A method and system for cooperative work of multi-robot and internet of things device

Abstract

The application provides a kind of multi-robot and internet of things equipment collaborative work method, system, it is related to warehousing logistics automation technical field, method includes: obtaining to be handled task and the real-time load state of multiple robots;According to the real-time environmental data of multiple internet of things equipment collected, the ability atlas of internet of things equipment group is generated;Based on dynamic task evaluation model, according to to be handled task, the real-time load state of multiple robots and the ability atlas of internet of things equipment group, the three-dimensional task priority matrix of to be handled task is generated;Using the resource matching algorithm of reinforcement learning, based on the three-dimensional task priority matrix of to be handled task, to be handled task is assigned target robot-internet of things equipment combination.The application matches target robot-internet of things equipment combination for to be handled task through the real-time environmental data of internet of things equipment, pre-established dynamic task evaluation model, realizes the collaborative work of multi-robot and internet of things equipment in multiple scenes.

Description

Technical Field

[0001] This invention relates to the field of warehousing and logistics automation technology, specifically to a method and system for collaborative work of multiple robots and Internet of Things (IoT) devices. Background Technology

[0002] With the rapid development of IoT technology, IoT robots have been widely used in various fields such as industry, agriculture, healthcare, and the military. IoT robots collect data from the work site in real time through sensors and other devices, and transmit this data to a remote control center via a network, enabling remote control of the robot.

[0003] However, in existing collaborative work scenarios between robots and IoT devices, due to factors such as the large number of robots, complex tasks, and variable environments, the IoT devices in existing distributed systems do not respond to real-time environmental data in a timely manner. Furthermore, existing algorithms are inefficient in dynamic task allocation, making it difficult to guarantee the accuracy and real-time performance of control. Consequently, they cannot meet the collaborative work needs of multiple robots and IoT devices in various scenarios. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, the present invention provides a method for collaborative work of multiple robots and Internet of Things (IoT) devices, comprising:

[0005] Obtain the tasks to be processed and the real-time load status of multiple robots;

[0006] Based on real-time environmental data collected from multiple IoT devices by a distributed sensor network, a capability map of the IoT device group is generated.

[0007] Based on a pre-established dynamic task evaluation model, a three-dimensional task priority matrix is ​​generated according to the tasks to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group.

[0008] A resource matching algorithm based on reinforcement learning is used to assign a target robot-IoT device combination to the task to be processed based on the three-dimensional task priority matrix of the task to be processed.

[0009] Preferably, the process of establishing a dynamic task evaluation model includes:

[0010] Based on the capability map of IoT device groups, the entropy value of the topology of environmental obstacles and the dynamic interference factor are calculated to obtain the real-time environmental complexity of IoT device groups.

[0011] A multi-dimensional capability vector is constructed based on the specifications and historical performance data of IoT devices to obtain the capability quantification vector of IoT devices.

[0012] A dynamic task evaluation model is established based on real-time environmental complexity and capability quantification vectors.

[0013] Preferably, the expression for the three-dimensional task priority matrix is ​​as follows:

[0014] P(t,s,r)=α·T(t)+β·S(s)+γ·R(r)

[0015] Where P(t,s,r) is the three-dimensional task priority matrix, T(t) is the time dimension priority function, S(s) is the space dimension priority function, R(r) is the resource dimension priority function, α is the time dimension priority weight coefficient, β is the space dimension priority weight coefficient, and γ is the resource dimension priority weight coefficient.

[0016] Preferably, the expression for the time-dimension priority function T(t) is as follows:

[0017]

[0018] Where, d i t represents the deadline for the pending tasks. now ω is the current system time, ε is the minimum value before zero, λ is the decay factor, and ω is the minimum value before zero. i The remaining workload for tasks to be processed;

[0019] The expression for the time dimension priority weight coefficient α is as follows:

[0020]

[0021] Where α0 is the initial time dimension weight coefficient, d i t represents the deadline for the pending tasks. now τ is the current system time. critical The preset critical time threshold;

[0022] The expression for the spatial dimension priority function S(s) is as follows:

[0023]

[0024] Among them, C path For dynamic path cost, (x r ,y r (x) represents the robot's position coordinates. t ,y t ) represents the coordinates of the mission target's location, and σ represents the spatial attenuation radius; v is the sum of all edges (path segments) on the path. r Let μ be the robot's moving speed, μ be the obstacle penalty coefficient, and obstacle_density(e) be the obstacle density in the region where path segment e is located. For time cost;

