Obstacle avoidance control method and system of underwater autonomous platform, electronic device and storage medium

Abstract

This application provides an obstacle avoidance control method, system, electronic device, and storage medium for an underwater autonomous platform. The obstacle avoidance control method includes: offline data processing of raw terrain data to obtain at least one terrain block data; based on the real-time positioning information of the underwater autonomous platform, acquiring target terrain block data for the underwater autonomous platform in the next motion cycle from the at least one terrain block data; determining terrain gradient obstacle data based on the target terrain block data; determining real-time sonar obstacle data based on the real-time sonar data of the underwater autonomous platform, and determining historical sonar obstacle data based on the historical sonar data of the underwater autonomous platform; determining multi-source obstacle data based on the terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data; determining the safe altitude of the underwater autonomous platform based on the multi-source obstacle data; and controlling the motion path of the underwater autonomous platform based on the safe altitude.

Description

Technical Field

[0001] This application relates to the technical field of obstacle avoidance control for underwater autonomous platforms, and more specifically, to an obstacle avoidance control method and system for underwater autonomous platforms, electronic devices, and storage media. Background Technology

[0002] With the development of marine resource exploration, scientific research, underwater engineering, and national defense, underwater autonomous platforms are playing an increasingly important role. As the operating environment extends to complex terrains (such as underwater canyons, reef areas, and areas near underwater infrastructure), higher demands are placed on the environmental perception and autonomous obstacle avoidance capabilities of underwater autonomous platforms.

[0003] However, the inventors of this application have discovered that current underwater autonomous platforms have at least the following technical problems in environmental perception and autonomous obstacle avoidance: 1. Strong reliance on offline methods: Current obstacle avoidance methods rely heavily on pre-programmed paths or prior maps, and are poorly adaptable to unknown and dynamic terrain.

[0004] 2. Slow data processing: The real-time reading, parsing and querying of high-precision, massive terrain data on embedded devices is inefficient and cannot meet the decision-making cycle requirements of high-speed underwater autonomous platforms.

[0005] 3. Single-source decision-making: Current obstacle avoidance algorithms are mostly based on information from a single sensor. In complex terrain or when sensors are limited, they lack a robust framework for multi-source fusion and intelligent decision-making.

[0006] 4. Rigid anti-bottoming strategy: The general approach is to use a fixed safety height, which wastes detection capabilities in flat areas and poses a risk of bottoming out in complex areas, lacking adaptive capabilities.

[0007] The content in the background section is merely technology known to the public and does not necessarily represent existing technology in this field. Summary of the Invention

[0008] This application provides an obstacle avoidance control method and system for an underwater autonomous platform, as well as electronic devices and storage media, aimed at solving at least one of the technical problems mentioned in the background art.

[0009] According to one aspect of this application, an obstacle avoidance control method for an underwater autonomous platform is provided, comprising: performing offline data processing on raw terrain data to obtain at least one terrain block data, and pre-storing the at least one terrain block data in a non-volatile storage device of the underwater autonomous platform; acquiring target terrain block data of the underwater autonomous platform in the next motion cycle from the at least one terrain block data based on the real-time positioning information of the underwater autonomous platform; determining terrain gradient obstacle data based on the target terrain block data; determining real-time sonar obstacle data based on real-time sonar data of the underwater autonomous platform, and determining historical sonar obstacle data based on historical sonar data of the underwater autonomous platform; determining multi-source obstacle data based on terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data; determining a safe altitude of the underwater autonomous platform based on the multi-source obstacle data; and controlling the motion path of the underwater autonomous platform based on the safe altitude.

[0010] According to another aspect of this application, an obstacle avoidance control system for an underwater autonomous platform is provided, including an offline data processing module, a real-time data processing module, and an obstacle avoidance control module. The offline data processing module processes raw terrain data offline to obtain at least one terrain block data, and pre-stores the at least one terrain block data in the non-volatile storage device of the underwater autonomous platform. The real-time data processing module, based on the real-time positioning information of the underwater autonomous platform, acquires target terrain block data for the underwater autonomous platform in the next movement cycle from the at least one terrain block data; determines terrain gradient obstacle data based on the target terrain block data; determines real-time sonar obstacle data based on the real-time sonar data of the underwater autonomous platform, and determines historical sonar obstacle data based on the historical sonar data of the underwater autonomous platform; and determines multi-source obstacle data based on the terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data. The obstacle avoidance control module determines the safe altitude of the underwater autonomous platform based on the multi-source obstacle data; and controls the movement path of the underwater autonomous platform based on the safe altitude.

[0011] Beneficial effects This application can determine multi-source obstacle data by deeply fusing offline high-precision raw terrain data with online sonar perception based on the "prior + real-time" fusion perception method. This can realize the "digital twin" of underwater navigation, enabling the underwater autonomous platform to have a forward-looking environmental understanding capability, which can significantly reduce the risk of bottoming out and collision, and is especially suitable for exploration tasks in complex and unknown terrain. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating an embodiment of the obstacle avoidance control method of this application is shown. Figure 2 This diagram illustrates yet another flow chart of the obstacle avoidance control method according to an embodiment of this application; Figure 3 This diagram illustrates yet another flow chart of the obstacle avoidance control method according to an embodiment of this application; Figure 4 This diagram illustrates yet another flow chart of the obstacle avoidance control method according to an embodiment of this application; Figure 5 A schematic diagram of the obstacle avoidance control system according to an embodiment of this application is shown.

[0014] Explanation of reference numerals in the attached figures: Offline data processing module 10; real-time data processing module 20; obstacle avoidance control module 30. Detailed Implementation

[0015] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that this application will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0016] The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of these specific details, or other methods, components, materials, devices, etc. In these cases, well-known structures, methods, devices, implementations, materials, or operations will not be shown or described in detail.

[0017] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0018] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, rather than to describe a specific order.

