An underwater unmanned vehicle emergency floating obstacle avoidance method, device, equipment and medium

Abstract

The application relates to the technical field of underwater unmanned vehicle control, and discloses an underwater unmanned vehicle emergency floating obstacle avoidance method, device, equipment and medium, environment observation data of a target object is acquired, collision risk prediction is carried out according to the environment observation data, and dynamic collision risk is obtained; risk-time collaborative path planning is carried out according to the dynamic collision risk and a floating time coefficient of the target object, and a current floating path is obtained; the target object is switched to a target floating mode, and a floating control amount is output according to the current floating path. The beneficial effect is that path planning is carried out on the basis of balancing collision risk and floating time, and the target object is controlled to perform emergency floating in the corresponding floating mode, actual floating requirements of the underwater unmanned vehicle are effectively adapted, safe and intelligent emergency floating in a complex underwater environment is realized, and the reliability and safety of the underwater unmanned vehicle are significantly improved.

Description

Technical Field

[0001] This application relates to the field of underwater unmanned vehicle control technology, and in particular to an emergency surfacing obstacle avoidance method, device, equipment and medium for underwater unmanned vehicles. Background Technology

[0002] Unmanned Underwater Vehicles (UUVs) are intelligent devices capable of autonomous navigation and mission execution underwater, playing an increasingly important role in areas such as marine environmental perception, resource exploration, underwater facility inspection, military reconnaissance, and anti-submarine warfare. Due to the complexity of the underwater environment, UUVs may encounter various emergency malfunctions during mission execution, requiring immediate initiation of emergency surfacing procedures to quickly return to the surface and ensure their own safety and the complete recovery of mission data. In related technologies, emergency surfacing of UUVs typically involves directly jettisoning ballast or surfacing by steering at a fixed angle / heading. However, these solutions lack the ability to avoid dynamic obstacles in the underwater environment and are difficult to adapt to actual surfacing requirements. Summary of the Invention

[0003] This application provides an emergency surfacing obstacle avoidance method, device, equipment, and medium for underwater unmanned vehicles. It performs path planning based on balancing collision risk and surfacing time, and controls the target object to surface in an emergency under the corresponding surfacing mode. It effectively adapts to the actual surfacing requirements of underwater unmanned vehicles, realizes safe and intelligent emergency surfacing in complex underwater environments, and significantly improves the reliability and safety of underwater unmanned vehicles.

[0004] To achieve the above objectives, the main technical solutions adopted in this application include: In a first aspect, embodiments of this application provide an emergency surfacing and obstacle avoidance method for an underwater unmanned vehicle, the method comprising: Obtain environmental observation data of the target object, and predict the collision risk of the target object in the environment based on the environmental observation data to obtain the dynamic collision risk of the target object; Based on the dynamic collision risk and the ascent time coefficient of the target object, risk-time coordinated path planning is performed to obtain the current ascent path of the target object; The target object is switched to the target ascent mode corresponding to the dynamic collision risk, and the ascent control amount of the target object is output according to the current ascent path in the target ascent mode to control the target object to perform emergency ascent.

[0005] The emergency ascent and obstacle avoidance method for underwater unmanned vehicles (UAVs) proposed in this application predicts the dynamic collision risk of the target object in its environment based on environmental observation data. Based on the dynamic collision risk and actual ascent requirements, collaborative path planning is performed to obtain a current ascent path that balances collision risk avoidance and ascent requirements. The target object is switched to a target ascent mode based on the dynamic collision risk, and in this mode, the target object is controlled to perform an emergency ascent according to the current ascent path. Compared with related technologies, this application performs collaborative path planning based on the dynamic collision risk of the environment and the target object's ascent time requirements, and controls the target object to perform an emergency ascent in the target ascent mode corresponding to the dynamic collision risk. This effectively meets the actual requirements of underwater UAVs for ascent time and safety, enabling rapid and safe ascent of underwater UAVs in complex underwater environments, and significantly improving the reliability and safety of underwater UAVs.

[0006] Optionally, the step of predicting the collision risk of the target object based on the environmental observation data in the environment where the target object is located, to obtain the dynamic collision risk of the target object, includes: Extract dynamic obstacles in the environment where the target object is located, estimate the motion of the dynamic obstacles based on the environmental observation data, and obtain the dynamic uncertainty range of the dynamic obstacles; Based on the dynamic uncertainty range, the collision risk of the target object is predicted to obtain the collision probability between the target object and any dynamic obstacle. The dynamic collision risk is obtained by jointly calculating the collision probabilities of all dynamic obstacles.

[0007] Optionally, the step of performing risk-time coordinated path planning based on the dynamic collision risk and the ascent time coefficient of the target object to obtain the current ascent path of the target object includes: Construct a global risk field for the environment in which the target object is located based on the dynamic collision risk; The global risk field is used to sample the guidance path in the environment where the target object is located, and the results of the guidance path sampling are optimized by risk accumulation to obtain a global reference path; At any moment when the target object is controlled to ascend in an emergency according to the global reference path, the local path optimization of the global reference path is performed based on the ascent time coefficient and the control state parameters of the target object, taking into account the ascent time, to obtain the current ascent path at any moment.

[0008] Optionally, the ascent time coefficient can be obtained in the following way: Obtain the remaining energy of the target object; Based on the preset energy range where the remaining energy is located, determine the buoyancy time coefficient corresponding to the preset energy range.

[0009] Optionally, the target ascent mode is determined by the following method: If the dynamic collision risk or the emergency state parameters of the target object, or the combination of parameters between the dynamic collision risk and the emergency state parameters, satisfies the first mode decision condition, the target ascent mode is the jettison ascent mode. If the dynamic collision risk, the emergency state parameters, or the combination of parameters meet the second mode decision conditions, the target ascent mode is the load adjustment ascent mode. If neither the dynamic collision risk nor the emergency state parameters nor the parameter combination satisfies the first mode decision condition nor the second mode decision condition, the target ascent mode is the obstacle avoidance ascent mode.

[0010] Optionally, the method further includes: When the target ascent mode is the payload jettison ascent mode, the current ascent path is locally optimized to obtain the path optimization result; If the path optimization result indicates that an optimized path is obtained through the local optimization that meets the preset safe ascent conditions within the preset confirmation time window, the target ascent mode is switched to the load adjustment ascent mode or the obstacle avoidance ascent mode. If the path optimization result indicates that the optimized path cannot be obtained through the local optimization, the target object is safely offset and the payload drop and float mode is executed.

