Multi-robot autonomous cooperation method and system

Abstract

The invention provides a multi-robot autonomous cooperation method, which comprises the steps that a communication network is established, a robot pool containing a plurality of robots is configured, and the robot pool selects at least one robot as a member to form a current working team according to a task target issued by a user; enabling members in the current working team to perceive task environment information and monitor task conditions in real time, and synchronizing local environment task state diagrams among the members through a communication network so as to update respective local environment task state diagrams; and each member evaluates the capability of the current working team according to the updated local environment task state diagram based on autonomous communication and a multi-round negotiation mechanism so as to judge whether team reorganization and / or task decomposition and distribution are carried out or not, so that the current working team and a task decomposition and distribution scheme are updated. And each member executes the allocated subtask based on the updated scheme. According to the method, the cooperative adaptive capacity, the task completion efficiency and the overall system robustness of a multi-robot team in a dynamic complex environment are improved.

Description

Technical Field

[0001] The present invention relates to the field of multi-agent collaboration technology, and in particular to a multi-robot autonomous collaboration method and system. Background Art

[0002] With the rapid development of robotics and its widespread application in industrial automation, disaster relief, and service industries, multi-robot systems (MRS) have become a key area of ​​intelligent robotics research. Compared to traditional single, fixed-function robots, flexible intelligent robots are capable of completing more complex tasks in dynamic environments. However, as the number of robots and the complexity of tasks increase, multi-robot collaboration presents numerous challenges in system organization and scheduling, resource allocation, and collaborative mechanisms.

[0003] Most existing multi-robot systems employ centralized or hierarchical organizational structures. While these structures can achieve a certain degree of task allocation and resource coordination, they suffer from sluggish responses to environmental changes or dynamic task adjustments, and their overall activity and adaptability are limited. Especially in dynamic or uncertain environments, robots require extensive information exchange and negotiation, which significantly increases system communication pressure, making them prone to information congestion and scheduling bottlenecks, impacting the overall efficiency and robustness of the system. Furthermore, centralized scheduling models rely heavily on central nodes or managers, presenting single-point failure risks and poor scalability.

[0004] Currently, some distributed multi-robot collaborative systems attempt to reduce system communication volume and improve task scheduling flexibility through mechanisms such as local self-organization, autonomous evaluation, and bulletin boards. For example, robots can proactively acquire task information during idle time and compete and decompose tasks based on their capabilities, thus reducing redundant communication and resource waste. Although these approaches have made some progress in reducing communication volume and improving resource utilization, most solutions still suffer from the following technical issues: First, they lack efficient dynamic reorganization mechanisms for complex, heterogeneous teams, making it difficult to flexibly adjust team structures based on actual task requirements, resulting in insufficient or redundant resource allocation in some scenarios. Second, current systems often passively adapt to member changes (such as failure exits and external assignments), making it difficult to foresee and proactively respond to potential challenges, and failing to achieve proactive self-adaptation based on team status and environmental awareness. Third, during task execution, communication and collaboration protocols often rely on pre-defined scripts or limited rules, making them difficult to meet the complex and ever-changing collaborative needs of multiple robots in open environments, thus hindering the improvement of the team's overall intelligence and robustness.

[0005] Furthermore, to address information sharing and task division during multi-robot collaboration, some research has established inter-robot communication platforms using cloud servers or local area networks to enable data synchronization, file transfer, and status broadcasting. However, relying on external servers not only incurs additional system maintenance and operation costs, but also presents challenges such as poor communication stability, network congestion, and security risks. For example, network bandwidth bottlenecks or signal interference can significantly impact task collaboration efficiency, and data transmission can also present security risks such as leakage, packet loss, or tampering.

[0006] In summary, existing collaboration and scheduling methods for multi-robot systems still face technical challenges that need to be overcome in the following areas: (1) How to achieve efficient dynamic team reorganization and resource allocation in heterogeneous multi-robot teams to improve the system's ability to adapt to complex environmental changes; (2) How to break through the limitations of passive and delayed responses and build an intelligent mechanism based on team autonomous perception and active collaboration to improve task completion efficiency and team robustness; (3) How to reduce the communication burden between multi-robot systems while ensuring efficient and secure communication and improve the collaborative ability of information processing and decision-making. The effective solution to these problems is the key to promoting the widespread application and intelligent upgrading of multi-robot systems. Summary of the Invention

[0007] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0008] To this end, the present invention proposes a multi-robot autonomous collaboration method and system for complex dynamic environments that supports team autonomous reorganization and efficient task collaboration. In response to the problems of limited collaboration capabilities, low team reorganization efficiency, insufficient active adaptability, and heavy communication burden of existing multi-robot systems in dynamic environments, the present invention can autonomously perceive team status and environmental changes, dynamically adjust team member structure, achieve efficient collaboration and task allocation, and effectively improve the collaborative adaptability, task completion efficiency and overall system robustness of multi-robot systems in heterogeneous tasks and complex environments, thereby promoting the widespread application and intelligent development of multi-robot systems in actual scenarios.

[0009] In order to achieve the above object, the present invention adopts the following technical solutions:

[0010] A first aspect of the present invention provides a multi-robot autonomous collaboration method, comprising:

[0011] Step S100: Build a communication network and configure a robot pool containing multiple robots. All robots share each other's basic configuration information. The robot pool selects at least one robot as a member to form the current work team based on the task objectives issued by the user.

[0012] Step S200: After the current work team enters the task environment, its members perceive the task environment information and monitor the task status in real time, obtain their respective local environment task status graphs, and synchronize the local environment task status graphs among the members via the communication network to update their respective local environment task status graphs;

[0013] Step S300: Each member evaluates the capabilities of the current work team based on the updated local environment task status diagram based on autonomous communication and multi-round negotiation mechanisms, and determines whether to reorganize the team and / or decompose and allocate tasks based on the evaluation results, thereby updating the current work team and task decomposition and allocation plan, and each member performs the assigned subtasks based on the updated plan.