[0019] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] Currently, obstacle avoidance methods for underwater autonomous platforms can be categorized as follows: 1. Rule-based obstacle avoidance system based on a single forward-looking sonar.

[0021] Such systems are typically equipped with a forward-looking sonar, which detects the nearest obstacle in the forward sector and, in conjunction with a preset safe distance threshold, generates obstacle avoidance commands using traditional rule-based algorithms such as the artificial potential field method and the vector field histogram method.

[0022] However, the inventors discovered that such rule-based obstacle avoidance systems have at least the following drawbacks: First, their perception dimension is limited, providing only two-dimensional profile information from the front and failing to construct a three-dimensional terrain model of the area beneath and to the sides of the platform, resulting in blind spots and making the platform prone to collisions with the seabed or side protrusions. Second, their decision-making mechanism is rigid; algorithms based on fixed rules are highly susceptible to getting stuck in local optima (such as oscillations or deadlocks) when encountering complex obstacle groups, lacking global foresight. Finally, they lack terrain understanding capabilities, failing to distinguish between different terrain features such as rocks, sand, or seaweed, often leading to unnecessary and ineffective obstacle avoidance actions.

[0023] 2. Offline terrain matching navigation based on multibeam sonar.

[0024] Multibeam sonar can acquire high-precision seabed topographic data, and by matching it with pre-stored digital topographic maps, it can achieve precise positioning and global path tracking.

[0025] However, the inventors discovered that such rule-based obstacle avoidance systems have at least the following drawbacks: they primarily serve surveying and long-range navigation, resulting in poor real-time performance. Multibeam bathymetry and terrain matching algorithms are computationally intensive, typically taking seconds or even minutes, which is completely insufficient to meet the millisecond-level response requirements of underwater autonomous platforms for real-time obstacle avoidance in high-speed navigation or complex terrain. Furthermore, they heavily rely on prior maps, becoming completely ineffective in completely unknown areas or areas with dynamically changing terrain (such as landslides or sand dune migration). Additionally, their functionality is separated from the obstacle avoidance decision-making system, lacking tightly coupled real-time closed-loop control.

[0026] 3. Perception system based on traditional 3D point cloud processing.

[0027] To overcome perception blind spots, perception systems based on traditional 3D point cloud processing attempt to fuse multi-source data such as multibeam sonar and forward-looking sonar to generate 3D point clouds, and identify obstacles through point cloud segmentation, clustering and other algorithms.

[0028] However, the inventors discovered that such rule-based obstacle avoidance systems have at least the following drawbacks: the inherent characteristics of underwater acoustic imaging (such as low signal-to-noise ratio, uneven data density, and severe multipath effects) lead to poor quality of 3D point cloud data. Real-time, efficient, and accurate processing of these massive amounts of noisy, unstructured point cloud data poses a significant challenge to the limited onboard computing resources of underwater autonomous platforms, making stable operation on low-power embedded platforms difficult. Furthermore, its parameters are highly dependent on specific environmental conditions (such as water turbidity and substrate type), and its performance deteriorates sharply in complex hydrological environments such as turbidity and strong scattering, resulting in severely insufficient environmental robustness.

[0029] 4. Introduce artificial intelligence perception methods.

[0030] Existing technologies attempt to use convolutional neural networks to perform target detection and recognition on forward-looking sonar images in order to improve perceptual intelligence.

[0031] However, the inventors discovered that the shortcomings of such rule-based obstacle avoidance systems include at least the following: most work is limited to single improvements in the perception stage (such as simply increasing the recognition rate of a certain type of target), failing to integrate the perception results of artificial intelligence with dynamic path planning and motion control in an end-to-end systematic manner. At the decision-making level, it still fails to fully utilize methods such as deep reinforcement learning to allow the platform to autonomously learn optimal obstacle avoidance strategies under complex and dynamic terrain in a simulation environment; its decision intelligence level and generalization ability remain insufficient. Furthermore, because the perception, planning, and control modules are usually loosely coupled by different protocols or middleware, the overall system latency is high, communication overhead is large, and it is difficult to guarantee the overall real-time performance and reliability of the system.

[0032] According to one aspect of this application, this application provides an obstacle avoidance control method for an underwater autonomous platform.

[0033] According to the example embodiment, the underwater autonomous platform can be any motion vehicle or operating system that can autonomously complete specific tasks in an underwater environment without real-time human control.

[0034] For example, underwater autonomous platforms include, but are not limited to, autonomous underwater vehicles, underwater robots, autonomous underwater gliders, and autonomous underwater unmanned systems, etc., and this application does not limit them.

[0035] like Figure 1As shown, the obstacle avoidance control method of the underwater autonomous platform may include steps S100-S700.

[0036] For example, this obstacle avoidance control method can be executed by the obstacle avoidance control system of an underwater autonomous platform. This obstacle avoidance control system can be located on the underwater autonomous platform or on an external server; this application does not impose any restrictions on this.

[0037] In S100, the obstacle avoidance control system performs offline data processing on the raw terrain data to obtain at least one terrain block data, and pre-stores the at least one terrain block data in the non-volatile storage device of the underwater autonomous platform.

[0038] For example, the obstacle avoidance control system can load raw terrain data (such as global or regional ocean data) covering the target operational sea area through a ground control station or external data interface. This raw terrain data can be in NetCDF (Network Common Data Form) format, which may include longitude grid range, latitude grid range, resolution, and elevation data matrix. The obstacle avoidance control system can then perform offline data processing on this raw terrain data to obtain at least one terrain block.

[0039] For example, terrain patch data can be sub-region terrain data obtained by spatially dividing the original large-scale, high-precision, and global terrain data according to certain rules.

[0040] The obstacle avoidance control system encapsulates the data of at least one terrain block obtained from the processing and pre-stores it in a custom format to the non-volatile storage device of the underwater autonomous platform.

