Moon polar region sun synchronization A* path planning method and system, electronic equipment and storage medium

Abstract

The invention provides a lunar polar region sun synchronization # imgabs0 # path planning method, which can be applied to the technical field of celestial body image processing and path planning. The method comprises the following steps: independently training a plurality of enhanced U-Net models by using training data with time sequence slices and fusion types and a mixed loss function; dividing and extracting the reasoning data with a time sequence slice type and a time sequence fusion type by using a plurality of trained enhanced U-Net models to obtain a lunar polar region friendly illumination region, and preprocessing the reasoning data with the lunar polar region friendly illumination region and gradient data corresponding to the lunar polar region friendly illumination region; and based on the reasoning data after time alignment and the gradient data after space alignment, constructing a heuristic function by using mixed cost, and performing heuristic path planning by using an improved # imgabs1 # path planning algorithm with a time one-dimensional function and an in-situ waiting function to obtain a sun synchronization planning path of the lunar polar region.

Description

Technical Field

[0001] The present invention relates to the technical field of celestial body image processing and path planning, and specifically relates to a sun-synchronous path planning method and system, an electronic device, and a storage medium for the lunar polar region. Background Art

[0002] The water ice and volatiles enriched in the lunar south pole provide important ways for scientific research and in-situ resource utilization, and have become the key areas of many international lunar exploration programs, including the Artemis program of the United States, the Chang'e-7 (CE-7) mission of China, as well as the future exploration missions such as the joint Japan-India lunar polar exploration mission and the International Lunar Research Station led by China. CE-7 is expected to be launched in 2026, and will use the rover to conduct research in geology and geophysics in the illuminated area, and detect the water ice and volatiles in the permanently shadowed area through the flyer. Due to the complex terrain occlusion and illumination conditions in the lunar polar region, it is urgent to carry out sun-synchronous path planning during the movement of the rover to maximize the use of solar energy to extend the detection mission time, while avoiding obstacles and efficiently reaching the target point. Therefore, carrying out path planning research based on the illumination-friendly area and combining slope safety constraints is of great significance for supporting the mission implementation and improving the detection efficiency.

[0003] At present, related research mainly relies on low-resolution data and static environmental factors for path planning. Therefore, it is difficult to directly process the high-spatial-resolution temporal illumination data in the polar region statically, and there is a lack of efficient and accurate methods and systems for carrying out sun-synchronous path planning of the rover based on dynamic-friendly temporal illumination conditions. Summary of the Invention

[0004] In view of the above problems, the present invention provides a sun-synchronous path planning method and system, an electronic device, and a storage medium for the lunar polar region, which are used to solve at least one of the above technical problems.

[0005] According to the first aspect of the present invention, there is provided a sun-synchronous path planning method for the lunar polar region, including:

[0006] Preprocessing the original temporal illumination data to obtain training data with temporal slice type and temporal fusion type, and inference data with temporal slice type and temporal fusion type;

[0007] Independently training multiple enhanced U-Net models using the training data with temporal slice type and temporal fusion type and a hybrid loss function to obtain multiple trained enhanced U-Net models;

[0008] Use multiple trained enhanced U-Net models to segment and extract inference data with temporal slice types and temporal fusion types to obtain the friendly illumination area in the lunar polar region, and based on the preset path planning requirements, align the inference data with the friendly illumination area in the lunar polar region in time and align the slope data corresponding to the friendly illumination area in the lunar polar region in space;

[0009] Based on the inference data after time alignment and the slope data after space alignment, use a hybrid cost based on illumination, slope, and Euclidean distance to construct a heuristic function, and an improved path planning algorithm with both one-dimensional time function and the function of waiting in place Perform heuristic path planning to obtain the sun-synchronous planning path in the lunar polar region.

[0010] According to the embodiments of the present invention, the above preprocessing of the temporal illumination raw data to obtain multi-type training data and multi-type inference data includes:

[0011] Automatically parse and extract the temporal illumination raw data to obtain temporal illumination slice training data and temporal illumination slice inference data;

[0012] Based on variable time intervals, respectively determine the grouping and fusion units of the temporal illumination slice training data and the temporal illumination slice inference data;

[0013] Use the grouping and fusion unit to label and group the temporal illumination slice training data to obtain temporal illumination fusion training data;

[0014] Use the grouping and fusion unit to label and group the temporal illumination slice inference data to obtain temporal illumination fusion inference data.

[0015] According to the embodiments of the present invention, the above preprocessing of the temporal illumination raw data to obtain multi-type training data and multi-type inference data further includes:

[0016] Perform automatic label definition operations based on one-hot encoding on the temporal illumination slice training data and the temporal illumination fusion training data respectively. Among them, the one-hot encoding of the temporal fusion illumination fusion data is obtained by key frame operation, and the key frame is to automatically obtain the intermediate frame mask data of each group of fusion units;

[0017] Perform automatic grouping operations on the temporal illumination slice training data with label information to obtain a temporal illumination slice training set, a temporal illumination slice validation set, and a temporal illumination slice test set;

[0018] Perform automatic grouping operations on the temporal illumination fusion training data with label information to obtain a temporal illumination fusion training set, a temporal illumination fusion validation set, and a temporal illumination fusion test set.

[0019] According to an embodiment of the present invention, the above-mentioned independent training of multiple enhanced U-Net models using training data with temporal slice types and temporal fusion types and a hybrid loss function to obtain multiple trained enhanced U-Net models includes:

[0020] Based on a temporal illumination slice training set, a temporal illumination slice validation set, and a temporal illumination slice test set, using a hybrid loss function based on cross-entropy loss and Dice loss to perform cross-iterative training, validation, and testing on the temporal slice enhanced U-Net model, obtaining a trained temporal slice enhanced U-Net model;

[0021] Based on a temporal illumination fusion training set, a temporal illumination fusion validation set, and a temporal illumination fusion test set, using a hybrid loss function based on cross-entropy loss and Dice loss to perform cross-iterative training, validation, and testing on the temporal fusion enhanced U-Net model, obtaining a trained temporal fusion enhanced U-Net model.

