Distributed robot cooperative control method and system based on edge computing

Abstract

The invention provides a distributed robot cooperative control method and system based on edge computing. Environmental space data and dynamic obstacle motion parameters of robots are collected, and cross-robot joint compression is carried out to obtain a distributed sensing data set. Then, based on the spatial density difference of the data set, hierarchical communication links in the group and the priorities of signal transmission paths of the hierarchical communication links are dynamically constructed and adjusted in real time. Then, the adjusted priority is converted into a communication weight parameter, a cooperative control instruction set is generated in combination with an obstacle motion parameter, and cross-node synchronization is carried out on the cooperative control instruction set and obstacle information sensed by each node; and finally, the synchronized instruction set is utilized to drive the robot to execute edge computing resource interaction, and a cooperative motion path is generated. According to the technical scheme provided by the invention, the communication efficiency and the adaptive capability of distributed robot cooperative control can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of distributed robot collaborative control, and in particular to a distributed robot collaborative control method and system based on edge computing. Background Art

[0002] In scenarios such as industrial automation, smart logistics, and disaster relief, multi-robot systems must collaborate to complete tasks in dynamic environments while avoiding moving obstacles. These scenarios place high demands on the systems for real-time performance, robustness, and low communication dependency. Robots must rapidly respond to changes in obstacle motion through local perception and distributed decision-making, while also implementing path planning and conflict avoidance without centralized control. Furthermore, the systems must adapt to complex and changing environmental uncertainties, such as unknown obstacle motion patterns and sensor noise interference, which poses significant challenges to the flexibility and reliability of collaborative control algorithms.

[0003] A typical approach to dynamic obstacle avoidance currently involves a distributed collaborative control algorithm based on a multi-agent system. This approach leverages local information exchange and a consensus protocol to enable robots to dynamically adjust their trajectories based on the status of neighboring nodes. Specifically, the robots use visual sensors to obtain information about the positions of surrounding obstacles and their companions. This information is combined with a time prediction model to estimate the future trajectory of obstacles, and then an optimized control strategy is used to generate an obstacle avoidance path. Simultaneously, the system transmits key status data via a distributed communication network, ensuring global behavioral consistency and local decision-making independence, thereby achieving efficient collaboration without the need for a central controller.

[0004] Although the above scheme demonstrates certain advantages in dynamic obstacle avoidance, it still has significant limitations. First, information exchange relies on a local communication network, which is susceptible to signal interference or delays in complex environments, leading to path planning delays and even conflicts. Second, this method requires high prediction accuracy of the obstacle motion model. If the actual obstacle behavior deviates from the predicted model, avoidance failure may occur. In addition, the distributed consensus protocol is prone to slow response speed due to competition for computing resources in high-density robot clusters, making it difficult to meet the real-time requirements of large-scale systems. These issues limit the practical application of this scheme in highly dynamic and complex scenarios. Summary of the Invention

[0005] The present application provides a distributed robot collaborative control method and system based on edge computing to solve the problems of low communication efficiency and poor adaptability of distributed robot collaborative control in the prior art.

[0006] In a first aspect, the present application provides a distributed robot collaborative control method based on edge computing, comprising:

[0007] Collecting environmental spatial data of distributed robots and motion parameters of dynamic obstacles, and performing cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set;

[0008] Based on the spatial density differences of the distributed perception data sets, dynamically constructing hierarchical communication links in the distributed robot group, and adjusting the priorities of signal transmission paths in the hierarchical communication links according to real-time changes in the spatial density differences to obtain adjusted signal transmission path priorities;

[0009] Converting the adjusted signal transmission path priority into a communication path weight parameter, and generating a collaborative control instruction set in combination with the motion parameter;

[0010] Synchronizing the collaborative control instruction set with the dynamic obstacle perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set;

[0011] The distributed robot is driven by the synchronized collaborative control instruction set to perform resource interaction behavior of edge computing to obtain the collaborative motion path of the distributed robot.

[0012] Optionally, converting the adjusted signal transmission path priority into a communication path weight parameter and generating a collaborative control instruction set in combination with the motion parameter includes:

[0013] Dividing the hierarchical communication links into intervals according to the adjusted level values ​​of the signal transmission path priorities, and converting the level values ​​within the intervals to obtain weight parameters corresponding to the hierarchical communication links;

[0014] Extracting a direction change amount and a speed change amount from the motion parameters, and superimposing and merging the direction change amount and the speed change amount into a dynamic threat coefficient;

[0015] Performing joint constraint processing on the weight parameter and the dynamic threat coefficient to obtain a weight parameter distribution result, and constructing an instruction generation template corresponding to each layer in the layered communication link based on the weight parameter distribution result;

[0016] The dynamic threat coefficient is input into the instruction generation template to generate a collaborative control instruction set including distributed robot motion constraint conditions.

[0017] Optionally, performing joint constraint processing on the weight parameter and the dynamic threat coefficient to obtain a weight parameter distribution result includes:

[0018] Dividing the variation range of the dynamic threat coefficient into continuous subintervals, each subinterval corresponding to a conversion function;

[0019] According to the hierarchical position corresponding to the weight parameter in the hierarchical communication link, obtaining an initial conversion factor associated with the hierarchical position;

[0020] In the subinterval, performing an association calculation on the dynamic threat coefficient and the initial conversion factor according to the conversion function to obtain an associated subinterval parameter;

[0021] The associated sub-interval parameters are subjected to interval boundary fusion to obtain fused parameters, and inter-level balance constraints are executed on the fused parameters to obtain a weight parameter distribution result.

[0022] Optionally, performing an associative calculation on the dynamic threat coefficient and the initial conversion factor according to the conversion function within the subinterval to obtain an associated subinterval parameter includes:

[0023] Determining a slope adjustment factor of the conversion function according to the span range of the subinterval, and scaling the dynamic threat coefficient according to the slope adjustment factor to generate a dynamic threat value;

[0024] In the subinterval, a level adjustment factor is obtained according to the level position corresponding to the weight parameter, and the dynamic threat value is cross-correlated with the level adjustment factor to obtain an initial conversion factor;

[0025] An interval accumulation process is performed on the initial conversion factor, and the accumulation process result is superimposed on the dynamic threat value to generate an associated sub-interval parameter.

[0026] Optionally, the collecting of environmental spatial data of the distributed robots and motion parameters of dynamic obstacles, and performing cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set includes:

[0027] Dividing the spatial perception range of the distributed robot into overlapping areas, and dividing the environmental spatial data into discrete block data sets containing redundant identifiers;

[0028] In the discrete block data set, repeated feature matching processing is performed on the overlapping area data of adjacent distributed robots, and redundant data points exceeding a set repetition threshold are deleted to generate a de-redundant block data set;

[0029] Extracting high-density feature points and static environment feature points in the motion parameters of dynamic obstacles from the de-redundant block data set, and separating the high-density feature points from the static environment feature points to generate a dynamic feature data set and a static feature data set;

[0030] The dynamic feature dataset is subjected to trajectory extension prediction according to the movement direction of the dynamic obstacle to obtain an extended dynamic feature sequence, and the extended dynamic feature sequence is spatially spliced ​​with the static feature dataset to generate a distributed perception dataset.

[0031] Optionally, performing trajectory extension prediction on the dynamic feature dataset according to the movement direction of the dynamic obstacle to obtain an extended dynamic feature sequence includes:

[0032] Extracting obstacle position coordinates for consecutive time periods from the dynamic feature data set, and calculating a motion vector of the dynamic obstacle based on a change in the obstacle position coordinates;

[0033] Taking the obstacle position coordinates in the current time period as a reference point, the velocity distance of the dynamic obstacle is extended along the direction of the motion vector by a preset multiple to generate a predicted trajectory point sequence;

[0034] A spatial credibility verification process is performed on the predicted trajectory point sequence, and the verified predicted trajectory point sequence is timestamp aligned and merged with the dynamic feature dataset to obtain an extended dynamic feature sequence.

[0035] Optionally, the step of driving the distributed robot to perform edge computing resource interaction behavior through the synchronized collaborative control instruction set to obtain a collaborative motion path of the distributed robot includes:

[0036] Decomposing the synchronized collaborative control instruction set into an action sequence corresponding to the distributed robot, wherein the action sequence includes a target displacement and an execution priority tag;

[0037] Allocating a resource occupancy ratio of edge computing to the distributed robot based on the execution priority label, and determining a computing coverage range of dynamic obstacles according to the resource occupancy ratio;

[0038] Dynamically adjusting the boundary conditions of the calculation coverage range according to the target displacement and the resource occupancy ratio, and calculating the overlapping buffer area of ​​the distributed robot in the moving direction based on the adjusted calculation coverage range;

[0039] The motion parameters of the dynamic obstacle and the displacement state of the distributed robot are aligned and then input into the overlapping buffer area to generate a coordinated motion path of the distributed robot.

