Dynamic obstacle avoidance path planning method and system based on obstacle detection

Abstract

The invention provides a dynamic obstacle avoidance path planning method and system based on obstacle detection, and relates to the technical field of path planning, and the method comprises the steps: carrying out the path pre-planning according to a current fault object recognition result and obstacle movement change characteristics, carrying out the moving region positioning of a dynamic obstacle, and carrying out the path planning; inserting a self-adaptive path rule in the dynamic moving area; and performing path execution according to a pre-planned path, activating the adaptive path rule when the dynamic moving area is reached, and performing matching according to the real-time sensor data and the adaptive path rule to generate adaptive path planning information of the dynamic moving area. The method aims to solve the technical problem that path planning is not flexible and accurate enough due to the fact that static obstacles and dynamic obstacles are difficult to effectively distinguish in a traditional method, path execution errors caused by dynamic changes of the obstacles are avoided through self-adaptive path rules and moving area positioning of the dynamic obstacles, and path planning accuracy is improved. And the flexibility and accuracy of path planning are obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a dynamic obstacle avoidance path planning method and system based on obstacle detection. Background Art

[0002] Most existing technologies use obstacle detection and recognition algorithms based on sensors (such as lidar, cameras, ultrasound, etc.). Although these algorithms can identify the location of obstacles, they still have many shortcomings in distinguishing between static obstacles and dynamic obstacles. The behavioral characteristics of dynamic obstacles (such as speed, direction, frequent changes, etc.) are different from those of static obstacles, but traditional algorithms often cannot accurately capture these dynamic characteristics, resulting in the movement trends of dynamic obstacles being ignored during path planning, which in turn affects the flexibility and accuracy of the planning results.

[0003] Therefore, how to effectively distinguish between static and dynamic obstacles and incorporate the behavioral characteristics of dynamic obstacles into path planning decisions has become a difficult problem that needs to be urgently solved in current path planning technology. Summary of the invention

[0004] The purpose of the present invention is to provide a dynamic obstacle avoidance path planning method and system based on obstacle detection, so as to solve the technical problem that the traditional method is difficult to effectively distinguish static obstacles from dynamic obstacles, resulting in insufficient flexibility and accuracy in path planning, including:

[0005] In a first aspect, the present invention provides a dynamic obstacle avoidance path planning method based on obstacle detection, comprising: performing obstacle identification on collected real-time sensor data to obtain a current fault object identification result; performing fault object type discrimination based on the current fault object identification result, marking the fault object type, wherein the fault object type includes static and dynamic; acquiring historical sensor data of the same area, and using the historical sensor data to match and analyze the current fault object identification result to obtain obstacle movement change characteristics; performing path pre-planning based on the current fault object identification result and obstacle movement change characteristics, and locating the moving area of ​​dynamic obstacles, and inserting adaptive path rules in the dynamic moving area; performing path execution according to the pre-planned path, activating the adaptive path rules when reaching the dynamic moving area, and generating adaptive path planning information for the dynamic moving area based on matching the real-time sensor data with the adaptive path rules.

[0006] Preferably, the dynamic obstacle avoidance path planning method based on obstacle detection further includes: acquiring multi-source sensor data, performing identification feature analysis on the multi-source sensor data to determine the identification types of each data source; based on the identification types of each data source, performing faulty object identification on the multi-source sensor data to obtain corresponding faulty object identification type features; integrating the corresponding faulty object identification type features of the identification types of each data source to obtain the current faulty object identification result.

[0007] Preferably, the dynamic obstacle avoidance path planning method based on obstacle detection further includes: performing faulty object identification evaluation on the multi-source sensor data to obtain a faulty object identification evaluation result; establishing a mapping relationship between the faulty object identification evaluation result and each data source; based on the mapping relationship, obtaining the identification types of each data source according to the faulty object identification evaluation result, where the identification type is the identification dimension corresponding to the feature identification evaluation value higher than the threshold.

[0008] Preferably, the identification dimensions included in the dynamic obstacle avoidance path planning method based on obstacle detection include: appearance features, distance, speed, size, and penetrability features.

[0009] Preferably, the dynamic obstacle avoidance path planning method based on obstacle detection further includes: dividing the faulty object distribution area according to the current faulty object identification result and the obstacle movement change characteristics to obtain a static faulty object area, a regular moving faulty object area, and a dynamic moving faulty object area; obtaining the faulty object characteristics of the static faulty object area and the regular moving faulty object area, and the path distribution characteristics of the dynamic moving faulty object area; performing preliminary path planning according to the faulty object characteristics of the static faulty object area and the regular moving faulty object area and the path distribution characteristics of the dynamic moving faulty object area to obtain a pre-planned path; setting the dynamic moving faulty object area as the dynamic moving area, and performing dynamic classification according to the path distribution characteristics and historical change records of the dynamic moving area to construct the adaptive path rule, and inserting the adaptive path rule into the dynamic moving area.

