Underwater robot operation mode self-adaptive processing method and device

Abstract

The application discloses a kind of underwater robot operation mode self-adaptive processing method and device, wherein the method comprises: obtaining multi-source environmental parameters based on the sensor array carried by underwater robot;The multi-source environmental parameters include water dynamics parameters, visibility, obstacle density;Using weighted fusion method and / or fuzzy reasoning method, multi-source environmental parameters are used for fusion analysis, to determine the operation mode of underwater robot;The operation mode includes energy-saving mode, high-precision mode, emergency obstacle avoidance mode;Based on the current operation mode of underwater robot, dynamically adjust underwater robot control parameters;The control parameters include at least one kind of motion control class parameters, operation tool control class parameters, sensor sampling frequency.This application can combine real-time environmental parameters, dynamically determine underwater robot operation mode and parameter adaptive adjustment, improve the adaptability of underwater robot in complex underwater environment.

Description

Technical Field

[0001] This invention relates to the field of underwater robot technology, and in particular to an adaptive processing method and apparatus for underwater robot operation modes. Background Technology

[0002] With the deepening of marine resource development and marine scientific research, ROVs (Remote Operated Vehicles) have become important underwater operational tools, widely used in fields such as marine resource exploration, subsea pipeline inspection, and underwater archaeology. When performing tasks in complex and ever-changing underwater environments, ROVs face the influence of various environmental factors such as changing water currents, varying visibility, and uneven obstacle distribution, which places higher demands on their operational capabilities. Currently, ROV operation mainly relies on preset programs or real-time manual remote control, resulting in significant shortcomings in overall system energy efficiency and intelligence.

[0003] Preset programs refer to instructions set by operators before the mission begins, such as fixed routes, robotic arm action sequences, and sensor start / stop logic, based on the environment and mission objectives. The ROV executes these instructions step by step while running underwater, without continuous human intervention. However, in this approach, the ROV lacks the ability to perceive and adjust to changes in the environment. The fixed instructions prevent the system from autonomously correcting its path or optimizing its actions, often resulting in ineffective navigation, repetitive operations, or even mission failure, leading to energy waste.

[0004] While manual remote control can provide temporary intervention, it is limited by high underwater communication latency, low bandwidth, and the operator's limited knowledge of on-site information, often resulting in problems such as delayed response, redundant operation, or over-adjustment, which further exacerbates energy consumption.

[0005] In addition, existing ROVs generally rely on a single sensor (such as a low-resolution camera) for environmental perception, which is insufficient in terms of information dimensions and makes it difficult to support efficient decision-making. The control system mostly adopts fixed parameter configurations, such as constant propulsion power and non-optimal attitude control strategies, which cannot be dynamically adjusted according to mission stage, load status or hydrodynamic conditions, resulting in the thruster operating under high load for a long time and the battery being consumed too quickly.

[0006] Overall, existing ROV solutions not only have poor adaptability to complex underwater environments, but also have low energy efficiency and lack energy-saving optimization mechanisms. Summary of the Invention

[0007] This invention provides an adaptive processing method for underwater robot operation modes, which combines real-time environmental parameters to dynamically determine the underwater robot's operation mode and adaptively adjust parameters, thereby improving the underwater robot's adaptability in complex underwater environments. The method includes: Multi-source environmental parameters are obtained based on a sensor array mounted on an underwater robot; these multi-source environmental parameters include hydrodynamic parameters, visibility, and obstacle density. The underwater robot's operating mode is determined by using a weighted fusion method and / or fuzzy inference method, which utilizes multi-source environmental parameters for fusion analysis. The operating modes include energy-saving mode, high-precision mode, and emergency obstacle avoidance mode. Based on the current operating mode of the underwater robot, the control parameters of the underwater robot are dynamically adjusted; the control parameters include at least one of the following: motion control parameters, tool control parameters, and sensor sampling frequency.

[0008] This invention also provides an underwater robot operation mode adaptive processing device, which combines real-time environmental parameters to dynamically determine the underwater robot's operation mode and adaptively adjust parameters, thereby improving the underwater robot's adaptability in complex underwater environments. The device includes: The data acquisition module is used to obtain multi-source environmental parameters based on the sensor array carried by the underwater robot; the multi-source environmental parameters include hydrodynamic parameters, visibility, and obstacle density. The operation mode determination module is used to determine the operation mode of the underwater robot by using a weighted fusion method and / or fuzzy reasoning method to perform fusion analysis using multi-source environmental parameters; the operation mode includes energy-saving mode, high-precision mode, and emergency obstacle avoidance mode; The dynamic control module is used to dynamically adjust the control parameters of the underwater robot based on the current operation mode of the underwater robot; the control parameters include at least one of the following: motion control parameters, tool control parameters, and sensor sampling frequency.

