Control method and system for working process of underwater robot

Abstract

The invention relates to the technical field of robot control, in particular to a control method and system for the working process of an underwater robot. The method comprises the following steps: step S100, acquiring a three-dimensional grid map of a to-be-rescued area; wherein the three-dimensional grid map comprises a plurality of grid blocks, and each grid block corresponds to one working path point and one preset collision risk coefficient; s102, acquiring a current working path point of the underwater robot in the three-dimensional grid map and environment parameters of a grid block at an adjacent position of the underwater robot; s104, determining a working path point corresponding to the next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map and the current working path point; step S106, circularly executing the steps S102 to S104 to obtain a plurality of working path points; and S108, determining the working path of the underwater robot based on the plurality of working path points. Therefore, intelligent planning of the working path of the underwater robot is realized, and the path planning precision in a complex dynamic environment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of robot control, and in particular to a method and system for controlling the working process of an underwater robot. Background Art

[0002] With the growing demand for marine resource development and underwater emergency rescue, underwater robots (AUVs), owing to their ability to operate in complex waters, have gradually become core equipment for underwater exploration and search and rescue. Path planning technology, crucial for autonomous navigation of AUVs, directly impacts mission efficiency and safety. However, existing technologies still have significant shortcomings in addressing the unique characteristics of underwater environments, multi-target coordination, and dynamic adaptability.

[0003] Traditional path planning algorithms are mostly based on static environment models and require pre-loading fixed obstacle maps. However, the underwater environment is highly dynamic, making the pre-set paths easily invalid.

[0004] Based on this, the present invention proposes a control method and system for the working process of an underwater robot to solve the above technical problems. Summary of the Invention

[0005] The present invention describes a control method and system for the working process of an underwater robot, which can improve the path planning accuracy in complex dynamic environments.

[0006] According to a first aspect, the present invention provides a method for controlling a working process of an underwater robot, the method comprising: Step S100: Obtain a three-dimensional grid map of the area to be rescued; wherein the three-dimensional grid map includes a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient; Step S102: Acquiring environmental parameters of the underwater robot's current working path point and adjacent grid blocks in the three-dimensional grid map; wherein the environmental parameters include water turbidity, water salinity, water flow velocity, and water flow direction; Step S104: determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map, and the current working path point; Step S106: looping through steps S102 to S104 to obtain multiple working path points; Step S108: Determine the working path of the underwater robot based on the multiple working path points.

[0007] According to a second aspect, the present invention provides a control system for an underwater robot working process, comprising: The first acquisition unit is configured to perform step S100: acquiring a three-dimensional grid map of the area to be rescued; wherein the three-dimensional grid map includes a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient; The second acquisition unit is configured to perform step S102: acquiring environmental parameters of the underwater robot's current working path point and adjacent grid blocks in the three-dimensional grid map; wherein the environmental parameters include water turbidity, water salinity, water flow speed, and water flow direction; The first data processing unit is configured to: determine, in step S104, a working path point corresponding to a next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map, and the current working path point; The second data processing unit is configured to: step S106: loop through steps S102 to S104 to obtain a plurality of working path points; The third data processing unit is configured to perform step S108: determining a working path of the underwater robot based on a plurality of working path points.

[0008] In a third aspect, an embodiment of this specification further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in any embodiment of this specification is implemented.

[0009] In a fourth aspect, an embodiment of this specification further provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the method described in any embodiment of this specification.

[0010] According to the present invention, a method and system for controlling the operation of an underwater robot first constructs a three-dimensional grid map of the rescue area using sonar mapping or pre-stored water area data. This map divides the underwater space into several grid blocks, each corresponding to a unique work path point. A preset collision risk coefficient (ranging from 0 to 1, with higher values ​​indicating a higher collision risk) is assigned based on prior information such as historical obstacle distribution and terrain complexity. Using an inertial navigation system and environmental sensors, the coordinates of the underwater robot's current work path point in the three-dimensional grid map, as well as multi-dimensional environmental parameters of adjacent grid blocks, including water turbidity, salinity, current velocity, and direction, are acquired in real time. Based on the environmental parameters, the three-dimensional grid map, and the current work path point, the underwater robot dynamically selects the work path point corresponding to the next grid block and controls the underwater robot to move to the corresponding work path point. By repeatedly executing the environmental parameter acquisition and path point decision process, multiple work path points are obtained. Based on these multiple work path points, the underwater robot's work path is determined. Through these steps, the present invention achieves intelligent work path planning for the underwater robot, significantly improving path planning accuracy and mission execution reliability in complex dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 A schematic flow chart showing a method for controlling an underwater robot during operation according to an embodiment is shown; Figure 2 A schematic block diagram of a control system for an underwater robot working process according to an embodiment is shown. DETAILED DESCRIPTION

[0013] The solution provided by the present invention is described below with reference to the accompanying drawings.