[0040] In a second aspect, the present application provides a distributed robot collaborative control system based on edge computing, including:

[0041] An acquisition module collects environmental spatial data of distributed robots and motion parameters of dynamic obstacles, and performs cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set;

[0042] an adjustment module, which dynamically constructs hierarchical communication links in the distributed robot group based on the spatial density differences of the distributed perception data set, and adjusts the priorities of signal transmission paths in the hierarchical communication links according to real-time changes in the spatial density differences to obtain adjusted signal transmission path priorities;

[0043] a generation module, converting the adjusted signal transmission path priority into a communication path weight parameter, and generating a collaborative control instruction set in combination with the motion parameter;

[0044] A synchronization module synchronizes the collaborative control instruction set with the dynamic obstacle perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set:

[0045] A driving module drives the distributed robot to perform resource interaction behavior of edge computing through the synchronized collaborative control instruction set to obtain a collaborative motion path of the distributed robot.

[0046] In a third aspect, an embodiment of the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a distributed robot collaborative control method based on edge computing as described in the first aspect above.

[0047] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, it implements a distributed robot collaborative control method based on edge computing as described in the first aspect.

[0048] In an embodiment of the present application, environmental spatial data of a distributed robot and motion parameters of dynamic obstacles are collected, and the environmental spatial data are subjected to cross-robot joint compression processing to obtain a distributed perception data set; based on the spatial density differences of the distributed perception data set, a hierarchical communication link is dynamically constructed in the group of distributed robots, and the priority of the signal transmission path in the hierarchical communication link is adjusted according to the real-time changes of the spatial density differences to obtain an adjusted signal transmission path priority; the adjusted signal transmission path priority is converted into a communication path weight parameter, and a collaborative control instruction set is generated in combination with the motion parameters; the collaborative control instruction set is synchronized with the dynamic obstacles perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set; the distributed robot is driven by the synchronized collaborative control instruction set to perform resource interaction behavior of edge computing to obtain a collaborative motion path of the distributed robot.

[0049] The technical solution of this application has the following beneficial effects:

[0050] This application uses cross-robot joint compression technology to eliminate the redundancy of multi-source environmental data, significantly reduce communication bandwidth occupancy, and improve the transmission efficiency and real-time performance of perception data. A hierarchical communication network is adaptively constructed based on spatial density differences, and the priority of signal transmission paths is dynamically adjusted to ensure the centralized allocation of communication resources in high-density areas and enhance the reliability of data transmission in key areas. Priority is quantified as a weight parameter, and constraints are generated in combination with dynamic obstacle motion parameters to achieve deep coupling of communication resource allocation and motion safety control. Through a distributed timestamp alignment mechanism, the execution delay differences of multi-node instructions are eliminated, ensuring the spatiotemporal consistency of multi-robot actions. Edge computing resources are dynamically allocated based on instructions to optimize the collaborative efficiency of local path planning and global obstacle avoidance, ultimately generating a highly secure collaborative motion path.

[0051] Furthermore, the adjusted signal transmission path priority is divided into intervals and converted into weight parameters. The directional and velocity changes of dynamic obstacles are extracted to generate dynamic threat coefficients. The weight parameters and dynamic threat coefficients are processed through joint constraints to construct an instruction generation template corresponding to the layered communication link. The dynamic threat coefficients are then input into the template to generate a collaborative control instruction set containing motion constraints. By quantifying the correlation between communication priority and dynamic threat factors, a deep integration of communication resource allocation and obstacle avoidance strategies is achieved. The joint constraint processing of the weight parameter distribution results and the dynamic threat coefficient enables the instruction generation template to dynamically adapt to environmental changes, improving the communication efficiency and obstacle avoidance response capabilities of the multi-robot system in complex scenarios. At the same time, the motion constraints in the instruction generation template ensure that the robot path planning takes into account both communication priority and safety margins, significantly reducing the risk of conflict and optimizing overall task execution efficiency.

[0052] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0054] Figure 1 A flow chart of a distributed robot collaborative control method based on edge computing provided by the present application is shown;

[0055] Figure 2 The following is a schematic diagram showing the structure of a distributed robot collaborative control system based on edge computing provided by the present application;

[0056] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0058] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0059] In scenarios such as industrial automation, intelligent logistics, and disaster relief, multi-robot systems must collaborate to complete tasks in dynamic environments while avoiding moving obstacles. Existing distributed collaborative control algorithms based on multi-agent systems achieve path planning through local perception and distributed decision-making. However, they rely on local communication networks to transmit obstacle motion states and neighboring node information, making them susceptible to signal interference or delays, leading to path planning delays and even conflicts. Furthermore, this method places high demands on the prediction accuracy of the obstacle motion model. If the actual obstacle behavior deviates from the predicted model, for example, through sudden acceleration or directional changes, avoidance failure may occur. In high-density robot clusters, distributed consensus protocols slow down response times due to competition for computing resources, making it difficult to meet the real-time requirements of large-scale systems. These issues limit the practical application of existing solutions in highly dynamic and complex scenarios. There is an urgent need for a collaborative control method that balances real-time performance, robustness, and low communication dependency.

[0060] To address the above issues, this application proposes a distributed robot collaborative control method based on edge computing. This method first generates a distributed perception dataset by jointly compressing and processing environmental spatial data and dynamic obstacle motion parameters across multiple robots, reducing communication bandwidth usage and improving data transmission efficiency. Subsequently, hierarchical communication links are dynamically constructed based on spatial density differences, and signal transmission path priorities are adjusted in real time to optimize communication resource allocation to accommodate signal interference and latency issues in complex environments. Furthermore, communication path priorities are converted into weight parameters, combined with dynamic obstacle motion parameters to generate a collaborative control instruction set. Data latency differences are eliminated through cross-node synchronization to ensure consistency between the instruction set and the real-time state of the obstacle. Finally, through edge computing-driven resource interaction, the overlapping buffer areas and computational coverage in the distributed robot path planning are dynamically adjusted, achieving high-precision obstacle avoidance and efficient collaboration. Leveraging the localized processing capabilities of edge computing, dynamic communication link optimization, and cross-node synchronization, this solution significantly improves the system's response speed and robustness to dynamic obstacles. This addresses the core shortcomings of existing technologies, such as high communication dependency, sensitive prediction errors, and sluggish response in large-scale clusters. This provides reliable technical support for multi-robot collaborative control in highly dynamic scenarios.

[0061] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0062] Figure 1 A flowchart of a distributed robot collaborative control method based on edge computing is provided for an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0063] 101. Collect environmental spatial data of the distributed robots and motion parameters of dynamic obstacles, and perform cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set;

[0064] In this step, distributed robots refer to a system composed of multiple robots with independent perception, decision-making and execution capabilities, which achieve mission goals through collaborative communication and data sharing.

[0065] Environmental spatial data includes three-dimensional spatial information of the static and dynamic environment within the perception range of the distributed robot, such as obstacle locations, terrain features, etc.

[0066] The motion parameters of dynamic obstacles are used to describe the real-time motion status of dynamic obstacles, including position coordinates, speed, direction change, and rate change.

[0067] Cross-robot joint compression processing refers to compressing the original environmental spatial data into a structured distributed perception dataset through multi-robot collaborative data de-redundancy and feature extraction technology.

[0068] The distributed perception dataset integrates the de-redundant environmental data and dynamic obstacle features to form a unified data set containing static environmental feature points and dynamic obstacle trajectory prediction information.

[0069] In the embodiments of the present application, each distributed robot first collects environmental spatial data using sensors such as lidar and cameras, and uses a target detection algorithm to identify dynamic obstacles and their motion parameters. Subsequently, all robots upload the environmental spatial data and the motion parameters of dynamic obstacles to the distributed data exchange layer, where data consistency is verified using a consensus algorithm. Next, a joint compression algorithm is used to reduce the dimensionality of the environmental spatial data, remove duplicate static obstacle information, and perform parameterized modeling of the trajectories of dynamic obstacles. Ultimately, a distributed perception dataset is generated that contains global environmental features and predicted trajectories of dynamic obstacles.

[0070] During warehousing and logistics, multiple mobile robots collaborate to carry goods. The robots use lidar to collect location data of shelves and dynamic obstacles (such as pedestrians). After cross-robot joint compression processing, they generate a distributed perception dataset that includes shelf layout and pedestrian motion trajectory predictions, providing accurate environmental information for subsequent communication and collaborative control.

