truck plans its driving information with itself as the center, the reasoning complexity does not increase superlinearly with the increase in the number of first mining trucks, thus saving computational costs.

[0061] Based on the above embodiments, as an optional embodiment, the first driving information of the first mine car is updated based on the first driving information of each second mine car and the reference line of the mine car, and the updated first driving information is used as the second driving information of the first mine car, specifically including: The first driving information of each second mining car is encoded by a preset encoder to obtain the first feature of each second mining car. The reference line of the first mining car is encoded by the preset encoder to obtain the second feature of the first mining car. The first feature and the second feature have the same dimension. The first and second features of each second mine car are merged to obtain the merged feature; The initial random Gaussian noise and the first diffusion time step are obtained. The initial random Gaussian noise is denoised based on the fusion features and the first diffusion time step to obtain the second driving information of the first mining truck.

[0062] In the embodiments of this application, a preset encoder can be used to encode information such as driving information or reference lines into corresponding feature forms. The initial random Gaussian noise is a purely random data matrix conforming to a Gaussian distribution (normal distribution) that serves as the starting point of the image generation process in generative artificial intelligence based on a diffusion model. The diffusion time step can be a core parameter in the diffusion model used to quantify and control the progress of the "noise addition" and "denoising" processes; it is a discrete step index that marks a specific stage in the data transformation between pure noise and structured data.

[0063] Specifically, in order to visualize driving information as data, this application embodiment can use a preset encoder to encode each piece of driving information, converting the driving information into a corresponding feature representation. Since the reference line and driving information contain different contents, in order to facilitate subsequent fusion, multiple identical preset encoders can be used to encode them into feature representations with the same dimension. After obtaining the first feature and the second feature with the same dimension, in order to combine each piece of first driving information and the reference line, each piece of first feature and the second feature can be fused to obtain a fused feature containing each piece of first driving information and reference line information.

[0064] In practice, the reference line of each first mining truck will be normalized to its own coordinate system, represented as... Where L represents the number of points on the reference line. The endpoint of the reference line is the target position of the first mining truck, and the starting point is set to be 1 meter backward from the projection point of its current position on the reference line, to ensure that the reference line contains sufficient points. The reference line and driving information are then input into the preset encoder. Previously, the reference line needed to be resampled first. To ensure a fixed input dimension, denoted as The driving information for each second mining car is represented as follows: Where M is the prediction time domain and the time interval is... Consistent with the current mining truck itself. Because the driving information prediction model captures game-theoretic behavioral patterns in the training data after training, the prediction time domains for the first mining truck and other second mining trucks can differ. The encoder used by each second mining truck is... The reference lines and the first driving information of each second mine car are encoded using the method described in the multilayer perceptron.

[0065] Each pre-defined encoder maps its input to a fixed-length latent representation (i.e., a first feature and a second feature). These latent representations are fused and spliced ​​to form a fused representation, which is then used by a projection layer to generate a unified conditional embedding. This conditional embedding can serve as a compact and information-rich descriptor of the mining truck's operating environment.

[0066] Next, based on the above-mentioned fusion features, the second driving information of the first mining truck is planned. In this embodiment, the second driving information is obtained by using a diffusion model and a step-by-step denoising method. Specifically, the initial random Gaussian noise that needs to be denoised can be obtained, and the number of times denoising needs to be performed (i.e., the first diffusion time step) can be set according to the requirements. Then, the initial random Gaussian noise is denoised for the number of times the first diffusion time step is performed according to the above-mentioned fusion features. Finally, the second driving information of the first mining truck can be obtained after the denoising is completed.

