A multi-sensor fusion underwater robot positioning method and system

Abstract

The present invention discloses a multi-sensor fusion underwater robot positioning method and system, which relates to the technical field of underwater positioning. By fusing multiple sensor information, including sonar data, accelerometers, gyroscopes, magnetic compasses and depth gauges, the positioning accuracy and stability of the underwater robot can be further improved; the sonar data provides three-dimensional information of the underwater environment, and the sensor integration provides the robot's own motion state. The combination of the two can further realize accurate monitoring of the robot's position and posture. By pre-constructing an underwater three-dimensional point cloud map and predicting the robot's state vector in advance, the real-time and adaptability of the positioning system can be enhanced. The pre-constructed point cloud map can be used as a reference to provide auxiliary information in the actual positioning process, and the state vector predicted in advance can be used to adjust the robot's motion trajectory to achieve more accurate navigation and control.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater positioning, and in particular to a multi-sensor fusion underwater robot positioning method and system. Background Art

[0002] Driven by modern science and technology, underwater robot technology has gradually become an indispensable tool for marine engineering, resource exploration and scientific research. In order to achieve the detection, monitoring and operation of underwater environment, researchers are constantly exploring new positioning methods to improve the navigation accuracy and efficiency of underwater robots. Underwater robot positioning technology is the key to underwater robot navigation and control. It needs to obtain accurate underwater robot position and posture information by fusing multiple sensor information to achieve efficient underwater environment survey and operation.

[0003] However, the current underwater robot positioning method based on multi-sensor fusion still has some shortcomings in practical applications. Existing methods usually rely on a single sensor or a simple combination of sensors. The data of a single sensor is often affected by environmental interference and noise, resulting in low positioning accuracy. In addition, for predicting the position state of the underwater robot at the next moment in advance, the traditional method needs to be improved, because sensor data usually has a certain delay, and in the actual positioning process, advance prediction can reduce the frequent sampling and data processing of sensors, thereby saving energy and system resources, and extending the working time and life of the underwater robot. Therefore, we need to continuously improve the precise positioning of underwater robots and continuously improve the prediction of the position state of underwater robots. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a multi-sensor fusion underwater robot positioning method and system, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A multi-sensor fusion underwater robot positioning method comprises the following steps:

[0006] S1. Detect and image the underwater environment in advance through relevant acoustic navigation instruments, generate sonar data information, construct an underwater three-dimensional point cloud map based on the sonar data information, and display the operation path of the underwater robot in the underwater three-dimensional point cloud map;

[0007] S2. several groups of sensors are arranged in the underwater robot to construct a sensor integration, wherein the sensor integration includes an accelerometer, a gyroscope, a magnetic compass and a depth gauge, and the relevant speed data information of the robot in the underwater three-dimensional space is measured by the accelerometer and the gyroscope, and the direction information and the depth information are obtained according to the magnetic compass and the depth gauge;

[0008] S3, preprocess the relevant speed data information, direction information and depth information obtained in step S2, and summarize them to generate a trajectory data set, then extract features from the trajectory data set, and obtain the speed value Vs in different directions and the angular velocity Ss of different rotation modes of the underwater robot during the monitoring period. Based on the trajectory data set after feature extraction, a state evaluation prediction model is constructed, and the state vector of the underwater robot at the future time i is predicted according to the trained state evaluation prediction model. ;

[0009] S4. Obtaining the state vector at time i in the future After that, wait for the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i, and correct the state vector at the future time i according to the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , to obtain the correction vector , when the correction vector When the preset threshold is exceeded, a self-check command is issued and the position status of the underwater robot in the underwater three-dimensional point cloud map is updated.

[0010] Preferably, underwater imaging is performed according to relevant acoustic navigation instruments to generate sonar data information, the noise signal contained in the sonar data is removed by median filtering, and effective signals are extracted, the echo signal is enhanced, the intensity of the sonar echo is analyzed, the characteristic information of the surfaces of different objects is extracted, and the sonar data obtained by the feature extraction is converted into an underwater three-dimensional point cloud map;

[0011] By matching sonar data with maps, the robot's path from the starting point to the target point is planned, and the underwater robot is positioned in real time.

[0012] Preferably, several groups of sensors are pre-installed in the underwater robot to form a sensor integration, wherein the sensor integration also includes a pressure sensor, a temperature sensor, an acoustic sensor and an underwater camera;

[0013] The relevant speed data information of the robot in the underwater three-dimensional space is measured by an accelerometer and a gyroscope, and the direction information and depth information are obtained according to the magnetic compass and the depth meter, wherein the relevant speed data information includes the speed value Vs of the underwater robot in different directions and the angular velocity Ss of different rotation modes during the monitoring period, wherein the speed value Vs in different directions includes the speed value Vx of the underwater robot in the x-axis direction, the speed value Vy of the underwater robot in the y-axis direction and the speed value Vz of the underwater robot in the z-axis direction; the angular velocity Ss of different rotation modes includes the angular velocity of the underwater robot in the longitudinal direction. , the angular velocity of the underwater robot's lateral rotation And the angular velocity of the underwater robot's vertical rotation ;

[0014] The direction information includes the longitudinal rotation angle Q, the lateral rotation angle R and the vertical rotation angle P of the underwater robot during the monitoring period;

[0015] The depth information includes the depth data of the underwater robot during the monitoring period.

