A deep learning-based path planning and obstacle avoidance method for underwater vehicles

Abstract

The present invention relates to the field of underwater robot technology and discloses a deep learning-based underwater vehicle path planning and obstacle avoidance method, comprising the following steps: S1, acquiring multimodal data of the underwater environment through underwater sensors and the Beidou satellite navigation system, and generating a high-precision environmental model in real time; S2, utilizing a multi-sensor fusion algorithm to fuse the global positioning information provided by the Beidou satellite with data from sonar, an inertial navigation system, and visual SLAM. By fusing high-definition sensors with the Beidou satellite navigation system and combining it with a deep learning model, the underwater vehicle's environmental perception and path planning capabilities are significantly improved. In complex and changing underwater environments, the high-definition sensors can collect accurate environmental data in real time, while the deep learning model accurately identifies and classifies obstacles based on the data, planning the optimal obstacle avoidance path. The accuracy and real-time performance of its perception enhance the vehicle's navigation capabilities.

Description

Technical Field

[0001] The present invention relates to the field of underwater robot technology, and in particular to a deep learning-based underwater vehicle path planning and obstacle avoidance method. Background Art

[0002] With the development of ocean development and underwater detection technology, underwater vehicles have been widely used as important tools for ocean exploration, rescue, military and scientific research. However, the underwater environment is complex and uncertain. Factors such as obstacle distribution, water flow changes and underwater terrain increase the difficulty of vehicle path planning and obstacle avoidance. Therefore, autonomous navigation technology of underwater vehicles has become a research focus.

[0003] Traditional path planning and obstacle avoidance methods are primarily based on classical algorithms, which rely primarily on discretized grid graphs to plan paths. For example, CN109470249A discloses an optimal path planning and obstacle avoidance design method for underwater vehicles. This method integrates a binary heap acceleration algorithm, a 26-neighborhood NA* algorithm, and an "arc-line-arc" turning strategy to optimize the path. This method considers the physical characteristics of the underwater vehicle and optimizes the path using an "arc-line-arc" turning strategy, effectively reducing inflection points and improving path smoothness. However, it still has the following shortcomings:

[0004] This method relies on the construction of a three-dimensional grid map to identify obstacles and does not integrate the real-time perception capabilities of high-definition sensors or multimodal data. The poor lighting and short viewing distance in underwater environments result in low environmental perception accuracy and are unable to cope with complex underwater terrain and dynamically changing obstacles.

[0005] The traditional NA* algorithm uses preset heuristic search rules for path planning and lacks adaptive capabilities. It cannot adjust the path when encountering unknown or complex environments. Furthermore, it lacks an autonomous learning mechanism in dynamically changing underwater environments, resulting in insufficient adaptability and flexibility.

[0006] Although this method optimizes the local path through turning strategies, it has shortcomings in the combination of global navigation and local obstacle avoidance, and is prone to falling into local optimality, resulting in the vehicle being unable to effectively respond to changes in the global environment.

[0007] Therefore, those skilled in the art provide a deep learning-based underwater vehicle path planning and obstacle avoidance method to solve the problems raised in the above background technology. Summary of the Invention

[0008] In view of the shortcomings of the existing technology, the present invention provides an underwater vehicle path planning and obstacle avoidance method based on deep learning to solve the problems raised in the above background technology.

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a deep learning-based underwater vehicle path planning and obstacle avoidance method, comprising the following steps: S1, acquiring multimodal data of the underwater environment through underwater sensors and the Beidou satellite navigation system, and generating a high-precision environment model in real time;

[0010] S2. Use multi-sensor fusion algorithms to fuse the global positioning information provided by BeiDou satellites with data from sonar, inertial navigation systems, and visual SLAM.

[0011] S3, processes environmental perception data through deep learning models, uses convolutional neural networks to identify and classify obstacles in the three-dimensional environment acquired by high-definition cameras and lidar, and predicts navigation paths;

[0012] S4. Based on the deep reinforcement learning model, the obstacle avoidance path is planned in real time in a dynamic environment. The obstacle avoidance strategy is adjusted by the reward function in reinforcement learning. The specific formula is:

[0013] R=-λ1·d obstacle +λ2·d goal -λ3·C collision ,

[0014] Among them, R is the reward value, d obstacle is the distance between the aircraft and the obstacle, d goal is the distance between the aircraft and the target point, C collision is the collision penalty coefficient, λ1, λ2, λ3 are weight parameters;

[0015] S5. Using the global position information provided by the BeiDou navigation system, if the vehicle deviates from the global planned path, the global path correction module is automatically activated to re-plan the path based on the BeiDou satellite data;

[0016] S6. Based on the turning radius and physical size of the spacecraft, the path output by deep learning is optimized using the “arc-straight-arc” path smoothing strategy.

