Intelligent perception obstacle avoidance planning method and system for underwater obstacles

Abstract

The application provides an intelligent perception obstacle avoidance planning method and system for underwater obstacles, and relates to the technical field of underwater navigation. First, real-time navigation parameters of an underwater navigation carrier and underwater environment trigger signals are received to dynamically define a perception range. A multi-mode perception device is driven to collect multi-dimensional data and perform fusion processing to generate dynamic correlation knowledge features of obstacles. An obstacle avoidance path evolution mechanism is constructed in combination with the dynamic correlation knowledge features of obstacles and the real-time navigation parameters to generate an initial obstacle avoidance path. The initial obstacle avoidance path is adjusted based on the motion state of the obstacles to obtain a target obstacle avoidance path. The target obstacle avoidance path is associated and mapped with the real-time navigation parameters to generate control instructions to drive the carrier to navigate, thereby significantly improving the obstacle avoidance capability and navigation safety of the underwater navigation carrier in a complex underwater environment.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more specifically, to an intelligent perception and obstacle avoidance planning method and system for underwater obstacles. Background Technology

[0002] With the deepening of marine exploration and development, underwater vehicles are increasingly used in marine scientific research, resource exploration, and military reconnaissance. However, the underwater environment is complex and changeable, with many unknown underwater obstacles, such as reefs, shipwrecks, and underwater building debris, which pose a serious threat to the safe navigation of underwater vehicles.

[0003] Currently, obstacle avoidance technology for underwater vehicles faces numerous limitations. On one hand, existing sensing methods often employ single-mode sensing devices, such as relying solely on acoustic detection or optical imaging equipment. Single-mode sensing devices acquire limited information; acoustic detection devices are prone to signal interference and misjudgment in complex underwater terrain or multi-target environments, while optical imaging equipment is significantly affected by factors such as water turbidity and lighting conditions, making it difficult to comprehensively and accurately acquire information about underwater obstacles. On the other hand, traditional obstacle avoidance planning methods lack real-time response capabilities to dynamic changes in the underwater environment. When faced with dynamic situations such as water flow disturbances and obstacle movement, they cannot adjust the obstacle avoidance path in a timely manner, resulting in poor obstacle avoidance performance and potentially even collisions. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide an intelligent perception and obstacle avoidance planning method for underwater obstacles, the method comprising: The system receives real-time navigation parameters and underwater environment trigger signals from an underwater vehicle. Based on the real-time navigation parameters and the underwater environment trigger signals, it dynamically defines the sensing range of the underwater environment and generates a dynamic definition result of the sensing range. The real-time navigation parameters include information related to the underwater vehicle's heading, attitude, and power output status. The underwater environment trigger signals include underwater water flow disturbance signals and ambient light change signals. Based on the dynamic definition of the sensing range, the multi-mode sensing devices carried by the underwater vehicle are driven to work together to collect multi-dimensional underwater environmental data within the sensing range. The multi-dimensional underwater environmental data is then correlated and fused to generate obstacle dynamic association knowledge features. The multi-dimensional underwater environmental data includes acoustic detection data and optical imaging data. The obstacle dynamic association knowledge features include spatial position association, motion state association, and material property association of obstacles. Combining the dynamic association knowledge features of obstacles and the real-time navigation parameters, an obstacle avoidance path evolution mechanism is constructed. Multiple initial obstacle avoidance paths are generated through the obstacle avoidance path evolution mechanism, which includes path initial generation rules, path dynamic adjustment basis, and path optimization direction. Based on the obstacle motion state association relationship in the obstacle dynamic association knowledge feature, multiple initial obstacle avoidance paths are dynamically evolved and adjusted to generate a target obstacle avoidance path that adapts to the real-time underwater environment, and the target obstacle avoidance path generation result is obtained. The target obstacle avoidance path generation result is correlated and mapped with the real-time navigation parameters of the underwater vehicle to generate real-time control commands for the underwater vehicle. The real-time control commands are transmitted to the execution control system of the underwater vehicle to drive the underwater vehicle to navigate along the target obstacle avoidance path. The real-time control commands include heading correction commands, speed adjustment commands, and power distribution commands.

[0005] Furthermore, embodiments of the present invention also provide an intelligent obstacle avoidance planning system for underwater obstacles, characterized in that it includes: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described intelligent perception and obstacle avoidance planning method for underwater obstacles by executing the machine-executable instructions.

[0006] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions stored in a computer-readable storage medium, wherein a processor of an intelligent perception and obstacle avoidance planning system for underwater obstacles reads the machine-executable instructions from the computer-readable storage medium, and the processor executes the machine-executable instructions, causing the intelligent perception and obstacle avoidance planning system for underwater obstacles to execute the aforementioned intelligent perception and obstacle avoidance planning method for underwater obstacles.

[0007] Based on the above, by receiving real-time navigation parameters of the underwater vehicle and underwater environment trigger signals to dynamically define the perception range, the perception area can be flexibly adjusted according to navigation status and environmental changes. This ensures the acquisition of the most critical underwater environmental information, drives multi-mode sensing devices to work collaboratively to collect multi-dimensional data and perform correlation and fusion processing, generating dynamic correlation knowledge features that include the spatial position, motion state, and material properties of obstacles. Combining the obstacle dynamic correlation knowledge features with real-time navigation parameters, an obstacle avoidance path evolution mechanism is constructed, generating multiple initial obstacle avoidance paths. Based on the correlation of obstacle motion state, the initial obstacle avoidance paths are dynamically evolved and adjusted, enabling real-time adaptation to dynamic changes in the underwater environment and generating target obstacle avoidance paths adapted to real-time conditions. Finally, the target obstacle avoidance path is correlated and mapped with real-time navigation parameters to generate real-time control commands, driving the underwater vehicle to accurately navigate along the target path, significantly improving the obstacle avoidance capability and navigation safety of the underwater vehicle in complex underwater environments. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the execution flow of the intelligent perception and obstacle avoidance planning method for underwater obstacles provided in an embodiment of the present invention.

[0009] Figure 2 This is a schematic diagram of exemplary hardware and software components of the intelligent perception and obstacle avoidance planning system for underwater obstacles provided in an embodiment of the present invention. Detailed Implementation

[0010] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating an intelligent perception and obstacle avoidance planning method for underwater obstacles provided in an embodiment of the present invention. The following is a detailed description of this intelligent perception and obstacle avoidance planning method for underwater obstacles.

[0011] Step S110: Receive real-time navigation parameters and underwater environment trigger signals from the underwater vehicle; dynamically define the sensing range of the underwater environment based on the real-time navigation parameters and the underwater environment trigger signals; and generate a dynamic definition result of the sensing range. The real-time navigation parameters include information related to the underwater vehicle's heading, attitude, and power output status. The underwater environment trigger signals include underwater current disturbance signals and ambient light change signals.

[0012] In this embodiment, firstly, real-time navigation parameters are continuously received by a sensor system mounted on the carrier. The navigation heading is acquired by a heading sensor and represented as three-dimensional azimuth angles; the navigation attitude is collected by an attitude sensor, including pitch, roll, and yaw angle parameters; the power output status is read from the power control system, encompassing information such as the thruster's output power and rotational speed. Simultaneously, underwater environment trigger signals are collected by a dedicated detection device; water flow disturbance signals include water flow velocity and direction information, and ambient light change signals reflect real-time changes in the surrounding ambient light intensity.

[0013] The sensing range is dynamically defined based on the aforementioned real-time navigation parameters and underwater environmental trigger signals. The navigation heading serves as the reference orientation for the sensing range, while the navigation attitude is used to adjust the spatial angle of the sensing range. The power parameter in the power output state is related to the initial size of the sensing range; higher power results in a larger initial sensing range. The water flow velocity and direction in the water flow disturbance signal are used to correct the shape of the sensing range; when the water flow velocity is high, the sensing range is appropriately reduced along the water flow direction and appropriately expanded against the water flow direction. The light intensity in the ambient light change signal determines the effective sensing range of the optical sensing device; high light intensity expands the optical sensing range, and vice versa, while the acoustic sensing range is adjusted accordingly. Combining these factors, and after data fusion processing, a dynamically defined sensing range result is generated, including spatial boundaries, depth range, and the device's working area.

[0014] Step S120: Based on the dynamic definition of the sensing range, drive the multi-mode sensing devices carried by the underwater vehicle to work collaboratively, collect multi-dimensional underwater environmental data within the sensing range, perform correlation and fusion processing on the multi-dimensional underwater environmental data, and generate obstacle dynamic correlation knowledge features. The multi-dimensional underwater environmental data includes acoustic detection data and optical imaging data, and the obstacle dynamic correlation knowledge features include spatial position correlation, motion state correlation, and material property correlation of obstacles.

[0015] Step S121: Analyze the dynamic definition result of the sensing range, extract the spatial boundary information, depth range information and environmental interference level information of the sensing range, and divide multiple collaborative sensing sub-regions based on the spatial boundary information, with each collaborative sensing sub-region corresponding to a specific spatial range.

[0016] In the scenario of exploring unknown sea areas, the dynamic definition result of the sensing range generated in step S110 is analyzed in detail. A boundary extraction algorithm is used to extract spatial boundary information from the dynamic definition result. This spatial boundary information is presented as a set of three-dimensional coordinate points, which, when connected, form the spatial outline of the sensing range. Depth range information is obtained through statistical analysis of the depth coordinates in the spatial boundary information, determining the minimum and maximum depths of the sensing range, as well as the division of different depth intervals. Environmental interference level information is assessed based on the intensity of water flow disturbance signals and ambient light variation signals within the sensing range, dividing the sensing range into different environmental interference level zones.

[0017] Based on the extracted spatial boundary information, a spatial grid partitioning method is used to divide the sensing range into multiple collaborative sensing sub-regions. During partitioning, an appropriate grid size is set according to the shape and size of the spatial boundaries. Each collaborative sensing sub-region has a clearly defined three-dimensional spatial coordinate range, identified by a sub-region number. After partitioning, each sub-region corresponds to an independent sensing task, ensuring that the multi-mode sensing device can perform comprehensive and detailed detection of the entire sensing range.

[0018] Step S122: Based on the environmental interference level information of multiple collaborative sensing sub-regions, assign the working mode of the multi-mode sensing device to each collaborative sensing sub-region. The working mode includes acoustic detection frequency and optical imaging resolution related parameters, and generate collaborative working parameters for the sensing devices.

[0019] In the scenario of exploring unknown sea areas, for each cooperative sensing sub-region divided in step S121, the operating mode of the multi-mode sensing device is determined based on its environmental interference level information. For sub-regions with low environmental interference levels, the acoustic detection device uses a higher detection frequency to obtain more detailed acoustic data; the optical imaging device sets a higher imaging resolution to obtain clear image information. For sub-regions with high environmental interference levels, the acoustic detection device reduces the detection frequency to improve signal penetration; the optical imaging device appropriately reduces the imaging resolution while increasing the exposure time to reduce noise interference.

[0020] After determining the operating mode of each collaborative sensing sub-region, the aforementioned operating mode parameters are integrated to generate collaborative operating parameters for the sensing devices. These collaborative operating parameters include detailed information such as the frequency and power of the acoustic detection device and the resolution and exposure time of the optical imaging device for each sub-region, and are used to guide the collaborative operation of multi-mode sensing devices in different sub-regions.

[0021] Step S123: Based on the collaborative working parameters of the sensing devices, activate the acoustic detection device in the multi-mode sensing device, transmit directional acoustic detection signals to each collaborative sensing sub-region, receive the reflected acoustic signals, and convert the reflected acoustic signals into digitized acoustic detection data. The acoustic detection data includes information related to signal reflection duration, signal amplitude change, and signal frequency shift.

[0022] Step S1231: Adjust the transmission power and signal waveform of the acoustic detection device, wherein the transmission power is set based on the environmental interference level of the corresponding cooperative sensing sub-region, and the signal waveform is selected based on the underwater propagation characteristics.

[0023] In the scenario of exploring unknown sea areas, for each cooperative sensing sub-region determined in step S122, the transmission power of the acoustic detection equipment is adjusted according to its environmental interference level. Sub-regions with high environmental interference levels require higher transmission power to ensure that the detection signal can effectively penetrate the interference area and be reflected back; sub-regions with low environmental interference levels can appropriately reduce transmission power to save energy and reduce interference to other equipment.

[0024] Simultaneously, an appropriate signal waveform is selected based on underwater propagation characteristics. Different signal waveforms exhibit different attenuation characteristics, resolution, and anti-interference capabilities during underwater propagation. For example, in turbid water with significant propagation loss, a signal waveform with a wider pulse width and concentrated energy is selected; when high-resolution detection is required, a signal waveform with a higher frequency and narrower pulse width is chosen. By adjusting the transmission power and selecting the signal waveform, the acoustic detection equipment can achieve optimal detection results in different cooperative sensing sub-regions.