[0025] The expression for the resource dimension priority function R(r) is as follows:

[0026]

[0027] Where n is the total number of resource types, ω i η represents the weight of the i-th resource type, and η is the excess capacity reward coefficient. Let i be the normalized task requirement value for the i-th type of resource. Let be the normalized device capability value of device j under the i-th type of resource. Let d be the total margin between the normalized equipment capacity of device j under resource type i and the normalized task requirement value under resource type i, and let sigmoid be the total margin mapped to the interval (0,1); i Let max(D) be the task's requirement value for the i-th type of resource, and max(D) be the task requirement vector. The maximum value in, For the demand values ​​of all resource types, cj i Let max(c) be the capability value of IoT device j for the i-th type of resource. j ) represents the capability vector of IoT devices. The maximum value, For device j, the capability values ​​for all resource types.

[0028] Preferably, after assigning a target robot-IoT device combination to the task to be processed, the method further includes:

[0029] In the process of the target robot-IoT device combination performing pending tasks, the status parameters of the IoT device group are monitored, and the resource allocation strategy is dynamically adjusted through an adaptive feedback mechanism.

[0030] Preferably, the resource allocation strategy is dynamically adjusted through an adaptive feedback mechanism, including:

[0031] Establish a composite evaluation index that includes task completion timeliness, equipment reliability, and energy efficiency;

[0032] Based on a composite evaluation index, control parameters are optimized using a deep deterministic strategy gradient algorithm, and resource allocation strategies are adjusted through these control parameters.

[0033] The preferred expression for the composite evaluation index is as follows:

[0034]

[0035] Where Q is the composite evaluation index value, and T completeE represents the timeframe for task completion. consumed S represents the energy efficiency value. reliability μ1 is the weighting coefficient for task completion timeliness, μ2 is the weighting coefficient for energy efficiency, and μ3 is the weighting coefficient for IoT device reliability.

[0036] The expression for the deep deterministic policy gradient algorithm is as follows:

[0037]

[0038] Where, π θ (a|s) is the parameterized policy function, and θ is the policy network parameter. Let θ be the objective function to be maximized through the policy network parameters θ, Ε[Q(s,a)] be the expected value of the action value function Q(s,a), λ be the policy update constraint coefficient, and KL(π) be the objective function to be maximized. θ ||π θold ) represents the KL divergence between the old and new strategies.

[0039] Based on the same inventive concept, the present invention also provides a collaborative working system for multiple robots and Internet of Things (IoT) devices, comprising:

[0040] The raw data acquisition module is used to acquire the tasks to be processed and the real-time load status of multiple robots;

[0041] The device group status acquisition module is used to generate a capability map of the IoT device group based on real-time environmental data of multiple IoT devices collected by the distributed sensor network.

[0042] The task priority matrix generation module is used to generate a three-dimensional task priority matrix for the tasks to be processed based on a pre-established dynamic task evaluation model, the real-time load status of multiple robots, and the capability map of the IoT device group.

[0043] The robot-device allocation module is used to allocate target robot-IoT device combinations to tasks based on a three-dimensional task priority matrix of the tasks to be processed, using a resource matching algorithm based on reinforcement learning.

[0044] Preferably, the process of establishing a dynamic task evaluation model includes:

[0045] Based on the capability map of IoT device groups, the entropy value of the topology of environmental obstacles and the dynamic interference factor are calculated to obtain the real-time environmental complexity of IoT device groups.

[0046] A multi-dimensional capability vector is constructed based on the specifications and historical performance data of IoT devices to obtain the capability quantification vector of IoT devices.

[0047] A dynamic task evaluation model is established based on real-time environmental complexity and capability quantification vectors.

[0048] Preferably, the expression for the three-dimensional task priority matrix is ​​as follows:

[0049] P(t,s,r)=α·T(t)+β·S(s)+γ·R(r)

[0050] Where P(t,s,r) is the three-dimensional task priority matrix, T(t) is the time dimension priority function, S(s) is the space dimension priority function, R(r) is the resource dimension priority function, α is the time dimension priority weight coefficient, β is the space dimension priority weight coefficient, and γ is the resource dimension priority weight coefficient.