[0067] Based on the above embodiments, as an optional embodiment, the initial random Gaussian noise is denoised based on the fusion features and the first diffusion time step to obtain the second driving information of the first mining truck, specifically including: The fused features, initial random Gaussian noise, and first diffusion time step are input into the pre-trained driving information prediction model of the first mining truck to obtain the initial driving information of the first mining truck. The driving information prediction model for each first mining truck is trained in the following way: Multiple training samples and corresponding training labels are obtained. Each training sample contains the fusion features of the first mining truck, the initial noisy driving information, and the third diffusion time step. The training label is the actual driving information of the first mining truck. The initial noisy driving information in each training sample is obtained by adding noise to the corresponding training label several times in the third diffusion time step. The first mining truck did not collide with any of the second mining trucks when driving according to the actual driving information. For each training sample, the training sample is input into the initial model, and the second driving information corresponding to the training sample is output. Based on the second driving information, the third diffusion time step, and the training label corresponding to each training sample, the training loss of the initial model is determined, and the model parameters of the initial model are adjusted based on the training loss until the preset training termination condition is met, thus obtaining the driving information prediction model.

[0068] In the embodiments of this application, the initial model can be an untrained denoising diffusion model, while the driving information prediction model can be a trained denoising diffusion model. The input of the model can be the initial noise that needs to be denoised (initial random Gaussian noise in this application), the number of times denoising needs to be performed (the first diffusion time step in this application), and the basis for the denoising process (fusion features in this application).

[0069] Specifically, since each first mining truck has a different kinematic model and a certain structural difference, in order to avoid the impact of the above differences on the driving information planning results, each first mining truck will be trained with its own driving information prediction model in this embodiment.

[0070] In some embodiments, each training sample includes the fusion features of the first mining vehicle, initial noisy driving information, and a third diffusion time step. The training label is the actual driving information of the first mining vehicle, and the initial noisy driving information in each training sample is obtained by adding noise to the corresponding training label several times in the third diffusion time step.

[0071] In this embodiment, the actual driving information can be a simulated driving path for the first mining truck that will not collide, based on practical experience. The initial noisy driving information can be the driving information obtained by adding noise to the actual driving information a certain number of times. The method used for adding noise can be random Gaussian noise, and the number of noise additions is the third diffusion time step. The third diffusion time step can reflect the degree of noise addition. Generally speaking, the larger the value, the larger the number of steps and the greater the added noise.

[0072] Then, the initial model is trained. The training method involves sequentially inputting the training samples into the initial model, which then outputs the result corresponding to each training sample (i.e., the second driving information). After each training iteration, the training loss of the initial model is determined using the second driving information, the third diffusion time step, and the training labels. The model parameters are then adjusted based on this training loss until training is complete, resulting in a well-trained driving information prediction model. The model loss in this process is calculated as follows:

[0073] in, Represents initial random Gaussian noise, After the third time step Initial noise-added driving information after noise addition.

[0074] Based on the above embodiments, as an optional embodiment, the fused features, initial random Gaussian noise, and first diffusion time step are input into a pre-trained prediction model for the driving information of the first mining truck to obtain the initial driving information of the first mining truck, specifically including: Based on the fusion features and the first diffusion time step, the initial random Gaussian noise is iterated for a preset number of rounds. The denoised Gaussian noise output in the last iteration is used as the second driving information. The preset number of rounds is the first diffusion time step. Each iteration includes: Obtain the denoised Gaussian noise for this iteration, where the denoised Gaussian noise for the first iteration is the initial random Gaussian noise; Based on the second diffusion time step, fusion features, and denoised Gaussian noise of this iteration, prediction is made to obtain the predicted noise of this iteration. The predicted noise is the part to be removed from the denoised Gaussian noise of this iteration. Based on the predicted noise of this iteration, the denoised Gaussian noise of this iteration is denoised to obtain the denoised Gaussian noise of the next iteration.

[0075] In the embodiments of this application, the denoised Gaussian noise can be an intermediate product in the process of obtaining the second driving information from the initial random Gaussian noise, that is, the Gaussian noise obtained by each step of the denoising process.