[0016] Preferably, according to the speed value Vs of the underwater robot in different directions during the monitoring period, combined with the statistical averaging algorithm, the average speed of the underwater robot in different directions during the monitoring period is obtained respectively. , based on the average speed of the underwater robot in different directions during the monitoring period The speed values ​​Vs of the underwater robot in different directions during the monitoring period simulate the speed of the underwater robot in different directions at the next time i :

[0017] ;

[0018] ;

[0019] ;

[0020] In the formula, , and They are respectively represented as the speed of the underwater robot in the x-axis direction at the next i moments, the speed in the y-axis direction at the next i moments, and the speed in the z-axis direction at the next i moments. , and They represent the average speed of the underwater robot in the x-axis direction, the average speed in the y-axis direction, and the average speed in the z-axis direction during the monitoring period, respectively. , and They are respectively represented as the speed value of the underwater robot in the x-axis direction at the i-1th monitoring time point, the speed value of the underwater robot in the y-axis direction at the i-1th monitoring time point, and the speed value of the underwater robot in the z-axis direction at the i-1th monitoring time point.

[0021] Preferably, a convolutional neural network is used to construct an initial orientation model, and the initial orientation model is trained and tested with relevant speed data information, direction information and depth information, and the trained initial orientation model is used as an orientation recognition model to obtain feature information therein, and the acquired feature information is used to train and test the orientation recognition model, and the speed of the underwater robot in different directions at the next i moments is simulated. , the trained orientation recognition model is used as the state evaluation prediction model.

[0022] Preferably, the trained state evaluation prediction model is used to predict the state vector of the underwater robot at the next time i :

[0023] ;

[0024] In the formula, , and They are respectively represented as the displacement in the x-axis direction at the i-1th monitoring time point, the displacement in the y-axis direction at the i-1th monitoring time point, and the displacement in the z-axis direction at the i-1th monitoring time point; , and They are respectively represented as the longitudinal rotation angle of the underwater robot at the i-1th monitoring time point, the lateral rotation angle of the underwater robot at the i-1th monitoring time point, and the vertical rotation angle of the underwater robot at the i-1th monitoring time point; Expressed as a time interval; , and It is expressed as the angular velocity of the longitudinal rotation of the underwater robot at the i-1th monitoring time point, the angular velocity of the lateral rotation of the underwater robot at the i-1th monitoring time point, and the angular velocity of the vertical rotation of the underwater robot at the i-1th monitoring time point.

[0025] Preferably, the actual measurement value is obtained based on the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , and the state vector at the future i moment Make corrections and obtain the correction vector :

[0026] .

[0027] Preferably, a threshold is preset and the threshold is combined with the correction vector A comparative analysis is performed to determine whether the position status of the underwater robot in the three-dimensional point cloud map needs to be updated. The specific contents are as follows:

[0028] If the correction vector When the threshold is exceeded, a self-check instruction will be issued to update the position status of the underwater robot in the three-dimensional point cloud map in time;

[0029] If the correction vector When the threshold is not exceeded, no self-check instruction will be issued, and there is no need to update the position state of the underwater robot in the three-dimensional point cloud map.

[0030] A multi-sensor fusion underwater robot positioning system, including an underwater environment control module, a sensor monitoring module, a position prediction and analysis module, and a correction module;

[0031] The underwater environment control module detects and images the underwater environment in advance through relevant acoustic navigation instruments, generates sonar data information, constructs an underwater three-dimensional point cloud map based on the sonar data information, and displays the running path of the underwater robot in the underwater three-dimensional point cloud map;

[0032] The sensor monitoring module is provided with several groups of sensors in the underwater robot to construct a sensor integration, wherein the sensor integration includes an accelerometer, a gyroscope, a magnetic compass and a depth gauge, and measures the relevant speed data information of the robot in the underwater three-dimensional space through the accelerometer and the gyroscope, and obtains the direction information and the depth information according to the magnetic compass and the depth gauge;

[0033] The position prediction and analysis module performs data preprocessing on the relevant speed data information, direction information and depth information obtained in step S2, and summarizes and generates a trajectory data set, and then extracts features from the trajectory data set to obtain the speed value Vs in different directions and the angular velocity Ss of different rotation modes of the underwater robot in the monitoring period, respectively. Based on the trajectory data set after feature extraction, a state evaluation prediction model is constructed, and the state vector of the underwater robot at the future time i is predicted according to the trained state evaluation prediction model. ;

[0034] The correction module obtains the state vector at the future time i After that, wait for the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i, and correct the state vector at the future time i according to the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , to obtain the correction vector , when the correction vector When the preset threshold is exceeded, a self-check command is issued and the position status of the underwater robot in the underwater three-dimensional point cloud map is updated.