[0071] 102. Based on the spatial density differences of the distributed perception data set, dynamically construct hierarchical communication links in the distributed robot group, and adjust the priorities of signal transmission paths in the hierarchical communication links according to real-time changes in the spatial density differences to obtain adjusted signal transmission path priorities;

[0072] In this step, spatial density difference refers to the data density distribution of each region in the distributed perception dataset, reflecting the complexity of the environment and the density of robot distribution.

[0073] Hierarchical communication links are communication hierarchical structures divided according to spatial density differences. High-density areas use high-frequency communication links, and low-density areas use low-frequency communication links.

[0074] The signal transmission path refers to the sequence of communication links that data or control instructions pass through from the sending end to the receiving end in a distributed robot system.

[0075] The adjusted signal transmission path priority is a communication link priority parameter that is dynamically optimized according to the environmental spatial density difference and task requirements, and is used to guide the efficient transmission of data in a distributed robot system.

[0076] In the embodiment of the present application, first, based on the spatial density differences of the distributed sensing data set, the system uses a density clustering algorithm to identify high-density areas and low-density areas, and divides the hierarchical communication links accordingly. Subsequently, the bandwidth and transmission rate of the hierarchical communication links are dynamically adjusted, and the priority of the signal transmission paths is assigned to different levels based on the current task requirements and the threat coefficient of dynamic obstacles. For example, high-density areas use high-frequency communication links and increase the transmission priority, while low-density areas reduce the use of communication resources. Finally, the adjusted signal transmission path priority is generated, providing a basis for the conversion of communication weight parameters.

[0077] During warehouse logistics, when multiple robots enter a narrow aisle (high-density area), the system dynamically increases the priority of the communication link in that area to ensure rapid transmission of obstacle avoidance commands between robots. In open areas (low-density areas), the communication link priority is lowered to conserve resources.

[0078] 103. Convert the adjusted signal transmission path priority into a communication path weight parameter, and generate a collaborative control instruction set in combination with the motion parameter;

[0079] In this step, the communication path weight parameter maps the adjusted signal transmission path priority to a numerical weight, which is used to quantify the importance of the communication link.

[0080] The dynamic threat coefficient is a threat assessment index generated by integrating the direction and rate changes of dynamic obstacles, and is used to constrain collaborative control instructions.

[0081] The collaborative control instruction set contains a set of instructions for the motion constraints of distributed robots, guiding the distributed robots to collaboratively avoid obstacles and plan paths.

[0082] In the embodiments of the present application, a weighted graph model is first used to convert the signal transmission path priority into a communication path weight parameter, where higher priority paths correspond to higher values, and lower priority paths correspond to lower values. Subsequently, the communication path weight parameters are combined with the motion parameters of dynamic obstacles through a distributed optimization algorithm. The weighted graph model is used to calculate the obstacle avoidance potential field between each distributed robot and the dynamic obstacle. The robot's motion trajectory is optimized through model predictive control, ensuring that robots with higher communication path weights perform key tasks first, ultimately generating a collaborative control instruction set.

[0083] In warehouse logistics, when multiple robots are transporting goods through narrow aisles, the system generates instructions requiring the robots to slow down and maintain a safe distance based on the weight parameters of high-priority communication links and the high threat factor of dynamic obstacles. In open areas, low weight parameters and low threat factors allow the robots to move faster.

[0084] 104. Synchronize the collaborative control instruction set with the dynamic obstacle perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set;

[0085] In this step, cross-node synchronization refers to ensuring that the collaborative control instruction sets received by the distributed robots are consistent in time and content through timestamp alignment and consensus algorithms.

[0086] The synchronized collaborative control instruction set refers to the instruction set that has been synchronized across nodes and can drive the robots to perform collaborative actions.

[0087] In the embodiments of the present application, a unified time base is assigned to the collaborative control instruction set through timestamp alignment technology, and a distributed consensus algorithm is used to verify instruction consistency, eliminating deviations caused by communication delays. The system distributes the verified instruction set to all robot nodes, ensuring that the instructions received by each node are completely synchronized in content and time, and ultimately generates a synchronized collaborative control instruction set. For example, during the warehousing process, multiple robots can simultaneously receive obstacle avoidance instructions and adjust their paths, avoiding the risk of collision caused by asynchronous instructions.

[0088] In the warehousing and logistics process, when multiple robots receive obstacle avoidance instructions, cross-node synchronization is used to ensure that all robots adjust their paths at the same time to avoid collisions caused by instruction delays.

[0089] 105. Drive the distributed robot to perform resource interaction behavior of edge computing through the synchronized collaborative control instruction set to obtain a collaborative motion path of the distributed robot.

[0090] In this step, the resource interaction behavior of edge computing refers to the distributed robot allocating computing resources on the local edge node and dynamically adjusting the obstacle coverage and path planning parameters.

[0091] A collaborative motion path refers to a robot motion path generated based on resource interaction behavior, taking into account both mission objectives and obstacle avoidance requirements.

[0092] In the embodiment of the present application, first, each distributed robot breaks down the synchronized collaborative control instruction set into an action sequence containing target displacement and execution priority tags, and generates a preliminary path based on the predicted trajectory of dynamic obstacles. Subsequently, the preliminary path and the distributed resource scheduling algorithm are used to coordinate resource allocation among multiple robots to avoid contention for computing resources. For example, robots with high-priority tasks take priority in occupying the computing resources of the edge server. Finally, the paths generated by all robots are verified by the conflict detection algorithm, and the final collaborative motion path is output.

[0093] During the warehousing and logistics process, robots dynamically adjust the resource allocation range based on resource interaction behavior, generating overlapping buffer areas in narrow passages to ensure that multiple robots can efficiently complete cargo handling tasks while slowing down and avoiding obstacles.

[0094] In summary, steps 101 to 105 achieve dynamic and optimal allocation of communication resources by collecting environmental data and dynamic obstacle parameters, combined with cross-robot joint compression and layered communication link construction. The generation and synchronization of collaborative control instruction sets ensures efficient obstacle avoidance and collaborative path planning for multiple robots in complex environments. The interactive behavior of edge computing resources further enhances the real-time and security of path generation, ultimately significantly improving the operational efficiency and task completion rate of robot swarms in scenarios such as warehousing and logistics.

[0095] To optimize communication link priorities and dynamic motion threat handling in a distributed robotic system, this solution divides communication link intervals into hierarchical sections and converts priority levels into weight parameters. This extracts the direction and rate changes in the motion parameters to generate dynamic threat coefficients. This combined constraint weight parameters and dynamic threat coefficients construct an instruction generation template, ultimately generating a collaborative control instruction set that includes motion constraints, enabling efficient multi-robot collaboration. In some embodiments, step 103 converts the adjusted signal transmission path priority into a communication path weight parameter and generates a collaborative control instruction set based on the motion parameters, including:

[0096] 201. Divide the hierarchical communication links into intervals according to the adjusted level values ​​of the signal transmission path priorities, convert the level values ​​within the intervals, and obtain weight parameters corresponding to the hierarchical communication links;

[0097] In step 201, the adjusted level value of the signal transmission path priority refers to the value used to describe the priority of communication requirements at each level in the hierarchical communication link. The level value is composed of data such as the stability of the communication link, bandwidth occupancy, and task urgency. The hierarchical communication link refers to the physical layer, data link layer, network layer, transport layer, and application layer divided based on the modern communication network architecture, and each layer corresponds to different communication functions and data transmission characteristics. Interval division refers to dividing the communication link into several sub-intervals according to the distribution range of the level value, and each sub-interval corresponds to a different communication resource allocation strategy. The weight parameter is a parameter obtained after numerical conversion, which is used to quantify the importance of each hierarchical communication link in the overall system. For example, a high-priority communication link may obtain a higher weight value to ensure priority transmission of critical data.

[0098] In an embodiment of the present application, first, the stability index, bandwidth occupancy rate, and task urgency data of the communication link are collected in real time through sensors and communication modules. Subsequently, the level values ​​of the adjusted signal transmission path priority are sorted and normalized using a dynamic programming algorithm, and the original data is mapped to the interval between zero and one. Next, the normalized level values ​​are divided into intervals using a mean clustering algorithm, and each sub-interval corresponds to a specific communication resource allocation strategy. Finally, the level value of each sub-interval is converted into a corresponding weight parameter through a weighted average algorithm to ensure that the weight parameter can reflect the dynamic demand priority of each layered communication link.