[0010] Preferably, the dynamic obstacle avoidance path planning method based on obstacle detection further includes: performing clustering according to the historical change records to obtain movement change categories; performing path planning feature analysis on each change category according to the movement change categories to obtain dynamic interference characteristics and path constraint conditions; configuring the adaptive path rule of the dynamic moving area according to the dynamic interference characteristics and path constraint conditions of the movement change categories.

[0011] Preferably, the dynamic obstacle avoidance path planning method based on obstacle detection further includes: determining the obstacle characteristics of the static obstacle area according to the obstacle occlusion edge size of the static obstacle area; determining the obstacle characteristics of the regularly moving obstacle area according to the moving area, frequency, and speed of the obstacles in the regularly moving obstacle area; and extracting the path distribution characteristics according to the historical path of the dynamically moving obstacle area.

[0012] In a second aspect, the present invention also provides a dynamic obstacle avoidance path planning system based on obstacle detection for performing the dynamic obstacle avoidance path planning method as described in the first aspect, including: an obstacle recognition module for recognizing obstacles from the collected real-time sensor data to obtain the current obstacle recognition result; a fault type discrimination module for discriminating the fault type according to the current obstacle recognition result and marking the fault type, where the fault type includes static and dynamic; an obstacle movement change characteristic acquisition module for obtaining the historical sensing data of the same area and performing matching analysis on the current obstacle recognition result using the historical sensing data to obtain the obstacle movement change characteristics; a movement area positioning module for performing path pre-planning according to the current obstacle recognition result and the obstacle movement change characteristics, positioning the movement area of the dynamic obstacle, and inserting an adaptive path rule into the dynamic movement area; and an adaptive path planning information generation module for performing path execution according to the pre-planned path, activating the adaptive path rule when reaching the dynamic movement area, and generating the adaptive path planning information of the dynamic movement area by matching the real-time sensor data with the adaptive path rule.

[0013] The embodiments of the present invention have the following advantages:

[0014] By recognizing obstacles from the collected real-time sensor data to obtain the current obstacle recognition result; discriminating the fault type according to the current obstacle recognition result and marking the fault type, where the fault type includes static and dynamic; obtaining the historical sensing data of the same area and performing matching analysis on the current obstacle recognition result using the historical sensing data to obtain the obstacle movement change characteristics; performing path pre-planning according to the current obstacle recognition result and the obstacle movement change characteristics, positioning the movement area of the dynamic obstacle, and inserting an adaptive path rule into the dynamic movement area; and performing path execution according to the pre-planned path, activating the adaptive path rule when reaching the dynamic movement area, and generating the adaptive path planning information of the dynamic movement area by matching the real-time sensor data with the adaptive path rule. That is to say, through the adaptive path rule and the movement area positioning of the dynamic obstacle, the path execution error caused by the dynamic change of the obstacle is avoided, and the flexibility and accuracy of path planning are significantly improved. Description of the Drawings

[0015] Figure 1 It is a flowchart of the steps of a method for dynamic obstacle avoidance path planning based on obstacle detection according to the present invention;

[0016] Figure 2 It is a schematic structural diagram of a system for dynamic obstacle avoidance path planning based on obstacle detection according to the present invention.

[0017] Description of the reference numerals:

[0018] Obstacle recognition module 11, malfunction type discrimination module 12, obstacle movement change feature acquisition module 13, moving area positioning module 14, adaptive path planning information generation module 15. Detailed implementation manners

[0019] By providing a method and system for dynamic obstacle avoidance path planning based on obstacle detection, the present invention solves the technical problem that traditional methods are difficult to effectively distinguish static obstacles and dynamic obstacles, resulting in inflexible and inaccurate path planning. Through adaptive path rules and the positioning of the moving area of dynamic obstacles, the error in path execution caused by the dynamic change of obstacles is avoided, and the flexibility and accuracy of path planning are significantly improved.

[0020] Next, the technical solutions in the present invention will be clearly and completely described with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited by the example embodiments described here. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention. In addition, it should be noted that for the sake of description, only parts related to the present invention are shown in the drawings rather than all.