[0025] Step S1232: According to the distribution order of the cooperative sensing sub-regions, align the emission direction of the acoustic detection device with the center position of each cooperative sensing sub-region in turn, and adjust the emission angle and emission range parameters of the acoustic detection device so that the coverage of the directional acoustic detection signal can completely include the current cooperative sensing sub-region.

[0026] In the scenario of exploring unknown sea areas, acoustic detection is performed on each sub-region according to the spatial distribution order of the cooperative sensing sub-regions divided in step S121. First, the transmitting probe is aligned sequentially with the center position of each cooperative sensing sub-region using the steering mechanism of the acoustic detection equipment. During the alignment process, the current position of the transmitting probe is monitored in real time using a position sensor and compared with the center position of the sub-region. The steering mechanism is adjusted through feedback control to ensure alignment accuracy.

[0027] After alignment, the emission angle and emission range parameters of the acoustic detection equipment are adjusted. The emission angle is set according to the spatial location and shape of the sub-region to ensure that the signal can cover the sub-region at an appropriate angle. The emission range parameters are adjusted according to the size of the sub-region to ensure that the coverage of the directional acoustic detection signal completely includes the current cooperative sensing sub-region, avoiding detection gaps. During the adjustment process, the rationality of the emission angle and emission range parameters is verified through a combination of simulation and actual testing.

[0028] Step S1233: Activate the signal transmission module of the acoustic detection device to transmit directional acoustic detection signals to the currently aligned cooperative sensing sub-region, and record the precise time point of the directional acoustic detection signal transmission.

[0029] In the scenario of exploring unknown sea areas, after completing the adjustments in step S1232, the signal transmission module of the acoustic detection equipment is activated. The transmission module generates a directional acoustic detection signal based on set parameters such as transmission power, signal waveform, and frequency, and transmits it to the currently aligned cooperative sensing sub-region. At the instant of signal transmission, a high-precision timestamp module records the exact time of transmission. This time point will be used to subsequently calculate the signal reflection duration, which is a crucial basis for determining the distance to obstacles. The timestamp accuracy must reach the microsecond level to ensure the accuracy of distance calculations.

[0030] Step S1234: Keep the transmission direction of the acoustic detection device unchanged, start the signal receiving module of the acoustic detection device, capture the acoustic reflection signal reflected back from the current cooperative sensing sub-region in real time, and continuously receive it for a preset time.

[0031] In unknown sea area exploration scenarios, after transmitting a directional acoustic detection signal, the signal receiving module is immediately activated while maintaining the transmission direction of the acoustic detection equipment. The receiving module operates in real-time monitoring, capturing acoustic reflection signals from obstacles within the current cooperative sensing sub-region. Based on the size of the cooperative sensing sub-region and the detection distance, a reasonable reception duration is preset to ensure that all possible reflection signals are received. During the reception process, the receiving module continuously acquires signals and temporarily stores the acquired analog signals in a buffer.

[0032] Step S1235: Record the precise time point of each received acoustic reflection signal, calculate the difference between the transmission time point of the directional acoustic detection signal and the reception time point of the acoustic reflection signal, and obtain the signal reflection duration corresponding to each acoustic reflection signal.

[0033] In the scenario of exploring unknown sea areas, when the signal receiving module captures an acoustic reflection signal, the timestamp module synchronously records the precise time point of each received signal. Then, the reception time point of each acoustic reflection signal is compared with the transmission time point recorded in step S1233, and the time difference between the two is calculated. This time difference is the signal reflection duration. The signal reflection duration reflects the time it takes for the sound wave to travel from transmission to reception and is a key parameter for subsequent calculations of obstacle distances. The above processing is performed on each received acoustic reflection signal to obtain the signal reflection duration data.

[0034] Step S1236: Perform amplitude detection on the received acoustic reflection signal, collect amplitude change data of the acoustic reflection signal during propagation, record the peak and valley values ​​of the amplitude change data and the corresponding time nodes, and obtain signal amplitude change information.

[0035] In unknown sea exploration scenarios, the amplitude of received acoustic reflection signals is detected. An amplitude detection circuit monitors the signal amplitude in real time. During signal propagation, the signal amplitude changes due to encountering different obstacles and variations in the propagation medium. Collecting this amplitude variation data, including the maximum (peak) and minimum (trough) amplitude values, as well as the time points at which peaks and troughs occur, reveals information reflecting the size, shape, and material characteristics of obstacles. This data is then processed to obtain the signal amplitude variation information.

[0036] Step S1237: Analyze the frequency characteristics of the acoustic reflection signal, compare it with the original frequency of the directional acoustic detection signal, calculate the offset of the acoustic reflection signal relative to the original frequency of the directional acoustic detection signal, and obtain the signal frequency offset information.

[0037] In unknown sea area exploration scenarios, frequency analysis is performed on the received acoustic reflection signals. Using spectrum analysis techniques, the acoustic reflection signals are decomposed into different frequency components, obtaining the signal's frequency spectrum. Then, this frequency spectrum is compared with the original frequency of the directional acoustic detection signal to identify the differences. The offset of each frequency component in the acoustic reflection signal relative to the original frequency is calculated; these offsets reflect information such as the motion state of obstacles. The offset data is then processed to generate signal frequency offset information.

[0038] Step S1238: The collected signal reflection time, signal amplitude change information and signal frequency offset information are digitally converted to convert the analog signal data into digital data in a standard format.

[0039] In the scenario of exploring unknown sea areas, the signal reflection time obtained in step S1235, the signal amplitude change information collected in step S1236, and the signal frequency offset information calculated in step S1237 are digitized. The analog signal data is converted into digital data using an analog-to-digital converter. During the conversion process, an appropriate sampling rate and quantization bit depth are set to ensure the accuracy and resolution of the digital data. The converted digital data is stored in a standard data format to facilitate subsequent computer processing and analysis.

[0040] Step S1239: According to the identifier of the collaborative sensing sub-region, associate and bind the digital signal data corresponding to each collaborative sensing sub-region to form an acoustic data subset divided by the collaborative sensing sub-region.

[0041] In the scenario of exploring unknown sea areas, the digitized signal data converted in step S1238 is associated and bound with the corresponding sub-region based on the identification information of the collaborative sensing sub-region. Each collaborative sensing sub-region has a unique identifier, which is used to integrate digitized data such as signal reflection time, signal amplitude variation information, and signal frequency offset information belonging to the same sub-region, forming an acoustic data subset. In this way, each acoustic data subset corresponds to a specific collaborative sensing sub-region, facilitating subsequent individual processing and analysis of acoustic data from different sub-regions.

[0042] Step S12310: Integrate the acoustic data subsets of all collaborative sensing sub-regions, supplement the acquisition parameters and environmental interference level information corresponding to each acoustic data subset, and generate the digital acoustic detection data.

[0043] In the scenario of exploring unknown sea areas, the acoustic data subsets of all collaborative sensing sub-regions formed in step S1239 are integrated. During the integration process, corresponding acquisition parameters, such as transmission power, signal frequency, and transmission angle, as well as environmental interference level information for that sub-region, are added to each acoustic data subset. This supplementary information helps to more comprehensively analyze the acoustic data and evaluate the detection effect. After integration, digital acoustic detection data is formed, which contains acoustic detection information of all collaborative sensing sub-regions within the sensing range.

[0044] Step S124: Simultaneously start the optical imaging device in the multi-mode sensing device to continuously acquire images of each cooperative sensing sub-region, obtain raw optical imaging data, perform light compensation processing on the raw optical imaging data, correct the imaging deviation caused by underwater light attenuation, and generate clear optical imaging data.

[0045] Step S1241: Based on the collaborative working parameters of the sensing device, adjust the imaging resolution, exposure time, and photosensitivity parameters of the optical imaging device so that the parameter settings of the optical imaging device are adapted to the illumination conditions and environmental interference levels of the corresponding collaborative sensing sub-region.

[0046] In unknown sea area exploration scenarios, for each collaborative sensing sub-region, the imaging resolution, exposure time, and sensitivity parameters of the optical imaging equipment are adjusted according to the requirements of the collaborative working parameters of the sensing devices. The imaging resolution setting is determined based on the size of the sub-region and the level of detail of the obstacles to be identified. For sub-regions that need to observe small obstacles, a higher resolution is set; for larger sub-regions, the resolution can be appropriately reduced to improve imaging speed.

[0047] Exposure time and ISO sensitivity parameters are adjusted according to the lighting conditions and environmental interference levels of the sub-regions. In sub-regions with good lighting conditions, the exposure time is shortened and the ISO is reduced to avoid overexposure and reduce noise; in sub-regions with poor lighting conditions, the exposure time is extended and the ISO is increased to ensure image brightness and sharpness. Through these parameter adjustments, the optical imaging device can achieve optimal imaging results in different cooperative sensing sub-regions.

[0048] Step S1242: Align the lens of the optical imaging device with each collaborative sensing sub-region in turn, adjust the focal length and shooting angle of the lens of the optical imaging device so that the entire range of the collaborative sensing sub-region falls within the imaging field of view of the optical imaging device, and fix the lens orientation of the optical imaging device.

[0049] In unknown sea area exploration scenarios, the lenses of the optical imaging equipment are sequentially aimed at each sub-region according to the order of collaborative sensing sub-regions. The lens focal length is adjusted by the lens drive mechanism to ensure clear imaging of objects within the sub-region. Simultaneously, the shooting angles of the optical imaging equipment are adjusted, including both horizontal and vertical angles, to ensure that the entire area of ​​the collaborative sensing sub-region falls within the imaging field of view. After adjustment, the lens attitude is fixed by a locking mechanism to prevent changes in lens position due to carrier shaking or other reasons during imaging, thus ensuring the stability and accuracy of the imaging.

[0050] Step S1243: Activate the continuous imaging function of the optical imaging device, continuously image the currently aligned cooperative sensing sub-region according to the preset acquisition interval, and acquire multiple frames of images as the original optical imaging data of the cooperative sensing sub-region.

[0051] In unknown sea exploration scenarios, after lens alignment and parameter adjustment, the continuous imaging function of the optical imaging equipment is activated. Based on a preset acquisition interval (e.g., capturing a certain number of images per second), continuous imaging is performed on the currently aligned collaborative sensing sub-region. The acquired multiple frames serve as the raw optical imaging data for that sub-region, reflecting the dynamic changes of obstacles within the sub-region. During imaging, image data is stored in a buffer in real time, and a timestamp and sub-region identifier are added to each frame.

[0052] Step S1244: Record the acquisition time, corresponding collaborative sensing sub-region identifier, and illumination conditions of each frame of raw optical imaging data, and establish a raw optical imaging data association table.

[0053] In unknown sea area exploration scenarios, for each frame of raw optical imaging data acquired, its acquisition time, corresponding collaborative sensing sub-region identifier, and the lighting conditions at that time are recorded in detail. Lighting condition information includes parameters such as light intensity and light source direction, which can be obtained through the light sensor built into the optical imaging device. This information is then correlated with the raw image data to establish a raw optical imaging data association table. This table facilitates subsequent image data retrieval, analysis, and processing, enabling rapid location of the required image data based on the acquisition time and sub-region identifier, and assessment of image quality in conjunction with lighting condition information.

[0054] Step S1245: Extract the brightness distribution information from the original optical imaging data, analyze the brightness differences in different areas of the original optical imaging data, and identify dark areas caused by underwater light attenuation.

[0055] In uncharted waters exploration scenarios, brightness analysis is performed on raw optical imaging data. Image analysis algorithms are used to extract brightness distribution information from each frame, obtaining the brightness value of each pixel. Then, the average brightness and variance of different regions are statistically analyzed to identify areas with significantly lower brightness than other areas; these are dark areas caused by underwater light attenuation. These dark areas may conceal obstacles and require subsequent light compensation processing to reveal their details.

[0056] Step S1246: Based on the underwater light attenuation model and the illumination conditions during data collection, calculate the light compensation coefficient for the dark area. The light compensation coefficient is dynamically adjusted according to the location and depth of the dark area.

[0057] In unknown sea area exploration scenarios, an underwater light attenuation model is used to describe the attenuation of light as it propagates underwater with distance and depth. By combining the illumination conditions information acquired during the acquisition of raw optical imaging data, such as light intensity and light source direction, a light compensation coefficient for dark areas is calculated. The magnitude of the light compensation coefficient is related to the location and depth of the dark area; the farther away from the light source and the greater the depth of the dark area, the more severe the light attenuation, and the larger the corresponding compensation coefficient. By dynamically adjusting the compensation coefficient, effective light compensation is ensured for dark areas at different locations and depths.

[0058] Step S1247: According to the calculated light compensation coefficient, perform brightness enhancement processing on the dark areas in the original optical imaging data, and adjust the pixel brightness values ​​in the dark areas so that the details in the dark areas can be clearly presented.

[0059] In unknown sea area exploration scenarios, the brightness of dark areas in the original optical imaging data is enhanced based on the light compensation coefficient calculated in step S1246. An image brightness adjustment algorithm is used to adjust the brightness value of each pixel within the dark area, increasing the pixel brightness according to the compensation coefficient. During the adjustment process, care is taken to avoid over-enhancement that could amplify image noise and cause loss of detail. Through brightness enhancement, details in the dark areas are clearly presented, improving the overall image quality.