[0014] In some embodiments, the robot pool is configured outside the task environment, and the robots contained in the robot pool are heterogeneous robots or homogeneous robots; each robot maintains communication with the robot pool and other members of the current work team through its own wireless communication module, supporting point-to-point communication and broadcast mechanisms.

[0015] In some embodiments, the robot pool is equipped with an AI model. For clear or vague task objectives issued by users, the AI ​​model selects members based on task understanding and capability matching to build an initial current work team.

[0016] In some embodiments, the task environment information perceived by the member includes structural information of the task execution space, object information related to the task goal, potential obstacles and operational constraints, and dynamic change information;

[0017] The task status monitored by the members in real time includes the execution stage of the current task assignment, the current execution status of the members, the completion mark of the task goal, whether the preset timeout threshold is triggered, the failure events that occur in the task collaboration, and whether there is an intermediate state that requires adjusting the task strategy or reorganizing the team.

[0018] In some embodiments, each member is driven by a local clock to periodically upload a local environment task status diagram, while receiving and integrating information from other members, thereby updating their respective local environment task status diagrams.

[0019] In some embodiments, the updated local environment task state diagram is a multimodal, layered semantic task state diagram, and the updating process includes:

[0020] Step S210, node extraction: extract the following four types of semantic nodes based on the task objectives and information synchronized by other members:

[0021] T-node: represents the subtask to be completed and its location;

[0022] A-node: represents the members who have joined the team and their current location, ability tags and status information;

[0023] E-node: represents the binding relationship between the currently executing subtask and the executor;

[0024] U-node: indicates an area or task point that has not been explored or has insufficient information, and is used to indicate areas that need to be sensed later;

[0025] Step S220, layer division: Divide the local environment task state graph into three semantic layers:

[0026] Task layer: includes all task nodes to be completed or being executed;

[0027] Spatial layer: contains the environment spatial structure nodes and the spatial dependency relationship between tasks and the environment;

[0028] Agent layer: includes team member nodes and their perception, execution, and communication status;

[0029] Step S230: Edge connection:

[0030] In the task layer, directed edges are established based on the dependency order between tasks;

[0031] In the spatial layer, undirected edges are established based on spatial adjacency;

[0032] In the proxy layer, edges connect member nodes with the task nodes they are responsible for, indicating the current task allocation situation;

[0033] Nodes with unassigned tasks or areas to be explored are connected to A-nodes or U-nodes through virtual edges to indicate scheduling requirements;

[0034] Step S240, graph fusion and update: After each member has constructed the environment task status graph, it synchronizes state changes through the communication network and performs graph merging and deduplication operations to obtain an updated local environment task status graph.

[0035] In some embodiments, step S300 includes:

[0036] Step S310: Each member of the current work team obtains self-assessment information based on the updated local environment task state diagram, combined with their own capabilities, and the matching degree between the task requirement dimension and the robot capability dimension. If the member's self-assessment information indicates that the member has the intention to communicate, the process proceeds to step S320; otherwise, the member will continue to perform the currently assigned subtask;

[0037] In step S320, members with communication intentions send their self-assessment information to other relevant members via the communication network, entering into an autonomous communication and multi-round negotiation process. During this process, the communicating members compare each other's adaptability, execution efficiency, and task load information to achieve one or a combination of the following negotiation goals:

[0038] When the current work team has a capacity bottleneck, reorganize the team, recruit new members with matching capabilities from the robot pool, and update the current task status;

[0039] When there is redundant member resources in the current work team, the member is marked as releasable, returned to the robot pool, and the current task status is updated;

[0040] When the current task status changes, each member divides the subsequent task objectives based on their updated local environment task status diagram and redistributes them to the corresponding members.

[0041] In some embodiments, the communication intention is generated when the member detects any one or a combination of the following situations: new observations of the task environment, updates to task-related information, updates to the status of task execution, and estimates of the effect of task execution.

[0042] In some embodiments, the autonomous communication and multi-round negotiation mechanism includes:

[0043] Step S321: Initialize members who have communication intentions into a waiting-to-speak queue;

[0044] Step S322: Determine whether there are any members in the waiting-to-speak queue. If not, exit the current round of negotiation, and each member continues to perform the currently assigned subtask until one or more members express the intention to communicate and enter the next round of negotiation. If yes, proceed to step S323.

[0045] Step S323: The first person in the waiting-to-speak queue is removed from the queue and becomes the current speaker. The current speaker creates a communication content based on their communication intention and selects a recipient to send the communication content to. The current speaker and recipient update their respective local communication message records. If the recipient is not in the current speaking queue, the recipient is added to the end of the waiting-to-speak queue, and the process proceeds to step S324.

[0046] Step S324: If the communication intention of the first member in the waiting-to-speak queue disappears, delete him from the waiting-to-speak queue and return to step S322; if not, return to step S323.

[0047] A second aspect of the present invention provides a multi-robot autonomous collaboration system, comprising:

[0048] A robot pool containing multiple robots. The robot pool is configured to select at least one robot from the multiple robots as a member to form an initial current work team based on the task objectives issued by the user. All robots in the robot pool share each other's basic configuration information;

[0049] A communication network, configured in the task environment, is used to achieve communication between members of the current work team and between the current work team and the robot pool;

[0050] The current work team is configured to evaluate the capabilities of the current work team based on the updated local environment task state graph formed during the execution of each member's task and based on autonomous communication and multi-round negotiation mechanisms, to form a new current work team and task decomposition and allocation plan. The robot pool responds to the new current work team plan and forms a new current work team in the task environment, wherein the updated local environment task state graph is obtained by each member of the current work team fusing their own local environment task state graph with the local environment task state graph synchronized with other members through the communication network.

[0051] Compared with the prior art, the present invention has the following significant features and beneficial effects:

[0052] (1) The present invention adopts a capability assessment mechanism based on team status and environmental perception, which realizes the autonomous optimization and adjustment of the work team structure, and greatly improves the adaptability and task completion efficiency of the multi-robot system in complex dynamic environments.

[0053] (2) The present invention realizes efficient division of labor and optimal resource allocation through autonomous task decomposition and allocation, and member capability modeling, effectively avoiding capability redundancy and resource vacancies.

