Precise target identification and path planning method and system based on point cloud segmentation

Abstract

The invention provides an accurate target identification and path planning method and system based on point cloud segmentation, and relates to the technical field of robot navigation and path planning, and the method comprises the steps: firstly, collecting three-dimensional point cloud data of a target region, and setting related parameters; then, segmenting the point cloud by using a region growing algorithm to obtain a plurality of target regions; thirdly, extracting target features based on a segmentation result, and performing target recognition by using a support vector machine (SVM); then, calculating an optimal path from the starting point to the target point through an improved dynamic weighting A * algorithm; and finally, introducing time sequence information in the path planning process to realize real-time path optimization. Through the technical scheme of the invention, the accuracy of target identification is improved, the efficiency of path planning is optimized, the robot can better navigate in a complex environment, and the method has important application value.

Description

Technical Field

[0001] The present invention relates to the technical field of robot navigation and path planning, and in particular to a method and system for accurate target recognition and path planning based on point cloud segmentation. Background Art

[0002] In application fields such as robot navigation and autonomous driving, target recognition and path planning are key technologies for achieving autonomous navigation. With the development of computer vision and artificial intelligence technology, three-dimensional point cloud data has gradually become the main way to describe the spatial information of objects. By segmenting and identifying the point cloud data of the target area, environmental information can be obtained more accurately, thus providing strong support for path planning. However, current point cloud segmentation and recognition methods often have problems such as insufficient accuracy and poor real-time performance. Especially in complex scenarios, it is difficult to effectively distinguish between targets and obstacles, resulting in limited accuracy and safety of path planning.

[0003] In the existing technology, target recognition methods based on support vector machines (SVM) or convolutional neural networks (CNN) have achieved good results in two-dimensional image recognition, but in point cloud data, due to data dimension and sparsity issues, the recognition accuracy still needs to be improved. At the same time, path planning mostly uses traditional graph search methods such as the A* algorithm, but these algorithms often ignore the real-time changes of the environment in a dynamic environment and lack dynamic adjustment capabilities, which leads to deviations or detours in the path planning process.

[0004] Therefore, there is an urgent need for an accurate target recognition and path planning method based on point cloud segmentation, which can improve the accuracy of target positioning through effective target segmentation and recognition on the basis of real-time acquisition of environmental changes, and improve the efficiency and safety of path selection in combination with an optimized path planning algorithm to meet the needs of applications such as robot navigation, unmanned driving and intelligent inspection. Summary of the invention

[0005] In order to solve the technical problems in the prior art of insufficient target recognition accuracy in complex environments, poor real-time path planning and weak adaptability to dynamic environments, the present invention provides a method and system for accurate target recognition and path planning based on point cloud segmentation.

[0006] The technical solution provided by the present invention is as follows:

[0007] First aspect:

[0008] The present invention provides a method for accurate target recognition and path planning based on point cloud segmentation, comprising:

[0009] S1, collecting the three-dimensional point cloud data of the target area, setting parameters P1 and P2, where P1 is the density parameter of the point cloud data, used to represent the number of points per unit volume, and P2 is the segmentation threshold, used to determine the similarity standard of point cloud segmentation;

[0010] S2. Use a region growing algorithm to segment the point cloud and obtain multiple target regions. The region growing algorithm is as follows:

[0011] R(G)={p∈P|d(p,R) <P2}

[0012] Where R(G) is the growth area, P is the point cloud dataset, d(p,R) is the distance from point p to region R, and if the distance is less than the threshold P2, point P is included;

[0013] S3. Based on the segmentation results, the target features are extracted and the target is recognized. The recognition method is based on the support vector machine (SVM) algorithm. The improved feature vector construction formula is:

[0014]

[0015] Among them, F v is the feature vector, W i is the weight of the i-th feature, V i is the value of the i-th feature, n is the total number of features, and m is the total number of weights;

[0016] S4. Calculate the optimal path from the starting point to the target point by using the improved dynamic weighted A* algorithm. The optimal path planning formula is:

[0017]

[0018] Among them, C(P) is the path cost, D k is the actual distance of the kth segment in the path, and N is the total number of path segments;

[0019] S5. Introduce timing information in the path planning process, through the formula:

[0020] T eff =T real +P1·T obstacle

[0021] Among them, T eff is the effective path time, T real is the actual elapsed time, T obstacle is the time delay caused by the obstacle.

