Cluster type water surface cleaning system and method based on dynamic role distribution and front line convergence supply

Abstract

The invention discloses a cluster type water surface cleaning system and method based on dynamic role distribution and front line convergence supply, and relates to the technical field of environmental protection and unmanned systems, the cluster type water surface cleaning system comprises a mother ship, a multifunctional sub-ship and an unmanned aerial vehicle cluster; the mother ship serves as a mobile command and supply center and is integrated with a garbage treatment assembly line; the sub-ship adopts a modular design, and the function can be dynamically reconfigured; the unmanned aerial vehicle cluster has garbage drift prediction and visual servo guidance; the mother ship is introduced to serve as a mobile command and supply center, a three-level intelligent framework is constructed, and the problems that in the prior art, the equipment role is fixed, time consumed for going to and fro a base is long, and the overall efficiency is low are fundamentally solved.

Description

Technical Field

[0001] This invention relates to the fields of environmental protection and unmanned systems technology, and in particular to a clustered water surface cleaning system and method based on dynamic role allocation and frontline rendezvous and replenishment. Background Technology

[0002] With industrialization and urbanization, the problem of floating garbage pollution in rivers, lakes, and coastal waters is becoming increasingly serious. Traditional manual boat-based salvage methods have inherent drawbacks, including low efficiency, dangerous working environments, high labor intensity, and high overall costs.

[0003] To improve cleaning efficiency, automation technology has been introduced into this field. Currently, existing technologies mostly focus on simple collaboration between drones and unmanned surface vessels (USVs). For example, Chinese patent document CN113443087A discloses a cleaning device that uses drones for reconnaissance and transmits location information to USVs for cleaning, while also featuring bidirectional charging to alleviate range anxiety. However, this solution has significant shortcomings: First, its collaboration mode is static and fixed; the drone is only responsible for identification, and the USV is only responsible for cleaning, making dynamic strategy adjustments impossible based on complex waste distribution. Second, when the USV is full or its battery is depleted, it must return to a fixed shore base; this long-distance "empty round trip" significantly reduces effective operating time, becoming a core bottleneck restricting efficiency.

[0004] Another Chinese patent document, CN109178308A, discloses a method for using drones to independently identify, collect, and dump waste onto a collection vessel. While this solution achieves automation, its practicality is limited by the payload capacity of drones, making it difficult to handle large-volume, heavy waste.

[0005] In summary, existing technologies generally suffer from three common problems: "rigid collaboration," "endurance bottlenecks," and "system isolation." Therefore, there is an urgent need in this field for a next-generation water surface cleaning solution capable of intelligent decision-making, long-term endurance, and efficient collaboration. Summary of the Invention

[0006] To address the problems existing in the prior art, the purpose of this invention is to provide a clustered surface clearing system and method based on dynamic role allocation and frontline rendezvous and resupply. This invention introduces a mother ship as a mobile command and resupply hub, constructing a three-level intelligent architecture, which fundamentally solves the problems of fixed equipment roles, long travel time to and from the base, and low overall efficiency in the prior art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is: a cluster-type water surface cleaning system based on dynamic role allocation and front-line rendezvous and resupply, comprising: a mother ship, multi-functional sub-ships, and a cluster of unmanned aerial vehicles (UAVs); the mother ship serves as a mobile command and resupply hub, integrating a waste processing pipeline; the sub-ships adopt a modular design, and their functions can be dynamically reconfigured; the UAV cluster is equipped with waste drift prediction and visual servo guidance.

[0008] Furthermore, the mother ship specifically includes: The central processing unit runs a global dynamic scheduling system to achieve task planning, resource allocation, and path optimization. The waste processing unit integrates a coarse crushing module, a magnetic separation module, a dewatering module, and a compression module connected in sequence, forming a highly efficient waste processing production line; Modular storage bins are used to store compressed, standardized waste blocks; The energy replenishment station, including generator sets, energy storage batteries and wireless charging platforms, is used to replenish energy for sub-boats and drones; Navigation and communication systems are used for high-precision positioning and to maintain stable communication with sub-ships, drones, and shore-based systems.

