Automatic ship obstacle avoidance path planning method and device and storage medium

Abstract

The invention discloses an automatic boat obstacle avoidance path planning method and device and a storage medium, and is applied to the technical field of path planning, and the method comprises the steps: collecting the feature information of obstacles around an automatic boat and the sea surface environment information in real time; the obstacle feature information comprises first feature information of a dynamic obstacle and second feature information of a static obstacle; when the automatic boat executes the task of the wind power plant, a global obstacle avoidance path is formulated for the automatic boat by adopting a genetic algorithm according to the second feature information of the static obstacle and the sea surface environment information; loading a scheduling model of reinforcement learning; when the automatic boat executes the task according to the global obstacle avoidance path, optimizing the global obstacle avoidance path into a local obstacle avoidance path by adopting a reinforcement learning scheduling model according to the first feature information of the dynamic obstacle and the sea surface environment information, so that the automatic boat avoids the dynamic obstacle; and the local obstacle avoidance path is smoothed to obtain the target obstacle avoidance path, so that the adaptive capability of path planning is improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of path planning, and in particular, to an automatic boat obstacle avoidance path planning method, device and storage medium. Background Art

[0002] With the transformation of the global energy structure, offshore wind power, as a clean and sustainable energy source, has an increasing demand in the global energy market. To ensure the normal operation and maintenance of offshore wind farms, automatic boats are required to inspect and maintain the wind farms. However, the environment of offshore wind farms is quite complex. There are dense obstacles in the wind farm, including wind turbine blades, buoys, other boats, and potential underwater obstacles. When an automatic boat inspects and maintains the wind farm, it needs to avoid these obstacles.

[0003] Traditional obstacle avoidance methods generally rely on methods such as preset routes, the A* algorithm, and the artificial potential field method. However, the paths formulated by these methods are usually fixed and unchangeable. When facing the real-time changing sea surface environment and obstacles, the ability of the paths formulated by traditional methods to adaptively adjust according to the dynamic environment is poor, which may lead to low accuracy of path planning or low timeliness of obstacle avoidance, resulting in low navigation efficiency and even possible collision accidents. Summary of the Invention

[0004] The present invention provides an automatic boat obstacle avoidance path planning method, device and storage medium to improve the ability of the automatic boat obstacle avoidance path to adaptively adjust.

[0005] In a first aspect, an embodiment of the present invention provides an automatic boat obstacle avoidance path planning method, including:

[0006] Real-time collect the obstacle feature information and sea surface environment information around the automatic boat; the obstacle feature information includes the first feature information of dynamic obstacles and the second feature information of static obstacles;

[0007] When the automatic boat executes the tasks of the wind farm, formulate a global obstacle avoidance path for the automatic boat using a genetic algorithm based on the second feature information of the static obstacles and the sea surface environment information;

[0008] Load the scheduling model of reinforcement learning;

[0009] When the automatic boat executes the task according to the global obstacle avoidance path, optimize the global obstacle avoidance path into a local obstacle avoidance path using the scheduling model of reinforcement learning based on the first feature information of the dynamic obstacles and the sea surface environment information, so that the automatic boat can avoid the dynamic obstacles;

[0010] Smooth the local obstacle avoidance path to obtain a target obstacle avoidance path.

[0011] In a second aspect, an embodiment of the present invention further provides an automatic boat obstacle avoidance path planning device, including:

[0012] An environmental information acquisition module, configured to acquire in real time the obstacle feature information and the sea surface environmental information around the automatic boat; the obstacle feature information includes the first feature information of dynamic obstacles and the second feature information of static obstacles;

[0013] A global obstacle avoidance path determination module, configured to, when the automatic boat executes tasks in a wind farm, determine a global obstacle avoidance path for the automatic boat by using a genetic algorithm according to the second feature information of the static obstacles and the sea surface environmental information;

[0014] A scheduling model loading module, configured to load a scheduling model of reinforcement learning;

[0015] A local obstacle avoidance path optimization module, configured to, when the automatic boat executes the task according to the global obstacle avoidance path, optimize the global obstacle avoidance path into a local obstacle avoidance path by using the scheduling model of reinforcement learning according to the first feature information of the dynamic obstacles and the sea surface environmental information, so that the automatic boat can avoid the dynamic obstacles;

[0016] A target obstacle avoidance path acquisition module, configured to smooth the local obstacle avoidance path to obtain a target obstacle avoidance path.

[0017] In a third aspect, an embodiment of the present invention further provides a computer device, where the computer device includes:

[0018] One or more processors;

[0019] A storage device, configured to store one or more programs;

[0020] When the one or more programs are executed by the one or more processors, the one or more processors implement the automatic boat obstacle avoidance path planning method provided in the first aspect of the present invention.

[0021] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the automatic boat obstacle avoidance path planning method provided in the first aspect of the present invention.

[0022] In a fifth aspect, an embodiment of the present invention further provides a computer program product, where the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the automatic boat obstacle avoidance path planning method provided in the first aspect of the present invention.