[0021] Embodiment 1. Please refer to the attached Figure 1 , the present invention provides a method for dynamic obstacle avoidance path planning based on obstacle detection, which is applied to a system for dynamic obstacle avoidance path planning based on obstacle detection, and specifically includes the following steps:

[0022] Step 101: Identify obstacles from the collected real-time sensor data to obtain the current fault object recognition result; Step 102: Discriminate the type of the fault object according to the current fault object recognition result and mark the fault object type, where the fault object type includes static and dynamic; Step 103: Obtain the historical sensing data of the same area, and use the historical sensing data to perform matching analysis on the current fault object recognition result to obtain the obstacle movement change characteristics; Step 104: Perform pre-path planning according to the current fault object recognition result and the obstacle movement change characteristics, locate the movement area of the dynamic obstacle, and insert adaptive path rules into the dynamic movement area; Step 105: Execute the path according to the pre-planned path. When reaching the dynamic movement area, activate the adaptive path rules, and generate the adaptive path planning information for the dynamic movement area by matching the real-time sensor data with the adaptive path rules.

[0023] Specifically, by collecting real-time sensor data, obstacle identification is first carried out. This process uses sensors (such as lidar, cameras, etc.) to detect obstacles in the surrounding environment and obtain the real-time fault object recognition result. Then, according to the real-time obstacle recognition result, the type of each obstacle is further determined. Obstacle types are usually divided into static obstacles (such as fixed objects like walls and columns) and dynamic obstacles (such as moving objects like pedestrians and vehicles). This discrimination helps to decide whether dynamic path adjustment is needed or how to plan the path. Then, the historical sensing data of the same area is obtained to analyze the past behavior of obstacles. These historical data help to infer the change characteristics such as the movement trajectory, speed, and direction of obstacles, so as to better predict the future behavior of dynamic obstacles.

[0024] Based on the current obstacle recognition result and the movement change characteristics of dynamic obstacles, pre-path planning is carried out. For dynamic obstacles, their movement area is predicted through historical data, and adaptive path rules are inserted into this area. These rules can adjust the path in real time according to the dynamic characteristics of the obstacles to ensure more flexible obstacle avoidance operations. During the path execution process, when the device (such as a robot, an autonomous vehicle, etc.) reaches the movement area of the dynamic obstacle, the preset adaptive path rules are activated, and new path planning information is generated by matching the real-time sensor data with these rules to dynamically adjust the current path to avoid obstacles.

[0025] This method effectively handles obstacles in a dynamic environment through real-time obstacle recognition, historical data analysis, and dynamic path pre-planning, combined with adaptive path rules. In this way, the system can achieve flexible and real-time path adjustment, ensure collision avoidance in a dynamically changing environment, and improve the accuracy and efficiency of path planning; this method is applicable to fields such as autonomous driving and robot navigation and has high application value in complex environments.

[0026] Furthermore, the present invention further includes the following steps:

[0027] Obtain multi-source sensor data, perform identification feature analysis on the multi-source sensor data, and determine the identification types of each data source; based on the identification types of each data source, perform fault object identification on the multi-source sensor data to obtain corresponding fault object identification type features; integrate the corresponding fault object identification type features of the identification types of each data source to obtain the current fault object identification result.

[0028] Specifically, collect data from multiple sensors, which may include lidar, cameras, infrared sensors, ultrasonic sensors, etc. Each sensor provides different types of information, such as distance, images, temperature, speed, etc. These data sources together provide a comprehensive perception of the environment. Perform identification feature analysis on the collected multi-source sensor data. The data of each sensor may have different characteristics and resolution capabilities. Through feature analysis, the identification type of each data source can be determined. For example, lidar may mainly identify the distance information of objects, cameras provide image recognition features, and ultrasonic sensors provide short-range detection, etc.

[0029] Based on the identification type of each data source, perform fault object identification. For example, use image processing methods to identify static obstacles, use lidar data to identify the position and shape of obstacles, or use temperature sensors to detect whether the equipment is overheated. Each data source classifies and extracts feature of the identification result of obstacles or fault objects according to its identification type. Integrate the identification type features from different data sources. This step requires fusing the features extracted by each data source under its identification type to obtain a comprehensive fault object identification result. This integration usually relies on multi-sensor data fusion algorithms, such as weighted average, Kalman filtering, or deep learning methods, to make full use of the advantages of each data source. By integrating the identification type features of each data source, the current fault object identification result is finally generated. This result contains obstacle information extracted from multiple data sources, can provide a more comprehensive and accurate fault object identification, and enhance the system's understanding and response ability to the environment.

[0030] This method realizes the accurate identification of fault objects through the feature analysis and integration of multi-source sensor data. Each sensor performs different types of fault object identification according to its characteristics, and then fuses their respective identification results to obtain a more comprehensive and accurate fault object identification result. This method can improve the accuracy of identification in a dynamic environment, especially suitable for fields such as autonomous driving, robot navigation, and industrial inspection, and helps to better identify and handle obstacles and fault objects in the environment.