[0076] Specifically, the driving information prediction model in the embodiments of this application mainly includes two processes: a forward noise addition process and a reverse noise reduction process. The forward noise addition process is achieved through... Internal driving information Gradually add noise until the driving information is completely covered by Gaussian noise. At this point, the data... Only compared with the data from the previous time point Related. The formula for the forward process is as follows: (1) in It is the variance scheduling parameter. After a sufficient number of noise-adding steps... Finally, the trajectory distribution converges to: .

[0077] In the reverse diffusion process, the diffusion model gradually denoises and restores the original driving information from a set of noisy driving information. The formula for the reverse process is as follows: (2) in, and All The function.

[0078] Returning to this embodiment, specifically, the initial random Gaussian noise needs to be denoised by the model for a first preset number of time steps to complete a complete denoising process. Each iteration can be seen as a denoising process on the initial random Gaussian noise. After all iterations are completed, the fully denoised second driving information can be obtained. In each iteration of denoising, the noise portion that needs to be removed from the Gaussian noise of that round is actually predicted, and then the predicted noise portion is removed from the denoised Gaussian noise to obtain the denoised Gaussian noise of the next round. This process continues until the last round of denoising is completed, at which point the second driving information is obtained.

[0079] Based on the above embodiments, as an optional embodiment, the driving information prediction model includes multiple cascaded layer structures, each of which contains hidden features for supplementing lost feature information. Based on the second diffusion time step, fusion features, and denoised Gaussian noise of this iteration, the predicted noise for this iteration is obtained, specifically including: For each layer, if it is the first layer, the hidden features of the layer are used to supplement the fused features to obtain the third feature of the layer. If it is not the first layer, the features of the non-overlapping part between the denoised Gaussian noise of the current iteration and the denoised Gaussian noise of the next iteration are predicted based on the second diffusion time step and the third feature of the previous layer. These features are then used as the third feature. The third features of each layer are fused to obtain the fourth feature. The predicted noise is then obtained based on the fourth feature. The third feature of the first layer structure is determined as follows: based on the second diffusion time step and the denoised Gaussian noise of the current iteration, the features of the non-overlapping part between the denoised Gaussian noise of the current iteration and the denoised Gaussian noise of the next iteration are predicted and used as the third feature.

[0080] Specifically, due to the excessive number of network layers during the training process of deep neural networks, errors can easily decrease or even disappear during backpropagation. Therefore, to address this issue, the driving information prediction model in this embodiment can be configured with multiple residual structure layers to recover and supplement lost feature information. Specifically, the second diffusion time step and the denoised Gaussian noise are first fed into the first layer structure. In the first layer structure, the difference between the denoised Gaussian noise that needs denoising in the current iteration and the denoised Gaussian noise after the current iteration's denoising operation (i.e., the third feature) is predicted. Then, the second diffusion time step, the third feature, and the denoised Gaussian noise obtained after denoising in the first layer structure are input into the second layer structure. The second layer structure again predicts the difference between the denoised Gaussian noise that needs denoising in the current iteration and the denoised Gaussian noise after the current iteration's denoising operation. This can be represented as follows:

[0081] in, For the first The third characterization of layer residual structure.

[0082] Then, the second diffusion time step, the third feature predicted by the second layer structure, and the denoised Gaussian noise are input into the third layer structure to continue prediction. This process is repeated until each residual structure layer has predicted its own third feature. Then, the third features are fused to obtain the fourth feature. The third features here are predicted by different residual layer structures. The residual layers are directly complementary, so the fourth feature obtained after fusion contains complete feature information. Then, the fourth feature is used to predict the noise, so that the noise to be removed can be obtained more accurately.

[0083] Based on the above embodiments, as an optional embodiment, the first and second features of each second mine car are fused to obtain a fused feature, specifically including: The first features of each second minecart are merged to obtain the second feature; The second feature is fused with the first feature of the first mine car to obtain the fused feature.