[0035] The present invention provides a multi-sensor fusion underwater robot positioning method and system, which has the following beneficial effects:

[0036] (1) By integrating information from multiple sensors, including sonar data, accelerometers, gyroscopes, magnetic compasses, and depth gauges, the positioning accuracy and stability of underwater robots can be further improved; sonar data provides three-dimensional information about the underwater environment, while sensor integration provides the robot's own motion state. The combination of the two can further achieve accurate monitoring of the robot's position and posture. By pre-constructing an underwater three-dimensional point cloud map and predicting the robot's state vector in advance, the real-time and adaptability of the positioning system can be enhanced. The pre-constructed point cloud map can be used as a reference to provide auxiliary information during the actual positioning process, while the predicted state vector can be used to adjust the robot's motion trajectory to achieve more accurate navigation and control. By processing and correcting sensor data, the frequent sampling and data processing of sensors in the later stage can be reduced, thereby saving energy and system resources and further extending the working time and life of the underwater robot. In addition, according to the correction vector Update the position status of the underwater robot in the underwater 3D point cloud map to further improve the efficiency of the system and save resources. The improvement of positioning accuracy and stability, as well as the enhancement of real-time and adaptability, will directly affect the task execution efficiency of the underwater robot.

[0037] (2) By predicting and correcting the position state of the underwater robot, the robot has a certain degree of autonomous decision-making and path planning capabilities, further realizing the autonomy and intelligence of the underwater robot. This method integrates multiple sensor information, comprehensively considers multiple factors of the underwater environment, improves the underwater robot's ability to adapt to complex environments, and has strong robustness and stability. By optimizing the collection and processing methods of sensor data, it saves energy and system resource consumption, reduces the operating cost of the underwater robot, and improves its use efficiency.

[0038] (3) Based on the speed value Vs of the underwater robot in different directions during the monitoring period, combined with the statistical averaging algorithm, the speed of the underwater robot in different directions at future moments is simulated. This real-time speed simulation and prediction method can further help the robot predict future movement trends, make adjustments and decisions in advance, and enhance the robot's autonomy and adaptability. By obtaining the average speed of the underwater robot in different directions , and based on the real-time speed simulation and prediction results, the path planning and navigation strategies can be optimized, enabling the robot to reach the target location more quickly and accurately, which helps to improve the efficiency and success rate of underwater operations and further reduce costs and risks.

[0039] The future position and state of the underwater robot can be predicted by analyzing historical data and pattern recognition. Sensor data usually has a certain delay, and advance prediction can partially offset the impact of this delay and improve the accuracy and stability of positioning. In the actual positioning process, advance prediction can reduce the frequent sampling and data processing of sensors, thereby saving energy and system resources and extending the working time and life of the underwater robot. Advance prediction can reduce the impact of sensor noise and interference by analyzing historical data, and improve the accuracy and stability of positioning.

[0040] (3) Using a convolutional neural network to construct an initial orientation model can fully utilize the advantages of deep learning and learn more abstract and effective feature representations from speed data, direction information, and depth information. Such an orientation model has stronger generalization ability and adaptability, and can more accurately identify the position of the underwater robot. Training and testing the relevant speed data, direction information, and depth information with the initial orientation model can comprehensively utilize the information of multiple sensors to improve the accuracy and robustness of the orientation recognition model. This helps to more accurately capture the characteristics of the underwater environment and the motion state of the robot. Combined with the simulation of the speed of the underwater robot in different directions at future times, the trained orientation recognition model is used as a state evaluation prediction model to accurately predict the state vector of the underwater robot at future times. Such prediction capabilities help to discover potential position deviations or motion trends in advance, providing an important reference for the real-time adjustment and path planning of the robot. By comprehensively utilizing deep learning models and multi-sensor information, the positioning accuracy and robustness of the underwater robot can be improved. The robot can more accurately understand and perceive the underwater environment, achieve more stable and reliable positioning and navigation, and thus improve work efficiency and success rate.