[0054] (3) By adopting a distributed autonomous communication and multi-round negotiation mechanism, members can actively initiate collaboration requests according to their needs, which improves the flexibility and robustness of multi-robot collaboration and breaks through the limitations of traditional methods such as communication congestion and slow response.

[0055] (4) Through dynamic self-learning and optimization of the team capability model based on negotiation results, the continuous improvement and adaptive optimization of the system's collaborative capabilities are achieved.

[0056] (5) The present invention supports flexible access to heterogeneous robot platforms, expanding the application space of multi-robot systems in different scenarios and tasks.

[0057] In summary, the present invention realizes autonomous perception, active collaboration and efficient task completion of multi-robot teams in complex dynamic environments, overcomes the shortcomings of existing technologies in terms of team flexibility, autonomy, collaborative efficiency and system robustness, and significantly improves the collaborative intelligence and application value of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of the overall flow of a multi-robot autonomous collaboration method provided for an embodiment of the first aspect of the present invention.

[0059] Figure 2 A schematic flow chart of an autonomous communication and multi-round negotiation mechanism employed in a multi-robot autonomous collaboration method provided in an embodiment of the first aspect of the present invention.

[0060] Figure 3 A schematic diagram of the structural framework of a multi-robot autonomous collaboration system provided in an embodiment of the second aspect of the present invention. DETAILED DESCRIPTION

[0061] In order to make the objectives, technical solutions, and advantages of this application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only intended to explain this application and are not intended to limit this application. The following scheme is merely an illustration of the inventive concept and the specific scheme is not limited thereto. In addition, for ease of description, the accompanying drawings only show the parts related to the present invention, rather than the entire process.

[0062] On the contrary, this application covers any alternatives, modifications, equivalents, and solutions made within the spirit and scope of this application as defined by the claims. Furthermore, to facilitate a better understanding of this application, certain specific details are described in detail below in the detailed description of this application. Those skilled in the art will be able to fully understand this application without these details.

[0063] See also Figure 1 The first embodiment of the present invention provides a multi-robot autonomous collaboration method, comprising the following steps:

[0064] Step S100: Build a communication network and configure a robot pool containing multiple robots. All robots share each other's basic configuration information. The robot pool selects at least one robot as a member of the current work team based on the task objectives issued by the user. At least one robot in the current work team has perception capabilities.

[0065] Step S200: After the current work team enters the task environment, its members respectively perceive the task environment information and monitor the task status in real time, obtain their respective local environment task status diagrams, and synchronize the local environment task status diagrams among the members via the communication network to update their respective local environment task status diagrams;

[0066] Step S300: Each member evaluates the capabilities of the current work team based on the updated local environment task state diagram using autonomous communication and a multi-round negotiation mechanism. Based on the evaluation results, the team determines whether to reorganize the team and / or decompose and allocate tasks. The current work team and task decomposition and allocation plan are updated, and each member performs the corresponding task based on the updated plan.

[0067] Step S400: Repeat steps S200 to S300 until the task is completed.

[0068] In some embodiments, in step S100, a robot pool is configured outside the task environment. The robots in the robot pool may be either heterogeneous or homogeneous robots. Heterogeneous robots refer to robots with different performance parameters, including different structural parameters, model parameters, motion parameters, or sensor parameters. In contrast, homogeneous robots are robots with identical performance parameters. Each robot in the robot pool can be an intelligent robot with reasoning capabilities or a standard rule-based robot without reasoning capabilities. The basic configuration information of a robot includes its performance parameters.

[0069] In a specific embodiment of the present application, the robot pool before task execution contains the following four types of robots:

[0070] Mobile operating robots with AI, navigation, and manipulation capabilities (such as the four-legged robot AlienGo launched by Yushu Technology);

[0071] Precision manipulation robots with wide-range robotic arm manipulation capabilities (such as Boston Dynamics' Stretch, a robot designed specifically for warehousing and logistics);

[0072] Cleaning robots dedicated to cleaning tasks (such as the Cleaner robot);

[0073] Transport robots with basic navigation and reasoning capabilities (such as compact reconfigurable automated robots CAR).

[0074] Each robot maintains data interaction with the robot pool and other robots in the current work team through its own wireless communication module, supporting point-to-point communication and broadcast mechanisms.

[0075] The robot pool is also equipped with a wireless communication module and an AI model. The wireless communication module communicates with the communication network to respond to team reorganization plans, waking up or powering on robots currently in the pool and dispatching them, or recalling and shutting down or hibernating members of the current work team. The AI ​​model utilizes a large language model, and based on its task understanding and capability matching mechanism, it is capable of intelligent reasoning and prediction even when task objectives are unclear. Furthermore, because members of the current work team exchange information through autonomous communication and multi-round negotiation, the task objectives received by the present method do not need to be clearly defined. User-issued instructions lacking task details are acceptable. This allows the robot team to complete the task by continuously exploring the task environment, adjusting team composition and task assignments, and ultimately completing the task even when details are lacking. Ideally, if the user-issued task objective is clear, the AI ​​model can directly arrange the most appropriate initial work team configuration and division of labor. For example, in the query "Please help me clean three rooms," the user does not need to provide further task details such as cleanliness level or room size. Based on this task objective, the AI ​​model in the robot pool can select three robots that are most suitable as members of the initial work team.

[0076] In some embodiments, the communication network is built in the task environment through a wireless local area network (WLAN) to achieve real-time information interaction and synchronization between members of the current work team. The robot pool creates a dedicated Wi-Fi network. After the work team members access it, they communicate through the DDS or MQTT protocol of ROS2, supporting multicast discovery, topic publishing and subscription, and achieving real-time synchronization of task status, perception data, and negotiation information. For example, in the “Cleaning Two Rooms and One Living Room” task, AlienGo, Stretch, and Cleaner are connected to the same LAN to collaboratively complete map construction, area cleaning, and item grabbing tasks, maintaining low-latency communication and efficient collaboration through the LAN throughout the process. The specific process is as follows:

[0077] Network initialization: After the robot pool determines the initial work team, it will assign a unique task identifier (Task ID) and communication channel to the team.