[0022] Second aspect:

[0023] The present invention provides a precise target recognition and path planning system based on point cloud segmentation, comprising:

[0024] processor;

[0025] A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the precise target recognition and path planning method based on point cloud segmentation as described in the first aspect is implemented.

[0026] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0027] (1) In the present invention, by introducing a point cloud segmentation method based on region growing, the target area can be segmented more accurately, which significantly improves the accuracy of target recognition and can effectively distinguish targets from obstacles even in complex environments;

[0028] (2) In the present invention, the support vector machine (SVM) algorithm and the improved feature extraction method are combined to achieve high-precision recognition of the target, and the rationality of path selection is improved through the dynamic weighted path planning algorithm, thereby improving the accuracy and safety of path planning;

[0029] (3) In the present invention, real-time environmental change information is incorporated into the path planning process, so that the system can dynamically adjust the path, enhance the adaptability of the system in a dynamic environment, effectively avoid detours and deviations, and improve the efficiency and safety of navigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 A schematic diagram of a flow chart of a method for accurate target recognition and path planning based on point cloud segmentation provided by an embodiment of the present invention;

[0032] Figure 2 A schematic diagram of the structure of a precise target recognition and path planning system based on point cloud segmentation provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0034] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0035] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0036] In the embodiments of the present invention, sometimes a subscript such as W1 may be mistakenly written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are consistent.

[0037] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0038] Reference Manual Attached Figure 1 , showing a flow chart of a method for accurate target recognition and path planning based on point cloud segmentation provided by an embodiment of the present invention.

[0039] The embodiment of the present invention provides a method for accurate target recognition and path planning based on point cloud segmentation. The method can be implemented by an accurate target recognition and path planning device based on point cloud segmentation. The accurate target recognition and path planning device based on point cloud segmentation can be a terminal or a server. The processing flow of the method for accurate target recognition and path planning based on point cloud segmentation can include the following steps:

[0040] S1. Collect the three-dimensional point cloud data of the target area and set the parameters P1 and P2, where P1 is the density parameter of the point cloud data, which is used to represent the number of points per unit volume, and P2 is the segmentation threshold, which is used to determine the similarity standard of point cloud segmentation.

[0041] It should be noted that the 3D point cloud data of the target area is collected and the density parameters and segmentation thresholds are set to ensure that spatial information with sufficient details is provided in the actual scene. The density parameters can be adjusted according to the density of the point cloud, so that the system can adapt to different point cloud distributions in complex or simple environments, while the segmentation threshold sets the similarity standard and determines the system's tolerance for adjacent points. Through these parameter settings, the quality and structure of the initial data can be controlled, making subsequent processing stable and accurate, laying a solid foundation for further segmentation and feature extraction.

[0042] S2. Use the region growing algorithm to segment the point cloud and obtain multiple target regions. The region growing algorithm is as follows:

[0043] R(G)={p∈P|d(p,R) <P2}

[0044] Among them, R(G) is the growth area, P is the point cloud dataset, d(p,R) is the distance from point p to region R, and if the distance is less than the threshold P2, point P is included.

[0045] It should be noted that the use of region growing algorithm for point cloud segmentation can automatically form continuous regions by utilizing the spatial relationship between points. This algorithm selects the initial seed point and gradually incorporates adjacent points into the same region to obtain the accurate distribution of the target region. The process of region growing can eliminate the segmentation error caused by discrete data, so that the extracted segmented region has spatial coherence and accuracy, thereby ensuring that the segmentation result is highly consistent with the actual target structure, which is convenient for subsequent recognition and path planning.

[0046] S3. Based on the segmentation results, the target features are extracted and the target is recognized. The recognition method is based on the support vector machine (SVM) algorithm. The improved feature vector construction formula is:

[0047]

[0048] Among them, F v is the feature vector, W i is the weight of the i-th feature, V i is the value of the ith feature, n is the total number of features, and m is the total number of weights.