[0011] Optionally, the method further includes: When the target floating mode switches between the obstacle avoidance floating mode, the load adjustment floating mode and the load shedding floating mode, the floating control amount of the target object before the switch and the floating control amount after the switch are obtained. Within a preset switching time window, the upward control quantity before switching and the upward control quantity after switching are smoothly calculated to obtain a smoothed control quantity. The smoothing control quantity is output to control the emergency buoyancy process of the target object within the preset switching time window.

[0012] Secondly, embodiments of this application provide an emergency surfacing and obstacle avoidance device for underwater unmanned vehicles, the device comprising: The collision risk prediction module is used to acquire environmental observation data of the target object, and to predict the collision risk of the target object in the environment based on the environmental observation data, so as to obtain the dynamic collision risk of the target object. The ascent path planning module is used to perform risk-time coordinated path planning based on the dynamic collision risk and the ascent time coefficient of the target object, so as to obtain the current ascent path of the target object. The buoyancy control output module is used to switch the target object to the target buoyancy mode corresponding to the dynamic collision risk, and output the buoyancy control amount of the target object according to the current buoyancy path in the target buoyancy mode, so as to control the target object to perform emergency buoyancy.

[0013] Thirdly, embodiments of this application provide a computer device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in any of the above embodiments.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions, which are used to cause a computer to perform the method described in any one of the above embodiments. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 A step diagram illustrating the emergency surfacing and obstacle avoidance method for an underwater unmanned vehicle provided in this application embodiment; Figure 2 This is a flowchart illustrating the steps of collision risk prediction in an embodiment of this application; Figure 3 This is a flowchart illustrating the steps of risk-time collaborative path planning in the embodiments of this application; Figure 4 This is a flowchart of the risk-time collaborative path planning in the embodiments of this application; Figure 5 This is a flowchart illustrating the steps for obtaining the buoyancy time coefficient in an embodiment of this application. Figure 6 This is a schematic diagram of the risk-time co-optimization model in the embodiments of this application; Figure 7 This is a flowchart illustrating the steps for determining the target buoyancy mode in an embodiment of this application. Figure 8 This is the state transition diagram of the floating mode decision engine in the embodiments of this application; Figure 9 This is a diagram illustrating the safety confirmation steps for the jettison and buoyancy mode in this application embodiment; Figure 10 This is a diagram illustrating the smooth switching steps between different buoyancy modes in the embodiments of this application; Figure 11 This is a schematic diagram of the smooth switching control timing in the embodiments of this application; Figure 12 A block diagram of an emergency surfacing and obstacle avoidance device for an underwater unmanned vehicle provided in an embodiment of this application; Figure 13 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. 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.

[0018] Unmanned Underwater Vehicles (UUVs) are intelligent devices capable of autonomous navigation and mission execution underwater, playing an increasingly important role in areas such as marine environmental perception, resource exploration, underwater facility inspection, military reconnaissance, and anti-submarine warfare. Due to the complexity of the underwater environment, UUVs may encounter various emergency malfunctions during mission execution, requiring immediate initiation of emergency surfacing procedures to quickly return to the surface and ensure their own safety and the complete recovery of mission data. In related technologies, emergency surfacing of UUVs typically involves directly jettisoning ballast or surfacing by steering at a fixed angle / heading. However, these solutions lack the ability to avoid dynamic obstacles in the underwater environment and are difficult to adapt to actual surfacing requirements.

[0019] Based on the above problems, this application provides an emergency surfacing obstacle avoidance method, device, equipment and medium for underwater unmanned vehicles. It acquires environmental observation data of the target object, performs collision risk prediction based on the environmental observation data to obtain dynamic collision risk; performs risk-time coordinated path planning based on dynamic collision risk and the surfacing time coefficient of the target object to obtain the current surfacing path; switches the target object to target surfacing mode, and outputs surfacing control quantity based on the current surfacing path.

[0020] The emergency ascent and obstacle avoidance method for underwater unmanned vehicles provided in this application predicts the dynamic collision risk of the target object in its environment based on environmental observation data; performs collaborative path planning for the target object based on the dynamic collision risk and actual ascent requirements to obtain a current ascent path that balances collision risk avoidance and ascent requirements; switches the target object to target ascent mode based on the dynamic collision risk, and controls the target object to perform emergency ascent based on the current ascent path in target ascent mode.

[0021] Compared with related technologies, this application performs collaborative path planning based on the dynamic collision risk of the environment and the ascent time requirement of the target object, and controls the target object to perform emergency ascent in the target ascent mode corresponding to the dynamic collision risk. This effectively meets the actual needs of underwater unmanned vehicles for ascent time and safety, realizes the rapid and safe ascent of underwater unmanned vehicles in complex underwater environments, and significantly improves the reliability and safety of underwater unmanned vehicles.

[0022] According to an embodiment of this application, an emergency surfacing and obstacle avoidance method for an underwater unmanned vehicle is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0023] Reference Figure 1 As shown in this embodiment, an emergency surfacing and obstacle avoidance method for an underwater unmanned vehicle is provided, the method comprising: S100. Obtain environmental observation data of the target object, and predict the collision risk in the environment where the target object is located based on the environmental observation data to obtain the dynamic collision risk of the target object.

[0024] S200. Perform risk-time coordinated path planning based on dynamic collision risk and the target object's ascent time coefficient to obtain the target object's current ascent path.

[0025] S300. Switch the target object to the target ascent mode corresponding to the dynamic collision risk, and output the ascent control amount of the target object according to the current ascent path in the target ascent mode, so as to control the target object to perform emergency ascent.

[0026] Specifically, the target object can be an underwater unmanned vehicle or other types of underwater vehicle, equipped with various types of sensors, including but not limited to forward-looking sonar, collision avoidance sonar, and altimeters. Environmental observation is conducted using the sensors on the target object, and point cloud construction is performed based on the sensor feedback data to obtain the observation point cloud of the target object's environment, serving as environmental observation data. It is understood that the location and type of obstacles in the target object's environment can be distinguished based on the environmental observation data. Obstacle types can include static obstacles whose positions do not change over time, and dynamic obstacles whose positions change over time. The environmental observation data for static obstacles can be represented using an octomap.

[0027] For any dynamic obstacle in the environment where the target object is located, the motion information of the dynamic obstacle is determined based on the changes in the observed point cloud of the dynamic obstacle between different frames. Motion prediction is then performed based on the position and motion information of the dynamic obstacle to obtain its predicted position at a future time. Based on the current position of the target object and the predicted positions of all dynamic obstacles, the probability of a collision between the target object and any dynamic obstacle is predicted, thus obtaining the dynamic collision risk of the target object.

