A control method and system for underwater robot working process

Abstract

The present application relates to the technical field of robot control, and particularly relates to a control method and system for underwater robot working process. The method comprises the following steps: S100: obtaining a three-dimensional grid map of a rescue area; wherein the three-dimensional grid map comprises a plurality of grid blocks, each of which corresponds to a working path point and a preset collision risk coefficient; S102: obtaining a current working path point of the underwater robot in the three-dimensional grid map and an environmental parameter of a grid block adjacent to the position of the underwater robot; S104: determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameter, the three-dimensional grid map and the current working path point; S106: cyclically executing steps S102 to S104 to obtain a plurality of working path points; and S108: determining a working path of the underwater robot based on the plurality of working path points. Therefore, the present application realizes intelligent planning of the working path of the underwater robot and improves the path planning accuracy in a complex dynamic environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot control, and in particular to a control method and system for underwater robot working process. BACKGROUND

[0002] With the growth of marine resource development and underwater emergency rescue demand, underwater robots gradually become the core equipment of underwater detection and search and rescue due to their operation ability in complex waters. Path planning technology, as the key of autonomous navigation of underwater robots, directly affects the efficiency and safety of task execution. However, the existing technology still has significant defects in the particularity of underwater environment, multi-target cooperation and dynamic adaptability.

[0003] Traditional path planning algorithms are mostly based on static environment models and need to load fixed obstacle maps in advance. However, the underwater environment has strong dynamics, which leads to the invalidation of the preset path.

[0004] Therefore, the present application provides a control method and system for underwater robot working process to solve the above technical problems. SUMMARY

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

[0006] According to a first aspect, the present application provides a control method for underwater robot working process, comprising:

[0007] Step S100: obtaining a three-dimensional grid map of a rescue area; wherein the three-dimensional grid map comprises a plurality of grid blocks, each grid block corresponding to a working path point and a preset collision risk coefficient;

[0008] Step S102: obtaining a current working path point of the underwater robot in the three-dimensional grid map and an environmental parameter of an adjacent position grid block of the underwater robot; wherein the environmental parameter comprises water turbidity, water salinity, water flow speed and water flow direction;

[0009] Step S104: determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameter, the three-dimensional grid map and the current working path point;

[0010] Step S106: cyclically executing steps S102 to S104 to obtain a plurality of working path points;

[0011] Step S108: determining a working path of the underwater robot based on the plurality of working path points.

[0012] According to a second aspect, the present application provides a control system for an underwater robot working process, comprising:

[0013] A first obtaining unit configured to obtain a three-dimensional grid map of a rescue area in step S100; wherein the three-dimensional grid map comprises a plurality of grid blocks, each of which corresponds to a working path point and a preset collision risk coefficient;

[0014] A second obtaining unit configured to obtain a current working path point of the underwater robot in the three-dimensional grid map and an environmental parameter of an adjacent position grid block of the underwater robot in step S102; wherein the environmental parameter comprises water turbidity, water salinity, water flow speed and water flow direction;

[0015] A first data processing unit configured to determine a working path point corresponding to a next grid block of the underwater robot based on the environmental parameter, the three-dimensional grid map and the current working path point in step S104;

[0016] A second data processing unit configured to cyclically execute steps S102 to S104 to obtain a plurality of working path points in step S106;

[0017] A third data processing unit configured to determine a working path of the underwater robot based on the plurality of working path points in step S108.

[0018] In a third aspect, the embodiments of the present application also provide an electronic device, including a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the method of any of the embodiments of the present application.

[0019] In a fourth aspect, the embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program, when executed in a computer, causes the computer to execute the method of any of the embodiments of the present application.

