Cloud side end collaborative amphibious robot system and collaborative operation method thereof

Abstract

The invention relates to the technical field of marine ranch intelligent operation and maintenance and robots, and discloses a cloud side end collaborative amphibious robot system and a collaborative operation method thereof. The system comprises an amphibious robot, an amphibious robot intelligent docking station and a cloud service platform comprising a central cloud and an edge cloud. The method comprises the steps that the robot collects and preprocesses data and returns the data to the docking station; the docking station carries out local identification and decision making and controls the robot to execute first physical operation; the complex task is reported to the edge cloud for enhancement processing, and a cooperative instruction is generated to control the robot to execute a second physical operation; and the central cloud carries out global model training and issuing. According to the invention, cross-medium continuous autonomous operation of single equipment on the shore base, the water surface and the underwater of the marine ranching is realized, the problems of insufficient end-side computing power and cloud dependence are solved through cloud edge-end cooperation, and the operation and maintenance intelligence level and the response efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance and robotics technology for marine ranches, specifically to an amphibious robot system with cloud-edge-device collaboration and its collaborative operation method. Background Technology

[0002] As a complex production system integrating ecological aquaculture, environmental monitoring and facility maintenance, marine ranches require daily operation and maintenance to span different spaces such as shore-based land, water surface areas and seabed layers. This places higher demands on the comprehensive capabilities of intelligent equipment, such as operational continuity, cross-media mobility, rapid response, intelligent decision-making and remote efficient collaboration, in order to support the long-term, unmanned, closed-loop autonomous operation and maintenance of marine ranches.

[0003] However, existing marine ranch monitoring and maintenance generally rely on dedicated platforms with limited functionality: underwater robots can only perform fixed-depth cruising and limited sensing, unable to go ashore or maneuver on the surface; unmanned surface vessels are limited to surface inspections or observation of the perimeter of net cages, lacking diving capabilities; and shore-based mobile equipment is entirely located outside the waterway. To compensate for these capability gaps, some solutions assemble multiple modules: integrating tracked or wheeled chassis, propellers, and vertical channel propulsion, and adding independent computing units or shore-based GPU servers for data processing. Such methods achieve preliminary automation in localized scenarios, replacing some manual operations. However, this solution still suffers from the following prominent problems:

[0004] 1. Lack of amphibious autonomous operation capability: Existing robots cannot seamlessly switch from shore-based deployment and surface inspection to underwater operations within the same mission cycle. Ranches still need to use multiple sets of equipment, including shore-based tracked vehicles, surface unmanned vessels, and underwater robots. Scheduling and coordination rely on manual labor, which not only significantly increases operation and maintenance costs but also leads to complex task scheduling and low collaboration efficiency due to the independent operation of multiple systems.

[0005] 2. Lack of cloud-edge-device collaborative architecture: Current systems generally adopt two modes: "independent operation on the device side" or "pure cloud processing". The device side is limited by the computing power of the embedded platform and cannot deploy complex AI models such as fish swarm analysis and disease identification. While cloud processing has powerful computing power, it is highly dependent on stable communication links and is prone to robot malfunction due to network interruption in complex marine environments.

[0006] Therefore, this invention proposes an amphibious robot system with cloud-edge-device collaboration and its collaborative operation method to solve the above problems. Summary of the Invention

[0007] This invention aims to overcome the shortcomings of existing technologies and provide a cloud-edge-device collaborative amphibious robot system and its collaborative operation method to achieve autonomous and intelligent operation and maintenance of single devices, multiple tasks, and cross-media in marine ranches.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] In a first aspect, the present invention provides a cloud-edge-device collaborative amphibious robot system, comprising:

[0010] Amphibious robots are used to perform inspection and operation tasks in land, water, and underwater environments, and to collect environmental and equipment status data.

[0011] An amphibious robot smart docking station is connected to the amphibious robot and is used to automatically deploy and retrieve the amphibious robot, and to perform local preprocessing and intelligent analysis on the data transmitted back by the amphibious robot.

