Marine surveying and mapping system based on deepwater exploration robot intelligent navigation

Abstract

The invention discloses a marine surveying and mapping system based on deepwater exploration robot intelligent navigation, and belongs to the technical field of marine surveying and mapping. Comprising a data acquisition module for synchronously acquiring navigation positioning data and submarine topography data through multi-source navigation equipment carried by a deepwater detection robot and submarine measurement equipment; the navigation fusion module is used for realizing self-adaptive fusion of multi-source data and outputting initial pose estimation; the quality control module is used for carrying out quality evaluation and abnormity identification on the real-time surveying and mapping data; the path planning module is used for planning an optimal surveying and mapping path in real time and carrying out online adjustment according to the re-measurement requirement of the quality control module; the fault-tolerant control module is used for realizing multi-level navigation source fault diagnosis, starting a degraded navigation strategy when navigation is abnormal or fails, and ensuring that basic navigation and surveying and mapping functions can still be maintained; and the topographic surveying and mapping module is used for compensating and splicing submarine topographic data in real time by using the pose information provided by the navigation fusion module.

Description

Technical Field

[0001] The present application relates to the field of ocean surveying and mapping technology, and more specifically, to an ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot. Background Art

[0002] Marine surveying and mapping, as an important means of obtaining marine geographic information, is widely used in the fields of seabed resource exploration, marine environmental monitoring, and waterway planning. With the continuous development of marine scientific research and engineering technology, the demand for surveying and mapping in deepwater areas has gradually increased. Traditional marine surveying and mapping methods mainly rely on manual operation or simple automated equipment, facing problems such as low surveying and mapping efficiency, insufficient accuracy, and poor adaptability. Especially in deepwater areas, due to the complex seabed terrain and diverse environmental changes, navigation and positioning are difficult and data quality is unstable.

[0003] In order to solve these problems, deep-water exploration robot technology has gradually developed in recent years. By integrating a variety of sensors and navigation equipment, deep-water exploration robots can perform autonomous navigation and mapping in complex seabed environments. However, since deep-water exploration robots need to face challenges such as real-time changes in the ocean environment, sensor errors, and data inconsistency when performing tasks, how to achieve high-precision and high-efficiency deep-water mapping is still a technical problem that needs to be solved urgently.

[0004] At present, although the existing marine surveying and mapping systems can provide automated surveying and mapping functions to a certain extent, there are still problems such as low accuracy of navigation source data fusion, lack of flexibility in path planning algorithms, and insufficient surveying and mapping quality control. Especially in deep-water exploration environments, the system has weak tolerance for navigation source failure, seabed terrain changes, and environmental interference, resulting in reduced accuracy during the surveying and mapping process and even mission failure.

[0005] In summary, how to achieve high-precision and high-reliability navigation and mapping in complex deep-water environments, how to effectively integrate multi-source navigation data, optimize path planning and ensure the fault tolerance of the system have become technical problems that need to be solved urgently. Summary of the invention

[0006] In order to overcome a series of defects in the prior art, the purpose of this application is to provide an ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot, including the following modules:

[0007] The data acquisition module collects navigation positioning data and seabed topography data synchronously through the multi-source navigation equipment and seabed measurement equipment carried by the deep-water exploration robot;

[0008] Navigation fusion module, which realizes adaptive fusion of multi-source data and outputs initial pose estimation;

[0009] The quality control module conducts quality assessment and anomaly identification on real-time surveying and mapping data, and generates local re-survey requirements based on the assessment results to ensure that the surveying and mapping accuracy meets the requirements;

[0010] The path planning module plans the optimal mapping path in real time based on the robot's current position, task requirements, and local terrain features, and makes online adjustments based on the re-measurement requirements of the quality control module;

[0011] Fault-tolerant control module, which implements multi-level navigation source fault diagnosis and initiates degraded navigation strategy when navigation is abnormal or fails, ensuring that basic navigation and mapping functions can still be maintained;

[0012] The terrain mapping module uses the posture information provided by the navigation fusion module to perform real-time compensation and splicing of seabed terrain data.

[0013] Furthermore, the data acquisition module includes the following components:

[0014] The navigation and positioning unit collects the original navigation data of the deep-water exploration robot through integrated GPS, inertial navigation system and sonar positioning equipment;

[0015] The seafloor topography detection unit uses multi-beam sonar and side-scan sonar technology to collect seafloor topography data in real time, including the depth, shape and characteristics of the seabed;

[0016] The environmental perception unit obtains marine environmental parameters through environmental sensors and provides real-time data support for environmental changes during surveying and mapping;

[0017] The posture monitoring unit uses a gyroscope and an accelerometer to collect the robot's posture data in real time;

[0018] The data preprocessing unit performs preliminary processing and formatting on the collected raw data to provide standardized data input for subsequent modules.