[0041] For example, non-volatile storage devices include, but are not limited to, eMMC chips, SD cards, and solid-state drives.

[0042] Optionally, such as Figure 2 As shown, step S100 may also include S110-S120.

[0043] In S110, the obstacle avoidance control system performs data parsing and reconstruction, data slicing and spatial index construction on the original terrain data to obtain segmented terrain processing data.

[0044] In the S120, the obstacle avoidance control system encapsulates the segmented terrain processing data in a custom format and pre-stores it in the non-volatile storage device of the underwater autonomous platform.

[0045] For example, the obstacle avoidance control system performs data parsing and reconstruction on the original terrain data, decrypts the original terrain data, and extracts metadata such as latitude and longitude boundary range, spatial resolution, and elevation data matrix.

[0046] To avoid inefficient retrieval of global terrain data during real-time navigation, the obstacle avoidance control system can slice the elevation data matrix. For example, the obstacle avoidance control system can slice the data according to a preset block size (e.g., The latitude and longitude grid divides the global terrain data into multiple non-overlapping terrain blocks.

[0047] To support rapid location of corresponding terrain patch data based on arbitrary latitude and longitude coordinates, the obstacle avoidance control system can construct a spatial index for all terrain patch data. For example, this spatial index can use an R-tree (a tree-like data structure) or a variant thereof. By mapping the spatial bounding box of each terrain patch data and its storage location to index entries in the R-tree, rapid spatial queries of the corresponding terrain patch data can be achieved.

[0048] To facilitate efficient access to the non-volatile storage devices of the underwater autonomous platform, the obstacle avoidance control system encapsulates the obtained segmented terrain processing data into custom-formatted files (such as...). Format, or rename For example, this custom format logically includes at least a file header, an index area, and a data area. The file header stores global metadata (such as coverage, chunk size, and projection method). The index area stores serialized spatial index data. The data area stores the elevation data matrix of each terrain block after encryption and compression. The obstacle avoidance control system then stores the encapsulated custom-formatted data file in the non-volatile storage device of the underwater autonomous platform.

[0049] Through the above embodiments, this application can convert the original terrain data into a local data format with block, spatial index and compressed storage by offline data preprocessing, which can lay the data foundation for underwater autonomous platforms to achieve low-latency and low-power terrain query and perception in real-time navigation.

[0050] This application divides the original terrain data into multiple terrain blocks and constructs a spatial index, enabling the underwater autonomous platform to quickly locate and load the target terrain block data based on its current position during navigation. This avoids traversing and searching through massive amounts of global data, reduces query time complexity, and thus meets the real-time requirements of the underwater autonomous platform for terrain perception during high-speed navigation.

[0051] This application reduces the storage space occupied by the data by using a custom format to store the index area and the data area separately, and by performing lossless compression on the terrain block data. Furthermore, during queries, only the index area and the necessary terrain block data need to be loaded, avoiding the reading of irrelevant data and reducing the number of I / O (data read / write) operations, thereby reducing the bandwidth pressure on the embedded storage medium.

[0052] In S200, the obstacle avoidance control system acquires the target terrain block data of the underwater autonomous platform in the next motion cycle from at least one terrain block data, based on the real-time positioning information of the underwater autonomous platform.

[0053] For example, when an underwater autonomous platform performs tasks in an underwater environment, its obstacle avoidance control system receives real-time positioning information from the integrated navigation system.

[0054] For example, the integrated navigation system includes one or more of GPS (Global Positioning System), INS (Inertial Navigation System), and DVL (Doppler Velocity Log).

[0055] The obstacle avoidance control system obtains the current position information (including longitude, latitude, and heading) of the underwater autonomous platform based on real-time positioning information, and calculates the predicted position that the underwater autonomous platform may reach in the next motion cycle based on the current motion state of the underwater autonomous platform (such as speed and heading angle).

[0056] For example, the obstacle avoidance control system uses the predicted location as a query condition and quickly retrieves the target terrain block data corresponding to the predicted location from multiple stored terrain block data through a pre-built spatial index, so as to provide a data foundation for subsequent elevation interpolation, obstacle avoidance decision-making and intelligent pre-reading.

[0057] For example, the next motion cycle can be the range of travel between the current position and the future position reached by the underwater autonomous platform after traveling a predetermined spatial distance (e.g., 10 kilometers) along the planned path.

[0058] Optionally, such as Figure 3 As shown, S200 may also include S210-S230.

[0059] In S210, the obstacle avoidance control system determines whether the target terrain block data is stored in the memory cache of the underwater autonomous platform.

[0060] In S220, if so, the obstacle avoidance control system retrieves the target terrain block data from the memory cache.

[0061] In S230, if not, the obstacle avoidance control system obtains the target terrain block data from the non-volatile storage device of the underwater autonomous platform.

[0062] For example, the obstacle avoidance control system maintains a cache queue of recently accessed terrain tile data in its memory cache (e.g., with a capacity of 10 terrain tile data). If the identified target terrain tile data is cached in this memory cache, the obstacle avoidance control system can directly access that target terrain tile data, significantly reducing I / O latency. If the identified target terrain tile data is not cached in this memory cache, the obstacle avoidance control system can quickly locate the target terrain tile data in the non-volatile storage device of the underwater autonomous platform using a preset spatial index, load it into the memory cache, and update the cache queue.

[0063] Through the above embodiments, this application can preferentially read target terrain block data from the memory cache of the underwater autonomous platform. The read speed of the memory cache (e.g., nanosecond to microsecond level) is significantly faster than that of non-volatile storage devices (e.g., eMMC chips or SD cards, which are in the millisecond range). Therefore, when the target terrain block data is already stored in the memory cache, the data can be obtained directly at a faster processing speed, avoiding low-speed I / O operations and thus meeting the real-time requirements of the underwater autonomous platform for obstacle avoidance decisions. Furthermore, this application can absorb most data query requests through the memory cache, reducing the access frequency to the underlying storage device, extending hardware lifespan, and reducing system power consumption.

