Distributed robot collaborative scheduling system and method in dynamic environment

Abstract

The invention relates to the technical field of robot scheduling, in particular to a distributed robot collaborative scheduling system and method in a dynamic environment, and the system comprises an environment sensing layer which is used for collecting environment dynamic data in real time; the distributed decision-making layer comprises local task scheduling modules of a plurality of robots; the cooperative communication layer is used for realizing task state synchronization and conflict detection among the robots based on a low-delay communication protocol; the dynamic weight calculation module is used for generating a real-time optimization weight according to the task emergency degree, the robot energy consumption and the path risk factor; according to the invention, by using a completely distributed collaborative scheduling architecture, through a decentralized task distribution mechanism and a distributed consensus protocol, a single-point fault risk existing in a traditional centralized scheduling system is thoroughly eliminated, and even if a part of robot nodes have faults or communication is interrupted, the system can work normally. And the system can still run continuously through autonomous negotiation of the remaining nodes, so that the reliability of the system in a complex environment is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot scheduling, in particular to a distributed robot collaborative scheduling system and method in dynamic environment. BACKGROUND

[0002] Due to the rapid development of industrial automation, intelligent warehousing and unmanned systems, robot scheduling technology has become a key means to improve work efficiency and reduce operating costs. In dynamic environments such as logistics and warehousing, intelligent manufacturing and disaster rescue, multi-robot systems need to be highly efficient in order to complete complex tasks such as cargo handling, path planning and resource allocation. Existing robot scheduling systems mainly use centralized control or static distributed strategies to achieve task allocation and path coordination through a central server or fixed rules. Such systems usually rely on global information, use optimization algorithms or heuristic rules for scheduling decisions, and have achieved certain results in some scenarios. However, as the complexity of application scenarios and the scale of robots increase, the existing technology gradually exposes many limitations in actual deployment.

[0003] The centralized scheduling system in the prior art relies heavily on the central control node. Once the server fails or the communication delay increases, the robustness and real-time performance of the entire system will be greatly reduced, making it difficult to adapt to the needs of high dynamic environments. Static distributed scheduling strategies usually use fixed rules or predefined weight allocation tasks, which cannot be dynamically adjusted according to changes in the environment and the priority of the task, resulting in low task execution efficiency and even frequent conflicts. Traditional collaborative algorithms have a sharp increase in communication overhead when the number of robots is large, which can easily cause network congestion and affect the real-time performance of scheduling decisions. Existing systems are not good at conflict detection and recovery mechanisms. Most solutions only provide simple collision avoidance strategies and lack efficient dynamic rescheduling capabilities, resulting in long recovery time after task interruption and affecting the overall system performance. Because of these problems, the existing robot collaborative scheduling system is difficult to operate stably in highly dynamic and large-scale distributed scenarios, limiting its promotion in actual industrial applications.

[0004] Therefore, in view of the above problems, the present application provides a distributed robot collaborative scheduling system and method in dynamic environment, which realizes efficient autonomous collaboration of robots in complex environments through a bidding task allocation mechanism, a dynamic weight optimization algorithm and a real-time conflict detection technology, supports local decision-making and distributed consensus of robots, and can dynamically adjust the scheduling strategy according to changes in the environment and task requirements, thereby significantly improving the robustness, real-time performance and scalability of the system. SUMMARY

[0005] In order to overcome the problem that the existing robot collaborative scheduling system is difficult to stably run in a highly dynamic and large-scale distributed scene, the application provides a distributed robot collaborative scheduling system and method in a dynamic environment.

[0006] The technical scheme of the application is as follows: a distributed robot collaborative scheduling system in a dynamic environment, comprising: An environment perception layer comprising a plurality of sensor nodes deployed in a work area, which collect dynamic environment data in real time; A distributed decision-making layer comprising a plurality of local task scheduling modules of robots, each module being internally provided with a space-time constraint reasoning engine and a bidding task allocation algorithm; A collaborative communication layer for realizing task state synchronization and conflict detection among robots based on a low-delay communication protocol; A dynamic weight calculation module for generating real-time optimization weights according to task urgency, robot energy consumption and path risk factors.

[0007] Preferably, the execution steps of the bidding task allocation algorithm comprise: decomposing a global task into a set of subtasks and broadcasting them to all robots, each robot generating a cost function value for each subtask based on its own state through local calculation and submitting a bid value, the system performing multiple rounds of verification and synchronization on all bids using a distributed consensus protocol, and finally determining the winning robot and updating the task allocation table by the robot group participating in the bidding.