[0009] Furthermore, the sub-ship specifically includes: The modular mission bay uses standardized mechanical and electrical interfaces for replacing different types of mission payloads; The status monitoring system is used to monitor the waste bin capacity, battery level, and equipment health status in real time. The autonomous docking system is used to achieve precise and safe docking with the mother ship during navigation in order to complete waste disposal and energy replenishment.

[0010] Furthermore, the drone swarm specifically includes: The integrated sensing system, equipped with a visible light camera and a depth sensor, is used for waste identification and three-dimensional positioning; Edge computing units are used to run image recognition algorithms, garbage drift prediction algorithms, and visual servo guidance algorithms; A high-precision positioning module is used to provide its own centimeter-level positioning information.

[0011] This invention also provides a clustered surface clearing method based on dynamic role allocation and frontline rendezvous and resupply, characterized in that it is implemented using the clustered surface clearing system based on dynamic role allocation and frontline rendezvous and resupply as described above, and the method specifically includes the following steps: Step 1: Wide-area reconnaissance and predictive situation map construction: The UAV swarm scans the target water area, identifies garbage and estimates its type and volume; at the same time, by analyzing continuous image frames and fusing environmental wind flow data, the drift trajectory of the garbage is predicted; the mother ship fuses all data to generate a garbage distribution probability density map with spatiotemporal prediction information. Step 2, Dynamic Task Allocation: The mothership broadcasts the cleanup task to all available sub-ships; each sub-ship calculates the cost of completing the task based on its own position, capabilities, role, and status, and reports it back to the mothership; the mothership selects the sub-ship with the lowest cost to execute the task based on the principle of global optimization. Step 3, Adaptive Collaborative Execution: For tasks requiring precise operation, the drone hovers above the garbage and uses a visual servo controller to calculate the image deviation between the robotic arm end and the garbage in real time. The calculated motion control commands are then sent to the sub-boat to drive its robotic arm to complete the precise grasping, forming a closed-loop control system. Step 4, Frontline Rendezvous and Resupply: When the waste bin capacity or battery power of a sub-ship falls below a preset threshold, it sends a resupply request to the mother ship. The mother ship models the resupply requests from multiple sub-ships as a path optimization problem, calculates the optimal patrol path for the mother ship to serve these sub-ships and the corresponding frontline rendezvous point for each sub-ship, and the mother ship and sub-ships move synchronously towards the rendezvous point to complete automatic docking, waste dumping, and energy resupply.

[0012] Furthermore, in step 1, the predicted drift trajectory of the waste is as follows: Record position: Record the current position of the trash in the camera frame (u_now, v_now), and its position t seconds ago (u_old, v_old); Calculate movement speed: Movement speed = ((u_now - u_old) / t, (v_now - v_old) / t); Receive environmental data: Obtain the current wind speed (wind_speed) and wind direction (wind_direction) from the broadcast information sent by the mother ship, and merge them into a wind flow vector W; Comprehensive prediction: When calculating the future direction and size of the waste, the movement of the waste itself and the impact of wind and waves are taken into account. Predicted speed = moving speed + α * W, where α is an empirical coefficient representing the degree of wind's impact on the garbage; Output prediction point: Predicted position = current actual position + prediction velocity * T, where T represents time. The output prediction point is the location of the garbage after T seconds.

[0013] Furthermore, step 2 is specifically as follows: Mothership issues mission: Mothership broadcasts mission T, including the location and type of trash; Sub-ship cost calculation: Each idle sub-ship that receives a task calculates its own cost to complete the task: Cost = Distance to the task point + (100 - Remaining power) * 0.1; Sub-ship bidding: Each sub-ship reports its own number and the cost of completing the task to the mother ship; Mother ship bidding: After collecting all bids, the mother ship selects the subsidiary ship with the lowest reported cost and issues a TDE (Task Execution) order to it; Status Update: The mothership marks the status of the child ships performing the mission as busy.