[0028] It is understood that the dynamic collision risk obtained in this embodiment can be in analytical form, including the specific path taken by the target object during emergency ascent. This specific path can be an unknown quantity or a preset ascent path before risk-time collaborative path planning. Therefore, the dynamic collision risk can serve as a guiding condition during risk-time collaborative path planning, guiding the target object to generate an ascent path with fewer collisions with dynamic obstacles, thus ensuring the safety of the target object during emergency ascent.

[0029] In some embodiments, based on the dynamic collision risk, the static collision risk between the target object and the static obstacle can also be predicted, and the static collision risk and dynamic collision risk can be combined for subsequent path planning and control output. In this embodiment, the prediction of static collision risk is not specifically limited.

[0030] Furthermore, the ascent time coefficient of the target object can be used to represent the importance of ascent time to the target object during emergency ascent. It can be determined based on the current state of the target object or based on the control objective of the target object. For example, when it is desired that the target object ascent to the surface faster, the ascent time coefficient is increased, enabling the target object to generate a current ascent path with shorter time. In some embodiments, when only ascent time and collision risk are considered during the emergency ascent of the target object, the target object can also have an ascent risk coefficient to represent the importance of ascent risk to the target object during emergency ascent. This coefficient can be determined based on the dynamic collision risk of the target object, and the ascent time coefficient and the ascent risk coefficient satisfy normalization constraints.

[0031] Dynamic collision risk and ascent time coefficient are used as cooperative guidance conditions. Path planning for the target object is performed based on these conditions to balance the ascent risk and ascent time, while simultaneously meeting the target object's actual ascent requirements, thus improving the target object's adaptability to complex underwater environments and the safety of emergency ascents. In some embodiments, the risk-time cooperative path planning process can also combine path optimization algorithms with the path planning algorithm to optimize the current ascent path in real time during emergency ascents, further enhancing the safety of the current ascent path.

[0032] For example, the path planning algorithm in this embodiment may include, but is not limited to, graph search-based algorithms, sampling-based algorithms, and intelligent optimization algorithms. Sampling-based algorithms may include Rapidly-exploring Random Tree (RRT) and Probabilistic Roadmap (PRM). The path optimization algorithm in this embodiment may include, but is not limited to, Model Predictive Control (MPC), trajectory optimization, and numerical optimization. Model Predictive Control may include Linear Model Predictive Control (LMPC) and Nonlinear Model Predictive Control (NMPC).

[0033] Furthermore, the target object's ascent mode is switched based on dynamic collision risk, placing it in the target ascent mode corresponding to the dynamic collision risk. The current ascent path is input to the target object's underlying controller. In target ascent mode, the underlying controller converts the current ascent path into control commands for actuators such as thrusters, serving as ascent control quantities. The underlying controller then outputs these ascent control quantities to the corresponding actuators, driving each actuator to control the target object to perform emergency ascent along the current ascent path.

[0034] The emergency ascent and obstacle avoidance method for underwater unmanned vehicles provided in this embodiment predicts the dynamic collision risk of the target object in its environment based on environmental observation data; performs collaborative path planning for the target object based on the dynamic collision risk and actual ascent requirements to obtain a current ascent path that balances collision risk avoidance and ascent requirements; switches the target object to target ascent mode based on the dynamic collision risk, and controls the target object to perform emergency ascent based on the current ascent path in target ascent mode.

[0035] Compared with related technologies, this application performs collaborative path planning based on the dynamic collision risk of the environment and the ascent time requirement of the target object, and controls the target object to perform emergency ascent in the target ascent mode corresponding to the dynamic collision risk. This effectively meets the actual needs of underwater unmanned vehicles for ascent time and safety, realizes the rapid and safe ascent of underwater unmanned vehicles in complex underwater environments, and significantly improves the reliability and safety of underwater unmanned vehicles.

[0036] Reference Figure 2 As shown, in one embodiment of this application, collision risk prediction is performed in the environment where the target object is located based on environmental observation data to obtain the dynamic collision risk of the target object, including: S110. Extract dynamic obstacles in the environment where the target object is located, estimate the motion of the dynamic obstacles based on environmental observation data, and obtain the dynamic uncertainty range of the dynamic obstacles.

[0037] S120. Based on the dynamic uncertainty range, predict the collision risk of the target object to obtain the collision probability between the target object and any dynamic obstacle.

[0038] S130. The collision probabilities of all dynamic obstacles are jointly calculated to obtain the dynamic collision risk.

[0039] Specifically, when environmental observation data is represented by an octree map, for any grid cell in the octree map, the environmental observation data in that grid cell is recursively updated based on the feedback data from the target object's sensors. Continuous inter-frame difference calculations are performed on each grid cell in the octree map to identify changing grid cells with point cloud variations, and these changing grid cells are clustered and segmented to obtain the dynamically moving obstacles in the environment where the target object is located. It is understood that other mature and conventional methods can also be used in other embodiments, and this embodiment does not impose specific limitations. For example, the motion state vector of any dynamic obstacle can be represented as: in, Let be the motion state vector of the i-th dynamic obstacle; , and Let i be the position coordinates of the i-th dynamic obstacle in different directions; , and Let be the speed of the i-th dynamic obstacle in different directions.

[0040] After separating dynamic obstacles, motion estimation is performed on any dynamic obstacle based on multi-frame environmental observation data to predict the position change and uncertainty of any dynamic obstacle at any future time, thereby determining the range of dynamic uncertainty of any dynamic obstacle. In some embodiments, the motion estimation process can be performed using an extended Kalman filter (EKF). In this embodiment, it is assumed that the dynamic obstacle follows a constant velocity motion model (CV) in local time. The discretized state equation and observation equation for motion estimation of any dynamic obstacle can be expressed as: in, Let be the motion state vector of the i-th dynamic obstacle at time k. Let be the motion state vector of the i-th dynamic obstacle at time k-1; This is the state transition matrix; The process noise at time k-1 follows a process noise probability distribution. , The process noise covariance matrix; The geometric center position measurement of the i-th dynamic obstacle at time k; The observation matrix; The observation noise at time k follows the observation noise probability distribution. , The observation noise covariance matrix. The state transition matrix. and observation matrix It can be represented as: in, It is a 3×3 identity matrix; The sensor sampling period can be set to 0.1s.

[0041] Based on the above formula, the extended Kalman filter is used to estimate the state of any dynamic obstacle. The iterative prediction step at any moment during the state estimation process can be expressed as: in, Let k be the prior state estimate at time k. This is the posterior state estimate at time k-1; Let be the prior covariance matrix at time k; Let be the posterior covariance matrix at time k-1. It can be understood that the iterative update step corresponding to any given time can be expressed as: in, The Kalman gain matrix; The posterior state estimate at time k; Let be the posterior covariance matrix at time k; It is an identity matrix.