[0049] It should be noted that the target features of the segmented area are extracted and identified. Under the optimization of the support vector machine (SVM) classifier, key features can be highlighted and the target type can be accurately identified. By optimizing the feature weights and structures through the improved feature vector construction method, the system pays more attention to key features, thereby effectively distinguishing important targets during the recognition process. This can improve recognition accuracy, ensure the accuracy of locating targets in complex scenes, and provide accurate target locations for path planning.

[0050] S4. Calculate the optimal path from the starting point to the target point through the improved dynamic weighted A* algorithm. The optimal path planning formula is:

[0051]

[0052] Among them, C(P) is the path cost, D k is the actual distance of the kth segment in the path, and N is the total number of path segments.

[0053] It should be noted that the optimal path from the starting point to the target is planned by using the improved dynamic weighted A* algorithm, so that the path is not only the shortest in space, but also comprehensively considers the impact of environmental changes on the path. This path planning combines weighted parameters to adjust the generation of the path in real time to adapt to different obstacles and path risks. This method can ensure the effectiveness of the path in complex terrain, improve the overall safety of the system, and shorten the execution time, so that the path planning is both in line with environmental constraints and efficient.

[0054] S5. Introduce timing information in the path planning process, through the formula:

[0055] T eff =T real +P1·T obstacle

[0056] Among them, T eff is the effective path time, T real is the actual elapsed time, T obstacle is the time delay caused by the obstacle.

[0057] It should be noted that the introduction of timing information in the path planning process ensures that the system can dynamically adapt to environmental changes and achieve more real-time path adjustments. Timing information takes into account the execution time of the effective path and the delay caused by factors such as obstacles, so that the system can estimate and handle potential time delays when selecting a path. This method improves the reliability of path planning, and can adjust the path in time even in emergencies, enhancing the system's responsiveness in a dynamic environment.

[0058] In a possible implementation, the region growing algorithm performs similarity judgment based on the normal vector of the point and the threshold P2:

[0059]

[0060] Among them, S(p i ,p j ) represents point p i and p j similarity, N(p) is the normal vector of point p.

[0061] In a possible implementation, feature extraction further includes:

[0062] The improved principal component analysis (PCA) method is used to dynamically adjust the weight matrix W to improve the extraction effect of key features. The calculation formula of the weight matrix W is:

[0063]

[0064] Among them, F importance Score the importance of each feature, W adapt Dynamically adjust according to feature importance, W adapt The formula used for feature transformation is:

[0065] Z=X·W adapt

[0066] Among them, Z is the feature matrix after dimension reduction, and X is the original feature matrix.

[0067] It should be noted that the improved principal component analysis (PCA) method is introduced and the weight matrix W is dynamically adjusted to effectively improve the extraction effect of key features. The traditional PCA method measures the importance of features by variance, but in complex environments, a single variance standard may lead to the loss of some key features. Through the dynamically adjusted weight matrix W, the system can flexibly assign weights according to the importance score of the features, so that key features are given a higher retention priority during the dimensionality reduction process, and important information related to the identification target is retained to the maximum extent.

[0068] In this process, the calculation formula of the weight matrix W uses the feature importance score to make the feature extraction process dynamically adaptable. When the target environment or data structure changes, the system automatically optimizes and adjusts the weight distribution, effectively improving the accuracy and robustness of feature extraction. Through this improved PCA method, the original feature matrix can be converted into a reduced-dimensional feature matrix, achieving efficient compression and extraction of information, ensuring the accurate use of key information in subsequent recognition and analysis processes, and improving the overall recognition capability of the system.

[0069] In one possible implementation, the heuristic function of the path planning algorithm is defined as:

[0070]

[0071] Among them, H(n) is the heuristic value of the current node n, and D(n,target) is the Euclidean distance from the current node to the target node.

[0072] It should be noted that the heuristic function of the path planning algorithm is used to evaluate the estimated distance between the current node and the target node, thereby providing a more reasonable direction for path planning. Guided by the heuristic function, the path planning algorithm can achieve optimal path planning with a smaller amount of calculation, which not only speeds up the path-finding process, but also avoids blind exploration of non-optimal paths.

[0073] In the process of path planning, the design of heuristic functions is introduced to enable the algorithm to approach the target node more quickly while taking into account the overall optimization of the path. This method can significantly improve the efficiency of path planning in practical applications, especially in scenarios that require fast response and path optimization, ensuring the accuracy of the path and the efficiency of execution.