[0022] According to an embodiment of the present invention, the above-mentioned temporal slice enhanced U-Net model includes a residual attention module, and the residual attention module includes a multi-layer perceptron, a channel attention layer, a spatial attention layer, and a temporal slice residual connection layer;

[0023] Among them, the temporal fusion enhanced U-Net model includes a multi-channel fusion input module, a residual convolution module, and a spatio-temporal fusion module. The residual convolution module includes a temporal fusion attention sub-module and a temporal residual connection layer, and the spatio-temporal fusion module includes a spatial convolution sub-module and a temporal convolution sub-module.

[0024] According to an embodiment of the present invention, the above-mentioned use of multiple trained enhanced U-Net models to perform segmentation and extraction on inference data with temporal slice types and temporal fusion types to obtain a lunar polar region friendly illumination area includes:

[0025] Using the trained temporal slice enhanced U-Net model to perform feature extraction operations and region segmentation operations on the temporal illumination slice inference data, obtaining temporal illumination slice inference data with a lunar polar region friendly illumination area;

[0026] Using the trained temporal fusion enhanced U-Net model to perform feature extraction operations and segmentation operations based on grouped fusion units on the temporal illumination fusion data, obtaining temporal illumination slice inference data of the lunar polar region friendly illumination area.

[0027] According to an embodiment of the present invention, the above-mentioned inference data after time alignment and the slope data after space alignment are constructed by using a mixed cost based on illumination, slope and Euclidean distance, and at the same time have the improvement of time one-dimensional function and the function of waiting in place The path planning algorithm performs heuristic path planning and obtains the sun-synchronous planning path in the lunar polar region, including:

[0028] Initialize the open set with the preset starting point, preset the closed set to the empty set, set the initial state of the heuristic function to the estimated cost of illumination, slope and Euclidean distance of the preset starting point and the actual cost of the preset starting point, and set the preset starting point to the current node;

[0029] If the current node is the preset end point, put the current node into the closed set, terminate the node search and path planning, and return the closed set as the planning path result;

[0030] If the current node is illuminated, put the current node into the closed set;

[0031] When the current node has no light, select the node with the smallest light, slope, and Euclidean distance cost from the neighboring nodes around the current node as the current node and put it into the closed set;

[0032] Before searching for a pre-selected node, the open set is set to empty to ensure the time one-dimensional function of path planning, wherein the search for the pre-selected node is to traverse the neighboring nodes around the current node;

[0033] Calculate the heuristic function value of the neighboring nodes, and put the neighboring nodes and the heuristic function values of the neighboring nodes into the open set, where the heuristic function value includes the mixed estimated cost consisting of the illumination cost of the neighboring nodes, the slope cost, the Euclidean distance cost with the preset end point, and the actual cost consisting of the Euclidean distance cost from the current node to the neighboring nodes and the accumulated actual cost of the planned path nodes;

[0034] After searching for the pre-selected nodes, the node with the smallest heuristic function value is selected from the open set as the pre-selected node for the next path planning;

[0035] After searching for the pre-selected nodes, the lighting conditions at the current moment are replaced by the lighting conditions at the next moment of the time-aligned inference data;

[0036] After searching for the pre-selected node, if the open set is an empty set, the path planning is set to a waiting state, that is, the current node is continued to be selected as the current node at the next moment, the path search operation at the current moment is stopped, and the path search at the next moment is entered;

[0037] After searching for preselected nodes, when the open set is not an empty set, the preselected nodes are used as the current nodes at the next moment, and the path search operation at the next moment is continued.

[0038] The second aspect of the present invention provides a lunar polar sun-synchronous path planning system, including:

[0039] A data preprocessing module for preprocessing the original time-series illumination data to obtain training data with time-series slice types and time-series fusion types and inference data with time-series slice types and time-series fusion types;

[0040] An enhanced U-Net model training module for independently training multiple enhanced U-Net models using the training data with time-series slice types and time-series fusion types and a mixed loss function to obtain multiple trained enhanced U-Net models;

[0041] A model inference and preprocessing module for segmenting and extracting the inference data with time-series slice types and time-series fusion types using the multiple trained enhanced U-Net models to obtain a lunar polar friendly illumination area, and based on the preset path planning requirements, performing temporal alignment on the inference data with the lunar polar friendly illumination area and spatial alignment on the slope data corresponding to the lunar polar friendly illumination area;

[0042] A path planning module for constructing a heuristic function using a mixed cost based on illumination, slope, and Euclidean distance based on the temporally aligned inference data and spatially aligned slope data, and performing heuristic path planning using an improved path planning algorithm with one-dimensional time function and in-place waiting function to obtain a sun-synchronous planning path for the lunar polar region.

[0043] The third aspect of the present invention provides an electronic device, including: one or more processors; a memory for storing one or more computer programs, wherein the above one or more processors execute the above one or more computer programs to implement the steps of the above method.

[0044] The fourth aspect of the present invention further provides a computer-readable storage medium having a computer program or instruction stored thereon, and the above computer program or instruction implements the steps of the above method when executed by a processor.

[0045] The above lunar polar sun-synchronous provided by the present invention Path planning method and system, which use training data with temporal slice types and temporal fusion types to train multiple enhanced U-Net models, improving the ability of the multiple enhanced U-Net models to extract details of temporal illumination data; by increasing the slope and illumination condition terms added in the heuristic function, it is possible to ensure the selection of appropriate slopes and illumination paths during the path planning process while maintaining the original The algorithm mainly uses the Euclidean distance for the best path selection feature, effectively reducing the complexity of the algorithm, so that the lunar polar sun-synchronous path planning method and system provided by the present invention can effectively support the research related to lunar polar sun-synchronous path planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Through the following description of the embodiments of the present invention with reference to the accompanying drawings, the above content and other objects, features and advantages of the present invention will become clearer. In the drawings:

[0047] Figure 1 is the application scenario diagram of the lunar polar sun-synchronous path planning method according to an embodiment of the present invention;

[0048] Figure 2 is the flowchart of the lunar polar sun-synchronous path planning method according to an embodiment of the present invention;

[0049] Figure 3 is the structural diagram of the lunar polar sun-synchronous path planning system according to an embodiment of the present invention;

[0050] Figure 4 is the block diagram of an electronic device suitable for implementing the lunar polar sun-synchronous path planning method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary and not intended to limit the scope of the present invention. In the following detailed description, for the purpose of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present invention. However, obviously, one or more embodiments can be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present invention.