[0060] Step S1248: Perform color correction on the enhanced optical imaging data to correct the color shift caused by underwater light refraction, so that the corrected image color is closer to the color of the real underwater environment.

[0061] In exploration scenarios in unknown sea areas, the refraction of underwater light can cause color shifts in optical imaging data, such as an overall bluish or greenish tint to the image. Therefore, color correction is necessary for the enhanced optical imaging data. Based on underwater optical characteristics, a color correction model is established. This model considers the differences in attenuation of light of different wavelengths in water and corrects color shifts by adjusting the gain and offset of the red, green, and blue color channels of the image. The corrected image colors are closer to the real underwater environment colors, which helps in subsequent obstacle identification and material determination.

[0062] Step S1249: Remove noise interference from the color-corrected optical imaging data, retain the object outline and texture details in the color-corrected optical imaging data, and obtain a clear single-frame optical image.

[0063] In unknown sea exploration scenarios, optical imaging data, even after light compensation and color correction, may still contain noise interference, such as salt-and-pepper noise and Gaussian noise. This noise can affect image quality and the accuracy of subsequent processing. Therefore, noise removal is necessary. Image denoising algorithms, such as median filtering and Gaussian filtering, are used to process the color-corrected image. While removing noise, it is crucial to preserve object contours and texture details as much as possible, avoiding excessive smoothing that could lead to detail loss. After processing, a clear single-frame optical image is obtained.

[0064] Step S12410: Integrate multiple clear optical images of all collaborative sensing sub-regions, associate the corresponding acquisition time with the collaborative sensing sub-region identifier, and generate the clear optical imaging data.

[0065] In unknown sea area exploration scenarios, multiple clear optical images obtained from all collaborative sensing sub-regions are integrated. The image data is sorted and organized according to the identifiers and acquisition times of the collaborative sensing sub-regions. Each image frame is assigned a corresponding sub-region identifier and acquisition time information to ensure the traceability of the image data. After integration, clear optical imaging data is generated, containing clear optical image information of all collaborative sensing sub-regions within the sensing range at different time points.

[0066] Step S125: Collect acoustic detection data and clear optical imaging data corresponding to all collaborative sensing sub-regions, construct a multi-dimensional underwater environment data set, and add corresponding collaborative sensing sub-region identifiers and acquisition time identifiers to each data entry.

[0067] In the scenario of exploring unknown sea areas, the digitized acoustic detection data generated in step S12310 and the clear optical imaging data generated in step S12410 are collected. According to the division of collaborative sensing sub-regions, the acoustic detection data and optical imaging data corresponding to each sub-region are grouped together. A corresponding collaborative sensing sub-region identifier and acquisition time identifier are added to each data entry (such as a segment of acoustic data or a frame of optical image) to ensure that the data can accurately correspond to a specific sub-region and acquisition time. In this way, a multi-dimensional underwater environment dataset is constructed, which includes acoustic and optical multi-dimensional detection data within the sensing range.

[0068] Step S126: Extract signal features from the acoustic detection data and image features from the clear optical imaging data in the underwater environment multi-dimensional data set. The signal features include reflection signal peak features and signal attenuation features, and the image features include object edge features and texture distribution features.

[0069] In unknown sea area exploration scenarios, feature extraction is performed on acoustic detection data and clear optical imaging data from a multi-dimensional underwater environment dataset. For acoustic detection data, signal features are extracted using signal processing algorithms. Peak characteristics of the reflected signal include parameters such as peak amplitude, occurrence time, and duration, which reflect the reflectivity and size of obstacles. Signal attenuation characteristics include the attenuation rate and attenuation pattern of the signal amplitude over time, reflecting the characteristics of the propagation medium and the distance to obstacles.

[0070] For clear optical imaging data, image feature extraction algorithms are used to extract image features. Object edge features are extracted using edge detection operators (such as Sobel operator, Canny operator, etc.), including information such as edge position, direction and length; texture distribution features are extracted using texture analysis methods (such as gray-level co-occurrence matrix, wavelet transform, etc.), reflecting the texture patterns and distribution rules of different regions in the image.

[0071] Step S127: Establish cross-modal association mapping between signal features and image features. By using the same collaborative sensing sub-region identifier and acquisition time identifier, the acoustic signal features and optical image features belonging to the same obstacle are associated and bound to generate a cross-modal feature set of the obstacle.

[0072] In unknown sea area detection scenarios, a cross-modal association mapping between signal features and image features is established based on the cooperative sensing sub-region identifier and the acquisition time identifier. For data detected in the same cooperative sensing sub-region within the same acquisition time, it is assumed that they may contain acoustic signal features and optical image features of the same obstacle. Using a feature matching algorithm, the positional information in the acoustic signal features (calculated from the signal reflection duration) is compared with the object positional information in the optical image features to find the feature pair with the highest matching degree. The acoustic signal features and optical image features belonging to the same obstacle are associated and bound to form a cross-modal feature description of the obstacle. This process is performed on all detected obstacles to generate a cross-modal feature set for obstacles, which contains the multimodal feature information of each obstacle.

[0073] Step S128: Based on the cross-modal feature set of the obstacles, extract the spatial coordinates of each obstacle, analyze the spatial distance and relative orientation between different obstacles, and generate the spatial position association relationship of the obstacles.

[0074] In unknown sea area exploration scenarios, the spatial coordinates of each obstacle are extracted from a cross-modal feature set. For acoustic signal features, the distance from the obstacle to the underwater vehicle is calculated based on the signal reflection time and the speed of sound propagation in water. This distance, combined with the emission angle of the acoustic detection equipment, determines the obstacle's three-dimensional spatial coordinates. For optical image features, the actual spatial coordinates corresponding to the obstacle's position in the image are calculated using stereo vision algorithms or based on camera calibration parameters. The spatial coordinates obtained from the two modalities are then fused to obtain more accurate obstacle spatial coordinates.

[0075] Next, the spatial relationships between different obstacles are analyzed. The straight-line distance between any two obstacles and their relative azimuth angles (such as azimuth and pitch angles) are calculated. Based on this distance and azimuth information, a spatial relationship model between obstacles is established. This model can describe the distribution of obstacles in space and their relative positions, such as which obstacles are close together and which obstacles are in the same direction.

[0076] Step S129: By continuously collecting multi-dimensional underwater environmental data, track the spatial position changes of each obstacle, calculate the position change trend of the obstacle, and combine the underwater water flow direction and velocity information to generate the motion state correlation of the obstacle.

[0077] In unknown sea area exploration scenarios, because multi-dimensional underwater environmental data is continuously collected, the spatial position of each obstacle can be tracked. By comparing the spatial coordinates of the same obstacle at different collection points, its positional change is calculated. Based on the positional change and the time interval, the obstacle's velocity and acceleration are calculated. Analyzing these motion parameters determines the obstacle's positional change trend, such as whether the direction of movement is stable and whether the velocity is uniform.

[0078] Simultaneously, by combining the water flow direction and velocity information from underwater water flow disturbance signals, the influence of water flow on the motion state of obstacles is analyzed. For obstacles whose motion may be affected by water flow (such as suspended organisms or floating objects), their motion state is correlated with the water flow direction and velocity. By establishing a correlation model between motion state and water flow parameters, the correlation relationship of obstacle motion state is generated, which can describe the interaction between obstacle motion and water flow.

[0079] Step S1210: Analyze the acoustic signal reflection characteristics and optical image texture features of each obstacle, associate the correspondence between the material of common underwater obstacles and the signal features, generate the material characteristic association relationship of the obstacles, integrate the spatial position association relationship of the obstacles, the motion state association relationship of the obstacles and the material characteristic association relationship of the obstacles, and generate the dynamic association knowledge features of the obstacles.

[0080] In the scenario of exploring unknown sea areas, a detailed analysis is conducted on the acoustic signal reflection characteristics and optical image texture features of each obstacle. Acoustic signal reflection characteristics include the intensity, frequency shift, and attenuation of the reflected signal. Obstacles of different materials have varying abilities to reflect sound waves; for example, metallic obstacles typically reflect stronger signals, while muddy obstacles reflect weaker signals. Optical image texture features include the coarseness, direction, and uniformity of the texture. Obstacles of different materials exhibit variations in surface texture; for example, rock surfaces are rough, while plastic surfaces are relatively smooth.

[0081] By establishing a database of correspondences between common underwater obstacle materials and their signal characteristics, the analyzed obstacle signal characteristics are compared with the data in the database to determine the possible material type of each obstacle. Then, based on the obstacle's material type, the correlations of material properties between different obstacles are analyzed, such as which obstacles have the same or similar materials and which have different materials. Finally, the spatial location correlations, motion state correlations, and material property correlations of obstacles are integrated to form a comprehensive dynamic correlation knowledge feature of obstacles that reflects various correlation information.

[0082] Step S130: Combining the obstacle dynamic association knowledge features and the real-time navigation parameters, construct an obstacle avoidance path evolution mechanism, and generate multiple initial obstacle avoidance paths through the obstacle avoidance path evolution mechanism. The obstacle avoidance path evolution mechanism includes path initial generation rules, path dynamic adjustment basis, and path optimization direction.

[0083] Step S131: Analyze the dynamic association knowledge features of the obstacles, extract the spatial position association, motion state association and material property association of the obstacles, and generate an obstacle environmental situation description.

[0084] In unknown sea area exploration scenarios, the dynamic correlation characteristics of obstacles are analyzed. Information such as obstacle distribution density, relative position, and distance are extracted from spatial location correlations; parameters such as obstacle velocity, direction, and acceleration are extracted from motion state correlations; and information such as obstacle material type and hardness is extracted from material property correlations. These information are integrated and comprehensively analyzed to generate an obstacle environmental situation description. This description, combining text and data, comprehensively reflects the overall distribution, movement trends, and material properties of obstacles in the current underwater environment.

[0085] Step S132: Analyze the real-time navigation parameters, extract the navigation heading, speed, power output status and attitude stability information of the underwater vehicle, determine the current navigation capability and subsequent navigation potential of the underwater vehicle, and generate navigation status assessment results.

[0086] In the scenario of exploration in unknown sea areas, real-time navigation parameters are analyzed. The course reflects the direction of travel of the vehicle; the speed indicates the current speed of movement of the vehicle; the power output status, including the power and speed of the thrusters, reflects the power reserve of the vehicle; attitude stability information is evaluated through the measurement data of attitude sensors, such as the amplitude and frequency of changes in pitch and roll angles, reflecting the stability of the vehicle during navigation.

[0087] Based on these parameters, the current navigation capabilities of the underwater vehicle are assessed, such as maximum turning angle, acceleration, and deceleration capabilities. Simultaneously, considering power output and energy reserves, the vehicle's future navigation potential is predicted, including the duration and distance it can sustain navigation. The above assessment results are then compiled to generate a navigation status assessment.

[0088] Step S133: Based on the obstacle environment situation description and the navigation status assessment results, set the initial path generation rules. The initial path generation rules include the definition criteria for the path start and end points, the safety distance criteria between the path and obstacles, and the smoothness criteria for path heading changes.

[0089] In unknown sea area exploration scenarios, initial path generation rules are established by comprehensively considering the obstacle environment situation description and navigation status assessment results. The starting point of the path is defined by the current position coordinates of the underwater vehicle; the ending point is defined according to the preset navigation mission objective, such as reaching a designated detection point or area. The safety distance standard between the path and obstacles is determined based on the size, material, and motion state of the obstacles. For large, rigid, or fast-moving obstacles, a larger safety distance is set; for small, soft, or stationary obstacles, the safety distance can be appropriately reduced.

[0090] The smoothness standard for path heading changes specifies the maximum rate of change and the maximum single angle of change of heading angle in the path to ensure the stability of the vehicle's attitude during navigation and avoid loss of control caused by sharp turns. The above standards together constitute the initial path generation rules to ensure that the generated initial obstacle avoidance path is feasible and safe.

[0091] Step S134: Based on the correlation between the motion state of obstacles and the characteristics of underwater water flow, set the dynamic adjustment criteria for the path. The dynamic adjustment criteria for the path include the response threshold of obstacle position change, the influence coefficient of water flow disturbance on the path, and the adaptation range of navigation attitude adjustment.

[0092] In unknown sea exploration scenarios, the correlation of obstacle motion states reflects the obstacle's motion trend, and underwater current characteristics include information such as current velocity and direction. Based on these factors, a dynamic path adjustment basis is established. The response threshold for obstacle position change refers to the threshold at which the obstacle avoidance path needs adjustment. The influence coefficient of current disturbance on the path is determined based on the angle between the current velocity and direction and the path; the greater the influence of the current, the higher the coefficient value, and the greater the adjustment range. The adaptation range for navigation attitude adjustment specifies the allowable range of attitude changes for the carrier when adjusting the path, such as the maximum allowable deviation of pitch and roll angles, to ensure the carrier's stability. The above criteria are used to determine whether and how to adjust the path.