[0041] (4) By correcting the position status in real time, the accumulation of positioning errors can be further reduced. The underwater environment is complex and changeable, and sensor data may be affected by various interferences, resulting in the gradual accumulation of positioning errors. Timely automatic correction can effectively avoid further accumulation of errors and effectively ensure the accuracy and reliability of positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of the steps of a multi-sensor fusion underwater robot positioning method of the present invention;

[0043] Figure 2 The figure is a flow chart of a multi-sensor fusion underwater robot positioning system of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Example 1

[0046] See also Figure 1 The present invention provides a multi-sensor fusion underwater robot positioning method, comprising the following steps:

[0047] S1. Detect and image the underwater environment in advance through relevant acoustic navigation instruments, generate sonar data information, construct an underwater three-dimensional point cloud map based on the sonar data information, and display the operation path of the underwater robot in the underwater three-dimensional point cloud map;

[0048] S2. several groups of sensors are arranged in the underwater robot to construct a sensor integration, wherein the sensor integration includes an accelerometer, a gyroscope, a magnetic compass and a depth gauge, and the relevant speed data information of the robot in the underwater three-dimensional space is measured by the accelerometer and the gyroscope, and the direction information and the depth information are obtained according to the magnetic compass and the depth gauge;

[0049] S3, preprocess the relevant speed data information, direction information and depth information obtained in step S2, and summarize them to generate a trajectory data set, then extract features from the trajectory data set, and obtain the speed value Vs in different directions and the angular velocity Ss of different rotation modes of the underwater robot during the monitoring period. Based on the trajectory data set after feature extraction, a state evaluation prediction model is constructed, and the state vector of the underwater robot at the future time i is predicted according to the trained state evaluation prediction model. ;

[0050] S4. Obtaining the state vector at time i in the future After that, wait for the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i, and correct the state vector at the future time i according to the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , to obtain the correction vector , when the correction vector When the preset threshold is exceeded, a self-check command is issued and the position status of the underwater robot in the underwater three-dimensional point cloud map is updated.

[0051] In this embodiment, the underwater three-dimensional point cloud map is constructed through sonar data, and the data provided by the accelerometer, gyroscope, magnetic compass and depth meter can be used to obtain the position, speed and posture information of the robot, thereby further achieving more accurate positioning. The robot's position status is updated in real time. This method can realize autonomous navigation and operation of underwater robots. The robot can plan paths and make autonomous decisions based on the predicted state vector, thereby completing complex underwater tasks. Real-time status updates can reduce frequent sampling and data processing of sensors, thereby saving energy and system resources and extending the working time and life of underwater robots. The multi-sensor fusion method can comprehensively consider the information of different sensors and improve the underwater robot's perception and adaptability to complex underwater environments. The robot can more accurately identify and respond to complex situations such as underwater obstacles and terrain changes, further improving the success rate and efficiency of task execution. By constructing an underwater three-dimensional point cloud map and displaying the operation path of the underwater robot, efficient surveying and operation of the underwater environment can be achieved. Underwater robots can accurately perceive and record information such as the terrain and object distribution of the underwater environment, providing important support for marine engineering, resource exploration and scientific research.

[0052] Example 2

[0053] Please refer to Figure 1 Specifically: Underwater imaging is performed according to relevant acoustic navigation instruments to generate sonar data information, the noise signal contained in the sonar data is removed through median filtering, and effective signals are extracted, echo signals are enhanced, and the resolution of reflection intensity is improved to better identify environmental features, analyze the intensity of sonar echoes, extract feature information of different object surfaces, and then convert the sonar data obtained by feature extraction into an underwater three-dimensional point cloud map;

[0054] By matching sonar data with maps, the robot's path from the starting point to the target point is planned, and the underwater robot is positioned in real time.

[0055] In this embodiment, by performing median filtering on the sonar data, the noise signal can be effectively removed, the effective signal can be extracted, and the echo signal can be enhanced, thereby improving the quality and resolution of the sonar data, which helps to capture the characteristics of the underwater environment more clearly, accurately analyze the intensity of the sonar echo, and extract the characteristic information of the surfaces of different objects. The sonar data obtained by feature extraction can be converted into an underwater three-dimensional point cloud map, which can not only accurately reflect the terrain and object distribution of the underwater environment, but also provide detailed environmental feature information, providing an important reference for the positioning and navigation of the underwater robot. By matching the sonar data with the map, the path of the underwater robot from the starting point to the target point can be planned. This path planning can take into account the actual situation of the underwater environment, avoid obstacles and dangerous areas, and achieve safe and efficient navigation. Using real-time sonar data to locate the real-time position of the underwater robot can continuously update the position state of the robot, achieve accurate positioning and stable navigation, which helps to improve the task execution efficiency and success rate of the underwater robot. By analyzing sonar data and constructing underwater 3D point cloud maps, we can have a more comprehensive understanding of the characteristics and structure of the underwater environment, providing richer information and support for the planning and execution of underwater missions.