[0009] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described adaptive processing method for underwater robot operation modes.

[0010] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described adaptive processing method for underwater robot operation modes.

[0011] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described adaptive processing method for underwater robot operation modes.

[0012] In this invention, multi-source environmental parameters, including hydrodynamic parameters, visibility, and obstacle density, are obtained based on the sensor array mounted on the underwater robot. A weighted fusion method and / or fuzzy inference method are used to perform fusion analysis of the multi-source environmental parameters to determine the underwater robot's operating mode. Based on the current operating mode, the underwater robot's control parameters are dynamically adjusted. This invention combines real-time environmental parameters to dynamically determine the underwater robot's operating mode, achieving adaptive adjustment of the underwater robot's control parameters and improving the underwater robot's adaptability in complex underwater environments. Experimental comparisons show that multi-parameter fusion avoids misjudgments from a single sensor, reducing the failure rate to below 5%. Regarding energy consumption optimization, using energy-saving mode can extend the endurance by 20%-30%. In terms of mission reliability, the obstacle avoidance success rate in emergency obstacle avoidance mode is increased to over 95%. Compared to the preset programs or manual intervention methods of existing technologies, this achieves truly intelligent adaptive operation. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating the adaptive processing method for underwater robot operation modes in an embodiment of the present invention. Figure 2 This is a specific example diagram of the underwater robot operation mode adaptive processing method in an embodiment of the present invention; Figure 3 This is another specific example of the adaptive processing method for underwater robot operation mode in the embodiments of the present invention; Figure 4 This is a schematic diagram of the underwater robot operation mode adaptive processing device in an embodiment of the present invention; Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0015] Existing underwater robot control methods still suffer from the following shortcomings: First, current ROV operation modes largely rely on preset programs or manual intervention, making it difficult to adapt in real time to changes in complex underwater environments, such as sudden changes in water flow and visibility. Second, existing technologies often rely on single sensors or simple data fusion methods for environmental perception, failing to comprehensively and accurately perceive complex underwater environments. Third, existing technologies exhibit lag in responding to environmental changes, making it difficult to adjust operation modes in a timely manner. Finally, existing technologies are relatively crude in adjusting control parameters, lacking refined parameter optimization strategies for different operation modes. These problems severely limit the intelligent operation capabilities of underwater robots in complex underwater environments.

[0016] To address the technical problem that existing ROV operation modes rely heavily on preset programs or manual intervention, making it difficult to adapt to complex underwater environments in real time, and to achieve significant improvements in environmental adaptability, energy consumption optimization, and mission reliability, this invention provides an adaptive processing method and apparatus for underwater robot operation modes, specifically a method and apparatus for adaptive operation mode switching of underwater ROVs.

[0017] Figure 1 This is a flowchart illustrating the adaptive processing method for underwater robot operation modes in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes: Step 101: Obtain multi-source environmental parameters based on the sensor array carried by the underwater robot; the multi-source environmental parameters include hydrodynamic parameters, visibility, and obstacle density; Step 102: Using a weighted fusion method and / or fuzzy inference method, multi-source environmental parameters are used for fusion analysis to determine the operation mode of the underwater robot; the operation mode includes energy-saving mode, high-precision mode, and emergency obstacle avoidance mode; Step 103: Based on the current operating mode of the underwater robot, dynamically adjust the control parameters of the underwater robot; the control parameters include at least one of the following: motion control parameters, tool control parameters, and sensor sampling frequency.

[0018] The following describes in detail the adaptive processing method for underwater robot operation modes in an embodiment of the present invention.

[0019] In practice, the method can be deployed on an embedded main control device with a certain computing power on the ROV itself.

[0020] First, multi-source environmental parameters are obtained based on the sensor array carried by the underwater robot. These multi-source environmental parameters include, but are not limited to, hydrodynamic parameters, visibility, and obstacle density. Hydrodynamic parameters are key indicators reflecting the movement state of water bodies, including flow velocity, flow direction, and their spatiotemporal variation characteristics.