[0074] In a possible implementation, it further includes a target recognition module based on deep learning, using a convolutional neural network (CNN) for target classification, and the loss function of the CNN is:

[0075]

[0076] Among them, L is the loss value, y i is the actual label, is the predicted label, β is the regularization coefficient, R j is the regularization term.

[0077] It should be noted that the deep learning-based target recognition module uses a convolutional neural network (CNN) to classify targets. The introduction of this method enables the system to process complex data, thereby improving the accuracy and robustness of target recognition. With its powerful feature extraction capabilities, convolutional neural networks can automatically identify and learn key features in input data, and are particularly suitable for tasks that process three-dimensional point cloud data and image data.

[0078] By using the CNN loss function, the system can effectively evaluate the gap between the model prediction and the actual label, and update the parameters based on this gap. This feedback mechanism ensures that the model gradually improves its classification performance during the training process, while the introduced regularization term helps prevent overfitting, thereby enhancing the generalization ability of the model. Ultimately, the integration of this deep learning module not only improves the accuracy of target recognition, but also provides more reliable target information for subsequent path planning, laying a solid foundation for the overall performance optimization of the system.

[0079] In a possible implementation, point cloud data collection further includes:

[0080] Adopting the adaptive sampling method given the environmental characteristics, the sampling interval T s According to the complexity of the environment C env Dynamic adjustment, the specific calculation method is:

[0081]

[0082] Among them, T base is the basic sampling interval, C max is the maximum value of the environment complexity to ensure more frequent sampling in complex environments.

[0083] It should be noted that point cloud data collection adopts an adaptive sampling method to improve sampling efficiency and data quality. By dynamically adjusting the sampling interval according to the complexity of the environment, the system can achieve more flexible point cloud data acquisition under different environmental conditions.

[0084] Specifically, the basic sampling interval is set to an initial value, and in areas with high environmental complexity, the system will shorten the sampling interval to ensure that more dense point cloud data is collected in these areas. This adjustment mechanism can effectively cope with possible obstacles and changes in the environment, ensuring that details in important feature areas are fully captured. On the contrary, in areas with low environmental complexity, the sampling interval can be appropriately increased to save computing resources.

[0085] Through this adaptive sampling method, the system not only improves the accuracy and completeness of point cloud data, but also improves the efficiency of subsequent processing stages, ultimately providing more reliable basic data support for accurate target recognition and path planning.

[0086] In a possible implementation, the path planning process includes dynamic detection of environmental obstacles, and the dynamic monitoring model is:

[0087] O={o|o∈obstacles∧D(o,P) <P2}

[0088] Where O is the obstacle set, and D(o,P) is the distance from obstacle o to path P.

[0089] It should be noted that the path planning process enhances the dynamic detection of environmental obstacles to ensure effective path optimization under real-time environmental changes. Through the dynamic monitoring model, the system can continuously track and update the location of obstacles and their relationship with the path, thereby improving the safety and reliability of path planning.

[0090] Specifically, the dynamic monitoring model uses an obstacle set to describe all potential obstacles in the environment and calculates the distance from each obstacle to the planned path. This distance information is an important reference factor for path planning because it can reflect the impact of obstacles on the path in real time. For example, when an obstacle is close to the path, the system can make timely adjustments to select a safer path to avoid collision risks.

[0091] This dynamic detection mechanism can not only adapt to changes in the environment and ensure the flexibility of path planning, but also improve the efficiency of path execution, making the final selected path not only effective but also safe. By continuously monitoring and updating obstacle information, the system can make decisions in real time, thereby optimizing the entire path planning process.

[0092] In one possible implementation, the target recognition process improves recognition accuracy by combining an ensemble learning method, and the output of the ensemble method is defined as:

[0093]

[0094] Among them, Y ensemble is the integration result, Y i is the output of the i-th model, and N is the number of models.

[0095] It should be noted that the target recognition process significantly improves the recognition accuracy by combining the ensemble learning method. Ensemble learning is a machine learning technique that combines multiple different models to obtain better prediction performance than a single model. Specifically, this method can combine the advantages of each model to reduce the deviations and errors that may be caused by a single model.