[0099] 202. Extracting a direction change and a speed change from the motion parameters, and superimposing and merging the direction change and the speed change into a dynamic threat coefficient;

[0100] In step 202, the directional change refers to the magnitude of the change in the robot's direction per unit time and is used to describe the instability of the robot's path. The velocity change refers to the range of fluctuation in the robot's velocity per unit time and is used to characterize the dynamics of the robot's motion. The dynamic threat coefficient is a comprehensive indicator derived by weighted superposition of the directional change and velocity change, used to assess the potential collision risk or path conflict between robots. A higher coefficient indicates greater uncertainty in the robot's motion and requires a more rigorous obstacle avoidance strategy.

[0101] In an embodiment of the present application, motion parameters, including direction change and rate change, are first extracted from the robot's sensor data. The direction change is calculated using a directional difference algorithm to obtain the angular difference between adjacent time points. The rate change uses a velocity change rate algorithm to calculate the absolute value of the velocity change. Next, a weighted superposition merging technique is applied to linearly combine the direction change and rate change according to a preset proportional coefficient to form a dynamic threat coefficient. A larger dynamic threat coefficient indicates a higher motion risk.

[0102] 203. Performing joint constraint processing on the weight parameter and the dynamic threat coefficient to obtain a weight parameter distribution result, and constructing an instruction generation template corresponding to each layer in the layered communication link based on the weight parameter distribution result;

[0103] In step 203, the weight parameter distribution result refers to the parameter distribution diagram formed after the weight parameters and the dynamic threat coefficient are jointly constrained. Joint constraint processing refers to the simultaneous processing of multiple interrelated constraints in optimization or decision-making problems, and ensuring that these constraints can be satisfied in an uncertain or dynamically changing environment. The instruction generation template refers to a structured framework built based on template engine technology, which is used to map the dynamic threat coefficient and weight parameters into a specific collaborative control instruction set. The template contains modules such as communication resource allocation rules, obstacle avoidance strategies, and path optimization algorithms, and can generate adaptive control instructions based on input parameters.

[0104] In the embodiment of the present application, first, the weight parameters and the dynamic threat coefficient are jointly constrained by applying a constrained optimization algorithm. The dynamic threat coefficient is used as a constraint boundary condition, and the weight parameters are adjusted through linear programming technology to ensure that the weight parameters maintain the optimal distribution under the dynamic threat coefficient limit. The processing results form a weight parameter distribution result, which reflects the adjusted importance of communication links at each level. Then, based on the weight parameter distribution result, a template construction technology is used to design an instruction generation template for each communication link level. Finally, the generated instruction generation template is passed to the next step for actual control instruction generation.

[0105] 204. Input the dynamic threat coefficient into the instruction generation template to generate a collaborative control instruction set including distributed robot motion constraint conditions.

[0106] In step 204, the collaborative control instruction set refers to a specific set of control commands generated based on the instruction generation template and the dynamic threat factor. This instruction set contains the motion constraints of the distributed robots, such as speed limits, path offsets, and obstacle avoidance strategies, and is used to guide the collaborative operation of the multi-robot system. The dynamic threat factor is used as an input parameter to adjust the constraints in the instruction generation template, ensuring that the instruction set can respond to environmental changes in real time.

[0107] In an embodiment of the present application, the dynamic threat coefficient is first injected into the instruction generation template as an input parameter. Next, by applying the instruction filling algorithm, the dynamic threat coefficient variable is replaced in the instruction generation template, and specific motion constraints are generated according to the preset logic of the instruction generation template. Then, through distributed collaborative processing technology, the output of the instruction generation template at all levels is integrated to form a complete collaborative control instruction set. Finally, the collaborative control instruction set is sent to the robot execution unit to achieve real-time control. The entire process ensures that the dynamic threat coefficient is effectively integrated into the control instructions, improving the system's responsiveness.

[0108] Here's a specific example:

[0109] In warehouse logistics, when multiple mobile robots collaborate to move goods, for example, three robots are moving boxes in a warehouse. The communication link is divided into three layers: the control layer, the coordination layer, and the execution layer. First, the system uses a constant-width interval partitioning algorithm to divide the communication priority values ​​into three intervals based on preset values ​​(e.g., high for the control layer, medium for the coordination layer, and low for the execution layer). Within each interval, linear normalization is performed to obtain weight parameters for the control layer, coordination layer, and execution layer. Second, data is extracted from the robot motion sensors, including the direction change and velocity change m. Using a weighted superposition merging technique, these are combined according to a proportional coefficient to obtain a dynamic threat coefficient. Next, the weight parameters are jointly constrained with the dynamic threat coefficient. Linear programming is used to optimize the weight distribution, ensuring that the control layer weights are adjusted within the threat coefficient constraints. Then, an instruction generation template is constructed, for example, the control layer template includes "collision avoidance" logic. Finally, the dynamic threat coefficient is input into the template to generate a collaborative control instruction set, such as "robot slows down and turns." This instruction set is used by all robots to coordinate and move goods, avoiding collisions and improving efficiency.

[0110] In summary, steps 201 to 204 achieve high efficiency and safety in robot collaborative control by prioritizing communication links and dynamically evaluating motion risks, improve the system's adaptability in complex environments, and reduce conflicts and delays in task execution.

[0111] To address the dynamic adaptation problem of the dynamic threat coefficient and communication link weight parameters in a multi-autonomous mobile robot system in a warehousing and logistics environment, the solution divides the dynamic threat coefficient into subintervals and designs a conversion function. This initial conversion factor is obtained by combining the dynamic threat coefficient and the conversion factor within the subinterval to generate parameters. After interval fusion and hierarchical balance constraints, an optimized weight parameter distribution result is obtained, thereby improving the dynamic adaptability of the communication weight. In some embodiments, the weight parameter and the dynamic threat coefficient are jointly constrained in step 203 to obtain the weight parameter distribution result, including:

[0112] 301. Divide the variation range of the dynamic threat coefficient into continuous sub-intervals, each sub-interval corresponding to a conversion function;

[0113] In step 301, the dynamic threat coefficient is a quantitative indicator representing the risk of robot motion. The range of variation refers to the upper and lower bounds of the coefficient's possible values. Subranges are continuous, smaller segments of the range. The conversion function is the rule used to mathematically transform values ​​within the subrange.

[0114] In this embodiment, the minimum and maximum values ​​of the dynamic threat coefficient are first determined through historical data analysis or real-time monitoring. Subsequently, an equal-width binning algorithm is used to divide the range of values ​​into several contiguous, non-overlapping subintervals, each corresponding to a different threat level, such as low, medium, or high. Next, a corresponding conversion function is designed for each subinterval. Finally, the dynamic threat coefficient value is input into the corresponding subinterval and conversion function to complete the initial mapping.

[0115] 302. According to the hierarchical position corresponding to the weight parameter in the hierarchical communication link, obtain an initial conversion factor associated with the hierarchical position;

[0116] In step 302, the hierarchical position of a layered communication link refers to the specific layer of the communication link within a layered architecture, such as the physical layer, data link layer, network layer, transport layer, and application layer. The initial conversion factor is a preset parameter associated with the layer position, used to quantify the basic weight of that layer in communication resource allocation. For example, the initial conversion factor for the physical layer may be lower, while the initial conversion factor for the application layer may be higher to reflect the impact of different layers on system performance.

[0117] In an embodiment of the present application, first, the layered architecture of the communication link is identified, such as the physical layer, data link layer, network layer, transport layer, and application layer, and the specific hierarchical position of the weight parameters of each layer is determined. Subsequently, an initial conversion factor is assigned to each layer through expert experience or historical data analysis. The setting of the initial conversion factor needs to comprehensively consider the contribution of each layer to communication stability and task urgency. For example, the application layer has a higher initial conversion factor because it is directly related to task execution, while the physical layer has a smaller initial conversion factor because of its lower basic transmission requirements.

[0118] 303. Within the subinterval, perform correlation calculation on the dynamic threat coefficient and the initial conversion factor according to the conversion function to obtain correlated subinterval parameters;

[0119] In step 303, the correlation calculation combines the dynamic threat coefficient with the initial conversion factor to generate parameters corresponding to the subintervals through a conversion function. The associated subinterval parameters are the intermediate results after mapping and are used for subsequent interval boundary fusion and inter-level balance constraints.

[0120] In the embodiments of the present application, the numerical value of the dynamic threat coefficient is first substituted into the subinterval and matched with the corresponding conversion function. For example, if the dynamic threat coefficient belongs to the medium threat interval, a piecewise function is used. Subsequently, the output value of the conversion function is combined with the initial conversion factor, and the associated subinterval parameters are calculated. For example, the output value of the conversion function is multiplied by the initial conversion factor to generate an intermediate parameter reflecting the association between the threat level and the communication layer. This process, through function matching and calculation, quantitatively associates the dynamic threat coefficient with the communication link layer weight, forming an operational intermediate parameter.