[0012] A cloud service platform is connected to the amphibious robot's intelligent docking station. The cloud service platform includes a central cloud platform and a cluster of edge cloud platforms.

[0013] Furthermore, the amphibious robot includes:

[0014] Integrated propeller and wheel, used to provide power output for both land and water scenarios;

[0015] The environmental monitoring module is equipped with a multi-source sensor array for collecting water quality, climate, image, and video data;

[0016] The pose control and navigation module is used to realize autonomous navigation and motion control;

[0017] A real-time communication module is used for data exchange with the amphibious robot's intelligent docking station;

[0018] The power management module is used to manage the power supply and energy consumption analysis of the amphibious robot.

[0019] Reserved expansion space for connecting functional expansion modules.

[0020] Furthermore, the integrated propeller adopts a propeller-leg composite mechanism and is driven by a brushless DC motor to achieve switching between underwater propulsion and land walking.

[0021] Furthermore, the amphibious robot intelligent docking station includes:

[0022] An amphibious robot lifting platform is used for the automatic deployment and retrieval of the amphibious robot.

[0023] The power management module is used to charge the amphibious robot and has an autonomous energy storage function;

[0024] The edge computing terminal is pre-deployed with a lightweight artificial intelligence model, which is used to perform local preprocessing, feature extraction and intelligent classification judgment on the data transmitted back by the amphibious robot.

[0025] Furthermore, the edge computing terminal is also used to maintain basic intelligent decision-making functions when communication is interrupted, and to upload processed structured data to the cloud service platform.

[0026] Furthermore, the central cloud platform is used to uniformly manage the edge cloud platform group, perform high-precision model training based on the global dataset, and distribute the optimized model.

[0027] The edge cloud platform in the edge cloud platform group is deployed in the regional data center and is used to perform semantic enhancement and structured processing on the data streams uploaded by the amphibious robot smart docking stations within the jurisdiction.

[0028] This invention also provides a cloud-edge-device collaborative operation method for amphibious robots, characterized by being executed collaboratively by an amphibious robot, an amphibious robot intelligent docking station, and a cloud service platform. The method includes the following steps:

[0029] S1: The amphibious robot collects raw environmental data and equipment pose data, and after filtering, generates reliable working data and transmits it back to the amphibious robot's smart dock station.

[0030] S2: The amphibious robot intelligent docking station calls a pre-deployed lightweight artificial intelligence model to identify the reliable working data; if the identification result meets the preset local event conditions, a first control command is generated and sent to the amphibious robot, which then performs the first physical operation action.

[0031] S3: If the recognition result meets the preset reporting event conditions, the data is forwarded to the edge cloud platform in the area. The edge cloud platform performs semantic enhancement and structured processing. If the global coordination conditions are met, a second control command is generated and issued, and the amphibious robot performs the second physical operation.

[0032] S4: The central cloud platform trains a high-precision model based on a global dataset and then distributes the optimized lightweight inference model to the edge cloud platform and the amphibious robot smart docking station via over-the-air download.

[0033] Furthermore, in step S2, the preset local event conditions include at least one of identifying netting damage, water quality parameters exceeding thresholds, and detecting abnormal fish aggregation; the first physical operation includes navigating to the event coordinates to acquire close-up images, or adjusting sensors to perform fixed-point continuous monitoring of the abnormal area.

[0034] Furthermore, in step S3, the preset global collaborative conditions include identifying at least one of the following: large-scale algal bloom, pollutant diffusion across multiple aquaculture areas, and abnormal stress in the cage structure; the second physical operation includes performing a surrounding inspection in collaboration with other robots according to collaborative operation instructions, or calling upon its installed functional expansion modules to perform cleaning or auxiliary operations.

[0035] Furthermore, in step S4, the central cloud platform distributes the lightweight inference model to the edge cloud platform and the amphibious robot smart docking station via over-the-air (OTA) download.