[0064] In the S300, the obstacle avoidance control system determines the terrain gradient obstacle data based on the target terrain block data.

[0065] For example, the obstacle avoidance control system can determine the terrain gradient obstacle data for the next movement cycle based on the identified target terrain block data. Based on this target terrain block data, the system calculates the terrain slope of each local area in the next movement cycle and marks local areas with terrain slopes greater than a preset safety threshold as obstacle areas, thus obtaining terrain gradient obstacle data based on all identified obstacle areas.

[0066] For example, the preset safety threshold can be customized according to the user's actual needs, and this application does not limit it. As one embodiment, the preset safety threshold can be 30°.

[0067] In the S400, the obstacle avoidance control system determines real-time sonar obstacle data based on the real-time sonar data of the underwater autonomous platform, and determines historical sonar obstacle data based on the historical sonar obstacle data of the underwater autonomous platform.

[0068] For example, real-time sonar data can be acquired in real time by forward-looking sonar / multibeam sonar. Based on the real-time sonar data, the obstacle avoidance control system determines the obstacle area in each local region for the next movement cycle, thus obtaining real-time sonar obstacle data based on all determined obstacle areas.

[0069] Historical sonar data can be sonar data from forward-looking sonar / multibeam sonar records, including obstacle areas historically detected by forward-looking sonar / multibeam sonar. The obstacle avoidance control system can also determine the historical obstacle areas for each local region in the next movement cycle based on the historical sonar data, thus obtaining historical sonar obstacle data based on all determined historical obstacle areas.

[0070] In the S500, the obstacle avoidance control system determines multi-source obstacle data based on terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data.

[0071] For example, the obstacle avoidance control system can fuse terrain gradient obstacle data, real-time sonar obstacle data and historical sonar obstacle data according to the weight allocation to obtain the final unified obstacle probability map, which is multi-source obstacle data.

[0072] It can be understood here that, in the absence of historical sonar data, the weight of historical sonar obstacle data is 0.

[0073] For example, multi-source obstacle data can be characterized as: ; in, These are the coordinates of the obstacles corresponding to the multi-source obstacle data. These are the obstacle coordinates corresponding to the terrain gradient obstacle data. These are the coordinates of the obstacle corresponding to real-time sonar obstacle data. These are the coordinates of the obstacles corresponding to historical sonar obstacle data. The first weight corresponding to the terrain gradient obstacle data. The second weight corresponding to real-time sonar obstacle data. The third weight corresponds to historical sonar obstacle data. .

[0074] Optionally, the weights corresponding to terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data can be dynamically adjusted according to preset rules.

[0075] For example, in the first environmental condition (such as clear water and high-quality sonar data), the second weight corresponding to the real-time sonar obstacle data is increased. In the second environmental condition (such as complex terrain and the possibility of multiple sonar reflections), the first weight corresponding to the terrain gradient obstacle data is increased.

[0076] For example, if a sensor temporarily fails, its weight is reduced accordingly so that the obstacle avoidance control system can still function normally by relying on other information sources.

[0077] This can be understood as follows: terrain data can detect potential risks such as steep slopes in advance, while sonar data can dynamically detect sudden obstacles. The complementarity of these two technologies can effectively eliminate the blind spots of a single sensor. Furthermore, when real-time sonar data becomes unreliable due to environmental factors such as water turbidity, the obstacle avoidance control system can still rely on pre-stored terrain data and historical sonar obstacle data to make decisions. This multi-source redundancy design can improve robustness in complex waters. Historical sonar obstacle data allows the underwater autonomous platform to "remember" obstacle areas that have been detected, avoiding the danger of misjudging safety due to temporary sonar obstruction. This improves the obstacle avoidance reliability and navigation safety of the underwater autonomous platform in unknown and complex environments.

[0078] In the S600, the obstacle avoidance control system determines the safe altitude of the underwater autonomous platform based on multi-source obstacle data.

[0079] For example, an obstacle avoidance control system can determine the current location and the terrain ahead based on multi-source obstacle data, and calculate the terrain roughness index (in meters). The obstacle avoidance control system can then determine the safe height based on the terrain roughness index and the basic safe height.

[0080] For example, a security height can be characterized as: ; in, For safety reasons, Based on the safety height, This is the terrain roughness index.

[0081] In the S700, the obstacle avoidance control system controls the movement path of the underwater autonomous platform based on the safe altitude.

[0082] For example, when an obstacle avoidance control system determines a safe height based on multi-source obstacle data, it uses optimization algorithms to plan or adjust a safe path in real time.

[0083] For example, the optimization algorithm includes, but is not limited to, the A algorithm, the D Lite algorithm, or the Model Predictive Control (MPC) algorithm, etc., and this application does not limit it.

[0084] Optionally, such as Figure 4 As shown, S700 may also include S710-S750.

[0085] In S710, the obstacle avoidance control system determines the elevation value of the target terrain block data.

[0086] For example, the elevation value can be the height of the sea surface above the seabed in the target terrain block data.

[0087] For example, the obstacle avoidance control system can calculate the accurate elevation value based on the bilinear interpolation method.

[0088] In the S720, the obstacle avoidance control system determines the predicted height of the underwater autonomous platform above the bottom based on the elevation value and the real-time positioning information of the underwater autonomous platform.

[0089] For example, the obstacle avoidance control system can obtain the predicted height of the underwater autonomous platform above the bottom based on the difference between the elevation value and the position of the underwater autonomous platform. .

[0090] In the S730, the obstacle avoidance control system controls the underwater autonomous platform to execute a first motion strategy when the predicted height above the bottom and the safe height meet the first preset conditions.