[0017] S7. Through online learning mechanisms, the vehicle can make adaptive adjustments based on new environmental data during execution and update the weights of the deep learning model;

[0018] S8. Generate motion control instructions for the aircraft based on the planned path and real-time obstacle avoidance output, and control the propellers and steering gear of the aircraft to execute motion;

[0019] S9. During the operation of the aircraft, continuously acquire environmental perception data and update the model in real time;

[0020] S10. Monitor the deviation between the position of the aircraft and the target path through the BeiDou navigation system and activate the emergency obstacle avoidance mode in an emergency.

[0021] Preferably, in step S2, the algorithm formula for correcting the positioning error is:

[0022] P t+1 =P t +α·(P BDS -P t )+β·(P INS -P t )+γ·(P SLAM -P t ),

[0023] Among them, P t+1 is the corrected position, P BDS 、P INS 、P SLAM The position information provided by Beidou navigation, inertial navigation and visual SLAM, α, β, γ are weighting coefficients.

[0024] Preferably, the multi-sensor fusion algorithm is used to combine the data of the Beidou navigation system, the inertial navigation system and the visual SLAM to enhance the positioning accuracy of the underwater vehicle. The following is the formula of the fusion algorithm:

[0025] P fused =α·P BDS +β·P INS +γ·P SLAM ,

[0026] Among them, P fused The final positioning result after fusion, P BDS The positioning information obtained by the Beidou satellite navigation system, P INS The positioning information obtained by the inertial navigation system, P SLAM Positioning information obtained through visual SLAM, weighting coefficients of α, β, and γ system data.

[0027] Preferably, the path planning uses a convolutional neural network combined with lidar data for environmental perception and dynamically adjusts the planned path. The following is the algorithm formula for the path planning:

[0028] The environment perception and path planning formula:

[0029]

[0030] Among them, P optimal The optimal path after convolutional neural network and lidar perception optimization, candidate points of P path, d goal The distance from the path point to the target point, dobstacle The distance from the path point to the surrounding obstacles, C smooth The smoothness penalty function of the P path, λ1, λ2, and λ3 control the weight coefficients of each factor.

[0031] Preferably, the Beidou navigation-assisted global path correction module is used to automatically re-plan the path when the aircraft deviates from the predetermined path. The following is the algorithm formula involved:

[0032] Global path correction formula:

[0033] P new =P current +α·(P BDS -P current )+β·(P planned -P current ),

[0034] Among them, P new The corrected new path point, P current The position of the spacecraft, P BDS The precise global location information obtained by the BeiDou navigation system, P planned The target point on the predetermined planning path, and the α and β weight coefficients.

[0035] Preferably, the path smoothing and optimization generates a smooth path of the physical characteristics of the spacecraft through the “arc-straight line-arc” strategy. The following is an algorithm formula for calculating the smooth path:

[0036] Arc-straight line-arc path optimization formula:

[0037] Arc segment calculation: Among them, θ arc The angle of the arc segment, L arc The arc length of the arc segment, r is the turning radius of the vehicle;

[0038] Calculation of straight line segment length:

[0039] Among them, L line The straight-line distance between arc segments; the coordinates of the starting and ending points of the x1, y1 and x2, y2 straight-line segments;

[0040] Path smoothness optimization function:

[0041] C smooth =λ1·θ arc +λ2·L line -λ3·(1 / r),

[0042] Among them, C smooth Path smoothness cost function, λ1, λ2, λ3 weight coefficients.

[0043] Preferably, the dynamic self-learning adjustment updates the weights of the deep learning model in real time through an online learning mechanism. The following is the algorithm formula of the process:

[0044] Online learning weight update formula:

[0045]

[0046] Among them, w t+1 The updated model weight at time t+1, w t The model weight at time t, η learning rate, Loss function L on weight w t The gradient of the model indicates that the model is t and the true label y t The error derivative on ;

[0047] Real-time environment input:

[0048] x t : Input data collected by the sensor or vision system in this environment;

[0049] y t : The actual obstacle avoidance path or successful navigation result is used as a label in supervised learning to correct the model's prediction;

[0050] Loss function:

[0051] Among them, f(w t , x t ) The deep learning model is trained on the input x t The predicted output, y t Actual navigation or obstacle avoidance results.