[0036] Compared with the prior art, the advantages of this invention are as follows:

[0037] (1) By integrating an amphibious robot with a single propeller, seamless switching and continuous operation of a single device on the shore, on the water surface and underwater are realized, fundamentally solving the problems of high operation and maintenance costs and complex scheduling caused by "multiple sets of equipment + manual connection" in the existing technology.

[0038] (2) By constructing a three-level collaborative architecture of “device end (robot) - edge end (smart dock station) - cloud end (edge ​​cloud + central cloud)”, computing tasks are rationally allocated. The edge end performs real-time data preprocessing and rapid decision-making, solving the problem of insufficient computing power on the edge side; the cloud end performs complex model training and global optimization, and pushes the capabilities down to the edge, reducing the absolute dependence on stable communication links and improving the robustness and intelligent decision-making efficiency of the system in the complex marine environment.

[0039] (3) Through modular design and reserved expansion space, the system can be flexibly equipped with various functional modules (such as cleaning and rescue devices), realizing the upgrade from "passive monitoring" to "monitoring + active disposal", and comprehensively improving the automation and intelligence level of marine ranch operation and maintenance. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments 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, wherein:

[0041] Figure 1 This is a block diagram of the overall structure of the system of the present invention.

[0042] Figure 2 This is a block diagram of the subsystem structure of the amphibious robot of the present invention.

[0043] Figure 3This is a block diagram of the subsystem structure of the amphibious robot intelligent docking station of the present invention.

[0044] Figure 4 This is a block diagram of the subsystem structure of the cloud service platform of the present invention.

[0045] Figure 5 This is a flowchart of the collaborative operation method of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] The terms "first," "second," etc., used in this specification are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, the first object can be one or more. Furthermore, in the specification, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0048] like Figure 1 As shown, the cloud-edge-device collaborative amphibious robot system provided by this invention mainly includes a device end, an edge end, and a cloud service end. The device end, namely the amphibious robot 11, serves as the execution entity, responsible for maneuvering, data collection, and operations in the marine ranch's shore-based, surface, and underwater spaces. The edge end, namely the amphibious robot intelligent docking station 12, serves as the robot's local base and intelligent hub, responsible for the robot's deployment and retrieval, energy replenishment, and local data processing. The cloud service end, namely the cloud service platform 13, consists of a central cloud platform 1301 and an edge cloud platform cluster 1302, responsible for global data aggregation, complex model training, and intelligent collaborative scheduling. The three components are connected via wired or wireless networks, forming a hierarchical, collaborative, and elastically distributed autonomous operating system.

[0049] like Figure 2 As shown, the amphibious robot 11 specifically includes the following modules:

[0050] The integrated propeller 1101 provides power output for both water and land scenarios. It utilizes a high-thrust brushless DC motor as its core power source, coupled with a propeller-leg hybrid mechanism: the brushless DC motor drives the propeller blades to push the tubes in underwater scenarios, while the outer wheels provide movement on land or in shallow water, meeting the stable navigation requirements of different water and land environments. The thrust can be dynamically adjusted according to operational needs. This design allows a single power mechanism to adapt to two media, effectively reducing system complexity and weight. It is key to achieving continuous cross-media operation and directly solves the mechanical challenge of robot water-land mobility switching.

[0051] Real-time communication module 1102: preferably using fiber optic micro-cable connection to ensure high-speed, low-latency, and highly reliable data exchange between amphibious robot 11 and amphibious robot smart dock station 12, providing a foundation for real-time control and high-definition video transmission.

[0052] Environmental monitoring module 1103: Equipped with a multi-source sensor array, it can collect environmental data around the amphibious robot 11. For example, the environmental monitoring module 1103 integrates a high-precision water quality sensor and a micro-climate sensor, which can simultaneously collect multiple key indicators such as water pH, dissolved oxygen content, turbidity, and water temperature. It uses a high-definition camera to capture the swimming trajectory, aggregation density, feeding status, and algae coverage and growth pattern of fish. Simultaneously, it switches to panoramic shooting mode to conduct a comprehensive inspection of the outside of the cage, focusing on capturing details such as the integrity of the cage frame structure, the degree of damage to the netting, and the sealing status of the interfaces. All data, after filtering, is transmitted back to the amphibious robot intelligent docking station 12 via the real-time communication module 1102.