[0052] The terms used herein are for describing specific embodiments only and are not intended to limit the present invention. The terms such as "including" and "comprising" used herein indicate the presence of the described features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0053] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0054] In the case of using expressions such as "at least one of A, B, and C", generally, it should be interpreted according to the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include, but is not limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.).

[0055] Figure 1 is the application scenario diagram of the lunar polar sun-synchronous path planning method according to an embodiment of the present invention.

[0056] As Figure 1 shown, the application scenario 100 according to this embodiment may include celestial image processing and path planning. The network 104 is used to provide a medium for communication links between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0057] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0058] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop portable computers, and desktop computers, etc.

[0059] Server 105 may be a server that provides various services. For example, it may be a background management server (for example only) that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server can analyze and process data such as user requests received, and feedback the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0060] It should be noted that the lunar polar sun-synchronous path planning method provided by the embodiments of the present invention can generally be executed by server 105. Correspondingly, the lunar polar sun-synchronous path planning system provided by the embodiments of the present invention can generally be set in server 105. The lunar polar sun-synchronous path planning method can also be executed by a server or a server cluster different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the lunar polar sun-synchronous path planning system can also be set in a server or a server cluster different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0061] It should be understood that Figure 1 the numbers of terminal devices, networks, and servers in

[0062] are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers.

[0062] Hereinafter, based on the Figure 1 scenario described below, the lunar polar sun-synchronous Figure 2 path planning method of the disclosed embodiments will be described in detail through ...

[0063] Figure 2 is a flowchart of the lunar polar sun-synchronous path planning method according to the embodiments of the present invention.

[0064] As Figure 2 shown, the above-mentioned lunar polar sun-synchronous path planning includes operations S210 to S240.

[0065] In operation S210, preprocess the temporal illumination raw data to obtain training data with temporal slice type and temporal fusion type, and inference data with temporal slice type and temporal fusion type.

[0066] Automatically parse and extract the original time-series illumination data to obtain time-series illumination slice data; fuse the time-series illumination slice data based on a variable time interval to obtain time-series illumination fusion data. The present invention processes subsequent models based on multi-type time-series illumination data, incorporates the time dimension into illumination modeling, and can accurately capture the complex periodic illumination changes in the lunar polar region.

[0067] In operation S220, independently train multiple enhanced U-Net models using training data with time-series slice types and time-series fusion types and a hybrid loss function to obtain multiple trained enhanced U-Net models.

[0068] The above multiple enhanced U-Net models include a time-series slice enhanced U-Net model and a time-series fusion enhanced U-Net model, which are respectively used to process time-series slice data (including training data and inference data) and time-series fusion data (including training data and inference data).

[0069] Using multiple enhanced U-Net models can effectively identify microscopic terrain details (e.g., craters / rocks) and macroscopic illumination trends.

[0070] The design of the hybrid loss function (e.g., cross-entropy loss function and / or Dice Loss, and those skilled in the art can choose other suitable loss functions according to actual needs) improves the recognition accuracy of multiple enhanced U-Net models for sparse-friendly regions and reduces misjudgment in the low-contrast environment of the lunar polar region.

[0071] In operation S230, use the multiple trained enhanced U-Net models to segment and extract inference data with time-series slice types and time-series fusion types to obtain a lunar polar region-friendly illumination area, and based on the preset path planning requirements, align the inference data with the lunar polar region-friendly illumination area in time and align the slope data corresponding to the lunar polar region-friendly illumination area in space.

[0072] In operation S240, based on the time-aligned inference data and the space-aligned slope data, construct a heuristic function using a hybrid cost based on illumination, slope, and Euclidean distance, and an improved path planning algorithm with both one-dimensional time function and in-situ waiting function Perform heuristic path planning to obtain a sun-synchronous planned path for the lunar polar region.

[0073] The above third-order cost function innovatively improves the existing Path planning algorithm, where the illumination cost can maximize the proportion of solar illumination time, the slope cost can minimize the motion energy consumption, and the Euclidean cost can optimize the total path length.

[0074] The one-dimensionality of time extends the existing path planning algorithm, and the in-situ waiting function enables the improved algorithm provided by the present invention to achieve shadow avoidance (e.g., waiting until the next light window) and energy replenishment (e.g., extending the residence time in the illuminated area).

[0075] The above-mentioned lunar polar sun-synchronous path planning method and system provided by the present invention use multi-type training data with temporal slice types and temporal fusion types to train multiple enhanced U-Net models, improving the ability of the multiple enhanced U-Net models to extract details of temporal illumination data; by increasing the slope and illumination condition terms added in the heuristic function, it enables the path planning process to maintain the original characteristics that the algorithm mainly uses Euclidean distance for optimal path selection, effectively reducing the complexity of the algorithm, making the above-mentioned lunar polar sun-synchronous path planning method and system provided by the present invention can effectively support research related to lunar polar sun-synchronous path planning.

[0076] According to an embodiment of the present invention, the preprocessing of the original temporal illumination data to obtain multi-type training data and multi-type inference data includes: automatically parsing and extracting the original temporal illumination data to obtain temporal illumination slice training data and temporal illumination slice inference data; respectively determining the grouped fusion units of the temporal illumination slice training data and the temporal illumination slice inference data based on variable time intervals; using the grouped fusion units to perform labeled grouping on the temporal illumination slice training data to obtain temporal illumination fusion training data; using the grouped fusion units to perform labeled grouping on the temporal illumination slice inference data to obtain temporal illumination fusion inference data.

[0077] The above preprocessing process of the original temporal illumination data realizes dynamic period feature capture, that is, through the grouped fusion unit with variable time intervals, it can adapt to the unique long-period (such as lunar phase change) and short-period (such as instantaneous shadow caused by terrain occlusion) illumination patterns in the lunar polar region. This dynamic grouping mechanism effectively solves the problem that it is difficult for traditional fixed time windows to take into account both circadian rhythms and sudden illumination interruptions, ensuring that the training data contains complete periodic illumination characteristics.