[0021] In one embodiment, the hydrodynamic parameters include water flow velocity, which is an important parameter for assessing the dynamic characteristics of the underwater environment and directly affects the navigation stability and energy consumption of the underwater robot; the sensor array includes, but is not limited to, a current meter, an optical camera, and a sonar, which are installed on the underwater robot to collect various parameter data of the underwater environment in real time.

[0022] Furthermore, multi-source environmental parameters obtained based on the sensor array carried by the underwater robot can include: The water flow velocity v (unit: m / s) is obtained by a water flow velocity meter; Underwater environment images are obtained through optical cameras, and visibility s (unit: m) is obtained in real time using a lightweight deep learning model. Point cloud data is obtained through sonar, and obstacle density d (unit: number of obstacles per unit volume) is obtained through real-time analysis of the local density of the point cloud data.

[0023] The process of acquiring underwater environment images through an optical camera and using a lightweight deep learning model to infer visibility s in real time can include: 1) acquiring underwater image sequences; 2) preprocessing the images, including denoising, color correction, and contrast enhancement; and 3) inputting the processed images into a lightweight deep learning model. 4) Deep learning models calculate underwater visibility range in real time based on image features.

[0024] In one embodiment, the lightweight deep learning model can be a lightweight CNN model. This lightweight CNN model uses MobileNet as its backbone architecture. The model takes underwater images as input and extracts visual features strongly related to visibility, such as turbidity features, contrast attenuation, and blur degree, through multiple convolution and pooling operations. Then, it performs regression mapping through fully connected layers to directly output quantifiable underwater visibility level or visibility distance values, thereby realizing online perception of underwater environmental visibility conditions.

[0025] The process involves acquiring point cloud data via sonar and then performing real-time analysis of the local density of this point cloud data to determine the obstacle density (d). This can be achieved by: acquiring real-time 3D point cloud data of the underwater environment using an onboard multibeam imaging sonar; performing noise filtering, outlier removal, and coordinate normalization preprocessing on the original point cloud; and then using a clustering algorithm to group the effective echo point clouds, aggregating point clouds with similar spatial locations and reflection characteristics into independent obstacle clusters. Based on the clustering results, features such as the spatial location, geometric dimensions, distribution dispersion, and relative distance of each obstacle can be extracted, and the number of obstacles per unit volume can be calculated. This enables a quantitative characterization of the spatial distribution pattern, density, and hazard level of underwater obstacles, providing reliable prior environmental information for underwater robots to perceive the complexity of the local environment in real time.

[0026] In step 102, a weighted fusion method and / or fuzzy inference method are used to perform fusion analysis using multi-source environmental parameters to determine the underwater robot's operating mode. In this embodiment of the invention, three operating modes are proposed: energy-saving mode, high-precision mode, and emergency obstacle avoidance mode.

[0027] Energy-saving mode refers to a working state in which the underwater robot's energy consumption is reduced by adjusting its operating parameters, shutting down unnecessary functions, or optimizing energy allocation, thereby extending its working time or reducing energy consumption. High-precision mode, for example, is a working state suitable for complex tasks such as high-precision assembly and micro-operations in an ideal environment. Emergency obstacle avoidance mode refers to a safety protection mechanism triggered in response to sudden obstacles or dangerous situations. Its core objective is to avoid collisions and damage to itself through rapid detection, judgment, and response, ensuring the safe completion of the mission.

[0028] In this embodiment of the invention, the underwater robot's operating mode is determined and adaptively switched by performing a fusion analysis of water flow speed, visibility, obstacle density, etc.

[0029] The weighted fusion method refers to the method of determining the operation mode based on the weighted fusion processing of multi-source environmental parameters, while the fuzzy inference method refers to the method of determining the operation mode based on the fuzzification, rule matching and defuzzification operation of multi-source environmental parameters.

[0030] Figure 2 This is a specific example diagram of the adaptive processing method for underwater robot operation modes in an embodiment of the present invention, as shown below. Figure 2 As shown, a weighted fusion method is used to perform fusion analysis using multi-source environmental parameters to determine the operating mode of the underwater robot, which may include: Step 201: Perform weighted fusion analysis using multi-source environmental parameters to obtain the comprehensive environmental assessment value; the comprehensive environmental assessment value characterizes the operability of the underwater robot in the underwater environment and is used to determine the operating mode of the underwater robot. Step 202: Determine the operating mode of the underwater robot based on the comprehensive environmental assessment value and the preset mode segmentation threshold; the mode segmentation threshold is used to characterize the boundary value of the comprehensive environmental assessment value of different operating modes.