[0096] In practical applications, ensemble methods integrate the output results of multiple base models to form a final ensemble result. This output can be calculated in a variety of ways, such as weighted average, voting mechanism or other ensemble strategies. This process ensures that when faced with complex data sets, the models can work together to provide more accurate and robust prediction results. For example, some models may perform well on specific types of data, while other models are more advantageous in different situations. Ensemble learning can effectively integrate these differences and make the final recognition results more representative.

[0097] In addition, ensemble learning is more robust when dealing with noisy data and abnormal situations, which enables the target recognition system to maintain high accuracy in a changing environment. This method not only improves the adaptability of the recognition system, but also provides higher stability for practical applications, ensuring that the target can be accurately identified. By making full use of the strengths of different models, the system can analyze the input data more comprehensively, thereby optimizing the final recognition effect.

[0098] In one possible implementation, path planning comprehensively evaluates path cost, path safety, and path efficiency, and adjusts path selection to minimize cost and risk while improving path execution efficiency. The optimization process also needs to consider real-time environmental changes to achieve dynamic path optimization.

[0099] It should be noted that path planning optimizes path selection by comprehensively evaluating path cost, path safety, and path efficiency. The purpose of this comprehensive evaluation is to minimize driving costs and risks while ensuring safety, while improving path execution efficiency.

[0100] Path cost assessment mainly considers the economic investment under different path selections, such as time and energy consumption, in order to find a route that is both economical and effective. Path safety focuses on potential obstacles, traffic conditions and other factors that may affect driving safety, in order to reduce the probability of accidents and ensure the safety of the driving process.

[0101] Path efficiency involves the smoothness of driving, including factors such as traffic congestion and road conditions. By evaluating these factors, the path planning system can more reasonably choose the best route, avoid unnecessary delays, and thus improve overall driving efficiency.

[0102] In addition, the optimization process needs to take environmental changes into account in real time. Changes in environmental conditions, such as weather, traffic conditions, or emergencies, can have a significant impact on path selection. Therefore, the system should have the ability to dynamically adjust and update path selection in a timely manner to cope with the changing environment. In this way, path planning not only reduces both cost and risk, but also improves the flexibility and adaptability of the path, ensuring that the system can complete tasks efficiently and safely under different conditions. Ultimately, this dynamic path optimization method helps to improve the overall performance and user experience of the target recognition and path planning system.

[0103] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0104] (1) In the present invention, by introducing a point cloud segmentation method based on region growing, the target area can be segmented more accurately, which significantly improves the accuracy of target recognition and can effectively distinguish targets from obstacles even in complex environments;

[0105] (2) In the present invention, the support vector machine (SVM) algorithm and the improved feature extraction method are combined to achieve high-precision recognition of the target, and the rationality of path selection is improved through the dynamic weighted path planning algorithm, thereby improving the accuracy and safety of path planning;

[0106] (3) In the present invention, real-time environmental change information is incorporated into the path planning process, so that the system can dynamically adjust the path, enhance the adaptability of the system in a dynamic environment, effectively avoid detours and deviations, and improve the efficiency and safety of navigation.

[0107] Reference Manual Attached Figure 2, showing a structural schematic diagram of a precise target recognition and path planning system based on point cloud segmentation provided by an embodiment of the present invention.

[0108] The present invention further provides a precise target recognition and path planning system 20 based on point cloud segmentation, which is applied to the precise target recognition and path planning method based on point cloud segmentation, and comprises:

[0109] Processor 201;

[0110] The memory 202 stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the method for accurate target recognition and path planning based on point cloud segmentation as described in the method embodiment is implemented.

[0111] The precise target recognition and path planning system based on point cloud segmentation provided by the present invention can execute the precise target recognition and path planning method based on point cloud segmentation and achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0112] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0113] (1) In the present invention, by introducing a point cloud segmentation method based on region growing, the target area can be segmented more accurately, which significantly improves the accuracy of target recognition and can effectively distinguish targets from obstacles even in complex environments;

[0114] (2) In the present invention, the support vector machine (SVM) algorithm and the improved feature extraction method are combined to achieve high-precision recognition of the target, and the rationality of path selection is improved through the dynamic weighted path planning algorithm, thereby improving the accuracy and safety of path planning;

[0115] (3) In the present invention, real-time environmental change information is incorporated into the path planning process, so that the system can dynamically adjust the path, enhance the adaptability of the system in a dynamic environment, effectively avoid detours and deviations, and improve the efficiency and safety of navigation.