[0084] Specifically, the first feature of each second mining car in this application embodiment is essentially to provide a reference "driving environment" for the first mining car. That is, the driving mode of the first mining car is determined according to the current "driving environment". Therefore, when fusing the first features of each mining car, it is not possible to simply fuse the first features of all mining cars directly. Instead, it is necessary to first fuse the first features of each second mining car except the first mining car to obtain the second feature used to describe the "driving environment". Then, the second feature used to describe the "driving environment" is fused with the first feature of the first mining car to make the planned driving information more accurate.

[0085] Based on the above embodiments, as an optional embodiment, the second driving information includes the driving direction of the first mine car when it reaches each point; the driving direction of each point is determined according to the reference line; After obtaining the second driving information of the first mining truck, the process also includes optimizing the second driving information: For each first mine car, the second driving information of the first mine car is divided into multiple driving information segments. The average value of the driving direction angle of each point in each driving information segment is determined, and the distance between the first point and the last point in each driving information segment is obtained. For each segment of driving information, the average speed value of the segment is determined based on distance and average value.

[0086] Specifically, since the driving information obtained from the driving information prediction model cannot guarantee control feasibility, additional path and speed planning is required. The point spacing obtained from the driving information prediction model consists of several points at equal time intervals, which implicitly contain speed information. However, directly using differential calculations to obtain the speed can lead to speed jumps or even negative speeds due to the scattered points. Therefore, the speed in the second driving information needs to be smoothed. The heading angle obtained by the diffusion model is basically stable. The average displacement in the heading direction can be calculated using a sliding window to obtain the velocity.

[0087] The following example will describe the processing procedure in this solution: First, calculate the average heading angle of each heading angle in the second driving information. The average heading angle can be calculated in the following way:

[0088] Where N represents the number of sites taken in the smoothing window (5 for example), and i represents the i-th site in the diffusion model, with a maximum of 40 sites. If there are fewer than 5 sites, the sliding window range is reduced until N=1.

[0089] Next, we calculate the displacement components in the heading direction:

[0090] in, This represents the displacement component in the horizontal direction. y represents the displacement component in the vertical direction. It represents the resultant velocity in two directions.

[0091] Further calculations can yield the velocity value (i.e., v). i ):

[0092] To avoid negative speed, amplitude limiting control is also implemented.

[0093] Finally, the array lengths are aligned using methods such as interpolation.

[0094] In another optional embodiment of this application, after obtaining the second driving information and speed information, the driving information can be further optimized using MPC (Model Predictive Control) in conjunction with the kinematic model of each first mining car to ensure that it meets the kinematic constraints: Discretizing and linearizing the kinematic model of the first mine car using a first-order Taylor expansion yields the following form:

[0095]

[0096]

[0097] Among them, A k B k These are the state matrices of the vehicle's kinematic model, and their specific values ​​are related to the heading angle, velocity, and front wheel steering angle of the first mining truck at time k, which are at the reference position. k Let be the error at time k.

[0098] For ease of solution, the above optimization process can be transformed into a quadratic programming problem. Taking the control variable U as the optimization variable, the following prediction equation can be obtained:

[0099]

[0100]

[0101] Where T, U, and V are the state transition matrices obtained from the state equations, D is the column vector composed of errors at all times, and U is the column vector composed of all control variables.

[0102] The first few terms of the error d after Taylor expansion can be expressed as:

[0103] Further optimization issues will be addressed as follows:

[0104] Where U is the control sequence, X is the state sequence, and X ref This is a sequence of reference values ​​for state variables. Let be the reference value at time N. Here, Q is the state error weight matrix of the adjustable weight parameter, R is the control quantity weight matrix of the adjustable weight parameter, and P is the terminal penalty matrix P of the adjustable weight parameter.