[0052] Preferably, the controller performs dynamic adjustment of the underwater vehicle through multi-channel control signals. The following is the controller algorithm formula:

[0053] Propeller speed control formula: v new =v current +K v ·(v desired -v current ),

[0054] Among them, v new Adjusted thruster speed, v current Propeller speed, v desired Desired thruster speed, K v Speed ​​adjustment gain coefficient;

[0055] Heading control formula (based on proportional-integral-derivative control, PID):

[0056]

[0057] Among them, θ new Adjusted heading angle, θ current Heading angle, e(t) heading error, K p , K i , K d are the proportional, integral, and differential coefficients of the PID controller;

[0058] Attitude control formula: φ new =φ current +K φ ·(φ desired -φ current ),

[0059] Among them, φ new Adjusted attitude angle, φ current Attitude angle, φ desired Desired attitude angle, K φ Attitude control gain coefficient;

[0060] Total control signal formula: u control =[v new ,θ new ,φ new ],

[0061] The total control signal formula represents the final control signal u control , is given by the propeller speed v new , heading angle θ new and attitude angle φ new The multi-channel control signals composed together are used to adjust the motion and attitude of the spacecraft.

[0062] Preferably, the data feedback and real-time optimization monitors the sensor data and path execution of the aircraft in real time, and dynamically optimizes the path planning and obstacle avoidance strategy through the feedback mechanism. The following is the algorithm formula for real-time data feedback and optimization:

[0063] Real-time optimization feedback formula (based on feedback control theory):

[0064] ΔP adjust =K f ·(P desired -P current ),

[0065] Where ΔP adjust Correction value used to adjust path planning, P desired Expected path point, P current Actual path point, Kf Feedback gain coefficient;

[0066] Path error calculation formula:

[0067]

[0068] Among them, e P (t) Path error, x desired 、y desired 、z desired The three-dimensional coordinates of the desired position, x current 、y current 、z current The three-dimensional coordinates of the actual position;

[0069] Real-time optimization adjustment formula:

[0070] P new =P current +ΔP adjust -η·e P (t),

[0071] Among them, P new The new path point after optimization, η learning rate.

[0072] Preferably, the path tracking and monitoring is based on the high-precision positioning data provided by the Beidou navigation system, and monitors the deviation between the aircraft and the target path in real time and makes timely adjustments. The following is the algorithm formula for path tracking and monitoring:

[0073] Path tracking error formula:

[0074]

[0075] Among them, e track (t) Tracking error between the vehicle and the target path point, x target 、y target 、z target The three-dimensional coordinates of the desired target point, x BDS 、y BDS 、z BDS The three-dimensional coordinates of the aircraft obtained through the Beidou navigation system;

[0076] Path correction formula: P adjust =P BDS +K p ·e track (t),

[0077] Among them, P adjust Adjusted waypoint, P BDS Beidou positioning coordinates of the spacecraft, K p Proportional gain;

[0078] Course adjustment formula: θ new =θ current +K θ ·(θ target -θ current ),

[0079] Among them, θ new Adjusted heading angle, θ current Vehicle heading angle, θ target Desired heading angle, K θ Proportional gain coefficient for heading adjustment;

[0080] Overall control update formula: u control =[P adjust ,θ new ],

[0081] Among them, u control Adjustment signals used to control the aircraft.

[0082] The present invention provides a deep learning-based underwater vehicle path planning and obstacle avoidance method. It has the following beneficial effects:

[0083] 1. This invention significantly improves the environmental perception and path planning capabilities of underwater vehicles by integrating high-definition sensors with the Beidou satellite navigation system and combining it with a deep learning model. In complex and changeable underwater environments, high-definition sensors can collect accurate environmental data in real time, and the deep learning model accurately identifies and classifies obstacles based on the data and plans the optimal obstacle avoidance path. The accuracy and real-time performance of its perception enhance the navigation capabilities of the vehicle, especially in low visibility and complex terrain.

[0084] 2. The present invention introduces a dynamic adaptive learning mechanism, which enables the deep learning model to continuously optimize path planning and obstacle avoidance strategies based on the environmental data obtained in real time during the operation of the vehicle. This method can autonomously learn and update model weights without relying on fixed algorithms or preset environmental information. It has strong environmental adaptability and ensures that the vehicle can maintain stable planning capabilities and obstacle avoidance performance even in unknown or changing underwater environments.

[0085] 3. The present invention intelligently integrates the global high-precision positioning information provided by Beidou satellites with local path planning based on deep learning to ensure that the aircraft has higher stability in global path navigation and flexible obstacle avoidance capabilities in local environments. The intelligent combination of global and local paths avoids the limitations of local optimality in traditional algorithms, ensuring that the aircraft always operates along the optimal route and can effectively cope with dynamically changing obstacles and environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1Flowchart of the present invention. DETAILED DESCRIPTION

[0087] To help those skilled in the art understand the present invention, the following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only partial embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0088] The present invention is described in detail below with reference to the accompanying drawings:

[0089] Example:

[0090] Please see the attached Figure 1 , an embodiment of the present invention provides a deep learning-based underwater vehicle path planning and obstacle avoidance method, comprising the following steps: S1, acquiring multimodal data of the underwater environment through underwater sensors and the Beidou satellite navigation system, and generating a high-precision environment model in real time;

[0091] S2. Use multi-sensor fusion algorithms to fuse the global positioning information provided by BeiDou satellites with data from sonar, inertial navigation systems, and visual SLAM.