[0014] Figure 1 The flowchart of the control method of the underwater robot working process according to one embodiment is shown. It is understood that the method can be executed by any system, device, platform, or device cluster with computing and processing capabilities. Figure 1 As shown, the method includes: Step S100: Obtain a three-dimensional grid map of the area to be rescued; wherein the three-dimensional grid map includes a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient; Step S102: Obtaining environmental parameters of the underwater robot's current working path point and adjacent grid blocks in the three-dimensional grid map; wherein the environmental parameters include water turbidity, water salinity, water flow velocity, and water flow direction; Step S104: determining the working path point corresponding to the next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map and the current working path point; Step S106: looping through steps S102 to S104 to obtain multiple working path points; Step S108: Determine the working path of the underwater robot based on the multiple working path points.

[0015] In this embodiment, the present invention first constructs a three-dimensional grid map of the rescue area using sonar mapping or pre-stored water area data. This map divides the underwater space into several grid blocks, each corresponding to a unique work path point. A preset collision risk coefficient (ranging from 0 to 1, with higher values ​​indicating a higher collision risk) is assigned based on prior information such as historical obstacle distribution and terrain complexity. Using an inertial navigation system and environmental sensors, the coordinates of the underwater robot's current work path point in the three-dimensional grid map, as well as multi-dimensional environmental parameters of adjacent grid blocks, including water turbidity, salinity, current velocity, and direction, are acquired in real time. Based on these environmental parameters, the three-dimensional grid map, and the current work path point, the underwater robot dynamically selects the work path point corresponding to the next grid block and controls the underwater robot to move to the corresponding work path point. By repeatedly executing the environmental parameter acquisition and path point decision process, multiple work path points are obtained. Based on these multiple work path points, the underwater robot's work path is determined. Through these steps, the present invention achieves intelligent work path planning for the underwater robot, significantly improving path planning accuracy and mission execution reliability in complex dynamic environments.

[0016] In one embodiment of the present invention, determining a working path point corresponding to a next grid block of the underwater robot based on environmental parameters, a three-dimensional grid map, and a current working path point includes: Based on the three-dimensional grid map and the current working path point, determine the preset collision risk coefficient of the grid blocks adjacent to the current working path point; Determine the comprehensive risk coefficient of adjacent grid blocks based on environmental parameters and corresponding preset collision risk coefficients; Screening the comprehensive risk coefficients of the adjacent grid blocks that are smaller than a first preset safety threshold value to obtain the adjacent grid blocks whose comprehensive risk coefficients are smaller than the first preset safety threshold value; A grid block with the smallest distance from a preset rescue point is selected from adjacent grid blocks with a comprehensive risk coefficient less than a first preset safety threshold as a working path point.

[0017] In this embodiment, the process of determining the working path point corresponding to the next grid block of the underwater robot is as follows: First, based on the three-dimensional grid map and the robot's current position, the robot obtains a preset collision risk value for each adjacent grid area (this value is pre-set based on historical obstacle distribution and terrain complexity, with higher values ​​indicating greater collision risk). Next, the robot calculates a comprehensive risk factor for each adjacent grid area, combining real-time environmental data (including water turbidity, salinity, current velocity, and direction) with the preset risk values ​​for each adjacent grid area. This factor comprehensively assesses the impact of collision risk and environmental interference on movement. A first preset safety threshold is then set to screen adjacent grid areas with a comprehensive risk factor below this threshold to ensure the safety of candidate paths. Finally, within the selected safe grid areas, the distance between the center of each area and the preset rescue point is calculated. The grid area closest to the rescue target is selected as the next working path point, optimizing rescue efficiency based on the shortest path principle. This method achieves a balance between safety and efficiency in path planning in complex underwater environments through a layered screening mechanism involving risk pre-assessment, multi-factor comprehensive calculation, safety filtering, and distance optimization.