[0053] The pose control and navigation module 1104 is used to regulate the motion posture, spatial position, and operational route of the amphibious robot 11. In autonomous inspection mode, the amphibious robot 11 can receive preset waypoints, operational area coordinates, and task instructions from the amphibious robot intelligent dock station 12 in real time through the real-time communication module 1102. Simultaneously, it collects equipment status data and environmental data from the attitude sensor, Beidou positioning module, power management module 1105, and environmental monitoring module 1103. The onboard autonomous navigation system performs real-time data fusion processing to precisely control the amphibious robot's attitude and speed. Attitude control is achieved by adjusting the output differential of the integrated propeller 1101, while speed control uses PWM technology. Meanwhile, the amphibious robot 11 continuously transmits key pose status information back to the operator. This transmitted data is dynamically displayed on the interface of the edge computing terminal 1203, forming a visualized trajectory and attitude model for panoramic monitoring. Furthermore, the edge computing terminal 1203 can perform real-time analysis of the pose data stream to predict abnormal drift or path deviation.

[0054] The power management module 1105 ensures the amphibious robot is ready for operation at any time. This includes, but is not limited to, a battery pack, a battery pack monitor, and an energy consumption analysis unit. The battery pack is charged by the battery management module 1202 deployed in the amphibious robot's smart docking station 12, which connects to a 220V power supply to provide sufficient power to the amphibious robot for normal operation. The battery pack monitor tracks the remaining battery power, while the energy consumption analysis unit analyzes various power consumption parameters of the amphibious robot 11 during its operation. When the amphibious robot 11 completes its task or its battery is low, it autonomously navigates back to the amphibious robot's smart docking station 12 for charging. Once the battery is fully charged, it automatically cuts off power and enters standby mode, protecting battery life and saving energy. Simultaneously, the amphibious robot 11's power information is fed back to the edge computing terminal 1203 in real-time via the real-time communication module 1102 for remote monitoring by staff.

[0055] 1106 Reserved for expandable space: Adopting a modular design, it can flexibly connect to various functional expansion modules according to the actual needs of aquaculture scenarios, thereby expanding the system functions from basic monitoring to an integrated "monitoring + active disposal" mode. The expansion modules include, but are not limited to: adding pollutant cleaning devices to realize marine debris recycling and treatment; adding rescue modules, configuring rescue equipment such as first aid kits and winches, and building a rapid response network for maritime emergency rescue.

[0056] In one specific embodiment, the reserved expandable space 1106 adopts standardized mechanical and electrical interfaces, supporting plug-and-play functional module expansion. Exemplary expansion modules include:

[0057] A robotic arm module for cleaning marine debris entangled in fishing nets;

[0058] Sampling bottle module, used for collecting water or sediment samples at fixed points;

[0059] The emergency rescue kit has a mounting module that includes equipment such as buoys and rescue ropes.

[0060] like Figure 3 As shown, the amphibious robot intelligent docking station 12 specifically includes:

[0061] Amphibious robot lifting platform 1201: Used to support the deployment and retrieval of amphibious robot 11. For example, when the amphibious robot lifting platform 1201 receives a task, the amphibious robot 11 will autonomously be deployed underwater via the amphibious robot lifting platform 1201 to perform the task; when it receives a return command from the amphibious robot 11 indicating task completion or low battery, the amphibious robot lifting platform 1201 automatically descends into the water, waits for the amphibious robot 11 to enter, switches to land-walking mode, and lifts the amphibious robot 11 back to its position. This achieves unmanned automatic deployment and retrieval of the robot, supporting long-term stationary operations.