[0056] By removing noise signals and enhancing echo signals, the quality and resolution of sonar data can be improved, making the characteristic information of the underwater environment more clearly visible. This helps to more accurately identify the characteristic information of the surfaces of different objects, such as seabed topography, underwater obstacles, etc. Based on the extracted clear sonar data, an underwater three-dimensional point cloud map can be generated more accurately and matched with the underwater environment, thereby achieving real-time positioning of the robot. This helps to improve the accuracy and reliability of the robot's path planning, allowing the robot to reach the target point more efficiently from the starting point. Accurate environmental feature recognition and precise path planning can improve the operating efficiency and safety of underwater robots. The robot can complete underwater tasks more quickly and accurately, while avoiding collisions or getting into trouble, ensuring the smooth progress of operations.

[0057] By removing noise and enhancing echo signals through median filtering, and then extracting effective feature information, it not only improves the recognition accuracy of underwater environmental features, but also effectively copes with the noise and interference in sonar data. Combining clear sonar data and underwater three-dimensional point cloud maps, it can achieve real-time positioning and path planning of underwater robots. This real-time positioning and planning method can more flexibly cope with changes in the underwater environment and improve the autonomy and adaptability of the robot.

[0058] Example 3

[0059] Please refer to Figure 1Specifically: several groups of sensors are pre-arranged in the underwater robot to form a sensor integration, wherein the sensor integration also includes a pressure sensor, a temperature sensor, an acoustic sensor and an underwater camera;

[0060] The relevant speed data information of the robot in the underwater three-dimensional space is measured by an accelerometer and a gyroscope, and the direction information and depth information are obtained according to the magnetic compass and the depth meter, wherein the relevant speed data information includes the speed value Vs of the underwater robot in different directions and the angular velocity Ss of different rotation modes during the monitoring period, wherein the speed value Vs in different directions includes the speed value Vx of the underwater robot in the x-axis direction, the speed value Vy of the underwater robot in the y-axis direction and the speed value Vz of the underwater robot in the z-axis direction; the angular velocity Ss of different rotation modes includes the angular velocity of the underwater robot in the longitudinal direction. , the angular velocity of the underwater robot's lateral rotation And the angular velocity of the underwater robot's vertical rotation ;

[0061] The direction information includes the longitudinal rotation angle Q, the lateral rotation angle R and the vertical rotation angle P of the underwater robot during the monitoring period;

[0062] The depth information includes the depth data of the underwater robot during the monitoring period.

[0063] According to the speed value Vs of the underwater robot in different directions during the monitoring period, combined with the statistical averaging algorithm, the average speed of the underwater robot in different directions during the monitoring period is obtained. , based on the average speed of the underwater robot in different directions during the monitoring period The speed values ​​Vs of the underwater robot in different directions during the monitoring period simulate the speed of the underwater robot in different directions at the next time i :

[0064] ;

[0065] ;

[0066] ;

[0067] In the formula, , and They are respectively represented as the speed of the underwater robot in the x-axis direction at the next i moments, the speed in the y-axis direction at the next i moments, and the speed in the z-axis direction at the next i moments. , and They represent the average speed of the underwater robot in the x-axis direction, the average speed in the y-axis direction, and the average speed in the z-axis direction during the monitoring period, respectively. , and They are respectively represented as the speed value of the underwater robot in the x-axis direction at the i-1th monitoring time point, the speed value of the underwater robot in the y-axis direction at the i-1th monitoring time point, and the speed value of the underwater robot in the z-axis direction at the i-1th monitoring time point.

[0068] In this embodiment, by setting up multiple groups of sensors in the underwater robot, including accelerometers, gyroscopes, magnetic compasses, depth gauges, pressure sensors, temperature sensors, acoustic sensors, and underwater cameras, etc., various sensor information can be comprehensively utilized to further improve the accuracy and reliability of positioning. These sensors can measure the speed, direction, depth, and environmental parameters of the underwater robot in three-dimensional space, and provide more comprehensive and accurate data support for positioning. By measuring the speed and angular velocity of the underwater robot with accelerometers and gyroscopes, and obtaining direction information and depth information based on the magnetic compass and depth gauge, the status of the underwater robot can be monitored in real time, which helps to promptly detect and handle abnormal situations of the robot during underwater movement, and ensure the accuracy and stability of positioning. Based on the speed value Vs of the underwater robot in different directions during the monitoring period, the average speed is obtained through a statistical averaging algorithm. , and simulate the speed at future moments, which helps predict the movement trend of the underwater robot in different directions, thereby optimizing path planning and enabling the robot to complete tasks more efficiently. This method of comprehensive utilization of multiple sensor information can give full play to the advantages of each sensor and improve the accuracy and stability of positioning. Acquiring sound and image information of the underwater environment through acoustic sensors and underwater cameras can enhance the underwater robot's ability to perceive the environment, which helps to identify underwater obstacles, detect seabed terrain and observe marine life, etc., providing more comprehensive environmental information for underwater operations.