[0008] Preferably, the calculation formula of the cost function is , wherein is a task execution time estimate, calculated based on the robot's movement speed and path length, is an energy consumption estimate, modeled by motor power and running time, is a path risk coefficient, generated by quantifying the density of environmental obstacles and the ground friction coefficient, and the weight coefficient is adjusted by the dynamic weight calculation module according to real-time environment data and robot state, and the weight coefficient satisfies .

[0009] Preferably, the conflict detection uses a space-time corridor model, which pre-allocates a time window and a spatial coordinate range for each robot's path, broadcasts the space-time occupation state in the collaborative communication layer, and when it detects that the pre-determined paths of multiple robots overlap in space within the same time window, triggers a conflict warning signal and starts a priority-based rescheduling process, wherein the priority is determined by the task urgency and the robot's current load, and the conflict resolution strategy includes path delay, local detour or task reallocation.

[0010] As preferred, the system further comprises an incremental map updating module, which broadcasts the obstacle coordinates when the sensor detects an environmental mutation, and the local path planner of each robot receives the updating data to update the path in real time.

[0011] As preferred, the task scheduling module comprises a task preemption mechanism: when a high-priority task is detected, the system generates an interrupt instruction and sends it to the robot executing a low-priority task, which immediately saves the current task state and releases the occupied resources, calculates the shortest path to the starting point of the new task, and broadcasts the task switching notification to the cooperative communication layer to trigger the state synchronization of the related robots, and the interrupted task is redistributed by the distributed decision layer after the preemption is completed.

[0012] As preferred, a distributed robot cooperative scheduling method in a dynamic environment comprises the following steps: S1, obtaining the dynamic obstacle distribution and task location information through the environment perception layer; S2, decomposing the task into an atomic operation sequence by the robot initiating the task and generating a bid request; S3, each competing robot calculates a bid value based on the energy state, position and load capacity; S4, a lightweight Byzantine fault tolerance protocol is used to reach a distributed consensus on the bid results.

[0013] As preferred, the specific implementation steps of the distributed consensus include: each robot participating in the bidding encapsulates the locally calculated bid value and digital signature into a blockchain transaction structure, broadcasts it to all nodes in the cluster through the network, and performs a three-stage verification process including pre-preparation, preparation and confirmation, and writes the local task allocation table when receiving confirmation messages from more than 2 / 3 nodes.

[0014] As preferred, during the task execution process, if an environmental mutation or robot failure is detected, dynamic rescheduling is triggered.

[0015] As preferred, the steps of the dynamic rescheduling include: first, the abnormal robot or the adjacent monitoring node broadcasts a state alarm containing the fault code and position, the system analyzes the affected sub-task set based on the task dependency graph, reinitiates the distributed bidding allocation process in the local range, and starts the task migration mechanism of the faulty robot to safely transfer its unfinished sub-tasks and context state to the replacement robot.

[0016] The present application has the following advantages: 1. The application adopts a fully distributed cooperative scheduling architecture, through a decentralized task allocation mechanism and a distributed consensus protocol, completely eliminating the single point failure risk existing in traditional centralized scheduling systems, even if part of the robot nodes fail or communication is interrupted, the system can continue to run through the autonomous negotiation of the remaining nodes, thereby significantly improving the reliability of the system in complex environments.

[0017] 2. By introducing a dynamic weight calculation module and a real-time environment perception mechanism, the system can automatically optimize the scheduling strategy according to dynamic factors such as sudden obstacles, changes in robot power, and task priority adjustments, improving the task completion efficiency in simulated warehouse environments and reducing the path conflict rate, thereby effectively solving the problem that static strategies are difficult to adapt to environmental changes.

[0018] 3. The lightweight consensus protocol and incremental map update mechanism designed by the application convert the global information exchange required by traditional distributed systems into local data synchronization, enabling the system to support larger-scale robot collaborative work.

[0019] 4. The conflict detection method based on the space-time corridor model and the multi-priority task preemption mechanism can complete conflict detection and generate an optimized solution within 50ms, significantly shortening the conflict resolution time, and through distributed logging, all operations are traceable, avoiding the common deadlock and resource competition problems in traditional systems. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The workflow diagram of the application is shown. DETAILED DESCRIPTION