[0023] In an embodiment of the present invention, the feature information of obstacles and the sea surface environment information around the autonomous boat are collected in real time; the feature information of obstacles includes the first feature information of dynamic obstacles and the second feature information of static obstacles; when the autonomous boat executes tasks in a wind farm, a global obstacle avoidance path is formulated for the autonomous boat by using a genetic algorithm based on the second feature information of static obstacles and the sea surface environment information; a scheduling model of reinforcement learning is loaded; when the autonomous boat executes the task according to the global obstacle avoidance path, the global obstacle avoidance path is optimized into a local obstacle avoidance path by using the scheduling model of reinforcement learning based on the first feature information of dynamic obstacles and the sea surface environment information, so that the autonomous boat can avoid dynamic obstacles; the local obstacle avoidance path is smoothed to obtain a target obstacle avoidance path. By collecting the feature information of obstacles and the sea surface environment information, the situation around the autonomous boat can be understood in real time, providing accurate data for subsequent path planning. When executing tasks in a wind farm, the genetic algorithm can explore multiple possible paths in a complex marine environment. The genetic algorithm adapts and adjusts the path by simulating the natural selection process to avoid local optimal solutions, and selects the best global obstacle avoidance path through multi-generation iteration optimization to cope with static obstacles and complex sea surface environments. On the basis of the global obstacle avoidance path, a scheduling model of reinforcement learning is introduced. The characteristic of reinforcement learning is that through continuous trial and error and reward mechanisms, the autonomous boat can continuously optimize its decision-making strategy in interaction with the environment. Especially when facing dynamic obstacles, it can continuously optimize and adjust the global obstacle avoidance path to obtain a local obstacle avoidance path, improving the adaptive ability of the autonomous boat to complex and dynamic environments, and ensuring that it can quickly and accurately avoid changing obstacles during actual operation. The local obstacle avoidance path is smoothed to remove sharp turns or unreasonable paths, making the navigation of the autonomous boat smoother and more fluent, thereby improving the navigation efficiency and enhancing the stability of the autonomous boat. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 FIG. is a flowchart of a method for planning an obstacle avoidance path of an autonomous boat provided in Embodiment 1 of the present invention;

[0025] Figure 2 FIG. is a structural block diagram of a device for planning an obstacle avoidance path of an autonomous boat provided in Embodiment 2 of the present invention;

[0026] Figure 3 FIG. is a structural schematic diagram of a computer device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] To enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned accompanying drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can cover arrangements other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] See Figure 1 , which shows a flowchart of an automatic boat obstacle avoidance path planning method provided in Embodiment 1 of the present invention. This embodiment is applicable to the case of planning a global obstacle avoidance path based on a genetic algorithm and adjusting the global obstacle avoidance path to a local obstacle avoidance path based on reinforcement learning. This method can be executed by an automatic boat obstacle avoidance path planning device, which can be implemented in the form of hardware and / or software, and the automatic boat obstacle avoidance path planning device can be configured in a computer device. As Figure 1 shown, the method includes:

[0031] Step 101, collect the obstacle feature information and sea surface environment information around the automatic boat in real time.

[0032] An autonomous boat refers to a vessel that can autonomously complete navigation and tasks without human intervention. Autonomous boats typically carry various sensors, cameras, radars, lidars, and other devices to sense the surrounding environment in real time and use advanced control systems and algorithms for path planning, obstacle avoidance, navigation, and other operations. Autonomous boats can navigate autonomously according to preset task objectives, such as transportation, monitoring, exploration, etc., and can make real-time adjustments according to environmental changes. For example, autonomous boats can avoid obstacles, cope with complex sea conditions, and even perform tasks such as wind farm maintenance and ocean surveys. The core advantages of autonomous boats are to improve efficiency, reduce labor costs, and replace manual operations in some high-risk environments to ensure safety. The hull of the autonomous boat is designed for wind and wave adaptability, made of high-strength composite materials, with excellent wind and wave resistance and impact resistance, and can operate stably in complex sea conditions, ensuring that the autonomous boat can continuously perform tasks for a long time in the harsh environment of the wind farm.

[0033] In this embodiment, by collecting the obstacle feature information and sea surface environment information around the autonomous boat in real time, real-time basic data is provided for subsequent obstacle avoidance path planning. The obstacle feature information includes the first feature information of dynamic obstacles and the second feature information of static obstacles, which can help the boat identify and avoid fixed or moving obstacles; the sea surface environment information, such as wind speed and wave height, can reflect the impact of the current navigation conditions on the behavior of the autonomous boat. By collecting this obstacle feature information and sea surface environment information in real time, the surrounding environment of the autonomous boat can be continuously and accurately understood, providing a basis for planning global and local obstacle avoidance paths, and ensuring that the autonomous boat can complete tasks safely and efficiently, especially in a complex wind farm environment.

[0034] In an embodiment of the present invention, step 101 may include the following steps:

[0035] Step 1011: Collect the object information and sea condition information in the surrounding environment of the autonomous boat.

[0036] In this embodiment, the object information includes the distance information between the object and the boat, the shape information of the object, and the image information of the object. These object information helps to judge whether potential obstacles will interfere with the navigation of the boat and determine the type of obstacles. The sea condition information includes sea surface wind speed and wave height, etc. These sea condition information affect the navigation stability and control strategy of the autonomous boat. For example, strong winds and high waves may cause the autonomous boat to sail unsteadily, thus affecting the planning of the obstacle avoidance path. Therefore, collecting the object information and sea condition information in the surrounding environment of the autonomous boat is to enable the autonomous boat to understand the changes in the surrounding environment in real time and provide support for path planning.

[0037] Exemplarily, through the collaborative work of multiple sensors such as radar, lidar, vision cameras, and ultrasonic sensors, the automatic boat collects object information and sea condition data in the surrounding environment. Radar can effectively detect distant obstacles, lidar provides high-precision three-dimensional environmental modeling, vision cameras are responsible for identifying object types and sea surface conditions, and ultrasonic sensors precisely detect nearby obstacles.

[0038] Step 1012: Perform preprocessing operations on the object information and sea condition information.

[0039] In this embodiment, after collecting the object information and sea condition information, these original object information and sea condition information usually contain noise or incomplete parts and need to be preprocessed before being used for subsequent analysis. The preprocessing operations usually include steps such as data denoising, outlier detection, and filling in missing data. Through these preprocessing operations, the accuracy and reliability of the object information and sea condition information can be improved, thus ensuring the accuracy of subsequent path planning and obstacle avoidance decisions. The purpose of preprocessing is to convert the original object information and sea condition information into a usable format and eliminate any unnecessary or incorrect information, enabling the automatic boat to perform environmental analysis more efficiently.