[0078] According to an embodiment of the present invention, the above preprocessing of the temporal illumination raw data to obtain multi-type training data and multi-type inference data further includes: performing label automatic definition operations based on one-hot encoding on the temporal illumination slice training data and the temporal illumination fusion training data respectively. Among them, the one-hot encoding of the temporal fusion illumination fusion data is obtained by taking key frame operations, and the key frame is to automatically obtain the intermediate frame mask data of each group of fusion units; performing an automatic grouping operation on the temporal illumination slice training data with label information to obtain a temporal illumination slice training set, a temporal illumination slice validation set, and a temporal illumination slice test set; performing an automatic grouping operation on the temporal illumination fusion training data with label information to obtain a temporal illumination fusion training set, a temporal illumination fusion validation set, and a temporal illumination fusion test set.

[0079] The one-hot encoding provided by the present invention uses a single grid illumination visibility coefficient greater than 75% as the true mask one-hot encoding threshold, as shown in formula (1):

[0080] (1),

[0081] Wherein, represents the mask of pixel , represents the sun visibility coefficient of pixel , and the value range is [0, 1].

[0082] The data automatic grouping is as shown in formula (2):

[0083] (2),

[0084] Wherein, represents the grouping set; is the grouping coefficient, and the values are 0.7, 0.2, and 0.1, indicating that the data is grouped and cut according to the serial number order in the proportions of 70%, 20%, and 10% to construct a training set, a validation set, and a test set and output them to the specified storage space; is the test data set for model training and validation.

[0085] The above embodiment can achieve enhanced dynamic feature capture, that is, by automatically extracting the intermediate frame mask data through key frames, effectively capturing the peak features of illumination changes. This dynamic sampling mechanism based on fusion units can not only retain the periodicity of the original temporal data (such as the law of day and night alternation), but also highlight the significant features of sudden illumination events (such as instantaneous shadow occlusion).

[0086] Meanwhile, a dual-path one-hot encoding strategy is adopted: the sliced data encoding preserves the original time resolution; the fused data encoding aggregates spatio-temporal correlations through key frames; this design enables the model to learn short-term local features (such as terrain detail reflections) and long-term global patterns (such as the change of solar altitude angle in the lunar polar region) simultaneously.

[0087] In addition, it is possible to improve the training stability, that is, the automatic grouping algorithm implements stratified sampling within the sliced and fused data sets to ensure time continuity (to avoid splitting data in the same continuous time period into different sets) and spatial coverage integrity (to ensure that all terrain feature types are distributed in the training / validation / test sets).

[0088] According to an embodiment of the present invention, the above-mentioned training of multiple enhanced U-Net models independently using training data with temporal slice types and temporal fusion types and a hybrid loss function includes: based on a temporal illumination slice training set, a temporal illumination slice validation set, and a temporal illumination slice test set, using a hybrid loss function based on cross-entropy loss and Dice loss to cross-iteratively train, validate, and test the temporal slice enhanced U-Net model to obtain a trained temporal slice enhanced U-Net model; based on a temporal illumination fusion training set, a temporal illumination fusion validation set, and a temporal illumination fusion test set, using a hybrid loss function based on cross-entropy loss and Dice loss to cross-iteratively train, validate, and test the temporal fusion enhanced U-Net model to obtain a trained temporal fusion enhanced U-Net model.

[0089] According to an embodiment of the present invention, the above-mentioned temporal slice enhanced U-Net model includes a residual attention module, and the residual attention module includes a multi-layer perceptron, a channel attention layer, a spatial attention layer, and a temporal slice residual connection layer; among them, the temporal fusion enhanced U-Net model includes a multi-channel fusion input module, a residual convolution module, and a spatio-temporal fusion module. The residual convolution module includes a temporal fusion attention sub-module and a temporal residual connection layer, and the spatio-temporal fusion module includes a spatial convolution sub-module and a temporal convolution sub-module.

[0090] The temporal slice enhanced U-Net model replaces the convolutional module for extracting each layer's feature map with a residual attention module to enhance the extraction of data features. The residual attention module is composed of regularization, a channel attention module, a spatial attention module, and a residual connection. Its core architecture is shown in formula (3):

[0091] (3),

[0092] Among them, is the data of the temporal slice type at time is the number of layers, represents the channel attention value, represents the spatial attention value, is the regularized feature map data, while represents the residual value of the input data, is the activation function, represents the residual attention module operation, represents the downsampling operation while represents the upsampling operation.

[0093] The temporal fusion enhanced U-Net model replaces the convolutional module for extracting the feature map of each original layer with a residual convolutional module to enhance the extraction of data features. The residual convolutional module is composed of regularization, a temporal attention module, and a residual connection. A spatio-temporal fusion module is added between the encoder and the decoder, and its core architecture is shown in Equation (4):

[0094] (4),

[0095] where, is a time series data composed of time slice data of grouped fusion units, that is, temporal fusion type data, represents the time series attention value, is the spatio-temporal fusion processing, and other symbols are consistent with the time series slice data.

[0096] The training of the above-mentioned multiple enhanced U-Net models uses a loss function that combines cross-entropy and Dice and establishes backpropagation to implement the training and update of model parameters. The expression of this combined loss function is shown in Equation (5):

[0097] (5),

[0098] where, represents the total loss function, is the cross-entropy loss function, is the Dice loss function, is a hyperparameter for balancing the weights of the two different loss functions, represents the pixel at the true mask for classification, the pixel at the predicted result for classification, is the number of pixels, is a small constant to avoid division by zero in the denominator.

[0099] According to an embodiment of the present invention, the above-mentioned segmentation and extraction of inference data with temporal slice type and temporal fusion type using multiple trained enhanced U-Net models to obtain the lunar polar friendly illumination area includes: using the trained temporal slice enhanced U-Net model to perform feature extraction operations and region segmentation operations on the temporal illumination slice inference data to obtain the temporal illumination slice inference data with the lunar polar friendly illumination area; using the trained temporal fusion enhanced U-Net model to perform feature extraction operations and segmentation operations based on the grouped fusion unit on the temporal illumination fusion data to obtain the temporal illumination slice inference data of the lunar polar friendly illumination area.