[0092] S3, processes environmental perception data through deep learning models, uses convolutional neural networks to identify and classify obstacles in the three-dimensional environment acquired by high-definition cameras and lidar, and predicts navigation paths;

[0093] S4. Based on the deep reinforcement learning model, the obstacle avoidance path is planned in real time in a dynamic environment. The obstacle avoidance strategy is adjusted by the reward function in reinforcement learning. The specific formula is:

[0094] R=-λ1·d obstacle +λ2·d goal -λ3·C collision ,

[0095] Among them, R is the reward value, d obstacle is the distance between the aircraft and the obstacle, d goal is the distance between the aircraft and the target point, C collision is the collision penalty coefficient, λ1, λ2, λ3 are weight parameters;

[0096] S5. Using the global position information provided by the BeiDou navigation system, if the vehicle deviates from the global planned path, the global path correction module is automatically activated to re-plan the path based on the BeiDou satellite data;

[0097] S6. Based on the turning radius and physical size of the spacecraft, the path output by deep learning is optimized using the “arc-straight-arc” path smoothing strategy.

[0098] S7. Through online learning mechanisms, the vehicle can make adaptive adjustments based on new environmental data during execution and update the weights of the deep learning model;

[0099] S8. Generate motion control instructions for the aircraft based on the planned path and real-time obstacle avoidance output, and control the propellers and steering gear of the aircraft to execute motion;

[0100] S9. During the operation of the aircraft, continuously acquire environmental perception data and update the model in real time;

[0101] S10. Monitor the deviation between the position of the aircraft and the target path through the BeiDou navigation system and activate the emergency obstacle avoidance mode in an emergency.

[0102] The benefit of S1 is that it can generate a high-precision environmental model in a real-time dynamic underwater environment, improving the vehicle's ability to recognize and respond to the environment; the benefit of S2 is that it enhances the reliability of the vehicle's navigation and obstacle avoidance in complex environments, especially in areas with unstable signals or many obstacles, ensuring continuous and accurate navigation; the benefit of S3 is that it can quickly identify and classify obstacles, automatically plan obstacle avoidance paths, and accurately predict future navigation paths; the benefit of S4 is that the vehicle can learn to avoid obstacles and move towards the target in complex environments, effectively reducing the risk of collision; the benefit of S5 is that it can effectively adjust the direction in complex or unforeseen environments to avoid getting lost; S6 The benefit of S7 is that it reduces sharp turns and discontinuities in the path, and improves the steering stability of the vehicle and the feasibility of the path; the benefit of S7 is that the dynamic learning mechanism enhances the vehicle's ability to cope with unknown environments and dynamic changes, enabling it to continuously optimize path planning and obstacle avoidance strategies; the benefit of S8 is that it provides a closed loop from perception to control, allowing the vehicle to flexibly cope with complex underwater environments and autonomously perform obstacle avoidance operations; the benefit of S9 is that it ensures that the vehicle always makes decisions based on the latest environmental information during long-term operation and maintains efficient navigation performance; the benefit of S10 is that once a deviation or emergency is detected, the emergency obstacle avoidance mode can be immediately enabled to quickly adjust the vehicle's heading and speed. In step S2, the algorithm formula for correcting the positioning error is:

[0103] P t+1 =P t +α·(P BDS -P t )+β·(P INS -P t )+γ·(P SLAM -P t ),

[0104] Among them, P t+1 is the corrected position, P BDS 、P INS 、P SLAM The position information provided by Beidou navigation, inertial navigation and visual SLAM, α, β, γ are weighting coefficients.

[0105] The core function of the formula is to integrate the advantages of each sensor by fusing data from multiple sensors and reduce the error caused by a single sensor. Beidou navigation provides high-precision global positioning, but cannot provide stable signals in some underwater environments. The inertial navigation system can provide stable positioning information in a short time, but it is prone to drift errors in long-term use. Visual SLAM provides real-time environmental perception through data such as cameras and lidar, but is limited in low-visibility environments. Through the fusion formula, the aircraft can integrate position information from different sources, compensate for their respective defects, and thereby improve the accuracy of overall positioning. By fusing multi-source data, the formula can effectively reduce the positioning error caused by the failure or insufficient accuracy of a single sensor, and enhance the robustness and stability of the aircraft. By weighted fusion of data from different sensors through this formula, it can ensure that the aircraft can maintain a high positioning accuracy in complex environments, thereby enhancing the accuracy of path planning and obstacle avoidance.