[0014] Furthermore, step 3 is specifically as follows: Deviation detection: The drone simultaneously identifies the center point (u_g, v_g) of the trash and the center point (u_arm, v_arm) of the robotic arm claw, and calculates the pixel difference between them: Deviation = (u_g - u_arm, v_g - v_arm); Conversion command: Use a preset conversion coefficient β to convert the pixel deviation into the actual distance and direction that the robotic arm should move: Motion command = ((u_g-u_arm)*β, (v_g-v_arm)*0.1, 0); Sending commands: The drone sends motion commands to the sub-ship in real time; Closed-loop execution: After receiving the instruction, the sub-ship immediately controls the robotic arm to move according to the instruction; the drone continuously observes the deviation and continuously sends new instructions until the deviation is less than the preset threshold, at which point it commands the capture.

[0015] Furthermore, step 4 is specifically as follows: List the requirements: The mother ship receives the locations of all the child ships that need resupply, forming a list to be served = [S1, S2, S3, ..., Sn], where n represents the number of child ships that need resupply; Set starting point: The mothership sets its current position P_home as the starting point; Find the nearest: From the list of pending services, find the child ship S_nearest that is closest to the mother ship's current position; Heading to Service: The mothership plans its route to S_nearest to rendezvous with it and resupply; Update status: Remove S_nearest from the list of services pending and update the mothership's current position to the position of S_nearest; Loop check: Check if the list of ships to be served is empty; if not empty, continue searching for the next nearest child ship; if empty, the mother ship returns to its initial position P_home or proceeds to a new work area. Output path: The final result is a resupply path for the mothership.

[0016] The beneficial effects of this invention are: 1. Significantly improve operational efficiency: The innovative "frontline rendezvous and resupply" mechanism minimizes the ineffective sailing time of sub-vessels, resulting in an order-of-magnitude increase in the proportion of effective operational time of the system.

[0017] 2. Extreme system flexibility: "Dynamic role allocation" and "modular task cabin" enable the system to adapt to various complex scenarios and achieve seamless switching from "large area cleaning" to "precise point grasping".

[0018] 3. High level of intelligence: Based on predictive situational awareness, distributed auction decision-making, and real-time visual servo control, the system has unprecedented environmental adaptability and task execution accuracy.

[0019] 4. Superior economic efficiency: The mother ship's waste disposal capacity reduces the frequency of docking; the clustered collaborative operation has a wide coverage area, significantly reducing the overall cost of a single cleanup task. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall architecture of the clustered water surface cleaning system in an embodiment of the present invention; Figure 2 This is a schematic diagram of the internal waste treatment line of the mother ship in an embodiment of the present invention; Figure 3 This is a schematic diagram of the interface and replacement process of the modular mission compartment of the sub-ship in an embodiment of the present invention; Figure 4 This is a block diagram illustrating the principle of precise grasping control based on visual servoing in an embodiment of the present invention. Figure 5 This is a flowchart of the algorithm for frontline rendezvous and supply path planning in an embodiment of the present invention. Detailed Implementation

[0021] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0022] Example 1

[0023] like Figure 1 As shown, a clustered surface clearing system based on dynamic role allocation and frontline rendezvous and resupply is described. This system consists of a mother ship, several sub-ships, and a cluster of unmanned aerial vehicles (UAVs) interconnected via a wireless communication network. 1. Mothership: Serving as the system's mobile command and supply hub, including: The central processing unit runs a global dynamic scheduling system to perform task planning, resource allocation, and path optimization.

[0024] Waste treatment units, such as Figure 2As shown, it integrates a coarse crushing module, a magnetic separation module, a dewatering module, and a compression module connected in sequence, forming a highly efficient waste processing line.

[0025] Modular storage bins are used to store compressed, standardized waste blocks.

[0026] The energy replenishment station, including generator sets, energy storage batteries, and a wireless charging platform, is used to replenish energy for sub-ships and drones.

[0027] Navigation and communication systems are used for high-precision positioning and to maintain stable communication with sub-ships, drones, and shore-based systems.

[0028] 2. Sub-ship: As a functionally reconfigurable execution unit, it includes: The modular mission cabin uses standardized mechanical and electrical interfaces, enabling quick replacement of different types of mission payloads, such as a "collection cabin" for large-area collection, a "grabbing cabin" for precise grasping, or a "cutting cabin" for handling entangled objects.