[0018] In one embodiment of the present invention, the comprehensive risk factor is determined by the following formula: Where t is the time, is the comprehensive risk factor, is the adaptive weight (dynamically adjusting the importance of each factor through deep learning), is the preset collision risk factor (high value in obstacle area and low value in free area), is the nonlinear mapping function of the jth environmental parameter, is the water turbidity mapping, is the water turbidity of grid block i, is the first preset turbidity threshold, is the second preset turbidity threshold, is the water salinity mapping, is the water salinity of grid block i, is the mean salinity, is the standard deviation of salinity, is the water velocity mapping, is the water flow velocity in grid block i, is the maximum design speed of the robot, is a constant, is the water flow direction mapping, is the angle between the water flow direction and the robot's target moving direction, is the parameter of the deep learning network at time t, is the learning rate, is the total reward function, is the gradient of the total reward, is the first preset weight coefficient, is the task completion time, is the second preset weight coefficient, For energy consumption, is the third preset weight coefficient, is the number of collision penalties, and n is the dimension of the environment parameters.

[0019] In this embodiment, the ingenuity of this system of equations lies in its innovative breakthrough in addressing the challenges of underwater robot path planning. By integrating multidimensional factors and incorporating cutting-edge technological concepts, it addresses the difficulty traditional methods face in navigating complex and dynamic underwater environments. The specific design and innovations are as follows: Multidimensional Fusion and Precise Modeling: Breaking through the limitations of traditional path planning, which relies solely on static maps or a single environmental factor, the system of equations incorporates multiple environmental parameters, including a preset collision risk factor, water turbidity, salinity, current velocity, and direction, to construct a high-dimensional environmental parameter vector and risk coefficient. These parameters encompass the key factors that influence underwater environmental conditions on robot movement. Through a cluster of nonlinear cost transformation functions, the raw environmental parameters are mapped into a normalized risk space, enabling precise mathematical modeling of complex underwater environments. For example, for water turbidity, a logarithmic transformation and a sigmoid function are used to fully account for its impact on the robot's vision and detection equipment, converting it into a quantifiable risk cost. Compared to traditional simple hierarchical assessments, this system more accurately reflects the environmental impact on movement costs, significantly improving the model's adaptability and accuracy to underwater environments. Dynamic Adaptive Weight Mechanism: The introduction of adaptive weight vectors and a deep reinforcement learning network breaks the traditional fixed or empirical weighting model used in path planning. Weights are no longer static values, but are dynamically generated through the deep reinforcement learning network based on real-time environmental parameters and historical environmental state memory. In different underwater scenarios, for example, in areas with dense obstacles, the network automatically increases the weight of the preset collision risk factor; in areas with turbulent water flow, it increases the weight of water speed and direction. This enables the robot to adaptively adjust its focus on different environmental factors, achieving more intelligent and realistic path planning decisions. This dynamic adaptive capability enables the robot to flexibly respond to environmental changes in complex and changing underwater environments, significantly improving the rationality and effectiveness of path planning. Multi-Objective Optimization and Continuous Learning: By constructing a multi-objective reward function centered on task completion time, accumulated energy consumption, and number of collisions, path planning is transformed from single-objective optimization to multi-objective collaborative optimization. Traditional path planning often focuses solely on obstacle avoidance or the shortest path. This system of equations balances multiple interrelated and potentially conflicting objectives, enabling the robot to reach its destination quickly while also balancing energy consumption and safety. At the same time, the reward function updates the parameters of the deep reinforcement learning network, forming a closed loop of continuous learning and optimization. As the robot continues to accumulate experience, it adjusts network parameters based on task performance, optimizes weight generation strategies, and continuously improves its path planning capabilities, achieving a leap from "pre-set rule execution" to "autonomous learning and optimization."

[0020] In this embodiment, the hyperbolic tangent function normalizes salinity to [-1, 1] and then maps it to [0, 1] after translation and scaling. The further away from the mean salinity, the higher the penalty. Both the mean and standard deviation of salinity are statistically derived based on historical data. The first preset turbidity threshold can be 10 NTU, and the second preset turbidity threshold can be 50 NTU. is a constant, greater than zero, used to adjust the steepness of the function.

[0021] In one embodiment of the present invention, the adaptive weight is determined by the following formula: Where, is the normalization function, For deep reinforcement learning networks, is the current environment parameter, For updated memory.

[0022] In this embodiment, adaptive weights are used to dynamically assign weights to the impact of various environmental parameters on the comprehensive risk factor, ensuring that the model adapts to different environments. A normalization function is used to ensure that the sum of the weights is 1 and non-negative.