[0106] The multi-sensor fusion algorithm is used to combine the data of the BeiDou navigation system, inertial navigation system and visual SLAM to enhance the positioning accuracy of underwater vehicles. The following is the formula of the fusion algorithm:

[0107] P fused =α·P BDS +β·P INS +γ·P SLAM ,

[0108] Among them, P fused The final positioning result after fusion, P BDS The positioning information obtained by the Beidou satellite navigation system, P INS The positioning information obtained by the inertial navigation system, P SLAM Positioning information obtained through visual SLAM, weighting coefficients of α, β, and γ system data.

[0109] By weightedly fusing data from different sensors and integrating positioning information from Beidou satellite navigation, inertial navigation, and visual SLAM systems, positioning accuracy can be significantly improved. The advantages of each sensor complement each other, minimizing errors caused by local failure or insufficient accuracy of a single sensor, ensuring accurate positioning of the vehicle in complex environments. Beidou navigation fails in certain deep-sea areas or when obstructed by obstacles, inertial navigation will produce drift errors over time, and visual SLAM will be limited in low light or unclear environments. Through this fusion algorithm, the defects of a single sensor are effectively reduced, allowing the vehicle to continuously provide accurate positioning.

[0110] Path planning uses a convolutional neural network combined with lidar data for environmental perception and dynamically adjusts the planned path. The following is the algorithm formula for this path planning:

[0111] Environmental perception and path planning formula:

[0112]

[0113] Among them, P optimal The optimal path after convolutional neural network and lidar perception optimization, candidate points of P path, d goal The distance from the path point to the target point, d obstacle The distance from the path point to the surrounding obstacles, C smooth The smoothness penalty function of the P path, λ1, λ2, and λ3 control the weight coefficients of each factor.

[0114] By using convolutional neural networks to process the three-dimensional environmental data obtained by lidar, obstacles can be automatically identified and environmental complexity can be assessed. The generated path planning results are more intelligent than traditional algorithms. Convolutional neural networks can efficiently perceive key features in complex underwater environments, enabling the algorithm to maintain stable navigation capabilities in dynamically changing environments.

[0115] The BeiDou navigation-assisted global path correction module is used to automatically re-plan the path when the vehicle deviates from the planned path. The following is the algorithm formula involved:

[0116] Global path correction formula:

[0117] P new =P current +α·(P BDS -P current )+β·(P planned -P current ),

[0118] Among them, P new The corrected new path point, P current The position of the spacecraft, P BDSThe precise global location information obtained by the BeiDou navigation system, P planned The target point on the predetermined planning path, and the α and β weight coefficients.

[0119] When the aircraft deviates from the planned path, the global path correction is automatically performed. Through the global precise positioning provided by the Beidou navigation system and combined with the original planned path, the aircraft can adjust the path in time after discovering the deviation, ensuring that it always follows the global optimal route and avoiding path deviations caused by local obstacles or environmental changes. It effectively integrates the real-time positioning information of Beidou navigation and the planned planned path, and combines the information of the two to perform path correction. Beidou navigation provides the precise real-time position of the aircraft, while the planned path provides the ideal navigation target. Through the weight coefficient, it ensures that the aircraft always moves towards the target in a dynamic environment and maintains the continuity of the path.

[0120] Path smoothing and optimization uses the "arc-line-arc" strategy to generate a smooth path that takes into account the physical characteristics of the vehicle. The following is the algorithm formula used to calculate the smooth path:

[0121] Arc-straight line-arc path optimization formula:

[0122] Arc segment calculation: Among them, θ arc The angle of the arc segment, L arc The arc length of the arc segment, r is the turning radius of the vehicle;

[0123] Calculation of straight line segment length:

[0124] Among them, L l ine The straight-line distance between arc segments; the coordinates of the starting and ending points of the x1, y1 and x2, y2 straight-line segments;

[0125] Path smoothness optimization function:

[0126] C smooth =λ1·θ arc +λ2·L l ine -λ3·(1 / r),

[0127] Among them, C smooth Path smoothness cost function, λ1, λ2, λ3 weight coefficients.

[0128] The weights of the deep learning model are updated in real time during actual navigation. During operation, the aircraft obtains new environmental data through sensors, makes predictions based on path planning and obstacle avoidance decisions, compares the model's predicted results with the actual results, calculates the error, and adjusts the model weights based on the gradient descent algorithm. This process ensures that the model can quickly learn and adapt to new environmental changes during actual operation. Updating the model weights through the gradient descent method and minimizing the prediction error means that the model will accurately predict the aircraft's path planning and obstacle avoidance decisions, improving the overall accuracy and performance of the deep learning model.