[0040] Step 1013: If the preprocessing operation is completed, use the Kalman filter algorithm to fuse the object information and sea condition information to determine the obstacle feature information of the obstacles around the automatic boat and the sea surface environment information when the automatic boat is traveling.

[0041] In this embodiment, if the preprocessing operation is completed, integrating data from different sensors (such as object information and sea condition information) through the Kalman filter algorithm can eliminate the possible errors or limitations of a single sensor. Through the complementarity of multi-source data, the accuracy of path decision-making can be improved. For example, a single sensor may not be able to accurately determine the shape of an object, but by fusing image information, distance information, etc., the object features can be more comprehensively identified. Similarly, the fusion of sea condition information helps to provide a comprehensive sea surface environment state, ensuring that the automatic boat can adjust its navigation strategy according to the latest sea conditions. Fusing the object information and sea condition information provides high-quality input data for subsequent global path planning and local obstacle avoidance optimization, ensuring that the automatic boat can efficiently and safely perform tasks in a complex environment.

[0042] The Kalman filter algorithm plays a role in prediction and update during the process of data fusion, calibrating the object information and sea condition information of multiple sensors. The Kalman filter algorithm can eliminate the noise and uncertainty in the data and optimize the estimation of states such as the target position and speed. At the same time, the Kalman filter algorithm also considers the signal-to-noise ratios of different data sources, weights the reliability of different information, and thus realizes the effective fusion of the object information and the sea condition information, obtaining the obstacle feature information of the obstacles around the autonomous boat and the sea surface environment information when the autonomous boat is traveling.

[0043] Step 102: When the autonomous boat executes the tasks of the wind farm, a global obstacle avoidance path is formulated for the autonomous boat according to the second feature information of the static obstacles and the sea surface environment information by using a genetic algorithm.

[0044] A wind farm refers to a wind power generation project built on the ocean, usually located in offshore or far - sea areas. It uses a group of wind turbines to utilize the strong and stable wind energy resources at sea and convert them into electrical energy. These wind turbines are usually installed on the seabed foundation, and the sea breeze drives the blades to rotate, thereby driving the generator to generate electricity. The wind farm provides one of the sources of renewable energy and can reduce the dependence on fossil fuels, which is one of the ways to cope with climate change and promote the development of green energy.

[0045] In this embodiment, using a genetic algorithm to formulate a global obstacle avoidance path for the autonomous boat is to ensure that the boat can effectively avoid static obstacles when executing the tasks of the wind farm. Static obstacles refer to objects that do not change their positions, such as seabed rocks, platform structures, etc. These static obstacles have a fixed impact on the navigation of the autonomous boat. According to the second feature information of the static obstacles and the sea surface environment information, a global obstacle avoidance path is formulated to ensure that the autonomous boat can safely avoid these static obstacles throughout the task process, avoiding collisions and navigation risks.

[0046] The genetic algorithm (GA) is an optimization method that simulates natural selection and genetic mechanisms. In the obstacle avoidance path planning of the autonomous boat, the genetic algorithm generates multiple driving paths (populations), and then evaluates each driving path according to its fitness. The driving path with a higher fitness is more likely to be selected. The algorithm continuously optimizes the driving paths through operations such as selection, crossover, and mutation until the optimal global obstacle avoidance path is found.

[0047] In an embodiment of the present invention, step 102 may include the following steps:

[0048] Step 1021: Randomly select multiple groups of driving paths from the predefined search space.

[0049] In this embodiment, within a predefined search space, multiple possible travel paths are provided for an autonomous boat to perform tasks in a wind farm. By randomly selecting multiple sets of travel paths, it is ensured that various possible situations can be covered, further increasing the chance of finding the optimal path. The search space is the set of all possible travel paths that the autonomous boat may adopt during the process of performing tasks in the wind farm, covering possible paths under various sea surface environment information and static obstacle conditions, enhancing the comprehensiveness and adaptability of the optimization results.

[0050] Exemplarily, the search space is the space composed of all possible travel paths from the starting point to the target point of the autonomous boat. This search space can be discrete or continuous. For a discrete search space, the travel path consists of a series of nodes, and each node represents a possible position or decision point that the autonomous boat may pass through; for a continuous search space, the travel path can be any continuous curve between the starting point and the target point. Multiple sets of travel paths are randomly selected, and these travel paths may be based on a certain connection method between the starting point and the target point (such as the grid method, rasterized map, etc.).

[0051] Step 1022: Initialize the population for multiple sets of travel paths.

[0052] In this embodiment, by initializing the population for multiple sets of travel paths, a starting point is provided for the genetic algorithm. Each individual in the population represents a travel path, and each travel path is represented by a set of codes. The coding method is usually the sequence of the path or the number sequence of the nodes. After these travel paths are adjusted and optimized in different ways, they can finally be selected as the optimal global obstacle avoidance path. Through this initialization, the genetic algorithm can start from multiple different paths and gradually evolve the optimal path that adapts to the environment and task requirements.

[0053] Step 1023: Calculate the fitness of the travel path for each individual according to the second feature information of the static obstacle, the sea surface environment information, and the smoothness of the travel path.

[0054] In this embodiment, the fitness of each travel path is evaluated by comprehensively considering the second feature information of the static obstacle, the sea surface environment information, and the smoothness of the travel path, ensuring that the selected travel path can not only avoid static obstacles but also adapt to the actual sea surface environment information, reduce unnecessary turns and mutations, and help the system identify which paths better meet the task requirements.