[0091] For example, the first preset condition is the predicted height from the bottom. Less than the first preset value (e.g., the first preset value is 1.2 times the safe height, i.e.: The first motion strategy is to control the underwater autonomous platform to decelerate.

[0092] In the S740, the obstacle avoidance control system controls the underwater autonomous platform to execute a second motion strategy when the predicted height above the bottom and the safe height meet the second preset conditions.

[0093] For example, the second preset condition is that the predicted height from the bottom is less than the second preset value (if the second preset value is the safe height, i.e.: The second motion strategy includes at least: cutting off the main propulsion power of the underwater autonomous platform and stopping its forward movement; and controlling the rudder surfaces and thrusters of the underwater autonomous platform to rise urgently at a preset maximum safe elevation angle until the preset height above the bottom is greater than a third preset value (e.g., the third preset value is the safe height / 1.5, i.e., ...). It also controls the underwater autonomous platform to enter a hovering state.

[0094] In the S750, the obstacle avoidance control system re-controls the movement path of the underwater autonomous platform based on updated multi-source obstacle data.

[0095] For example, after the obstacle avoidance control system puts the underwater autonomous platform into a hovering state, it can redetermine and update multi-source obstacle data based on the latest environmental information, thereby replanning the detour route.

[0096] Through the above embodiments, this application dynamically compares the real-time predicted seabed clearance with the adaptive safety height, triggering different motion strategies based on different comparison results. Based on the elevation values ​​of target terrain data and real-time positioning information, this application can accurately calculate the predicted seabed clearance, enabling it to detect the risk of bottom contact before the underwater autonomous platform actually approaches the seabed, thus achieving a shift from passive response to proactive avoidance.

[0097] This application distinguishes different levels of danger by setting first and second preset conditions, and executes corresponding movement strategies (such as deceleration, turning, or emergency surfacing) respectively. This can achieve risk classification and handling, which can avoid oversensitivity leading to frequent ineffective avoidance, and ensure decisive intervention when there is real danger.

[0098] When the environment continues to change, this application can replan the path based on the updated multi-source obstacle data, forming a closed-loop decision-making mechanism of "perception-judgment-execution-re-perception", thereby comprehensively improving the navigation safety and environmental adaptability of the underwater autonomous platform in complex terrain.

[0099] This application can determine multi-source obstacle data by deeply fusing offline high-precision raw terrain data with online sonar perception based on the "prior + real-time" fusion perception method. This can realize the "digital twin" of underwater navigation, enabling the underwater autonomous platform to have a forward-looking environmental understanding capability, which can significantly reduce the risk of bottoming out and collision, and is especially suitable for exploration tasks in complex and unknown terrain.

[0100] According to another aspect of this application, this application provides an obstacle avoidance control system for an underwater autonomous platform.

[0101] like Figure 5 As shown, the obstacle avoidance control system of the underwater autonomous platform may include an offline data processing module 10, a real-time data processing module 20, and an obstacle avoidance control module 30.

[0102] The offline data processing module 10 performs offline data processing on the raw terrain data to obtain at least one terrain block data, and pre-stores the at least one terrain block data in the non-volatile storage device of the underwater autonomous platform.

[0103] For example, the offline data processing module 10 can load raw topographic data (such as global or regional oceanographic data) covering the target operational sea area through a ground control station or external data interface. This raw topographic data can be in NetCDF (Network Common Data Form) format, which may include longitude grid range, latitude grid range, resolution, and elevation data matrix. The offline data processing module 10 performs offline data processing on this raw topographic data to obtain at least one topographic block data.

[0104] For example, terrain patch data can be sub-region terrain data obtained by spatially dividing the original large-scale, high-precision, and global terrain data according to certain rules.

[0105] The offline data processing module 10 encapsulates at least one terrain block data obtained from the processing and pre-stores it in a custom format to the non-volatile storage device of the underwater autonomous platform.

[0106] For example, non-volatile storage devices include, but are not limited to, eMMC chips, SD cards, and solid-state drives.

[0107] Optionally, the offline data processing module 10 performs data parsing and reconstruction, data slicing and spatial index construction on the original terrain data to obtain segmented terrain processing data.

[0108] The offline data processing module 10 encapsulates the segmented terrain processing data in a custom format and pre-stores it in the non-volatile storage device of the underwater autonomous platform.

[0109] For example, the offline data processing module 10 performs data parsing and reconstruction on the original terrain data, decrypts the original terrain data, and extracts metadata such as latitude and longitude boundary range, spatial resolution, and elevation data matrix.

[0110] To avoid inefficient retrieval of global terrain data during real-time navigation, the offline data processing module 10 can slice the elevation data matrix. For example, the offline data processing module 10 can slice the data according to a preset block size (e.g., ...). The latitude and longitude grid divides the global terrain data into multiple non-overlapping terrain blocks.

[0111] To support the rapid location of corresponding terrain block data based on arbitrary latitude and longitude coordinates, the offline data processing module 10 can construct a spatial index for all terrain block data. For example, this spatial index can adopt an R-tree (a tree-like data structure) or a variant thereof. By mapping the spatial bounding box of each terrain block data and its storage location to index entries of the R-tree, fast spatial queries of the corresponding terrain block data can be achieved.

[0112] To facilitate efficient access to the non-volatile storage devices of the underwater autonomous platform, the offline data processing module 10 encapsulates the obtained segmented terrain processing data into a custom-formatted file (such as...). Format, or rename For example, this custom format logically includes at least a file header, an index area, and a data area. The file header stores global metadata (such as coverage, chunk size, and projection method). The index area stores serialized spatial index data. The data area stores the elevation data matrix of each terrain block after encryption and compression. Afterward, the offline data processing module 10 stores the encapsulated custom-formatted data file in the non-volatile storage device of the underwater autonomous platform.