[0020] According to the underwater robot working process control method and system provided by the application, the three-dimensional grid map of the rescue area is first constructed by sonar surveying or pre-stored water area data. The map divides the underwater space into a plurality of grid blocks, each grid block corresponds to a unique working path point, and a preset collision risk coefficient (the value range is 0-1, and the higher the value is, the higher the collision risk is) is given based on prior information such as historical obstacle distribution and terrain complexity. Using the inertial navigation system and the environment sensor, the current working path point coordinates of the underwater robot in the three-dimensional grid map and the multi-dimensional environment parameters of the adjacent position grid block, including water turbidity, water salinity, water flow velocity and water flow direction, are obtained in real time. Based on the environment parameters, the three-dimensional grid map and the current working path point, the working path point corresponding to the next grid block of the underwater robot is dynamically selected, and the underwater robot is controlled to move to the corresponding working path point. Through the cyclic execution of the environment parameter acquisition and path point decision process, a plurality of working path points are obtained; based on the plurality of working path points, the working path of the underwater robot is determined. Through the above steps, the intelligent planning of the working path of the underwater robot is realized, and the path planning accuracy and task execution reliability in the complex dynamic environment are significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0022] Figure 1 A flowchart of the underwater robot working process control method according to one embodiment is shown;

[0023] Figure 2 A schematic block diagram of the underwater robot working process control system according to one embodiment is shown. DETAILED DESCRIPTION

[0024] The scheme provided by the present application will be described below in combination with the drawings.

[0025] Figure 1 A flowchart of the underwater robot working process control method according to one embodiment is shown. It can be understood that the method can be executed by any system, device, platform, device cluster with computing and processing capability. As Figure 1 shown, the method comprises:

[0026] Step S100: acquire a three-dimensional grid map of a rescue area; wherein the three-dimensional grid map comprises a plurality of grid blocks, each of which corresponds to a working path point and a preset collision risk coefficient;

[0027] Step S102: acquire an environmental parameter of a grid block adjacent to a current working path point of the underwater robot in the three-dimensional grid map; wherein the environmental parameter comprises water turbidity, water salinity, water flow speed and water flow direction;

[0028] Step S104: determine a working path point corresponding to a next grid block of the underwater robot based on the environmental parameter, the three-dimensional grid map and the current working path point;

[0029] Step S106: cyclically execute steps S102 to S104 to obtain a plurality of working path points;

[0030] Step S108: determine a working path of the underwater robot based on the plurality of working path points.

[0031] In the embodiment, the three-dimensional grid map of the rescue area is first constructed by sonar surveying or pre-stored water area data. The map divides the underwater space into a plurality of grid blocks, each of which corresponds to a unique working path point, and is assigned a preset collision risk coefficient (value range 0-1, the higher the value, the higher the collision risk) based on prior information such as historical obstacle distribution and terrain complexity. Using an inertial navigation system and environmental sensors, the current working path point coordinate of the underwater robot in the three-dimensional grid map and the multi-dimensional environmental parameters of the adjacent position grid block, including water turbidity, water salinity, water flow speed and water flow direction, are acquired in real time. Based on the environmental parameter, the three-dimensional grid map and the current working path point, the working path point corresponding to the next grid block of the underwater robot is dynamically selected, and the underwater robot is controlled to move to the corresponding working path point. Through the cyclic execution of the environmental parameter acquisition and path point decision process, a plurality of working path points are obtained; based on the plurality of working path points, the working path of the underwater robot is determined. Through the above steps, the intelligent planning of the working path of the underwater robot is realized, and the path planning accuracy and task execution reliability in complex dynamic environment are significantly improved.

[0032] In an embodiment of the present application, based on the environmental parameter, the three-dimensional grid map and the current working path point, the working path point corresponding to the next grid block of the underwater robot is determined, comprising:

[0033] determining the preset collision risk coefficient of the grid block adjacent to the current working path point based on the three-dimensional grid map and the current working path point;

[0034] determining the comprehensive risk coefficient of the adjacent grid block according to the environmental parameter and the corresponding preset collision risk coefficient;

[0035] screening the comprehensive risk coefficients of the adjacent grid blocks that are less than the first preset safety threshold to obtain the adjacent grid blocks with the comprehensive risk coefficients less than the first preset safety threshold;

[0036] selecting the grid block with the minimum distance to the preset rescue point from the adjacent grid blocks with the comprehensive risk coefficients less than the first preset safety threshold as the working path point.