[0055] Exemplarily, the fitness is expressed as

[0056]

[0057] where f(x) is the fitness, f i (x) is an evaluation index related to the travel path, n is the number of evaluation indexes, ωi is the weight of the evaluation index; the evaluation index is calculated by weighted summation and other methods using parameters such as the second feature information of static obstacles, sea surface environment information, and the smoothness of the driving path (such as the height of sea waves, etc.). The evaluation index includes path length, obstacle avoidance ability, path smoothness, real-time sea conditions, etc. In the evaluation of fitness, if the driving path crosses a static obstacle or collides with a static obstacle, the fitness value will be relatively low. In some cases, some driving paths may not be suitable for the autonomous boat to perform tasks due to strong winds or large waves. The fitness function will consider this sea surface environment information and reduce the fitness value of the unsuitable driving paths. The smoothness of the driving path is considered in the fitness function to avoid overly sharp turns and overly long path segments to ensure that the driving path conforms to the dynamic characteristics of the autonomous boat.

[0058] Step 1024: Determine whether the preset optimization condition is satisfied; if so, execute Step 1025; if not, execute Step 1026.

[0059] In this embodiment, one or more optimization conditions can be preset. For example, the number of iterations reaches the upper limit, the change range of fitness in consecutive iterations is lower than a certain threshold, etc. In each iteration, it can be determined whether at least one optimization condition is satisfied.

[0060] If at least one optimization condition is satisfied, the genetic algorithm can be stopped.

[0061] If all optimization conditions are not satisfied, the genetic algorithm can continue, generate the next generation, and enter the next iteration.

[0062] Step 1025: Use the individual with the highest fitness as the global obstacle avoidance path when the autonomous boat performs the wind farm task.

[0063] In this embodiment, when the genetic algorithm is stopped, the individual with the highest fitness is selected from the population optimized through multiple iterations as the global obstacle avoidance path when the autonomous boat performs the wind farm task. This driving path can avoid static and dynamic obstacles to the greatest extent and adapt to sea condition changes to ensure the smooth completion of the autonomous boat task.

[0064] Step 1026: Perform selection operation, crossover operation, and mutation operation on the individuals to update the individuals, and return to execute Step 1023.

[0065] When generating the next generation, the following operations can be sequentially performed on the individuals:

[0066] 1. Selection operation: Adopt strategies such as roulette wheel selection and tournament selection to select a certain proportion of individuals from the current population to enter the next generation, preferring individuals with high fitness.

[0067] 2. Crossover: Through single-point crossover or uniform crossover operations, exchange some components or parameters of the selected individuals to simulate gene recombination in the biological genetic process.

[0068] 3. Mutation: Randomly change the genes in individuals to increase diversity.

[0069] When completing the next generation of inheritance, steps 1023 - 1024 can be re-executed to continue iterative training until at least one optimization condition is met.

[0070] Step 103: Load the scheduling model of reinforcement learning.

[0071] In this embodiment, the scheduling model of reinforcement learning is loaded into the system of the autonomous boat. The scheduling model of reinforcement learning is used to optimize the obstacle avoidance behavior of the autonomous boat when performing tasks. By continuously updating the scheduling model, the scheduling model can adjust the driving path of the autonomous boat according to the actual situation, thereby improving the efficiency and accuracy of obstacle avoidance and ensuring that the autonomous boat performs tasks in the wind farm in a complex sea environment.

[0072] Step 104: When the autonomous boat performs tasks according to the global obstacle avoidance path, optimize the global obstacle avoidance path into a local obstacle avoidance path based on the first feature information of the dynamic obstacle and the sea environment information by using the scheduling model of reinforcement learning, so that the autonomous boat can avoid the dynamic obstacle.

[0073] In this embodiment, the global obstacle avoidance path is formulated based on the second feature information of static obstacles and sea environment information collected in real time. However, during actual navigation, dynamic obstacles (such as other ships, floating objects, etc.) may suddenly appear on the path. The global path cannot respond to these dynamic changes in real time. Therefore, it is necessary to adjust the global obstacle avoidance path. The scheduling model of reinforcement learning can adjust the behavior of the autonomous boat according to the current state and environment information through real-time learning and adaptation, avoiding collisions.

[0074] The planning of the global obstacle avoidance path and the planning of the local obstacle avoidance path are seamlessly connected through an end-to-end optimization process to ensure that the autonomous boat always maintains the optimal driving path in a complex environment.

[0075] Reinforcement Learning (RL) is a machine learning method. It enables an agent to interact with the environment and uses reward feedback to gradually learn how to take actions. Its core idea is to guide the agent to select appropriate behaviors according to the state through rewards and punishments, thereby maximizing its cumulative return.

[0076] An Agent is a subject that executes tasks and makes decisions. In reinforcement learning, it is responsible for interacting with the environment and taking actions to achieve goals. In this invention, the autonomous boat is the Agent of reinforcement learning. The autonomous boat executes tasks according to the global obstacle avoidance path and optimizes the local obstacle avoidance path through reinforcement learning based on the obstacle information in the environment.

[0077] The Environment is the external system in which the Agent is located. The Agent perceives information and takes actions in the environment, and the environment will feedback information according to the Agent's behavior. The environment is the sea environment where the autonomous boat is located, including the first feature information of dynamic obstacles and the sea environment information (such as wind speed, wave height, etc.). The boat adjusts its navigation strategy by collecting this information in real time.

[0078] The State describes the current situation of the Agent in the environment, and it contains all the information required for the Agent to execute actions. In this invention, the State is the current navigation state of the autonomous boat. This information is used as input into the reinforcement learning model to optimize the path planning of the autonomous boat.

[0079] An Action is an operation taken by the Agent in a certain State, aiming to change the state of the environment or optimize the decision-making goal. In the reinforcement learning scheduling model, the action of the autonomous boat is to avoid obstacles and adjust the heading.