[0093] Step S135: Based on the power output status and energy reserve of the underwater vehicle, set the path optimization direction, which includes the optimization direction with the lowest energy consumption, the optimization direction with the highest navigation efficiency, and the optimization direction with the most stable attitude.

[0094] In uncharted waters exploration scenarios, the power output of an underwater vehicle reflects its current power usage, while its energy reserves determine its ability to continue navigating. Based on this information, different path optimization directions are established. The lowest energy consumption optimization direction aims to select a path that minimizes the vehicle's energy consumption by rationally planning speed and steering to reduce power output. The highest navigation efficiency optimization direction aims to reach the target point in the shortest time, maximizing navigation speed while ensuring safety. The most stable attitude optimization direction focuses on maintaining the vehicle's stable navigation attitude by selecting paths with gentle course changes and minimal current impact, reducing energy consumption and equipment wear caused by attitude adjustments. Depending on mission requirements and actual conditions, a single optimization direction can be selected, or multiple optimization directions can be considered comprehensively.

[0095] Step S136: Integrate the initial path generation rules, the dynamic path adjustment basis, and the path optimization direction to construct the obstacle avoidance path evolution mechanism and clarify the core principles to be followed in the path generation and adjustment process.

[0096] In the scenario of exploring unknown sea areas, the initial path generation rules set in step S133, the dynamic path adjustment criteria set in step S134, and the path optimization direction set in step S135 are integrated to construct an obstacle avoidance path evolution mechanism. During the integration process, the priority and interrelationships among these rules, criteria, and directions are clarified. For example, in the path generation phase, the initial path generation rules are primarily followed; during path execution, when encountering changes in obstacle positions or water flow disturbances, path adjustments are made based on the dynamic path adjustment criteria; throughout the entire process, optimization is consistently aimed at achieving the optimal path direction.

[0097] Simultaneously, the core principles to be followed in the path generation and adjustment process are clearly defined, such as the safety priority principle (no collision with obstacles under any circumstances) and the feasibility principle (the generated path must be within the vehicle's navigation capabilities). Through the integration of the above, an obstacle avoidance path evolution mechanism is formed to guide the generation and adjustment of obstacle avoidance paths.

[0098] Step S137: Based on the path initial generation rules in the obstacle avoidance path evolution mechanism, determine the current position of the underwater vehicle as the path starting point, combine the preset navigation target direction to determine the path ending area, and generate path start and end information.

[0099] Step S1371: Analyze the path initial generation rules in the obstacle avoidance path evolution mechanism, extract the definition criteria of the path start point and the path end point. The path start point is completely consistent with the current position of the underwater vehicle, and the path end point is located in a specific area in the direction of the preset navigation target.

[0100] In unknown sea exploration scenarios, the path initialization rules in the obstacle avoidance path evolution mechanism are analyzed. The defining criteria for the path start and end points are extracted. According to the rules, the path start point must be completely consistent with the current real-time position of the underwater vehicle to ensure the accuracy of the path planning start point. The defining criterion for the path end point is a specific area located in the preset navigation target direction. The size and shape of this area are determined based on the accuracy requirements of the navigation mission and the characteristics of the target area. For example, if the target is to reach a large exploration area, the end point area can be set larger; if the target is a precise point, the end point area can be set smaller.

[0101] Step S1372: Extract the current position coordinates of the underwater vehicle from the real-time navigation parameters. The current position coordinates of the underwater vehicle are determined based on the underwater global coordinate system. The current position coordinates of the underwater vehicle are directly used as the path start coordinates, and the path start coordinate information is recorded.

[0102] In uncharted waters exploration scenarios, real-time navigation parameters contain the current position coordinates of the underwater vehicle. These coordinates are determined based on an underwater global coordinate system, a fixed three-dimensional coordinate system used to describe the position of points in underwater space. The specific numerical values ​​of the current position coordinates, including the X, Y, and Z coordinates, are extracted from the real-time navigation parameters. These coordinate values ​​are directly used as the starting point coordinates of the path and recorded in detail, including the coordinate values, coordinate system type, and extraction time, to ensure the accuracy and traceability of the path starting point.

[0103] Step S1373: Obtain the preset underwater vehicle navigation target direction, and determine the safe area range in the preset underwater navigation target direction in combination with the navigation mission requirements of the underwater vehicle. The safe area range in the preset underwater navigation target direction is far away from known fixed obstacle areas.

[0104] In unknown sea exploration scenarios, the preset underwater navigation target direction is determined based on the pre-set navigation mission. Combining mission requirements, such as exploring a specific area or reaching a designated location, the safe zone range in that target direction is determined. When determining the safe zone range, known information on the distribution of fixed obstacles needs to be considered to ensure the area is far from these obstacles, avoiding setting the endpoint near or inside obstacles. Determining the safe zone range also requires considering the vehicle's navigation capabilities and mission feasibility to ensure the vehicle can reach the area.

[0105] Step S1374: Spatial division of the safe zone range in the preset underwater navigation target direction, and determination of the core area within the preset safe zone range in the underwater navigation target direction as the path endpoint area. The size of the path endpoint area is set according to the size of the underwater navigation vehicle and the navigation accuracy requirements.

[0106] In unknown sea exploration scenarios, a pre-defined safe zone is spatially divided. Using grid partitioning or region segmentation methods, the safe zone is divided into multiple sub-regions. Based on the importance and accuracy requirements of the navigation mission, the core region is determined as the path endpoint region. The size of the path endpoint region needs to consider the dimensions of the underwater vehicle to ensure that the vehicle can fully enter the region; simultaneously, the boundary range of the region is determined according to the navigation accuracy requirements—a smaller region for higher accuracy requirements and a larger region for lower accuracy requirements.

[0107] Step S1375: Extract the boundary coordinate information of the path endpoint region, determine the center coordinates of the path endpoint region as the reference endpoint coordinates, and record the boundary coordinates and reference endpoint coordinate information of the path endpoint region.

[0108] In unknown sea area exploration scenarios, after the path endpoint region is determined, the boundary coordinate information of this region is extracted. The boundary coordinate information is represented by a series of 3D coordinate points, which, when connected, form the boundary contour of the endpoint region. By calculating the average value of these boundary coordinate points, the center coordinates of the path endpoint region are determined and used as the reference endpoint coordinates. Detailed records of the boundary coordinates of the path endpoint region and the reference endpoint coordinates are kept, including the numerical value and coordinate type of each coordinate point.

[0109] Step S1376: Calculate the straight-line distance between the starting point coordinates and the reference ending point coordinates. Based on the underwater vehicle's speed and energy reserves, make a preliminary estimate of the time required for straight-line travel between the starting point coordinates and the reference ending point coordinates.

[0110] In uncharted waters exploration scenarios, the straight-line distance between the starting point and reference endpoint coordinates is calculated using the distance formula between the two points. Then, considering the underwater vehicle's current speed and energy reserves, a preliminary estimate of the time required for straight-line travel is made. The current average speed is used, and the energy reserves are used to determine whether straight-line travel is feasible at that speed. The estimated time serves as a reference for subsequent path planning and optimization, helping to determine a reasonable path length and speed.

[0111] Step S1377: Analyze the distribution of obstacles around the path starting point, and combine the safe distance standard in the path initial generation rules to determine the feasible initial heading range from the path starting point. The feasible initial heading range from the path starting point avoids obstacles near the path starting point.

[0112] In unknown sea area exploration scenarios, based on the dynamic association characteristics of obstacles, the distribution of obstacles within a certain range around the path starting point is analyzed, including the location, size, and type of obstacles. Combining this with the safety distance standard in the initial path generation rules, a feasible initial course range from the path starting point is determined. The feasible initial course range refers to the course selected within this range that ensures the vehicle maintains a safe distance from surrounding obstacles during the departure phase. By eliminating courses that would lead to collisions with obstacles, a feasible range of initial course angles is obtained.

[0113] Step S1378: Analyze the distribution of obstacles around the path endpoint area, and combine the safety distance standard in the initial path generation rules to determine the feasible termination course range for entering the path endpoint area. The feasible termination course range for entering the path endpoint area enables the underwater vehicle to smoothly enter the path endpoint area.

[0114] In unknown sea exploration scenarios, the distribution of obstacles around the path endpoint area is analyzed based on the dynamic correlation knowledge characteristics of obstacles. Combining this with the safety distance standard in the initial path generation rules, a feasible termination course range for entering the path endpoint area is determined. This feasible termination course range ensures that the vehicle can enter the endpoint area with a stable attitude, avoiding collisions with surrounding obstacles due to improper heading. The determination of this range considers the shape and size of the endpoint area, as well as the position and movement of surrounding obstacles, enabling the vehicle to reach the endpoint safely and accurately.

[0115] Step S1379: Integrate the coordinates of the path start point, the feasible initial heading range from the path start point, the boundary coordinates of the path end point area, the reference end point coordinates, and the feasible termination heading range for entering the path end point area.

[0116] In unknown sea area exploration scenarios, the information obtained from the previous steps, including the coordinates of the path start point, feasible initial course range, boundary coordinates of the path end area, reference end point coordinates, and feasible termination course range, is integrated. This information is organized according to a specific data format to form a complete path start-end information dataset. During the integration process, consistency and correlation between the various pieces of information are ensured; for example, the feasible initial course range and feasible termination course range are matched with the locations of the path start and end areas.

[0117] Step S13710: Supplement the environmental information related to the start and end of the path, including the water flow direction, water flow velocity, and illumination conditions of the start and end of the path, and generate the path start and end information.

[0118] In unknown sea area exploration scenarios, to more comprehensively describe the path's start and end points, supplementary environmental information related to these points is required. The direction and velocity of water flow in the path's starting area are acquired using a water flow sensor near the starting point; illumination information is obtained using a light sensor. Similarly, the direction, velocity, and illumination information in the path's ending area are obtained through corresponding sensors or inferred from known environmental data. Adding this environmental information to the path start and end information dataset generates complete path start and end information.

[0119] Step S138: Based on the path start and end information and the spatial location relationship of the obstacles, plan multiple initial path structures from the path start point to the path end point area. During the planning process, compare the path trajectory coordinates with the spatial occupancy coordinates of the obstacles so that the planned path trajectory is outside the spatial occupancy range of all obstacles.

[0120] Step S1381: Extract the path start coordinates, feasible initial heading range from the path start, boundary coordinates of the path end area, and feasible termination heading range into the path end area from the path start and end information to determine the core constraints of the path planning.

[0121] In unknown sea area exploration scenarios, key parameters are extracted from the path start and end information, including the path start coordinates, feasible initial heading range, path end area boundary coordinates, and feasible termination heading range. These parameters constitute the core constraints of path planning. The path start coordinates restrict the starting position of the path; the feasible initial heading range specifies the heading selection range when starting the path; the path end area boundary coordinates limit the path's endpoint to be located within this area; and the feasible termination heading range requires that the path's heading when entering the endpoint area must be within this range. These constraints together ensure the feasibility and safety of the planned path.

[0122] Step S1382: Analyze the spatial location association relationship in the obstacle dynamic association knowledge feature, extract the current position coordinates and outline size information of all obstacles, determine the spatial occupancy range of each obstacle, and generate an obstacle spatial distribution dataset.

[0123] In unknown sea area exploration scenarios, the spatial location associations within the dynamic association knowledge features of obstacles contain the location information of the obstacles. This association is parsed to extract the current location coordinates of all obstacles, including coordinate values ​​in the X, Y, and Z directions. Simultaneously, the outline dimensions of each obstacle are obtained, such as length, width, and height. Based on the obstacle's location coordinates and outline dimensions, the spatial occupancy of each obstacle is determined, typically approximated by a cube or sphere. The spatial occupancy information of all obstacles is then integrated to generate an obstacle spatial distribution dataset.

[0124] Step S1383: Based on the coordinates of the path start point and the feasible initial heading range from the path start point, generate multiple different initial heading angles. Each initial heading angle is within the feasible initial heading range from the path start point, and adjacent initial heading angles are evenly spaced.

[0125] In unknown sea exploration scenarios, after determining the starting coordinates of the path and the feasible initial heading range, multiple different initial heading angles are generated within this range. These initial heading angles are selected using a uniform interval method, meaning the difference between any two adjacent initial heading angles is equal. The size of the interval is determined by the number of initial paths to be generated; the more paths to generate, the smaller the interval. Each initial heading angle serves as the starting direction of an initial path, ensuring the diversity of the generated initial paths.

[0126] Step S1384: For each initial heading angle, starting from the coordinates of the path starting point, plan the initial path segment according to the initial heading angle. The length of the initial path segment is determined based on the distance from the path starting point to the nearest obstacle. When planning the initial path segment, compare the trajectory coordinates of the initial path segment with the spatial occupancy coordinates of the obstacles in the obstacle spatial distribution data to ensure that the trajectory coordinates of the initial path segment are all outside the spatial occupancy coordinates of the obstacles.