[0078] LAN establishment: Using communication modules deployed within the robot pool, a task-specific Wi-Fi network is created based on wireless local area network (WLAN) protocols (e.g., IEEE 802.11) (either in router mode or hotspot direct connection mode).

[0079] Member joining: After receiving the task joining instruction, the awakened or dispatched robot accesses the communication network through the specified Wi-Fi SSID and pre-shared key;

[0080] Identity registration and heartbeat maintenance: After successfully connecting, each robot must register its identity information, capability model, and current status with the communication network and maintain its online status through a periodic heartbeat mechanism.

[0081] Communication method: Team members use lightweight communication middleware such as ROS2 DDS (Data Distribution Service) or MQTT for data exchange, supporting topic publish / subscribe and service request (request / reply) models to ensure low-latency and high-reliability transmission of task status, perception data, control commands, and negotiation messages.

[0082] Network fault tolerance: If a member loses connection or communication is interrupted, the communication anomaly detection mechanism will be triggered, and network resources will be reallocated or the team will be reorganized as needed.

[0083] In some embodiments, in step S200, members of the current work team are enabled to perceive task environment information and monitor task status in real time to form their own multimodal local environment task status diagrams.

[0084] Task environment information includes, but is not limited to: structural information about the task execution space, such as the number, size, and layout of rooms; object information related to the task objective, such as the type and location of cleaned items and the spatial state of the target object; potential obstacles and operational constraints, such as slippery floors and crowded spaces; and dynamically changing information, such as the current location and status feedback of other team members. Each robot acquires environmental perception data within its local field of view through its onboard sensors (e.g., depth camera, RGB camera, LiDAR, IMU, etc.) and encodes this data into a structured state vector or task feature graph. Specifically, the robot collects raw perception data using its onboard sensors and employs existing visual recognition and spatial mapping methods (e.g., YOLOv5, Mask R-CNN, ORB-SLAM, or PointNet++) to model the location, type, confidence, and spatial layout of objects in the environment. The model output is encoded into a structured state vector (e.g., target ID, location coordinates, category label, confidence) or a task feature graph (e.g., nodes representing task objectives and edges representing spatial or operational dependencies) for subsequent task planning and team collaboration. The above-mentioned perception and encoding methods are commonly used methods disclosed in the art. Those skilled in the art can directly call existing models and tools to implement the above-mentioned functions.

[0085] The task status includes but is not limited to: the execution stage of the current task assignment (subtask) (such as not started, in progress, completed), the current execution status of each member (such as busy, idle, abnormal), the completion mark of the task goal, whether the preset timeout threshold is triggered, failure events that occur in task collaboration (such as dropped objects, blocked paths), and whether there are intermediate states that require adjustment of task strategies or reorganization of the team.

[0086] In some embodiments, in step S200, the work team members first obtain their own status, task execution progress and environmental information in real time through the sensors they carry, thereby constructing a local environment task status diagram, and then periodically broadcast or synchronize the local environment task status diagrams obtained by each member through the communication network to ensure that all members have a consistent understanding of the current task status and provide data support for subsequent capability assessment and collaborative decision-making. In order to achieve the effectiveness and real-time nature of communication, an asynchronous synchronization mechanism is adopted: each member is driven by a local clock and regularly uploads the local environment task status diagram, while receiving and integrating information from other members, thereby updating their respective local environment task status diagrams. Among them, each member parses and integrates the local environment task status diagram received through the communication network, and combines it with its own original local environment task status diagram to construct a multimodal, hierarchical semantic task status diagram. The task status diagram is a spatial-semantic coupling structured diagram used to uniformly represent the current task environment, target distribution and member status, and serves as the input basis for subsequent capability assessment and collaborative planning. The specific update process includes the following steps:

[0087] Step S210, node extraction: extract the following four types of semantic nodes based on the task objectives and information synchronized by other members:

[0088] T-nodes (TaskNodes): represent the subtasks to be completed and their locations, such as "clean room A" and "move object B";

[0089] A-Nodes (Agent Nodes): Indicates members who have joined the team and their current location, capability tags, and status information;

[0090] E-Nodes (ExecutingNodes): Indicates the binding relationship between the currently executing subtask and the executor;

[0091] U-nodes (Unexplored Nodes): Indicates areas or task points that have not been explored or have insufficient information, and are used to indicate areas that need to be sensed later;

[0092] Step S220, layer division: Divide the local environment task state graph into three semantic layers:

[0093] Goal Layer: includes all task nodes to be completed or being executed;

[0094] Spatial Layer: Contains environment spatial structure nodes such as rooms and corridors, as well as the spatial dependency relationship between tasks and the environment;

[0095] Agent Layer: includes team member nodes and their perception, execution, and communication status;

[0096] Step S230: Edge connection:

[0097] In the task layer, directed edges are established based on the dependency order between tasks (e.g., “clean first, then carry”);

[0098] In the spatial layer, undirected edges are established based on spatial adjacency relationships (e.g., “there is a passage between room A and room B”);

[0099] In the agent layer, edges connect robot nodes with the task nodes they are responsible for, indicating the current task allocation situation;

[0100] Nodes with unassigned tasks or areas to be explored are connected to A-nodes or U-nodes through virtual edges to indicate scheduling requirements.

[0101] Step S240, graph fusion and update: After each member has built the local environment task status graph, it synchronizes state changes through the communication network, and performs graph merging (that is, merging its own local environment task status graph with the local environment task status graphs of other members obtained through the communication network) and deduplication operations to obtain an updated local environment task status graph. The updated local environment task status graph provides data support for subsequent capability assessment and collaboration.

[0102] The above update method is based on the task graph modeling and multi-robot state fusion framework known in the art (such as Semantic Graphs, Task Allocation Graphs), combined with task semantics and spatial constraints. It has a clear structure, is scalable, and easy to reproduce. Those skilled in the art can implement the graph update of the embodiment of the present invention according to the above process.