[0069] Example 4

[0070] Please refer to Figure 1 Specifically: Use convolutional neural network to build an initial orientation model, and train and test the initial orientation model with relevant speed data information, direction information and depth information, and use the trained initial orientation model as an orientation recognition model to obtain feature information therein, and use the obtained feature information to train and test the orientation recognition model, and combine it to simulate the speed of the underwater robot in different directions at the next i moments , the trained orientation recognition model is used as the state evaluation prediction model.

[0071] Use the trained state evaluation prediction model to predict the state vector of the underwater robot at time i in the future :

[0072] ;

[0073] In the formula, , and They are respectively represented as the displacement in the x-axis direction at the i-1th monitoring time point, the displacement in the y-axis direction at the i-1th monitoring time point, and the displacement in the z-axis direction at the i-1th monitoring time point; , and They are respectively represented as the longitudinal rotation angle of the underwater robot at the i-1th monitoring time point, the lateral rotation angle of the underwater robot at the i-1th monitoring time point, and the vertical rotation angle of the underwater robot at the i-1th monitoring time point; Expressed as a time interval; , and It is expressed as the angular velocity of the longitudinal rotation of the underwater robot at the i-1th monitoring time point, the angular velocity of the lateral rotation of the underwater robot at the i-1th monitoring time point, and the angular velocity of the vertical rotation of the underwater robot at the i-1th monitoring time point.

[0074] In this embodiment, by using a convolutional neural network to construct an initial orientation model, and combining relevant speed data information, direction information and depth information for training and testing, efficient orientation recognition and prediction can be further achieved; the neural network can automatically learn and extract features, and the trained model is applied to the actual scene, which can quickly and accurately identify the orientation state of the underwater robot, and provide reliable support for subsequent navigation and control. The orientation recognition model not only takes into account the position information of the underwater robot, but also combines multi-dimensional information such as speed, angle and angular velocity. Such multi-dimensional data fusion can more comprehensively describe the motion state of the underwater robot and improve the accuracy and robustness of orientation recognition. Through the trained orientation recognition model, the characteristics and changes of different underwater environments can be adapted. In short, based on the trained model, the real-time prediction and update of the underwater robot state vector can be further achieved. This real-time and flexibility can meet the needs of underwater operation tasks for timely feedback and adjustment, and improve the autonomy and adaptability of the robot. Through accurate position identification and prediction, we can prepare for reducing the positioning error of underwater robots in the later stage and improve the accuracy and stability of positioning, which will help improve the navigation efficiency and operation success rate of underwater robots and reduce operation risks and costs.

[0075] Example 5

[0076] Please refer to Figure 1Specifically: Get the actual measurement value based on the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , and the state vector at the future i moment Make corrections and obtain the correction vector :

[0077] .

[0078] A threshold is set in advance, and the threshold is combined with the correction vector A comparative analysis is performed to determine whether the position status of the underwater robot in the three-dimensional point cloud map needs to be updated. The specific contents are as follows:

[0079] If the correction vector When the threshold is exceeded, a self-check instruction will be issued to update the position status of the underwater robot in the three-dimensional point cloud map in time;

[0080] If the correction vector When the threshold is not exceeded, no self-check instruction will be issued, and there is no need to update the position state of the underwater robot in the three-dimensional point cloud map.

[0081] In this embodiment, the state vector at the future time i is corrected, and the correction vector is obtained according to the actual measurement value. , can further realize the real-time correction of the underwater robot's position state. Such real-time correction can correct the positioning error in time and further ensure the accuracy of the underwater robot's position state in the three-dimensional point cloud map. Pre-set the threshold and compare it with the correction vector Comparative analysis can realize the automatic self-check mechanism. When the preset threshold is exceeded, the system will automatically issue a self-check instruction, prompting the underwater robot to update its position status; otherwise, it does not need to be updated temporarily. Such an automated self-check mechanism can reduce manual intervention and improve the autonomy and intelligence of the system. Real-time status correction and update can reduce positioning errors and uncertainties, and improve positioning reliability and stability. In the process of performing tasks, the robot can more reliably locate the target position, reduce the possibility of misjudgment and wrong navigation, and improve the success rate and efficiency of task execution.

[0082] In short, by comparing and analyzing the correction vector and the threshold, the system can adjust the position state in time according to the actual situation, avoid unnecessary misoperation and misjudgment, and maintain the stability and reliability of the system operation.