[0062] Power management module 1202: responsible for safe and energy-saving power management. For example, its intelligent charging pile is used to charge the amphibious robot 11, while monitoring the charging status of the amphibious robot 11. It will automatically perform a power-off operation when the battery is fully charged or when a work instruction is received. It has an autonomous energy storage device and can use photovoltaic power generation technology and wind power generation technology to charge the backup power supply to ensure power supply in emergency situations such as grid disconnection.

[0063] Edge computing terminal 1203: Pre-deploys a lightweight artificial intelligence model to perform local preprocessing and intelligent classification judgment on the raw data transmitted back by the amphibious robot 11. For example, for simple tasks or urgent tasks with high time requirements (including but not limited to water pollution alarms, netting damage, and emergency oxygenation for low oxygen), edge computing terminal 1203 can call the pre-deployed lightweight artificial intelligence model to generate rapid decision-making instructions and simultaneously display the working data and processing results of the amphibious robot 11 on the screen for operation and maintenance personnel to configure, monitor, and remotely take over. At the same time, the cleaned, fused, and structured key data is uploaded to the cloud service platform 13 via the network for the upper-layer system to perform global analysis and model training. This design completes a large number of real-time, simple intelligent judgments and responses locally, significantly reducing dependence on cloud communication and data transmission latency, and improving emergency response speed.

[0064] It should be further explained that the "lightweight artificial intelligence model" refers to a machine learning model optimized to adapt to the limited computing power of edge computing terminals. Typical implementations include, but are not limited to, convolutional neural networks (CNNs), lightweight vision transformers (ViT-Tiny), or MobileNet series models that have undergone pruning, quantization, or knowledge distillation, for completing image recognition, anomaly detection, or classification tasks locally.

[0065] like Figure 4 As shown, the cloud service platform 13 specifically includes:

[0066] Central Cloud Platform 1301: Unifies and manages multiple edge cloud platforms nationwide (collectively referred to as Edge Cloud Platform Group 1302), primarily undertaking four core functions: centralized management and control of devices and nodes, remote operation and maintenance auditing, big data-driven model training, and collaborative distribution of AI capabilities. Specifically, the central cloud platform relies on a high-performance GPU cluster to perform high-precision machine learning and deep learning training based on a comprehensive historical dataset, constructing computational models for complex scenarios such as marine ranching yield prediction, ecological evolution simulation, and disaster early warning. After training, the platform generates lightweight inference models through model pruning, quantization, and knowledge distillation techniques, and pushes them to each edge cloud platform via OTA (Over-The-Air) updates, achieving continuous iteration and collaborative evolution of the entire system's AI capabilities. This continuous iteration and optimization of the entire system's AI capabilities ensures that the intelligence level of the edge and terminals continuously improves with data accumulation.

[0067] In one specific embodiment, the central cloud platform 1301 sends model update files to the edge cloud platform and the amphibious robot smart docking station via over-the-air download technology. Specifically, it adopts a differential upgrade protocol to reduce transmission load and supports breakpoint resume and version rollback mechanisms to ensure that model iteration can still be reliably completed in the unstable network environment of the ocean.

[0068] Edge cloud platform cluster 1302: Composed of multiple edge cloud platforms deployed in regional data centers in different sea areas. Each edge cloud platform is responsible for data aggregation, intelligent enhancement, and collaborative scheduling of one or more marine ranches within its jurisdiction. The edge cloud platforms can synchronize data and coordinate commands through the central cloud platform (1301). This platform integrates an expert system for the marine aquaculture field. Its knowledge base incorporates local hydrological characteristics, aquaculture species habits, historical disease cases, and operational experience rules. This knowledge base is used to semantically enhance, logically verify, and classify event types of preliminary inference results at the edge, reporting structured data to the central cloud platform 1301 and simultaneously returning the calculation results to the edge computing terminal 1203 in the amphibious robot intelligent dock station 12. Optionally, the edge cloud platform can utilize newly collected data to fine-tune and incrementally learn the existing expert system, achieving self-training. By introducing domain knowledge, the rationality of decision-making and global collaborative capabilities are improved.