[0042] Based on the posterior state estimate and posterior covariance matrix at time k, the state and uncertainty of the dynamic obstacle after a preset number of steps are predicted, where the prediction result at any time can be expressed as: in, The state prediction result at time k+j; The covariance prediction result at time k+j; Preset number of steps In any previous prediction step, , We can take 50. The predicted distribution of the position components of each dynamic obstacle at time k+j, expressed using the Gaussian distribution form, can be represented as: in, Let be the position component of the i-th dynamic obstacle at time k+j; Let Gaussian distribution represent the predicted distribution. For covariance prediction results Components at corresponding positions The submatrix. It is understandable that the Gaussian distribution... It is denoted as an uncertain Gaussian distribution, which constitutes the uncertainty ellipsoid of the i-th dynamic obstacle at time k+j. As the dynamic uncertainty range of the i-th dynamic obstacle, it can be used for subsequent collision risk prediction.

[0043] Furthermore, at the corresponding future moment, the safety envelope range of the target object is determined based on its location, and the collision probability between the target object and any dynamic obstacle is determined by integrating the safety envelope range and the dynamic uncertainty range of any dynamic obstacle. For example, the collision probability can be expressed as: in, Let be the collision probability between the target object and the i-th dynamic obstacle. The safe outer volume of the target object; for The time uncertainty is a Gaussian distribution at point The value at that location.

[0044] Furthermore, the collision probabilities between the target object and each dynamic obstacle are jointly calculated to obtain the dynamic collision risk of the target object by calculating the probability of the target object colliding with at least one of the dynamic obstacles. The specific calculation process of the dynamic collision risk may include: calculating a first safety probability that the target object will not collide with any dynamic obstacle based on the collision probability between the target object and any dynamic obstacle; multiplying the first safety probabilities between the target object and all dynamic obstacles to calculate a second safety probability that the target object will not collide with any dynamic obstacle; and calculating the probability of the target object colliding with at least one dynamic obstacle based on the second safety probability, as the dynamic collision risk. For example, the dynamic collision risk can be expressed as: in, For the dynamic collision risk of the target object; This represents the total number of dynamic obstacles. This represents a product operation. For ease of calculation, the above equation can be simplified using the exponential form of the Mahalanobis distance. The simplified dynamic collision risk can be expressed as: in, This is a normalized constant related to the size of the dynamic obstacle; For the target object in The position at any given moment.

[0045] Reference Figure 3 As shown in one embodiment of this application, risk-time coordinated path planning is performed based on dynamic collision risk and the ascent time coefficient of the target object to obtain the current ascent path of the target object, including: S210. Construct a global risk field for the environment in which the target object is located based on dynamic collision risk.

[0046] S220. Use the global risk field to sample the guiding path in the environment where the target object is located, and perform risk accumulation optimization on the results of the guiding path sampling to obtain the global reference path.

[0047] S230. At any moment when the target object is controlled to rise in an emergency according to the global reference path, the local path optimization of the global reference path is performed based on the rise time coefficient and the control state parameters of the target object, taking into account the rise time, to obtain the current rise path at any moment.

[0048] Reference Figure 4 As shown, the dynamic collision risk of the target object and an octree map containing environmental observation data are input into the global planning layer. The dynamic collision risk of the target object in its environment, as well as the static collision risk with static obstacles, are fused to determine the global risk at any spatial point in the target object's environment. This constructs a global risk field for guiding sampling. For example, the global risk field can be represented as: in, For spatial points The size of the risk potential field; For spatial points Static risks; For spatial points Dynamic risks. Static risks. It can be obtained from an octree map, and is inversely proportional to the distance between the target object and static obstacles, which can be expressed as: in, This is the static risk gain coefficient, which can be set to 1. For spatial points Euclidean distance between the nearest static obstacle surface; To prevent division by zero of constants; A safe distance threshold can be set at 2 meters. Dynamic risk. It can be composed of the superposition of the dynamic uncertainty ranges of all dynamic obstacles, and can be expressed as: in, This is the time decay factor, which can be set to 0.9.

[0049] Furthermore, an initial reference path for the target object is generated through guided path sampling. At any sampling moment during guided path sampling, a sampling guidance factor is determined based on the dynamic collision risk of the target object at that sampling moment. Based on the sampling guidance factor, the initial reference path is guided to a safe area of ​​the target object's environment, and multiple spatial points are selected from the safe area as candidate samples. For example, the sampling guidance factor can be expressed as: in, for The sampling guidance factor at any given time can range from 0.1 to 0.5. In this embodiment... ; for The maximum instantaneous risk at a given time can be expressed as: .

[0050] Based on this, a risk assessment is performed on each spatial point in the candidate sample using a global risk field. The sampling probability density is obtained based on the collision risk of the target object with obstacles at each spatial point. This sampling probability density is then used to extract new path nodes from the candidate sample. It can be understood that the sampling probability density can gradually decrease as the collision risk at each spatial point increases; that is, the higher the collision risk of any spatial point, the lower the probability of extracting that spatial point as a path node, thus guiding the initial reference path towards open water with fewer obstacles. For example, the sampling probability density can be expressed as: in, This represents the sampling probability density; This is a risk scaling factor used to control the sensitivity to risk differences during sampling. Its value can range from 1 to 2. In this embodiment... .

[0051] The initial reference path is updated based on the new path nodes, and the risk cost of each path node in the initial reference path is calculated to determine the risk cost of each path node. It can be understood that the risk cost of any path node can include the risk cost of its parent node, the path distance between the path node and its parent node, and the path collision risk of the connecting path between the parent node and the path node. The parent node risk cost is used to improve the global nature of the risk cost; the higher the parent node risk cost, the higher the risk cost of any path node. The path distance represents the movement time of the target object between adjacent path nodes, used in risk accumulation optimization to represent the ascent time of the target object, achieving a balance between ascent time and ascent risk; the longer the path distance, the higher the risk cost of any path node. Path collision risk can be used to represent the collision risk between the target object and dynamic obstacles when traversing this path segment; the higher the path collision risk, the higher the risk cost of any path node.

[0052] For example, the cumulative risk cost function used for calculating risk costs can be expressed as: in, For spatial points The parent node of the node on the path above; For spatial points The cumulative risk cost function value of the upstream path nodes; For spatial points The cumulative risk cost function value of the parent node of the upstream path node; This is the initial reference path; and These are the weighting coefficients.