[0083] Example 6

[0084] Please refer to Figure 1 and Figure 2,Specifically: A multi-sensor fusion underwater robot positioning system includes an underwater environment control module, a sensor monitoring module, a position prediction and analysis module and a correction module;

[0085] The underwater environment control module detects and images the underwater environment in advance through relevant acoustic navigation instruments, generates sonar data information, constructs an underwater three-dimensional point cloud map based on the sonar data information, and displays the running path of the underwater robot in the underwater three-dimensional point cloud map;

[0086] The sensor monitoring module is provided with several groups of sensors in the underwater robot to construct a sensor integration, wherein the sensor integration includes an accelerometer, a gyroscope, a magnetic compass and a depth gauge, and measures the relevant speed data information of the robot in the underwater three-dimensional space through the accelerometer and the gyroscope, and obtains the direction information and the depth information according to the magnetic compass and the depth gauge;

[0087] The position prediction and analysis module performs data preprocessing on the relevant speed data information, direction information and depth information obtained in step S2, and summarizes and generates a trajectory data set, and then extracts features from the trajectory data set to obtain the speed value Vs in different directions and the angular velocity Ss of different rotation modes of the underwater robot in the monitoring period, respectively. Based on the trajectory data set after feature extraction, a state evaluation prediction model is constructed, and the state vector of the underwater robot at the future time i is predicted according to the trained state evaluation prediction model. ;

[0088] The correction module obtains the state vector at the future time i After that, wait for the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i, and correct the state vector at the future time i according to the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , to obtain the correction vector , when the correction vector When the preset threshold is exceeded, a self-check command is issued and the position status of the underwater robot in the underwater three-dimensional point cloud map is updated.

[0089] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-sensor fusion underwater robot positioning method, characterized in that: The following steps are included: S1. Detect and image the underwater environment in advance through relevant acoustic navigation instruments, generate sonar data information, construct an underwater three-dimensional point cloud map based on the sonar data information, and display the operation path of the underwater robot in the underwater three-dimensional point cloud map; S2. several groups of sensors are arranged in the underwater robot to construct a sensor integration, wherein the sensor integration includes an accelerometer, a gyroscope, a magnetic compass and a depth gauge, and the relevant speed data information of the robot in the underwater three-dimensional space is measured by the accelerometer and the gyroscope, and the direction information and the depth information are obtained according to the magnetic compass and the depth gauge; S3, preprocess the relevant speed data information, direction information and depth information obtained in step S2, and summarize them to generate a trajectory data set, then extract features from the trajectory data set, and obtain the speed value Vs in different directions and the angular velocity Ss of different rotation modes of the underwater robot during the monitoring period. Based on the trajectory data set after feature extraction, a state evaluation prediction model is constructed, and the state vector of the underwater robot at the future time i is predicted according to the trained state evaluation prediction model. ; The convolutional neural network is used to construct the initial orientation model, and the initial orientation model is trained and tested with relevant speed data information, direction information and depth information. The trained initial orientation model is used as the orientation recognition model to obtain the feature information therein, and the acquired feature information is used to train and test the orientation recognition model, and the speed of the underwater robot in different directions at the next time i is simulated. , the trained orientation recognition model is used as a state evaluation prediction model; Use the trained state evaluation prediction model to predict the state vector of the underwater robot at time i in the future : ； In the formula, , and They are respectively represented as the displacement in the x-axis direction at the i-1th monitoring time point, the displacement in the y-axis direction at the i-1th monitoring time point, and the displacement in the z-axis direction at the i-1th monitoring time point; , and They are respectively represented as the longitudinal rotation angle of the underwater robot at the i-1th monitoring time point, the lateral rotation angle of the underwater robot at the i-1th monitoring time point, and the vertical rotation angle of the underwater robot at the i-1th monitoring time point; Expressed as a time interval; , and It is expressed as the angular velocity of the underwater robot's longitudinal rotation at the i-1th monitoring time point, the angular velocity of the underwater robot's lateral rotation at the i-1th monitoring time point, and the angular velocity of the underwater robot's vertical rotation at the i-1th monitoring time point; S4. Obtaining the state vector at time i in the future After that, wait for the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i, and correct the state vector at the future time i according to the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , to obtain the correction vector , when the correction vector When the preset threshold is exceeded, a self-check command is issued and the position status of the underwater robot in the underwater three-dimensional point cloud map is updated.

2. The underwater robot positioning method of multi-sensor fusion according to claim 1 is characterized by: Underwater imaging is performed based on relevant acoustic navigation instruments to generate sonar data information. The noise signal contained in the sonar data is removed through median filtering, and effective signals are extracted to enhance the echo signal. The intensity of the sonar echo is analyzed to extract the characteristic information of the surfaces of different objects. The sonar data obtained by feature extraction is then converted into an underwater three-dimensional point cloud map. By matching sonar data with maps, the robot's path from the starting point to the target point is planned, and the underwater robot is positioned in real time.