[0105] Combining the above sequences, we can further obtain the Hession and f matrices, which characterize the driving information of the first mining truck:

[0106]

[0107] The driving information planning method proposed in the embodiments of this application will be described in general below. Figure 2 This is a schematic diagram of the overall process of a driving information generation method provided in an embodiment of this application, as shown below. Figure 2 As shown, this method can be roughly divided into two parts. The first part is the driving information generation part. This part mainly generates the driving information of the first mining truck by using model denoising and combining it with the driving information of each second mining truck. This includes the process of encoding the reference line of the first mining truck and the first driving information predicted by each second mining truck into feature vectors, and the process of fusing the above features. Then, the initial random Gaussian noise is iteratively denoised based on the fused features. After the iterative denoising is completed, the second driving information of the first mining truck can be generated. The main purpose of this part is to enable each first mining truck to plan its own driving route by taking itself as the center and treating the driving information predicted by each second mining truck as the "driving environment".

[0108] The second part involves optimizing the second driving information generated in the first part. This part mainly involves smoothing the second driving information generated in the first part and calculating the speed values ​​of multiple points in the second driving information. After the smoothing and speed calculation steps are completed, the optimal driving mode for the mining truck is determined by linear programming, and this mode is used as the optimized second driving information. The main purpose of this part is to correct the previously planned driving information and further improve the safety of the driving route.

[0109] Figure 4 A structural block diagram of a driving information planning device provided in an embodiment of this application is shown below. Figure 4 As shown, the driving information planning device 400 may include: a location acquisition module 401, a reference line construction module 402, a first driving information determination module 403, and a second driving information determination module 404, wherein, The location acquisition module 401 is used to acquire the initial and target positions of each mining vehicle in the mining area; The reference line construction module 402 is used to construct a reference line for each first mining car based on the initial position and the target position of the first mining car; the reference line is used to characterize the path of the first mining car from the initial position to the target position in the mining area; the first mining car can be any mining car in the mining area; The first driving information determination module 403 is used to obtain the first driving information of each first mining car through its own kinematic model and its own reference line. The second driving information determination module 404 is used to update the first driving information of each first mine car based on the first driving information of each second mine car and the reference line of the mine car, and use the updated first driving information as the second driving information of the first mine car; both the first driving information and the second driving information include the reference line of the corresponding mine car and the motion state at multiple points on the reference line; the second driving information of each mine car does not have the same point at the same time; the second mine car is any other mine car in the mining area other than the first mine car.

[0110] The solution provided in this application embodiment is that, firstly, each first mining car can determine its own first driving information based on its own kinematic model and position information, and send this first driving information to each second mining car, so that each first mining car can know the driving status of each second mining car so as to adjust its own driving status in the future.

[0111] Secondly, each first mining truck can plan its own reference line based on its current position and target position, and then combine its own reference line with the first driving information of each second mining truck. In this way, it can avoid collisions with each other when planning its own second driving information.

[0112] The solution provided in this application embodiment allows each first mining truck to obtain driving information predicted by surrounding second mining trucks, and then combine the driving information predicted by each second mining truck with its own reference line to plan actual driving information that does not collide with other mining trucks. This eliminates the need to set up a separate intelligent center for mining truck scheduling in the mining area. Furthermore, since each first mining truck plans its driving information with itself as the center, the reasoning complexity does not increase superlinearly with the increase in the number of first mining trucks, thus saving computational costs.

[0113] Based on the above embodiments, as an optional embodiment, the second driving information determination module is specifically used for: The first driving information of each second mining car is encoded by a preset encoder to obtain the first feature of each second mining car. The reference line of the first mining car is encoded by the preset encoder to obtain the second feature of the first mining car. The first feature and the second feature have the same dimension. The first and second features of each second mine car are merged to obtain the merged feature; The initial random Gaussian noise and the first diffusion time step are obtained. The initial random Gaussian noise is denoised based on the fusion features and the first diffusion time step to obtain the second driving information of the first mining truck.