[0031] Specifically, using multi-source environmental parameters for weighted fusion analysis to obtain a comprehensive environmental assessment value may include: preprocessing the multi-source environmental parameters to obtain preprocessed multi-source environmental parameters; the preprocessing includes normalization; and using the preprocessed multi-source environmental parameters for weighted fusion calculation to obtain a comprehensive environmental assessment value.

[0032] For example, based on the ROV equipped with a multimodal sensor array (including but not limited to water flow velocity meter, optical camera, sonar, inertial measurement unit IMU), parameters such as water flow velocity v, visibility s, obstacle density d, and built-in or received task priority t (supports user setting, 1-5 levels, 1 is the lowest and 5 is the highest) can be collected in real time.

[0033] The normalization process for nonlinear parameters (such as visibility) is as follows: ; Among them, S min S max S represents the minimum and maximum values ​​of the preset visibility threshold. norm This is the normalized visibility value.

[0034] Define the comprehensive environmental assessment value E using the weighted summation formula: ; The weight coefficients w1, w2, w3, and w4 were calibrated experimentally (e.g., w1=0.3, w2=0.2, w3=0.3, w4=0.2). The weight coefficients were determined through multi-stage experiments, including environmental data collection, initial selection based on expert experience, optimization through comparative experiments, and reinforcement learning iterations, ultimately verifying their adaptability and robustness.

[0035] Job modes are determined based on the value of E: Mode 1 (Energy Saving Mode): E < θ1 (e.g., mode segmentation threshold θ1 = 1.5); Mode 2 (high precision mode): θ1 ≤ E < θ2 (e.g., mode segmentation threshold θ2 = 3.0); Mode 3 (Emergency Obstacle Avoidance Mode): E ≥ θ2.

[0036] Furthermore, to improve the reliability of the operation mode division, the operation mode of the underwater robot is determined based on the comprehensive environmental assessment value and the preset mode segmentation threshold. This can include: when the difference between the comprehensive environmental assessment value and the preset mode segmentation threshold is less than the set minimum numerical interval, fuzzy inference is used to perform fusion analysis using multi-source environmental parameters to determine the operation mode of the underwater robot.

[0037] When using fuzzy inference, the underwater robot's operating mode is determined by fusion analysis of multi-source environmental parameters. This process may include: fuzzifying the multi-source environmental parameters to obtain the membership degree of each parameter; performing fuzzy inference based on the membership degree of each parameter using a fuzzy rule base; and defuzzifying the fuzzy set output by the fuzzy inference to determine the underwater robot's operating mode. The fuzzy rule base includes the mapping relationship between the membership degree combination of each parameter and the operating mode.

[0038] The following describes a fuzzy inference software processing method deployed on the ROV body in an embodiment of the present invention (which can form a fuzzy inference processing module).

[0039] (1) Fusion of dynamic membership function and environmental perception.

[0040] Water flow velocity (v): The water flow velocity is divided into three fuzzy states: low, medium, and high. Each state corresponds to a different membership value. Triangular membership functions are used in implementation. The low, medium, and high thresholds of different fuzzy states can be dynamically calibrated through real-time water depth data. For example, the "medium" threshold for deep-water operations is increased from 1.5 m / s to 2.0 m / s.

[0041] Visibility(s): Visibility is divided into blurry states such as clear, blurry, and hazy. Based on image processing data, a trapezoidal membership function is generated, and the "hazy" range is dynamically corrected by combining historical visibility statistics. For example, the visibility change trend is predicted by using a deep learning model, so that the judgment is more in line with the actual environment.

[0042] Obstacle density (d): Using an S-shaped membership function, the obstacle distribution characteristics are updated in real time through sonar point cloud data, such as cluster analysis of obstacle spatial distribution.

[0043] (2) Hybrid rule base and adaptive weights.

[0044] For example, the constructed fuzzy rule base is a fuzzy rule base that combines expert experience base and data-driven rules. It dynamically adjusts the weights of the rules based on their successful performance in actual tasks. That is, the rule weight adaptive adjustment dynamically adjusts the rule weights based on the success rate of mode switching, thereby improving the accuracy of job mode judgment.

[0045] (3) Fuzzy reasoning and multi-objective collaboration.

[0046] The inference method adopts Mamdani-type fuzzy inference, which calculates the rule activation intensity through least product operation and generates the output fuzzy set.