[0080] The Reward is the feedback obtained by the Agent according to the current State and the actions taken, reflecting the degree of achievement of the goal by this action. The Reward is the degree to which the autonomous boat avoids dynamic obstacles, maintains a safe distance, and completes tasks during navigation. For each successful obstacle avoidance or path optimization, the autonomous boat will obtain a positive reward; while if the boat collides or deviates from the optimal path, it may receive a negative reward. The reinforcement learning model gradually optimizes the decision-making process of the autonomous boat through this reward mechanism.

[0081] In an embodiment of the present invention, step 104 may include the following steps:

[0082] Step 1041: According to the first feature information of dynamic obstacles and the sea environment information, collect the state information generated by the autonomous boat during navigation to construct the state space in the reinforcement learning model.

[0083] In this embodiment, based on the first feature information of dynamic obstacles collected in real time and the sea surface environment information, when the autonomous boat is traveling on the global obstacle avoidance path, the state information generated during the travel of the autonomous boat is collected to ensure that the scheduling model of reinforcement learning can fully understand the complex situations and dynamically changing environment faced by the autonomous boat during actual navigation. The first feature information of dynamic obstacles includes changes such as the movement trajectory, speed, and direction of dynamic obstacles, and the sea surface environment information covers factors such as wind speed, sea waves, and ocean currents, which affect the movement state of the autonomous boat.

[0084] The state space is an integral part of reinforcement learning, which determines the various situations that the agent, here the autonomous boat, may face at each time step. By collecting state information related to the travel of the autonomous boat (such as dynamic obstacle and sea surface environment information, etc.), the reinforcement learning model can perceive the environment where the autonomous boat is currently located and make corresponding decisions based on this.

[0085] Step 1042: Set execution actions for the autonomous boat to avoid dynamic obstacles on the global obstacle avoidance path to construct the action space in the reinforcement learning model.

[0086] In this embodiment, in reinforcement learning, the action space contains all the actions that the agent may choose. In the obstacle avoidance path planning of the autonomous boat, the action space defines the execution actions of the autonomous boat, such as the autonomous boat turning left, the autonomous boat turning right, the autonomous boat accelerating, and the autonomous boat decelerating. Through these execution actions, the autonomous boat can flexibly avoid dynamic obstacles.

[0087] Step 1043: Input the state space into the scheduling model of reinforcement learning, and select an execution action adapted to the state space for the autonomous boat in the action space to enable the autonomous boat to avoid dynamic obstacles.

[0088] In this embodiment, after inputting the state space into the scheduling model of reinforcement learning, the scheduling model will evaluate which execution actions are most suitable for the current situation based on the current state information. The scheduling model needs to make a decision based on the state information and select an action that can best avoid obstacles and ensure the smooth completion of the task.

[0089] Reinforcement learning is a continuously optimizing process. Whenever the autonomous boat executes a certain action, the system will adjust the parameters of the scheduling model according to the feedback, so that the autonomous boat can optimize its travel path and execution actions according to the real-time sea surface environment information and obstacle changes. Continuously improve the decision-making ability of the autonomous boat in a complex dynamic environment. By repeatedly selecting appropriate execution actions and evaluating the effects, the agent (autonomous boat) can gradually learn how to more effectively respond to dynamic obstacles and sea surface environment information, and improve the efficiency of path planning and obstacle avoidance.

[0090] Exemplarily, the Q-learning algorithm is used to update the scheduling model of reinforcement learning. The Q-learning algorithm is a value-based reinforcement learning algorithm that learns the optimal policy by updating the Q-value. The Q-value represents the expected return of a state-action pair, and the purpose of updating the Q-value is to enable the autonomous boat to select those execution actions with the best long-term returns. Through Q-learning, the autonomous boat can adaptively learn how to avoid obstacles in a dynamic environment and continuously optimize the obstacle avoidance strategy.

[0091] When reinforcement learning is used for local obstacle avoidance path generation, it mainly learns how to select appropriate execution actions from the current state through the interaction between the autonomous boat and the environment to adjust the local driving path. After each adjustment, the autonomous boat will optimize the selection of the next driving path according to the obtained reward. When the autonomous boat moves forward along the global obstacle avoidance path, the local obstacle avoidance path planning will adjust the driving path in real time according to the sensor data. The core feature of the Q-learning algorithm is that the agent does not need to know the complete model of the environment in advance, but learns the strategy through interaction with the environment. The autonomous boat continuously executes actions during driving and adjusts the strategy according to the feedback, gradually improving the obstacle avoidance efficiency. After each optimization, the autonomous boat can make better decisions in future similar situations. Q-learning helps the agent avoid getting stuck in local optima by repeatedly iterating and updating the Q-value. This process ensures that the autonomous boat can make optimal path planning and obstacle avoidance decisions according to the latest environmental information in a complex and dynamic environment.

[0092] Exemplarily, the Q-learning algorithm is expressed as:

[0093]

[0094] where Q(s,a) is the expected reward value generated when executing action a in the state space s, α is the learning rate, r is the immediate reward obtained after executing the action, γ is the discount factor, is the maximum expected reward value for executing action a' in the next state space s'. Whenever the autonomous boat successfully avoids a dynamic obstacle and moves towards the target, the immediate reward is positive; if it collides with a dynamic obstacle or deviates from the target path, the immediate reward is negative. Then, after each execution of the action is completed, the Q-value is updated according to the current state, the executed action, and the obtained immediate reward. The updated Q-value will guide the selection of the next driving path. For example, if the Q-value of the currently selected driving path is high, it means that this driving path is relatively superior, and the autonomous boat is more inclined to move forward along this driving path; conversely, if the Q-value of a certain path is low, it means that this driving path is not ideal, and the autonomous boat will choose another driving path.

[0095] In the present invention, the system collects the surrounding obstacle feature information and sea surface environment information in real time through multiple sensors (such as radar, LiDAR, vision cameras, and ultrasonic sensors), and analyzes it in real time to ensure a rapid response to changes in the dynamic environment. For the real-time dynamic obstacle and sea surface environment information, the global obstacle avoidance path is gradually optimized and adjusted through a scheduling algorithm based on reinforcement learning, so that the autonomous boat can efficiently execute the tasks of the wind farm.