[0114] Based on the above embodiments, as an optional embodiment, the second driving information determination module is further configured to: The fused features, initial random Gaussian noise, and first diffusion time step are input into the pre-trained driving information prediction model of the first mining truck to obtain the initial driving information of the first mining truck. The driving information prediction model for each first mining truck is trained in the following way: Multiple training samples and corresponding training labels are obtained. Each training sample contains the fusion features of the first mining truck, the initial noisy driving information, and the third diffusion time step. The training label is the actual driving information of the first mining truck. The initial noisy driving information in each training sample is obtained by adding noise to the corresponding training label several times in the third diffusion time step. The first mining truck did not collide with any of the second mining trucks when driving according to the actual driving information. For each training sample, the training sample is input into the initial model, and the second driving information corresponding to the training sample is output. Based on the second driving information, the third diffusion time step, and the training label corresponding to each training sample, the training loss of the initial model is determined, and the model parameters of the initial model are adjusted based on the training loss until the preset training termination condition is met, thus obtaining the driving information prediction model.

[0115] Based on the above embodiments, as an optional embodiment, the second driving information determination module can also be used for: Based on the fusion features and the first diffusion time step, the initial random Gaussian noise is iterated for a preset number of rounds. The denoised Gaussian noise output in the last iteration is used as the second driving information. The preset number of rounds is the first diffusion time step. Each iteration includes: Obtain the denoised Gaussian noise for this iteration, where the denoised Gaussian noise for the first iteration is the initial random Gaussian noise; Based on the second diffusion time step, fusion features, and denoised Gaussian noise of this iteration, prediction is made to obtain the predicted noise of this iteration. The predicted noise is the part to be removed from the denoised Gaussian noise of this iteration. Based on the predicted noise of this iteration, the denoised Gaussian noise of this iteration is denoised to obtain the denoised Gaussian noise of the next iteration.

[0116] Based on the above embodiments, as an optional embodiment, the driving information prediction model includes multiple cascaded layer structures, each of which contains hidden features for supplementing lost feature information. The second driving information determination module can also be used for: For each layer, if it is the first layer, the hidden features of the layer are used to supplement the fused features to obtain the third feature of the layer. If it is not the first layer, the features of the non-overlapping part between the denoised Gaussian noise of the current iteration and the denoised Gaussian noise of the next iteration are predicted based on the second diffusion time step and the third feature of the previous layer. These features are then used as the third feature. The third features of each layer are fused to obtain the fourth feature. The predicted noise is then obtained based on the fourth feature. The third feature of the first layer structure is determined as follows: based on the second diffusion time step and the denoised Gaussian noise of the current iteration, the features of the non-overlapping part between the denoised Gaussian noise of the current iteration and the denoised Gaussian noise of the next iteration are predicted and used as the third feature.

[0117] Based on the above embodiments, as an optional embodiment, the device further includes a feature fusion module, specifically used for: The first features of each second minecart are merged to obtain the second feature; The second feature is fused with the first feature of the first mine car to obtain the fused feature.

[0118] Based on the above embodiments, as an optional embodiment, the device further includes a driving information optimization module, specifically used for: For each first mine car, the second driving information of the first mine car is divided into multiple driving information segments. The average value of the driving direction angle of each point in each driving information segment is determined, and the distance between the first point and the last point in each driving information segment is obtained. For each segment of driving information, the average speed value of the segment is determined based on distance and average value.

[0119] The following is for reference. Figure 5 It illustrates an electronic device suitable for implementing embodiments of this application (e.g., performing...). Figure 1 The diagram shows the structure of the terminal device or server 500 of the method shown. The electronic devices in the embodiments of this application may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (e.g., vehicle navigation terminals), wearable devices, and fixed terminals such as digital TVs and desktop computers. Figure 5 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0120] The electronic device includes a memory and a processor. The memory stores a program for executing the methods described in the various method embodiments above. The processor is configured to execute the program stored in the memory. The processor may be referred to as processing device 501 as described below. The memory may include at least one of read-only memory (ROM) 502, random access memory (RAM) 503, and storage device 508 as described below, as follows: like Figure 5 As shown, the electronic device 500 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0121] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0122] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 509, or installed from storage device 508, or installed from ROM 502. When the computer program is executed by processing device 501, it performs the functions defined in the methods of embodiments of this application.