[0047] Defuzzification strategy: The centroid method is used to transform the fuzzy set into deterministic operation mode suggestions, where different operation modes are assigned priority weights, for example, the emergency obstacle avoidance mode has a weight of 0.85. The formula is: ; in, Let x be the activation strength of the i-th fuzzy rule, representing the degree of matching of the rule under the current input conditions. i The priority values ​​(weights) for the corresponding modes are x1, x2, and x3, which correspond to the dynamic priority values ​​(weights) of mode 1 (energy saving mode), mode 2 (high precision mode), and mode 3 (emergency obstacle avoidance mode), respectively.

[0048] The priorities of the three task modes are dynamically changed. In this embodiment of the invention, the priorities of the three task modes are dynamically adjusted through environmental parameter fusion, fuzzy inference, and reinforcement learning. Initially, they are set by a weighted evaluation value E, and during runtime, they are adaptively optimized based on real-time data and feedback mechanisms.

[0049] In fuzzy inference and multi-objective collaboration, the feedback mechanism dynamically adjusts priorities based on real-time task data (such as obstacle avoidance success rate and energy efficiency). For example, if the system detects a high failure rate in emergency obstacle avoidance mode, the feedback triggers a weight update in the fuzzy rule base: the rule weights for the "high water flow speed and low visibility" scenario are recalculated using a fuzzy clustering algorithm (FCM, fuzzy C-means clustering algorithm), making fuzzy inference more inclined to output the emergency obstacle avoidance mode (M3), and simultaneously optimizing the membership function parameters (such as expanding the threshold range of "high water flow speed"), thereby improving the priority and decision accuracy of operation mode switching in complex environments.

[0050] Finally, based on the underwater robot's current operating mode, the underwater robot's control parameters are dynamically adjusted. These control parameters include at least one of the following: motion control parameters, tool control parameters, and sensor sampling frequency. Motion control parameters include propulsion power, and tool control parameters include tool load.

[0051] For example, based on the determined work mode, the following parameters are dynamically adjusted: Propulsion power P: ; in, P base The reference power is k, and the water flow velocity compensation coefficient is k. To adjust the amount of power.

[0052] Tool load L (e.g., robotic arm gripping force): ; Among them, L min It is the minimum allowable value of tool load (such as the lower limit of the gripping force of a robotic arm), L norminal This is the standard tool load under normal working conditions (such as the default gripping force of a robotic arm), L max The maximum allowable load on the tool (e.g., the upper limit of the gripping force of a robotic arm), where d is the obstacle density in the current environment. max It is a preset upper limit for obstacle density, used to determine whether the environment is in a high-risk state.

[0053] Sensor sampling frequency f: ; For example, a basic sampling frequency is preset for a specific device in a specific scenario, and the sampling frequency is adjusted by... Used for adjusting the sensor sampling frequency.

[0054] In one embodiment, dynamically adjusting the underwater robot's control parameters based on the underwater robot's current operating mode may include: adaptively optimizing the underwater robot's control parameters using a reinforcement learning algorithm; wherein the current operating mode, environmental parameters, and remaining battery power are used as state inputs, the adjustment amount of control parameters is used as action outputs, and the strategy is iteratively updated using task completion, energy efficiency, and safety factor as reward functions.

[0055] During implementation, reinforcement learning is introduced to optimize long-term performance: State space S: contains current operating mode, environmental parameters, remaining power, etc.

[0056] Action space: the differential increment of the adjustment parameter (e.g.) , ).

[0057] Reward function R: R = α × task completion degree + β × energy efficiency + γ × safety coefficient, where α, β, and γ are weighting coefficients, which are set through experiments (e.g., α = 0.4, β = 0.3, γ = 0.3).

[0058] Using the Q-learning algorithm, ROV updates its strategy after each mode switch to maximize cumulative rewards.

[0059] , The formula is dynamically generated by Q-learning and superimposed onto the baseline formula while retaining the original formula structure. The two work together to achieve adaptive parameter adjustment: mode determination determines the formula's applicable range (e.g., energy-saving, emergency obstacle avoidance modes), while Q-learning... , The parameters were fine-tuned, and the optimization was finally completed using a combination of fixed formulas and dynamic increments.

[0060] Furthermore, fuzzy reasoning and reinforcement learning can be optimized together. The synergy between fuzzy reasoning and reinforcement learning involves using the pattern suggestions (e.g., M3) output by fuzzy reasoning as action space constraints for the reinforcement learning algorithm (e.g., priority exploration). , (Parameter adjustment). Dynamic feedback correction involves feeding back indicators such as task completion, energy efficiency, and safety factor to the fuzzy inference processing module after each mode switch, and then updating the rule base and membership functions using fuzzy clustering algorithms (such as FCM).