[0113] Through the above embodiments, this application can convert the original terrain data into a local data format with block, spatial index and compressed storage by offline data preprocessing, which can lay the data foundation for underwater autonomous platforms to achieve low-latency and low-power terrain query and perception in real-time navigation.

[0114] This application divides the original terrain data into multiple terrain blocks and constructs a spatial index, enabling the underwater autonomous platform to quickly locate and load the target terrain block data based on its current position during navigation. This avoids traversing and searching through massive amounts of global data, reduces query time complexity, and thus meets the real-time requirements of the underwater autonomous platform for terrain perception during high-speed navigation.

[0115] This application reduces the storage space occupied by the data by using a custom format to store the index area and the data area separately, and by performing lossless compression on the terrain block data. Furthermore, during queries, only the index area and the necessary terrain block data need to be loaded, avoiding the reading of irrelevant data and reducing the number of I / O (data read / write) operations, thereby reducing the bandwidth pressure on the embedded storage medium.

[0116] Based on the real-time positioning information of the underwater autonomous platform, the real-time data processing module 20 acquires the target terrain block data of the underwater autonomous platform in the next motion cycle from at least one terrain block data.

[0117] For example, during the underwater autonomous platform's mission in the underwater environment, the real-time data processing module 20 receives real-time positioning information from the integrated navigation system.

[0118] For example, the integrated navigation system includes one or more of GPS (Global Positioning System), INS (Inertial Navigation System), and DVL (Doppler Velocity Log).

[0119] The real-time data processing module 20 obtains the current location information (including longitude, latitude and heading) of the underwater autonomous platform based on the real-time positioning information, and calculates the predicted location that the underwater autonomous platform may reach in the next motion cycle based on the current motion state of the underwater autonomous platform (such as speed and heading angle).

[0120] For example, the real-time data processing module 20 uses the predicted location as a query condition and quickly retrieves the target terrain block data corresponding to the predicted location from multiple stored terrain block data through a pre-built spatial index, so as to provide a data foundation for subsequent elevation interpolation, obstacle avoidance decision-making and intelligent pre-reading.

[0121] For example, the next motion cycle can be the range of travel between the current position and the future position reached by the underwater autonomous platform after traveling a predetermined spatial distance (e.g., 10 kilometers) along the planned path.

[0122] Optionally, the real-time data processing module 20 determines whether the target terrain block data is stored in the memory cache of the underwater autonomous platform.

[0123] If so, the real-time data processing module 20 retrieves the target terrain block data from the memory cache.

[0124] If not, the real-time data processing module 20 obtains the target terrain block data from the non-volatile storage device of the underwater autonomous platform.

[0125] For example, the obstacle avoidance control system maintains a cache queue of recently accessed terrain tile data in its memory cache (e.g., with a capacity of 10 terrain tile data). If the identified target terrain tile data is cached in this memory cache, the real-time data processing module 20 can directly access the target terrain tile data, greatly reducing I / O latency. If the identified target terrain tile data is not cached in this memory cache, the real-time data processing module 20 can quickly locate the target terrain tile data in the non-volatile storage device of the underwater autonomous platform using a preset spatial index, load it into the memory cache, and update the cache queue.

[0126] Through the above embodiments, this application can preferentially read target terrain block data from the memory cache of the underwater autonomous platform. The read speed of the memory cache (e.g., nanosecond to microsecond level) is significantly faster than that of non-volatile storage devices (e.g., eMMC chips or SD cards, which are in the millisecond range). Therefore, when the target terrain block data is already stored in the memory cache, the data can be obtained directly at a faster processing speed, avoiding low-speed I / O operations and thus meeting the real-time requirements of the underwater autonomous platform for obstacle avoidance decisions. Furthermore, this application can absorb most data query requests through the memory cache, reducing the access frequency to the underlying storage device, extending hardware lifespan, and reducing system power consumption.

[0127] The real-time data processing module 20 determines the terrain gradient obstacle data based on the target terrain block data.

[0128] For example, the real-time data processing module 20 can determine the terrain gradient obstacle data for the next movement cycle based on the determined target terrain block data. The real-time data processing module 20 calculates the terrain slope of each local area in the next movement cycle based on the target terrain block data, and marks local areas with terrain slopes greater than a preset safety threshold as obstacle areas, thereby obtaining terrain gradient obstacle data based on all determined obstacle areas.

[0129] For example, the preset safety threshold can be customized according to the user's actual needs, and this application does not limit it. As one embodiment, the preset safety threshold can be 30°.

[0130] The real-time data processing module 20 determines real-time sonar obstacle data based on the real-time sonar data of the underwater autonomous platform, and determines historical sonar obstacle data based on the historical sonar obstacle data of the underwater autonomous platform.

[0131] For example, real-time sonar data can be acquired in real time by forward-looking sonar / multibeam sonar. Based on the real-time sonar data, the real-time data processing module 20 determines the obstacle area of ​​each local region in the next motion cycle in real time, so as to obtain real-time sonar obstacle data based on all determined obstacle areas.

[0132] Historical sonar data can be sonar data from forward-looking sonar / multibeam sonar history records, including obstacle areas historically detected by forward-looking sonar / multibeam sonar. The real-time data processing module 20 can also determine the historical obstacle areas for each local region in the next motion cycle based on the historical sonar data, thus obtaining historical sonar obstacle data based on all determined historical obstacle areas.

[0133] The real-time data processing module 20 determines multi-source obstacle data based on terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data.

[0134] For example, the real-time data processing module 20 can fuse terrain gradient obstacle data, real-time sonar obstacle data and historical sonar obstacle data according to the weight allocation to obtain the final unified obstacle probability map, that is, multi-source obstacle data.

[0135] It can be understood here that, in the absence of historical sonar data, the weight of historical sonar obstacle data is 0.