[0051] Preferably, the expression for the time-dimension priority function T(t) is as follows:

[0052]

[0053] Where, d i t represents the deadline for the pending tasks. now ω is the current system time, ε is the minimum value before zero, λ is the decay factor, and ω is the minimum value before zero. i The remaining workload for tasks to be processed;

[0054] The expression for the time dimension priority weight coefficient α is as follows:

[0055]

[0056] Where α0 is the initial time dimension weight coefficient, d i t represents the deadline for the pending tasks. now τ is the current system time. critical The preset critical time threshold;

[0057] The expression for the spatial dimension priority function S(s) is as follows:

[0058]

[0059] Among them, C path For dynamic path cost, (x r ,y r (x) represents the robot's position coordinates. t ,y t ) represents the coordinates of the mission target's location, and σ represents the spatial attenuation radius; v is the sum of all edges (path segments) on the path. r Let μ be the robot's moving speed, μ be the obstacle penalty coefficient, and obstacle_density(e) be the obstacle density in the region where path segment e is located. For time cost;

[0060] The expression for the resource dimension priority function R(r) is as follows:

[0061]

[0062]

[0063] Where n is the total number of resource types, ω i η represents the weight of the i-th resource type, and η is the excess capacity reward coefficient. Let i be the normalized task requirement value for the i-th type of resource. Let be the normalized device capability value of device j under the i-th type of resource. Let d be the total margin between the normalized equipment capacity of device j under resource type i and the normalized task requirement value under resource type i, and let sigmoid be the total margin mapped to the interval (0,1); i Let max(D) be the task's requirement value for the i-th type of resource, and max(D) be the task requirement vector. The maximum value in, For the demand values ​​of all resource types, cj i Let max(c) be the capability value of IoT device j for the i-th type of resource. j ) represents the capability vector of IoT devices. The maximum value, For device j, the capability values ​​for all resource types.

[0064] Preferably, the system also includes a dynamic resource allocation adjustment module, used for:

[0065] In the process of the target robot-IoT device combination performing pending tasks, the status parameters of the IoT device group are monitored, and the resource allocation strategy is dynamically adjusted through an adaptive feedback mechanism.

[0066] Preferably, the dynamic resource allocation adjustment module is specifically used for:

[0067] Establish a composite evaluation index that includes task completion timeliness, equipment reliability, and energy efficiency;

[0068] Based on a composite evaluation index, control parameters are optimized using a deep deterministic strategy gradient algorithm, and resource allocation strategies are adjusted through these control parameters.

[0069] The preferred expression for the composite evaluation index is as follows:

[0070]

[0071] Where Q is the composite evaluation index value, and T complete E represents the timeframe for task completion. consumed S represents the energy efficiency value. reliability μ1 is the weighting coefficient for task completion timeliness, μ2 is the weighting coefficient for energy efficiency, and μ3 is the weighting coefficient for IoT device reliability.

[0072] The expression for the deep deterministic policy gradient algorithm is as follows:

[0073]

[0074] Where, π θ (a|s) is the parameterized policy function, and θ is the policy network parameter. Let θ be the objective function to be maximized through the policy network parameters θ, Ε[Q(s,a)] be the expected value of the action value function Q(s,a), λ be the policy update constraint coefficient, and KL(π) be the objective function to be maximized. θ ||π θold ) represents the KL divergence between the old and new strategies.

[0075] Based on the same inventive concept, the present invention also provides an electronic device, comprising: at least one processor and a memory; wherein the memory and the processor are connected via a bus;

[0076] The memory is used to store one or more programs;

[0077] When the one or more programs are executed by the at least one processor, a collaborative working method for multiple robots and Internet of Things devices as described above is implemented.

[0078] Based on the same inventive concept, the present invention also provides a readable storage medium having an executable program stored thereon, which, when executed, implements the collaborative working method of multiple robots and Internet of Things devices as described above.

[0079] Compared with the closest existing technology, the present invention has the following beneficial effects:

[0080] This invention provides a method for collaborative work between multiple robots and IoT devices, comprising: acquiring a task to be processed and the real-time load status of multiple robots; generating a capability map of an IoT device group based on real-time environmental data of multiple IoT devices collected by a distributed sensor network; generating a three-dimensional task priority matrix of the task to be processed based on a pre-established dynamic task evaluation model, according to the task to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group; and employing a reinforcement learning resource matching algorithm to assign target robot-IoT device combinations to the task to be processed based on the three-dimensional task priority matrix. This invention generates a capability map of an IoT device group by collecting real-time environmental data of IoT devices, and simultaneously inputs this map, along with the task to be processed and the real-time load status of multiple robots, into a pre-established dynamic task evaluation model to obtain a three-dimensional task priority matrix. Then, based on the three-dimensional task priority matrix, target robot-IoT device combinations are matched to the task to be processed, enabling collaborative work between multiple robots and IoT devices in multiple scenarios. Attached Figure Description