[0127] In unknown sea exploration scenarios, for each initial heading angle, starting from the path origin coordinates, an initial path segment is planned according to that heading angle. The length of the initial path segment is not fixed but determined based on the distance from the path origin to the nearest obstacle. By querying the obstacle spatial distribution dataset, the obstacle closest to the path origin is found, and a certain percentage (e.g., 80%) of that distance is used as the length of the initial path segment to ensure sufficient safety distance from the obstacle. When planning the trajectory of the initial path segment, the trajectory coordinates are compared in real time with the spatial occupancy coordinates of the obstacles to ensure that the trajectory coordinates do not fall within the spatial occupancy of any obstacle. If a potential collision is detected, the length or direction of the initial path segment is adjusted.

[0128] Step S1385: At the end position of the initial path segment, based on the obstacle spatial distribution dataset, analyze the obstacle distribution around the end position of the initial path segment, determine the feasible turning course range at the end position of the initial path segment, and avoid obstacles around the end position of the initial path segment within the feasible turning course range.

[0129] In an unknown sea exploration scenario, after the initial path segment is planned, the destination position is reached. At this position, the distribution of obstacles within a certain range is analyzed using an obstacle spatial distribution dataset. By querying the dataset, obstacles around the destination position are located, and their positions and spatial occupancy are determined. Based on this information, a feasible turning course range at the destination position is determined, i.e., a range of heading angles that can avoid surrounding obstacles. The method for determining the feasible turning course range is similar to that for the feasible initial course range, obtained by eliminating heading angles that would lead to collisions.

[0130] Step S1386: Select multiple turning angles from the feasible turning heading range at the end position of the initial path segment, and plan subsequent path segments for each turning angle. The length of the subsequent path segments is also determined based on the distance from the starting point of the subsequent path segment to the nearest obstacle. When planning the subsequent path segments, compare the trajectory coordinates of the subsequent path segments with the spatial occupancy coordinates of the obstacles in the obstacle spatial distribution dataset, so that the trajectory coordinates of the subsequent path segments are all outside the spatial occupancy coordinates of the obstacles.

[0131] In unknown sea exploration scenarios, multiple turning angles are selected at even intervals from the feasible turning heading range starting from the endpoint of the initial path segment. For each turning angle, subsequent path segments are planned, using the endpoint of the initial path segment as the starting point. The length of subsequent path segments is determined in the same way as the initial path segment, based on the distance from the starting point of the segment to the nearest obstacle. During the planning process, the trajectory coordinates of subsequent path segments are also compared with the spatial occupancy coordinates of obstacles to ensure trajectory safety. Through this method, the path is gradually extended, approaching the endpoint area.

[0132] Step S1387: Repeat the above turning planning process, gradually extending the path segment towards the path endpoint area. Each extension is based on the obstacle distribution at the current path endpoint to determine the feasible heading at the current path endpoint and the path segment length at the current path endpoint.

[0133] In unknown sea exploration scenarios, the turning planning process in steps S1385 and S1386 is repeated continuously. After each planned path segment reaches a new endpoint, the distribution of obstacles around that endpoint is analyzed to determine the feasible turning course range, the turning angle is selected, and the next path segment is planned. This process is repeated gradually, extending the path towards the endpoint area until it is nearly there. During each extension, the feasible course and path segment length are determined based on the actual obstacle distribution at the current endpoint location, ensuring the safety and flexibility of the path.

[0134] Step S1388: When the planned path segment extends to the boundary of the path endpoint area, adjust the heading angle of the last path segment to make it conform to the feasible termination heading range for entering the path endpoint area, so that the path can smoothly enter the path endpoint area.

[0135] In uncharted waters exploration scenarios, when the planned path segment extends to a certain distance from the boundary of the path's endpoint area, preparations begin to enter the endpoint area. At this point, the heading angle of the final path segment needs to be adjusted to conform to the feasible termination heading range for entering the endpoint area. By calculating the distance and direction from the current path endpoint to the endpoint area boundary, as well as the feasible termination heading range, the heading angle of the final path segment is determined. Adjusting this heading angle ensures a smooth and safe entry into the endpoint area, completing the entire path planning.

[0136] Step S1389: Connect the planned multiple path segments sequentially to form a complete path trajectory from the starting point to the ending point of the path, which serves as the initial path structure.

[0137] In uncharted waters exploration scenarios, after multiple turns, planning, and path extensions, a series of path segments are obtained. Connecting these path segments sequentially according to the planned order forms a complete path trajectory from the starting point to the ending point. This trajectory is the initial path structure, composed of multiple segments, each with a clearly defined starting point, ending point, and heading angle. The initial path structure satisfies the core constraints of path planning, avoiding the spatial occupancy of all obstacles.

[0138] Step S13810: Repeat the above steps to generate multiple different initial path structures based on different initial heading angles and turning angles. The trajectory coordinates of each initial path structure are outside the spatial occupancy range of all obstacles.

[0139] In the scenario of exploring unknown sea areas, for each different initial heading angle generated in step S1383, path planning is performed according to the process of steps S1384 to S1389 to generate an initial path structure. Due to the different initial heading angles and the different turning angles selected during the turning planning process, the generated initial path structures are also different. Each initial path structure undergoes a comparison of its trajectory coordinates with the coordinates of the space occupied by obstacles to ensure it remains outside the space occupied by all obstacles, thus ensuring safety. Multiple different initial path structures are generated in this way.

[0140] Step S139: Based on the path optimization direction in the obstacle avoidance path evolution mechanism, perform preliminary optimization on each initial path structure, adjust the heading change nodes and path length of the initial path structure, so that the initial path structure meets the basic requirements of the path optimization direction.

[0141] In unknown sea exploration scenarios, the path optimization direction in the obstacle avoidance path evolution mechanism includes minimizing energy consumption, maximizing navigation efficiency, and ensuring the most stable attitude. For each initial path structure, preliminary optimization is performed based on the preset path optimization direction. For the minimum energy consumption optimization direction, energy consumption is reduced by adjusting the heading change nodes to decrease the number of turns and the turning angle; at the same time, the path length is shortened as much as possible to reduce the navigation distance. For the maximum navigation efficiency optimization direction, while ensuring safety, the length of the path segments and the heading are adjusted to make the path straighter, reduce unnecessary turns, and increase navigation speed. For the most stable attitude optimization direction, the position and turning angle of the heading change nodes are adjusted to make the heading change smoother and reduce the fluctuation of the vehicle's attitude. Through these adjustments, the initial path structure meets the basic requirements of the path optimization direction, resulting in the preliminarily optimized initial obstacle avoidance path.

[0142] Step S1310: Perform trajectory connection processing on each path after preliminary optimization to achieve a smooth transition between path segments, thereby forming a continuous path extending from the path start point to the path end point area, generating multiple initial obstacle avoidance paths that meet the initial path generation rules and basic optimization requirements.

[0143] In unknown sea exploration scenarios, the initially optimized path may suffer from insufficient smoothness in the connections between path segments, requiring frequent attitude adjustments by the vehicle during navigation. Therefore, trajectory connection processing is necessary for each path. Curve fitting or path smoothing algorithms are used to smooth the turning points in the path, connecting adjacent path segments with smooth curves. For example, Bézier curves or spline curves are used to fit the path trajectory, making the transitions between path segments more natural and smooth. After processing, a continuous path extending from the path's starting point to its ending point is formed. This continuous path conforms to the initial path generation rules and basic optimization requirements, serving as the initial obstacle avoidance path. Through the above processing, multiple initial obstacle avoidance paths are generated for subsequent dynamic evolution and adjustment.

[0144] Step S140: Based on the obstacle motion state association relationship in the obstacle dynamic association knowledge feature, dynamically evolve and adjust multiple initial obstacle avoidance paths to generate a target obstacle avoidance path that adapts to the real-time underwater environment, and obtain the target obstacle avoidance path generation result.

[0145] Step S141: Extract the obstacle motion state association relationship from the obstacle dynamic association knowledge features, obtain the motion direction, motion speed and motion trend prediction information of each obstacle, and generate obstacle dynamic prediction dataset.

[0146] In unknown sea area exploration scenarios, the motion state relationships of obstacles are extracted from the dynamic association knowledge features of obstacles. These motion state relationships include the motion direction (represented by azimuth angle), motion speed (including speed magnitude and direction), and motion trend prediction information (such as acceleration and trajectory prediction) for each obstacle. After processing this information, a dynamic obstacle prediction dataset is generated. This dataset can reflect the real-time motion of obstacles and their motion trends over a future period.

[0147] Step S142: Extract the path trajectory information of multiple initial obstacle avoidance paths. The path trajectory information of each initial obstacle avoidance path includes all coordinate points on the initial obstacle avoidance path, the heading changes between adjacent coordinate points on the initial obstacle avoidance path, and the path length information of the initial obstacle avoidance path, and generate an initial path set.

[0148] In an unknown sea exploration scenario, multiple initial obstacle avoidance paths are analyzed, and trajectory information for each path is extracted. Trajectory information includes all coordinate points on the path (arranged in navigation order), the heading change angles and rates of change between adjacent coordinate points, and the total length of the entire path. This information is then organized according to path numbers to generate an initial path set. This initial path set contains detailed trajectory data for all initial obstacle avoidance paths.

[0149] Step S143: For each initial obstacle avoidance path, based on the obstacle dynamic prediction dataset, predict the overlap between the path trajectory of the initial obstacle avoidance path and the space occupied by the obstacle within a preset time period in the future, and generate a path conflict prediction result.

[0150] For example, step S1431: Extract the complete trajectory information of an initial obstacle avoidance path, sort the coordinate points on the trajectory of the initial obstacle avoidance path according to the navigation order, and generate a path time coordinate sequence for each coordinate point corresponding to an estimated arrival time.

[0151] In an unknown sea exploration scenario, an initial obstacle avoidance path is selected, and its complete trajectory information is extracted. The coordinate points in the trajectory information are sorted according to the order of the vehicle's voyage, forming an ordered sequence of coordinate points. Based on the vehicle's preset voyage speed and the distance between each coordinate point, the estimated time to reach each coordinate point is calculated. The coordinate points are then associated with their corresponding estimated arrival times to generate a path time coordinate sequence. This path time coordinate sequence describes the location the vehicle will reach at different time points.

[0152] Step S1432: Extract obstacle information related to the trajectory range of the initial obstacle avoidance path from the obstacle dynamic prediction dataset, including the current position of the relevant obstacle, the direction of movement of the relevant obstacle, the speed of movement of the relevant obstacle, and the predicted movement trend information of the relevant obstacle.

[0153] In unknown sea exploration scenarios, the trajectory range of the initial obstacle avoidance path is a specific spatial region. From the obstacle dynamic prediction dataset, obstacles located near this trajectory range or whose trajectories may intersect with it are identified; these obstacles are considered relevant to the initial obstacle avoidance path. Information about these relevant obstacles is extracted, including their current position coordinates, direction of movement, speed, and predicted movement trends (such as predicted position over a future period). This information is used to predict the obstacle's future position and assess the risk of conflict with the path trajectory.

[0154] Step S1433: Based on the movement direction and speed of the relevant obstacles, calculate the predicted position coordinates of the relevant obstacles at different time points within a preset time period in the future, and generate an obstacle time position sequence, with each time point corresponding one-to-one with the time point of the path time coordinate sequence.

[0155] In the scenario of exploring unknown sea areas, for each relevant obstacle, based on its direction and speed of movement and the current time, the predicted position coordinates at different points in a preset future time period are calculated. The selected time points correspond one-to-one with the time points in the path time coordinate sequence for spatiotemporal matching. For example, if the path time coordinate sequence contains time points t1, t2, t3, etc., the predicted position coordinates of the obstacle at time points t1, t2, t3, etc., are calculated. These predicted position coordinates are then arranged in chronological order to generate an obstacle time position sequence.

[0156] Step S1434: Combine the outline size information of the relevant obstacles to determine the spatial occupancy range of the relevant obstacles corresponding to each predicted position coordinate, and generate a sequence of dynamic occupancy ranges of obstacles.

[0157] In unknown sea area exploration scenarios, the outline dimensions of relevant obstacles include parameters such as length, width, and height. For each predicted position coordinate in the obstacle's time position sequence, combined with its outline dimensions, the spatial occupancy range of that obstacle at that time point is determined. The spatial occupancy range is typically represented by a cube or sphere, with its center at the predicted position coordinates, and its size determined by the outline dimensions. Arranging the spatial occupancy ranges at different time points in chronological order generates a dynamic obstacle occupancy range sequence. This dynamic obstacle occupancy range sequence describes the change in the spatial occupancy of obstacles over a predetermined future time period.

[0158] Step S1435: Extract each coordinate point and its corresponding arrival time from the path time coordinate sequence, and find the spatial occupancy range of related obstacles at the same time point in the obstacle dynamic occupancy range sequence.