[0037] In the embodiment, the process of determining the working path point corresponding to the next grid block of the underwater robot is as follows:

[0038] First, according to the three-dimensional grid map and the current position of the robot, the preset collision risk value of the adjacent grid region is obtained (the value is pre-set according to the historical obstacle distribution and the terrain complexity, and the higher the value, the greater the collision risk). Then, combined with real-time environmental data (including water turbidity, salinity, water flow speed and direction) and the preset risk value of each adjacent grid, the comprehensive risk coefficient of each adjacent grid is calculated (comprehensive evaluation of the impact of collision risk and environmental interference on movement). Subsequently, a first preset safety threshold is set to screen out the adjacent grid regions with a comprehensive risk coefficient lower than the threshold to ensure the safety of the candidate path. Finally, in the screened safe grid region, the distance between the center position of each region and the preset rescue point is calculated, and the grid region closest to the rescue target is selected as the next working path point to improve the rescue efficiency according to the shortest path principle. Through the layer-by-layer screening mechanism of risk pre-evaluation, multi-factor comprehensive calculation, safety filtering and distance optimization, the safety and efficiency balance of path planning in complex underwater environment is realized.

[0039] In an embodiment of the present application, the comprehensive risk coefficient is determined by the following formula:

[0040]

[0041]

[0042]

[0043]

[0044] In the formula, t is the time, is the comprehensive risk coefficient, is the adaptive weight (the importance of each factor is dynamically adjusted through deep learning), is the preset collision risk coefficient (obstacle region is high value, free region is low value), is the nonlinear mapping function of the jth environmental parameter, is the water turbidity mapping, is the water turbidity of grid i, is a first preset turbidity threshold, is a second preset turbidity threshold, is a water body salinity map, is a water body salinity of grid i, is a mean value of salinity, is a standard deviation of salinity, is a water flow velocity map, is a water flow velocity of grid i, is a maximum design speed of the robot, is a constant, is a water flow direction map, is an included angle between the water flow direction and the target moving direction of the robot, is a parameter of the deep learning network at time t, is a learning rate, is a total reward function, is a gradient of the total reward, is a first preset weight coefficient, is a task completion time, is a second preset weight coefficient, is an energy consumption, is a third preset weight coefficient, is a collision frequency penalty, n is a dimension of the environment parameter.

[0045] In this embodiment, the creation of the equation set reflects an innovative breakthrough in the path planning problem of underwater robots. By integrating multi-dimensional factors and introducing cutting-edge technology concepts, it solves the problem that traditional methods are difficult to cope with complex dynamic underwater environments. The specific ideas and creativity are as follows: Multi-dimensional fusion and precise modeling: Breaking through the limitations of traditional path planning relying only on static maps or single environmental factors, the equation set takes into account multiple environmental parameters such as preset collision risk coefficients, water turbidity, salinity, water flow speed and direction, etc., and constructs a high-dimensional environmental parameter vector and risk coefficient. These parameters cover the key elements of underwater environments that affect robot movement. Through a cluster of nonlinear cost conversion functions, the original environmental parameters are mapped to a normalized risk space, achieving precise mathematical modeling of complex underwater environments. For example, for water turbidity, the log transformation and Sigmoid function are used to fully consider its impact on robot vision and detection equipment, and it is converted into a quantifiable risk cost. Compared with traditional simple hierarchical evaluation, it can more subtly reflect the impact of the environment on the movement cost, greatly improving the adaptability and accuracy of the model to underwater environments. Dynamic adaptive weight mechanism: Introducing adaptive weight vectors and deep reinforcement learning networks breaks the pattern of fixed weights or experience weights in traditional path planning. The weight is no longer a static value, but is dynamically generated by the deep reinforcement learning network according to real-time environmental parameters and historical environmental state memory. In different underwater environmental scenarios, such as in obstacle-dense areas, the network automatically increases the weight of the preset collision risk coefficient; in areas with turbulent water flow, the weight of water flow speed and direction is increased, so that the robot can adaptively adjust the attention to different environmental factors, making more intelligent and more realistic path planning decisions. This dynamic adaptive capability enables the robot to flexibly respond to environmental changes in complex and variable 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, cumulative energy consumption and collision frequency, the path planning is transformed from single-objective optimization to multi-objective collaborative optimization. Traditional path planning often focuses only on single objectives such as obstacle avoidance or shortest path, while this equation set balances multiple interrelated and possibly conflicting objectives, enabling the robot to not only quickly reach the destination, but also to consider energy consumption and safety. At the same time, the reward function is used to update the deep reinforcement learning network parameters, forming a closed loop of continuous learning and optimization. The robot accumulates experience during task execution and adjusts network parameters based on task execution results to optimize weight generation strategies, continuously improving path planning capabilities and achieving a leap from "preset rule execution" to "autonomous learning and optimization".