[0023] In one embodiment of the present invention, the updated memory is determined by the following formula: Where, is a long short-term memory network, Memory of the previous moment In this embodiment, a long short-term memory (LSTM) network is used to construct a time series model of environmental states. This model accurately captures dynamic trends through the following mechanisms: Time series feature extraction: Environmental parameters (such as water velocity and direction) at consecutive time steps are encoded as a sequence input. The LSTM selectively retains historical state information through gating units (input gate, forget gate, and output gate), effectively identifying complex patterns such as periodic changes in water flow and sudden fluctuations. Dynamic weight adjustment: Based on the environmental trend prediction output by the LSTM, the influence weight of each environmental parameter is adjusted in real time. For example, if an upward trend in water velocity is detected, the weight of the flow-related cost function is automatically increased to plan a robust path. Multi-scale temporal dependency modeling: The LSTM network structure supports simultaneous capture of short-term changes (such as turbulent fluctuations) and long-term trends (such as tidal cycles). Through the state update mechanism of the memory units, nonlinear correlations across time steps are established, avoiding the information loss problem of traditional sliding window methods. Leveraging the adaptive nature of neural networks, this approach overcomes the limitations of traditional linear models such as the Kalman filter in nonlinear and non-stationary water flow environments, enabling in-depth perception of dynamic underwater environments and proactive decision-making for path planning.

[0024] In one embodiment of the present invention, when the number of adjacent grid blocks having a comprehensive risk coefficient less than a first preset safety threshold is less than one, the comprehensive risk coefficients of the adjacent grid blocks having a comprehensive risk coefficient less than a second preset safety threshold are screened to obtain adjacent grid blocks having a comprehensive risk coefficient less than the second preset safety threshold; wherein the second preset safety threshold is greater than the first preset safety threshold; Sort adjacent grids with comprehensive risk coefficients less than a second preset safety threshold from small to large to determine a risk priority list; Select the adjacent grid block with the smallest comprehensive risk factor in the risk priority list as the working path point.

[0025] In this embodiment, when the number of adjacent grid blocks with a comprehensive risk coefficient below a first preset safety threshold is less than one, the following emergency response process is executed: First, a second preset safety threshold with a higher fault tolerance level is activated (this threshold is greater than the first preset value, allowing for the inclusion of relatively high-risk path candidates). The comprehensive risk coefficients of adjacent grid blocks are rescreened, and all grid blocks with a risk coefficient below the second preset safety threshold are extracted to form a relaxed candidate set. The grid blocks in the candidate set are then sorted from smallest to largest by comprehensive risk coefficient to generate a risk priority list. Finally, the grid block with the lowest comprehensive risk coefficient in the list is selected as the next working path point. This ensures that, in extreme situations where an ideal safe path is unavailable, the robot can still choose the direction with the lowest relative risk, balancing mission execution requirements with environmental constraints. This mechanism, through hierarchical threshold screening and risk ranking strategies, enhances the path robustness of the underwater robot in complex environments and avoids mission interruptions due to local environmental obstructions.

[0026] In one embodiment of the present invention, when the number of adjacent grid blocks having a comprehensive risk coefficient less than a second preset safety threshold is less than one, the underwater robot is controlled to return to the original path.

[0027] In this embodiment, if no adjacent grid blocks meet the criteria after screening for the second preset safety threshold (i.e., the number of grid blocks with a comprehensive risk factor below the second preset safety threshold is zero), an extreme environment emergency mechanism is triggered: the underwater robot is directed to return along its previously planned path. This mechanism, leveraging the navigation system's path backtracking function, drives the robot in reverse direction along the sequence of path points it arrived at. This ensures that in high-risk environments where advancement is completely impossible, the robot can safely retreat to the starting area or a designated safe location, avoiding equipment failure or complete mission interruption caused by forced exploration of unknown high-risk areas. This design further improves the fault-tolerant nature of path planning, enhancing the underwater robot's survivability and mission reliability in complex and extreme environments.