[0031] Furthermore, the present invention further includes the following steps:

[0032] Perform a fault object recognition evaluation on the multi-source sensor data to obtain a fault object recognition evaluation result; establish a mapping relationship between the fault object recognition evaluation result and each data source; based on the mapping relationship, obtain the recognition types of the respective data sources according to the fault object recognition evaluation result, where the recognition type is the recognition dimension corresponding to the feature recognition evaluation value being higher than the threshold.

[0033] Specifically, after performing a fault object recognition on the collected multi-source sensor data, an evaluation is carried out. This evaluation process is based on the recognition results of each data source, taking into account factors such as recognition accuracy, stability, and reliability. For example, lidar may be more accurate in distance measurement, while image recognition may be more precise in analyzing appearance features. By quantitatively evaluating these results, a fault object recognition evaluation result for each data source is obtained. Then, according to the fault object recognition evaluation results of different data sources, a mapping relationship is established. This mapping relationship helps to associate the fault object recognition evaluation result with the data source characteristics by analyzing the recognition capabilities of each data source. For example, some data sources (such as cameras) may have a higher evaluation value for dynamic obstacle recognition, while other data sources (such as ultrasonic sensors) may have a higher evaluation value for static obstacle recognition.

[0034] Based on the established mapping relationship, for the fault object recognition evaluation result of each data source, obtain its corresponding recognition type. Specifically, if the fault object recognition evaluation value of a certain data source is higher than a preset threshold, then the recognition dimension corresponding to this data source is used as the recognition type of this data source. The recognition type indicates the fault object characteristics or types that each data source is best at or most accurately recognizes. The recognition type of each data source will be associated with its corresponding fault object recognition evaluation value and threshold. In this way, the most suitable recognition task can be assigned to each data source to ensure efficient and accurate fault object recognition. For example, if the recognition evaluation value of a certain data source is high and exceeds the threshold, then this data source can be used as the main data source for fault object recognition; conversely, it may be necessary to combine the results of other data sources or adopt supplementary means for optimization.

[0035] This method dynamically determines the recognition type according to the recognition capabilities of different data sources through fault object recognition evaluation and the establishment of a data source mapping relationship, optimizing the application of multi-source sensor data in fault object recognition; by mapping the evaluation results of each data source to the recognition type, a more efficient and accurate fault object recognition process can be provided for the system, ensuring the pertinence and accuracy of recognition; this method has strong application prospects in scenarios of multi-sensor fusion, especially in fields such as autonomous driving, industrial automation, and intelligent monitoring.

[0036] Furthermore, the present invention further includes the following steps:

[0037] The recognition dimensions include: appearance features, distance, speed, size, and penetrability features.

[0038] Specifically, the recognition dimensions include appearance features, distance, speed, size, and penetrability features. For these dimensions, the faulty objects can be recognized and analyzed in more detail. Among them, the features such as the shape, color, and texture of the object are usually captured by a camera or a vision sensor, which helps to distinguish different types of obstacles (such as static obstacles and dynamic obstacles); a lidar (LiDAR) or an ultrasonic sensor can measure the distance between the object and the sensor, thereby helping the system to determine the position and size of the obstacle; by continuously analyzing the data of a radar, a lidar, or a vision sensor, the moving speed of the object is calculated, which is crucial for the tracking and prediction of dynamic obstacles; the size information is usually obtained by combining and analyzing multi-source data. For example, through the data fusion of a vision sensor and a lidar, the size of the object can be judged more accurately; it usually involves the material or light transmissibility features of the object. By using special sensors such as an infrared sensor or an X-ray imaging technology, the penetrability of the object can be analyzed to help distinguish different types of obstacles (for example, a wall and a penetrable object). Through the comprehensive analysis of these dimensions, the system can more accurately identify and classify the faulty objects, and optimize the path planning or other task executions according to the characteristics of each dimension.

[0039] Furthermore, the present invention further includes the following steps:

[0040] According to the current faulty object recognition result and the obstacle movement change features, divide the faulty object distribution area to obtain a static faulty object area, a regularly moving faulty object area, and a dynamically moving faulty object area; obtain the faulty object features of the static faulty object area and the regularly moving faulty object area, and the path distribution features of the dynamically moving faulty object area; according to the faulty object features of the static faulty object area and the regularly moving faulty object area and the path distribution features of the dynamically moving faulty object area, perform preliminary path planning to obtain a pre-planned path; set the dynamically moving faulty object area as the dynamic moving area, and perform dynamic classification according to the path distribution features and historical change records of the dynamic moving area to construct the adaptive path rule, and insert the adaptive path rule into the dynamic moving area.