[0136] For example, multi-source obstacle data can be characterized as: ; in, These are the coordinates of the obstacles corresponding to the multi-source obstacle data. These are the obstacle coordinates corresponding to the terrain gradient obstacle data. These are the coordinates of the obstacle corresponding to real-time sonar obstacle data. These are the coordinates of the obstacles corresponding to historical sonar obstacle data. The first weight corresponding to the terrain gradient obstacle data. The second weight corresponding to real-time sonar obstacle data. The third weight corresponds to historical sonar obstacle data. .

[0137] Optionally, the weights corresponding to terrain gradient obstacle data, real-time sonar obstacle data, and historical sonar obstacle data can be dynamically adjusted according to preset rules.

[0138] For example, in the first environmental condition (such as clear water and high-quality sonar data), the second weight corresponding to the real-time sonar obstacle data is increased. In the second environmental condition (such as complex terrain and the possibility of multiple sonar reflections), the first weight corresponding to the terrain gradient obstacle data is increased.

[0139] For example, if a sensor temporarily fails, its weight is reduced accordingly so that the obstacle avoidance control system can still function normally by relying on other information sources.

[0140] This can be understood as follows: terrain data can detect potential risks such as steep slopes in advance, while sonar data can dynamically detect sudden obstacles. The complementarity of these two technologies can effectively eliminate the blind spots of a single sensor. Furthermore, when real-time sonar data becomes unreliable due to environmental factors such as water turbidity, the obstacle avoidance control system can still rely on pre-stored terrain data and historical sonar obstacle data to make decisions. This multi-source redundancy design can improve robustness in complex waters. Historical sonar obstacle data allows the underwater autonomous platform to "remember" obstacle areas that have been detected, avoiding the danger of misjudging safety due to temporary sonar obstruction. This improves the obstacle avoidance reliability and navigation safety of the underwater autonomous platform in unknown and complex environments.

[0141] The obstacle avoidance control module 30 determines the safe altitude of the underwater autonomous platform based on multi-source obstacle data.

[0142] For example, the obstacle avoidance control module 30 can determine the current location and the terrain ahead based on multi-source obstacle data, and calculate the terrain roughness index (in meters). The obstacle avoidance control system can determine the safe height based on the terrain roughness index and the basic safe height.

[0143] For example, a security height can be characterized as: ; in, For safety reasons, Based on the safety height, This is the terrain roughness index.

[0144] The obstacle avoidance control module 30 controls the movement path of the underwater autonomous platform according to the safe altitude.

[0145] For example, when the obstacle avoidance control module 30 determines the safe height based on multi-source obstacle data, it uses an optimization algorithm to plan or adjust the safe path in real time.

[0146] For example, the optimization algorithm includes, but is not limited to, the A algorithm, the D Lite algorithm, or the Model Predictive Control (MPC) algorithm, etc., and this application does not limit it.

[0147] Optionally, the obstacle avoidance control module 30 determines the elevation value of the target terrain block data.

[0148] For example, the elevation value can be the height of the sea surface above the seabed in the target terrain block data.

[0149] For example, the obstacle avoidance control module 30 can calculate the accurate elevation value based on bilinear interpolation.

[0150] The obstacle avoidance control module 30 determines the predicted height of the underwater autonomous platform above the bottom based on the elevation value and the real-time positioning information of the underwater autonomous platform.

[0151] For example, the obstacle avoidance control module 30 can obtain the predicted height of the underwater autonomous platform above the bottom based on the difference between the elevation value and the position of the underwater autonomous platform. .

[0152] When the predicted height from the bottom and the safe height meet the first preset conditions, the obstacle avoidance control module 30 controls the underwater autonomous platform to execute the first motion strategy.

[0153] For example, the first preset condition is the predicted height from the bottom. Less than the first preset value (e.g., the first preset value is 1.2 times the safe height, i.e.: The first motion strategy is to control the underwater autonomous platform to decelerate.

[0154] When the predicted height from the bottom and the safe height meet the second preset conditions, the obstacle avoidance control module 30 controls the underwater autonomous platform to execute the second motion strategy.

[0155] For example, the second preset condition is that the predicted height from the bottom is less than the second preset value (if the second preset value is the safe height, i.e.: The second motion strategy includes at least: cutting off the main propulsion power of the underwater autonomous platform and stopping its forward movement; and controlling the rudder surfaces and thrusters of the underwater autonomous platform to rise urgently at a preset maximum safe elevation angle until the preset height above the bottom is greater than a third preset value (e.g., the third preset value is the safe height / 1.5, i.e., ...). It also controls the underwater autonomous platform to enter a hovering state.

[0156] The obstacle avoidance control module 30 re-controls the movement path of the underwater autonomous platform based on the updated multi-source obstacle data.

[0157] For example, after the obstacle avoidance control module 30 controls the underwater autonomous platform to enter a hovering state, it can redetermine and update the multi-source obstacle data based on the latest environmental information, thereby replanning the detour path.

[0158] Through the above embodiments, this application dynamically compares the real-time predicted seabed clearance with the adaptive safety height, triggering different motion strategies based on different comparison results. Based on the elevation values ​​of target terrain data and real-time positioning information, this application can accurately calculate the predicted seabed clearance, enabling it to detect the risk of bottom contact before the underwater autonomous platform actually approaches the seabed, thus achieving a shift from passive response to proactive avoidance.

[0159] This application distinguishes different levels of danger by setting first and second preset conditions, and executes corresponding movement strategies (such as deceleration, turning, or emergency surfacing) respectively. This can achieve risk classification and handling, which can avoid oversensitivity leading to frequent ineffective avoidance, and ensure decisive intervention when there is real danger.

[0160] When the environment continues to change, this application can replan the path based on the updated multi-source obstacle data, forming a closed-loop decision-making mechanism of "perception-judgment-execution-re-perception", thereby comprehensively improving the navigation safety and environmental adaptability of the underwater autonomous platform in complex terrain.