[0028] The foregoing description describes specific embodiments of the present invention. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0029] According to another embodiment, the present invention provides a control system for a working process of an underwater robot. Figure 2 A schematic block diagram of a control system for an underwater robot working process according to an embodiment is shown. It is understood that the system can be implemented by any system, device, platform, or device cluster with computing and processing capabilities. Figure 2 As shown, the system includes: a first acquisition unit 200, a second acquisition unit 202, a first data processing unit 204, a second data processing unit 206 and a third data processing unit 208. The main functions of each component unit are as follows: The first acquisition unit 200 is configured to perform step S100: acquiring a three-dimensional grid map of the area to be rescued; wherein the three-dimensional grid map includes a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient; The second acquisition unit 202 is configured to perform step S102: acquiring environmental parameters of the underwater robot's current working path point and adjacent grid blocks in the three-dimensional grid map; wherein the environmental parameters include water turbidity, water salinity, water flow speed, and water flow direction; The first data processing unit 204 is configured to perform step S104: determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map, and the current working path point; The second data processing unit 206 is configured to perform step S106: looping through steps S102 to S104 to obtain a plurality of working path points; The third data processing unit 208 is configured to perform step S108: determining a working path of the underwater robot based on a plurality of working path points.

[0030] In one embodiment of the present invention, the first data processing unit 204 is configured to perform the following operations: Determining, based on the three-dimensional grid map and the current working path point, a preset collision risk coefficient of a grid block adjacent to the current working path point; Determining a comprehensive risk coefficient of adjacent grid blocks based on the environmental parameters and the corresponding preset collision risk coefficients; Screening the comprehensive risk coefficients of the adjacent grid blocks that are smaller than a first preset safety threshold value to obtain adjacent grid blocks whose comprehensive risk coefficients are smaller than the first preset safety threshold value; A grid block with the smallest distance from a preset rescue point is selected from the adjacent grid blocks with a comprehensive risk coefficient less than a first preset safety threshold as the working path point.

[0031] In one embodiment of the present invention, the system further includes a fourth data processing unit, and the fourth data processing unit is configured to perform the following operations: When the number of adjacent grid blocks having a comprehensive risk coefficient less than the first preset safety threshold is less than one, screening the adjacent grid blocks having comprehensive risk coefficients less than a second preset safety threshold among the comprehensive risk coefficients of the adjacent grid blocks to obtain adjacent grid blocks having comprehensive risk coefficients less than the second preset safety threshold; wherein the second preset safety threshold is greater than the first preset safety threshold; Sorting the adjacent grids whose comprehensive risk coefficients are less than the second preset safety threshold from small to large to determine a risk priority list; The adjacent grid block with the smallest comprehensive risk coefficient is selected from the risk priority list as the working path point.

[0032] In one embodiment of the present invention, the comprehensive risk factor is determined by the following formula: Where t is the time, is the comprehensive risk factor, is the adaptive weight, is the preset collision risk coefficient, is the nonlinear mapping function of the jth environmental parameter, is the water turbidity mapping, is the water turbidity of grid block i, is the first preset turbidity threshold, is the second preset turbidity threshold, is the water salinity mapping, is the water salinity of grid block i, is the mean salinity, is the standard deviation of salinity, is the water velocity mapping, is the water flow velocity in grid block i, is the maximum design speed of the robot, is a constant, is the water flow direction mapping, is the angle between the water flow direction and the robot's target moving direction, is the parameter of the deep learning network at time t, is the learning rate, is the total reward function, is the gradient of the total reward, is the first preset weight coefficient, is the task completion time, is the second preset weight coefficient, For energy consumption, is the third preset weight coefficient, is the number of collision penalties, and n is the dimension of the environment parameters.

[0033] In one embodiment of the present invention, the adaptive weight is determined by the following formula: Where, is the normalization function, For deep reinforcement learning networks, is the current environment parameter, For updated memory.

[0034] In one embodiment of the present invention, the updated memory is determined by the following formula: Where, is a long short-term memory network, Memory of the previous moment.

[0035] In one embodiment of the present invention, the system further includes a fifth data processing unit, and the fifth data processing unit is configured to perform the following operations: When the number of adjacent grid blocks having a comprehensive risk coefficient less than a second preset safety threshold is less than one, the underwater robot is controlled to return to the original path.

[0036] According to another embodiment, there is also provided a computer readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute a combination of Figure 1 The method described.

[0037] According to another embodiment, an electronic device is provided, comprising a memory and a processor, wherein the memory stores an executable code, and when the processor executes the executable code, the system realizes the combination of Figure 1 The method described.

[0038] The various embodiments of the present invention are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are described briefly because they are generally similar to the method embodiments. For relevant portions, refer to the description of the method embodiments.