3. The underwater robot positioning method of multi-sensor fusion according to claim 1 is characterized by: Several groups of sensors are pre-installed in the underwater robot to form a sensor integration, wherein the sensor integration also includes a pressure sensor, a temperature sensor, an acoustic sensor and an underwater camera; The relevant speed data information of the robot in the underwater three-dimensional space is measured by an accelerometer and a gyroscope, and the direction information and depth information are obtained according to the magnetic compass and the depth meter, wherein the relevant speed data information includes the speed value Vs of the underwater robot in different directions and the angular velocity Ss of different rotation modes during the monitoring period, wherein the speed value Vs in different directions includes the speed value Vx of the underwater robot in the x-axis direction, the speed value Vy of the underwater robot in the y-axis direction and the speed value Vz of the underwater robot in the z-axis direction; the angular velocity Ss of different rotation modes includes the angular velocity of the underwater robot in the longitudinal direction. , the angular velocity of the underwater robot's lateral rotation And the angular velocity of the underwater robot's vertical rotation ; The direction information includes the longitudinal rotation angle Q, the lateral rotation angle R and the vertical rotation angle P of the underwater robot during the monitoring period; The depth information includes the depth data of the underwater robot during the monitoring period.

4. The underwater robot positioning method of multi-sensor fusion according to claim 3 is characterized by: According to the speed value Vs of the underwater robot in different directions during the monitoring period, combined with the statistical averaging algorithm, the average speed of the underwater robot in different directions during the monitoring period is obtained. , based on the average speed of the underwater robot in different directions during the monitoring period The speed value Vs of the underwater robot in different directions during the monitoring period simulates the speed of the underwater robot in different directions at the next time i : ； ； ； In the formula, , and They are respectively represented as the speed of the underwater robot in the x-axis direction at the next i moments, the speed in the y-axis direction at the next i moments, and the speed in the z-axis direction at the next i moments. , and They represent the average speed of the underwater robot in the x-axis direction, the average speed in the y-axis direction, and the average speed in the z-axis direction during the monitoring period, respectively. , and They are respectively represented as the speed value of the underwater robot in the x-axis direction at the i-1th monitoring time point, the speed value of the underwater robot in the y-axis direction at the i-1th monitoring time point, and the speed value of the underwater robot in the z-axis direction at the i-1th monitoring time point.

5. The underwater robot positioning method of multi-sensor fusion according to claim 1, characterized in that: Get the actual measurement value based on the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , and the state vector at the future i moment Make corrections and obtain the correction vector : 。 6. The underwater robot positioning method of multi-sensor fusion according to claim 1, characterized in that: A threshold is set in advance, and the threshold is combined with the correction vector A comparative analysis is performed to determine whether the position status of the underwater robot in the three-dimensional point cloud map needs to be updated. The specific contents are as follows: If the correction vector When the threshold is exceeded, a self-check instruction will be issued to update the position status of the underwater robot in the three-dimensional point cloud map in time; If the correction vector When the threshold is not exceeded, no self-check instruction will be issued, and there is no need to update the position state of the underwater robot in the three-dimensional point cloud map.

7. A multi-sensor fusion underwater robot positioning system, comprising a multi-sensor fusion underwater robot positioning method according to any one of claims 1 to 6, characterized in that: It includes underwater environment control module, sensor monitoring module, position prediction and analysis module and correction module; The underwater environment control module detects and images the underwater environment in advance through relevant acoustic navigation instruments, generates sonar data information, constructs an underwater three-dimensional point cloud map based on the sonar data information, and displays the running path of the underwater robot in the underwater three-dimensional point cloud map; The sensor monitoring module is provided with several groups of sensors in the underwater robot to construct a sensor integration, wherein the sensor integration includes an accelerometer, a gyroscope, a magnetic compass and a depth gauge, and measures the relevant speed data information of the robot in the underwater three-dimensional space through the accelerometer and the gyroscope, and obtains the direction information and the depth information according to the magnetic compass and the depth gauge; The position prediction and analysis module performs data preprocessing on the relevant speed data information, direction information and depth information obtained in step S2, and summarizes and generates a trajectory data set, and then extracts features from the trajectory data set to obtain the speed value Vs in different directions and the angular velocity Ss of different rotation modes of the underwater robot in the monitoring period, respectively. Based on the trajectory data set after feature extraction, a state evaluation prediction model is constructed, and the state vector of the underwater robot at the future time i is predicted according to the trained state evaluation prediction model. ; The correction module obtains the state vector at the future time i After that, wait for the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i, and correct the state vector at the future time i according to the trajectory data set obtained by the speedometer, gyroscope, magnetic compass and depth gauge at the future time i , to obtain the correction vector , when the correction vector When the preset threshold is exceeded, a self-check command is issued and the position status of the underwater robot in the underwater three-dimensional point cloud map is updated.

Images