[0129] Dynamic self-learning adjustment uses an online learning mechanism to update the weights of the deep learning model in real time. The following is the algorithm formula for this process:

[0130] Online learning weight update formula:

[0131]

[0132] Among them, w t+1 The updated model weight at time t+1, w t The model weight at time t, η learning rate, Loss function L on weight w t The gradient of the model indicates that the model is t and the true label y t The error derivative on ;

[0133] Real-time environment input:

[0134] x t : Input data collected by the sensor or vision system in this environment;

[0135] y t : The actual obstacle avoidance path or successful navigation result is used as a label in supervised learning to correct the model's prediction;

[0136] Loss function:

[0137] Among them, f(w t , x t ) The deep learning model is trained on the input x t The predicted output, y t Actual navigation or obstacle avoidance results.

[0138] The deep learning model's weights are updated in real time during actual flight. During flight, the vehicle acquires new environmental data through sensors and makes predictions based on path planning and obstacle avoidance decisions. By comparing the model's predictions with the actual results, the program ensures that the model can quickly learn and adapt to new environmental changes during actual flight. In a dynamic environment, environmental conditions are constantly changing, such as the sudden appearance of new obstacles or changes in the target's position. Through online learning mechanisms, the vehicle can continuously update the model's weights during real-time operation and dynamically adjust its obstacle avoidance strategy.

[0139] The controller performs dynamic adjustment of the underwater vehicle through multi-channel control signals. The following is the controller algorithm formula:

[0140] Propeller speed control formula: v new =v current +K v ·(v desired -v current ),

[0141] Among them, v new Adjusted thruster speed, v current Propeller speed, v desired Desired thruster speed, K v Speed ​​adjustment gain coefficient;

[0142] Heading control formula (based on proportional-integral-derivative control, PID):

[0143]

[0144] Among them, θ new Adjusted heading angle, θ current Heading angle, e(t) heading error, K p , K i , K d are the proportional, integral, and differential coefficients of the PID controller;

[0145] Attitude control formula: φ new =φ current +K φ ·(φ desired -φ current ),

[0146] Among them, φ new Adjusted attitude angle, v current Attitude angle, φ desired Desired attitude angle, K φ Attitude control gain coefficient;

[0147] Total control signal formula: u control =[v new ,θnew ,φ new ],

[0148] The total control signal formula represents the final control signal u control , is given by the propeller speed v new , heading angle θ new and attitude angle φ new The multi-channel control signals composed together are used to adjust the motion and attitude of the spacecraft.

[0149] Speed ​​control ensures that the vehicle moves at the optimal speed, prevents excessive speed fluctuations, and optimizes energy consumption. Heading control ensures that the vehicle navigates along the planned path and adjusts the heading through PID control to prevent the vehicle from deviating from the target route. Attitude control maintains the vehicle's attitude stable by dynamically adjusting the attitude angle, ensuring motion stability in complex underwater environments. The total control signal integrates various control signals to ensure that the vehicle can flexibly and smoothly adjust its speed, direction, and attitude, enabling it to operate efficiently in a dynamically changing underwater environment and achieve autonomous obstacle avoidance and path planning.

[0150] Data feedback and real-time optimization monitor the vehicle's sensor data and path execution in real time, and dynamically optimize path planning and obstacle avoidance strategies through feedback mechanisms. The following is the algorithm formula for real-time data feedback and optimization:

[0151] Real-time optimization feedback formula (based on feedback control theory):

[0152] ΔP adjust =K f ·(P desired -P current ),

[0153] Where ΔP adjust Correction value used to adjust path planning, P desired Expected path point, P current Actual path point, K f Feedback gain coefficient;

[0154] Path error calculation formula:

[0155]

[0156] Among them, e P (t) Path error, x desired 、y desired 、z desired The three-dimensional coordinates of the desired position, x current 、y current 、z current The three-dimensional coordinates of the actual position;

[0157] Real-time optimization adjustment formula:

[0158] P new =P current +ΔP adjust -η·e P (t),

[0159] Among them, P new The new path point after optimization, η learning rate.

[0160] Through the feedback control mechanism, the formula can adjust the path planning in time according to the actual operation of the vehicle, ensuring that the vehicle is quickly corrected when it deviates from the expected path. The dynamic adaptive adjustment mechanism can cope with the path deviation caused by uncertain factors in complex underwater environments, ensuring that the vehicle can navigate along the optimal path, and clearly measure the deviation between the vehicle position and the expected path point. The larger the error value, the farther the vehicle deviates from the planned path. The system makes corresponding adjustments based on the error to ensure that the navigation system can always monitor the path execution in real time to prevent serious deviations.