[0081] Figure 1 A flowchart illustrating a collaborative working method for multiple robots and IoT devices provided by the present invention;

[0082] Figure 2 A structural diagram of a collaborative working system of multiple robots and Internet of Things devices provided by the present invention;

[0083] Figure 3 A schematic diagram of the electronic device provided by the present invention. Detailed Implementation

[0084] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0085] Example 1:

[0086] This invention provides a method for collaborative work between multiple robots and Internet of Things (IoT) devices, specifically... Figure 1 A flowchart illustrating the collaborative working method of multiple robots and IoT devices provided in this embodiment of the invention is shown in the figure, including the following steps:

[0087] S1: Obtain the tasks to be processed and the real-time load status of multiple robots;

[0088] S2: Generate a capability map of the IoT device group based on real-time environmental data collected from multiple IoT devices by a distributed sensor network;

[0089] S3: Based on a pre-established dynamic task evaluation model, a three-dimensional task priority matrix is ​​generated for the tasks to be processed, according to the real-time load status of multiple robots and the capability map of the IoT device group.

[0090] S4: Employ a resource matching algorithm based on reinforcement learning, and assign a target robot-IoT device combination to the task to be processed based on the three-dimensional task priority matrix of the task to be processed.

[0091] This invention generates a capability map of IoT device groups by collecting real-time environmental data from IoT devices. Simultaneously, it inputs this map, along with the tasks to be processed and the real-time load status of multiple robots, into a pre-established dynamic task evaluation model to obtain a three-dimensional task priority matrix. Then, based on the three-dimensional task priority matrix, it matches target robot-IoT device combinations for the tasks to be processed, enabling collaborative work of multiple robots and IoT devices in multiple scenarios.

[0092] In this invention, the pending tasks and the real-time load status of multiple robots are acquired. The pending tasks can be tasks that the robots need to perform, such as handling, inspection, or processing. In some optional embodiments, the pending tasks can be characterized by a task description tuple. For example, the constructed task description tuple can be T = {TaskID, Type, Priority, ResourceReq, Location, Deadline}. Here, TaskID is the task authentication information, Type is the task type (which can be handling, inspection, or processing), Priority is the task priority, ResourceReq is the task source sequence, Location is the task location, and Deadline is the task completion deadline. It is understood that the above task description tuple is merely an example and does not constitute a limitation of this invention.

[0093] The robot's real-time load status can be collected by an embedded monitoring module installed on the robot. The collected real-time load status of the robot includes, but is not limited to: power load: battery SOC (State of Charge) and battery SOH (State of Health); mechanical load: joint torque sensor data; computational load: processor utilization; task queue depth: a circular buffer is used to record tasks to be processed.

[0094] Next, real-time environmental data from multiple IoT devices is collected using a distributed sensor network. Specifically, intelligent sensor nodes with multimodal sensing capabilities can be deployed, a data fusion processing channel based on a Kalman filter can be established, and a dynamic sampling frequency adjustment mechanism can be configured to adapt to environmental changes. The real-time environmental data collected from IoT devices includes, but is not limited to: indoor positioning data, environmental monitoring data, and dynamic obstacle detection data. Specifically, indoor positioning data can be obtained through UWB (Ultra-Wideband) + TDOA (Time Difference of Arrival) hybrid positioning; environmental monitoring data can be obtained by collecting data such as temperature, humidity, and light intensity through Modbus RTU (Modbus Remote Terminal Unit); and dynamic obstacle detection data can be obtained by combining millimeter-wave radar point cloud data with the running trajectories of multiple robots.

[0095] Furthermore, based on real-time environmental data collected from multiple IoT devices, a capability map of the IoT device group is generated. Specifically, device capability characteristics can be extracted from real-time environmental data to construct a five-dimensional capability vector. For example, the constructed five-dimensional capability vector can be C = {Processing, Storage, Communication, Mobility, Sensing}, where Processing is the device's computing power, Storage is the device's storage space, Communication is the communication bandwidth, Mobility is the device's mobility (0, 0-1 scale for static devices), and Sensing is the sensor accuracy (normalized to the 0-1 range). It is understood that the above five-dimensional capability vector is an example of the generated capability map of the IoT device group and does not constitute a limitation of the present invention.

[0096] The aforementioned tasks to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group are all initial preparation data. After obtaining the initial preparation data, based on the pre-established dynamic task evaluation model, a three-dimensional task priority matrix for the tasks to be processed is generated according to the tasks to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group.