[0053] Based on the risk cost of each path node on the initial reference path, topology optimization is performed on the initial reference path. Connecting edges are regenerated between existing path nodes, resulting in an optimized path with the minimum total risk cost of all path nodes. This optimized path is then used as the initial reference path. It should be noted that during the optimization path generation process, dynamic collision detection can be performed on the newly generated connecting edges between path nodes. Specifically, this includes: obtaining the dynamic uncertainty range of any dynamic obstacle at any future time, comparing the spatial relationship between the dynamic uncertainty range and the connecting edge; if the connecting edge falls within the dynamic uncertainty range, it is determined to be invalid. After being determined to be invalid, the connecting edge is discarded.

[0054] Understandably, once the initial reference path reaches the water surface, the path sampling process is considered complete, and this initial reference path is used as the global reference path. This global reference path is then input into the local planning layer as the benchmark for the target object's ascent path, and is used in subsequent actual ascent processes to reread and optimize the local path to obtain the optimal current ascent path.

[0055] Furthermore, the emergency ascent process of the target object can be divided into multiple moments. At any given moment, the control cost of the target object within a preset optimization time window after that moment is calculated based on the current global reference path, serving as the data basis for local path optimization. It can be understood that the control cost may include the operational cost at each moment within the preset optimization time window except for the last moment, the terminal cost at the last moment within the preset optimization time window, and the overall risk cost of the global reference path. The overall risk cost can be calculated based on the ascent risk coefficient, and the length of the preset optimization time window may include... That moment.

[0056] For example, the spatial motion of the target object can be represented by a discrete-time nonlinear model, which can be expressed as: in, Let be the motion state vector of the target object at time k+1. Let k be the motion state vector of the target object at time k; Let k be the upward control value of the target object at time k; This is a discrete-time nonlinear model. The motion state vector of the target object can be represented as: in, , and The coordinates of the target object in different directions; , and The attitude angles of the target object in different directions; , and The linear velocity of the target object in different directions; , and Let be the angular velocity of the target object in different directions. The upward control variable of the target object can be expressed as: in, The rudder angle for the target object; The elevator angle of the target object; Let be the thruster rotational speed of the target object. Let any given moment be time k. Based on the target object's motion state vector, the finite-time optimal control problem for the target object at each moment can be expressed as: in, The control sequence to be optimized is the global reference path; The predicted value of the motion state vector of the target object at time k+j; The state value of the global reference path at time k+j; The predicted value of the upward control quantity of the target object at time k+j; for Predicted value of the motion state vector of the target object at any given time; for The status value of the global reference path at any given time; This is the state tracking error weighting matrix, used to penalize the deviation between the predicted value of the motion state vector and the state value of the global reference path. The larger the corresponding element in the state tracking error weighting matrix, the more serious the deviation of the target object from the global reference path. The control weighting matrix is ​​used to penalize based on the magnitude of the predicted upward control value. The larger the corresponding element in the control weighting matrix, the greater the penalty will be when the target object changes its motion state vector by a large amount. At this time, the control action of the target object is smoother, which can save energy and reduce actuator wear, but will reduce the response speed of the target object. The terminal state error weighting matrix is ​​used to penalize the state error of the target object in the last moment of the preset optimization time window, so as to ensure the closed-loop stability of local path optimization by approximating infinite time domain optimization. This is a risk penalty item; The predicted position components of the dynamic obstacle at time k+j; This is the coefficient for the risk penalty item. Let be the buoyancy risk coefficient at time k. It should be noted that in the finite-time optimal control problem, time k+j and... All times are within the preset optimization time window, and time k is the last time before the start of the preset optimization time window.

[0057] The constraints of the above finite-time optimal control problem can be expressed as: in, Let k+j be the position component of the target object at time k+j; This is the safe distance threshold; and The extreme value of the control variable for the target object; and The extreme values ​​of the control variables of the target object. The position components of the target object at time k+j. It can be represented as: in, , and Let k+j be the coordinates of the target object in different directions. Safety distance threshold. It can be represented based on the chi-square distribution, and can be expressed as: Understandably, the safe distance threshold Used to enforce a safe distance between the target object and each dynamic obstacle.

[0058] It should be noted that in the above-mentioned finite-time optimal control problem, The running cost can include the state tracking error and control quantity of the target object in each time step other than the last time step within the preset optimization time window. The terminal cost can include the final motion state vector of the target object at the last moment within the preset optimization time window. This represents the global risk cost, and its structure can be the same as the dynamic collision risk represented by Mahalanobis distance. Its value is scaled by the upward risk coefficient; the more sensitive the local path optimization is to risk, the larger the upward risk coefficient. It is understandable that the upward risk coefficient and the upward time coefficient satisfy a normalization constraint. The upward risk coefficient can be determined based on the value of the upward time coefficient, thus considering the target object's upward time requirement during the local path optimization process.

[0059] Under the aforementioned constraints, the finite-time optimal control problem is solved to obtain the optimized path, which serves as the current ascent path for controlling the target object to perform emergency ascent in the next time step. In the next time step, the output value of the first element in the ascent control sequence corresponding to the optimized path is executed by the actuator. The current ascent path is then used as the global reference path, and local path optimization is performed again on the global reference path, thereby achieving closed-loop feedback and ensuring that the target object can perform emergency ascent along the optimal path at each time step.

[0060] Reference Figure 5 As shown, in one embodiment of this application, the buoyancy time coefficient is obtained in the following manner: S232. Obtain the remaining energy of the target object.

[0061] S234. Determine the buoyancy time coefficient corresponding to the preset energy range based on the preset energy range where the remaining energy is located.

[0062] Reference Figure 6 As shown, the remaining energy of the target object is obtained based on sensors within the target object. This remaining energy can be any energy source within the target object capable of supplying the actuator to perform control operations on the target object. After obtaining the remaining energy, it is compared with the boundary values ​​of a preset energy range to determine the preset energy range in which the remaining energy lies. Then, an upward buoyancy coefficient is determined based on the coefficient correspondence within this preset energy range. For example, the upward buoyancy coefficient can be expressed as: in, This is the buoyancy time coefficient; This represents the remaining energy. The buoyancy risk coefficient can be calculated based on the normalization constraint between the buoyancy time coefficient and the buoyancy risk coefficient. The buoyancy risk coefficient can be expressed as: in, This represents the risk factor for upward adjustment.