[0019] Furthermore, the navigation fusion module includes the following components:

[0020] The credibility assessment unit, based on a deep learning algorithm, evaluates the data reliability of each navigation source in real time and outputs a credibility score for each navigation source;

[0021] The weight adjustment unit dynamically adjusts the weight of each navigation source in the fusion calculation according to the credibility score;

[0022] The error modeling unit uses an improved Kalman filter algorithm to model the error characteristics of each navigation source;

[0023] The data synchronization and fusion unit synchronizes and integrates navigation data from different sources, and calculates the preliminary pose estimation results based on the dynamic weight and error model;

[0024] The pose output unit outputs the fused preliminary pose estimation results to provide high-precision navigation information for subsequent path planning and surveying compensation.

[0025] Furthermore, the credibility score of the navigation source is expressed as: in, is the credibility score of the i-th navigation source, with a value between [0,1], indicating the reliability of the navigation source; W i is the weight matrix corresponding to the i-th navigation source; x i is the feature vector of the i-th navigation source; b i is the bias term;

[0026] The weight of each navigation source in the fusion calculation is dynamically adjusted by the following formula: w i is the weight of the i-th navigation source and its corresponding credibility score is proportional to; It is the sum of the credibility scores of all navigation sources. It is normalized to ensure that the sum of all weights is 1.

[0027] Furthermore, the preliminary pose estimation result is expressed as: in, represents the preliminary pose estimation result after fusion; p i (t i ) represents the i-th navigation source at time t i Navigation data at the moment; is the i-th navigation source at t i The rate of change at a moment; t i Indicates the timestamp of the i-th navigation source; t sync Indicates the target synchronization time; e i (t sync ) represents the error model of the t-th navigation source at synchronization time t sync The error estimate under i Represents the error model of the i-th navigation source.

[0028] Furthermore, the path planning module includes the following components:

[0029] The task requirement analysis unit analyzes the task requirements according to the objectives and constraints of the surveying and mapping task, and provides task-oriented constraints for path planning;

[0030] Environmental constraint analysis unit, which analyzes local terrain features based on real-time seabed terrain data and combines environmental perception data to provide environmental constraint information for path planning;

[0031] The path generation unit uses a reinforcement learning algorithm to learn and generate the optimal mapping path in real time based on the current posture, task requirements and terrain characteristics;

[0032] The path optimization unit dynamically adjusts the current path planning according to the re-survey requirements or feedback information generated by the quality control module to ensure that the path meets the surveying and mapping accuracy and task requirements;

[0033] The safety monitoring unit adjusts the path in real time based on terrain analysis results and real-time environmental perception to avoid collisions or entering dangerous areas, ensuring the safety of the robot;

[0034] The path output unit converts the optimized path into a control instruction sequence to control the robot execution.

[0035] Furthermore, the optimal mapping path is obtained through the reinforcement learning strategy π * (s T ) is generated step by step, and is expressed as: ,

[0036] Among them, a T is the action at the current time T, that is, the path planning decision selected by the robot; w1, w2, and w3 are the weight coefficients of task requirements, terrain adaptability, and progress optimization, respectively; s T represents the state of the robot at time T; R task (s T , a T ) means in state s T Next, take action a T The subsequent task reward is used to evaluate the completion of the current task requirements; R terrain (s T , a T ) means in state s T Next, take action a T The terrain adaptability reward after is used to evaluate the adaptability of the current path planning on the seabed terrain; R progress (s T , a T ) means in state s T Next, take action a T The progress reward after is used to evaluate the contribution of the current path planning to the task progress; γ is the discount factor, which is used to balance the relationship between the current reward and the future reward; V(s T+1 ) in state s T+1 The value function of the robot after taking the optimal strategy is expressed as T+1 The maximum long-term reward that can be obtained by setting out.

[0037] Furthermore, the fault-tolerant control module includes the following components:

[0038] The navigation status monitoring unit monitors the stability, accuracy and consistency of each navigation source data in real time and captures potential abnormal signals or failure signs;

[0039] The fault diagnosis unit analyzes the type and severity of navigation source anomalies based on a multi-level diagnostic algorithm to determine the cause of the fault and the scope of impact;

[0040] Redundant strategy management unit, which manages backup navigation sources and redundant strategy libraries, and quickly switches to the suboptimal navigation source to maintain navigation function when the primary navigation source fails;

[0041] Degraded navigation execution unit, in case of navigation source failure or abnormality, activates backup navigation scheme to maintain basic navigation function;

[0042] Abnormal data processing unit, which uniformly processes abnormal data from each module, including data elimination, abnormal correction and interference suppression;

[0043] The Navigation Performance Evaluation Unit evaluates the performance in the degraded navigation mode in real time to ensure that it meets the minimum mission requirements and attempts to restore the normal navigation mode when conditions permit.