[0097] The process of pre-establishing a dynamic task evaluation model includes: calculating the entropy value of the topology of environmental obstacles and the dynamic interference factor based on the capability map of the IoT device group to obtain the real-time environmental complexity of the IoT device group; constructing a multi-dimensional capability vector based on the specification parameters and historical performance data of the IoT devices to obtain the capability quantification vector of the IoT devices; and establishing a dynamic task evaluation model based on the real-time environmental complexity and the capability quantification vector.

[0098] The three-dimensional task priority matrix generated in the above manner includes time-dimension priority, spatial-dimension priority, and resource-dimension priority, along with their respective weight coefficients. Specifically, the time-dimension priority is used to calculate the urgency coefficient between the task's deadline and its current progress; the spatial-dimension priority is used to assess the topological relationship between the task's location and the distribution of IoT devices; and the resource-dimension priority is used to obtain the matching degree between the task's requirements and the capabilities of the IoT devices.

[0099] Specifically, the expression for the three-dimensional task priority matrix is ​​as follows:

[0100] P(t,s,r)=α·T(t)+β·S(s)+γ·R(r)

[0101] Where P(t,s,r) is the three-dimensional task priority matrix, T(t) is the time dimension priority function, S(s) is the space dimension priority function, R(r) is the resource dimension priority function, α is the time dimension priority weight coefficient, β is the space dimension priority weight coefficient, and γ is the resource dimension priority weight coefficient.

[0102] Calculating time-dimensional priority requires obtaining the time-dimensional priority function and the time-dimensional priority coefficient. The expression for the time-dimensional priority function T(t) is as follows:

[0103]

[0104] Where, d i t represents the deadline for the pending tasks. now ω is the current system time, ε is the minimum value before zero, λ is the decay factor, and ω is the minimum value before zero. i The remaining workload for tasks to be processed;

[0105] When the difference between the deadline of the pending task and the current system time is less than a preset critical time threshold, the time dimension weight coefficient is triggered to increase exponentially. The expression for the time dimension priority weight coefficient α is as follows:

[0106]

[0107] Where α0 is the initial time dimension weight coefficient, d i t represents the deadline for the pending tasks. now τ is the current system time. critical The preset critical time threshold;

[0108] Dijkstra's algorithm is a classic algorithm for finding the single-source shortest path in a weighted graph. Its core objective is to find the shortest path or one of the shortest paths from the origin to all other nodes in the graph. It is applicable to graph structures with non-negative weights.

[0109] The calculation process for spatial dimension priority is as follows: An improved Dijkstra algorithm (proposed by Dutch computer scientist Edsger W. Dijkstra) is used to calculate the dynamic path cost C. path The expression for dynamic path cost is as follows:

[0110]

[0111] Among them, v r For the robot's moving speed, To sum over all edges (path segments) on the path, ||e|| is the Euclidean length of edge e, μ is the obstacle penalty coefficient, and obstacle_density(e) is the obstacle density of the region containing path segment e. For time cost, μ·obstacle_density(e) is the obstacle density cost.

[0112] The expression for the spatial dimension priority function S(s) is as follows:

[0113]

[0114] Among them, (x r ,y r (x) represents the robot's position coordinates. t ,y t ) represents the coordinates of the mission target location, and σ represents the spatial attenuation radius.

[0115] The calculation process of the resource dimension priority function is as follows:

[0116] The task requirement vector and equipment capability vector are normalized as follows:

[0117]

[0118] Among them, the task requirement vector Device Capability Vector

[0119] The improved expression for cosine similarity calculation (i.e., the resource dimension priority function) is as follows:

[0120]

[0121] Where n is the total number of resource types, ω iη represents the weight of the i-th resource type, and η is the excess capacity reward coefficient. Let i be the normalized task requirement value for the i-th type of resource. Let be the normalized device capability value of device j under the i-th type of resource. Let d be the total margin between the normalized equipment capacity of device j under resource type i and the normalized task requirement value under resource type i, and let sigmoid be the total margin mapped to the interval (0,1); i Let max(D) be the task's requirement value for the i-th type of resource, and max(D) be the task requirement vector. The maximum value in, For the demand values ​​of all resource types, cj i Let max(c) be the capability value of IoT device j for the i-th type of resource. j ) represents the capability vector of IoT devices. The maximum value, For device j, the capability values ​​for all resource types.