[0061] Dynamic parameter adjustments provide immediate responses through the fuzzy inference processing module and achieve long-term optimization through reinforcement learning. The synergy between these two mechanisms enables the ROV to adaptively adjust propulsion power, sensor sampling frequency, and tool load in complex underwater environments, thereby improving mission reliability, reducing energy consumption, and enhancing safety.

[0062] Figure 3 This is another specific example diagram of the adaptive processing method for underwater robot operation modes in the embodiments of the present invention, as shown in the figure. Figure 3 As shown, the method includes: Step 301, Environmental Parameter Acquisition and Preprocessing: Real-time acquisition of water flow velocity, visibility, obstacle density, and task priority using a multimodal sensor array.

[0063] Step 302, Dynamic Environment Assessment and Mode Determination: Calculate the comprehensive environmental assessment value E based on the weighted fusion algorithm. Divide the operation mode into energy-saving mode, high-precision mode, and emergency obstacle avoidance mode according to E and the preset mode segmentation threshold. When the difference between the comprehensive environmental assessment value and the preset mode segmentation threshold is less than the set minimum numerical interval, the fuzzy inference method is used to perform fusion analysis using multi-source environmental parameters to determine the operation mode of the underwater robot.

[0064] Step 303, Parameter Adaptive Adjustment: Based on the determined operating mode, dynamically adjust the propulsion power, sensor sampling frequency, and tool load.

[0065] For example, energy-saving mode:

[0066] High-precision mode:

[0067] Emergency obstacle avoidance mode:

[0068] Where K is the water flow velocity compensation coefficient.

[0069] Step 304, Multi-objective optimization and dynamic feedback: Combining the pattern suggestions from the fuzzy inference processing module, the parameter adjustment amount is dynamically adjusted through reinforcement learning to form a closed-loop optimization.

[0070] The dynamically adjusted parameters include: Propulsion power: In energy-saving mode, the base power is reduced to save energy consumption; in emergency obstacle avoidance mode, the power is increased to enhance mobility. Sensor sampling frequency: Reduce sampling frequency in energy-saving mode to reduce energy consumption, and increase sampling frequency in emergency obstacle avoidance mode to enhance environmental perception accuracy; Tool load: The robotic arm's gripping force is dynamically adjusted according to the obstacle density to ensure mission safety.

[0071] The collaborative relationship between the fuzzy reasoning processing module and reinforcement learning is as follows: Pattern suggestions for fuzzy inference processing modules: as action space constraints for reinforcement learning (e.g., priority exploration) , (adjustment direction) Dynamic feedback mechanism: After each mode switch, the task completion rate, energy efficiency, safety factor and other indicators are fed back to the fuzzy inference processing module. The rule base and membership function are updated using fuzzy clustering algorithm (such as FCM), and the reward function weights (α, β, γ) of reinforcement learning are adjusted at the same time.

[0072] Key parameters (such as k, K, ...) in the adjustment formulas for propulsion power, sensor sampling frequency, and tool load. P base Dynamic optimization is achieved through reinforcement learning; The roles of ΔP and Δf: They serve as parameter adjustment values ​​generated by reinforcement learning, which are then added to the baseline values ​​in the formula to achieve the final parameter adjustment (e.g., P = ...). P base ×1.5+K×v+ΔP).

[0073] Through the above steps, the underwater robot can adaptively adjust its operating mode and control parameters based on multi-source environmental parameters, achieving efficient and safe operation in complex underwater environments. The dynamic adjustment of parameters provides immediate response through a fuzzy inference processing module and achieves long-term optimization through reinforcement learning. The synergy between these two mechanisms enables the underwater robot to adaptively adjust propulsion power, sensor sampling frequency, and tool load in complex underwater environments, thereby improving mission reliability, reducing energy consumption, and enhancing safety.

[0074] This invention also provides an adaptive processing device for underwater robot operation modes, as described in the following embodiments. Since the principle behind this device's problem-solving is similar to that of the adaptive processing method for underwater robot operation modes, its implementation can refer to the implementation of the adaptive processing method for underwater robot operation modes; repeated details will not be elaborated further.