[0029] The status monitoring system is used to monitor the waste bin capacity, battery level, and equipment health status in real time.

[0030] The autonomous docking system is used to achieve precise and safe docking with the mother ship during navigation in order to complete waste disposal and energy replenishment.

[0031] 3. Unmanned Aerial Vehicles (UAVs): As intelligent sensing and guidance units, including: The integrated sensing system, equipped with a visible light camera and a depth sensor, is used for waste identification and three-dimensional positioning.

[0032] Edge computing units are used to run image recognition algorithms, garbage drift prediction algorithms, and visual servo guidance algorithms.

[0033] A high-precision positioning module is used to provide its own centimeter-level positioning information.

[0034] This embodiment also provides a clustered water surface clearing method based on dynamic role allocation and frontline rendezvous and resupply, including the following steps: S1. Wide-area reconnaissance and predictive situation map construction steps: The UAV swarm scans the target water area, identifies garbage and estimates its type and volume; at the same time, by analyzing continuous image frames and fusing environmental wind flow data, the drift trajectory of the garbage is predicted; the mother ship fuses all data to generate a garbage distribution probability density map with spatiotemporal prediction information.

[0035] The garbage drift prediction algorithm (for drones) aims to solve the problem of enabling drones not only to see where the garbage is now, but also to predict where it will drift in the future, thereby directing drones to "intercept" rather than "chase" it, greatly improving the efficiency of garbage collection.

[0036] The core of the algorithm is to combine the image motion of the debris with the real wind flow data measured by the mother ship to make predictions.

[0037] Algorithm steps: Record position: Record the current position of the trash in the camera frame (u_now, v_now), and its position 0.5 seconds ago (u_old, v_old).

[0038] Calculate the movement speed: Movement speed = ((u_now-u_old) / 0.5, (v_now-v_old) / 0.5), which gives how fast the trash moves in each direction on the screen.

[0039] Receive environmental data: Obtain the current wind speed (wind_speed) and wind direction (wind_direction) from the broadcast information sent by the mother ship, and merge them into a wind flow vector W.

[0040] Comprehensive prediction: When calculating the future direction and size of waste movement, both its own movement and the impact of wind and waves are taken into account.

[0041] Predicted speed = moving speed + 0.6 * W (where 0.6 is an empirical coefficient representing the degree of wind's impact on the garbage).

[0042] Output prediction point: Predicted position = current actual position + prediction speed * 60. Here, we assume that we want to predict the location of the garbage in 60 seconds.

[0043] S2. Dynamic Task Allocation Steps: The mother ship "broadcasts" the cleanup task to all available sub-ships; each sub-ship calculates the "cost" of completing the task based on its own position, capabilities, role, and status, and reports it back to the mother ship; based on the principle of global optimization, the mother ship selects the sub-ship with the lowest "cost" to execute the task.

[0044] The problem the algorithm aims to solve is: when multiple pieces of trash appear in a body of water, how can the mother ship quickly and efficiently decide "who to send" to deal with them, much like an efficient "auction"?

[0045] The core of the algorithm is to mimic an auction mechanism, allowing the child ships to calculate their own "costs" and bid, with the mother ship choosing the most cost-effective option.

[0046] Algorithm steps: Mothership issues mission: Mothership broadcast: "Mission T detected (location in the East District, type is a pile of plastic bottles)".

[0047] Sub-ship cost calculation: Each idle sub-ship that receives a task calculates its own "cost" of completing the task: My cost = distance from the task point + (100 - my remaining power) * 0.1. It can be seen that the farther the distance and the lower the power, the higher the cost.

[0048] Sub-ship bids: Each sub-ship reports to the mother ship: "I am sub-ship X, and the cost to complete this task is Y."

[0049] Mothership bidding: After collecting all bids, the mothership selects the sub-ship with the lowest reported cost and issues the order to it: "Sub-ship X, please execute mission T".

[0050] Status Update: The mothership marks the status of the child ship X as "busy".