[0103] In a specific embodiment of the present application, the members of the current work team use a communication network to ensure that the perception information is effectively integrated among multiple members and reach cognitive consensus. For example, in the task of cleaning the room:

[0104] AlienGo can identify the boundaries of room areas and door locations;

[0105] Cleaner can identify the spatial distribution of water stains and dirt on the ground;

[0106] Stretch senses the position of high objects;

[0107] Each member will publish information on the status of targets, obstacles, channels and areas within its field of view on the communication network;

[0108] Each member will process the received communication information and construct a multimodal local environment task state diagram.

[0109] In some embodiments, step S300 includes:

[0110] In step S310, each member of the current work team will obtain self-assessment information based on the updated local environment task state diagram, combined with their own capabilities (such as structural parameters, operating range, grasping type, navigation efficiency, etc.), and comprehensively consider the match between the task requirement dimension and the robot capability dimension. If the member's self-assessment information indicates that the member has communication intention (i.e., needs to negotiate with one or more other members of the current work team), then the process proceeds to step S320; otherwise, the member will continue to perform the currently assigned subtask;

[0111] In step S320, members with communication intent send their self-assessment information to other relevant team members via the communication network, entering a self-communication and multi-round negotiation process. During this process, communicating members compare information such as adaptability, execution efficiency, and task load to achieve one or a combination of the following negotiation goals:

[0112] If the current work team has capacity bottlenecks, including insufficient operational capabilities (e.g., no member has sufficient arm span or degrees of freedom to complete high-position object grasping; no member has dual-arm coordination or stable grasping capabilities to complete large objects; inability to complete tasks that require delicate operations such as unscrewing bottle caps and plugging in power; inability to perform actions that require large torque such as opening and pushing doors; inability to complete tasks that grasp multiple objects simultaneously), lack of operational capabilities, limited navigation capabilities, lack of perception capabilities, etc., resulting in the inability to continue task execution, then after multiple rounds of negotiation, the negotiation goal is to reorganize the team, recruit new members with matching capabilities (e.g., Stretch) from the robot pool, and update the current task status;

[0113] If there is redundant member resources in the current work team, for example, a member currently has no or expected task assignments, the negotiation goal is to mark the member as releasable, issue a "return" command to the robot pool, return the marked member to the robot pool, and update the current task status.

[0114] If the current task status changes, such as completing part of the room cleaning, the members of the current work team will re-decompose the task during the discussion, divide the subsequent task goals based on the updated local environment task status diagram, and redistribute them to the remaining team members.

[0115] During the negotiation process, tasks can be reallocated to various members of the current work team and the negotiation process can be automatically terminated to ensure that the current work team can resume execution in a timely manner after reaching an agreement on the status.

[0116] For example, in a room cleaning task, when team members recognize that the floor in some areas is slippery and AlienGo cannot effectively perform the cleaning task, and Cleaner has left the current work team, the current work team, through autonomous communication and multiple rounds of negotiations, realizes that there is a capacity gap in the current work team, and the current members take the initiative to initiate a re-recruitment request for Cleaner.

[0117] Further, see Figure 2 The autonomous communication and multi-round negotiation mechanism provided by the embodiment of the present invention includes the following steps:

[0118] Step S321: Initialize members with communication intentions into a waiting-to-speak queue (for example, if the current work team contains members A, B, C, and D, the initialized waiting-to-speak queue contains members A, B, and C). Establish a communication preparation phase and determine the initial set of members to participate in the communication. A member generates a communication intention when it detects any one or a combination of the following: new observations of the task environment, updates to task-related information, updates to the task status, estimates of task performance, etc. For example, in a room cleaning task, a robot may generate a communication intention to seek communication or help from other members due to discovering new stains, successfully completing exploration of an area, or failing to grab an item to be cleaned.

[0119] Step S322: Determine whether there are any members in the waiting-to-speak queue. If not, exit the current round of negotiation. Each member continues to perform the currently assigned subtask until one or more members express the intention to communicate and enter the next round of negotiation. If yes, proceed to step S323.

[0120] Step S323: The first member in the queue to speak (i.e., member A) is removed from the queue and becomes the current speaker. The current speaker creates communication content based on his / her communication intention and selects recipients (e.g., member A selects members C and D as recipients) to send the communication content to them. The current speaker and recipient update their respective local communication message records. If the recipient is not in the current speaking queue (e.g., member D is not in the queue to speak at this time), the recipient (i.e., member D) is added to the end of the queue to speak (at this time, the queue to speak has members B, C, and D) so that he / she may initiate communication later, and the process proceeds to step S324.

[0121] Step S324: If the communication intention of the first member in the queue to speak disappears, delete him from the queue to speak (for example, when member C becomes the first member in the queue to speak, since member C has received the communication content sent by member A, and the communication content contains content that can resolve member C's communication intention, the communication intention of member C disappears, and member C is deleted from the queue to speak), and return to step S322; if it has not disappeared, return to step S323.

[0122] Furthermore, in step S322, if there are members in the waiting-to-speak queue, the first member in the waiting-to-speak queue is designated as the current speaker. The current speaker generates structured communication content based on the updated local environment task state graph and collaboration requirements, and determines the scope of recipients based on the message type: if the message is a task request (such as a collaboration request or capability fill), one or a small number of target members are selected as recipients; if the message is a capability inquiry or information broadcast, all or some team members are selected as recipients, using a broadcast mechanism. The current speaker can call upon a large language model deployed on their own to assist in determining the target recipients of the current message based on their own requirements, the team capability model, and the updated local environment task graph structure. This determination can be based on reasoning based on the following information: the capability requirements of the target task, the current status of team members, their spatial location relationships, and historical task assignment records. If the current speaker does not have a large language model deployed on their own, they can implement the same recipient screening logic through task feature matching and capability screening rules, ensuring a clear and executable communication path for the member negotiation process.