[0041] Specifically, according to the current identification results of faulty objects and the movement change characteristics of obstacles, the distribution areas of faulty objects in the environment are divided to obtain static faulty object areas (these areas contain obstacles with fixed positions), regular moving faulty object areas (these areas contain obstacles that move according to specific rules, usually these obstacles follow specific laws or periodic changes), and dynamic moving faulty object areas (these areas contain obstacles that change in real time, and their positions and movement trajectories have a high degree of randomness). Then, the faulty object characteristics of the static faulty object areas and regular moving faulty object areas are obtained, such as size, position, shape, etc.; the path distribution characteristics of the dynamic moving faulty object areas, that is, the movement trajectories and change rules of the obstacles.

[0042] According to the faulty object characteristics of the static faulty object areas and regular moving faulty object areas, and the path distribution characteristics of the dynamic moving faulty object areas, preliminary path planning is carried out. Through the analysis of the characteristics of these areas, a pre-planned path can be generated to avoid static obstacles and consider the influence of known regularly moving obstacles. The dynamic moving faulty object areas are set as dynamic moving areas. According to the path distribution characteristics and historical change records of the obstacles in this area, the dynamic obstacles are dynamically classified. This means that the system will analyze the movement patterns of the obstacles (such as speed, acceleration, etc.) and use historical data to predict their future behaviors. Based on the dynamic classification results, adaptive path rules are constructed. The adaptive path rules can adapt to the changes of dynamic obstacles in real time and adjust the path planning according to new sensor data to ensure that the path can avoid new dynamic obstacles. The constructed adaptive path rules are inserted into the dynamic moving areas. When the system runs, the path planning of the dynamic obstacle areas will be controlled by these rules, and new path planning that adapts to dynamic obstacles will be generated according to the changes of real-time sensor data.

[0043] By classifying the obstacles in the environment into three categories: static, regular moving, and dynamic moving, and comprehensively considering the characteristics and dynamic changes of different types of obstacles, this method designs a pre-planned path. At the same time, for dynamic obstacles, adaptive path rules are used for real-time adjustment to ensure the flexibility and accuracy of path planning; this method improves the response ability of the path planning system in dynamic and complex environments and is widely applicable to fields such as autonomous driving, robot navigation, and drone flight, ensuring the effectiveness, real-time performance, and safety of path planning.

[0044] Furthermore, the present invention further includes the following steps:

[0045] Cluster according to the historical change records to obtain mobile change categories; analyze the path planning characteristics of each change category according to the mobile change categories to obtain dynamic interference characteristics and path constraint conditions; configure the adaptive path rules of the dynamic mobile area according to the dynamic interference characteristics and path constraint conditions of the mobile change categories.

[0046] Specifically, according to the collected historical change records, perform a clustering analysis on the motion changes of dynamic obstacles. This clustering analysis is based on parameters such as the motion trajectory, speed, and acceleration of the obstacles within a certain period of time. Through clustering, the motion behaviors of the obstacles can be divided into different categories (such as regular movement, random change, periodic change, etc.). The division of these categories helps to understand the behavior patterns of dynamic obstacles. According to the obtained mobile change categories, perform a path planning characteristic analysis on each category of dynamic obstacles. By analyzing the characteristics of each category, the specific impacts of the obstacles on path planning can be identified.

[0047] Based on the path planning characteristic analysis, extract the dynamic interference characteristics and path constraint conditions of each category. The dynamic interference characteristics describe the impacts of the changes of the obstacles on the path (such as sudden acceleration or sharp turning), and the path constraint conditions describe the path restrictions that need to be observed due to the movement of the obstacles (such as safety distance, speed limit, passable channels, etc.). According to the interference characteristics and path constraint conditions of various dynamic obstacles, configure the adaptive path rules for the dynamic mobile area. These rules can help the system adjust the path planning in real time to adapt to the changes of dynamic obstacles. For example, based on real-time sensor data and the changes of the obstacles, adjust the planned path to avoid interference; adjust the speed of the vehicle or robot according to the movement changes of the obstacles to ensure a safe avoidance distance; for irregularly moving or randomly changing obstacles, design multiple alternative paths and select the most suitable path according to real-time data.

[0048] This method designs flexible adaptive path rules for the obstacles in the dynamic mobile area through clustering analysis and path planning characteristic analysis based on historical change records. These rules can respond to the interference caused by dynamic obstacles in real time, optimize the path planning, and ensure the real-time performance, safety, and effectiveness of the system in a complex environment. It is particularly suitable for scenarios such as autonomous driving, robot navigation, and intelligent transportation that need to consider dynamic obstacles, and can significantly improve the adaptability and accuracy of path planning.