[0051] S3, such as Figure 2 and Figure 3 As shown, the adaptive collaborative execution steps are as follows: For tasks requiring precise operation, the drone hovers above the garbage and uses a visual servo controller to calculate the image deviation between the end of the robotic arm and the garbage in real time. The calculated motion control commands are then sent to the sub-boat to drive its robotic arm to complete the precise grasping, forming a closed-loop control system.

[0052] The algorithm aims to solve the problem of the sub-boat's own camera having a poor field of view, causing the robotic arm to constantly "wander" around the trash without being able to grasp it accurately. The solution is to have an aerial drone act as its "eyes," directly instructing the robotic arm to "move a little to the left, a little down."

[0053] The core of the original algorithm: establishing a set of real-time conversion rules from "image deviation seen by the drone" to "robotic arm movement commands".

[0054] Algorithm steps: Deviation detection: The drone simultaneously identifies the center point (u_g, v_g) of the trash and the center point (u_arm, v_arm) of the robotic arm claw, and calculates the pixel difference between them: Deviation = (u_g - u_arm, v_g - v_arm).

[0055] Conversion command: Use a preset "conversion coefficient" (for example, every 10 pixel deviations correspond to the robotic arm moving 1 cm) to convert the pixel deviation into the actual distance and direction that the robotic arm should move: Motion command = ((u_g-u_arm)*0.1, (v_g-v_arm)*0.1, 0); Here, 0.1 is the conversion coefficient, and 0 in the z direction means that it will move only on the horizontal plane first.

[0056] Sending instructions: The drone sends this motion instruction to the sub-ship in real time.

[0057] Closed-loop execution: After receiving the instruction, the sub-ship immediately controls the robotic arm to move according to the instruction. The drone continuously observes the deviation and continuously sends new instructions until the deviation is small enough to be negligible, at which point it issues an order to grab.

[0058] S4, such as Figure 5 As shown, the frontline rendezvous and resupply steps are as follows: When the waste bin capacity or battery power of a sub-ship is lower than a preset threshold, it sends a resupply request to the mother ship; the mother ship models the resupply requests of multiple sub-ships as a path optimization problem, calculates the optimal patrol path for the mother ship to serve these sub-ships and the corresponding frontline rendezvous point for each sub-ship; the mother ship and sub-ships move synchronously towards the rendezvous point to complete automatic docking, waste dumping and energy replenishment.

[0059] The algorithm aims to solve the following problem: When multiple child ships need resupply at the same time, in what order should the mother ship meet them to minimize the total distance and the waiting time for all child ships? The algorithm's original core: It adopts a "nearest neighbor first" greedy strategy to achieve fast and efficient path planning.

[0060] Algorithm steps: List the requirements: The mother ship receives the locations of all the child ships that need resupply, forming a service list = [S1, S2, S3].

[0061] Set starting point: The mothership sets its current position P_home as the starting point.

[0062] Find the nearest: From the list of pending services, find the child ship S_nearest that is closest to the mother ship's current position.

[0063] Heading to Service: The mothership plans its route to S_nearest to rendezvous with it and resupply.

[0064] Status Update: Removed S_nearest from the service list. The mothership's current position is updated to the position of S_nearest.

[0065] Loop check: Check if the service list is empty. If not empty, jump back to step 3 and continue searching for the next nearest child ship. If empty, the mother ship returns to its initial position P_home or proceeds to a new work area.

[0066] Output path: The final result is a resupply path for the mothership, for example: P_home->S2->S1->S3->P_home.

[0067] Example 2

[0068] A cluster-based surface clearing method based on dynamic role allocation and frontline rendezvous and resupply includes the following steps: S1. Wide-area reconnaissance and predictive situation map construction steps: The UAV swarm scans the target water area, identifies garbage, and estimates its type and volume; simultaneously, by analyzing continuous image sequences, the movement direction and speed of the garbage in the images are calculated, and this visual motion data is fused with real-time wind and current data provided by the mother ship; the mother ship adopts a physically guided trajectory prediction method, treating the garbage as a point mass driven by wind and waves, and extrapolates its movement trajectory in the next few minutes; finally, the mother ship integrates the identification and prediction results of all UAVs to generate a garbage distribution prediction situation map that not only marks the current state but also predicts future changes.