[0044] Furthermore, the terrain mapping module includes the following components:

[0045] The terrain data compensation unit compensates the collected seabed terrain data in real time according to the current posture information of the robot, and corrects the data deviation caused by positioning error or sensor deviation;

[0046] The data splicing unit splices the seabed topographic data collected at different locations and time periods to form a continuous topographic map, ensuring seamless connection between data;

[0047] SLAM algorithm implementation unit, which optimizes the relationship between the robot's position and terrain data in real time based on the SLAM algorithm, and establishes and updates an accurate map of the seabed terrain;

[0048] The closed-loop optimization unit optimizes the accumulated error of the robot's position through closed-loop detection and correction of the SLAM algorithm, ensuring continuous improvement of map and positioning accuracy;

[0049] The terrain feature analysis unit extracts and analyzes the features of the seabed terrain data and provides terrain feature information for other modules;

[0050] The accuracy assessment and output unit performs accuracy assessment on the spliced ​​terrain data, outputs the final seabed topographic map, and provides the positioning optimization results to other modules.

[0051] Furthermore, the data deviation caused by positioning error or sensor deviation is corrected by the following formula: Among them, D final is the corrected seafloor topography data; (x raw ,y raw ) is the coordinate of the original seabed topography data, which represents the initial coordinate of the seabed topography in the measured data without correction; e x , e y are positioning errors, representing the horizontal positioning error and vertical positioning error of the robot respectively; e θ is the posture error, indicating the robot's orientation error; θ is the robot's orientation; R(e θ ) is the rotation matrix, which is used to correct the posture error e θ The terrain data deviation caused by x ,s y is the sensor deviation, which represents the systematic deviation or error of the sensor itself in the horizontal and vertical directions respectively.

[0052] Compared with the prior art, this application has the following beneficial effects:

[0053] This application achieves efficient and accurate seabed topography mapping by integrating the multi-source navigation equipment carried by the deep-water exploration robot with the seabed measurement equipment. It also ensures the continuity and reliability of the mapping task through intelligent navigation, path planning and fault-tolerant control, while improving the accuracy and integrity of the mapping data. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a schematic diagram of the structure of an ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical scheme and advantages of the implementation of the present invention clearer, the technical scheme in the embodiment of the present invention will be described in more detail below in conjunction with the drawings in the embodiment of the present invention. In the drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described embodiments are part of the embodiments of the present invention, not all of the embodiments.

[0056] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0057] The embodiments and directional terms described below with reference to the accompanying drawings are exemplary and intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0058] like Figure 1As shown, an ocean surveying and mapping system based on intelligent navigation of a deep-water detection robot includes the following modules:

[0059] The data acquisition module collects navigation positioning data and seabed topography data synchronously through the multi-source navigation equipment and seabed measurement equipment carried by the deep-water exploration robot;

[0060] Navigation fusion module, which realizes adaptive fusion of multi-source data and outputs initial pose estimation;

[0061] The quality control module conducts quality assessment and anomaly identification on real-time surveying and mapping data, and generates local re-survey requirements based on the assessment results to ensure that the surveying and mapping accuracy meets the requirements;

[0062] The path planning module plans the optimal mapping path in real time based on the robot's current position, task requirements, and local terrain features, and makes online adjustments based on the re-measurement requirements of the quality control module;

[0063] Fault-tolerant control module, which implements multi-level navigation source fault diagnosis and initiates degraded navigation strategy when navigation is abnormal or fails, ensuring that basic navigation and mapping functions can still be maintained;

[0064] The terrain mapping module uses the posture information provided by the navigation fusion module to perform real-time compensation and splicing of seabed terrain data.

[0065] The data acquisition module is the basic part of the system, responsible for synchronously collecting data from multiple sensors on the deep-water exploration robot. The robot collects data in real time by carrying multi-source navigation equipment, such as inertial navigation system (INS), sonar positioning system (SSBL), depth meter, etc., as well as seabed measurement equipment, such as sonar imaging system, laser scanner, etc. These devices can provide important information including navigation and positioning data (such as the position and posture of the robot) and seabed terrain data (such as seabed surface morphology, obstacles, etc.). The parallel collection of multi-source data not only improves the integrity and accuracy of the data, but also effectively overcomes the errors and blind spots that may be caused by a single device, providing high-quality raw data for subsequent navigation and mapping.

[0066] The navigation fusion module is one of the core technologies in the system. Its task is to effectively fuse the navigation data from different sensors and output the initial pose estimation of the robot. The data from multiple sensors usually have certain errors. How to comprehensively utilize these data through intelligent algorithms to improve positioning accuracy is the key to this module. Commonly used methods include Kalman filtering and particle filtering. By dynamically adjusting the weights of different sensors, the positioning accuracy after fusion is guaranteed. This module can not only provide the robot's position information in real time, but also adaptively adjust the weights of the sensors according to different environmental conditions to ensure that the robot can still be accurately positioned in complex underwater environments, providing an accurate basis for subsequent path planning and terrain mapping.