[0122] After generating the three-dimensional task priority matrix of the tasks to be processed in the above manner, a resource matching algorithm based on reinforcement learning is used to assign a target robot-IoT device combination to the tasks to be processed.

[0123] After obtaining the three-dimensional task priority matrix, matrix fusion and updating can be performed. Specifically, the three-dimensional tensor representation of the three-dimensional task priority matrix is ​​as follows: Where T represents the time dimension priority, S represents the spatial dimension priority, and R represents the resource dimension priority. The spatial dimension priority data is updated at preset intervals, and a local matrix refresh is triggered when the task status changes.

[0124] In the architecture of the reinforcement learning resource matching algorithm, the input features are slices of a three-dimensional priority matrix. (The 3D data of the current time slice) is used to extract spatial-resource features through 3D CNN (Convolutional Neural Network) and LSTM (Long Short-Term Memory) is used to capture temporal dependencies.

[0125] After assigning a target robot-IoT device combination to a task to be processed, in order to further improve the accuracy of the assignment, in some optional implementations, the method further includes: monitoring the status parameters of the IoT device group while the target robot-IoT device combination is performing the task to be processed, and dynamically adjusting the resource allocation strategy through an adaptive feedback mechanism.

[0126] The establishment of the adaptive feedback mechanism includes: building a knowledge graph of the collaborative efficiency of IoT devices; designing a long-term memory model based on the Transformer architecture to achieve online incremental learning and optimization of task allocation strategies.

[0127] Specifically, the resource allocation strategy is dynamically adjusted through an adaptive feedback mechanism, including: establishing a composite evaluation index that includes task completion time, equipment reliability, and energy efficiency; optimizing control parameters based on the composite evaluation index and a deep deterministic strategy gradient algorithm; and adjusting the resource allocation strategy through the control parameters.

[0128] The expression for the composite evaluation index is as follows:

[0129]

[0130] Where Q is the composite evaluation index value, and T complete E represents the timeframe for task completion. consumed S represents the energy efficiency value. reliability μ1 is the weighting coefficient for task completion timeliness, μ2 is the weighting coefficient for energy efficiency, and μ3 is the weighting coefficient for IoT device reliability.

[0131] The expression for the deep deterministic policy gradient algorithm is as follows:

[0132]

[0133] Where, π θ (a|s) is the parameterized policy function, and θ is the policy network parameter. Let θ be the objective function to be maximized through the policy network parameters θ, Ε[Q(s,a)] be the expected value of the action value function Q(s,a), λ be the policy update constraint coefficient, and KL(π) be the objective function to be maximized. θ ||π θold ) represents the KL divergence between the old and new strategies.

[0134] This invention provides a method for collaborative work between multiple robots and IoT devices, comprising: acquiring a task to be processed and the real-time load status of multiple robots; generating a capability map of an IoT device group based on real-time environmental data of multiple IoT devices collected by a distributed sensor network; generating a three-dimensional task priority matrix of the task to be processed based on a pre-established dynamic task evaluation model, according to the task to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group; and employing a reinforcement learning resource matching algorithm to assign a target robot-IoT device combination to the task to be processed based on the three-dimensional task priority matrix. This invention generates a capability map of an IoT device group by collecting real-time environmental data of IoT devices, and simultaneously inputs this map, along with the task to be processed and the real-time load status of multiple robots, into a pre-established dynamic task evaluation model to obtain a three-dimensional task priority matrix. Then, based on the three-dimensional task priority matrix, a target robot-IoT device combination is matched to the task to be processed, enabling collaborative work between multiple robots and IoT devices in multiple scenarios.

[0135] Example 2:

[0136] Based on the same inventive concept, this invention also provides a collaborative working system 200 for multiple robots and Internet of Things (IoT) devices, the system structure of which is as follows: Figure 2 As shown, the system includes:

[0137] The raw data acquisition module 201 is used to acquire the tasks to be processed and the real-time load status of multiple robots;

[0138] The device group status acquisition module 202 is used to generate a capability map of the IoT device group based on real-time environmental data of multiple IoT devices collected by the distributed sensor network.

[0139] The task priority matrix generation module 203 is used to generate a three-dimensional task priority matrix of the task to be processed based on a pre-established dynamic task evaluation model, according to the task to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group.

[0140] The robot-device allocation module 204 is used to allocate a target robot-IoT device combination to the task to be processed based on the three-dimensional task priority matrix of the task to be processed using a resource matching algorithm of reinforcement learning.