[0121] 304. Perform interval boundary fusion on the associated sub-interval parameters to obtain fused parameters, and perform inter-level balance constraints on the fused parameters to obtain a weight parameter distribution result.

[0122] In step 304, interval boundary fusion eliminates parameter abrupt changes between subintervals due to differences in transfer functions, ensuring smooth parameter transitions within adjacent intervals. Inter-level balancing constraints coordinate the weight parameters of different levels through a multi-objective optimization algorithm to ensure that communication resource allocation meets overall system requirements.

[0123] In the embodiment of the present application, all associated subinterval parameters output in step 303 are first received. Next, a boundary smoothing fusion algorithm is used to interpolate at the subinterval boundaries and merge the parameter values ​​into fused parameters. Then, constrained optimization solving techniques are applied to adjust the parameter values ​​using a linear programming algorithm to ensure that the distribution of weight parameters at each level meets the balance condition. Finally, the weight parameter distribution result is output.

[0124] Here's a specific example:

[0125] In warehousing logistics, when multiple mobile robots collaborate to move goods, for example, three robots are collaboratively moving boxes in a warehouse. The hierarchical communication link includes the control layer, the coordination layer, and the execution layer. First, the dynamic threat coefficient range is divided from minimum to maximum into three consecutive subintervals using an equal-width interval partitioning algorithm. A transformation function is assigned to each subinterval, such as a linear transformation function for the first subinterval. Second, an initial transformation factor is obtained based on the hierarchical position of the weight parameter. For example, the control layer corresponds to a larger initial transformation factor, the coordination layer to a medium one, and the execution layer to a smaller one. Next, the subinterval in which the current dynamic threat coefficient value belongs is determined based on the subinterval in which it resides. The transformation function for that subinterval is applied, and the dynamic threat coefficient is weighted multiplied by the initial transformation factor to obtain the associated subinterval parameters. All subinterval parameters are then fused at the boundaries, for example, by using an interpolation algorithm to smoothly merge them at the boundaries. Inter-hierarchical optimization with balance constraints is then performed to ensure a balanced distribution of weight parameters across the control, coordination, and execution layers. The resulting weight parameter distribution is then output for subsequent collaborative control.

[0126] In summary, steps 301 to 304 achieve dynamic adaptation of communication resource allocation to robot motion threats by dividing the dynamic threat coefficient into subintervals and designing a conversion function, combined with the initial conversion factors of the hierarchical communication links and inter-layer balance constraints. This method can improve the priority responsiveness of communication links in warehousing and logistics environments, reduce the collision risk caused by communication delays, and ensure the rational distribution of weights at each layer through a multi-objective optimization algorithm, thereby improving the collaborative efficiency and safety of multi-robot systems.

[0127] In order to solve the problem of dynamic adaptation of the dynamic threat coefficient and the communication link weight parameter in the multi-autonomous mobile robot system in the warehousing and logistics environment, the solution optimizes the sub-interval parameters by cross-correlating the dynamic threat coefficient and the weight parameter. The slope of the conversion function is adjusted according to the sub-interval span, the dynamic threat coefficient is scaled to generate the dynamic threat value, the adjustment factor is obtained in combination with the hierarchical position for cross-correlation, and the associated sub-interval parameters are generated through interval accumulation and superposition processing to enhance the responsiveness of the parameter distribution to dynamic threats. In some embodiments, in the sub-interval described in step 303, the dynamic threat coefficient and the initial conversion factor are correlated and calculated according to the conversion function to obtain the associated sub-interval parameters, including:

[0128] 401. Determine a slope adjustment factor of the conversion function according to the span of the subinterval, and scale the dynamic threat coefficient according to the slope adjustment factor to generate a dynamic threat value.

[0129] In step 401, the subinterval span refers to the difference between the upper and lower limits of the dynamic threat coefficient for that subinterval. The slope adjustment factor is a scaling factor calculated based on the subinterval span and used to adjust the sensitivity of the transfer function. Scaling refers to numerically adjusting the dynamic threat coefficient according to the scaling factor, and the dynamic threat magnitude refers to the normalized threat index obtained after scaling.

[0130] In this embodiment, historical data analysis is first used to determine the span of the subintervals of the dynamic threat coefficient. A linear interpolation algorithm is then used to calculate the slope adjustment factor of the conversion function based on the span of the subintervals (e.g., a larger span corresponds to a smaller slope adjustment factor). The dynamic threat coefficient is then input into a scaling process, where it is multiplied by the slope adjustment factor to generate a dynamic threat value.

[0131] 402. Within the subinterval, obtain a level adjustment factor according to the level position corresponding to the weight parameter, and cross-correlate the dynamic threat value with the level adjustment factor to obtain an initial conversion factor;

[0132] In step 402, the hierarchical position refers to the identifier of the communication link level to which the weight parameter belongs within the hierarchical structure. The hierarchical adjustment factor is a preset weight coefficient associated with the hierarchical position, used to reflect the importance of different levels. Cross-correlation refers to the mathematical correlation between the dynamic threat value and the hierarchical adjustment factor. The initial conversion factor is the intermediate adjustment parameter generated after the correlation operation.

[0133] In this embodiment, the hierarchical position of the weight parameter is first determined based on the layered architecture of the communication link, for example, the physical layer and the application layer. Subsequently, a hierarchical adjustment factor is assigned to each layer based on expert experience or historical data analysis. Next, the dynamic threat value is multiplied by the hierarchical adjustment factor using a multiplicative correlation algorithm to obtain an initial conversion factor value.

[0134] 403. Perform interval accumulation processing on the initial conversion factor, superimpose the accumulation processing result with the dynamic threat value, and generate associated sub-interval parameters.

[0135] In step 403, interval accumulation refers to the process of summing all initial conversion factors within the current subinterval. The accumulation result is the accumulated sum. Addition refers to the addition of the accumulation result to the dynamic threat value. The associated subinterval parameter is the parameter generated after addition and represents the final adjustment effect for the subinterval.

[0136] In this embodiment, the initial conversion factors are first applied using an accumulator technique, summing all initial conversion factors belonging to the same subinterval to produce a cumulative result. The dynamic threat value is then added to the cumulative result to generate a correlated subinterval parameter value. This parameter is ultimately output as a comprehensive adjustment for that subinterval at a given threat level and hierarchical importance.

[0137] Here's a specific example:

[0138] In warehouse logistics, when multiple mobile robots collaborate to move goods, suppose an execution-level robot detects an obstacle during transport, and its dynamic threat coefficient falls within a predefined subrange. First, the span of this subrange is calculated. A slope adjustment factor is determined based on an inverse proportional relationship, and the dynamic threat coefficient is multiplied by this factor to obtain the dynamic threat magnitude. Secondly, based on the execution-level position of the robot's weight parameter, a mapping table is retrieved to obtain a smaller level adjustment factor. The dynamic threat magnitude is multiplied by this factor to obtain an initial conversion factor. Next, the initial conversion factors of all robots within the subrange are cumulatively summed, and the sum is added to the dynamic threat magnitude to generate the associated subrange parameter. This parameter is used to adjust the motion priority of the execution-level robot, ensuring that it decelerates appropriately when avoiding obstacles while ensuring that global scheduling commands from the control-level robot are efficiently transmitted, maintaining the coordination of the overall handling task.

[0139] In summary, steps 401 to 403 achieve dynamic adaptation of communication resource allocation to robot motion threats through scaling the slope adjustment factor of the dynamic threat coefficient, cross-correlating the hierarchical adjustment factors, and performing interval accumulation and superposition processing. This method can improve the priority responsiveness of communication links in warehousing and logistics environments, reduce the collision risk caused by communication delays, and eliminate local fluctuations through accumulation processing, ensuring the stability of parameter distribution, thereby improving the collaborative efficiency and safety of multi-robot systems.

[0140] In order to solve the problem of low environmental modeling accuracy caused by redundant perception data and confusion between dynamic and static features in multi-distributed robot systems in warehousing and logistics environments, the solution improves the efficiency of environmental data perception through cross-robot joint compression processing, divides the overlapping areas of the perception range and segments the environmental data into discrete blocks, removes redundant data points to generate de-redundant blocks, separates dynamic obstacles from static environmental features, combines trajectory prediction to expand dynamic features, and splices them with static features to generate a distributed perception data set, providing accurate data support for collaborative control. In some embodiments, the environmental spatial data and motion parameters of dynamic obstacles of distributed robots are collected in step 101, and the environmental spatial data is subjected to cross-robot joint compression processing to obtain a distributed perception data set, including:

[0141] 501. Divide the spatial perception range of the distributed robot into overlapping areas, and segment the environmental spatial data into discrete block data sets containing redundant identifiers;

[0142] In step 501, the spatial perception range refers to the three-dimensional spatial area detectable by a single robot's sensor. Overlapping region partitioning refers to the process of marking the boundaries of the overlapping portions of the perception ranges of multiple robots. Environmental spatial data refers to the raw environmental information collected by all robots. Redundant identifiers are labels that indicate data repetitiveness. Discrete block datasets refer to the identified data subsets formed by segmenting the environmental space according to overlapping regions.