[0067] The main function of the quality control module is to conduct quality assessment on the surveying and mapping data collected in real time and identify abnormal data. Due to the complex seabed surveying environment, affected by various factors such as water flow, robot motion state, sensor failure, etc., the collected data may contain errors or be incomplete. This module designs a data quality assessment algorithm to perform accuracy analysis and anomaly detection on each frame of data, and can promptly detect and mark potential erroneous data. At the same time, the module generates local re-survey requirements based on the quality assessment results. When it is found that the surveying and mapping accuracy of certain areas does not meet the standards, it will trigger the robot to re-survey these areas to ensure the accuracy of the final data. This technology effectively guarantees the overall surveying and mapping accuracy of the system, especially in deepwater environments, ensuring the reliability and safety of the mission.

[0068] The path planning module is responsible for planning the optimal surveying and mapping path in real time based on the robot's current position, task requirements, and local terrain features. The goal of this module is to ensure that the robot is surveying and mapping in a complex seabed environment using the optimal path, which can not only complete the task requirements, but also avoid obstacles and reduce surveying and mapping time. At the same time, the path planning module will also make dynamic adjustments based on the local re-survey requirements provided by the quality control module. For example, when there is a problem with the surveying and mapping accuracy of a certain area, the path planning module will automatically plan a re-survey path through the area to improve the accuracy and completeness of the data. The design of this module can greatly improve the robot's operating efficiency, reduce unnecessary path duplication, and improve the overall benefits of surveying and mapping operations.

[0069] The fault-tolerant control module is a key guarantee module of the system, which mainly realizes multi-level navigation source fault diagnosis and emergency response. When any navigation equipment fails or is abnormal, the fault-tolerant control module can detect in real time and start the corresponding fault emergency mechanism. For example, when the sonar positioning system fails, the module will automatically switch to the inertial navigation system and correct the position of the robot through a supplementary algorithm to maintain the continuity of navigation. The module uses a multi-level redundant design to enable the robot to maintain basic navigation and mapping functions even in complex environments, reduce the risk of mission interruption due to equipment failure, and ensure the high availability of the system.

[0070] The topographic mapping module uses the posture information provided by the navigation fusion module to perform real-time compensation and splicing of the seabed topographic data. Due to the complexity of the seabed environment, the robot may be affected by water flow or obstacles during movement, resulting in displacement or deviation during data collection. Therefore, after collecting the topographic data, the module will perform spatial compensation based on the real-time posture of the robot, so that the collected topographic data can accurately reflect the real situation of the seabed. At the same time, the splicing algorithm will seamlessly splice the measurement data from different locations to generate a complete seabed topographic map. The high efficiency of this module ensures the continuity and accuracy of the seabed topographic map, providing a reliable basis for subsequent data analysis and decision-making.

[0071] In general, the ocean mapping system based on the intelligent navigation of the deep-water exploration robot can achieve high-precision and high-efficiency seabed topography mapping in a complex seabed environment through the collaborative work of multiple technical modules. The data acquisition module provides rich raw data, the navigation fusion module ensures accurate positioning information, the quality control module improves the reliability of the data, the path planning module optimizes the efficiency of the mapping operation, the fault-tolerant control module ensures the stability of the system, and the topography mapping module ultimately generates an accurate seabed topographic map. The organic combination of these technical modules enables the system to demonstrate strong adaptability and reliability in deep-water exploration missions, which is of great significance to ocean mapping operations.

[0072] Furthermore, the data acquisition module includes the following components:

[0073] The navigation and positioning unit collects the original navigation data of the deep-water exploration robot through integrated GPS, inertial navigation system and sonar positioning equipment;

[0074] The seafloor topography detection unit uses multi-beam sonar and side-scan sonar technology to collect seafloor topography data in real time, including the depth, shape and characteristics of the seabed;

[0075] The environmental perception unit obtains marine environmental parameters through environmental sensors and provides real-time data support for environmental changes during surveying and mapping;

[0076] The posture monitoring unit uses a gyroscope and an accelerometer to collect the robot's posture data in real time;

[0077] The data preprocessing unit performs preliminary processing and formatting on the collected raw data to provide standardized data input for subsequent modules.

[0078] In summary, through the analysis of each component in the data acquisition module, it can be seen that the technical effect of each component plays a vital role in the intelligent navigation of the deep-water exploration robot. From the precise location information provided by the navigation and positioning unit, to the high-precision terrain data provided by the seabed terrain detection unit, to the real-time monitoring of the environmental perception unit and the attitude monitoring unit, all components work together to ensure that tasks can be performed efficiently and stably in a complex and changeable marine environment. The data preprocessing unit provides optimized data input for subsequent modules, ensuring the data fluency and accuracy of the entire system. The organic coordination of various components enables the data acquisition module to provide comprehensive and reliable data support for the deep-water exploration robot, ensuring the efficiency and accuracy of the system, and promoting the smooth completion of the seabed mapping task.