[0123] In some embodiments, in step S400, the current work team repeatedly executes steps S200 to S300 in a fixed cycle or event-triggered manner until the task is completed. Each round of the perception-monitoring-negotiation-evaluation-reperception process constitutes a complete collaboration cycle, and the team status maintains continuity and adaptability between cycles.

[0124] It can be understood that the multi-robot active collaboration method provided by the embodiment of the present invention can automatically conduct a comprehensive assessment of team member capabilities, task requirements and task environment changes based on the team and environmental information synchronized by each member. When it is detected that the team capabilities are insufficient or there is redundant member resources, it will actively decide whether to recruit new members from the robot pool or release redundant members based on the task requirements and member capability differences, thereby realizing dynamic optimization of the current work team structure; after the team members are adjusted, the overall task is automatically decomposed and reasonably allocated to each member based on the task goals and member capability characteristics. When encountering task changes or new tasks, the task allocation can also be adjusted in real time to maintain the optimal collaboration state of the team. Members perform operations autonomously according to the assigned tasks, during which they continuously monitor member status, task progress and environmental changes. If the task cannot be completed on time or a failure occurs, the current work team automatically restarts the capability assessment to realize dynamic closed-loop management of task execution.

[0125] It should be noted that existing multi-robot collaboration methods mainly rely on centralized scheduling or static polling task allocation mechanisms. The interactions between robots mostly use sequential polling or fixed voice processes. When faced with dynamic tasks and heterogeneous teams, negotiation deadlocks are prone to occur (such as multiple members waiting for responses and no one actively advancing) or frequent deviations from the core issues of the task due to lack of contextual understanding, resulting in low negotiation efficiency and even collaboration failure. In the multi-robot autonomous collaboration method provided by the embodiment of the present invention, each member of the current work team uses autonomous communication and a multi-round negotiation mechanism to achieve information exchange and progress feedback during task execution, which has the following significant advantages:

[0126] 1. Avoid deadlock and blind waiting: Each member can proactively speak based on task status and their own needs. Through a queue mechanism and negotiation round control, structured and controllable multi-round interactions are achieved, significantly reducing negotiation stalls.

[0127] 2. Always focus on the core of the task: Combine the updated local environment task state diagram with the large model to assist in generating communication content, ensuring that the negotiation topic is closely related to the current decomposed task, avoiding topic drift and ineffective discussion;

[0128] 3. Improve the flexibility and adaptability of multi-robot systems: No central control node is required, and members can be dynamically added and removed to meet the needs of real-time task changes in complex environments.

[0129] 4. Enhance the intelligence and efficiency of collaboration: Support advanced collaboration content such as semantic expression, capability matching, and task decomposition, so that the negotiation process has contextual understanding and goal orientation.

[0130] Therefore, the communication and negotiation mechanism proposed in the present invention effectively overcomes the problems of frequent deadlocks, divergent discussions, and rigid scheduling in existing solutions, and improves the stability, efficiency, and intelligent collaboration capabilities of multi-robot systems in complex collaborative tasks.

[0131] In addition, the present invention can also support the access of robots of different types and capabilities, realize collaborative operations under heterogeneous platforms, and further expand the application scope of multi-robot collaboration.

[0132] A specific embodiment of the multi-robot autonomous collaboration method of the present invention is given below:

[0133] In a specific embodiment of the present invention, the user issues a fuzzy task goal to the multi-robot system: "Please help me clean up this room." According to step S100, a robot pool containing multiple types of functional robots is first configured, including: Robot-A (such as AlienGo) with navigation and pushing capabilities, Robot-B (such as Stretch) with fine grasping capabilities, Robot-C (such as Cleaner) with cleaning capabilities, and Robot-Car (Transporter) with basic navigation and transportation capabilities. The communication module in the robot pool and the deployed AI model parse the task goal, identify that the task belongs to the "room cleaning" task, and infer that it may include decomposition tasks such as "cleaning the floor", "moving obstacles", and "tidying up the desktop". Since the user did not provide environmental details, the AI ​​model estimated that the task space was a "two-bedroom and one-living room" structure, and accordingly selected Robot-A, Robot-C and Robot-Car from the robot pool to form the current work team.

[0134] In step S200, each member of the current work team activates their proprioception module to perceive the local task environment. Robot-A detects a tabletop and a low obstacle in the left area; Robot-C detects that the floor in the upper right corner is obstructed by a pile of bottles; Robot-Car detects that the path is clear but lacks fine-grained recognition capabilities. Each member broadcasts their local perception results via a wireless LAN communication network, integrating their data to construct an updated local environment task state map.

[0135] In step S300, each team member conducts capability assessments and negotiates communication based on the updated local environment task state map. Robot-C first assesses that its cleaning area is obstructed by a pile of bottles, making it unable to complete the task. It then proactively requests assistance from other team members via a broadcast broadcast over the communication network. Upon receiving this information, Robot-A assesses that while it can push the bottles, the space is limited and there's a risk of them tipping over. It explicitly responds that it cannot do so safely. Robot-Car, in turn, responds that it lacks the grasping capability. After several rounds of negotiation, Robot-A, based on the task status and historical experience, proactively states that the current team lacks high-degree-of-freedom maneuvering capabilities and suggests recruiting Robot-B from the robot pool for assistance. Once the team reaches consensus, Robot-B is activated via the communication network to join the current team. After joining the team, Robot-B activates its grasping module and successfully grasps the stacked bottles, sorting and placing them in a safe area. Robot-C detects that the cleaning area has been restored and resumes its cleaning task. Robot-A addresses the obstacle in the left area, while Robot-Car assists in transporting lightweight items.

[0136] During the task, in step S400, team members continuously synchronize perception, update tasks, and negotiate capabilities through the communication network, autonomous communication, and multi-round negotiation mechanisms. Task assignments and progress status are updated in real time based on their respective local environment task state graphs. Ultimately, the task is considered complete when all task nodes are marked as completed. The task log and collaborative state graph are archived to each robot's memory module, and each member is recalled to the robot pool and switched to standby or hibernation mode.