[0159] In unknown sea exploration scenarios, each coordinate point and its corresponding arrival time are extracted sequentially from the path time coordinate sequence. For each arrival time, the spatial occupancy range of related obstacles at the same time point is searched in the obstacle dynamic occupancy range sequence. In this way, the position of the vehicle at that time point can be compared with the spatial occupancy range of the obstacles at the same time point to determine whether there is a risk of collision.

[0160] Step S1436: Determine whether the path coordinate point in the path time coordinate sequence is within the spatial occupancy range of the corresponding related obstacle. If the path coordinate point in the path time coordinate sequence is within the spatial occupancy range of the corresponding related obstacle, mark the path coordinate point as a conflict point and record the coordinates of the conflict point and the time point corresponding to the conflict point.

[0161] In unknown sea area exploration scenarios, the path coordinates in the path time coordinate sequence are compared with the spatial occupancy of obstacles at the same time point. If the path coordinates are located within the spatial occupancy of obstacles, the point is considered to have a collision risk and is marked as a conflict point. The coordinates (path coordinates) and corresponding time points of the conflict points are recorded for subsequent conflict analysis.

[0162] Step S1437: Count the number and distribution of conflict points on the entire initial obstacle avoidance path. If there are conflict points, further analyze the continuous distribution length of the conflict points to determine whether a continuous conflict section has been formed.

[0163] In unknown sea exploration scenarios, the number of conflict points marked along the initial obstacle avoidance path and their distribution along the path are statistically analyzed. If conflict points exist, their continuous distribution is further analyzed. If multiple conflict points appear consecutively along the path and the distance between them is less than a preset threshold, these conflict points are considered to form a continuous conflict segment. A continuous conflict segment indicates a high collision risk within that segment of the path, requiring focused adjustments.

[0164] Step S1438: If there are continuous conflict segments, generate a path conflict prediction result, specifying the starting coordinates of the continuous conflict segments, the ending coordinates of the continuous conflict segments, the time range corresponding to the continuous conflict segments, and the obstacle information involved in the continuous conflict segments; if there are no conflict points or only isolated non-continuous conflict points, determine that the initial obstacle avoidance path will not have significant path conflicts in the future preset time period, and generate a conflict-free prediction result.

[0165] In unknown sea area exploration scenarios, if analysis reveals continuous conflict segments, a detailed path conflict prediction result is generated. This result includes the start and end coordinates of the continuous conflict segments, specifying their spatial extent; the corresponding time range, indicating the time period during which the conflict occurred; and information about the obstacles involved, such as obstacle number, type, and movement status. If there are no conflict points on the path, or only isolated, discontinuous conflict points exist (the distance between these points is greater than a preset threshold, preventing the formation of continuous conflict segments), then the initial obstacle avoidance path is determined to have no significant path conflict within a preset time period, generating a conflict-free prediction result.

[0166] Step S1439: Repeat the above steps to predict path conflicts for all initial obstacle avoidance paths and generate path conflict prediction results for each initial obstacle avoidance path.

[0167] In the scenario of exploring unknown sea areas, following steps S1431 to S1438, path conflict prediction is performed on multiple initial obstacle avoidance paths one by one. Each path generates a corresponding path conflict prediction result, which is either a conflict prediction result with continuous conflict segments or a conflict-free prediction result.

[0168] Step S144: For an initial obstacle avoidance path with path conflict prediction results, based on the path dynamic adjustment basis in the obstacle avoidance path evolution mechanism, determine the starting node and adjustment direction of the path adjustment for the initial obstacle avoidance path with path conflict prediction results. The starting node of the path adjustment is selected as the location of the closest pre-conflict location to the current position of the underwater vehicle.

[0169] In unknown sea area exploration scenarios, initial obstacle avoidance paths with predicted path conflicts require dynamic adjustments. Based on the dynamic adjustment criteria within the obstacle avoidance path evolution mechanism, the starting node and direction of the adjustment are determined. The starting node is selected as the closest pre-conflict location to the current position of the underwater vehicle; the pre-conflict location refers to a safe position before the conflict segment begins. The adjustment direction is determined based on the obstacle's movement direction and spatial distribution, selecting a direction that avoids the conflict segment and conforms to the path optimization direction. For example, if the obstacle moves to the right, the adjustment direction might be to the left, up, or down, depending on the specific circumstances.

[0170] Step S145: Based on the adjustment direction of the path adjustment and the obstacle movement trend prediction information, replan the path segment after the starting node of the path adjustment. When replanning, ensure that the new path segment is outside the predicted obstacle space occupancy range. At the same time, smooth the heading change nodes in the replanned path segment to achieve a smooth heading transition between each path segment.

[0171] In unknown sea exploration scenarios, after determining the starting node and direction of path adjustment, the path segment is replanned from the starting node. During replanning, obstacle movement trend prediction information is used to predict future obstacle position changes, ensuring that the newly planned path segment is outside the predicted obstacle space. Simultaneously, heading change nodes in the replanned path segment are smoothed using a path smoothing algorithm, making heading changes between adjacent path segments more gradual, reducing the number and magnitude of attitude adjustments required by the vehicle, and ensuring navigation stability.

[0172] Step S146: Smoothly connect the path segment after the starting node of the replanned path adjustment with the path trajectory of the initial obstacle avoidance path before the starting node of the path adjustment, update the trajectory information of the initial obstacle avoidance path, and generate the adjusted obstacle avoidance path.

[0173] In unknown sea exploration scenarios, after replanning the path segment after the starting node of the adjusted path, it needs to be connected to the initial obstacle avoidance path trajectory before the starting node. A smooth transition is used during the connection to ensure a smooth heading change at the connection point and avoid sharp turns. A path smoothing algorithm is used to optimize the trajectory near the connection point, making the entire path trajectory continuous and smooth. The trajectory information of the initial obstacle avoidance path is updated, and the adjusted path segment replaces the original conflicting sections, generating the adjusted obstacle avoidance path.

[0174] Step S147: For an initial obstacle avoidance path without path conflict prediction results, based on the path optimization direction in the obstacle avoidance path evolution mechanism, perform detailed optimization on the path trajectory of the initial obstacle avoidance path without path conflict prediction results, adjust the heading change node positions on the path of the initial obstacle avoidance path without path conflict prediction results, and shorten the total path length of the initial obstacle avoidance path without path conflict prediction results.

[0175] In unknown sea exploration scenarios, for initial obstacle avoidance paths without path conflict prediction results, although there is currently no collision risk, detailed optimization can still be performed based on the path optimization direction. For example, for the lowest energy consumption optimization direction, the positions of heading change nodes on the path can be adjusted to reduce the number of turns and turning angles, thereby reducing energy consumption; for the highest navigation efficiency optimization direction, the total path length can be shortened and navigation efficiency improved by optimizing the length and direction of path segments. During the adjustment process, it is ensured that the path trajectory still avoids the space occupied by all obstacles and does not introduce new conflict risks.

[0176] Step S148: Extract the key parameters of the adjusted obstacle avoidance path and the optimized obstacle avoidance path, including the total path length of the adjusted obstacle avoidance path, the number of heading changes of the adjusted obstacle avoidance path, the minimum distance between the adjusted obstacle avoidance path and the obstacle, the estimated navigation energy consumption of the adjusted obstacle avoidance path, the total path length of the optimized obstacle avoidance path, the number of heading changes of the optimized obstacle avoidance path, the minimum distance between the optimized obstacle avoidance path and the obstacle, and the estimated navigation energy consumption of the optimized obstacle avoidance path, and generate a path evaluation parameter set.

[0177] In the scenario of exploring unknown sea areas, parameters are extracted from the adjusted and optimized obstacle avoidance paths. Key extracted parameters include total path length (reflecting the path's length), number of heading changes (reflecting the path's smoothness), minimum distance to obstacles (measuring path safety), and estimated navigation energy consumption (evaluating path economics). These parameters are extracted separately for the adjusted and optimized obstacle avoidance paths, and then integrated to generate a path evaluation parameter set.

[0178] Step S149: Based on the path evaluation parameter set, sort all adjusted and optimized obstacle avoidance paths in multiple dimensions. The sorting dimensions include total path length, number of heading changes, minimum distance to obstacles, and estimated navigation energy consumption. Select the path with the highest ranking.

[0179] In unknown sea exploration scenarios, each parameter in the path evaluation parameter set reflects the path's performance from different dimensions. Based on these parameters, all adjusted and optimized obstacle avoidance paths are ranked across multiple dimensions. During ranking, each dimension (total path length, number of heading changes, minimum distance to obstacles, and estimated navigation energy consumption) is assigned a certain weight, determined according to the priority of mission requirements. For example, if safety is prioritized, the minimum distance to obstacles is given a larger weight; if efficiency is prioritized, the total path length is given a larger weight. A comprehensive score is calculated for each path, and paths are ranked according to their scores. The top-ranked paths are selected as candidate obstacle avoidance paths.

[0180] Step S1410: Determine the final target obstacle avoidance path from the top-ranked paths, record the complete trajectory information of the target obstacle avoidance path and the evaluation parameters of the target obstacle avoidance path, and obtain the target obstacle avoidance path generation result.

[0181] In unknown sea exploration scenarios, paths ranked higher typically exhibit better overall performance. From these paths, the final obstacle avoidance path is determined based on pre-defined decision rules (e.g., selecting the path with the highest overall score, or choosing the safest path when scores are similar). The complete trajectory information of the target obstacle avoidance path is recorded, including all coordinate points and heading changes; simultaneously, evaluation parameters such as total path length, number of heading changes, minimum distance to obstacles, and estimated navigation energy consumption are also recorded. Integrating this information yields the target obstacle avoidance path generation result.

[0182] Step S150: Associate and map the target obstacle avoidance path generation result with the real-time navigation parameters of the underwater vehicle to generate real-time control commands for the underwater vehicle. Transmit the real-time control commands to the execution control system of the underwater vehicle to drive the underwater vehicle to navigate along the target obstacle avoidance path. The real-time control commands include heading correction commands, speed adjustment commands, and power distribution commands.

[0183] For example, step S151: extract the complete trajectory information of the target obstacle avoidance path from the target obstacle avoidance path generation result, decompose the trajectory information of the target obstacle avoidance path into multiple continuous path segments, each path segment corresponding to a fixed heading angle, and generate a path segment heading sequence.

[0184] In unknown sea exploration scenarios, the complete trajectory information of a target obstacle avoidance path is a continuous curve or broken line. This trajectory information is decomposed into multiple continuous path segments according to the changes in heading angle. Each path segment has a fixed heading angle, within which the vehicle maintains a constant heading. The heading angles of these path segments are arranged in sequence to generate a path segment heading sequence. This path segment heading sequence describes the heading angles that the vehicle needs to maintain sequentially during navigation.

[0185] Step S152: Analyze the real-time navigation parameters, extract the current heading angle, current speed, and current power output distribution of the underwater vehicle, and determine the deviation between the current navigation state of the underwater vehicle and the target obstacle avoidance path.

[0186] In unknown sea exploration scenarios, real-time navigation parameters contain information about the vehicle's current navigation status. These parameters are analyzed to extract the current heading angle, current speed, and current power output distribution (e.g., the output power ratio of each thruster). The current heading angle is compared with the heading angle of the first path segment in the path sequence to calculate the heading deviation; the current speed is compared with the expected speed at the corresponding position in the target obstacle avoidance path to calculate the speed deviation. Simultaneously, it is analyzed whether the current power output distribution can meet the power requirements of the target path. By combining these deviations, the overall deviation between the vehicle's current navigation status and the target obstacle avoidance path is determined.

[0187] Step S153: For each path segment, calculate the difference between the current heading angle of the underwater vehicle and the target heading angle of the path segment. Based on this difference and the path dynamic adjustment basis in the obstacle avoidance path evolution mechanism, determine the heading correction amount.

[0188] In unknown sea exploration scenarios, for each path segment in the path segment heading sequence (starting from the current path segment to be executed), the difference between the vehicle's current heading angle and the target heading angle is calculated. Based on the magnitude and direction of this difference, and combined with the path dynamic adjustment criteria in the obstacle avoidance path evolution mechanism (such as the maximum allowable amplitude and rate of change of heading), the heading correction amount is determined. The heading correction amount includes the correction direction (left or right) and the correction magnitude (angle value), ensuring a smooth and safe heading adjustment process.

[0189] Step S154: Generate a heading correction command based on the heading correction amount. The heading correction command includes information on the correction direction and correction magnitude. This heading correction command is used to drive the heading adjustment mechanism of the underwater vehicle to adjust the heading of the underwater vehicle to the heading angle of the corresponding path segment.

[0190] In unknown sea exploration scenarios, a course correction command is generated based on a determined course correction amount. The command specifies the correction direction (e.g., left or right rudder) and the correction magnitude (e.g., rudder angle). This command is transmitted to the course adjustment mechanism (e.g., servo system) of the underwater vehicle, driving the servo to change the vehicle's course. During the adjustment process, changes in the course angle are monitored in real time, and adjustment stops when the target course angle is reached.