[0079] Furthermore, the navigation fusion module includes the following components:

[0080] The credibility assessment unit, based on a deep learning algorithm, evaluates the data reliability of each navigation source in real time and outputs a credibility score for each navigation source;

[0081] The weight adjustment unit dynamically adjusts the weight of each navigation source in the fusion calculation according to the credibility score;

[0082] The error modeling unit uses an improved Kalman filter algorithm to model the error characteristics of each navigation source;

[0083] The data synchronization and fusion unit synchronizes and integrates navigation data from different sources, and calculates the preliminary pose estimation results based on the dynamic weight and error model;

[0084] The pose output unit outputs the fused preliminary pose estimation results to provide high-precision navigation information for subsequent path planning and surveying compensation.

[0085] The navigation fusion module achieves efficient fusion of different navigation source data through the collaborative work of multiple key components such as credibility assessment, weight adjustment, error modeling, data synchronization fusion and posture output, providing accurate posture estimation for deep-water exploration robots. By dynamically adjusting the weights of each navigation source, evaluating its credibility in real time, and accurately modeling the error characteristics of the sensor, the module can maintain high-precision and high-reliability navigation in complex and changing marine environments. This module not only improves the robot's adaptability in complex seabed environments, but also provides strong data support for subsequent path planning and terrain mapping.

[0086] Furthermore, the credibility score of the navigation source is expressed as: in, is the credibility score of the i-th navigation source, with a value between [0,1], indicating the reliability of the navigation source; W i is the weight matrix corresponding to the i-th navigation source; x i is the feature vector of the i-th navigation source; b i is the bias term;

[0087] The weight of each navigation source in the fusion calculation is dynamically adjusted by the following formula: w i is the weight of the i-th navigation source and its corresponding credibility score is proportional to; It is the sum of the credibility scores of all navigation sources. It is normalized to ensure that the sum of all weights is 1.

[0088] The core idea of ​​the above formula is to dynamically evaluate the credibility of each navigation source and adjust its weight in the fusion calculation according to its reliability. This strategy based on credibility scoring and weight adjustment can flexibly respond to changes in navigation sources under different environments and conditions, ensuring that the fused pose estimation is more accurate and reliable. In this way, the navigation fusion module can maintain high-precision navigation in complex marine environments, greatly improving the overall operating efficiency and accuracy of the deep-water exploration robot.

[0089] Furthermore, the preliminary pose estimation result is expressed as: in, represents the preliminary pose estimation result after fusion; p i (t i ) represents the i-th navigation source at time t i Navigation data at the moment; is the i-th navigation source at t i The rate of change at a moment; t i Indicates the timestamp of the i-th navigation source; t sync Indicates the target synchronization time; ei (t sync ) represents the error model of the i-th navigation source at synchronization time t sync The error estimate under i Represents the error model of the i-th navigation source.

[0090] The calculation of the preliminary pose estimation results is achieved by fusing the data of multiple navigation sources and taking into account factors such as the time deviation, rate of change and error model of each navigation source. This method can provide high-precision pose estimation results for deep-water exploration robots in complex marine environments by dynamically adjusting the weights of each navigation source and correcting its errors. The estimation results provide reliable navigation support for subsequent path planning and seabed mapping tasks, improving the work efficiency and mapping accuracy of the entire system.

[0091] Furthermore, the path planning module includes the following components:

[0092] The task requirement analysis unit analyzes the task requirements according to the objectives and constraints of the surveying and mapping task, and provides task-oriented constraints for path planning;

[0093] Environmental constraint analysis unit, which analyzes local terrain features based on real-time seabed terrain data and combines environmental perception data to provide environmental constraint information for path planning;

[0094] The path generation unit uses a reinforcement learning algorithm to learn and generate the optimal mapping path in real time based on the current posture, task requirements and terrain characteristics;

[0095] The path optimization unit dynamically adjusts the current path planning according to the re-survey requirements or feedback information generated by the quality control module to ensure that the path meets the surveying and mapping accuracy and task requirements;

[0096] The safety monitoring unit adjusts the path in real time based on terrain analysis results and real-time environmental perception to avoid collisions or entering dangerous areas, ensuring the safety of the robot;

[0097] The path output unit converts the optimized path into a control instruction sequence to control the robot execution.

[0098] The path planning module ensures that the deep-water exploration robot can efficiently and safely perform mapping tasks in complex seabed environments through the collaboration of multiple components such as task requirement analysis, environmental constraint analysis, path generation, path optimization, safety monitoring and path output. The task requirement analysis unit provides task-oriented constraints for path planning, the environmental constraint analysis unit provides necessary environmental constraints in combination with terrain and environmental data, the path generation unit generates the optimal path using a reinforcement learning algorithm, the path optimization unit makes dynamic adjustments based on quality control feedback, the safety monitoring unit ensures the safety of the robot, and the path output unit ensures that the path is accurately executed. Through the coordinated work of these modules, the path planning module can maximize the robot's work efficiency and task completion while ensuring mapping accuracy and task requirements.