[0161] This application can determine multi-source obstacle data by deeply fusing offline high-precision raw terrain data with online sonar perception based on the "prior + real-time" fusion perception method. This can realize the "digital twin" of underwater navigation, enabling the underwater autonomous platform to have a forward-looking environmental understanding capability, which can significantly reduce the risk of bottoming out and collision, and is especially suitable for exploration tasks in complex and unknown terrain.

[0162] According to another aspect of this application, an electronic device is also provided. The electronic device includes: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, enable the one or more processors to perform the methods described above.

[0163] According to another aspect of this application, a non-volatile computer-readable storage medium is also provided. This storage medium stores a computer program that, when executed by a processor, can perform the methods described above.

[0164] According to another aspect of this application, this application also provides a computer program product. The computer program product includes: a computer program stored on a computer-readable storage medium; the computer program includes program instructions that, when executed by a computer, cause the computer to perform the methods described above.

[0165] Finally, it should be noted that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions of the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for obstacle avoidance control of an underwater autonomous platform, characterized in that, include: The raw terrain data is processed offline to obtain at least one terrain block data, and the at least one terrain block data is pre-stored in the non-volatile storage device of the underwater autonomous platform; Based on the real-time positioning information of the underwater autonomous platform, the target terrain block data of the underwater autonomous platform in the next motion cycle is obtained from the at least one terrain block data. Based on the target terrain block data, terrain gradient obstacle data is determined; Real-time sonar obstacle data is determined based on the real-time sonar data of the underwater autonomous platform, and historical sonar obstacle data is determined based on the historical sonar data of the underwater autonomous platform. Multi-source obstacle data is determined based on the terrain gradient obstacle data, the real-time sonar obstacle data, and the historical sonar obstacle data; The safe altitude of the underwater autonomous platform is determined based on the multi-source obstacle data. The movement path of the underwater autonomous platform is controlled according to the safe altitude.

2. The obstacle avoidance control method according to claim 1, characterized in that, The offline data processing of the raw terrain data to obtain at least one terrain block data, and the pre-storage of the at least one terrain block data in the non-volatile storage device of the underwater autonomous platform, includes: The original terrain data is parsed and reconstructed, slicing data and constructing spatial indexes to obtain segmented terrain processing data; The segmented terrain processing data is encapsulated in a custom format and pre-stored in the non-volatile storage device of the underwater autonomous platform.

3. The obstacle avoidance control method according to claim 1, characterized in that, The step of obtaining target terrain block data of the underwater autonomous platform in the next motion cycle from the at least one terrain block data based on the real-time positioning information of the underwater autonomous platform includes: Determine whether the target terrain patch data is stored in the memory cache of the underwater autonomous platform; If so, the target terrain block data is retrieved from the memory cache; If not, the target terrain data is obtained from the non-volatile storage device of the underwater autonomous platform.

4. The obstacle avoidance control method according to claim 1, characterized in that, The step of controlling the movement path of the underwater autonomous platform according to the safe altitude includes: Determine the elevation value of the target terrain block data; Based on the elevation value and the real-time positioning information of the underwater autonomous platform, the predicted height above the bottom of the underwater autonomous platform is determined. If the predicted height above the bottom and the safe height meet the first preset condition, the underwater autonomous platform is controlled to execute the first motion strategy. If the predicted height above the bottom and the safe height meet the second preset condition, the underwater autonomous platform is controlled to execute the second motion strategy. Based on the updated multi-source obstacle data, the movement path of the underwater autonomous platform is recontrolled.

5. An obstacle avoidance control system for an underwater autonomous platform, characterized in that, include: The offline data processing module performs offline data processing on the raw terrain data to obtain at least one terrain block data, and pre-stores the at least one terrain block data in the non-volatile storage device of the underwater autonomous platform; The real-time data processing module, based on the real-time positioning information of the underwater autonomous platform, acquires the target terrain block data of the underwater autonomous platform in the next motion cycle from the at least one terrain block data; determines the terrain gradient obstacle data based on the target terrain block data; determines the real-time sonar obstacle data based on the real-time sonar data of the underwater autonomous platform; and determines the historical sonar obstacle data based on the historical sonar data of the underwater autonomous platform. Multi-source obstacle data is determined based on the terrain gradient obstacle data, the real-time sonar obstacle data, and the historical sonar obstacle data; The obstacle avoidance control module determines the safe height of the underwater autonomous platform based on the multi-source obstacle data; and controls the movement path of the underwater autonomous platform based on the safe height.

6. The obstacle avoidance control system according to claim 5, characterized in that, The offline data processing module performs data parsing and reconstruction, data slicing and spatial index construction on the original terrain data to obtain segmented terrain processing data; The offline data processing module encapsulates the segmented terrain processing data in a custom format and pre-stores it in the non-volatile storage device of the underwater autonomous platform.

7. The obstacle avoidance control system according to claim 5, characterized in that, The real-time data processing module determines whether the target terrain block data is stored in the memory cache of the underwater autonomous platform; If so, the real-time data processing module retrieves the target terrain block data from the memory cache; If not, the real-time data processing module obtains the target terrain block data from the non-volatile storage device of the underwater autonomous platform.

8. The obstacle avoidance control system according to claim 5, characterized in that, The obstacle avoidance control module determines the elevation value of the target terrain block data; The obstacle avoidance control module determines the predicted height of the underwater autonomous platform above the bottom based on the elevation value and the real-time positioning information of the underwater autonomous platform. When the predicted height from the bottom and the safe height meet the first preset condition, the obstacle avoidance control module controls the underwater autonomous platform to execute the first motion strategy. When the predicted height from the bottom and the safe height meet the second preset condition, the obstacle avoidance control module controls the underwater autonomous platform to execute the second motion strategy. The obstacle avoidance control module re-controls the movement path of the underwater autonomous platform based on the updated multi-source obstacle data.

9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-4.

10. A non-volatile computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-4.

Images