[0143] In an embodiment of the present application, first, environmental space data is collected through distributed robot sensors to obtain the coverage of the surrounding environment of each robot. Subsequently, a rasterization algorithm is used to divide the environmental space into regular discrete blocks, and each block records the spatial coordinates and perception data. Next, a spatial intersection detection algorithm is used to identify the overlapping areas of the perception ranges of adjacent robots, and redundant identifiers are added to the discrete blocks in the overlapping areas to mark the block as containing duplicate perception data from multiple robots. Finally, all discrete blocks are classified according to the redundant identifiers to generate a discrete block data set containing redundant identifiers. This process provides a structured data foundation for subsequent data deduplication through spatial division and redundant marking.

[0144] 502. Performing repeated feature matching processing on the overlapping area data of adjacent distributed robots within the discrete block data set, deleting redundant data points exceeding a set repetition threshold, and generating a de-redundant block data set;

[0145] In step 502, overlapping area data refers to block data simultaneously sensed by multiple robots. Repeat feature matching refers to the process of identifying and comparing identical block data collected by different robots. The repetition threshold refers to the upper limit of the number of repetitions required to determine data redundancy. The de-redundant block data set is a streamlined data set after redundant data has been removed.

[0146] In this embodiment, first, data from overlapping regions with redundant identifiers is extracted from discrete block datasets. Subsequently, a feature hashing algorithm is used to perform a similarity comparison between the perception data of adjacent distributed robots in the overlapping regions to identify duplicate feature points, such as the edges of obstacles at the same location. Next, the number of duplicate feature points within each discrete block dataset is counted. If the number exceeds a preset duplication threshold, the redundant data points are deleted, retaining only the single valid data. Finally, the processed discrete block datasets are integrated to generate a redundancy-free, deduplicated block dataset.

[0147] 503. Extracting high-density feature points and static environment feature points in the motion parameters of the dynamic obstacles from the de-redundant block data set, and separating the high-density feature points from the static environment feature points to generate a dynamic feature data set and a static feature data set;

[0148] In step 503, dynamic obstacles refer to moving people or objects. Motion parameters include speed and direction. High-density feature points refer to the densely distributed key points along the trajectory of a dynamic obstacle. Static environmental feature points refer to the contour feature points of a stationary object. The dynamic feature dataset and the static feature dataset are two separate feature sets.

[0149] In the embodiment of the present application, the motion parameters of dynamic obstacles, such as speed and directional change rate, are first extracted from the de-redundant block dataset. Subsequently, a density clustering algorithm is used to identify high-density feature points in the dynamic obstacle trajectory, such as densely distributed moving path points. Static environmental feature points, such as fixed shelves and walls, are also distinguished through time series analysis. The high-density feature points and static environmental feature points are then stored in the dynamic feature dataset and the static feature dataset, respectively. This process provides data support for trajectory prediction and environmental modeling.

[0150] 504. Perform trajectory extension prediction on the dynamic feature dataset according to the movement direction of the dynamic obstacle to obtain an extended dynamic feature sequence, and spatially concatenate the extended dynamic feature sequence with the static feature dataset to generate a distributed perception dataset.

[0151] In step 504, trajectory extension prediction involves using an algorithm to predict the potential future trajectory of a dynamic obstacle based on its historical motion parameters. The extended dynamic feature sequence is a dynamic feature data set containing the predicted trajectory points. Spatial splicing involves spatially fusing the extended dynamic feature sequence with the static feature data set to generate a distributed perception dataset.

[0152] In this embodiment, a Kalman filter algorithm is first used to predict the trajectory extension of high-density feature points in a dynamic feature dataset, generating future trajectory points of dynamic obstacles and forming an extended dynamic feature sequence. Subsequently, the extended dynamic feature sequence is joined with the static feature dataset using a spatial coordinate alignment algorithm, integrating the dynamic obstacle trajectories with the static environment information to generate a distributed perception dataset. This process, through trajectory prediction and spatial fusion, achieves dynamic updating and global consistency of the environmental model, providing complete perception data for multi-robot collaborative decision-making.

[0153] Here's a specific example:

[0154] During warehouse logistics, three mobile robots collaborate to move large shelves, using their lidar sensors to scan the environment. First, the system detects that the perception areas of Robot One and Robot Two overlap at the shelf aisle. This area is then divided into discrete blocks and marked with redundant identifiers. Next, the system compares the shelf point cloud data collected by the two robots for this overlapping block, removing duplicate shelf support point clouds to generate a de-redundant dataset. Next, the system identifies high-density feature points of the mobile forklift and static shelf feature points from the dataset, separating them into two separate datasets. Finally, the forklift's future trajectory is predicted based on its direction of motion, generating an extended feature sequence that is then combined with the shelf's static features to form a fused environment model. Each robot adjusts its path based on this model to avoid collisions with the forklift and improve collaborative efficiency.

[0155] In summary, steps 501 to 504 achieve efficient modeling and dynamic updating of environmental data by dividing redundant identifiers within the distributed robot perception range, deduplicating duplicate data, separating dynamic and static features, and combining trajectory predictions. This method can reduce the storage and computational burden of multi-robot systems caused by data redundancy, improve the accuracy of dynamic obstacle trajectory prediction, and ensure consistency in global environmental perception by integrating static features with dynamic trajectories, thereby improving the safety and efficiency of multi-robot collaborative operations in warehousing and logistics.

[0156] In order to solve the path conflict problem caused by inaccurate prediction of dynamic obstacle trajectories in multi-autonomous mobile robot systems in warehousing and logistics environments, the solution revolves around the prediction and verification of dynamic obstacle trajectories. The motion vector is calculated by extracting the obstacle position coordinates of continuous time periods, and a sequence of predicted trajectory points is generated by extending along the vector direction. After spatial credibility verification, it is aligned and merged with the dynamic feature data set to generate an extended dynamic feature sequence, providing future state information of dynamic obstacles for path planning. In some embodiments, the step 504 of performing trajectory extension prediction on the dynamic feature data set according to the motion direction of the dynamic obstacle to obtain an extended dynamic feature sequence includes:

[0157] 601. Extracting obstacle position coordinates for consecutive time periods from the dynamic feature data set, and calculating a motion vector of the dynamic obstacle based on a change in the obstacle position coordinates;

[0158] In step 601, the dynamic feature dataset refers to a high-density feature point set of dynamic obstacles, including the obstacle position coordinates for consecutive time periods. A consecutive time period refers to a sequence of obstacle positions recorded at fixed time intervals. The change in obstacle position coordinates is a vector representation of the coordinate difference between adjacent time periods, reflecting the direction and distance of the obstacle's displacement. A motion vector is a composite vector of the obstacle's velocity and direction over consecutive time periods, used to characterize its motion state.

[0159] In this embodiment, the obstacle's position coordinates are first obtained from a dynamic feature dataset over multiple consecutive time periods. Subsequently, the change in the obstacle's position coordinates between adjacent time periods is calculated to generate a displacement vector. A Kalman filter algorithm is then used to smooth the displacement vector and eliminate noise interference, ultimately yielding the dynamic obstacle's motion vector. This process enables accurate modeling of the obstacle's motion state.

[0160] 602. Using the obstacle position coordinates in the current time period as a reference point, extending the velocity distance of the dynamic obstacle by a preset multiple along the direction of the motion vector to generate a predicted trajectory point sequence;

[0161] In step 602, the reference point refers to the obstacle's position coordinates during the current time period, serving as the starting point for the predicted trajectory. The preset multiplier is a parameter dynamically adjusted based on the obstacle's historical motion trends and the environment, controlling the length of the trajectory. The velocity distance characterizes the relationship between the speed of an object or signal's movement and its coverage under specific conditions. The predicted trajectory point sequence refers to multiple future position points generated along the direction of the motion vector, reflecting the obstacle's likely motion path.

[0162] In this embodiment, the obstacle's position coordinates for the current time period are first used as the starting point for the predicted trajectory. Subsequently, the directional angle and velocity values ​​along the motion vector are extracted, and the obstacle's velocity distance is extended by a preset multiple to generate predicted trajectory points for the future. Next, a linear interpolation algorithm is used to generate a continuous sequence of predicted trajectory points. This process achieves dynamic prediction of the obstacle's trajectory through directional extension and velocity adjustment.