[0039] Those skilled in the art will appreciate that, in one or more of the above examples, the functions described herein may be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions may be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0040] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for controlling the working process of an underwater robot, characterized in that: The method comprises: Step S100: Obtain a three-dimensional grid map of the area to be rescued; wherein the three-dimensional grid map includes a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient; Step S102: Acquiring environmental parameters of the underwater robot's current working path point and adjacent grid blocks in the three-dimensional grid map; wherein the environmental parameters include water turbidity, water salinity, water flow velocity, and water flow direction; Step S104: determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map, and the current working path point; Step S106: looping through steps S102 to S104 to obtain multiple working path points; Step S108: Determine the working path of the underwater robot based on the multiple working path points.

2. The method according to claim 1, characterized in that Determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map, and the current working path point includes: Determining, based on the three-dimensional grid map and the current working path point, a preset collision risk coefficient of a grid block adjacent to the current working path point; Determining a comprehensive risk coefficient of adjacent grid blocks based on the environmental parameters and the corresponding preset collision risk coefficients; Screening the comprehensive risk coefficients of the adjacent grid blocks that are smaller than a first preset safety threshold value to obtain adjacent grid blocks whose comprehensive risk coefficients are smaller than the first preset safety threshold value; A grid block with the smallest distance from a preset rescue point is selected from the adjacent grid blocks with a comprehensive risk coefficient less than a first preset safety threshold as the working path point.

3. The method according to claim 2, characterized in that The comprehensive risk factor is determined by the following formula: Where t is the time, is the comprehensive risk factor, is the adaptive weight, is the preset collision risk coefficient, is the nonlinear mapping function of the jth environmental parameter, For water turbidity mapping, is the water turbidity of grid block i, is the first preset turbidity threshold, is the second preset turbidity threshold, For water salinity mapping, is the water salinity of grid block i, is the mean salinity, is the standard deviation of salinity, is the water velocity mapping, is the water flow velocity in grid block i, is the maximum design speed of the robot, is a constant, is the water flow direction mapping, is the angle between the water flow direction and the robot’s target moving direction, is the parameter of the deep learning network at time t, is the learning rate, is the total reward function, is the gradient of the total reward, is the first preset weight coefficient, is the task completion time, is the second preset weight coefficient, For energy consumption, is the third preset weight coefficient, is the number of collision penalties, and n is the dimension of the environment parameters.

4. The method according to claim 3, characterized in that The adaptive weight is determined by the following formula: Where, is the normalization function, For deep reinforcement learning networks, is the current environment parameter, For updated memory.

5. The method according to claim 4, characterized in that The updated memory is determined by the following formula: Where, is a long short-term memory network, Memory of the previous moment.

6. The method according to claim 2, characterized in that Also includes: When the number of adjacent grid blocks having a comprehensive risk coefficient less than the first preset safety threshold is less than one, screening the adjacent grid blocks having comprehensive risk coefficients less than a second preset safety threshold among the comprehensive risk coefficients of the adjacent grid blocks to obtain adjacent grid blocks having comprehensive risk coefficients less than the second preset safety threshold; wherein the second preset safety threshold is greater than the first preset safety threshold; Sorting the adjacent grids whose comprehensive risk coefficients are less than the second preset safety threshold from small to large to determine a risk priority list; The adjacent grid block with the smallest comprehensive risk coefficient is selected from the risk priority list as the working path point.

7. The method according to claim 6, characterized in that Also includes: When the number of adjacent grid blocks having a comprehensive risk coefficient less than the second preset safety threshold is less than one, the underwater robot is controlled to return to the original path.

8. A control system for the working process of an underwater robot, characterized in that: include: The first acquisition unit is configured to perform step S100: acquiring a three-dimensional grid map of the area to be rescued; wherein the three-dimensional grid map includes a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient; The second acquisition unit is configured to perform step S102: acquiring environmental parameters of the underwater robot's current working path point and adjacent grid blocks in the three-dimensional grid map; wherein the environmental parameters include water turbidity, water salinity, water flow speed, and water flow direction; The first data processing unit is configured to: determine, in step S104, a working path point corresponding to a next grid block of the underwater robot based on the environmental parameters, the three-dimensional grid map, and the current working path point; The second data processing unit is configured to: step S106: loop through steps S102 to S104 to obtain a plurality of working path points; The third data processing unit is configured to perform step S108: determining a working path of the underwater robot based on a plurality of working path points.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

Images