[0100] According to an embodiment of the present invention, the above-mentioned improvement that constructs a heuristic function using a mixed cost based on illumination, slope, and Euclidean distance for the inference data after time alignment and the slope data after space alignment, and has both one-dimensional time function and in-place waiting function The path planning algorithm performs heuristic path planning to obtain a sun-synchronous planned path in the lunar polar region, including: initializing the open set with a preset starting point, presetting the closed set as an empty set, setting the initial state of the heuristic function as the estimated cost of the illumination, slope, and Euclidean distance of the preset starting point and the actual cost of the preset starting point, and setting the preset starting point as the current node; in the case where the current node is the preset end point, putting the current node into the closed set, terminating the node search and path planning, and returning the closed set as the result of the planned path; in the case where the current node has illumination, putting the current node into the closed set; in the case where the current node has no illumination, selecting the node with the minimum illumination, slope, and Euclidean distance cost from the surrounding neighboring nodes of the current node as the current node and putting it into the closed set; emptying the open set before searching for the preselected nodes to ensure the time one-dimensional function of the path planning, where searching for the preselected nodes is to traverse the neighboring nodes around the current node; calculating the heuristic function value of the neighboring nodes, and putting the neighboring nodes and their heuristic function values into the open set, where the heuristic function value includes the mixed estimated cost composed of the illumination cost, slope cost, and Euclidean distance cost to the preset end point of the neighboring node and the actual cost composed of the Euclidean distance cost from the current node to the neighboring node and the cumulative actual cost of the planned path nodes; after searching for the preselected nodes, selecting the node with the minimum heuristic function value from the open set as the preselected node for the next step of path planning; after searching for the preselected nodes, replacing the illumination condition at the current moment with the illumination condition at the next moment of the time-aligned inference data; after searching for the preselected nodes, in the case where the open set is an empty set, setting the path planning to a waiting state, that is, continuing to select the current node as the current node at the next moment, stopping the path search operation at the current moment and entering the path search at the next moment; after searching for the preselected nodes, in the case where the open set is not an empty set, taking the preselected node as the current node at the next moment and continuing the path search operation at the next moment.

[0101] The improved sun synchronization provided by the present invention When the path planning algorithm is initialized, the open set OpenSet is the starting point coordinates, the closed set ClosedSet is empty. Before selecting the next optimal neighboring node, first empty OpenSet to ensure that the planned nodes in the previous moment are fixed and no longer participate in the path planning, and perform node search according to formula (6):

[0102] (6),

[0103] where, is expressed as the neighboring node, is the current node, is At moment or the illumination data in the time period. is the total cost of neighboring nodes, is the actual cost from the starting point to the neighboring node, is the heuristic cost from the neighboring node to the target point; is obtained by adding the actual cost of the current point to the Euclidean distance from the current point to the neighboring node; is obtained by adding the Euclidean distance from the neighboring node to the target point to the slope heuristic value of the neighboring node plus the illumination value of the neighboring node ; The value is as shown in formula (7):

[0104] (7),

[0105] wherein, is the slope value of, is positive infinity; The value is as shown in formula (8):

[0106] (8).

[0107] The improved algorithm provided by the present invention will be further described in detail below through specific embodiments.

[0108] The improved The path planning algorithm first initializes the open set with the preset starting point, presets the closed set to the empty set, sets the initial state of the heuristic function to the estimated cost consisting of the Euclidean distance cost from the preset starting point to the preset end point, the slope cost of the preset starting point, and the illumination cost of the preset starting point plus the actual cost set to 0, and sets the preset starting point to the current node, then starts the node search operation and the path planning traversal operation, where the slope cost means that when the slope of the current node is less than 20 degrees, the slope cost is 0, otherwise it is infinite, and the illumination cost means that when the illumination condition of the current node at the current moment is 0, it is infinite, otherwise it is 0; when the current node is the preset end point, In this case, the current node is placed in a closed set, the node search operation and the path planning operation are terminated, and the closed set is returned as the planning path result; based on the current moment temporal illumination slice data or temporal illumination fusion data, if the current node is illuminated, the current node is placed in a closed set, and based on the slope data, the current moment temporal slice illumination data or temporal fusion data, if the current node is not illuminated, a node with the smallest illumination, slope, and Euclidean distance cost is selected from the surrounding neighboring nodes of the current node as the new current node and placed in a closed set; by setting the open set to empty, it is used to ensure that the nodes on the path planning no longer participate in the subsequent path planning. The method can realize the one-dimensional time planning of path planning; based on the slope data, the time-series sliced illumination data of the current moment or the time-series fusion data, the neighboring nodes around the current node are traversed, the heuristic function values of all the neighbors are calculated, and the neighboring nodes and the heuristic function values of the neighboring nodes are put into the open set, wherein the heuristic function value includes the mixed estimated cost composed of the illumination cost, the slope cost, the Euclidean distance cost between the current node and the preset end point, and the actual cost composed of the actual Euclidean distance cost from the current node to the neighboring nodes and the accumulated actual cost of the planned path nodes; the node with the smallest heuristic function value is selected from the open set as the pre-selected node for the next step of path planning; in the open set, the node with the smallest heuristic function value is selected as the pre-selected node for the next step of path planning; When the open set is an empty set, based on the temporal illumination slicing data or temporal illumination fusion data of the next moment, the illumination conditions of the current moment are replaced by the illumination conditions of the time-aligned reasoning data at the next moment, and the path planning is set to a waiting state, that is, the current node is continued to be selected as the current node at the next moment, the path search operation at the current moment is stopped and the path search at the next moment is entered; when the open set is not an empty set, based on the temporal illumination slicing data or temporal illumination fusion data of the next moment, the illumination conditions of the time-aligned reasoning data are forcibly updated to the illumination conditions of the next moment, the pre-selected node is used as the current node at the next moment, and the path search operation at the next moment is continued.

[0109] Figure 3 It is a lunar polar region sun synchronization according to an embodiment of the present invention. Block diagram of the path planning system.