[0046] In the embodiment, the hyperbolic tangent function normalizes the salinity to [-1, 1], which is mapped to [0, 1] after translation and scaling. The further away from the mean of salinity, the higher the cost. Both the mean of salinity and the standard deviation of salinity are based on historical data statistics. 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, which is a parameter for adjusting the steepness of the function.

[0047] In an embodiment of the application, the adaptive weight is determined by the following formula:

[0048]

[0049] In the formula, is a normalization function, is a deep reinforcement learning network, is the current environmental parameter, is the updated memory.

[0050] In the embodiment, the adaptive weight is used to dynamically allocate the influence weight of each environmental parameter on the comprehensive risk coefficient, ensuring that the model adapts to different environments. The normalization function ensures that the weight sum is 1 and non-negative.

[0051] In an embodiment of the application, the updated memory is determined by the following formula:

[0052]

[0053] In the formula, is a long short-term memory network, is the previous moment memory

[0054] In this embodiment, a long short-term memory network (LSTM) is used to construct an environment state time sequence model, and the following mechanisms are used to accurately capture the dynamic change trend: time sequence feature extraction: the environment parameters (water flow speed, direction, etc.) of continuous time steps are encoded as sequence input, and the LSTM selectively retains historical state information through the gating unit (input gate, forget gate, output gate) to effectively identify complex patterns such as periodic changes in water flow and sudden water flow fluctuations. Dynamic weight adjustment: based on the environment change trend prediction output by the LSTM, the influence weight of each environment parameter is adjusted in real time. For example, when an upward trend in water flow speed is identified, the weight of the water flow-related cost function is automatically increased to plan an anti-interference path in advance. Multi-scale time-dependent modeling: the LSTM network structure supports the simultaneous capture of short-term changes (such as turbulent pulsations) and long-term trends (such as tidal cycles), and through the state update mechanism of the memory unit, a nonlinear association across time steps is established to avoid the information loss problem of traditional sliding window methods. This method breaks through the limitations of traditional linear models such as Kalman filter in nonlinear and non-stationary water flow environments through the adaptability of neural networks, and realizes deep perception of the dynamic changes of the underwater environment and forward-looking decision-making of path planning.

[0055] In an embodiment of the present application, when the number of adjacent grid blocks with a comprehensive risk coefficient less than the first preset safety threshold is less than one, the comprehensive risk coefficients less than the second preset safety threshold are screened from the comprehensive risk coefficients of the adjacent grid blocks, to obtain adjacent grid blocks with 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.

[0056] The adjacent grid blocks with a comprehensive risk coefficient less than the second preset safety threshold are sorted from small to large to determine a risk priority list.

[0057] The adjacent grid block with the smallest comprehensive risk coefficient in the risk priority list is selected as a working path point.

[0058] In the embodiment, when the number of adjacent grid blocks with the integrated risk coefficient less than the first preset safety threshold is less than one, the following emergency processing procedure is performed: first, a second preset safety threshold of a higher fault tolerance level is enabled (the threshold is greater than the first preset value, allowing the path candidate with relatively high risk to be included), the integrated risk coefficient of the adjacent grid blocks is re-screened, all grid blocks lower than the second preset safety threshold are extracted to form a candidate set after the condition is relaxed. Then, the grid blocks in the candidate set are sorted according to the integrated risk coefficient from small to large to generate a risk priority list. Finally, the grid block with the smallest integrated risk coefficient in the list is selected as the next working path point, ensuring that the robot can still choose the direction with the lowest relative risk to balance the task execution demand and environmental constraints in the extreme case of no ideal safety path. The mechanism enhances the path robustness of the underwater robot in complex environments through the hierarchical threshold screening and risk sorting strategy, avoiding task interruption due to local environmental blockage.