[0161] Path tracking and monitoring is based on the high-precision positioning data provided by the Beidou navigation system. It monitors the deviation between the vehicle and the target path in real time and makes timely adjustments. The following is the algorithm formula for path tracking and monitoring:

[0162] Path tracking error formula:

[0163]

[0164] Among them, e track (t) Tracking error between the vehicle and the target path point, x target 、y target 、z target The three-dimensional coordinates of the desired target point, x BDS 、y BDS 、z BDS The three-dimensional coordinates of the aircraft obtained through the Beidou navigation system;

[0165] Path correction formula: P adjust =P BDS +K p ·e track (t),

[0166] Among them, P adjust Adjusted waypoint, P BDS Beidou positioning coordinates of the spacecraft, K p Proportional gain;

[0167] Course adjustment formula: θ new =θ current +K θ ·(θ target -θ current ),

[0168] Among them, θ new Adjusted heading angle, θ current Vehicle heading angle, θ target Desired heading angle, K θ Proportional gain coefficient for heading adjustment;

[0169] Overall control update formula: u control =[P adjust ,θ new ],

[0170] Among them, u control Adjustment signals used to control the aircraft.

[0171] The path tracking error formula monitors the distance deviation between the aircraft and the target path point in real time to ensure that the aircraft can detect whether it has deviated from the planned route in time. The error value provides feedback information, indicating the extent of the path correction required by the aircraft; the path correction formula generates adjusted path points based on the error, dynamically corrects the position of the aircraft, and ensures that the aircraft returns to the correct route. Through the proportional gain, the strength of the correction can be controlled to avoid excessive correction or slow response; the heading adjustment formula is used to ensure that the direction of the aircraft is always moving towards the target. If the heading of the aircraft differs from the expected heading, the system adjusts the heading angle through this formula to keep the aircraft direction correct and ensure the accuracy of path tracking; the total control update formula integrates the path position and heading adjustment to generate a multi-channel control signal, which synchronously controls the position and heading of the aircraft, so that the aircraft can accurately execute the planned path and maintain the correct navigation state through real-time adjustment.

[0172] While 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 these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A deep learning-based underwater vehicle path planning and obstacle avoidance method, characterized in that: The steps include: , obtain multimodal data of the underwater environment through underwater sensors and the Beidou satellite navigation system, and generate high-precision environmental models in real time; , using multi-sensor fusion algorithms to combine the global positioning information provided by BeiDou satellites with sonar, inertial navigation systems and vision fusion of data; , process environmental perception data through deep learning models, use convolutional neural networks to identify and classify obstacles in the three-dimensional environment obtained by high-definition cameras and lidar, and predict navigation paths; Based on the deep reinforcement learning model, the obstacle avoidance path is planned in real time in a dynamic environment. The obstacle avoidance strategy is adjusted by the reward function in reinforcement learning. The specific formula is: ， in, is the reward value, is the distance between the aircraft and the obstacle, is the distance between the aircraft and the target point, is the collision penalty coefficient, 、 、 is the weight parameter; Utilizing the global position information provided by the BeiDou navigation system, if the vehicle deviates from the globally planned path, the global path correction module is automatically activated to re-plan the path based on BeiDou satellite data; The Beidou navigation-assisted global path correction module is used to automatically re-plan the path when the aircraft deviates from the predetermined path. The following is the algorithm formula involved: Global path correction formula: ， in, is the new path point after correction, is the position of the aircraft, Provides location information for Beidou navigation. is the target point on the planned path. 、 is the weighting coefficient; Based on the turning radius and physical size of the aircraft, the "arc-straight line-arc" path smoothing strategy is used to optimize the path output by deep learning; The path smoothing and optimization described above uses an "arc-straight-arc" strategy to generate a smooth path that takes into account the physical characteristics of the vehicle. The following is the algorithm formula for calculating the smooth path: Arc-straight line-arc path optimization formula: Arc segment calculation: ,in, is the angle of the arc segment, is the arc length of the arc segment, is the turning radius of the aircraft; Calculation of straight line segment length: , in, is the straight-line distance between arc segments; 、 and 、 are the coordinates of the start and end points of the straight line segment; Path smoothness optimization function: ， in, is the path smoothness cost function, 、 、 is the weight parameter; Through online learning mechanisms, the vehicle can make adaptive adjustments based on new environmental data during execution and update the weights of the deep learning model; , based on the results of the planned path and real-time obstacle avoidance output, generate motion control instructions for the aircraft, and control the propellers and steering gear of the aircraft to execute the movement; , During the operation of the aircraft, continuously obtain environmental perception data and update the model in real time; , monitor the deviation of the vehicle's position from the target path through the Beidou navigation system, and enable the emergency obstacle avoidance mode in an emergency.