[0049] Furthermore, the present invention further includes the following steps:

[0050] Determine the characteristics of the static fault objects based on the occlusion edge size of the static fault object area; determine the characteristics of the regularly moving fault objects in the regularly moving fault object area according to the moving area, frequency, and speed of the fault objects in the regularly moving fault object area; extract the path distribution characteristics based on the historical path of the dynamically moving fault object area.

[0051] Specifically, determine the characteristics of the static fault objects based on the occlusion edge size of the static fault object area. Static obstacles usually do not move, so their characteristics are mainly reflected in aspects such as shape and size. By calculating the occlusion edge size of the static fault object area (i.e., the perimeter length or area of the outer edge of the obstacle), the geometric characteristics of the static obstacle can be determined. These characteristics help the system understand the space occupied by the static obstacle and the possible impact on path planning. Through the size of the occlusion edge, the influence range of the obstacle and its relationship with other obstacles can be further evaluated.

[0052] Determine the characteristics of the regularly moving fault objects in the regularly moving fault object area according to the moving area, frequency, and speed of the fault objects in the regularly moving fault object area. Regularly moving obstacles move according to certain rules or paths and usually have relatively regular movement trajectories. By obtaining the moving areas of these obstacles, their activity ranges and possible interference areas can be understood. In addition to determining the moving area, the movement frequency and speed of the regularly moving obstacles also need to be considered. The frequency refers to the movement cycle of the obstacle within a certain period of time, and the speed describes the movement rate of the obstacle. This information is important for predicting the future movement trajectory of the obstacle and dynamic path planning.

[0053] Extract the path distribution characteristics based on the historical path of the dynamically moving fault object area. Dynamically moving obstacles usually have relatively irregular movement patterns and may be affected by changes in the external environment. Therefore, it is necessary to extract their movement characteristics through the historical path. By analyzing the historical movement trajectories of dynamically moving obstacles, characteristics such as their movement patterns, directions, and accelerations can be identified. The historical path of dynamically moving obstacles not only helps predict future paths but also can identify their possible movement trends (such as whether there are repeated paths, whether there is a trend of acceleration or deceleration, etc.). For the dynamically moving fault object area, through further analysis of its historical path, the path distribution characteristics can be extracted, that is, the spatial area passed by the obstacle within a specific time period. The path distribution characteristics can help the system predict the possible future path range of the obstacle and provide a basis for dynamic path planning.

[0054] By extracting the features of static, regularly moving, and dynamically moving fault object areas, this method can more accurately describe the behaviors and impacts of different types of obstacles. Specifically, by analyzing the occlusion edge size of static fault objects, the moving areas, frequencies, and speeds of regularly moving obstacles, as well as the historical paths and path distribution characteristics of dynamic obstacles, the system can provide rich dynamic information for path planning. This information can help the system make more flexible and accurate path planning in complex environments, improving the system's obstacle avoidance ability and real-time response ability.

[0055] In summary, a dynamic obstacle avoidance path planning method based on obstacle detection provided by the present invention has the following technical effects:

[0056] By performing obstacle recognition on the collected real-time sensor data, the current fault object recognition result is obtained; according to the current fault object recognition result, the fault object type is discriminated, and the fault object type is marked, where the fault object type includes static and dynamic; the historical sensing data in the same area is obtained, and the historical sensing data is used to perform matching analysis on the current fault object recognition result to obtain the obstacle movement change characteristics; according to the current fault object recognition result and the obstacle movement change characteristics, path pre-planning is performed, and the moving area of the dynamic obstacle is located, and an adaptive path rule is inserted into the dynamic moving area; the path is executed according to the pre-planned path. When reaching the dynamic moving area, the adaptive path rule is activated, and the adaptive path planning information of the dynamic moving area is generated by matching the real-time sensor data with the adaptive path rule. That is to say, through the adaptive path rule and the moving area positioning of the dynamic obstacle, the path execution error caused by the dynamic change of the obstacle is avoided, and the flexibility and accuracy of path planning are significantly improved.