[0021] To make the purpose, technical scheme and advantages of the embodiments of the application clearer, the technical scheme in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, not all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0022] The application provides an embodiment: a distributed robot cooperative scheduling system in dynamic environment: The system specifically comprises the collaborative work of an environment perception layer, a distributed decision layer, a collaborative communication layer and a dynamic weight calculation module, the environment perception layer collects dynamic data of the environment in real time through various sensors (such as laser radar, visual camera, infrared sensor) arranged in the working area, including the position of obstacles, the trajectory of moving objects and the terrain features of the environment; the distributed decision layer is composed of task scheduling modules carried by each robot, each module is internally provided with a space-time constraint reasoning engine and a bidding task allocation algorithm, wherein the space-time constraint reasoning engine performs task time sequence analysis based on an improved space-time Petri net model, and each robot can independently calculate an optimal task scheme; the collaborative communication layer adopts a wireless Mesh network architecture based on the IEEE 802.11ac protocol, supports multi-hop transmission and a TDMA time division multiplexing mechanism, so as to realize millisecond-level synchronization and conflict detection of the task state among the robots, wherein the conflict detection adopts a distributed ledger technology based on a block chain to ensure that the data cannot be tampered with; the dynamic weight calculation module dynamically generates optimization weights in real time according to the task urgency, the remaining power of the robot and the path risk factor; and the application can significantly improve the overall performance of the multi-robot system in a dynamic environment.

[0023] Further, the bidding task allocation algorithm is described as follows: firstly, the global task is intelligently disassembled into a set of atomic subtasks that can be executed in parallel, and is distributed to all robot nodes in the cluster through a low-delay broadcast protocol; after receiving the task package, the local scheduling module of each robot starts multi-dimensional state evaluation immediately, comprehensively calculates key parameters such as the path length of the current position, the current battery remaining power and the load capacity of the execution mechanism, and generates an accurate bid evaluation value using a cost function formula; after all bid values are verified for rationality by each node independently, when more than 2 / 3 of the nodes receive the confirmation message, an unforgeable winning result is generated and written into the block chain task allocation table of each node; and the application can still maintain 100% scheduling function when the central server fails.

[0024] The calculation formula of the cost function is , wherein is the task execution time estimation, calculated based on the robot movement speed and the path length, is the energy consumption estimation, modeled by the motor power and the running time, is the path risk coefficient, generated by quantifying the environmental obstacle density and the ground friction coefficient, and the weight coefficient is adjusted by the dynamic weight calculation module according to real-time environmental data and the state of the robot, and the weight coefficient satisfies .

[0025] Further, the calculation of the cost function is realized by a dynamic weight calculation module for multi-objective optimization scheduling, and the working process is specifically as follows: when the robot receives a task to be allocated, the local scheduler first calculates the path length and motion parameters through the navigation system, and accurately estimates the task execution time in combination with the motor performance curve At the same time, the real-time current and voltage data of the battery management system are read, and the estimated energy consumption is calculated based on the power integration model The environmental perception system synchronously scans the work area, and the path risk coefficient is quantified by comprehensively analyzing three elements of obstacle density, ground friction coefficient detection and moving object trajectory prediction The dynamic weight calculation module dynamically adjusts the weight distribution of alpha, beta and gamma according to real-time monitoring of environmental mutation signals such as sudden obstacle approaching distance less than 2 meters, robot remaining power and task priority label, and adopts a fuzzy logic controller to dynamically adjust the weight distribution of alpha, beta and gamma, and forcibly satisfies alpha+beta+gamma=1, and finally outputs the comprehensive cost value after weighted summation; compared with the fixed weight scheme, the average task completion time of the robot is shortened by 32% in the logistics and warehousing, the collision avoidance success rate in the sudden obstacle scene is improved to 99.7%, and the battery endurance time is prolonged by 28%, through the weight dynamic adjustment mechanism, the system can flexibly switch between the energy efficiency priority mode (beta=0.6) or the time efficiency priority mode (alpha=0.7) under the premise of maintaining the path safety.

[0026] The conflict detection adopts a space-time corridor model, and by pre-allocating a time window and a spatial coordinate range for each robot, a space-time occupation state is broadcast in the cooperative communication layer, when it is detected that the scheduled paths of multiple robots exist spatial overlap in the same time window, a conflict warning signal is triggered and a priority-based rescheduling process is started, wherein the priority is determined by the task urgency and the current load of the robot, and the conflict resolution strategy includes path delay, local detour or task reassignment.

[0027] Further, each robot discretizes the scheduled motion trajectory into a space-time four-dimensional data packet containing three-dimensional coordinates (X, Y, Z) and a time stamp (T) in the path planning stage, and periodically sends it to the neighbor nodes through broadcasting, each robot locally maintains a dynamically updated space-time occupation map, when receiving the path data packet of other nodes, the conflict detection engine immediately performs four-dimensional space intersection operation, and judges the collision risk by comparing the spatial coordinate overlap degree in the time window, if the overlap degree>85%, the system automatically triggers a multi-level response strategy: for low priority tasks, 5-15% of the path speed adjustment is performed, for equal priority tasks, local path re-planning based on A* algorithm is started, and for emergency tasks, the preemption mechanism is activated, and all conflict decisions need to be confirmed by at least 3 neighbor nodes through a distributed verification protocol before execution.