[0116] The above contents are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0117] There are a few points to note:

[0118] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention, and other structures may refer to the general design.

[0119] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0120] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0121] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A precise target recognition and path planning method based on point cloud segmentation, characterized in that: include: S1, collecting the three-dimensional point cloud data of the target area, setting parameters P1 and P2, where P1 is the density parameter of the point cloud data, used to represent the number of points per unit volume, and P2 is the segmentation threshold, used to determine the similarity standard of point cloud segmentation; S2. Use a region growing algorithm to segment the point cloud and obtain multiple target regions. The region growing algorithm is as follows: R(G)={p∈P|d(p,R) <P2} Where R(G) is the growth area, P is the point cloud dataset, d(p,R) is the distance from point p to region R, and if the distance is less than the threshold P2, point P is included; S3. Based on the segmentation results, the target features are extracted and the target is recognized. The recognition method is based on the support vector machine (SVM) algorithm. The improved feature vector construction formula is: Among them, F v is the feature vector, W i is the weight of the i-th feature, V i is the value of the i-th feature, n is the total number of features, and m is the total number of weights; S4. Calculate the optimal path from the starting point to the target point by using the improved dynamic weighted A* algorithm. The optimal path planning formula is: Among them, C(P) is the path cost, D k is the actual distance of the kth segment in the path, and N is the total number of path segments; S5. Introduce timing information in the path planning process, through the formula: T eff =T real +P1·T obstacle Among them, T eff is the effective path time, T real is the actual elapsed time, T obstacle is the time delay caused by the obstacle.

2. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: The region growing algorithm makes similarity judgment based on the normal vector of the point and the threshold P2: Among them, S(p i ,p j ) represents point p i and p j similarity, N(p) is the normal vector of point p.

3. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: The feature extraction further comprises: The improved principal component analysis (PCA) method is used to improve the extraction effect of key features by dynamically adjusting the weight matrix W. The calculation formula of the weight matrix W is: Among them, F importance Score the importance of each feature, W adapt Dynamically adjust according to feature importance, W adapt The formula used for feature transformation is: Z=X·W adapt Among them, Z is the feature matrix after dimension reduction, and X is the original feature matrix.

4. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: The heuristic function of the path planning algorithm is defined as: Among them, H(n) is the heuristic value of the current node n, and D(n,target) is the Euclidean distance from the current node to the target node.

5. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: It further includes a target recognition module based on deep learning, using a convolutional neural network (CNN) for target classification, and the loss function of the CNN is: Among them, L is the loss value, y i is the actual label, is the predicted label, β is the regularization coefficient, R j is the regularization term.

6. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: The point cloud data collection further includes: Adopting an adaptive sampling method given the environmental characteristics, the sampling interval T s According to the complexity of the environment C env Dynamic adjustment, the specific calculation method is: Among them, T base is the basic sampling interval, C max is the maximum value of the environment complexity to ensure more frequent sampling in complex environments.

7. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: include: The path planning process includes dynamic detection of environmental obstacles. The dynamic monitoring model is: O={o|o∈obstacles∧D(o,P) <P2} Where O is the obstacle set, and D(o,P) is the distance from obstacle o to path P.

8. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: include: The Mubao recognition process improves the recognition accuracy by combining the ensemble learning method, and the output of the ensemble method is defined as: Among them, Y ensemble is the integration result, Y i is the output of the i-th model, and N is the number of models.

9. The method for accurate target recognition and path planning based on point cloud segmentation according to claim 1, characterized in that: include: The path planning comprehensively evaluates the path cost, path safety and path efficiency, adjusts the path selection to minimize the cost and risk, and improves the execution efficiency of the path. The optimization process also needs to consider real-time environmental changes to achieve dynamic path optimization.

10. A precise target recognition and path planning system based on point cloud segmentation, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for accurate target recognition and path planning based on point cloud segmentation as described in any one of claims 1 to 9 is implemented.

Images