[0096] Step 105: Smooth the local obstacle avoidance path to obtain the target obstacle avoidance path.

[0097] In this embodiment, during the actual navigation process, the autonomous boat may encounter many sharp turns or irregular paths, especially in a complex sea surface environment. Sharp turns and sharp changes in the path will increase the risk of the autonomous boat rolling over and even affect the stability of the equipment. By smoothing the local obstacle avoidance path, these sharp turns and irregular trajectories can be reduced, thereby improving the stability and safety of navigation.

[0098] In an embodiment of the present invention, Step 105 may include the following steps:

[0099] Step 1051: Collect key control points for the local obstacle avoidance path.

[0100] In this embodiment, the key control points at least include path turning points and path intersection points. Path turning points are usually places where the autonomous boat needs to change direction during navigation, while path intersection points are places where multiple paths intersect or bifurcate. The collection of these key control points helps in the subsequent smoothing of the local obstacle avoidance path.

[0101] Exemplarily, through path tracking technology, key control points such as path turning points and path intersection points in the local obstacle avoidance path can be identified. These key control points are usually places where the path curvature changes greatly and can be automatically detected by calculating the curvature or derivative of the path.

[0102] Step 1052: Construct a B-spline curve based on the key control points.

[0103] In this embodiment, the B-spline curve is a smoothing curve method widely used in computer graphics and path planning. Its advantage is that it can ensure the smoothness and continuity of the path while avoiding sharp corners. By constructing a B-spline curve based on the key control points, a natural and smooth transition can be generated in the navigation path, thereby preventing the autonomous boat from becoming unstable or consuming too much energy due to path turning points and path intersection points during task execution.

[0104] Exemplarily, the B-spline curve is expressed as:

[0105]

[0106] Among them, P(t) is the position of the B-spline curve at the given position parameter t, n is the number of key control points, N i,k (t) is the B-spline basis function, P i is the key control point, and k is the order of the B-spline curve.

[0107] Step 1053: Smooth the sharp-turn path and the abnormal path in the local obstacle avoidance path through the B-spline curve to obtain the target obstacle avoidance path.

[0108] In this embodiment, during the actual navigation process, the sharp-turn path or the abnormal path often causes the navigation of the autonomous boat to be unstable, and may even cause the autonomous boat to lose control, increase energy consumption or damage equipment. Smoothing these sharp-turn paths or abnormal paths using the B-spline curve can enable the autonomous boat to avoid sharp direction changes or inappropriate path adjustments during the obstacle avoidance process, improving the safety and reliability of navigation.

[0109] When the autonomous boat sails on the sea surface along the target obstacle avoidance path, adjust the navigation parameters (such as speed, heading, and acceleration) of the autonomous boat according to the obstacle characteristic information and the sea surface environment information to ensure the safety and efficiency of navigation. This adjustment can help the boat better adapt to the dynamically changing environment, avoid obstacles, and optimize its operating performance.

[0110] In an embodiment of the present invention, the characteristic information of obstacles and the sea surface environment information around the autonomous boat are collected in real time; the characteristic information of obstacles includes the first characteristic information of dynamic obstacles and the second characteristic information of static obstacles; when the autonomous boat executes the task of the wind farm, a global obstacle avoidance path is formulated for the autonomous boat by using the genetic algorithm according to the second characteristic information of static obstacles and the sea surface environment information; a scheduling model of reinforcement learning is loaded; when the autonomous boat executes the task according to the global obstacle avoidance path, the global obstacle avoidance path is optimized into a local obstacle avoidance path by using the scheduling model of reinforcement learning according to the first characteristic information of dynamic obstacles and the sea surface environment information, so that the autonomous boat can avoid dynamic obstacles; the local obstacle avoidance path is smoothed to obtain a target obstacle avoidance path. By collecting the characteristic information of obstacles and the sea surface environment information, the situation around the autonomous boat can be understood in real time, providing accurate data for subsequent path planning. When executing the wind farm task, the genetic algorithm can explore multiple possible paths in a complex marine environment. The genetic algorithm simulates the natural selection process, adaptively adjusts the path to avoid local optimal solutions, and selects the best global obstacle avoidance path through multi-generation iteration to cope with static obstacles and complex sea surface environments. On the basis of the global obstacle avoidance path, a scheduling model of reinforcement learning is introduced. The characteristic of reinforcement learning is that through continuous trial and error and reward mechanisms, the autonomous boat can continuously optimize its decision-making strategy in interaction with the environment. Especially when facing dynamic obstacles, it can continuously optimize and adjust the global obstacle avoidance path to obtain a local obstacle avoidance path, improving the adaptive ability of the autonomous boat to complex and dynamic environments, and ensuring that it can quickly and accurately avoid changing obstacles during actual operation. The local obstacle avoidance path is smoothed to remove sharp turns or unreasonable paths, making the navigation of the autonomous boat smoother and more fluent, thereby improving the navigation efficiency and enhancing the stability of the autonomous boat.