[0069] S2. Dynamic Task Allocation Steps: The mothership broadcasts newly discovered cleanup tasks, including their location and type, to all available sub-ships. Upon receiving the broadcast, each sub-ship initiates an autonomous benefit assessment mechanism: First, the sub-ship obtains a basic capability score based on the match between its equipment and the task type; then, it calculates the impact of its travel distance to the task location and its current remaining battery power on performing the task; finally, through a comprehensive benefit calculation formula, it quantifies capability, distance, and battery power into a single benefit value and feeds it back to the mothership. The mothership, acting as the decision-making center, selects the sub-ship with the highest benefit value from all feedback, using the principle of highest benefit, and formally assigns the task to it.

[0070] S3. Adaptive Collaborative Execution Steps: For tasks requiring precise grasping, the drone hovers above the trash and initiates a visual servo closed-loop guidance process: The drone continuously captures images of the trash and the end effector of the robotic arm on the sub-boat, and calculates the pixel coordinate deviation between them; subsequently, a visual servo controller, based on this deviation and known camera depth information, calculates in real time the direction and distance of motion that the end effector of the robotic arm should make in the world coordinate system through the image Jacobian matrix relationship, and sends motion control commands to the sub-boat; the sub-boat drives the robotic arm to execute the commands, thereby dynamically eliminating the positional deviation between itself and the trash until precise grasping is achieved.

[0071] S4. Frontline Rendezvous and Resupply Steps: When the waste bin capacity or battery power of a sub-ship falls below a preset threshold, it automatically sends a resupply request to the mother ship. Upon receiving multiple requests, the mother ship models this problem as a multi-objective path optimization problem and employs a path planning strategy based on a cost-saving algorithm. This strategy first calculates the total distance required for the mother ship to serve each sub-ship individually. Then, by evaluating the mileage "saved" by connecting the resupply points of different sub-ships onto a single path, it prioritizes merging the resupply points with the largest savings, thus planning a resupply route with the shortest total distance for the mother ship. Each docking point on this route becomes the frontline rendezvous point for the corresponding sub-ship. The mother ship and sub-ships move synchronously towards the rendezvous point according to instructions, completing efficient automatic docking, waste dumping, and energy resupply.

[0072] The following is a further explanation of this embodiment: First, the system is assembled. The mother ship can be a medium-displacement catamaran platform to ensure stability. A waste processing line is installed above the deck, with modular storage compartments below. The energy replenishment station is located on the stern deck. The subsidiary ships are pre-equipped with different payloads, such as "collection compartments" and "grabbing compartments," according to mission requirements. The UAVs must have RTK positioning capabilities and sufficient onboard computing power.

[0073] Suppose a cleanup task is being carried out on a lake, the specific workflow is as follows: Initial reconnaissance: After the mother ship arrives at the center of the operational lake area, it releases all drones. The drones fly along preset routes, and their onboard AI identifies debris such as plastic bottles, foam, and dead branches on the lake surface. At the same time, the drones calculate the movement speed of the debris using optical flow and transmit this data to the mother ship via broadcast, generating a dynamic "details distribution prediction map".

[0074] Task Allocation: The mothership's global scheduling system detected a high-density cluster of plastic bottles in the northwest area (Task A) and an abandoned fishing net in the southeast area (Task B). The system then initiated an "auction": Task A requires "collection" capabilities, while Task B requires "grabbing" capabilities. Idle vessels 1 (collection) and 2 (grabbing) each calculated their costs to reach the task points and submitted bids. The mothership determined that both were the lowest-cost options and assigned Task A to vessel 1 and Task B to vessel 2.

[0075] Collaborative Execution: As the second sub-boat approached the fishing net, a drone arrived and hovered ahead. The drone used its camera to identify the ArUco codes on the fishing net and the end effector of the robotic arm on the second sub-boat, initiating visual servo control. It continuously calculated image deviations and used the image Jacobian matrix to determine the appropriate motion increment commands for the robotic arm, sending these commands to the second sub-boat. The second sub-boat's control system then drove the robotic arm according to these commands, ultimately achieving a precise grab of the fishing net.