[0063] In some embodiments, the comprehensive cost function of the target object can also be obtained based on the uplift time coefficient and the uplift risk coefficient, which can be expressed as: in, The comprehensive cost function; The estimated total ascent time for the target object; This represents the allowable ascent time limit for the target object. It's understandable that when the target object has ample remaining energy, the ascent risk coefficient is greater than the ascent time coefficient, and the target object prioritizes reducing collision risk, tending to ascend in a safer manner. Conversely, when the target object has low remaining energy, the ascent risk coefficient is less than the ascent time coefficient, and the target object prioritizes shortening the ascent time, appropriately reducing its safety margin. (Comprehensive cost function) It can also be used in processes such as floating mode switching and floating path planning, for example as a data condition for determining the floating mode, or as a cost function in the floating path planning process to guide the floating path planning process.

[0064] Reference Figure 7 As shown, in one embodiment of this application, the target buoyancy mode is determined in the following manner: S310. When the dynamic collision risk or the emergency state parameters of the target object, or the combination of parameters between the dynamic collision risk and the emergency state parameters, satisfies the decision conditions of the first mode, the target ascent mode is the jettison ascent mode.

[0065] S320. When the dynamic collision risk or emergency state parameters or parameter combinations meet the decision conditions of the second mode, the target ascent mode is the load adjustment ascent mode.

[0066] S330. When the dynamic collision risk and emergency state parameters and parameter combinations do not meet the decision conditions of the first mode and the decision conditions of the second mode, the target ascent mode is the obstacle avoidance ascent mode.

[0067] Reference Figure 8 As shown, when a target object needs to perform an emergency ascent, the ascent mode is first set to obstacle avoidance ascent mode, and the emergency state function of the target object is collected, including but not limited to remaining energy, normalized fault level, and time urgency. The time urgency indicates whether the target object needs to accelerate its ascent speed based on the total available ascent time. In the first and second mode decision conditions, dynamic collision risk, emergency state parameters, and their combinations each correspond to their respective condition thresholds. The dynamic collision risk, emergency state parameters, and their combinations are compared with their respective condition thresholds, and the comparison results determine whether the first or second mode decision conditions are met. Specifically, if the dynamic collision risk, emergency state parameters, and their combinations satisfy any of the condition thresholds included in the first mode decision conditions, the first mode decision condition is deemed met; if the dynamic collision risk, emergency state parameters, and their combinations satisfy any of the condition thresholds included in the second mode decision conditions, the second mode decision condition is deemed met.

[0068] When the dynamic collision risk, the emergency status parameters of the target object, or a combination of both meet the first mode decision conditions, it is determined that the target object faces a significant collision risk, is in a state of energy depletion, or has a serious malfunction, requiring it to surface as quickly as possible. In this case, a jettisoning surfacing mode can be adopted to improve the target object's survival probability. When the dynamic collision risk, the emergency status parameters of the target object, or a combination of both meet the second mode decision conditions, it is determined that the target object faces a moderate collision risk, requiring an increase in surfacing speed or repair of its surfacing attitude to further ensure its safety. In this case, a ballast adjustment surfacing mode can be adopted, adjusting the ballast within the target object to adjust its surfacing speed and attitude. When neither the dynamic collision risk, the emergency status parameters of the target object, nor a combination of both meet the first or second mode decision conditions, the original obstacle avoidance surfacing mode is maintained.

[0069] For example, the target floating mode can be represented as: in, Target buoyancy mode; This is the jettison and buoyancy mode; This is the load adjustment and floating mode; This is the obstacle avoidance and ascent mode. The decision conditions for the first mode can be expressed as: in, The first mode decision condition; To normalize the fault level, ; The urgency of time can be expressed as: in, The total available ascent time for the target object can be obtained based on the remaining energy. The reference total ascent time for the target object. The decision condition for the second mode can be expressed as: in, This is the decision condition for the second mode.

[0070] It should be noted that in obstacle avoidance and surfacing modes, the surfacing control quantity can be output through the steering obstacle avoidance sliding mode controller. Taking depth-tip control in the vertical plane as an example, a linear sliding surface is first constructed, which can be expressed as: in, is the depth function of the linear sliding surface; The depth change rate of the target object; For depth error; The pitch angle function of the linear sliding surface; The rate of change of the pitch angle of the target object; This refers to the pitch angle error; and These are the design parameters for a linear sliding surface. Depth error. It can be represented as: in, The actual depth value of the target object; This represents the expected depth of the target object. (Pitch angle error) It can be represented as: in, The actual value of the pitch angle of the target object; This represents the expected pitch angle of the target object.

[0071] Differentiating the linear sliding surface described above and substituting it into the vertical dynamic model of the target object, we obtain the rate of change of the vertical velocity and the rate of change of the pitch angle of the target object, which can be expressed as: in, The rate of change of the vertical velocity of the target object; The nonlinear dynamic term for the vertical velocity of the target object; The control gain coefficient for the rate of change of vertical velocity with respect to the elevator angle; The rate of change of the pitch angle of the target object; The nonlinear dynamic term for the pitch angular velocity of the target object; The control gain coefficient of elevator angle on pitch rate; The elevator control law for the target object can be expressed as: in, This is the equivalent control term when the target object is on a linear sliding surface; To switch the gain; It is a linear sliding surface; Boundary layer thickness; This represents a saturation function used to reduce chattering. Understandably, the target object's rudder and horizontal tracking could also employ a similar design, with the thruster speed determined by a PI control based on the desired longitudinal velocity.

[0072] In the ballast adjustment and buoyancy mode, based on the net buoyancy auxiliary command given by the upper system of the target object, the rate of change of mass of movable ballast or ballast tanks within the target object is adjusted via a PI control law. For example, the rate of change of mass of the ballast tanks can be expressed as: in, The rate of change of the mass of the ballast water tank of the target object; The desired ballast water tank mass; To measure the mass of the ballast water tank; This is the proportionality coefficient; is the integral coefficient.

[0073] In the jettison and surfacing mode, a jettison safety confirmation must be completed beforehand. Once it is confirmed that jettisoning is necessary and the target object is in a safe jettison position, the jettison mechanism is triggered, causing the jettisonable ballast within the target object to detach rapidly, generating a significant positive buoyancy force on the target object. It is understandable that the jettison command in the jettison and surfacing mode is an irreversible discrete command. After executing the jettison command, any active control commands to the target object are disabled until the target object rises to the surface.

[0074] Reference Figure 9 As shown in one embodiment of this application, the method further includes: S312. When the target ascent mode is the load-drop ascent mode, perform local optimization on the current ascent path to obtain the path optimization result.

[0075] S314. If the path optimization result indicates that an optimized path is obtained through local optimization that meets the preset safe ascent conditions within the preset confirmation time window, the target ascent mode is switched to load adjustment ascent mode or obstacle avoidance ascent mode.

[0076] S316. If the path optimization result indicates that no optimized path has been obtained through local optimization, after a safe offset of the target object, execute the payload drop and float mode.