[0111] Embodiment 2

[0112] Figure 2 The following is a schematic structural diagram of an obstacle avoidance path planning device for an autonomous boat provided in Embodiment 2 of the present invention, as Figure 2 shown, the device includes:

[0113] An environmental information collection module 201, configured to collect the characteristic information of obstacles and the sea surface environment information around the autonomous boat in real time; the characteristic information of obstacles includes the first characteristic information of dynamic obstacles and the second characteristic information of static obstacles;

[0114] A global obstacle avoidance path formulation module 202, configured to, when the autonomous boat executes the task of the wind farm, formulate a global obstacle avoidance path for the autonomous boat by using the genetic algorithm according to the second characteristic information of the static obstacles and the sea surface environment information;

[0115] A scheduling model loading module 203, configured to load a scheduling model for reinforcement learning;

[0116] A local obstacle avoidance path optimization module 204, configured to, when the autonomous boat executes the task according to the global obstacle avoidance path, optimize the global obstacle avoidance path into a local obstacle avoidance path by using the scheduling model of the reinforcement learning according to the first feature information of the dynamic obstacle and the sea surface environment information, so that the autonomous boat can avoid the dynamic obstacle;

[0117] A target obstacle avoidance path obtaining module 205, configured to smooth the local obstacle avoidance path to obtain a target obstacle avoidance path.

[0118] In an embodiment of the present invention, the environment information acquisition module 201 includes:

[0119] A boat surrounding environment information acquisition module, configured to acquire object information and sea condition information in the surrounding environment of the autonomous boat; the object information includes distance information between the object and the autonomous boat, shape information of the object, and image information of the object; the sea condition information at least includes sea surface wind speed and wave height;

[0120] A preprocessing module, configured to perform a preprocessing operation on the object information and the sea condition information;

[0121] An information fusion module, configured to, if the preprocessing operation is completed, perform data fusion on the object information and the sea condition information by using a Kalman filtering algorithm to determine obstacle feature information of obstacles around the autonomous boat and sea surface environment information when the autonomous boat is traveling.

[0122] In an embodiment of the present invention, the global obstacle avoidance path determination module 202 includes:

[0123] A travel path selection module, configured to randomly select multiple groups of travel paths from a predefined search space; the search space is a set of all the travel paths when the autonomous boat executes a wind farm task;

[0124] A population initialization module, configured to initialize a population for multiple groups of the travel paths; each individual in the population represents one of the travel paths;

[0125] A fitness calculation module, configured to calculate the fitness of each of the travel paths according to the second feature information of the static obstacle, the sea surface environment information, and the smoothness of the travel path;

[0126] An optimization judgment module, configured to judge whether a preset optimization condition is satisfied; if so, execute an individual selection module; if not, execute an individual update module;

[0127] An individual selection module, configured to use the individual with the highest fitness as the global obstacle avoidance path when the autonomous boat performs the wind farm task;

[0128] An individual update module, configured to perform selection operations, crossover operations, and mutation operations on the individual to update the individual, and return to execute the fitness calculation module.

[0129] In an embodiment of the present invention, the fitness is expressed as:

[0130]

[0131] where f(x) is the fitness, f i (x) is an evaluation index related to the driving path, n is the number of evaluation indexes, and ω i is the weight of the evaluation index.

[0132] In an embodiment of the present invention, the local obstacle avoidance path optimization module 204 includes:

[0133] A state space loading module, configured to collect state information generated by the autonomous boat during driving according to the first feature information of the dynamic obstacle and the sea surface environment information, so as to construct a state space in the reinforcement learning model;

[0134] An action space setting module, configured to set execution actions for the autonomous boat to avoid the dynamic obstacle on the global obstacle avoidance path, so as to construct an action space in the reinforcement learning model; the execution actions at least include turning left, turning right, accelerating, and decelerating;

[0135] An execution action selection module, configured to input the state space into the scheduling model of reinforcement learning, and select, in the action space, the execution action adapted to the state space for the autonomous boat, so that the autonomous boat avoids the dynamic obstacle.

[0136] In an embodiment of the present invention, the local obstacle avoidance path optimization module 204 further includes:

[0137] A scheduling model update module, configured to update the scheduling model of the reinforcement learning by using the Q-learning algorithm.

[0138] The Q-learning algorithm is expressed as:

[0139]

[0140] where Q(s,a) is the expected reward value generated when the execution action a is performed in the state space s, α is the learning rate, r is the immediate reward obtained after performing the execution action, and γ is the discount factor, is the maximum expected reward value for executing the execution action a' in the next state space s'.

[0141] In one embodiment of the present invention, the target obstacle avoidance path acquisition module 205 includes:

[0142] A key control point acquisition module, configured to acquire key control points for the local obstacle avoidance path; the key control points at least include path turning points and path intersection points;

[0143] A curve construction module, configured to construct a B-spline curve based on the key control points;

[0144] A path smoothing module, configured to smooth the sharp turn paths and abnormal paths in the local obstacle avoidance path through the B-spline curve to obtain a target obstacle avoidance path;

[0145] The B-spline curve is expressed as:

[0146]

[0147] where P(t) is the position of the B-spline curve at a given position parameter t, n is the number of the key control points, N i,k (t) is the B-spline basis function, and P i is the key control point, and k is the order of the B-spline curve.

[0148] In one embodiment of the present invention, the device further includes:

[0149] A navigation parameter adjustment module, configured to adjust the navigation parameters of the autonomous boat according to the obstacle feature information and the sea surface environment information when the autonomous boat sails on the sea along the target obstacle avoidance path; the navigation parameters at least include speed, heading, and acceleration.

[0150] The autonomous boat obstacle avoidance path planning device provided by the embodiments of the present invention can execute the autonomous boat obstacle avoidance path planning method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the autonomous boat obstacle avoidance path planning method.

[0151] Embodiment III

[0152] See Figure 3, which shows a schematic structural diagram of a computer device provided by an embodiment of the present invention. The computer device is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, blade servers, mainframe computers, and other suitable computers. The computer device can also represent various forms of mobile devices, such as, personal digital processors, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described herein and / or claimed.

[0153] As Figure 3 shown, the computer device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. Among them, the memory stores a computer program executable by the at least one processor. The processor 11 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the computer device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0154] Multiple components in the computer device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disc, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the computer device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0155] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the automatic boat obstacle avoidance path planning method.