[0076] Frontline Resupply: Approximately one hour later, the first sub-ship's waste storage capacity reached 85%. It sent a request to the mother ship. At this time, the mother ship discovered that the second sub-ship's battery was also about to run out. Therefore, the mother ship's scheduling system ran a conservation algorithm, calculating an optimal path: the mother ship first proceeded to rendezvous point C with the first ship to dispose of its waste, and then proceeded to rendezvous point D with the second ship to recharge it. The mother ship and the two sub-ships moved synchronously, efficiently completing the resupply at the designated locations, minimizing the total downtime for both sub-ships.

[0077] The embodiments described above are merely illustrative of specific implementations of the present invention, and while the descriptions are detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A clustered water surface clearing system based on dynamic role allocation and frontline rendezvous and resupply, characterized in that, include: The system consists of a mother ship, multi-functional sub-ships, and a cluster of drones; the mother ship serves as a mobile command and supply hub and integrates a waste processing line. The sub-ships adopt a modular design and their functions can be dynamically reconfigured; the drone swarm has the ability to predict debris drift and provide visual servo guidance.

2. The clustered water surface clearing system based on dynamic role allocation and frontline rendezvous and resupply as described in claim 1, characterized in that, The mother ship specifically includes: The central processing unit runs a global dynamic scheduling system to perform task planning, resource allocation, and path optimization. The waste processing unit integrates a coarse crushing module, a magnetic separation module, a dewatering module, and a compression module connected in sequence, forming a highly efficient waste processing production line; Modular storage bins are used to store compressed, standardized waste blocks; The energy replenishment station, including generator sets, energy storage batteries, and a wireless charging platform, is used to replenish energy for sub-boats and drones. Navigation and communication systems are used for high-precision positioning and to maintain stable communication with sub-ships, drones, and shore-based systems.

3. The clustered water surface clearing system based on dynamic role allocation and frontline rendezvous and resupply as described in claim 1 or 2, characterized in that, The sub-ships specifically include: The modular mission bay uses standardized mechanical and electrical interfaces for replacing different types of mission payloads; The status monitoring system is used to monitor the waste bin capacity, battery level, and equipment health status in real time. The autonomous docking system is used to achieve precise and safe docking with the mother ship during navigation in order to complete waste disposal and energy replenishment.

4. The clustered water surface clearing system based on dynamic role allocation and frontline rendezvous and resupply as described in claim 3, characterized in that, The drone swarm specifically includes: The integrated sensing system, equipped with a visible light camera and a depth sensor, is used for waste identification and three-dimensional positioning; Edge computing units are used to run image recognition algorithms, garbage drift prediction algorithms, and visual servo guidance algorithms; A high-precision positioning module is used to provide its own centimeter-level positioning information.

5. A cluster-based water surface clearing method based on dynamic role allocation and frontline rendezvous and resupply, characterized in that, The method employs a clustered surface clearing system based on dynamic role allocation and frontline rendezvous and resupply, as described in any one of claims 1-4, and specifically includes the following steps: Step 1: Wide-area reconnaissance and predictive situation map construction: The UAV swarm scans the target water area, identifies garbage and estimates its type and volume; at the same time, by analyzing continuous image frames and fusing environmental wind flow data, the drift trajectory of the garbage is predicted; the mother ship fuses all data to generate a garbage distribution probability density map with spatiotemporal prediction information. Step 2, Dynamic Task Allocation: The mothership broadcasts the cleanup task to all available sub-ships; each sub-ship calculates the cost of completing the task based on its own position, capabilities, role, and status, and reports it back to the mothership; the mothership selects the sub-ship with the lowest cost to execute the task based on the principle of global optimization. Step 3, Adaptive Collaborative Execution: For tasks requiring precise operation, the drone hovers above the garbage and uses a visual servo controller to calculate the image deviation between the robotic arm end and the garbage in real time. The calculated motion control commands are then sent to the sub-boat to drive its robotic arm to complete the precise grasping, forming a closed-loop control system. Step 4, Frontline Rendezvous and Resupply: When the waste bin capacity or battery power of a sub-ship falls below a preset threshold, it sends a resupply request to the mother ship. The mother ship models the resupply requests from multiple sub-ships as a path optimization problem, calculates the optimal patrol path for the mother ship to serve these sub-ships and the corresponding frontline rendezvous point for each sub-ship, and the mother ship and sub-ships move synchronously towards the rendezvous point to complete automatic docking, waste dumping, and energy resupply.