[0059] In an embodiment of the present application, when the number of adjacent grid blocks with the integrated risk coefficient less than the second preset safety threshold is less than one, the underwater robot is controlled to return along the original path.

[0060] In the embodiment, after the screening of the second preset safety threshold is performed, if no adjacent grid block meeting the condition (i.e., the number of grid blocks with the integrated risk coefficient lower than the second preset safety threshold is zero) is found, an extreme environment emergency mechanism is triggered: the underwater robot is controlled to return along the original path according to the planned historical path. The mechanism drives the robot to move in reverse along the sequence of working path points of the coming time through the path backtracking function of the navigation system, ensuring that the robot can safely evacuate to the starting area or a designated safe location in a completely inadvisable high-risk environment, avoiding equipment failure or complete task interruption due to forced exploration of unknown high-risk areas. This design further improves the fault tolerance system of path planning, and improves the survival ability and task reliability of the underwater robot in complex extreme environments.

[0061] The above describes specific embodiments of the present application. 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 other than that described in the embodiments and still achieve desirable results. In addition, the processes depicted in the figures do not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous or possible.

[0062] According to another aspect, embodiments of the present application provide a control system for an underwater robot working process. Figure 2A schematic block diagram of a control system for an underwater robot working process is shown according to an embodiment. It can be understood that the system can be implemented by any system, device, platform and device cluster with computing and processing capabilities. 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: Figure 2

[0063] The first acquisition unit 200 is configured to step S100: acquire a three-dimensional grid map of a rescue area; wherein the three-dimensional grid map includes a plurality of grid blocks, each of which corresponds to a working path point and a preset collision risk coefficient;

[0064] The second acquisition unit 202 is configured to step S102: acquire the current working path point of the underwater robot in the three-dimensional grid map and the environmental parameters of the adjacent position grid block of the underwater robot; wherein the environmental parameters include water turbidity, water salinity, water flow speed and water flow direction;

[0065] The first data processing unit 204 is configured to step S104: based on the environmental parameters, the three-dimensional grid map and the current working path point, determine the working path point corresponding to the next grid block of the underwater robot;

[0066] The second data processing unit 206 is configured to step S106: cyclically execute steps S102 to S104 to obtain a plurality of working path points;

[0067] The third data processing unit 208 is configured to step S108: based on a plurality of working path points, determine the working path of the underwater robot.

[0068] In an embodiment of the present application, the first data processing unit 204 is configured to perform the following operations:

[0069] Based on the three-dimensional grid map and the current working path point, determine the preset collision risk coefficient of the adjacent grid block of the current working path point;

[0070] According to the environmental parameters and the corresponding preset collision risk coefficient, determine the comprehensive risk coefficient of the adjacent grid block;

[0071] Screen the comprehensive risk coefficients of the adjacent grid blocks that are less than the first preset safety threshold to obtain the adjacent grid blocks whose comprehensive risk coefficients are less than the first preset safety threshold;

[0072] ​selecting, from the adjacent grid blocks with the comprehensive risk coefficient less than the first preset safety threshold, a grid block with a minimum distance to a preset rescue point as the work path point.

[0073] In an embodiment of the present application, the system further comprises a fourth data processing unit, configured to perform the following operations:

[0074] When the number of the adjacent grid blocks with the comprehensive risk coefficient less than the first preset safety threshold is less than one, screening the comprehensive risk coefficient less than a second preset safety threshold from the comprehensive risk coefficients of the adjacent grid blocks to obtain the adjacent grid blocks with the comprehensive risk coefficient less than the second preset safety threshold; wherein the second preset safety threshold is greater than the first preset safety threshold.

[0075] sorting the adjacent grid blocks with the comprehensive risk coefficient less than the second preset safety threshold from small to large to determine a risk priority list.