[0110] As Figure 3 shown, the above lunar polar sun-synchronous path planning system includes a data preprocessing module 1, an enhanced U-Net model training module 2, a model inference and preprocessing module 3, and a path planning module 4.

[0111] The data preprocessing module 1 is used to preprocess the time-series illumination raw data to obtain training data with time-series slice type and time-series fusion type, as well as inference data with time-series slice type and time-series fusion type.

[0112] The enhanced U-Net model training module 2 is used to independently train multiple enhanced U-Net models by using the training data with time-series slice type and time-series fusion type and a mixed loss function, and obtain multiple trained enhanced U-Net models.

[0113] The model inference and preprocessing module 3 is used to segment and extract the inference data with time-series slice type and time-series fusion type by using multiple trained enhanced U-Net models to obtain a lunar polar friendly illumination area, and based on the preset path planning requirements, align the inference data with the lunar polar friendly illumination area in time and align the slope data corresponding to the lunar polar friendly illumination area in space.

[0114] The path planning module 4 is used to construct a heuristic function by using a mixed cost based on illumination, slope, and Euclidean distance based on the inference data aligned in time and the slope data aligned in space, and perform heuristic path planning by using an improved path planning algorithm with one-dimensional time function and in-situ waiting function to obtain the sun-synchronous planning path of the lunar polar region.

[0115] The above data preprocessing module 1 will be further described in detail below through specific embodiments.

[0116] The data preprocessing module 1 acquires and analyzes the time-series sun-synchronous illumination and slope basic data, and realizes data grouping and label definition; specifically, it includes a data storage module 11, a data parsing module 12, a data label definition module 13, and a data grouping module 14.

[0117] Among them, the data storage module 11 establishes a data storage repository and management function, and transfers all the original data into the storage space;

[0118] Among them, the data parsing module 12 parses and extracts the original data, extracts the data layer by layer according to the data stacking storage order, transfers and stores it as CSV and RGB files, and distributes it to different storage spaces;

[0119] Among them, the data parsing module 13 automatically defines tags for the data. It takes the visibility coefficient of a single grid's illumination being greater than 75% as the threshold for the true mask one-hot encoding, as shown in formula (1):

[0120] (1),

[0121] Among them, represents the mask of pixel ; represents the solar visibility coefficient of pixel , and its value range is [0, 1];

[0122] The data grouping module 14 automatically processes the data that needs to be classified and grouped, as shown in formula (2):

[0123] (2),

[0124] Among them, represents the grouping set; is the grouping coefficient, with values of 0.7, 0.2, and 0.1, indicating that the data is grouped and sliced in the order of the serial number according to the ratios of 70%, 20%, and 10% to construct the training set, validation set, and test set and output them to the specified storage space; is the test data set for model training and validation.

[0125] The following further elaborates on the above enhanced U-Net model training module 2 through specific implementation manners.

[0126] The enhanced U-Net training module 2 uses the training set and validation set data to train the enhanced U-Net model for feature extraction and semantic segmentation of the sun-synchronous basic illumination data, including the segmentation and extraction training of time-series sliced illumination data and time-series fused illumination data, and realizes the output of the extraction model for friendly illumination data; specifically, it includes a model training module 21 and a model output module 22

[0127] The model training module 21 trains the enhanced U-Net model for the segmentation and extraction of friendly illumination data, including two types: time-series sliced illumination data and time-series fused illumination data.

[0128] The model output module 22 exports the enhanced U-Net model for the segmentation and extraction of time-series illumination data, including two types: time-series sliced illumination data and time-series fused illumination data, and saves the exported models into the specified storage space respectively;

[0129] The time-series sliced light data enhanced U-Net segmentation and extraction model replaces the convolutional module for extracting each layer's feature map with a residual attention module to enhance the extraction of data features. The residual attention module consists of regularization, a channel attention module, a spatial attention module, and a residual connection. Its core architecture is shown in Equation (3):

[0130] (3),

[0131] where, is the time-series sliced type data at time , is the number of layers, represents the channel attention value, represents the spatial attention value, is the regularization feature map data, and represents the residual value of the input data, is the activation function, represents the residual attention module operation, represents the downsampling operation and represents the upsampling operation.

[0132] The time-series fusion light data enhanced U-Net segmentation and extraction model replaces the convolutional module for extracting each layer's feature map with a residual convolutional module to enhance the extraction of data features. The residual convolutional module consists of regularization, a time-series attention module, and a residual connection. A spatio-temporal fusion module is added between the encoder and the decoder. Its core architecture is shown in Equation (4):

[0133] (4),

[0134] where, is a time series data composed of grouped fusion unit time slice data, i.e., time-series fusion type data, represents the time series attention value, is the spatio-temporal fusion processing, and other symbols are consistent with the time-series sliced data;

[0135] The model training uses a loss function that combines cross-entropy and Dice and establishes backpropagation to realize the training and update of model parameters. The expression of this combined loss function is shown in Equation (5):

[0136] (5),

[0137] where, represents the total loss function, is the cross-entropy loss function, is the Dice loss function, is the hyperparameter for balancing the weights of the two different loss functions, Represents a pixel In The classified true mask Pixel In The predicted result of classification Is the number of pixels Is a small constant to avoid division by zero in the denominator.

[0138] The above model inference and preprocessing module 3 will be further described in detail through specific implementation manners below.

[0139] The model inference and preprocessing module 3 uses the trained model to extract friendly lighting areas from the time-series slices or fused lighting basic data, and completes the preprocessing of lighting and slope data according to the path planning requirements; specifically, it includes a model inference module 31 and a data preprocessing module 32.

[0140] The model inference module 31 imports the model, infers the basic lighting data set, completes the segmentation and extraction of friendly lighting data and outputs the results, including two types: time-series slice lighting data and time-series fused lighting data;

[0141] The data preprocessing module 32 groups and labels the time-series lighting data source data and the segmented and extracted data, imports the slope data, and exports the result data to the target storage space.

[0142] The above path planning module 4 will be further described in detail through specific implementation manners below.