[0137] This example demonstrates the complete closed-loop execution of the proposed method in a real-world task, highlighting its advantages in coping with ambiguous objectives, dynamic scenarios, and collaborative tasks involving heterogeneous members. The method supports members initiating active negotiations, identifying capability bottlenecks, and reorganizing teams. It also exhibits strong semantic focus and collaborative stability, effectively avoiding communication deadlocks and information redundancy, and implementing a highly adaptable and interpretable multi-robot proactive collaboration mechanism.

[0138] See also Figure 3 The second embodiment of the present invention provides a multi-robot autonomous collaboration system, comprising:

[0139] The robot pool contains multiple robots. The robot pool is configured to select at least one robot from multiple robots as a member to form an initial current work team based on the task objectives issued by the user. At least one robot in the current work team has perception capabilities. All robots in the robot pool share each other's basic configuration information;

[0140] A communication network, configured in the task environment, is used to achieve communication between members of the current work team and between the current work team and the robot pool;

[0141] The current work team is configured to evaluate the capabilities of the current work team based on the updated local environment task state graph formed during the execution of each member's task and based on autonomous communication and multi-round negotiation mechanisms, to form a new current work team and task decomposition and allocation plan. The robot pool responds to the new current work team plan and forms a new current work team in the task environment, wherein the updated local environment task state graph is obtained by each member of the current work team fusing their own local environment task state graph with the local environment task state graph synchronized with other members through the communication network.

[0142] In some embodiments, each robot is equipped with a memory module, an autonomous communication module, a perception module, a hierarchical execution module, and a state assessment module. The functions and specific implementation processes of each module are described as follows:

[0143] 1. Memory Module

[0144] Functional description: Used to continuously record and retrieve key historical information during task execution, such as global perception maps, state transitions, communication history, and task feedback, to provide a basis for subsequent action adjustments and long-term collaboration strategies.

[0145] Specific implementation: Build a task memory cache indexed by timestamp, including graph structure state memory and language interaction log; graph structure state memory contains global perception graph and task state graph, and the task state graph includes records of each round of task planning, path selection, negotiation response and execution feedback; provide read / write interfaces for perception module, execution module and its own large model to call.

[0146] 2. Selective Communication Module

[0147] Function description: Supports robots to independently determine the timing, content and target recipients of communication during task execution, achieving efficient and low-redundancy information exchange.

[0148] Specific implementation: The autonomous communication module determines whether to speak based on the current task state diagram, robot capabilities, and collaborative context. It uses large models (such as the GPT series) to embed the state into the encoded input prompt words to generate structured communication content. Based on the context matching strategy, it selects the recipient (single or multiple) from the current work team and completes the transmission.

[0149] 3. Perception Module

[0150] Functional description: Acquire environmental data within the local field of view of the main body, identify task-related objects and spatial information, and provide input for task mapping and status assessment.

[0151] Specific implementation: The perception module includes multi-source sensors such as RGB camera, depth camera, lidar and IMU; uses target recognition and semantic segmentation algorithms (such as YOLOv5+SAM) to detect object type and location; constructs local grid map and instance segmentation map, and synchronizes them to the global perception map.

[0152] 4. Hierarchical Execution Module

[0153] Function description: Realize the mapping from high-level task objectives to low-level action control, and support the continuous execution and interrupt recovery of complex tasks.

[0154] Specific implementation: Tasks are represented by natural language input or structured instructions (such as "clean the desktop") and decomposed into sub-skill chains through LLM or FSM; mid-level skills call specific modules (such as grasp, navigate, push) to form an orderly execution plan; the bottom-level control connects to the ROS controller, grasping action planner and mobile navigation module to realize physical actions; support behavior tree structure organization plan, with failure recovery and dynamic replacement functions.

[0155] 5. State Evaluation Module

[0156] Function description: Real-time analysis of robot execution status, task completion, and environmental changes to determine whether negotiation is triggered or capability supplementation is required.

[0157] Specific implementation: Sources include: task success mark, object status detection, time limit exceedance, and failure mark (such as graspfailure); based on the state diagram node information and perception results, the current task progress is updated; if an abnormality or bottleneck (such as insufficient capacity, target loss) is detected, the status mark is synchronized to the communication module to initiate negotiation; support linkage with the memory module to achieve behavior mode adjustment.

[0158] It should be noted that the aforementioned explanation of the embodiment of the multi-robot autonomous collaboration method is also applicable to the multi-robot autonomous collaboration system of this embodiment, and will not be repeated here.

[0159] In summary, this implementation fully demonstrates the entire process of the present invention in a complex, real-world environment, from task setting, team initialization, collaborative execution, dynamic reorganization, and system closed-loop control. Through the detailed design and modular implementation described above, those skilled in the art can directly replicate the present invention's technical solutions and achieve efficient, autonomous collaboration in heterogeneous multi-robot systems.

[0160] The above-described embodiments and / or implementations are merely intended to illustrate the preferred embodiments and / or implementations of the present technology, and are not intended to limit the embodiments of the present technology in any form. Any person skilled in the art can make some changes or modifications to other equivalent embodiments without departing from the scope of the technology disclosed in the present disclosure, but should be considered as substantially the same technology or embodiments of the present disclosure.

[0161] The principles and implementations of the present application are described herein using specific examples. The above description of the embodiments is only intended to help understand the method and its core idea of the present application. The above description is only the preferred embodiments of the present application. It should be noted that due to the limited nature of the language, there are objectively infinite specific structures. For ordinary skilled persons in the art, without departing from the principles of the present application, some improvements, refinements or changes can be made, or the above technical features can be combined in an appropriate manner. These improvements, refinements, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, shall be considered as the protection scope of the present application.

Claims

1. A multi-robot autonomous collaboration method, characterized in that: include: Step S100: Build a communication network and configure a robot pool containing multiple robots. All robots share each other's basic configuration information. The robot pool selects at least one robot as a member to form the current work team based on the task objectives issued by the user. Step S200: After the current work team enters the task environment, its members perceive the task environment information and monitor the task status in real time, obtain their respective local environment task status graphs, and synchronize the local environment task status graphs among the members via the communication network to update their respective local environment task status graphs; Step S300: Each member evaluates the capabilities of the current work team based on the updated local environment task status diagram based on autonomous communication and multi-round negotiation mechanisms, and determines whether to reorganize the team and / or decompose and allocate tasks based on the evaluation results, thereby updating the current work team and task decomposition and allocation plan, and each member performs the assigned subtasks based on the updated plan.