[0057] Embodiment 2. Based on the same inventive concept as the dynamic obstacle avoidance path planning method based on obstacle detection in the foregoing embodiment, the present invention also provides a dynamic obstacle avoidance path planning system based on obstacle detection. Please refer to the attached Figure 2, including: an obstacle recognition module 11 for recognizing obstacles from the collected real-time sensor data to obtain the current faulty object recognition result; a faulty object type discrimination module 12 for discriminating the type of the faulty object according to the current faulty object recognition result and marking the faulty object type, where the faulty object types include static and dynamic; an obstacle movement change feature obtaining module 13 for obtaining historical sensing data in the same area, using the historical sensing data to perform matching analysis on the current faulty object recognition result to obtain the obstacle movement change feature; a moving area positioning module 14 for performing path pre-planning according to the current faulty object recognition result and the obstacle movement change feature, and performing moving area positioning on the dynamic obstacle, and inserting an adaptive path rule into the dynamic moving area; an adaptive path planning information generation module 15 for performing path execution according to the pre-planned path, when reaching the dynamic moving area, activating the adaptive path rule, and generating adaptive path planning information for the dynamic moving area by matching the real-time sensor data with the adaptive path rule.

[0058] Furthermore, the dynamic obstacle avoidance path planning system based on obstacle detection is also used for: obtaining multi-source sensor data, performing recognition feature analysis on the multi-source sensor data to determine the recognition types of each data source; based on the recognition types of each data source, performing faulty object recognition on the multi-source sensor data to obtain the corresponding faulty object recognition type features; integrating the corresponding faulty object recognition type features of the recognition types of each data source to obtain the current faulty object recognition result.

[0059] Furthermore, the dynamic obstacle avoidance path planning system based on obstacle detection is also used for: performing faulty object recognition evaluation on the multi-source sensor data to obtain a faulty object recognition evaluation result; establishing a mapping relationship between the faulty object recognition evaluation result and each data source; based on the mapping relationship, obtaining the recognition types of each data source according to the faulty object recognition evaluation result, where the recognition type is the recognition dimension corresponding to the feature recognition evaluation value higher than the threshold.

[0060] Furthermore, the dynamic obstacle avoidance path planning system based on obstacle detection is also used for: the recognition dimensions include: appearance features, distance, speed, size, penetrability features.

[0061] Further, the dynamic obstacle avoidance path planning system based on obstacle detection is further configured to: divide the distribution area of the faulty objects according to the current faulty object recognition result and the characteristics of the movement and change of the obstacles, so as to obtain a static faulty object area, a regularly moving faulty object area, and a dynamically moving faulty object area; obtain the faulty object characteristics of the static faulty object area and the regularly moving faulty object area, and the path distribution characteristics of the dynamically moving faulty object area; perform preliminary path planning according to the faulty object characteristics of the static faulty object area and the regularly moving faulty object area and the path distribution characteristics of the dynamically moving faulty object area, so as to obtain a pre-planned path; set the dynamically moving faulty object area as the dynamic moving area, and perform dynamic classification according to the path distribution characteristics and historical change records of the dynamic moving area, construct the adaptive path rule, and insert the adaptive path rule into the dynamic moving area.

[0062] Further, the dynamic obstacle avoidance path planning system based on obstacle detection is further configured to: perform clustering according to the historical change records to obtain movement change categories; perform path planning feature analysis on each change category according to the movement change categories to obtain dynamic interference characteristics and path constraint conditions; configure the adaptive path rule of the dynamic moving area according to the dynamic interference characteristics and path constraint conditions of the movement change categories.

[0063] Further, the dynamic obstacle avoidance path planning system based on obstacle detection is further configured to: determine the faulty object characteristics of the static faulty object area according to the size of the occlusion edge of the faulty objects in the static faulty object area; determine the faulty object characteristics of the regularly moving faulty object area according to the moving area, frequency, and speed of the faulty objects in the regularly moving faulty object area; extract the path distribution characteristics according to the historical paths of the dynamically moving faulty object area.

[0064] The various embodiments in this specification are described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The dynamic obstacle avoidance path planning method and specific examples in the foregoing Embodiment 1 are equally applicable to the dynamic obstacle avoidance path planning system in this embodiment. Through the foregoing detailed description of the dynamic obstacle avoidance path planning method, those skilled in the art can clearly know the dynamic obstacle avoidance path planning system in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated herein. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0065] The foregoing description of the disclosed embodiments enables those skilled in the art to practice or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Thus, the invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0066] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A dynamic obstacle avoidance path planning method based on obstacle detection, characterized in that: include: Perform obstacle identification on the collected real-time sensor data to obtain the current fault object identification result; According to the current fault object identification result, the fault object type is determined and the fault object type is marked, where the fault object type includes static and dynamic; Acquire historical sensor data of the same area, use the historical sensor data to match and analyze the current fault object identification result, and obtain the obstacle movement change characteristics; Pre-planning the path according to the current fault object identification result and the obstacle movement change characteristics, positioning the moving area of ​​the dynamic obstacle, and inserting the adaptive path rule in the dynamic moving area; The path is executed according to the pre-planned path. When the dynamic moving area is reached, the adaptive path rule is activated, and adaptive path planning information of the dynamic moving area is generated by matching the real-time sensor data with the adaptive path rule.