[0069] It should be further explained that "semantic enhancement" refers to using a domain knowledge base (such as local hydrological features, aquaculture rules, and historical event database) to supplement the original recognition results with context, classify events, and perform logical verification; "structured processing" refers to organizing the enhanced information into a machine-readable standard format according to a preset event template (such as time, location, type, confidence level, and suggested action), which is convenient for further analysis and storage by the upper-level system.

[0070] like Figure 5As shown, the specific process of the collaborative operation method is as follows:

[0071] Step S1 (Data Acquisition and Primary Filtering): During inspection, the environmental monitoring module 1103 of the amphibious robot 11 continuously collects raw data (which may contain noise). The pose control and navigation module 1104 collects pose data. The embedded processor at the device end runs a Kalman filter or median filter algorithm to obtain reliable data, which is then transmitted back to the amphibious robot's intelligent docking station 12. This step reduces data redundancy and noise at the source, lowering the communication load.

[0072] Step S2 (Edge Intelligent Recognition and Rapid Response): The edge computing terminal 1203 receives data and calls the lightweight model for recognition. If the recognition result meets the preset local event conditions, a first control command is generated, which reads "Navigate to the coordinates and perform close-up photography." After receiving the command, the robot's pose control and navigation module 1104 plans the path and controls the integrated propeller 1101 to perform the action. This "close-up image acquisition" action provides first-hand high-definition detailed images for solving the specific maintenance problem of accurate positioning and assessment of mesh damage. This step achieves second-level perception and response to local emergencies.

[0073] Step S3 (Cloud-based Enhanced Analysis and Collaborative Scheduling): If the identification result meets the reporting event conditions (e.g., identifying a "suspicious large-area algal bloom," but requiring historical data comparison for confirmation), the data packet is forwarded to the edge cloud platform. The expert system on the edge cloud platform combines local hydrological historical data for reasoning. If the global collaborative conditions are met (confirmed as a cross-regional algal bloom), a second control command is generated, such as "directing robots A and B to conduct encirclement-style cruise sampling of the algal bloom area." The command is relayed to the relevant robots via the amphibious robot intelligent dock station 12 for execution. This "collaborative encirclement-style cruise sampling" action provides systematic data support for solving the global operational challenge of accurately assessing the scope and spread trend of algal blooms. Simultaneously, the edge cloud platform uploads a structured "algal bloom event report" to the central cloud platform 1301. This step handles complex tasks requiring global knowledge, enabling collaborative operations among multiple agents.

[0074] Step S4 (Global Model Optimization and Capability Deployment): The central cloud platform 1301 aggregates event reports from various edge clouds to form a global dataset. A high-precision algal bloom prediction model is trained using this dataset. After training, a lightweight version is obtained through model pruning and quantization, and deployed via OTA to the edge cloud platform 1302 and the edge computing terminal 1203 of the amphibious robot intelligent docking station 12, updating their original models. This step enables the system to learn patterns from historical algal bloom events, thereby predicting similar risks earlier and more accurately in the future, directly improving preventative maintenance capabilities. This step allows the entire system's intelligent decision-making capabilities to continuously evolve, becoming "smarter" with use.

[0075] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A cloud-edge-device collaborative amphibious robot system, characterized in that, include: Amphibious robot (11) is used to perform inspection and operation tasks in land, water and underwater environments and collect environmental and equipment status data; An amphibious robot smart docking station (12) is connected to the amphibious robot (11) and is used to automatically deploy and retrieve the amphibious robot (11) and perform local preprocessing and intelligent analysis on the data transmitted back by the amphibious robot (11). The cloud service platform (13) is connected to the amphibious robot smart docking station (12), and the cloud service platform (13) includes a central cloud platform (1301) and an edge cloud platform group (1302).