[0191] Step S155: Calculate the optimal speed for each path segment by combining the length of each path segment and the heading change of adjacent path segments. The optimal speed for each path segment balances navigation efficiency and attitude stability.

[0192] In unknown sea exploration scenarios, the length of each path segment and the magnitude of course changes between adjacent path segments vary, thus requiring different navigation speeds. When the path segment is long and the magnitude of course changes between adjacent path segments is small, a higher navigation speed can be set to improve navigation efficiency. Conversely, when the path segment is short or the magnitude of course changes between adjacent path segments is large, a lower navigation speed is needed to ensure attitude stability and facilitate timely course adjustments. Taking all these factors into account, the optimal navigation speed for each path segment is calculated.

[0193] Step S156: Calculate the difference between the current speed of the underwater vehicle and the optimal speed corresponding to the path segment. Based on the difference and the dynamic response characteristics of the underwater vehicle, determine the speed adjustment amount.

[0194] In unknown sea exploration scenarios, the current speed of the underwater vehicle is compared with the optimal speed corresponding to the current path segment, and the speed difference is calculated. Based on the sign (current speed is higher or lower than the optimal speed) and magnitude of this difference, combined with the dynamic response characteristics of the underwater vehicle (such as the time constants of acceleration and deceleration, maximum acceleration and deceleration, etc.), the speed adjustment amount is determined. The speed adjustment amount includes the adjustment direction (acceleration or deceleration) and the adjustment rate (the change in speed per unit time).

[0195] Step S157: Generate a speed adjustment command based on the speed adjustment amount. The speed adjustment command includes an acceleration or deceleration command and adjustment rate information. Drive the power adjustment mechanism of the underwater vehicle through the speed adjustment command to transition the underwater vehicle's travel speed to the optimal travel speed corresponding to the path segment.

[0196] In unknown sea exploration scenarios, a speed adjustment command is generated based on a determined speed adjustment amount. The command specifies whether it's acceleration or deceleration, and the rate of adjustment (e.g., the speed increase or decrease per second). This command is transmitted to the underwater vehicle's power regulation mechanism (e.g., the thruster control system), which controls the thruster's output power to transition the vehicle's speed to the optimal speed corresponding to that path segment at a specified rate. During speed adjustment, speed changes are monitored in real time to ensure a smooth speed transition.

[0197] Step S158: Based on the requirements of the course correction command and speed adjustment command, and combined with the power system structure of the underwater vehicle, allocate the power output ratio of each power output unit of the underwater vehicle and generate a power distribution command.

[0198] In uncharted waters exploration scenarios, course correction and speed adjustment commands place specific demands on the vehicle's power output. Course correction may require adjusting the power difference between the left and right thrusters, while speed adjustment requires adjusting the total power output. Considering the underwater vehicle's propulsion system structure (such as the number, arrangement, and power characteristics of the thrusters), the power output ratio of each power output unit (such as the main thruster and auxiliary thrusters) is calculated based on the needs of course correction and speed adjustment. For example, when a left turn is required, the power of the left thruster is reduced or the power of the right thruster is increased; when acceleration is required, the power of all thrusters is increased simultaneously. This power output ratio information is then used to generate a power distribution command.

[0199] Step S159: Integrate the heading correction command, speed adjustment command, and power distribution command, add the time node information for command execution, and generate the real-time control command.

[0200] In uncharted waters exploration scenarios, course correction commands, speed adjustment commands, and power distribution commands are integrated. Each command is assigned execution time information according to its execution order and timing requirements to ensure correct execution sequence. For example, the course correction command is executed first, followed by the speed adjustment command after the course is adjusted. After integration, a complete real-time control command is generated, containing all the control information required for the vehicle's navigation.

[0201] Step S1510: Real-time control commands are transmitted to the execution control system of the underwater vehicle through the internal data transmission link of the underwater vehicle. The execution control system of the underwater vehicle drives the corresponding actuators to move according to the real-time control commands, so that the underwater vehicle travels along the target obstacle avoidance path.

[0202] In uncharted waters exploration scenarios, once real-time control commands are generated, they are transmitted to the execution control system via a high-speed data transmission link (such as a CAN bus or Ethernet) within the underwater vehicle. The execution control system parses the received real-time control commands and drives the corresponding actuators to perform actions based on the command content. For example, a course correction command is sent to the servo control system, driving the servo to rotate and adjust the course; speed adjustment and power distribution commands are sent to the propulsion control system, controlling the propulsion speed and output power. Through the coordinated actions of these actuators, the underwater vehicle navigates along the target obstacle avoidance path.

[0203] In one exemplary embodiment, an intelligent obstacle avoidance planning system for underwater obstacles is provided. This system can be a terminal, server, etc., and its internal structure diagram can be as follows: Figure 2As shown, the intelligent obstacle avoidance planning system for underwater obstacles includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements an intelligent obstacle avoidance planning method for underwater obstacles. The display unit is used to generate a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the shell of an intelligent obstacle avoidance planning system for underwater obstacles, or an external keyboard, touchpad, or mouse, etc.

[0204] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. An intelligent obstacle avoidance planning method for underwater obstacles, characterized in that, The method includes: Receive real-time navigation parameters of the underwater vehicle and underwater environment trigger signals, dynamically define the sensing range of the underwater environment based on the real-time navigation parameters and the underwater environment trigger signals, and generate a dynamic definition result of the sensing range; Based on the dynamic definition of the perception range, the multi-mode perception devices carried by the underwater vehicle are driven to work together to collect multi-dimensional underwater environmental data within the perception range, and the multi-dimensional underwater environmental data is correlated and fused to generate dynamic association knowledge features of obstacles. By combining the dynamic association knowledge features of the obstacles and the real-time navigation parameters, an obstacle avoidance path evolution mechanism is constructed, and multiple initial obstacle avoidance paths are generated through the obstacle avoidance path evolution mechanism; Based on the obstacle motion state association relationship in the obstacle dynamic association knowledge feature, multiple initial obstacle avoidance paths are dynamically evolved and adjusted to generate a target obstacle avoidance path that adapts to the real-time underwater environment, and the target obstacle avoidance path generation result is obtained. The target obstacle avoidance path generation result is correlated and mapped with the real-time navigation parameters of the underwater vehicle to generate real-time control commands for the underwater vehicle. The real-time control commands are transmitted to the execution control system of the underwater vehicle to drive the underwater vehicle to navigate along the target obstacle avoidance path. The real-time control commands include heading correction commands, speed adjustment commands, and power distribution commands. Based on the dynamic definition of the sensing range, the multi-mode sensing devices mounted on the underwater vehicle are driven to work collaboratively to collect multi-dimensional underwater environmental data within the sensing range. This multi-dimensional underwater environmental data is then correlated and fused to generate dynamic obstacle association knowledge features, including: The dynamic definition result of the sensing range is analyzed, and the spatial boundary information, depth range information and environmental interference level information of the sensing range are extracted. Based on the spatial boundary information, multiple collaborative sensing sub-regions are divided, and each collaborative sensing sub-region corresponds to a clear spatial range. Based on the environmental interference level information of multiple collaborative sensing sub-regions, the working mode of the multi-mode sensing device is assigned to each collaborative sensing sub-region. The working mode includes parameters related to acoustic detection frequency and optical imaging resolution, and collaborative working parameters of the sensing device are generated. Based on the collaborative working parameters of the sensing devices, the acoustic detection device in the multi-mode sensing device is activated to transmit directional acoustic detection signals to each collaborative sensing sub-region, receive the reflected acoustic signals, and convert the reflected acoustic signals into digital acoustic detection data. The acoustic detection data includes information related to signal reflection time, signal amplitude change, and signal frequency shift. The optical imaging device in the multi-mode sensing device is started synchronously to continuously image and acquire images of each cooperative sensing sub-region to obtain raw optical imaging data. The raw optical imaging data is then processed for light compensation to correct the imaging deviation caused by underwater light attenuation and generate clear optical imaging data. Collect acoustic detection data and clear optical imaging data corresponding to all collaborative sensing sub-regions, construct a multi-dimensional underwater environment dataset, and add corresponding collaborative sensing sub-region identifiers and acquisition time identifiers to each data entry; The signal features of acoustic detection data and the image features of clear optical imaging data are extracted from the underwater environment multi-dimensional data set. The signal features include reflected signal peak features and signal attenuation features, and the image features include object edge features and texture distribution features. Establish a cross-modal association mapping between signal features and image features. By using the same collaborative sensing sub-region identifier and acquisition time identifier, the acoustic signal features and optical image features belonging to the same obstacle are associated and bound to generate a cross-modal feature set of the obstacle. Based on the cross-modal feature set of the obstacles, the spatial coordinates of each obstacle are extracted, the spatial distance and relative orientation between different obstacles are analyzed, and the spatial position correlation of the obstacles is generated. By continuously collecting multi-dimensional underwater environmental data, the spatial position changes of each obstacle are tracked, the position change trend of the obstacle is calculated, and the motion state correlation of the obstacle is generated by combining the underwater water flow direction and velocity information. The acoustic signal reflection characteristics and optical image texture features of each obstacle are analyzed, and the correspondence between the material and signal features of common underwater obstacles is correlated to generate the material property correlation of obstacles. The spatial position correlation of obstacles, the motion state correlation of obstacles, and the material property correlation of obstacles are integrated to generate dynamic correlation knowledge features of obstacles.

2. The intelligent perception and obstacle avoidance planning method for underwater obstacles according to claim 1, characterized in that, The obstacle avoidance path evolution mechanism is constructed by combining the dynamic association knowledge features of the obstacles and the real-time navigation parameters. Multiple initial obstacle avoidance paths are generated through this mechanism, including: The dynamic association knowledge features of the obstacles are analyzed, and the spatial position association, motion state association, and material property association of the obstacles are extracted to generate an environmental situation description of the obstacles; The real-time navigation parameters are analyzed to extract the navigation heading, speed, power output status and attitude stability information of the underwater vehicle, determine the current navigation capability and subsequent navigation potential of the underwater vehicle, and generate navigation status assessment results. Based on the obstacle environment situation description and the navigation status assessment results, initial path generation rules are set. The initial path generation rules include the definition criteria for the path start and end points, the safety distance criteria between the path and obstacles, and the smoothness criteria for path heading changes. By combining the motion state correlation of obstacles and the characteristics of underwater water flow, a dynamic path adjustment basis is set. The dynamic path adjustment basis includes the response threshold of obstacle position change, the influence coefficient of water flow disturbance on the path, and the adaptation range of navigation attitude adjustment. Based on the power output status and energy reserves of the underwater vehicle, a path optimization direction is set, which includes the optimization direction with the lowest energy consumption, the optimization direction with the highest navigation efficiency, and the optimization direction with the most stable attitude. By integrating the initial path generation rules, the dynamic path adjustment criteria, and the path optimization direction, the obstacle avoidance path evolution mechanism is constructed, clarifying the core principles to be followed in the path generation and adjustment process; Based on the path initial generation rules in the obstacle avoidance path evolution mechanism, the current position of the underwater vehicle is determined as the path starting point, and the path ending area is determined in combination with the preset navigation target direction to generate path start and end information. Based on the path start and end information and the spatial relationship of the obstacles, multiple initial path structures are planned from the path start point to the path end point area. During the planning process, the path trajectory coordinates are compared with the spatial occupancy coordinates of the obstacles so that the planned path trajectory is outside the spatial occupancy of all obstacles. Based on the path optimization direction in the obstacle avoidance path evolution mechanism, each initial path structure is initially optimized by adjusting the heading change nodes and path length of the initial path structure to make the initial path structure meet the basic requirements of the path optimization direction. Each path after preliminary optimization is processed to connect the paths to achieve a smooth transition between the path segments, thereby forming a continuous path extending from the starting point to the ending point of the path, generating multiple initial obstacle avoidance paths that meet the initial path generation rules and basic optimization requirements.