[0156] In some embodiments, the automatic boat obstacle avoidance path planning method may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed onto the computer device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by the processor 11, one or more steps of the automatic boat obstacle avoidance path planning method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the automatic boat obstacle avoidance path planning method by any other suitable means (e.g., by means of firmware).

[0157] The various embodiments of the systems and techniques described above in this document may be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs executable and / or interpretable on a programmable system including at least one programmable processor, which may be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0158] The computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine, or entirely on the remote machine or server.

[0159] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0160] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0161] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0162] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The relationship between the client and the server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0163] Embodiment 4

[0164] The embodiment of the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the automatic boat obstacle avoidance path planning method provided in any embodiment of the present invention.

[0165] In the process of implementing the computer program product, the computer program code for performing the operations of the present invention can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN) - or can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0166] It should be understood that the various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is imposed herein.

[0167] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An automatic boat obstacle avoidance path planning method, characterized in that, Including: Real-time collecting the obstacle feature information and sea surface environment information around the autonomous boat; The obstacle feature information includes the first feature information of dynamic obstacles and the second feature information of static obstacles; When the autonomous boat executes the task of the wind farm, formulating a global obstacle avoidance path for the autonomous boat by using a genetic algorithm according to the second feature information of the static obstacles and the sea surface environment information; Loading a scheduling model of reinforcement learning; When the autonomous boat executes the task according to the global obstacle avoidance path, optimizing the global obstacle avoidance path into a local obstacle avoidance path by using the scheduling model of reinforcement learning according to the first feature information of the dynamic obstacles and the sea surface environment information, so that the autonomous boat can avoid the dynamic obstacles; Smoothing the local obstacle avoidance path to obtain a target obstacle avoidance path.

2. The method according to claim 1, wherein The real-time collecting the obstacle feature information and sea surface environment information around the autonomous boat includes: Collecting the object information and sea condition information in the environment around the autonomous boat; the object information includes the distance information between the object and the autonomous boat, the shape information of the object, and the image information of the object; the sea condition information at least includes the sea surface wind speed and the wave height; Performing a preprocessing operation on the object information and the sea condition information; If the preprocessing operation is completed, fusing the object information and the sea condition information through a Kalman filtering algorithm to determine the obstacle feature information of the obstacles around the autonomous boat and the sea surface environment information when the autonomous boat is traveling.

3. The method according to claim 2, wherein The formulating a global obstacle avoidance path for the autonomous boat by using a genetic algorithm according to the second feature information of the static obstacles and the sea surface environment information includes: Randomly selecting multiple groups of travel paths from a predefined search space; the search space is the set of all the travel paths when the autonomous boat executes the task of the wind farm; Initializing a population for multiple groups of the travel paths; each individual in the population represents a travel path; Calculating the fitness of each travel path according to the second feature information of the static obstacles, the sea surface environment information, and the smoothness of the travel path; Judging whether a preset optimization condition is satisfied; if so, taking the individual with the highest fitness as the global obstacle avoidance path when the autonomous boat executes the task of the wind farm; if not, performing a selection operation, a crossover operation, and a mutation operation on the individual to update the individual, and returning to execute the calculating the fitness of each travel path according to the second feature information of the static obstacles, the sea surface environment information, and the smoothness of the travel path.

4. The method according to claim 3, wherein The fitness is expressed as: Among them, f(x) is the fitness, and f i (x) is an evaluation index related to the driving path, n is the number of the evaluation indexes, and ω i is the weight of the evaluation index.

5. The method according to claim 1, characterized in that When the autonomous boat executes the task according to the global obstacle avoidance path, optimizing the global obstacle avoidance path into a local obstacle avoidance path by using the scheduling model of reinforcement learning according to the first feature information of the dynamic obstacles and the sea surface environment information, so that the autonomous boat can avoid the dynamic obstacles, includes: Collect the state information generated by the autonomous boat during driving based on the first characteristic information of the dynamic obstacle and the sea surface environment information to construct the state space in the reinforcement learning model; Set execution actions for the autonomous boat to avoid the dynamic obstacle on the global obstacle avoidance path to construct the action space in the reinforcement learning model; the execution actions at least include turning left, turning right, accelerating, and decelerating; Input the state space into the scheduling model of reinforcement learning, and select the execution action adapted to the state space for the autonomous boat in the action space so that the autonomous boat can avoid the dynamic obstacle.

6. The method according to claim 5, characterized in that It further includes: Update the scheduling model of reinforcement learning using the Q-learning algorithm; The Q-learning algorithm is expressed as: Among them, Q(s,a) is the expected reward value generated when performing the execution action a in the state space s, α is the learning rate, r is the immediate reward obtained after performing the execution action, and γ is the discount factor. It is the maximum expected reward value for performing the execution action a' in the next state space s'.

7. The method according to claim 1, wherein The smoothing of the local obstacle avoidance path to obtain the target obstacle avoidance path includes: Collect key control points for the local obstacle avoidance path; the key control points at least include path turning points and path intersection points; Construct a B-spline curve based on the key control points; Smooth the sharp turning path and the abnormal path in the local obstacle avoidance path through the B-spline curve to obtain the target obstacle avoidance path; The B-spline curve is expressed as: Among them, P(t) is the position of the B-spline curve at the given position parameter t, n is the number of the key control points, N i,k (t) is the B-spline basis function, P i is the key control point, and k is the order of the B-spline curve.

8. The method according to any one of claims 1-7, characterized in that, It further includes: When the autonomous boat sails on the sea surface along the target obstacle avoidance path, adjust the navigation parameters of the autonomous boat according to the obstacle characteristic information and the sea surface environment information; the navigation parameters at least include speed, heading, and acceleration.

9. A computer device, characterized in that, The computer device includes: One or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the autonomous boat obstacle avoidance path planning method according to any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the autonomous boat obstacle avoidance path planning method according to any one of claims 1-8.

Images