2. The cloud-edge-device collaborative amphibious robot system according to claim 1, characterized in that, The amphibious robot (11) includes: An integrated propeller (1101) is used to provide power output for both land and water scenarios; The environmental monitoring module (1103) is equipped with a multi-source sensor array for collecting water quality, climate, image and video data; The pose control and navigation module (1104) is used to realize autonomous navigation and motion control; The real-time communication module (1102) is used to exchange data with the amphibious robot smart docking station (12); The power management module (1105) is used to manage the power supply and energy consumption analysis of the amphibious robot (11); Reserved expansion space (1106) for connecting functional expansion modules.

3. The cloud-edge-device collaborative amphibious robot system according to claim 2, characterized in that, The integrated propeller (1101) adopts a propeller-leg composite mechanism and is driven by a brushless DC motor to achieve the switching between underwater propulsion and land walking.

4. The cloud-edge-device collaborative amphibious robot system according to claim 1, characterized in that, The amphibious robot smart docking station (12) includes: An amphibious robot lifting platform (1201) is used for automatically deploying and retrieving the amphibious robot (11). The power management module (1202) is used to charge the amphibious robot (11) and has an autonomous energy storage function; The edge computing terminal (1203) is pre-deployed with a lightweight artificial intelligence model, which is used to perform local preprocessing, feature extraction and intelligent classification judgment on the data transmitted back by the amphibious robot (11).

5. The cloud-edge-device collaborative amphibious robot system according to claim 4, characterized in that, The edge computing terminal (1203) is also used to maintain basic intelligent decision-making functions when communication is interrupted, and to upload processed structured data to the cloud service platform (13).

6. The cloud-edge-device collaborative amphibious robot system according to claim 1, characterized in that, The central cloud platform (1301) is used to uniformly manage the edge cloud platform group (1302), perform high-precision model training based on the global dataset, and distribute the optimized model. The edge cloud platform in the edge cloud platform group (1302) is deployed in the regional data center and is used to perform semantic enhancement and structuring processing on the data stream uploaded by the amphibious robot smart dock station (12) within the jurisdiction.

7. A cloud-edge-device collaborative amphibious robot operation method based on the cloud-edge-device collaborative amphibious robot system according to any one of claims 1-6, characterized in that, The method, executed collaboratively by an amphibious robot (11), an amphibious robot smart docking station (12), and a cloud service platform (13), includes the following steps: S1: The amphibious robot (11) collects raw environmental data and equipment pose data, generates reliable working data after filtering, and sends it back to the amphibious robot smart dock station (12). S2: The amphibious robot smart docking station (12) calls the pre-deployed lightweight artificial intelligence model to identify the reliable working data; if the identification result meets the preset local event conditions, a first control command is generated and sent to the amphibious robot (11) to perform the first physical operation action; S3: If the recognition result meets the preset reporting event conditions, the data will be forwarded to the edge cloud platform in the area. The edge cloud platform will perform semantic enhancement and structured processing. If the global collaboration conditions are met, a second control command will be generated and issued, and the amphibious robot (11) will perform the second physical operation action. S4: The central cloud platform (1301) trains a high-precision model based on the global dataset and distributes the optimized lightweight inference model to the edge cloud platform and the amphibious robot smart docking station (12) via over-the-air download.

8. The method according to claim 7, characterized in that, In step S2, the preset local event conditions include at least one of the following: identifying netting damage, water quality parameters exceeding thresholds, and detecting abnormal fish aggregation; the first physical operation includes navigating to the event coordinates to acquire close-up images, or adjusting sensors to perform fixed-point continuous monitoring of the abnormal area.

9. The method according to claim 7, characterized in that, In step S3, the preset global collaborative conditions include identifying at least one of the following: large-scale algal bloom, pollutant diffusion across multiple aquaculture areas, and abnormal stress in the cage structure; the second physical operation includes performing a surrounding inspection in collaboration with other robots according to the collaborative operation instructions, or calling upon its added functional expansion modules to perform cleaning or auxiliary operations.

10. The method according to claim 7, characterized in that, In step S4, the central cloud platform (1301) downloads the lightweight inference model to the edge cloud platform and the amphibious robot smart docking station (12) via over-the-air download.

Images