[0163] 603. Perform spatial credibility verification on the predicted trajectory point sequence, and perform timestamp alignment and merging on the verified predicted trajectory point sequence and the dynamic feature dataset to obtain an extended dynamic feature sequence.

[0164] In step 603, spatial credibility verification involves comparing the predicted trajectory points with surrounding environmental data, such as the collision probability of static and other dynamic obstacles, to determine the plausibility of the trajectory points. Timestamp alignment involves synchronizing the predicted trajectory point sequence with the dynamic feature dataset based on time periods to ensure data consistency. The extended dynamic feature sequence is the result of merging the verified predicted trajectory points with the historical dynamic feature points.

[0165] In the embodiment of the present application, first, an environmental collision detection algorithm is used to compare the predicted trajectory point sequence with the static map, and invalid trajectory points that overlap with obstacles are eliminated. At the same time, kinematic feasibility analysis technology is applied to filter trajectory points that exceed the robot's dynamic constraints. If the verification passes, the trajectory point is retained. Subsequently, the verified predicted trajectory point sequence is synchronized with the dynamic feature dataset according to the time period. Finally, the two are merged into an extended dynamic feature sequence. This process ensures the reliability and data consistency of the predicted trajectory.

[0166] Here's a specific example:

[0167] During a warehouse logistics operation, two mobile robots were collaboratively moving cargo boxes when a LiDAR detected a moving forklift. First, the forklift's position coordinates for the most recent frames were extracted. The difference between the coordinates of adjacent frames was calculated to obtain a motion vector, with a direction of a certain angle north-east and a certain speed. Next, using the latest coordinate point as a reference, several predicted trajectory points were generated along that direction, spaced a certain distance apart, at a preset multiple of time units. Next, the predicted points were verified to see if they overlapped with the shelf area. At a certain point in time, the predicted point was found to be in a restricted area within the shelf aisle, so that point and subsequent points were deleted. The remaining predicted points were timestamped and merged with the forklift's actual trajectory points to form an extended sequence. The robot then adjusted its path accordingly, avoiding the forklift's predicted path in advance.

[0168] In summary, steps 601 to 603 achieve accurate obstacle trajectory prediction and integrated update of dynamic feature data through dynamic obstacle position coordinate extraction, motion vector calculation, trajectory extension prediction, and spatial credibility verification. This method can enhance the multi-robot system's dynamic obstacle perception and avoidance capabilities in warehousing and logistics environments, reduce the risk of path conflicts caused by trajectory prediction errors, and ensure data consistency through timestamp alignment, thereby improving the safety and efficiency of collaborative handling.

[0169] To address the path conflicts and resource allocation issues associated with collaborative cargo handling by multiple mobile robots during warehousing and logistics, this solution optimizes collaborative motion paths through resource interaction, breaking down collaborative control instruction sets into action sequences and allocating edge computing resource ratios. Dynamically adjusting the computational coverage boundaries based on displacement, generating overlapping buffer zones in the movement direction, aligning dynamic obstacle parameters with the robot displacement states and inputting them into the buffer zones, and generating collaborative motion paths that balance resource allocation and obstacle avoidance. In some embodiments, the resource interaction behavior of driving the distributed robots to perform edge computing through the synchronized collaborative control instruction set in step 105 to obtain the distributed robots' collaborative motion paths includes:

[0170] 701. Decompose the synchronized collaborative control instruction set into an action sequence corresponding to the distributed robot, wherein the action sequence includes a target displacement and an execution priority tag;

[0171] In step 701, the synchronized collaborative control instruction set is a set of target actions for multiple robots. An action sequence is a process of breaking down instructions into specific execution steps. The target displacement represents the distance and direction the robots need to move. The execution priority tag identifies the urgency of the action.

[0172] In this embodiment, a collaborative control instruction set is first received from a central control system, containing the target positions and coordination rules for all participating robots. Subsequently, a task decomposition algorithm is used to decompose the collaborative control instruction set into action sequences for each robot. Each action sequence includes a target displacement generated by the path planning module based on task requirements, and an execution priority tag dynamically assigned based on task urgency and resource requirements.

[0173] 702. Allocate a resource occupancy ratio of edge computing to the distributed robot based on the execution priority tag, and determine a computing coverage range of dynamic obstacles according to the resource occupancy ratio;

[0174] In step 702, edge computing refers to distributed computing node resources. Resource utilization refers to the percentage of computing resources allocated to each robot. Computational coverage refers to the maximum spatial area within which a single robot can process obstacle data in real time.

[0175] In an embodiment of the present application, first, the execution priority label is used to allocate the edge computing resource occupancy ratio for each distributed robot using a weighted resource allocation algorithm. High-priority robots obtain a higher resource share, and low-priority robots obtain a lower share. Subsequently, the computational coverage of dynamic obstacles is determined by a load balancing model in combination with real-time motion data of dynamic obstacles, such as the position and speed of a forklift. For example, when a high-priority robot needs to avoid a high-speed moving forklift, its corresponding computational coverage will be expanded to include more predicted trajectory points to ensure that the edge computing node can process obstacle avoidance decisions in a timely manner.

[0176] 703. Dynamically adjust the boundary conditions of the calculation coverage range according to the target displacement and the resource occupancy ratio, and calculate the overlapping buffer area of ​​the distributed robot in the moving direction based on the adjusted calculation coverage range;

[0177] In step 703, the boundary conditions refer to the spatial constraint parameters of the calculated coverage area. The overlapping buffer area is a common obstacle avoidance area formed by the intersection of the calculated coverage areas of multiple robots.

[0178] In this embodiment, a directional weighting algorithm is first applied based on the target displacement and resource occupancy ratio to expand the coverage boundary conditions in the target displacement direction and shrink them in the opposite direction. Next, based on the adjusted calculated coverage, a spatial overlap detection technique is implemented to calculate the intersection of the coverage areas in the distributed robot's movement direction. Finally, a safety margin extension algorithm is used to expand the intersection area by a preset buffer distance to generate an overlapping buffer area.

[0179] 704. Align the motion parameters of the dynamic obstacle and the displacement state of the distributed robot and input them into the overlapping buffer area to generate a coordinated motion path of the distributed robot.

[0180] In step 704, the motion parameters of the dynamic obstacle include speed and direction. The displacement state of the distributed robot refers to the real-time position and velocity of multiple robots with independent perception, decision-making, and execution capabilities. Alignment is the operation of synchronizing the time between the two types of data. The collaborative motion path is the resulting collision-avoidance trajectory.

[0181] In this embodiment, the motion parameters of dynamic obstacles are first time-stamped and aligned with the real-time displacement states of the distributed robots. This aligned data is then fed into an overlapping buffer. A multi-agent potential field obstacle avoidance algorithm is then employed to calculate the obstacle repulsion trajectory for each robot within the overlapping buffer. Distributed collaborative optimization techniques are then used to balance path conflicts and ensure global motion consistency. Finally, a collision-free collaborative motion path is output, and each robot executes its displacement action according to this path.

[0182] Here's a specific example:

[0183] In a warehouse logistics process, while three mobile robots were collaboratively moving goods, a lidar detected a high-speed forklift entering the work area. The central control system first decomposed the synchronized collaborative control instruction set into motion sequences for the three robots. Robot 1's urgent need to avoid the forklift was marked as high priority, while robots 2 and 3 were assigned medium priority. A weighted resource allocation algorithm then allocated resource usage percentages for edge computing nodes to the distributed robots based on the execution priority labels. In combination with real-time motion data from dynamic obstacles, the computational coverage of robot 1 was dynamically expanded to include the predicted trajectory of the forklift for a certain period of time. Based on the target displacement and resource usage percentage, a spatial topology analysis algorithm was used to adjust the boundary conditions of the computational coverage. An overlapping buffer zone was generated at the intersection of the paths of robots 1 and 2, preserving a safe distance. Finally, the motion parameters of the dynamic obstacle were timestamped with the displacement states of the distributed robots and then entered into the overlapping buffer zone. This generated a collaborative motion path. Robot 1 moved along the outer edge of the buffer zone to avoid the forklift, while robots 2 and 3 passed through the buffer zone at staggered intervals, ensuring efficient multi-robot collaboration and preventing collisions.

[0184] In summary, steps 701 to 704 achieve efficient multi-robot collaboration in a complex dynamic environment by decomposing coordination instructions, dynamically allocating resources, adjusting coverage boundaries, and generating buffer zones. In warehousing and logistics scenarios, this method significantly reduces the risk of collisions between robots and dynamic obstacles, improves the real-time performance and resource utilization of path planning, and ensures the prioritization of high-priority tasks, thereby enhancing overall handling efficiency and system stability.