[0141] Preferably, the process of establishing a dynamic task evaluation model includes:

[0142] Based on the capability map of IoT device groups, the entropy value of the topology of environmental obstacles and the dynamic interference factor are calculated to obtain the real-time environmental complexity of IoT device groups.

[0143] A multi-dimensional capability vector is constructed based on the specifications and historical performance data of IoT devices to obtain the capability quantification vector of IoT devices.

[0144] A dynamic task evaluation model is established based on real-time environmental complexity and capability quantification vectors.

[0145] Preferably, the expression for the three-dimensional task priority matrix is ​​as follows:

[0146] P(t,s,r)=α·T(t)+β·S(s)+γ·R(r)

[0147] Where P(t,s,r) is the three-dimensional task priority matrix, T(t) is the time dimension priority function, S(s) is the space dimension priority function, R(r) is the resource dimension priority function, α is the time dimension priority weight coefficient, β is the space dimension priority weight coefficient, and γ is the resource dimension priority weight coefficient.

[0148] Preferably, the expression for the time-dimension priority function T(t) is as follows:

[0149]

[0150] Where, d i t represents the deadline for the pending tasks. now ω is the current system time, ε is the minimum value before zero, λ is the decay factor, and ω is the minimum value before zero. i The remaining workload for tasks to be processed;

[0151] The expression for the time dimension priority weight coefficient α is as follows:

[0152]

[0153] Where α0 is the initial time dimension weight coefficient, d i t represents the deadline for the pending tasks. now τ is the current system time. critical The preset critical time threshold;

[0154] The expression for the spatial dimension priority function S(s) is as follows:

[0155]

[0156] Among them, C path For dynamic path cost, (x r ,y r (x) represents the robot's position coordinates. t ,y t ) represents the coordinates of the mission target's location, and σ represents the spatial attenuation radius; v is the sum of all edges (path segments) on the path. r Let μ be the robot's moving speed, μ be the obstacle penalty coefficient, and obstacle_density(e) be the obstacle density in the region where path segment e is located. For time cost;

[0157] The expression for the resource dimension priority function R(r) is as follows:

[0158]

[0159] Where n is the total number of resource types, ω i η represents the weight of the i-th resource type, and η is the excess capacity reward coefficient. Let i be the normalized task requirement value for the i-th type of resource. Let be the normalized device capability value of device j under the i-th type of resource. Let d be the total margin between the normalized equipment capacity of device j under resource type i and the normalized task requirement value under resource type i, and let sigmoid be the total margin mapped to the interval (0,1); i Let max(D) be the task's requirement value for the i-th type of resource, and max(D) be the task requirement vector. The maximum value in, For the demand values ​​of all resource types, cj i Let max(c) be the capability value of IoT device j for the i-th type of resource. j ) represents the capability vector of IoT devices. The maximum value, For device j, the capability values ​​for all resource types.

[0160] Preferably, the system also includes a dynamic resource allocation adjustment module, used for:

[0161] In the process of the target robot-IoT device combination performing pending tasks, the status parameters of the IoT device group are monitored, and the resource allocation strategy is dynamically adjusted through an adaptive feedback mechanism.

[0162] Preferably, the dynamic resource allocation adjustment module is specifically used for:

[0163] Establish a composite evaluation index that includes task completion time, equipment reliability, and energy efficiency;

[0164] Based on a composite evaluation index, control parameters are optimized using a deep deterministic strategy gradient algorithm, and resource allocation strategies are adjusted through these control parameters.

[0165] The preferred expression for the composite evaluation index is as follows:

[0166]

[0167] Where Q is the composite evaluation index value, and T complete E represents the timeframe for task completion. consumed S represents the energy efficiency value. reliability μ1 is the weighting coefficient for task completion timeliness, μ2 is the weighting coefficient for energy efficiency, and μ3 is the weighting coefficient for IoT device reliability.

[0168] The expression for the deep deterministic policy gradient algorithm is as follows:

[0169]

[0170] Where, π θ (a|s) is the parameterized policy function, and θ is the policy network parameter. Let θ be the objective function to be maximized through the policy network parameters θ, Ε[Q(s,a)] be the expected value of the action value function Q(s,a), λ be the policy update constraint coefficient, and KL(π) be the objective function to be maximized. θ ||π θold ) represents the KL divergence between the old and new strategies.

[0171] Example 3:

[0172] Based on the same inventive concept, such as Figure 3 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0173] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in a readable storage medium to implement the corresponding method flow or corresponding function, so as to realize the steps of the collaborative working method of multiple robots and Internet of Things devices in the above embodiments.