[0077] Specifically, this embodiment provides a detailed process for confirming the safety of payload jettisoning in the jettisoning and ascent mode. After determining that the target ascent mode is jettisoning and ascent mode, a path optimization attempt is first made through the local planning layer within the target object to confirm whether an optimized path that meets the preset safe ascent conditions can be generated within a preset confirmation time window. The path optimization result is obtained after a preset attempt time. For example, the preset safe ascent conditions may be that the target object can safely ascent without colliding with any obstacle.

[0078] If the path optimization result indicates that the local planning layer has obtained an optimized path that can meet the preset safe ascent conditions within the preset confirmation time window, it means that the target object can safely ascent without jettisoning. At this time, the jettisoning command of the target object is rejected, and the target ascent mode is switched to either the load adjustment ascent mode or the obstacle avoidance ascent mode based on whether the dynamic collision risk and / or emergency state parameters meet the second mode decision conditions.

[0079] If the path optimization results indicate that the local planning layer failed to obtain the optimized path, it means that the target object must jettison its ballast for safe ascent. At this point, based on the octree map and the dynamic uncertainty range, the target object's actuators are controlled to make a safe offset, causing the target object to move laterally for a short time to move away from potential overhead obstacles. After completing the safe offset, the jettison ascent mode is executed, allowing the target object to gain positive buoyancy by releasing ballast and achieve rapid ascent. This reduces the risk of collision between the target object and overhead obstacles, ensuring the safety of the target object during emergency ascent. Furthermore, it reduces the collision risk between ballast and obstacles after jettisoning, reducing the complexity of the underwater environment and further ensuring the safety of the target object during emergency ascent.

[0080] Reference Figure 10 As shown in one embodiment of this application, the method further includes: S340. When the target floating mode switches between obstacle avoidance floating mode, load adjustment floating mode and load dropping floating mode, obtain the floating control amount of the target object before the switch and the floating control amount after the switch.

[0081] S350. Perform smooth calculations on the floating control quantity before and after the switch within the preset switching time window to obtain the smoothed control quantity.

[0082] S360 outputs a smooth control quantity to control the emergency buoyancy process of the target object within a preset switching time window.

[0083] Reference Figure 11 As shown, the floating mode before the switch is taken as mode A, and the floating mode after the switch is taken as mode B. The floating control value of mode A is obtained as the floating control value before the switch, and the floating control value of mode B is obtained as the floating control value after the switch.

[0084] Furthermore, within a preset switching time window, a weighted average is calculated on the pre-switching and post-switching upward control quantities to obtain a smooth control quantity that changes over time, enabling a smooth transition from the pre-switching upward control quantity to the post-switching upward control quantity. For example, the smooth control quantity can be expressed as: in, This is the smoothing control quantity at time t within the preset switching time window; The smoothing switching coefficient at time t increases linearly from 0 to 1 within a preset switching time window; This is the upward control value before switching; This is the upward control value after switching; This refers to the time point when the mode is switched. The preset switching time window duration.

[0085] Furthermore, within a preset switching time window, a smooth control quantity is output to the actuator of the target object. The smooth control quantity is used to control the emergency ascent of the target object, thereby reducing the motion shock caused by the step change in the control quantity during the mode switching process and improving the reliability and stability of the target object during the emergency ascent.

[0086] Accordingly, please refer to Figure 12 This application provides an emergency surfacing and obstacle avoidance device for underwater unmanned vehicles, the device comprising: The collision risk prediction module 1210 is used to acquire environmental observation data of the target object, and to predict the collision risk in the environment where the target object is located based on the environmental observation data, so as to obtain the dynamic collision risk of the target object.

[0087] The ascent path planning module 1220 is used to perform risk-time collaborative path planning for the target object based on dynamic collision risk, and obtain the current ascent path of the target object.

[0088] The buoyancy control output module 1230 is used to switch the target object to the target buoyancy mode corresponding to the dynamic collision risk, and output the buoyancy control quantity of the target object according to the current buoyancy path in the target buoyancy mode, so as to control the target object to perform emergency buoyancy.

[0089] In some alternative implementations, the collision risk prediction module 1210 includes: The obstacle motion estimation unit is used to extract dynamic obstacles in the environment where the target object is located, and to estimate the motion of the dynamic obstacles based on environmental observation data to obtain the dynamic uncertainty range of the dynamic obstacles.

[0090] The collision risk prediction unit is used to predict the collision risk of the target object based on the dynamic uncertainty range, and obtain the collision probability between the target object and any dynamic obstacle.

[0091] The joint probability calculation unit is used to jointly calculate the collision probability of all dynamic obstacles to obtain the dynamic collision risk.

[0092] In some alternative implementations, the buoyancy path planning module 1220 includes: The risk field construction unit is used to construct a global risk field for the environment in which the target object is located based on dynamic collision risks.

[0093] The path sampling optimization unit is used to perform guided path sampling in the environment where the target object is located using the global risk field, and to perform risk accumulation optimization on the results of guided path sampling to obtain a global reference path.

[0094] The local path optimization unit is used to optimize the local path of the global reference path at any moment when the target object is controlled to rise in an emergency based on the global reference path, taking into account the rise time, based on the rise time coefficient and the control state parameters of the target object, so as to obtain the current rise path at any moment.

[0095] In some alternative implementations, the local path optimization unit includes: The remaining energy detection subunit is used to obtain the remaining energy of the target object.

[0096] The target coefficient determination sub-unit is used to determine the buoyancy time coefficient corresponding to the preset energy range based on the preset energy range where the remaining energy is located.

[0097] In some alternative implementations, the buoyancy control output module 1230 includes: The first mode determination unit is used to determine the target ascent mode as the jettison ascent mode when the dynamic collision risk or the emergency state parameters of the target object, or the combination of parameters between the dynamic collision risk and the emergency state parameters, meet the first mode decision conditions.

[0098] The second mode determination unit is used to determine the target ascent mode as the load adjustment ascent mode when the dynamic collision risk or emergency state parameters or parameter combinations meet the second mode decision conditions.

[0099] The third mode determination unit is used to determine the obstacle avoidance ascent mode when the dynamic collision risk and emergency state parameters and parameter combinations do not meet the decision conditions of the first mode and the decision conditions of the second mode.

[0100] In some optional implementations, the first mode determination unit includes: The path local optimization subunit is used to perform local optimization on the current ascent path when the target ascent mode is the load-drop ascent mode, and obtain the path optimization result.