[0099] Furthermore, the optimal mapping path is obtained through the reinforcement learning strategy π * (s T ) is generated step by step, and is expressed as: ,

[0100] Among them, a T is the action at the current time T, that is, the path planning decision selected by the robot; w1, w2, and w3 are the weight coefficients of task requirements, terrain adaptability, and progress optimization, respectively; s T represents the state of the robot at time T; R task (s T , a T ) means in state s T Next, take action a T The subsequent task reward is used to evaluate the completion of the current task requirements; R terrain (s T , a T ) means in state s T Next, take action a T The terrain adaptability reward after is used to evaluate the adaptability of the current path planning on the seabed terrain; R progress (s T , a T ) means in state s T Next, take action a T The progress reward after is used to evaluate the contribution of the current path planning to the task progress; γ is the discount factor, which is used to balance the relationship between the current reward and the future reward; V(s T+1 ) in state s T+1 The value function of the robot after taking the optimal strategy is expressed as T+1 The maximum long-term reward that can be obtained by setting out.

[0101] The generation of the optimal mapping path is gradually realized through the strategy of reinforcement learning. In each state, the path selection is based on multiple factors such as task requirements, terrain adaptability and progress. By setting a reasonable reward function and weight coefficient, and combining the dynamic adjustment of the discount factor, the reinforcement learning algorithm can help the robot generate the optimal path that meets the task requirements. This method can not only handle complex seabed terrain environments, but also flexibly adjust the path according to real-time task requirements and progress requirements, ensuring that the mapping task can be completed efficiently and safely.

[0102] Furthermore, the fault-tolerant control module includes the following components:

[0103] The navigation status monitoring unit monitors the stability, accuracy and consistency of each navigation source data in real time and captures potential abnormal signals or failure signs;

[0104] The fault diagnosis unit analyzes the type and severity of navigation source anomalies based on a multi-level diagnostic algorithm to determine the cause of the fault and the scope of impact;

[0105] Redundant strategy management unit, which manages backup navigation sources and redundant strategy libraries, and quickly switches to the suboptimal navigation source to maintain navigation function when the primary navigation source fails;

[0106] Degraded navigation execution unit, in case of navigation source failure or abnormality, activates backup navigation scheme to maintain basic navigation function;

[0107] Abnormal data processing unit, which uniformly processes abnormal data from each module, including data elimination, abnormal correction and interference suppression;

[0108] The Navigation Performance Evaluation Unit evaluates the performance in the degraded navigation mode in real time to ensure that it meets the minimum mission requirements and attempts to restore the normal navigation mode when conditions permit.

[0109] The fault-tolerant control module ensures that the deep-water exploration robot can continue its mission and maintain basic functions when encountering navigation source failure or abnormality through multi-level monitoring, diagnosis, redundant switching and data processing. The navigation status monitoring unit captures abnormal signals of the navigation source in real time, the fault diagnosis unit provides accurate fault analysis, and the redundant strategy management unit and the degraded navigation execution unit ensure that the system can quickly switch to the backup navigation source and maintain mission execution. The abnormal data processing unit and the navigation performance evaluation unit ensure the validity of the data and the stability of the system performance. Through the collaboration of these components, the fault-tolerant control module enhances the robustness of the system, allowing the robot to always maintain efficient navigation capabilities in complex and changing environments, ensuring the smooth progress of the seabed mapping mission.

[0110] Furthermore, the terrain mapping module includes the following components:

[0111] The terrain data compensation unit compensates the collected seabed terrain data in real time according to the current posture information of the robot, and corrects the data deviation caused by positioning error or sensor deviation;

[0112] The data splicing unit splices the seabed topographic data collected at different locations and time periods to form a continuous topographic map, ensuring seamless connection between data;

[0113] SLAM algorithm implementation unit, which optimizes the relationship between the robot's position and terrain data in real time based on the SLAM algorithm, and establishes and updates an accurate map of the seabed terrain;

[0114] The closed-loop optimization unit optimizes the accumulated error of the robot's position through closed-loop detection and correction of the SLAM algorithm, ensuring continuous improvement of map and positioning accuracy;

[0115] The terrain feature analysis unit extracts and analyzes the features of the seabed terrain data and provides terrain feature information for other modules;

[0116] The accuracy assessment and output unit performs accuracy assessment on the spliced ​​terrain data, outputs the final seabed topographic map, and provides the positioning optimization results to other modules.

[0117] SLAM (Simultaneous Localization and Mapping) algorithm is a technology that uses sensors to build maps in real time and simultaneously locate itself in an unknown environment. It is widely used in robot navigation, autonomous driving, drone flight, and other fields that need to perform tasks in unknown or dynamic environments. The core task of SLAM is to enable the robot to locate its own position and build a map of the surrounding environment without external positioning support (such as GPS).