[0075] Figure 4 This is a schematic diagram of the underwater robot operation mode adaptive processing device in an embodiment of the present invention, as shown below. Figure 4 As shown, the device 400 applies the underwater robot operation mode adaptive processing method, and the device 400 includes: The data acquisition module 401 is used to obtain multi-source environmental parameters based on the sensor array carried by the underwater robot; the multi-source environmental parameters include hydrodynamic parameters, visibility, and obstacle density; The operation mode determination module 402 is used to determine the operation mode of the underwater robot by using a weighted fusion method and / or fuzzy reasoning method to perform fusion analysis using multi-source environmental parameters; the operation mode includes energy-saving mode, high-precision mode, and emergency obstacle avoidance mode; The dynamic control module 403 is used to dynamically adjust the control parameters of the underwater robot based on the current operation mode of the underwater robot; the control parameters include at least one of the following: motion control parameters, operation tool control parameters, and sensor sampling frequency.

[0076] In one embodiment, the sensor array includes a water flow velocity meter, an optical camera, and a sonar; the hydrodynamic parameters include water flow velocity. The data acquisition module 401 is specifically used for: The water flow velocity is obtained using a flow velocity meter; Underwater environment images are obtained through optical cameras, and visibility is obtained in real time using a lightweight deep learning model; Point cloud data is obtained through sonar, and obstacle density is obtained through real-time analysis of the local density of the point cloud data.

[0077] In one embodiment, the job mode determination module 402 includes a weighted fusion threshold processing module and a fuzzy inference processing module. The weighted fusion threshold processing module is specifically used for: A weighted fusion analysis of multi-source environmental parameters is used to obtain a comprehensive environmental assessment value. The comprehensive environmental assessment value characterizes the operability of the underwater robot in the underwater environment and is used to determine the operating mode of the underwater robot. The underwater robot's operating mode is determined based on the comprehensive environmental assessment value and a preset mode segmentation threshold; the mode segmentation threshold is used to characterize the boundary value of the comprehensive environmental assessment value for different operating modes.

[0078] In one embodiment, the weighted fusion threshold processing module is specifically used for: The multi-source environmental parameters are preprocessed to obtain preprocessed multi-source environmental parameters; the preprocessing includes normalization. The preprocessed multi-source environmental parameters are used for weighted fusion calculation to obtain the comprehensive environmental assessment value.

[0079] In one embodiment, the operation mode determination module 402 is specifically used for: When the difference between the comprehensive environmental assessment value and the preset pattern segmentation threshold is less than the set minimum numerical interval, the underwater robot's operation mode is determined by using fuzzy inference and multi-source environmental parameters for fusion analysis.

[0080] In one embodiment, the fuzzy inference processing module is specifically used for: The multi-source environmental parameters are fuzzified to obtain the membership degree of each parameter. Fuzzy reasoning is performed based on the membership degree of each parameter using a fuzzy rule base. The fuzzy set output by the fuzzy reasoning is then defuzzified to determine the operation mode of the underwater robot. The fuzzy rule base includes the mapping relationship between the membership degree combination of each parameter and the operation mode.

[0081] In one embodiment, the dynamic control module 403 is specifically used for: A reinforcement learning algorithm is used to adaptively optimize the control parameters of the underwater robot. The current operation mode, environmental parameters, and remaining battery power are used as state inputs, the adjustment of control parameters is used as action outputs, and the reward function is used to iteratively update the strategy.

[0082] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described adaptive processing method for underwater robot operation modes.

[0083] Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention, such as... Figure 5 As shown, this embodiment of the invention also provides a computer device 500, including a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, it implements the above-mentioned adaptive processing method for underwater robot operation mode.

[0084] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described adaptive processing method for underwater robot operation modes.

[0085] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described adaptive processing method for underwater robot operation modes.

[0086] In this embodiment of the invention, the priorities of the three operating modes are dynamically adjusted through environmental parameter fusion, fuzzy inference, and reinforcement learning. Initially, the priority is set by a comprehensive environmental evaluation value, and during runtime, it is adaptively optimized based on real-time data and a feedback mechanism. In fuzzy inference and multi-objective collaboration, the feedback mechanism dynamically adjusts priorities using real-time task data (such as obstacle avoidance success rate and energy efficiency). For example, if the system detects a high obstacle avoidance failure rate in emergency mode, the feedback triggers a weight update in the fuzzy rule base: the rule weights for the "high water flow speed and low visibility" scenario are recalculated using a fuzzy clustering algorithm, making the fuzzy inference system more inclined to output the emergency obstacle avoidance mode, and simultaneously optimizing the membership function parameters. Fuzzy inference and reinforcement learning work together, with the former providing immediate response and the latter achieving long-term optimization. This synergy enables the underwater robot to adaptively adjust control parameters in complex underwater environments, thereby improving task reliability, reducing energy consumption, and enhancing safety.