2. The multi-robot autonomous collaboration method according to claim 1, characterized in that: The robot pool is configured outside the task environment, and the robots contained in the robot pool are heterogeneous robots or homogeneous robots; each robot maintains communication with the robot pool and other members of the current work team through its own wireless communication module, supporting point-to-point communication and broadcast mechanisms.

3. The multi-robot autonomous collaboration method according to claim 1, characterized in that: The robot pool is equipped with an AI model. For clear or vague task objectives issued by users, the AI ​​model selects members based on task understanding and ability matching to build an initial current work team.

4. The multi-robot autonomous collaboration method according to claim 1, characterized in that: The task environment information perceived by the members includes structural information of the task execution space, object information related to the task goal, potential obstacles and operational constraints, and dynamic change information; The task status monitored by the members in real time includes the execution stage of the current task assignment, the current execution status of the members, the completion mark of the task goal, whether the preset timeout threshold is triggered, the failure events that occur in the task collaboration, and whether there is an intermediate state that requires adjusting the task strategy or reorganizing the team.

5. The multi-robot autonomous collaboration method according to claim 1, characterized in that: Each member is driven by the local clock and regularly uploads the local environment task status graph. At the same time, it receives and integrates information from other members to update its own local environment task status graph.

6. The multi-robot autonomous collaboration method according to claim 1, characterized in that: The updated local environment task state diagram is a multimodal, layered semantic task state diagram, and the updating process includes: Step S210, node extraction: extract the following four types of semantic nodes based on the task objectives and information synchronized by other members: T-node: represents the subtask to be completed and its location; A-node: represents the members who have joined the team and their current location, ability tags and status information; E-node: represents the binding relationship between the currently executing subtask and the executor; U-node: indicates an area or task point that has not been explored or has insufficient information, and is used to indicate areas that need to be sensed later; Step S220, layer division: Divide the local environment task state graph into three semantic layers: Task layer: includes all task nodes to be completed or being executed; Spatial layer: contains the environment spatial structure nodes and the spatial dependency relationship between tasks and the environment; Agent layer: includes team member nodes and their perception, execution, and communication status; Step S230: Edge connection: In the task layer, directed edges are established based on the dependency order between tasks; In the spatial layer, undirected edges are established based on spatial adjacency; In the proxy layer, edges connect member nodes with the task nodes they are responsible for, indicating the current task allocation situation; Nodes with unassigned tasks or areas to be explored are connected to A-nodes or U-nodes through virtual edges to indicate scheduling requirements; Step S240, graph fusion and update: After each member has constructed the environment task status graph, it synchronizes state changes through the communication network and performs graph merging and deduplication operations to obtain an updated local environment task status graph.

7. The multi-robot autonomous collaboration method according to claim 1, characterized in that: Step S300 includes: Step S310: Each member of the current work team obtains self-assessment information based on the updated local environment task state diagram, combined with their own capabilities, and the matching degree between the task requirement dimension and the robot capability dimension. If the member's self-assessment information indicates that the member has the intention to communicate, the process proceeds to step S320; otherwise, the member will continue to perform the currently assigned subtask; In step S320, members with communication intentions send their self-assessment information to other relevant members via the communication network, entering into an autonomous communication and multi-round negotiation process. During this process, the communicating members compare each other's adaptability, execution efficiency, and task load information to achieve one or a combination of the following negotiation goals: When the current work team has a capacity bottleneck, reorganize the team, recruit new members with matching capabilities from the robot pool, and update the current task status; When there is redundant member resources in the current work team, the member is marked as releasable, returned to the robot pool, and the current task status is updated; When the current task status changes, each member divides the subsequent task objectives based on their updated local environment task status diagram and redistributes them to the corresponding members.

8. The multi-robot autonomous collaboration method according to claim 7, characterized in that: The communication intention is generated when the member detects any one of the following situations or a combination of multiple situations: new observation of the task environment, update of task-related information, update of the task execution status and estimation of the effect of task execution.

9. The multi-robot autonomous collaboration method according to claim 1, characterized in that: The autonomous communication and multi-round negotiation mechanism includes: Step S321: Initialize members who have communication intentions into a waiting-to-speak queue; Step S322: Determine whether there are any members in the waiting-to-speak queue. If not, exit the current round of negotiation, and each member continues to perform the currently assigned subtask until one or more members express the intention to communicate and enter the next round of negotiation. If yes, proceed to step S323. Step S323: The first person in the waiting-to-speak queue is removed from the queue and becomes the current speaker. The current speaker creates a communication content based on their communication intention and selects a recipient to send the communication content to. The current speaker and recipient update their respective local communication message records. If the recipient is not in the current speaking queue, the recipient is added to the end of the waiting-to-speak queue, and the process proceeds to step S324. Step S324: If the communication intention of the first member in the waiting-to-speak queue disappears, delete him from the waiting-to-speak queue and return to step S322; if not, return to step S323.

10. A multi-robot autonomous collaborative system, characterized in that: include: A robot pool containing multiple robots. The robot pool is configured to select at least one robot from the multiple robots as a member to form an initial current work team based on the task objectives issued by the user. All robots in the robot pool share each other's basic configuration information; A communication network is configured in the task environment and is used to realize communication between members of the current work team and between the current work team and the robot pool; The current work team is configured to evaluate the capabilities of the current work team based on the updated local environment task state graph formed during the execution of each member's task and based on autonomous communication and multi-round negotiation mechanisms, to form a new current work team and task decomposition and allocation plan. The robot pool responds to the new current work team plan and forms a new current work team in the task environment, wherein the updated local environment task state graph is obtained by each member of the current work team fusing their own local environment task state graph with the local environment task state graph synchronized with other members through the communication network.

Images