[0076] selecting, from the risk priority list, an adjacent grid block with the minimum comprehensive risk coefficient as the work path point.

[0077] In an embodiment of the present application, the comprehensive risk coefficient is determined by the following formula:

[0078]

[0079]

[0080]

[0081]

[0082] wherein t is a time, is the comprehensive risk coefficient, is an adaptive weight, is the preset collision risk coefficient, is a nonlinear mapping function of the jth environmental parameter, is a water turbidity mapping, is a water turbidity of the grid block i, is a first preset turbidity threshold, is a second preset turbidity threshold, is a water salinity mapping, is a water salinity of the grid block i, is a mean value of the salinity, is a standard deviation of the salinity, is a water flow speed mapping, is a water flow speed of the grid block i, is a maximum design speed of the robot, is a constant, is a water flow direction map, is an angle between the water flow direction and a target moving direction of the robot, is a parameter of a deep learning network at time t, is a learning rate, is a total reward function, is a gradient of the total reward, is a first preset weight coefficient, is a task completion time, is a second preset weight coefficient, is an energy consumption, is a third preset weight coefficient, is a collision frequency penalty, and n is a dimension of an environment parameter.

[0083] In an embodiment of the present application, the adaptive weight is determined by the following formula:

[0084]

[0085] In the formula, is a normalization function, is a deep reinforcement learning network, is a current environment parameter, is an updated memory.

[0086] In an embodiment of the present application, the updated memory is determined by the following formula:

[0087]

[0088] In the formula, is a long short-term memory network, is a memory at a previous time.

[0089] In an embodiment of the present application, the system further comprises a fifth data processing unit, which is configured to perform the following operations:

[0090] When the number of adjacent grid blocks with the comprehensive risk coefficient less than the second preset safety threshold is less than one, the underwater robot is controlled to return along the original route.

[0091] According to another aspect, embodiments also provide a computer readable storage medium having stored thereon a computer program which, when executed in a computer, causes the computer to perform the method described in conjunction with Figure 1 the above description.

[0092] According to an embodiment of still another aspect, an electronic device is also provided, including a memory having executable code stored therein and a processor that, when executing the executable code, implements a method as described above. Figure 1 the method.

[0093] Each of the embodiments of the present application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, the system embodiments are described more simply because they are basically similar to the method embodiments, and the relevant parts can be referred to the description of the method embodiments.

[0094] Those skilled in the art should appreciate that the functions described in the above one or more examples can be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, the functions can be stored in or transmitted as one or more instructions or code on a computer-readable medium.

[0095] The above detailed description merely describes the specific implementation of the present application. It is to be understood that the above description is only a specific implementation of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solutions of the present application shall be included in the protection scope of the present application.

Claims

1. A method of controlling a work process of an underwater robot, characterized in that, The method comprises: Step S100: acquiring a three-dimensional grid map of a region to be rescued; wherein the three-dimensional grid map comprises a plurality of grid blocks, each of which corresponds to a working path point and a preset collision risk coefficient; Step S102: acquiring a current working path point of an underwater robot in the three-dimensional grid map and an environmental parameter of a grid block adjacent to a position of the underwater robot; wherein the environmental parameter comprises water turbidity, water salinity, water flow speed 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 parameter, the three-dimensional grid map and the current working path point; Step S106: cyclically executing steps S102 to S104 to obtain a plurality of working path points; Step S108: determining a working path of the underwater robot based on the plurality of working path points; Determining a working path point corresponding to a next grid block of the underwater robot based on the environmental parameter, the three-dimensional grid map and the current working path point comprises: Determining a preset collision risk coefficient of a grid block adjacent to the current working path point based on the three-dimensional grid map and the current working path point; Determining a comprehensive risk coefficient of the adjacent grid block according to the environmental parameter and the corresponding preset collision risk coefficient; Screening a comprehensive risk coefficient smaller than a first preset safety threshold from the comprehensive risk coefficients of the adjacent grid blocks to obtain an adjacent grid block with a comprehensive risk coefficient smaller than the first preset safety threshold; Selecting, from the adjacent grid blocks with a comprehensive risk coefficient smaller than the first preset safety threshold, a grid block with a minimum distance from a preset rescue point as the working path point; The comprehensive risk coefficient is determined by the following formula: where t is a time instant, is the comprehensive risk coefficient, 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 the grid i, is the first preset turbidity threshold, is the second preset turbidity threshold, is the water salinity mapping, is the water salinity of the grid i, is the mean of salinity, is the standard deviation of salinity, is the water flow velocity mapping, is the water flow velocity of the grid 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 target moving direction of the robot, 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, is the energy consumption, is the third preset weight coefficient, is the collision frequency penalty, and n is the dimension of the environmental parameters.