[0028] The system also comprises an incremental map updating module, which broadcasts obstacle coordinates when the sensor detects environmental mutations such as newly added obstacles with a size exceeding 30cm*30cm or a change in the width of an original passage exceeding 20%, and each robot updates its local path planner in real time based on the RRT* algorithm after receiving the updated data, wherein the algorithm introduces a dynamic step adjustment mechanism to automatically reduce the search step to 5cm in narrow areas to ensure the feasibility of the path.

[0029] The task scheduling module comprises a task preemption mechanism: when a high-priority task is detected, the system generates an interrupt instruction and sends it to a robot executing a low-priority task, which immediately saves the current task state and releases the occupied resources, calculates the shortest path to the starting point of the new task, and broadcasts a task switching notification to the collaborative communication layer to trigger state synchronization of related robots, and the interrupted task is redistributed by the distributed decision layer after preemption is completed.

[0030] Please refer to Figure 1 Further, the present application provides an embodiment, a distributed robot collaborative scheduling method for dynamic environment, which specifically comprises the following steps: The system obtains dynamic obstacle distribution and task location information through the environment perception layer, and a robot initiating a task uses a topological decomposition algorithm to decompose a complex task into atomic operation sequences, such as "picking up goods-transporting-placing goods" basic units, each sub-task is attached with time and space constraint conditions and priority labels, and a standardized bidding request data packet is generated and broadcast to all available robot nodes in the cluster through a low-latency communication network.

[0031] Each receiving robot runs a local cost calculation engine based on multi-dimensional state detection data, generates an accurate bid value by referencing a dynamic weight formula, and sends a bidding data packet containing digital signature, robot ID and available time window to the task initiation node and neighborhood verification node through a multicast protocol, wherein the multi-dimensional state detection data includes remaining power, current position and robot arm load.

[0032] A lightweight Byzantine fault tolerance protocol is used to perform distributed verification on the bidding results, which specifically implements a three-stage processing flow: in the pre-preparation stage, the task initiation node packs valid bids to generate a proposal block, in the preparation stage, the verification node confirms the validity of the bid through threshold signature technology, and in the confirmation stage, after collecting more than 2 / 3 of the nodes' agreement response, the final allocation result is written into the blockchain task table of each node, and the whole process takes an average of 120ms.

[0033] During task execution, the robot failure or environmental mutation is detected by the real-time health monitoring module, when the gyroscope is abnormal, the motor is overloaded or the sudden obstacle distance is less than the safety threshold, the faulty robot immediately broadcasts an encrypted alarm data packet containing abnormal code and position information, and starts the local safety protection program.

[0034] The system analyzes the affected subtask set based on the pre-constructed task dependency graph, automatically identifies the job units that must be reassigned and their associated constraints before and after, prioritizes the neighborhood nodes of the original execution robot when restarting the bidding process in a local range to reduce communication overhead, and synchronously migrates the task context data through the distributed ledger technology, so as to ensure that the rescheduling process is completed within 300ms and maintains data consistency, the context data includes the completed subtask state and resource occupation.

[0035] An embodiment of the present application provides a system for intelligent warehouse logistics scenario. Deploying the system in a large e-commerce warehouse center, configuring 150 AGV robots with a load of 1 ton, when the central management system receives a customer order, the system automatically disassembles the composite order into atomic tasks such as "shelf positioning-goods picking-turnover transportation-packaging table distribution", through a distributed bidding mechanism, each AGV generates differentiated bids based on real-time power and current location, and finally three AGVs cooperatively complete a single order, the actual measurement shows that the order processing efficiency is improved by 55% compared with the traditional WMS system, and at the same time, due to the distributed architecture, the risk of system paralysis caused by single point failure of the central server is avoided.

[0036] An embodiment of the present application provides a system for intelligent manufacturing production line scenario. The present application is applied to a flexible production line of automobile parts, 32 collaborative robots form a dynamic production unit, when a new model workpiece enters the production line, the system automatically generates a task chain including processes such as "clamping-machining-detection-transportation" according to the workpiece process route, and completes the process allocation within 80ms through a practical Byzantine fault tolerance protocol, wherein the high-precision machining process is preferentially allocated to a robot with a repeated positioning accuracy of ±0.02mm, and the heavy-load transportation task is won by a robot with a load capacity of 500kg, cooperating with a dynamic rescheduling mechanism, when a robot is automatically offline due to tool wear, its unfinished process is reallocated to the adjacent two robots within 200ms to ensure the continuous operation of the production line.