2. The dynamic obstacle avoidance path planning method based on obstacle detection according to claim 1, characterized in that: The obstacle identification is performed on the collected real-time sensor data to obtain the current fault object identification result, including: Acquire multi-source sensor data, perform identification feature analysis on the multi-source sensor data, and determine the identification type of each data source; Based on the identification type of each data source, fault object identification is performed on the multi-source sensor data to obtain corresponding fault object identification type features; The fault object identification type features corresponding to the identification types of the data sources are integrated to obtain the current fault object identification result.

3. The dynamic obstacle avoidance path planning method based on obstacle detection according to claim 2 is characterized in that: Performing identification feature analysis on the multi-source sensor data to determine the identification type of each data source includes: Performing fault object identification and evaluation on the multi-source sensor data to obtain a fault object identification and evaluation result; Establishing a mapping relationship between the fault object identification and evaluation result and each data source; Based on the mapping relationship and according to the fault object identification evaluation result, the identification type of each data source is obtained, and the identification type is the identification dimension corresponding to the feature identification evaluation value being higher than the threshold.

4. The dynamic obstacle avoidance path planning method based on obstacle detection according to claim 3 is characterized in that: Identification dimensions include: appearance characteristics, distance, speed, size, and penetration characteristics.

5. The dynamic obstacle avoidance path planning method based on obstacle detection according to claim 1, characterized in that: According to the current fault object identification result and the obstacle movement change characteristics, path pre-planning is performed, and the moving area of ​​the dynamic obstacle is located, and an adaptive path rule is inserted in the dynamic moving area, including: According to the current fault object identification result and the obstacle movement change characteristics, the fault object distribution area is divided to obtain a static fault object area, a regularly moving fault object area, and a dynamically moving fault object area; Obtaining fault object characteristics of the static fault object area, the regularly moving fault object area, and the path distribution characteristics of the dynamically moving fault object area; Perform preliminary path planning according to the fault object characteristics of the static fault object area, the regularly moving fault object area, and the path distribution characteristics of the dynamically moving fault object area to obtain a pre-planned path; The dynamic moving fault area is set as the dynamic moving area, and dynamic classification is performed according to the path distribution characteristics and historical change records of the dynamic moving area, the adaptive path rule is constructed, and the adaptive path rule is inserted into the dynamic moving area.

6. The dynamic obstacle avoidance path planning method based on obstacle detection according to claim 5, characterized in that: Dynamic classification is performed according to the path distribution characteristics and historical change records of the dynamic movement area to construct the adaptive path rule, including: Clustering is performed according to the historical change records to obtain movement change categories; Performing path planning feature analysis on each change category according to the mobile change category to obtain dynamic interference features and path constraint conditions; According to the dynamic interference characteristics and path constraints of the movement change category, an adaptive path rule of the dynamic movement area is configured.

7. The dynamic obstacle avoidance path planning method based on obstacle detection according to claim 6, characterized in that: Obtaining the fault object features of the static fault object area, the regularly moving fault object area, and the path distribution features of the dynamically moving fault object area includes: Determining the fault object characteristics of the static fault object area according to the fault object occlusion edge size of the static fault object area; Determine the fault object characteristics of the regular moving fault object area according to the fault object moving area, frequency, and speed of the regular moving fault object area; According to the historical path of the dynamically moving fault object area, the path distribution feature is extracted.

8. Dynamic obstacle avoidance path planning system based on obstacle detection, characterized in that: The steps for implementing the dynamic obstacle avoidance path planning method based on obstacle detection according to any one of claims 1 to 7 include: The obstacle recognition module is used to identify obstacles based on the collected real-time sensor data and obtain the current fault object recognition result; A fault object type identification module, used to identify the fault object type according to the current fault object identification result and mark the fault object type, wherein the fault object type includes static and dynamic; The obstacle movement change feature acquisition module is used to obtain the historical sensor data of the same area, and use the historical sensor data to match and analyze the current fault object identification result to obtain the obstacle movement change feature; A mobile area positioning module, used to perform path pre-planning according to the current fault object identification result and obstacle movement change characteristics, and to perform mobile area positioning for dynamic obstacles, and to insert adaptive path rules in the dynamic movement area; The adaptive path planning information generation module is used to execute the path according to the pre-planned path. When reaching the dynamic movement area, the adaptive path rule is activated, and the adaptive path planning information of the dynamic movement area is generated by matching the real-time sensor data with the adaptive path rule.

Images