[0118] The terrain mapping module achieves high-precision seabed terrain mapping through the collaborative work of multiple precision components. The terrain data compensation unit and the data stitching unit ensure the accuracy and seamless connection of the data, while the SLAM algorithm implementation unit and the closed-loop optimization unit ensure the continuous accuracy of the map by optimizing the posture and map in real time. The terrain feature analysis unit provides the system with valuable environmental information, and the accuracy assessment and output unit ensures that the mapping results meet the task requirements. Through these technical means, the terrain mapping module can provide high-quality, real-time updated seabed terrain maps in complex seabed environments, providing a solid guarantee for deep-water exploration robots to complete accurate seabed mapping tasks.

[0119] Furthermore, the data deviation caused by positioning error or sensor deviation is corrected by the following formula: Among them, D finalis the corrected seafloor topography data; (x raw ,y raw ) is the coordinate of the original seabed topography data, which represents the initial coordinate of the seabed topography in the measured data without correction; e x , e y are positioning errors, representing the horizontal positioning error and vertical positioning error of the robot respectively; e θ is the posture error, indicating the robot's orientation error; θ is the robot's orientation; R(e θ ) is the rotation matrix, which is used to correct the posture error e θ The terrain data deviation caused by x ,s y is the sensor deviation, which represents the systematic deviation or error of the sensor itself in the horizontal and vertical directions respectively.

[0120] Through the above correction formula and process, the deep-water exploration robot can effectively correct the seabed terrain data deviation caused by positioning error, attitude error and sensor deviation. These correction measures ensure the accuracy of the collected seabed terrain data, provide a reliable basis for subsequent path planning, map construction and accuracy evaluation, and thus improve the performance and operability of the entire ocean mapping system.

[0121] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, a person skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot, characterized in that: Includes the following modules: The data acquisition module collects navigation positioning data and seabed topography data synchronously through the multi-source navigation equipment and seabed measurement equipment carried by the deep-water exploration robot; Navigation fusion module, which realizes adaptive fusion of multi-source data and outputs initial pose estimation; The quality control module conducts quality assessment and anomaly identification on real-time surveying and mapping data, and generates local re-survey requirements based on the assessment results to ensure that the surveying and mapping accuracy meets the requirements; The path planning module plans the optimal mapping path in real time based on the robot's current posture, task requirements, and local terrain features, and makes online adjustments based on the re-measurement requirements of the quality control module; Fault-tolerant control module, which implements multi-level navigation source fault diagnosis and initiates degraded navigation strategy when navigation is abnormal or fails, ensuring that basic navigation and mapping functions can still be maintained; The terrain mapping module uses the posture information provided by the navigation fusion module to perform real-time compensation and splicing of seabed terrain data.

2. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 1 is characterized in that: The data acquisition module includes the following components: The navigation and positioning unit collects the original navigation data of the deep-water exploration robot through integrated GPS, inertial navigation system and sonar positioning equipment; The seafloor topography detection unit uses multi-beam sonar and side-scan sonar technology to collect seafloor topography data in real time, including the depth, shape and characteristics of the seabed; The environmental perception unit obtains marine environmental parameters through environmental sensors and provides real-time data support for environmental changes during surveying and mapping; The posture monitoring unit uses a gyroscope and an accelerometer to collect the robot's posture data in real time; The data preprocessing unit performs preliminary processing and formatting on the collected raw data to provide standardized data input for subsequent modules.

3. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 1 is characterized in that: The navigation fusion module includes the following components: The credibility assessment unit, based on a deep learning algorithm, evaluates the data reliability of each navigation source in real time and outputs a credibility score for each navigation source; The weight adjustment unit dynamically adjusts the weight of each navigation source in the fusion calculation according to the credibility score; The error modeling unit uses an improved Kalman filter algorithm to model the error characteristics of each navigation source; The data synchronization and fusion unit synchronizes and integrates navigation data from different sources, and calculates the preliminary pose estimation results based on the dynamic weight and error model; The pose output unit outputs the fused preliminary pose estimation results to provide high-precision navigation information for subsequent path planning and surveying compensation.

4. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 3 is characterized in that: The credibility score of the navigation source is expressed as: in, is the credibility score of the i-th navigation source, with a value between [0, 1], indicating the reliability of the navigation source; W i is the weight matrix corresponding to the i-th navigation source; x i is the feature vector of the i-th navigation source; b i is the bias term; The weight of each navigation source in the fusion calculation is dynamically adjusted by the following formula: w i is the weight of the i-th navigation source and its corresponding credibility score is proportional to; It is the sum of the credibility scores of all navigation sources. It is normalized to ensure that the sum of all weights is 1.

5. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 4 is characterized in that: The preliminary pose estimation result is expressed as: in, represents the preliminary pose estimation result after fusion; p i (t i ) represents the i-th navigation source at time t i Navigation data at the moment; is the i-th navigation source at t i The rate of change at a moment; t i Indicates the timestamp of the i-th navigation source; t sync Indicates the target synchronization time; e i (t sync ) represents the error model of the i-th navigation source at synchronization time t sync The error estimate under i Represents the error model of the i-th navigation source.

6. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 1, characterized in that: The path planning module includes the following components: The task requirement analysis unit analyzes the task requirements according to the objectives and constraints of the surveying and mapping task, and provides task-oriented constraints for path planning; Environmental constraint analysis unit, which analyzes local terrain features based on real-time seabed terrain data and combines environmental perception data to provide environmental constraint information for path planning; The path generation unit uses a reinforcement learning algorithm to learn and generate the optimal mapping path in real time based on the current position, task requirements and terrain characteristics; The path optimization unit dynamically adjusts the current path planning according to the re-survey requirements or feedback information generated by the quality control module to ensure that the path meets the surveying and mapping accuracy and task requirements; The safety monitoring unit adjusts the path in real time based on terrain analysis results and real-time environmental perception to avoid collisions or entering dangerous areas, ensuring the safety of the robot; The path output unit converts the optimized path into a control instruction sequence to control the robot execution.

7. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 6, characterized in that: Optimal mapping path through reinforcement learning policy π * (s T ) is generated step by step, which can be expressed as: , where a T is the action at the current time T, that is, the path planning decision selected by the robot; w1, w2, and w3 are the weight coefficients of task requirements, terrain adaptability, and progress optimization, respectively; s T represents the state of the robot at time T; R task (s T , a T ) means in state s T Next, take action a T The subsequent task reward is used to evaluate the completion of the current task requirements; R terrain (s T , a T ) means in state s T Next, take action a T The terrain adaptability reward after is used to evaluate the adaptability of the current path planning on the seabed terrain; R progress (s T , a T ) means in state s T Next, take action a T The progress reward after is used to evaluate the contribution of the current path planning to the task progress; γ is the discount factor, which is used to balance the relationship between the current reward and the future reward; V(s T+1 ) in state s T+1 Under this condition, the value function after the robot adopts the optimal strategy is expressed as T+1 The maximum long-term reward that can be obtained by setting out.

8. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 1, characterized in that: The fault-tolerant control module includes the following components: The navigation status monitoring unit monitors the stability, accuracy and consistency of each navigation source data in real time and captures potential abnormal signals or failure signs; The fault diagnosis unit analyzes the type and severity of navigation source anomalies based on a multi-level diagnostic algorithm to determine the cause of the fault and the scope of impact; Redundant strategy management unit, which manages backup navigation sources and redundant strategy libraries, and quickly switches to the suboptimal navigation source to maintain navigation function when the primary navigation source fails; Degraded navigation execution unit, in case of navigation source failure or abnormality, activates backup navigation scheme to maintain basic navigation function; Abnormal data processing unit, which uniformly processes abnormal data from each module, including data elimination, abnormal correction and interference suppression; The Navigation Performance Evaluation Unit evaluates the performance in the degraded navigation mode in real time to ensure that it meets the minimum mission requirements and attempts to restore the normal navigation mode when conditions permit.

9. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 1, characterized in that: The terrain mapping module includes the following components: The terrain data compensation unit compensates the collected seabed terrain data in real time according to the current posture information of the robot, and corrects the data deviation caused by positioning error or sensor deviation; The data splicing unit splices the seabed topographic data collected at different locations and time periods to form a continuous topographic map, ensuring seamless connection between data; SLAM algorithm implementation unit, which optimizes the relationship between the robot's position and terrain data in real time based on the SLAM algorithm, and establishes and updates an accurate map of the seabed terrain; The closed-loop optimization unit optimizes the accumulated error of the robot's position through closed-loop detection and correction of the SLAM algorithm, ensuring continuous improvement of map and positioning accuracy; The terrain feature analysis unit extracts and analyzes the features of the seabed terrain data and provides terrain feature information for other modules; The accuracy assessment and output unit performs accuracy assessment on the spliced ​​terrain data, outputs the final seabed topographic map, and provides the positioning optimization results to other modules.

10. The ocean surveying and mapping system based on intelligent navigation of a deep-water exploration robot according to claim 9, characterized in that: The following formula is used to correct the data deviation caused by positioning error or sensor deviation: Among them, D final is the corrected seafloor topography data; (x raw ,y raw ) is the coordinate of the original seabed topography data, which represents the initial coordinate of the seabed topography in the measured data without correction; e x , e y are positioning errors, representing the horizontal positioning error and vertical positioning error of the robot respectively; e θ is the posture error, indicating the orientation error of the robot; θ is the orientation of the robot; R(e θ ) is the rotation matrix, which is used to correct the posture error e θ The terrain data deviation caused by x ,s y is the sensor deviation, which represents the systematic deviation or error of the sensor itself in the horizontal and vertical directions respectively.

Images