[0037] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can modify or change the disclosed technical content into equivalent embodiments with equivalent changes, and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the technical solution of the present application, still falls within the protection scope of the present application.

Claims

1. A distributed robot collaborative scheduling system for dynamic environments, characterized in that, Comprise: An environmental perception layer comprising a plurality of sensor nodes deployed in the work area to collect real-time environmental dynamic data; A distributed decision-making layer comprising a plurality of local task scheduling modules for robots, each module having a built-in spatiotemporal constraint reasoning engine and a bidding task allocation algorithm; A collaborative communication layer based on a low-latency communication protocol to achieve task state synchronization and conflict detection among robots; A dynamic weight calculation module to generate real-time optimization weights based on task urgency, robot energy consumption, and path risk factors.

2. The dynamic environment distributed robot collaborative scheduling system according to claim 1, wherein, The execution steps of the bidding task allocation algorithm include: decomposing the global task into a set of subtasks and broadcasting them to all robots, each robot generating a cost function value for each subtask based on its own state and submitting a bid value, the system uses a distributed consensus protocol to verify and synchronize all bids in multiple rounds, and finally determines the winning robot and updates the task allocation table through the collective efforts of the bidding robot group.

3. The dynamic environment distributed robot collaborative scheduling system according to claim 2, wherein: The calculation formula of the cost function is wherein is the task execution time estimate, is the task energy consumption estimate, is the path risk coefficient, generated by quantifying the environmental obstacle density and the ground friction, is the dynamic weight coefficient, and the weight coefficient is adjusted by the dynamic weight calculation module according to real-time environmental data and the robot state, and the weight coefficient satisfies .

4. The distributed robotic coordination system for dynamic environments of claim 1, wherein: The conflict detection uses a spatiotemporal corridor model, which pre-allocates a time window and spatial coordinate range for each robot's path, and broadcasts the spatiotemporal occupancy state in the collaborative communication layer. When detecting that multiple robots' scheduled paths overlap in the same time window, a conflict warning signal is triggered and a priority-based rescheduling process is initiated, where the priority is determined by the task urgency and the robot's current load. Conflict resolution strategies include path delay, local detour, or task redistribution.

5. The distributed robotic coordination system for dynamic environments of claim 1, wherein: The system also includes an incremental map update module that broadcasts obstacle coordinates when the sensor detects environmental mutations. Each robot's local path planner updates the path in real time after receiving the update data.

6. The dynamic environment distributed robot collaborative scheduling system of claim 1, wherein, The task scheduling module includes a task preemption mechanism: when a high-priority task is detected, the system generates an interrupt instruction and sends it to the robot executing a low-priority task. The robot immediately saves the current task state and releases the occupied resources, calculates the shortest path to the new task starting point, and broadcasts the task switching notification to the collaborative communication layer to trigger the state synchronization of related robots. After preemption is complete, the distributed decision-making layer redistributes the interrupted task.

7. A method for distributed robot collaborative scheduling in dynamic environment, using the distributed robot collaborative scheduling system in dynamic environment according to any one of claims 1-6. Comprise the following steps: S1, obtain dynamic obstacle distribution and task location information through the environmental perception layer; S2, decompose the task into atomic operation sequences and generate a bid request from the robot that initiates the task; S3, each competing robot calculates a bid value based on energy state, location, and load capacity; S4, use a lightweight Byzantine fault-tolerant protocol to achieve distributed consensus on the bid results.

8. The method of claim 7, wherein, The specific implementation steps of the distributed consensus include: each bidding robot encapsulates the locally calculated bid value and digital signature into a blockchain transaction structure, broadcasts it to all nodes in the cluster through the network, and performs a three-stage verification process including pre-preparation, preparation, and confirmation. When receiving confirmation messages from more than 2 / 3 nodes, write to the local task allocation table.

9. The method of claim 7, wherein: During task execution, if an environmental mutation or robot failure is detected, dynamic rescheduling is triggered.

10. The method of claim 9, wherein, The dynamic rescheduling step includes: first, by the abnormal robot or the adjacent monitoring node, a state alarm containing a fault code and a position is broadcasted, the system analyzes the affected sub-task set based on the task dependency graph, re-initiates the distributed bidding allocation process in a local range, and starts the task migration mechanism of the faulty robot to safely transfer its unfinished sub-tasks and context state to the replacement robot.

Images