2. The method of claim 1, wherein, The adaptive weight is determined by the following formula: wherein is a normalization function, is a deep reinforcement learning network, is a current environment parameter, is an updated memory.

3. The method of claim 2, wherein, The updated memory is determined by the following formula: In the formula, is a long short-term memory network, is a memory of the previous time.

4. The method of claim 1, wherein, Further comprising: When the number of the adjacent grid blocks with a comprehensive risk coefficient smaller than the first preset safety threshold is less than one, screening a comprehensive risk coefficient smaller than a second preset safety threshold from the comprehensive risk coefficients of the adjacent grid blocks to obtain an adjacent grid block with a comprehensive risk coefficient smaller than the second preset safety threshold; wherein the second preset safety threshold is greater than the first preset safety threshold; Ordering the adjacent grid blocks with a comprehensive risk coefficient smaller than the second preset safety threshold from small to large to determine a risk priority list; Selecting, from the risk priority list, an adjacent grid block with a minimum comprehensive risk coefficient as the working path point.

5. The method of claim 4, wherein, Further comprising: When the number of the adjacent grid blocks with a comprehensive risk coefficient smaller than the second preset safety threshold is less than one, returning the underwater robot to the original path.

6. A control system for underwater robotic work processes, characterized in that, Comprise: A first acquisition unit configured to acquire a three-dimensional grid map of a region to be rescued in step S100; wherein the three-dimensional grid map comprises a plurality of grid blocks, each of which corresponds to a working path point and a preset collision risk coefficient; The second acquisition unit is configured to step S102: acquire an environmental parameter of a current working path point of the underwater robot in the three-dimensional grid map and an adjacent position grid block of the underwater robot; wherein the environmental parameter comprises water turbidity, water salinity, water flow velocity and water flow direction; The first data processing unit is configured to step S104: determine a working path point corresponding to a next grid block of the underwater robot based on the environmental parameter, the three-dimensional grid map and the current working path point; The second data processing unit is configured to step S106: cyclically execute steps S102 to S104 to obtain a plurality of working path points; The third data processing unit is configured to step S108: determine a working path of the underwater robot based on the plurality of working path points; The first data processing unit 204 is configured to perform the following operations: determine a preset collision risk coefficient of an adjacent grid block based on the three-dimensional grid map and the current working path point; determine a comprehensive risk coefficient of the adjacent grid block according to the environmental parameter and the corresponding preset collision risk coefficient; screen the comprehensive risk coefficients of the adjacent grid blocks which are less than the first preset safety threshold to obtain the adjacent grid blocks with the comprehensive risk coefficients less than the first preset safety threshold; select a grid block with the minimum distance to a preset rescue point from the adjacent grid blocks with the comprehensive risk coefficients less than the first preset safety threshold as the working path point; The comprehensive risk coefficient is determined by the following formula: where t is a time instant, is the comprehensive risk coefficient, 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 the grid i, is the first preset turbidity threshold, is the second preset turbidity threshold, is the water salinity mapping, is the water salinity of the grid i, is the mean of salinity, is the standard deviation of salinity, is the water flow velocity mapping, is the water flow velocity of the grid 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 target moving direction of the robot, 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, is the energy consumption, is the third preset weight coefficient, is the collision frequency penalty, and n is the dimension of the environmental parameters.

7. An electronic device, comprising: The device comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the method in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer program is stored on the computer and makes the computer execute the method in any one of claims 1-5 when the computer program is executed in the computer.

Images