[0143] The path planning module 4 is based on the segmented data and adopts an improved Algorithm, obtains the neighboring planning points with the minimum sun-synchronous lighting data, slope data and Euclidean distance cost through a heuristic function, and fixes the planned path, and finally outputs the sun-synchronous planned path; specifically, it includes a data preparation module 41, a path planning module 42, and a result analysis and output module 43.

[0144] The data preparation module 41 is used to integrate the data from different storage spaces into a unified data set, align the data from different data sources in time or space, and perform data checks to ensure that the data meets the temporal requirements at time points or spatial positions;

[0145] The path planning module 42 is used to carry out path planning using a data aggregator and a sun-synchronous Algorithm kernel;

[0146] The result analysis and output module 43 is used to output the step distributions of the dark, dim, and bright areas in the planned path and the step distributions of the prohibited and non-prohibited areas in the slope by recording and analyzing the situation of each planned path point, and support the output of the cumulative path map at the last moment;

[0147] The data collector circulates and retrieves data from the specified storage space, sends it to the queue for the algorithm kernel to process, and after obtaining the processing result of the algorithm kernel, the collector distributes the data result to the specified storage space;

[0148] Sun-synchronous When the algorithm kernel is initialized, the open set OpenSet is the starting point coordinate, and the closed set ClosedSet is empty. Before selecting the next optimal neighboring node, first set OpenSet to empty to ensure that the planned nodes at the previous moment have been fixed and no longer participate in the path planning, and conduct node search according to the following formula (6):

[0149] (6),

[0150] Among them, represents the neighboring node, is the current node, is

[0151] At moment or the illumination data in the time period. is the total cost of the neighboring node, is the actual cost from the starting point to the neighboring node, is the heuristic cost from the neighboring node to the target point; is obtained by adding the actual cost of the current point to the Euclidean distance from the current point to the neighboring node; is obtained by adding the Euclidean distance from the neighboring node to the target point to the slope heuristic value of the neighboring node plus the illumination value of the neighboring node ; The value of

[0152] (7),

[0153] Among them, is the slope value of, is positive infinity; The value of

[0154] (8).

[0155] The result analysis and output module 43 supports outputting the cumulative path diagram at the last moment, superimposing the historical path with the slope diagram or the time-series illumination data diagram at the last moment, and obtaining the final path planning effect diagram.

[0156] Figure 4 is a block diagram of an electronic device suitable for implementing a lunar polar sun-synchronous path planning method according to an embodiment of the present invention.

[0157] As Figure 4 shown, the electronic device 400 according to an embodiment of the present invention includes a processor 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage section 408 into a random access memory (RAM) 403. The processor 401 may include, for example, a general microprocessor (such as a CPU), an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application specific integrated circuit (ASIC)), etc. The processor 401 may also include on-board memory for caching purposes. The processor 401 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0158] In the RAM 403, various programs and data required for the operation of the electronic device 400 are stored. The processor 401, the ROM 402, and the RAM 403 are connected to each other through a bus 404. The processor 401 performs various operations of the method flow according to an embodiment of the present invention by executing the programs in the ROM 402 and / or the RAM 403. It should be noted that the programs may also be stored in one or more memories other than the ROM 402 and the RAM 403. The processor 401 may also perform various operations of the method flow according to an embodiment of the present invention by executing the programs stored in the one or more memories.

[0159] According to an embodiment of the present invention, the electronic device 400 may further include an input / output (I / O) interface 405, and the input / output (I / O) interface 405 is also connected to the bus 404. The electronic device 400 may further include one or more of the following components connected to the input / output (I / O) interface 405: an input portion 406 including a keyboard, a mouse, etc.; an output portion 407 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage portion 408 including a hard disk, etc.; and a communication portion 409 including a network interface card such as a LAN card, a modem, etc. The communication portion 409 performs communication processing via a network such as the Internet. The drive 410 is also connected to the input / output (I / O) interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage portion 408 as needed.

[0160] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist separately without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the methods according to the embodiments of the present invention are implemented.

[0161] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in conjunction with an instruction execution system, device, or device. For example, according to an embodiment of the present invention, the computer-readable storage medium may include the above-described ROM 402 and / or RAM 403 and / or one or more memories other than ROM 402 and RAM 403.

[0162] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code, and the above-mentioned module, segment of a program, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order from that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, as well as combinations of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0163] Those skilled in the art can understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments of the present invention can be combined and / or combined in various ways. All such combinations and / or combinations fall within the scope of the present invention.

[0164] The above describes the embodiments of the present invention. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. Although the embodiments are described separately above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Without departing from the scope of the present invention, those skilled in the art can make various substitutions and modifications, and all such substitutions and modifications should fall within the scope of the present invention.

Claims

1. A sun-synchronous path planning method for the lunar polar region, characterized in that The method includes: Preprocessing the raw temporal illumination data to obtain training data with temporal slice types and temporal fusion types, and inference data with temporal slice types and temporal fusion types; Independently training multiple enhanced U-Net models using the training data with temporal slice types and temporal fusion types and a mixed loss function to obtain multiple trained enhanced U-Net models; Using the multiple trained enhanced U-Net models to segment and extract the inference data with temporal slice types and temporal fusion types to obtain a lunar polar region friendly illumination area, and based on preset path planning requirements, performing temporal alignment on the inference data with the lunar polar region friendly illumination area and spatial alignment on the slope data corresponding to the lunar polar region friendly illumination area; Based on the inference data after time alignment and the slope data after spatial alignment, a heuristic function is constructed using a hybrid cost based on illumination, slope, and Euclidean distance, with improvements in both the one-dimensional time function and the ability to wait in place. A path planning algorithm performs heuristic path planning to obtain a sun-synchronous planned path in the lunar polar region.

2. The method according to claim 1, wherein Preprocessing the raw temporal illumination data to obtain multi-type training data and multi-type inference data, including: Automatically parsing and extracting the raw temporal illumination data to obtain temporal illumination slice training data and temporal illumination slice inference data; Respectively determining the grouped fusion units of the temporal illumination slice training data and the temporal illumination slice inference data based on variable time intervals; Using the grouped fusion units to perform labeled grouping on the temporal illumination slice training data to obtain temporal illumination fusion training data; Using the grouped fusion units to perform labeled grouping on the temporal illumination slice inference data to obtain temporal illumination fusion inference data.