[0087] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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.

[0088] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] 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.

[0090] 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.

[0091] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An adaptive processing method for underwater robot operation modes, characterized in that, include: Multi-source environmental parameters are obtained based on the sensor array carried by the underwater robot; the multi-source environmental parameters include hydrodynamic parameters, visibility, and obstacle density; The underwater robot's operating mode is determined by using a weighted fusion method and / or fuzzy inference method, which utilizes multi-source environmental parameters for fusion analysis. The operating modes include energy-saving mode, high-precision mode, and emergency obstacle avoidance mode. Based on the current operating mode of the underwater robot, the control parameters of the underwater robot are dynamically adjusted; the control parameters include at least one of the following: motion control parameters, tool control parameters, and sensor sampling frequency.

2. The method as described in claim 1, characterized in that, The sensor array includes a water flow velocity meter, an optical camera, and a sonar; The hydrodynamic parameters include water flow velocity; Multi-source environmental parameters are obtained based on the sensor array carried by the underwater robot, including: The water flow velocity is obtained using a flow velocity meter; Underwater environment images are obtained through optical cameras, and visibility is obtained in real time using a lightweight deep learning model; Point cloud data is obtained through sonar, and obstacle density is obtained through real-time analysis of the local density of the point cloud data.

3. The method as described in claim 1, characterized in that, A weighted fusion method is used to perform fusion analysis of multi-source environmental parameters to determine the operating mode of the underwater robot, including: A weighted fusion analysis of multi-source environmental parameters is used to obtain a comprehensive environmental assessment value. The comprehensive environmental assessment value characterizes the operability of the underwater robot in the underwater environment and is used to determine the operating mode of the underwater robot. The underwater robot's operating mode is determined based on the comprehensive environmental assessment value and a preset mode segmentation threshold; the mode segmentation threshold is used to characterize the boundary value of the comprehensive environmental assessment value for different operating modes.

4. The method as described in claim 3, characterized in that, A weighted fusion analysis of multi-source environmental parameters is used to obtain a comprehensive environmental assessment value, including: The multi-source environmental parameters are preprocessed to obtain preprocessed multi-source environmental parameters; the preprocessing includes normalization. The preprocessed multi-source environmental parameters are used for weighted fusion calculation to obtain the comprehensive environmental assessment value.

5. The method as described in claim 4, characterized in that, Based on the comprehensive environmental assessment values ​​and preset pattern segmentation thresholds, the operating modes of the underwater robot are determined, including: When the difference between the comprehensive environmental assessment value and the preset pattern segmentation threshold is less than the set minimum numerical interval, the underwater robot's operation mode is determined by using fuzzy inference and multi-source environmental parameters for fusion analysis.

6. The method as described in claim 1, characterized in that, The underwater robot's operational mode is determined by using fuzzy reasoning and multi-source environmental parameters for fusion analysis, including: The multi-source environmental parameters are fuzzified to obtain the membership degree of each parameter. Fuzzy reasoning is performed based on the membership degree of each parameter using a fuzzy rule base. The fuzzy set output by the fuzzy reasoning is then defuzzified to determine the operation mode of the underwater robot. The fuzzy rule base includes the mapping relationship between the membership degree combination of each parameter and the operation mode.

7. The method as described in claim 1, characterized in that, Based on the underwater robot's current operating mode, dynamically adjust the underwater robot's control parameters, including: A reinforcement learning algorithm is used to adaptively optimize the control parameters of the underwater robot. The current operation mode, environmental parameters, and remaining battery power are used as state inputs, the adjustment of control parameters is used as action outputs, and the reward function is used to iteratively update the strategy.

8. An adaptive processing device for underwater robot operation modes, characterized in that, include: The data acquisition module is used to obtain multi-source environmental parameters based on the sensor array carried by the underwater robot; the multi-source environmental parameters include hydrodynamic parameters, visibility, and obstacle density; The operation mode determination module is used to determine the operation mode of the underwater robot by using a weighted fusion method and / or fuzzy reasoning method to perform fusion analysis using multi-source environmental parameters; the operation mode includes energy-saving mode, high-precision mode, and emergency obstacle avoidance mode; The dynamic control module is used to dynamically adjust the control parameters of the underwater robot based on the current operation mode of the underwater robot; the control parameters include at least one of the following: motion control parameters, tool control parameters, and sensor sampling frequency.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images