2. The method for underwater vehicle path planning and obstacle avoidance based on deep learning according to claim 1, characterized in that: described In the step, the algorithm formula for correcting the positioning error is: ， in, is the corrected position, 、 、 For Beidou navigation, inertial navigation and vision Provided location information, 、 、 is the weighting coefficient.

3. The underwater vehicle path planning and obstacle avoidance method based on deep learning according to claim 1, characterized in that: The multi-sensor fusion algorithm is used to combine the Beidou navigation system, inertial navigation system and vision The data is used to enhance the positioning accuracy of the underwater vehicle. The following is the formula of the fusion algorithm: ， in, is the final positioning result after fusion, 、 、 For Beidou navigation, inertial navigation and vision Provided location information, 、 、 is the weighting coefficient.

4. The underwater vehicle path planning and obstacle avoidance method based on deep learning according to claim 1, characterized in that: The path planning uses a convolutional neural network combined with lidar data for environmental perception and dynamically adjusts the planned path. The following is the algorithm formula for this path planning: The environment perception and path planning formula: ， in, is the optimal path after optimization, is the candidate point of the path, is the distance between the aircraft and the obstacle, is the distance between the aircraft and the target point, is the path smoothness cost function, 、 、 is the weight parameter.

5. The underwater vehicle path planning and obstacle avoidance method based on deep learning according to claim 1, characterized in that: The dynamic self-learning adjustment uses an online learning mechanism to update the weights of the deep learning model in real time. The following is the algorithm formula for this process: Online learning weight update formula: ， in, For in time The model weights after updating at any moment, For in time The model weight at time t, is the learning rate, is the loss function Weight The gradient of the model indicates that the model and the true label The error derivative on ; Real-time environment input: : Input data collected by the sensor or vision system in this environment; : The actual obstacle avoidance path or successful navigation result is used as a label in supervised learning to correct the model's prediction; Loss function: , in, Input for deep learning models The predicted output of is the actual navigation or obstacle avoidance result.

6. The underwater vehicle path planning and obstacle avoidance method based on deep learning according to claim 1, characterized in that: The controller performs dynamic adjustment of the underwater vehicle through multi-channel control signals. The following is the controller algorithm formula: Propeller speed control formula: , in, is the adjusted propeller speed, is the propeller speed, is the desired propeller speed, Adjust the gain factor for speed; Heading control formula, : ， in, is the adjusted heading angle, is the heading angle, is the heading error, 、 、 for The proportional, integral and differential coefficients of the controller, is the integral variable; Posture control formula: , in, is the attitude angle after adjustment, is the attitude angle, is the desired attitude angle, is the attitude control gain coefficient; The total control signal formula is: , The total control signal formula represents the final control signal , is given by the propeller speed , heading angle and attitude angle The multi-channel control signals composed together are used to adjust the motion and attitude of the spacecraft.

7. The underwater vehicle path planning and obstacle avoidance method based on deep learning according to claim 1, characterized in that: Data feedback and real-time optimization monitor the vehicle's sensor data and path execution in real time, and dynamically optimize path planning and obstacle avoidance strategies through feedback mechanisms. The following is the algorithm formula for real-time data feedback and optimization: Real-time optimization feedback formula, based on feedback control theory: ， in, is the correction value used to adjust the path planning, is the desired path point, is the position of the aircraft, is the feedback gain coefficient; Path error calculation formula: ， in, is the path error, 、 、 is the three-dimensional coordinate of the desired position, 、 、 is the three-dimensional coordinate of the actual position; Real-time optimization adjustment formula: ， in, is the new path point after optimization, is the learning rate.

8. The underwater vehicle path planning and obstacle avoidance method based on deep learning according to claim 1, characterized in that: Path tracking and monitoring is based on the high-precision positioning data provided by the Beidou navigation system. It monitors the deviation between the vehicle and the target path in real time and makes timely adjustments. The following is the algorithm formula for path tracking and monitoring: Path tracking error formula: ， in, is the tracking error between the vehicle and the target path point, 、 、 is the three-dimensional coordinate of the desired target point, 、 、 The three-dimensional coordinates of the aircraft obtained through the BeiDou navigation system; Path correction formula: , in, is the adjusted path point, Provides location information for Beidou navigation. is the proportional gain; Course adjustment formula: , in, is the adjusted heading angle, is the heading angle of the aircraft, is the desired heading angle, Proportional gain coefficient for heading adjustment; Overall control update formula: , in, Adjustment signals used to control the aircraft.

Images