[0101] The mode switching subunit is used to switch the target ascent mode to load adjustment ascent mode or obstacle avoidance ascent mode when the path optimization result indicates that the optimized path obtained through local optimization meets the preset safe ascent conditions within the preset confirmation time window.

[0102] The safety offset sub-unit is used to perform a safe offset on the target object and then execute the payload drop and float mode when the path optimization result indicates that no optimized path has been obtained through local optimization.

[0103] In some alternative implementations, the buoyancy control output module 1230 further includes: The control quantity acquisition unit is used to acquire the target object's floating control quantity before and after the switching when the target floating mode switches between obstacle avoidance floating mode, load adjustment floating mode and load dropping floating mode.

[0104] The smoothing calculation unit is used to perform smoothing calculations on the floating control quantity before and after the switch within a preset switching time window to obtain the smoothed control quantity.

[0105] The control output unit is used to output a smooth control quantity to control the emergency buoyancy process of the target object within a preset switching time window.

[0106] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0107] In this embodiment, the underwater unmanned vehicle emergency surfacing obstacle avoidance device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0108] Please see Figure 13 , Figure 13 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application, such as... Figure 13 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 13 Take a processor 10 as an example.

[0109] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0110] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0111] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0112] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0113] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0114] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the above embodiments are implemented.

[0115] This application provides a computer program product including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method of any embodiment of this application.

[0116] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and all such modifications and variations fall within the scope defined by the appended claims.

[0117] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0118] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0119] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0124] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0125] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0126] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for emergency surfacing and obstacle avoidance of an underwater unmanned vehicle, characterized in that, The method includes: Obtain environmental observation data of the target object, and predict the dynamic collision risk of the target object based on the environmental observation data in the environment where the target object is located. Risk-time coordinated path planning is performed based on the dynamic collision risk and the ascent time coefficient of the target object to obtain the current ascent path of the target object. This includes: constructing a global risk field of the environment in which the target object is located based on the dynamic collision risk; using the global risk field to sample the guiding path in the environment in which the target object is located, and performing risk accumulation optimization on the results of the guiding path sampling to obtain a global reference path; at any moment when the target object is controlled to perform emergency ascent according to the global reference path, local path optimization considering the ascent time is performed on the global reference path based on the ascent time coefficient and the control state parameters of the target object to obtain the current ascent path at that moment. The target object is switched to the target ascent mode corresponding to the dynamic collision risk, and in the target ascent mode, an ascent control quantity for the target object is output according to the current ascent path to control the target object to perform emergency ascent; wherein, the target ascent mode is determined in the following ways: when the dynamic collision risk or the emergency state parameter of the target object or the parameter combination between the dynamic collision risk and the emergency state parameter meets the first mode decision condition, the target ascent mode is a load shedding ascent mode; when the dynamic collision risk or the emergency state parameter or the parameter combination meets the second mode decision condition, the target ascent mode is a load adjustment ascent mode; when the dynamic collision risk, the emergency state parameter, and the parameter combination do not meet the first mode decision condition and the second mode decision condition, the target ascent mode is an obstacle avoidance ascent mode.

2. The method according to claim 1, characterized in that, The step of predicting the collision risk of the target object based on the environmental observation data in the environment where the target object is located, to obtain the dynamic collision risk of the target object, includes: Extract dynamic obstacles in the environment where the target object is located, estimate the motion of the dynamic obstacles based on the environmental observation data, and obtain the dynamic uncertainty range of the dynamic obstacles; Based on the dynamic uncertainty range, the collision risk of the target object is predicted to obtain the collision probability between the target object and any dynamic obstacle. The dynamic collision risk is obtained by jointly calculating the collision probabilities of all dynamic obstacles.

3. The method according to claim 1, characterized in that, The ascent time coefficient is obtained in the following manner: Obtain the remaining energy of the target object; Based on the preset energy range where the remaining energy is located, determine the buoyancy time coefficient corresponding to the preset energy range.

4. The method according to claim 1, characterized in that, The method further includes: When the target ascent mode is the payload jettison ascent mode, the current ascent path is locally optimized to obtain the path optimization result; If the path optimization result indicates that an optimized path is obtained through the local optimization that meets the preset safe ascent conditions within the preset confirmation time window, the target ascent mode is switched to the load adjustment ascent mode or the obstacle avoidance ascent mode. If the path optimization result indicates that the optimized path cannot be obtained through the local optimization, the target object is safely offset and the payload drop and float mode is executed.

5. The method according to claim 1, characterized in that, The method further includes: When the target floating mode switches between the obstacle avoidance floating mode, the load adjustment floating mode and the load shedding floating mode, the floating control amount of the target object before the switch and the floating control amount after the switch are obtained. Within a preset switching time window, the upward control quantity before switching and the upward control quantity after switching are smoothly calculated to obtain a smoothed control quantity. The smoothing control quantity is output to control the emergency buoyancy process of the target object within the preset switching time window.

6. An emergency surfacing and obstacle avoidance device for an underwater unmanned vehicle, characterized in that, The device includes: The collision risk prediction module is used to acquire environmental observation data of the target object, and to predict the collision risk of the target object in the environment based on the environmental observation data, so as to obtain the dynamic collision risk of the target object. The ascent path planning module is used to perform risk-time coordinated path planning based on the dynamic collision risk and the ascent time coefficient of the target object to obtain the current ascent path of the target object. This includes: constructing a global risk field of the environment in which the target object is located based on the dynamic collision risk; using the global risk field to sample the guiding path in the environment in which the target object is located, and performing risk accumulation optimization on the results of the guided path sampling to obtain a global reference path; at any moment when the target object is controlled to perform emergency ascent according to the global reference path, performing local path optimization on the global reference path considering the ascent time based on the ascent time coefficient and the control state parameters of the target object to obtain the current ascent path at that moment. The buoyancy control output module is used to switch the target object to the target buoyancy mode corresponding to the dynamic collision risk, and outputs a buoyancy control quantity for the target object according to the current buoyancy path in the target buoyancy mode to control the target object to perform emergency buoyancy; wherein, the target buoyancy mode is determined by the following methods: when the dynamic collision risk or the emergency state parameter of the target object or the parameter combination between the dynamic collision risk and the emergency state parameter satisfies the first mode decision condition, the target buoyancy mode is a load shedding buoyancy mode; when the dynamic collision risk or the emergency state parameter or the parameter combination satisfies the second mode decision condition, the target buoyancy mode is a load adjustment buoyancy mode; when the dynamic collision risk, the emergency state parameter, and the parameter combination do not satisfy the first mode decision condition and the second mode decision condition, the target buoyancy mode is an obstacle avoidance buoyancy mode.

7. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the method of any one of claims 1 to 5.

Images