3. The intelligent perception and obstacle avoidance planning method for underwater obstacles according to claim 1, characterized in that, The step of activating the acoustic detection device in the multi-mode sensing device based on the collaborative working parameters of the sensing device, transmitting directional acoustic detection signals to each collaborative sensing sub-region, receiving the reflected acoustic signals, and converting the reflected acoustic signals into digital acoustic detection data includes: Adjust the transmission power and signal waveform of the acoustic detection equipment. The transmission power is set based on the environmental interference level of the corresponding cooperative sensing sub-region, and the signal waveform is selected based on the underwater propagation characteristics. According to the distribution order of the collaborative sensing sub-regions, the emission direction of the acoustic detection device is aligned with the center position of each collaborative sensing sub-region in turn, and the emission angle and emission range parameters of the acoustic detection device are adjusted so that the coverage of the directional acoustic detection signal can completely include the current collaborative sensing sub-region. The signal transmission module of the acoustic detection equipment is activated to transmit directional acoustic detection signals to the currently aligned cooperative sensing sub-region, while simultaneously recording the precise time point of the directional acoustic detection signal transmission. Keeping the transmission direction of the acoustic detection device unchanged, the signal receiving module of the acoustic detection device is activated to capture the acoustic reflection signal reflected back from the current cooperative sensing sub-region in real time and continuously receive it for a preset duration; Record the precise time point of each received acoustic reflection signal, calculate the difference between the time point of transmission of the directional acoustic detection signal and the time point of reception of the acoustic reflection signal, and obtain the signal reflection duration corresponding to each acoustic reflection signal; The amplitude of the received acoustic reflection signal is detected, and the amplitude change data of the acoustic reflection signal during the propagation process is collected. The peak and valley values ​​of the amplitude change data and the corresponding time nodes are recorded to obtain the signal amplitude change information. Analyze the frequency characteristics of the acoustic reflection signal, compare it with the original frequency of the directional acoustic detection signal, calculate the offset of the acoustic reflection signal relative to the original frequency of the directional acoustic detection signal, and obtain the signal frequency offset information. The collected signal reflection time, signal amplitude variation information and signal frequency offset information are digitally converted to convert analog signal data into digital data in standard format. According to the identifier of the collaborative sensing sub-region, the digital signal data corresponding to each collaborative sensing sub-region is associated and bound to form an acoustic data subset divided by the collaborative sensing sub-region; The acoustic data subsets of all collaborative sensing sub-regions are integrated, and the acquisition parameters and environmental interference level information corresponding to each acoustic data subset are supplemented to generate the digital acoustic detection data.

4. The intelligent perception and obstacle avoidance planning method for underwater obstacles according to claim 1, characterized in that, The optical imaging device in the synchronously activated multi-mode sensing device continuously acquires images of each collaborative sensing sub-region to obtain raw optical imaging data. The raw optical imaging data undergoes light compensation processing to correct imaging deviations caused by underwater light attenuation, generating clear optical imaging data, including: Based on the collaborative working parameters of the sensing devices, the imaging resolution, exposure time, and photosensitivity parameters of the optical imaging devices are adjusted so that the parameter settings of the optical imaging devices are adapted to the lighting conditions and environmental interference levels of the corresponding collaborative sensing sub-regions. Align the lens of the optical imaging device with each collaborative sensing sub-region in turn, adjust the focal length and shooting angle of the lens of the optical imaging device so that the entire range of the collaborative sensing sub-region falls within the imaging field of view of the optical imaging device, and fix the lens posture of the optical imaging device. The continuous imaging function of the optical imaging device is activated, and the currently aligned cooperative sensing sub-region is continuously imaged according to the preset acquisition interval, and multiple frames of images are acquired as the raw optical imaging data of the cooperative sensing sub-region. Record the acquisition time, corresponding collaborative sensing sub-region identifier, and illumination conditions of each frame of raw optical imaging data, and establish a raw optical imaging data association table. Extract brightness distribution information from the raw optical imaging data, analyze the brightness differences in different regions of the raw optical imaging data, and identify dark areas caused by underwater light attenuation. Based on the underwater light attenuation model and the illumination conditions during data collection, the light compensation coefficient for the dark area is calculated, and the light compensation coefficient is dynamically adjusted according to the location and depth of the dark area. Based on the calculated light compensation coefficient, the brightness of the dark areas in the original optical imaging data is enhanced, and the pixel brightness values ​​in the dark areas are adjusted so that the details in the dark areas can be clearly presented. Color correction is performed on the enhanced optical imaging data to correct the color shift caused by underwater light refraction, so that the corrected image color is closer to the color of the real underwater environment. Noise interference in the color-corrected optical imaging data is removed, while object outlines and texture details are preserved to obtain a clear single-frame optical image. The clear optical imaging data is generated by integrating multiple clear optical images from all collaborative sensing sub-regions, associating them with the corresponding acquisition time and collaborative sensing sub-region identifiers.

5. The intelligent perception and obstacle avoidance planning method for underwater obstacles according to claim 2, characterized in that, The path initial generation rule based on the obstacle avoidance path evolution mechanism determines the current position of the underwater vehicle as the path starting point, and combines it with the preset navigation target direction to determine the path ending area, generating path start and end information, including: The path initial generation rules in the obstacle avoidance path evolution mechanism are analyzed, and the criteria for defining the path start and end are extracted. The path start is completely consistent with the current position of the underwater vehicle, and the path end is located in a specific area in the direction of the preset navigation target. The current position coordinates of the underwater vehicle are extracted from the real-time navigation parameters. The current position coordinates of the underwater vehicle are determined based on the underwater global coordinate system. The current position coordinates of the underwater vehicle are directly used as the path start coordinates, and the path start coordinate information is recorded. Obtain the preset underwater vehicle navigation target direction, and in combination with the underwater vehicle navigation mission requirements, determine the safe zone range in the preset underwater navigation target direction. The safe zone range in the preset underwater navigation target direction is far away from known fixed obstacle areas. The safe zone in the direction of the preset underwater navigation target is spatially divided, and the core area within the safe zone in the direction of the preset underwater navigation target is determined as the path endpoint area. The size of the path endpoint area is set according to the size of the underwater navigation vehicle and the navigation accuracy requirements. Extract the boundary coordinate information of the path endpoint region, determine the center coordinates of the path endpoint region as the reference endpoint coordinates, and record the boundary coordinates and reference endpoint coordinates of the path endpoint region. Calculate the straight-line distance between the starting point coordinates and the reference ending point coordinates. Combine the underwater vehicle's speed and energy reserves to make a preliminary estimate of the time required for straight-line travel between the starting point coordinates and the reference ending point coordinates. Analyze the distribution of obstacles around the path start point, and combine the safety distance standard in the path initial generation rules to determine the feasible initial course range from the path start point. The feasible initial course range from the path start point avoids obstacles near the path start point. Analyze the distribution of obstacles around the path endpoint area, and combine the safety distance standard in the path initial generation rules to determine the feasible termination course range for entering the path endpoint area. The feasible termination course range for entering the path endpoint area enables the underwater vehicle to smoothly enter the path endpoint area. The coordinates of the path start point, the feasible initial heading range from the path start point, the boundary coordinates of the path end point area, the reference end point coordinates, and the feasible termination heading range for entering the path end point area are integrated. Supplement the environmental information related to the start and end of the path, including the water flow direction, water flow velocity, and lighting conditions of the start and end areas of the path, and generate the path start and end information.

6. The intelligent perception and obstacle avoidance planning method for underwater obstacles according to claim 2, characterized in that, The step of planning multiple initial path structures from the path start point to the path end point region based on the path start and end information and the spatial location association of the obstacles includes: Extract the path start coordinates, feasible initial heading range from the path start, boundary coordinates of the path end area, and feasible termination heading range into the path end area from the path start and end information to determine the core constraints of path planning; The spatial location relationships in the dynamic association knowledge features of the obstacles are analyzed, the current position coordinates and outline size information of all obstacles are extracted, the spatial occupancy range of each obstacle is determined, and an obstacle spatial distribution dataset is generated. Based on the coordinates of the path start point and the feasible initial heading range from the path start point, multiple different initial heading angles are generated. Each initial heading angle is within the feasible initial heading range from the path start point, and adjacent initial heading angles are evenly spaced. For each initial heading angle, starting from the coordinates of the path start point, an initial path segment is planned according to that initial heading angle. The length of the initial path segment is determined based on the distance from the path start point to the nearest obstacle. When planning the initial path segment, the trajectory coordinates of the initial path segment are compared with the spatial occupancy coordinates of the obstacles in the obstacle spatial distribution data, so that the trajectory coordinates of the initial path segment are all outside the spatial occupancy coordinates of the obstacles. At the end of the initial path segment, based on the obstacle spatial distribution dataset, the distribution of obstacles around the end of the initial path segment is analyzed to determine the feasible turning course range at the end of the initial path segment. The feasible turning course range at the end of the initial path segment avoids the obstacles around the end of the initial path segment. Multiple turning angles are selected from the feasible turning heading range at the end position of the initial path segment. For each turning angle, subsequent path segments are planned. The length of the subsequent path segments is also determined based on the distance from the starting point of the subsequent path segment to the nearest obstacle. When planning the subsequent path segments, the trajectory coordinates of the subsequent path segments are compared with the spatial occupancy coordinates of the obstacles in the obstacle spatial distribution dataset, so that the trajectory coordinates of the subsequent path segments are all outside the spatial occupancy coordinates of the obstacles. Repeat the above turning planning process, gradually extending the path segment towards the path endpoint area. Each extension is based on the obstacle distribution at the current path endpoint to determine the feasible heading at the current path endpoint and the path segment length at the current path endpoint. When the planned path segment extends to the boundary of the path endpoint area, adjust the heading angle of the last path segment to make it conform to the feasible termination heading range for entering the path endpoint area, so that the path can smoothly enter the path endpoint area. The planned path segments are connected sequentially to form a complete path trajectory from the starting point to the ending point of the path, which serves as the initial path structure. Repeat the above steps, and generate multiple different initial path structures based on different initial heading angles and turning angles. The trajectory coordinates of each initial path structure are outside the spatial occupancy range of all obstacles.

7. The intelligent perception and obstacle avoidance planning method for underwater obstacles according to claim 1, characterized in that, The method involves dynamically evolving and adjusting multiple initial obstacle avoidance paths based on the obstacle motion state associations in the obstacle dynamic association knowledge features, generating a target obstacle avoidance path adapted to the real-time underwater environment, and obtaining the target obstacle avoidance path generation result, including: Extract the obstacle motion state association relationship from the obstacle dynamic association knowledge features, obtain the motion direction, motion speed and motion trend prediction information of each obstacle, and generate obstacle dynamic prediction dataset; Extract the path trajectory information of multiple initial obstacle avoidance paths. The path trajectory information of each initial obstacle avoidance path includes all coordinate points on the initial obstacle avoidance path, the heading changes between adjacent coordinate points on the initial obstacle avoidance path, and the path length information of the initial obstacle avoidance path, and generate an initial path set. For each initial obstacle avoidance path, the overlap between the path trajectory of the initial obstacle avoidance path and the space occupied by the obstacle is predicted within a preset time period, based on the obstacle dynamic prediction dataset, and a path conflict prediction result is generated. For an initial obstacle avoidance path with path conflict prediction results, based on the path dynamic adjustment basis in the obstacle avoidance path evolution mechanism, the starting node and adjustment direction of the path adjustment of the initial obstacle avoidance path with path conflict prediction results are determined. The starting node of the path adjustment is selected as the conflict precursor position closest to the current position of the underwater vehicle. Based on the adjustment direction of the path adjustment and the obstacle movement trend prediction information, the path segment after the starting node of the path adjustment is replanned. During the replanning, the new path segment is ensured to be outside the predicted obstacle space occupation range. At the same time, the heading change nodes in the replanned path segment are smoothed to achieve a smooth heading transition between each path segment. The path segment after the starting node of the replanned path adjustment is smoothly connected to the path trajectory of the initial obstacle avoidance path before the starting node of the path adjustment, the trajectory information of the initial obstacle avoidance path is updated, and the adjusted obstacle avoidance path is generated. For an initial obstacle avoidance path without path conflict prediction results, based on the path optimization direction in the obstacle avoidance path evolution mechanism, the path trajectory of the initial obstacle avoidance path without path conflict prediction results is optimized in detail, the heading change node positions on the path of the initial obstacle avoidance path without path conflict prediction results are adjusted, and the total path length of the initial obstacle avoidance path without path conflict prediction results is shortened. Extract key parameters from the adjusted and optimized obstacle avoidance paths, including the total path length of the adjusted obstacle avoidance path, the number of heading changes of the adjusted obstacle avoidance path, the minimum distance between the adjusted obstacle avoidance path and the obstacle, the estimated navigation energy consumption of the adjusted obstacle avoidance path, the total path length of the optimized obstacle avoidance path, the number of heading changes of the optimized obstacle avoidance path, the minimum distance between the optimized obstacle avoidance path and the obstacle, and the estimated navigation energy consumption of the optimized obstacle avoidance path, to generate a path evaluation parameter set; Based on the path evaluation parameter set, all adjusted and optimized obstacle avoidance paths are ranked in multiple dimensions, including total path length, number of heading changes, minimum distance to obstacles, and estimated navigation energy consumption. Paths ranked higher are selected. From the top-ranked paths, determine the final target obstacle avoidance path, record the complete trajectory information and evaluation parameters of the target obstacle avoidance path, and obtain the target obstacle avoidance path generation result.

8. An intelligent obstacle avoidance planning system for underwater obstacles, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the intelligent perception and obstacle avoidance planning method for underwater obstacles as described in any one of claims 1 to 7 by executing the machine-executable instructions.

9. A computer program product, characterized in that, The computer program product includes machine-executable instructions stored in a computer-readable storage medium. The processor of the intelligent perception and obstacle avoidance planning system for underwater obstacles reads the machine-executable instructions from the computer-readable storage medium and executes the machine-executable instructions, causing the intelligent perception and obstacle avoidance planning system for underwater obstacles to perform the intelligent perception and obstacle avoidance planning method for underwater obstacles as described in any one of claims 1 to 7.

Images