[0123] It should be noted that the computer-readable storage medium described above in this application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0124] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0125] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0126] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: The initial and target positions of each mining car within the mining area are obtained. For each first mining car, a reference line is constructed based on its initial and target positions. The reference line represents the path taken by the first mining car from its initial position to its target position within the mining area. The first mining car can be any mining car in the mining area. For each first mining car, its first driving information is obtained through its own kinematic model and its own reference line. For each first mining car, its first driving information is updated based on the first driving information of each second mining car and the reference line of the mining car. The updated first driving information is used as the second driving information of the first mining car. Both the first and second driving information include the reference line of the corresponding mining car and the motion state at multiple points on the reference line. The second driving information of each mining car does not have the same points at the same time. The second mining cars are any mining cars in the mining area other than the first mining car.

[0127] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0129] The modules or units described in the embodiments of this application can be implemented in software or hardware. The names of modules or units do not necessarily limit the specific unit; for example, a first constraint acquisition module can also be described as a "module for acquiring the first constraint".

[0130] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0131] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0132] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0133] The above description is only a partial embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A driving information planning method, characterized in that, include: Obtain the initial and target positions of each mining vehicle within the mining area; For each first mining car, a reference line for the mining car is constructed based on the initial position and the target position of the first mining car; The reference line is used to characterize the path taken by the first mining car in the mining area from the initial position to the target position; the first mining car can be any mining car in the mining area; For each first mining car, the first driving information of the first mining car is obtained through its own kinematic model and its own reference line; For each first mining car, the first driving information of the first mining car is updated based on the first driving information of each second mining car and the reference line of the mining car, and the updated first driving information is used as the second driving information of the first mining car. Both the first driving information and the second driving information include a reference line for the corresponding mine car and the movement status at multiple points on the reference line; the second driving information of each mine car does not have the same point at the same time; the second mine car is any other mine car in the mining area besides the first mine car.

2. The method according to claim 1, characterized in that, The step of updating the first driving information of the first mine car based on the first driving information of each second mine car and the reference line of the mine car, and using the updated first driving information as the second driving information of the first mine car, includes: The first driving information of each second mining car is encoded by a preset encoder to obtain the first feature of each second mining car. The reference line of the first mining car is encoded by the preset encoder to obtain the second feature of the first mining car. The first feature and the second feature have the same dimension. The first and second features of each second mine car are merged to obtain the merged feature; The initial random Gaussian noise and the first diffusion time step are obtained. Based on the fusion feature and the first diffusion time step, the initial random Gaussian noise is denoised to obtain the second driving information of the first mining truck.

3. The method according to claim 2, characterized in that, The step of denoising the initial random Gaussian noise based on the fusion features and the first diffusion time step to obtain the second driving information of the first mining truck includes: The fusion features, the initial random Gaussian noise, and the first diffusion time step are input into the pre-trained driving information prediction model of the first mining truck to obtain the initial driving information of the first mining truck. The driving information prediction model for each first mining truck is trained in the following way: Multiple training samples and corresponding training labels are obtained. Each training sample contains the fusion features of the first mining truck, initial noisy driving information, and the third diffusion time step. The training label is the actual driving information of the first mining truck. The initial noisy driving information in each training sample is obtained by adding noise to the corresponding training label multiple times according to the third diffusion time step. The first mining truck does not collide with any of the second mining trucks when driving according to the actual driving information. For each training sample, the training sample is input into the initial model, and the second driving information corresponding to the training sample is output. Based on the second driving information corresponding to each training sample, the third diffusion time step, and the training label, the training loss of the initial model is determined, and the model parameters of the initial model are adjusted based on the training loss until the preset training termination condition is met, thereby obtaining the driving information prediction model.