3. The method according to claim 2, characterized in that, It also includes: Performing label automatic definition operations based on one-hot encoding on the temporal illumination slice training data and the temporal illumination fusion training data respectively, wherein the one-hot encoding of the temporal fusion illumination fusion data is obtained by key frame operations, and the key frame is to automatically obtain the intermediate frame mask data of each group of fusion units; Performing an automatic grouping operation on the temporal illumination slice training data with label information to obtain a temporal illumination slice training set, a temporal illumination slice validation set, and a temporal illumination slice test set; Performing an automatic grouping operation on the temporal illumination fusion training data with label information to obtain a temporal illumination fusion training set, a temporal illumination fusion validation set, and a temporal illumination fusion test set.

4. The method according to claim 3, wherein Independently training multiple enhanced U-Net models using the training data with temporal slice types and temporal fusion types and a mixed loss function to obtain multiple trained enhanced U-Net models, including: Based on the temporal illumination slice training set, the temporal illumination slice validation set, and the temporal illumination slice test set, using a mixed loss function based on cross-entropy loss and Dice loss to perform cross-iterative training, validation, and testing on the temporal slice enhanced U-Net model to obtain a trained temporal slice enhanced U-Net model; Based on the temporal illumination fusion training set, the temporal illumination fusion verification set and the temporal illumination fusion test set, the temporal fusion enhanced U-Net model is cross-iteratively trained, verified and tested using the hybrid loss function based on cross entropy loss and Dice loss to obtain a trained temporal fusion enhanced U-Net model.

5. The method according to claim 4, wherein The temporal slicing enhanced U-Net model includes a residual attention module, and the residual attention module includes a multi-layer perceptron, a channel attention layer, a spatial attention layer, and a temporal slicing residual connection layer; Among them, the temporal fusion enhanced U-Net model includes a multi-channel fusion input module, a residual convolution module and a spatiotemporal fusion module, the residual convolution module includes a temporal fusion attention submodule and a temporal residual connection layer, and the spatiotemporal fusion module includes a spatial convolution submodule and a temporal convolution submodule.

6. The method according to claim 4, wherein The trained multiple enhanced U-Net models are used to segment and extract the inference data with time series slicing type and time series fusion type, and the polar friendly illumination areas of the moon are obtained, including: Using the trained temporal slicing enhanced U-Net model, feature extraction and region segmentation operations are performed on the temporal illumination slicing inference data to obtain the temporal illumination slicing inference data having the lunar polar friendly illumination area; The trained temporal fusion enhanced U-Net model is used to perform feature extraction and segmentation operations based on the grouping fusion unit on the temporal illumination fusion data to obtain temporal illumination slice inference data of the lunar polar friendly illumination area.

7. The method according to claim 4, characterized in that, Based on the inference data after time alignment and the slope data after space alignment, a heuristic function is constructed using a mixed cost based on illumination, slope, and Euclidean distance, and improvements are made to have both one-dimensional time functionality and the ability to wait in place The path planning algorithm performs heuristic path planning to obtain a sun-synchronous planned path in the lunar polar region, including: Initialize the open set with a preset starting point, preset the closed set to an empty set, set the initial state of the heuristic function to the estimated costs of illumination, slope and Euclidean distance of the preset starting point and the actual cost of the preset starting point, and set the preset starting point as the current node; In the case where the current node is the preset end point, the current node is placed in the closed set, node search and path planning are terminated, and the closed set is returned as a planned path result; When the current node is illuminated, placing the current node into the closed set; When the current node has no illumination, a node with the smallest illumination, slope, and Euclidean distance cost is selected from the neighboring nodes around the current node and is placed in the closed set as the current node; Before searching for a pre-selected node, the open set is set to empty to ensure the time one-dimensional function of the path planning, wherein the searching for the pre-selected node is to traverse the neighboring nodes around the current node; Calculate the heuristic function value of the neighboring node, and put the neighboring node and the heuristic function value of the neighboring node into an open set, wherein the heuristic function value includes the mixed estimated cost consisting of the illumination cost of the neighboring node, the slope cost, the Euclidean distance cost with the preset end point, and the actual cost consisting of the Euclidean distance cost from the current node to the neighboring node and the accumulated actual cost of the planned path nodes; After searching for the preselected node, select the node with the smallest heuristic function value from the open set as the preselected node for the next path planning. After searching for the preselected node, replace the lighting condition at the current moment with the lighting condition at the next moment of the inference data after time alignment. After searching for the preselected node, when the open set is an empty set, set the path planning to a waiting state, that is, continue to select the current node as the current node at the next moment, stop the path search operation at the current moment and enter the path search at the next moment. After searching for the preselected node, when the open set is not an empty set, use the preselected node as the current node at the next moment and continue the path search operation at the next moment.

8. A lunar polar sun-synchronous path planning system, characterized in that The system includes: A data preprocessing module for preprocessing the temporal lighting raw data to obtain training data with temporal slice type and temporal fusion type and inference data with temporal slice type and temporal fusion type. An enhanced U-Net model training module for independently training multiple enhanced U-Net models using the training data with temporal slice type and temporal fusion type and a hybrid loss function to obtain multiple trained enhanced U-Net models. A model inference and preprocessing module for segmenting and extracting the lunar polar region friendly lighting area from the inference data with temporal slice type and temporal fusion type using the multiple trained enhanced U-Net models, and based on the preset path planning requirements, performing time alignment on the inference data with the lunar polar region friendly lighting area and spatial alignment on the slope data corresponding to the lunar polar region friendly lighting area. The path planning module is used to construct a heuristic function by using a hybrid cost based on illumination, slope, and Euclidean distance based on the inference data after time alignment and the slope data after spatial alignment, and has an improved one-dimensional time function and an in-place waiting function The path planning algorithm performs heuristic path planning to obtain a sun-synchronous planned path in the lunar polar region.

9. An electronic device, comprising: One or more processors; A memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, The computer program or instruction, when executed by the processor, implements the steps of the method according to any one of claims 1 to 7.

Images