6. The clustered water surface clearing method based on dynamic role allocation and frontline rendezvous and resupply as described in claim 5, characterized in that, In step 1, the predicted drift trajectory of the waste is as follows: Record position: Record the current position of the trash in the camera frame (u_now, v_now), and its position t seconds ago (u_old, v_old); Calculate movement speed: Movement speed = ((u_now - u_old) / t, (v_now - v_old) / t); Receive environmental data: Obtain the current wind speed (wind_speed) and wind direction (wind_direction) from the broadcast information sent by the mother ship, and merge them into a wind flow vector W; Comprehensive prediction: When calculating the future direction and size of the waste, the movement of the waste itself and the impact of wind and waves are taken into account. Predicted speed = moving speed + α * W, where α is an empirical coefficient representing the degree of wind's impact on the garbage; Output prediction point: Predicted position = current actual position + prediction velocity * T, where T represents time. The output prediction point is the location of the garbage after T seconds.

7. The clustered water surface clearing method based on dynamic role allocation and frontline rendezvous and resupply as described in claim 5, characterized in that, Step 2 is described in detail below: Mothership issues mission: Mothership broadcasts mission T, including the location and type of trash; Sub-ship cost calculation: Each idle sub-ship that receives a task calculates its own cost to complete the task: Cost = Distance to the task point + (100 - Remaining power) * 0.1; Sub-ship bidding: Each sub-ship reports its own number and the cost of completing the task to the mother ship; Mother ship bidding: After collecting all bids, the mother ship selects the subsidiary ship with the lowest reported cost and issues a TDE (Task Execution) order to it; Status Update: The mothership marks the status of the child ships performing the mission as busy.

8. The clustered water surface clearing method based on dynamic role allocation and frontline rendezvous and resupply as described in claim 5, characterized in that, Step 3 is as follows: Deviation detection: The drone simultaneously identifies the center point (u_g, v_g) of the trash and the center point (u_arm, v_arm) of the robotic arm claw, and calculates the pixel difference between them: Deviation = (u_g - u_arm, v_g - v_arm); Conversion command: Use a preset conversion coefficient β to convert the pixel deviation into the actual distance and direction that the robotic arm should move: Motion command = ((u_g-u_arm)*β, (v_g-v_arm)*0.1, 0); Sending commands: The drone sends motion commands to the sub-ship in real time; Closed-loop execution: After receiving the instruction, the sub-ship immediately controls the robotic arm to move according to the instruction; the drone continuously observes the deviation and continuously sends new instructions until the deviation is less than the preset threshold, at which point it commands the capture.

9. The clustered water surface clearing method based on dynamic role allocation and frontline rendezvous and resupply as described in claim 5, characterized in that, Step 4 is as follows: List the requirements: The mother ship receives the locations of all the child ships that need resupply, forming a list to be served = [S1, S2, S3, ..., Sn], where n represents the number of child ships that need resupply; Set starting point: The mothership sets its current position P_home as the starting point; Find the nearest: From the list of pending services, find the child ship S_nearest that is closest to the mother ship's current position; Heading to Service: The mothership plans its route to S_nearest to rendezvous with it and resupply; Update status: Remove S_nearest from the list of services pending and update the mothership's current position to the position of S_nearest; Loop check: Check if the list of ships to be served is empty; if not empty, continue searching for the next nearest child ship; if empty, the mother ship returns to its initial position P_home or proceeds to a new work area. Output path: The final result is a resupply path for the mothership.

Images