4. The method according to claim 3, characterized in that, The step of inputting the fused features, the initial random Gaussian noise, and the first diffusion time step into the pre-trained driving information prediction model of the first mining truck to obtain the initial driving information of the first mining truck includes: Based on the fusion features and the first diffusion time step, the initial random Gaussian noise is iterated for a preset number of rounds, and the denoised Gaussian noise output in the last iteration is used as the second driving information, wherein the preset number of rounds is the first diffusion time step; Each iteration includes: Obtain the denoised Gaussian noise for this iteration, wherein the denoised Gaussian noise for the first iteration is the initial random Gaussian noise; Based on the second diffusion time step of this iteration, the fusion feature, and the denoised Gaussian noise of this iteration, a prediction noise for this iteration is obtained, wherein the prediction noise is the part to be removed in the denoised Gaussian noise of this iteration; Based on the predicted noise of this iteration, the denoised Gaussian noise of this iteration is denoised to obtain the denoised Gaussian noise of the next iteration.

5. The method according to claim 4, characterized in that, The driving information prediction model includes multiple cascaded layer structures, and each layer structure contains hidden features used to supplement lost feature information. The prediction noise for this iteration is obtained by predicting based on the second diffusion time step of this iteration, the fusion feature, and the denoised Gaussian noise of this iteration, including: For each layer, if the layer is the first layer, the fused features are supplemented by the hidden features of the layer to obtain the third feature of the layer; if the layer is not the first layer, the features of the non-overlapping part between the denoised Gaussian noise of the current iteration and the denoised Gaussian noise of the next iteration are predicted based on the second diffusion time step and the third feature of the previous layer, and used as the third feature. The third features of each layer are fused to obtain the fourth feature, and the predicted noise is obtained based on the fourth feature. The third feature of the first layer structure is determined by predicting the non-overlapping features between the denoised Gaussian noise of the current iteration and the denoised Gaussian noise of the next iteration based on the second diffusion time step and the denoised Gaussian noise of the current iteration, as the third feature.

6. The method according to claim 2, characterized in that, The process of fusing the first and second features of each second mine car to obtain a fused feature includes: The first features of each second minecart are merged to obtain the second feature; The second feature is fused with the first feature of the first mining vehicle to obtain the fused feature.

7. The method according to claim 1, characterized in that, The second driving information includes the driving direction of the first mine car when it reaches each point; the driving direction of each point is determined according to the reference line; After obtaining the second driving information of the first mining truck, the process further includes a step of optimizing the second driving information: For each first mining car, the second driving information of the first mining car is divided into multiple driving information segments. The average value of the driving direction angle of each point in each driving information segment is determined, and the distance between the first point and the last point in each driving information segment is obtained. For each travel information segment, the average speed value of the travel information segment is determined based on the distance and the average value.

8. A driving information planning device, characterized in that, include: The location acquisition module is used to obtain the initial and target locations of each mining truck within the mining area; A reference line construction module is used to construct a reference line for each first mining car based on its initial position and target position. The reference line is used to characterize the path taken by the first mining car in the mining area from the initial position to the target position; the first mining car can be any mining car in the mining area; The first driving information determination module is used to obtain the first driving information of each first mining car by using its own kinematic model and its own reference line. The second driving information determination module is used to update the first driving information of each first mining car based on the first driving information of each second mining car and the reference line of the mining car, and use the updated first driving information as the second driving information of the first mining car. Both the first driving information and the second driving information include a reference line for the corresponding mine car and the movement status at multiple points on the reference line; the second driving information of each mine car does not have the same point at the same time; the second mine car is any other mine car in the mining area besides the first mine car.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the method of any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-7.

Images