[0185] Figure 2 The present invention provides a schematic diagram of a distributed robot collaborative control system based on edge computing, as shown in FIG. Figure 2 As shown, the system includes:

[0186] The acquisition module 21 collects the environmental spatial data of the distributed robots and the motion parameters of the dynamic obstacles, and performs cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set;

[0187] an adjustment module 22 for dynamically constructing hierarchical communication links in the distributed robot group based on the spatial density differences of the distributed perception data set, and adjusting the priorities of signal transmission paths in the hierarchical communication links according to real-time changes in the spatial density differences to obtain adjusted signal transmission path priorities;

[0188] A generating module 23 converts the adjusted signal transmission path priority into a communication path weight parameter, and generates a collaborative control instruction set in combination with the motion parameter;

[0189] A synchronization module 24 synchronizes the collaborative control instruction set with the dynamic obstacle perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set;

[0190] The driving module 25 drives the distributed robot to perform resource interaction behavior of edge computing through the synchronized collaborative control instruction set to obtain the collaborative motion path of the distributed robot.

[0191] Figure 2 The distributed robot collaborative control system based on edge computing can execute Figure 1 The implementation principle and technical effects of the distributed robot collaborative control method based on edge computing described in the illustrated embodiment will not be repeated here. The specific manner in which each module and unit performs operations in the distributed robot collaborative control system based on edge computing in the above embodiment has been described in detail in the embodiment of the method and will not be elaborated here.

[0192] In one possible design, Figure 2 A distributed robot collaborative control system based on edge computing of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0193] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0194] The processing component 32 is used for the above Figure 1 The embodiment provides a distributed robot collaborative control method based on edge computing.

[0195] The processing component 32 may include one or more processors to execute computer instructions to perform all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0196] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0197] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0198] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0199] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0200] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0201] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 A distributed robot collaborative control method based on edge computing in the illustrated embodiment.

[0202] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0203] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0204] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0205] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distributed robot collaborative control method based on edge computing, characterized in that: include: Collecting environmental spatial data of distributed robots and motion parameters of dynamic obstacles, and performing cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set; Based on the spatial density differences of the distributed perception data sets, dynamically constructing hierarchical communication links in the distributed robot group, and adjusting the priorities of signal transmission paths in the hierarchical communication links according to real-time changes in the spatial density differences to obtain adjusted signal transmission path priorities; Converting the adjusted signal transmission path priority into a communication path weight parameter, and generating a collaborative control instruction set in combination with the motion parameter; Synchronizing the collaborative control instruction set with the dynamic obstacle perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set; The synchronized collaborative control instruction set is used to drive the distributed robot to perform resource interaction behavior of edge computing to obtain a collaborative motion path of the distributed robot.

2. The method according to claim 1, characterized in that The converting the adjusted signal transmission path priority into a communication path weight parameter and generating a collaborative control instruction set in combination with the motion parameter includes: Dividing the hierarchical communication links into intervals according to the adjusted level values ​​of the signal transmission path priorities, and converting the level values ​​within the intervals to obtain weight parameters corresponding to the hierarchical communication links; Extracting a direction change amount and a speed change amount from the motion parameters, and superimposing and merging the direction change amount and the speed change amount into a dynamic threat coefficient; Performing joint constraint processing on the weight parameter and the dynamic threat coefficient to obtain a weight parameter distribution result, and constructing an instruction generation template corresponding to each layer in the layered communication link based on the weight parameter distribution result; The dynamic threat coefficient is input into the instruction generation template to generate a collaborative control instruction set including distributed robot motion constraint conditions.

3. The method according to claim 2, characterized in that The step of performing joint constraint processing on the weight parameter and the dynamic threat coefficient to obtain a weight parameter distribution result includes: Dividing the variation range of the dynamic threat coefficient into continuous subintervals, each subinterval corresponding to a conversion function; According to the hierarchical position corresponding to the weight parameter in the hierarchical communication link, obtaining an initial conversion factor associated with the hierarchical position; In the subinterval, performing an association calculation on the dynamic threat coefficient and the initial conversion factor according to the conversion function to obtain an associated subinterval parameter; The associated sub-interval parameters are subjected to interval boundary fusion to obtain fused parameters, and inter-level balance constraints are executed on the fused parameters to obtain a weight parameter distribution result.

4. The method according to claim 3, characterized in that The step of performing an associated calculation on the dynamic threat coefficient and the initial conversion factor within the subinterval according to the conversion function to obtain associated subinterval parameters includes: Determining a slope adjustment factor of the conversion function according to the span range of the subinterval, and scaling the dynamic threat coefficient according to the slope adjustment factor to generate a dynamic threat value; In the subinterval, a level adjustment factor is obtained according to the level position corresponding to the weight parameter, and the dynamic threat value is cross-correlated with the level adjustment factor to obtain an initial conversion factor; An interval accumulation process is performed on the initial conversion factor, and the accumulation process result is superimposed on the dynamic threat value to generate an associated sub-interval parameter.

5. The method according to claim 1, wherein The distributed robot environment space data and motion parameters of dynamic obstacles are collected, and the environment space data is subjected to cross-robot joint compression processing to obtain a distributed perception data set, including: Dividing the spatial perception range of the distributed robot into overlapping areas, and dividing the environmental spatial data into discrete block data sets containing redundant identifiers; In the discrete block data set, repeated feature matching processing is performed on the overlapping area data of adjacent distributed robots, and redundant data points exceeding a set repetition threshold are deleted to generate a de-redundant block data set; Extracting high-density feature points and static environment feature points in the motion parameters of dynamic obstacles from the de-redundant block data set, and separating the high-density feature points from the static environment feature points to generate a dynamic feature data set and a static feature data set; The dynamic feature dataset is subjected to trajectory extension prediction according to the movement direction of the dynamic obstacle to obtain an extended dynamic feature sequence, and the extended dynamic feature sequence is spatially spliced ​​with the static feature dataset to generate a distributed perception dataset.

6. The method according to claim 5, characterized in that The step of performing trajectory extension prediction on the dynamic feature dataset according to the movement direction of the dynamic obstacle to obtain an extended dynamic feature sequence includes: Extracting obstacle position coordinates for consecutive time periods from the dynamic feature data set, and calculating a motion vector of the dynamic obstacle based on a change in the obstacle position coordinates; Taking the obstacle position coordinates in the current time period as a reference point, the velocity distance of the dynamic obstacle is extended along the direction of the motion vector by a preset multiple to generate a predicted trajectory point sequence; A spatial credibility verification process is performed on the predicted trajectory point sequence, and the verified predicted trajectory point sequence is timestamp aligned and merged with the dynamic feature dataset to obtain an extended dynamic feature sequence.

7. The method according to claim 1, characterized in that The step of driving the distributed robot to perform edge computing resource interaction through the synchronized collaborative control instruction set to obtain a collaborative motion path of the distributed robot includes: Decomposing the synchronized collaborative control instruction set into an action sequence corresponding to the distributed robot, wherein the action sequence includes a target displacement and an execution priority tag; Allocating a resource occupancy ratio of edge computing to the distributed robot based on the execution priority label, and determining a computing coverage range of dynamic obstacles according to the resource occupancy ratio; Dynamically adjusting the boundary conditions of the calculation coverage range according to the target displacement and the resource occupancy ratio, and calculating the overlapping buffer area of ​​the distributed robot in the moving direction based on the adjusted calculation coverage range; The motion parameters of the dynamic obstacle and the displacement state of the distributed robot are aligned and then input into the overlapping buffer area to generate a coordinated motion path of the distributed robot.

8. A distributed robot collaborative control system based on edge computing, characterized in that: include: An acquisition module collects environmental spatial data of distributed robots and motion parameters of dynamic obstacles, and performs cross-robot joint compression processing on the environmental spatial data to obtain a distributed perception data set; an adjustment module, which dynamically constructs hierarchical communication links in the distributed robot group based on the spatial density differences of the distributed perception data set, and adjusts the priorities of signal transmission paths in the hierarchical communication links according to real-time changes in the spatial density differences to obtain adjusted signal transmission path priorities; a generation module, converting the adjusted signal transmission path priority into a communication path weight parameter, and generating a collaborative control instruction set in combination with the motion parameter; a synchronization module, synchronizing the collaborative control instruction set with the dynamic obstacle perceived by the distributed robot across nodes to obtain a synchronized collaborative control instruction set; A driving module drives the distributed robot to perform resource interaction behavior of edge computing through the synchronized collaborative control instruction set to obtain a collaborative motion path of the distributed robot.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a distributed robot collaborative control method based on edge computing as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, a distributed robot collaborative control method based on edge computing as described in any one of claims 1 to 7 is implemented.

Images