[0174] Example 4:

[0175] Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the readable storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the storage medium to implement the steps of the collaborative working method for multiple robots and IoT devices described in the above embodiments.

[0176] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0177] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0178] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0179] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0180] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the application, but these changes, modifications or equivalent substitutions are all within the scope of protection of the claims pending approval.

Claims

1. A method for collaborative work of multiple robots and Internet of Things (IoT) devices, characterized in that, include: Obtain the tasks to be processed and the real-time load status of multiple robots; A capability map of an IoT device group is generated based on real-time environmental data collected from multiple IoT devices by a distributed sensor network; wherein, the capability map is based on the real-time environmental data and is represented by multi-dimensional capability vectors of the IoT devices. Based on a pre-established dynamic task evaluation model, a three-dimensional task priority matrix is ​​generated for the task to be processed, according to the task to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group. A resource matching algorithm based on reinforcement learning is used to assign a target robot-IoT device combination to the task to be processed based on the three-dimensional task priority matrix of the task to be processed. The expression for the three-dimensional task priority matrix is ​​as follows: in, This is a three-dimensional task priority matrix. For time-dimension priority functions, For spatial dimension priority function, For resource-level priority functions, This refers to the priority weight coefficient for the time dimension. This refers to the priority weight coefficient for the spatial dimension. This refers to the priority weight coefficient for the resource dimension. The time-dimensional priority function The expression is as follows: in, The deadline for pending tasks. The current system time. To prevent the minimum value from being divided by zero, As the attenuation factor, The remaining workload for tasks to be processed; The time dimension priority weight coefficient The expression is as follows: in, These are the initial time dimension weighting coefficients. The deadline for pending tasks. The current system time. The preset critical time threshold; The spatial dimension priority function The expression is as follows: in, For dynamic path cost, For the robot's position coordinates, The coordinates of the mission target location. The spatial attenuation radius; The sum of all edges on the path. For the robot's moving speed, The obstacle penalty coefficient. Path segment Obstacle density in the area For time cost; The resource dimension priority function The expression is as follows: Where n is the total number of resource types, Let i be the weight of the i-th resource type. This is the excess capacity reward coefficient; Let i be the normalized task requirement value for the i-th type of resource. Let be the normalized device capability value of device j under the i-th type of resource. Let $\frac{j}{i}$ be the total margin between the normalized equipment capacity of equipment j under the $i$-th resource type and the normalized task requirement value under the $i$-th resource type. Map the total surplus to the interval (0,1); Let i be the resource requirement value for the task. Task requirement vector The maximum value in, For the demand values ​​of all resource types, , Let j be the capability value of IoT device j for the i-th type of resource. Capability vectors for IoT devices The maximum value, For device j, the capability values ​​for all resource types. .

2. The method according to claim 1, characterized in that, The process of establishing a dynamic task evaluation model includes: Based on the capability map of the IoT device group, the entropy value of the topology of environmental obstacles and the dynamic interference factor are calculated to obtain the real-time environmental complexity of the IoT device group. A multi-dimensional capability vector is constructed based on the specifications and historical performance data of the IoT device to obtain the capability quantification vector of the IoT device. A dynamic task evaluation model is established based on the real-time environment complexity and the capability quantification vector.

3. The method according to claim 1, characterized in that, After assigning the target robot-IoT device combination to the task to be processed, the method further includes: During the execution of the task by the target robot-IoT device combination, the status parameters of the IoT device group are monitored, and the resource allocation strategy is dynamically adjusted through an adaptive feedback mechanism.

4. The method according to claim 3, characterized in that, The method of dynamically adjusting the resource allocation strategy through an adaptive feedback mechanism includes: Establish a composite evaluation index that includes task completion timeliness, equipment reliability, and energy efficiency; Based on the composite evaluation index, the control parameters are optimized using a deep deterministic strategy gradient algorithm, and the resource allocation strategy is adjusted using the control parameters.

5. The method according to claim 4, characterized in that, The expression for the composite evaluation index is as follows: in, The composite evaluation index value, This is the timeframe for task completion. This is the energy efficiency value. This is the reliability index for IoT devices. The weighting coefficient for task completion timeliness. This is the energy efficiency weighting coefficient. This is a reliability weighting coefficient for IoT devices; The expression for the deep deterministic policy gradient algorithm is as follows: in, For parameterized policy functions, For policy network parameters, To pass through policy network parameters Maximize the objective function, Action value function Expected value Update the constraint coefficients for the strategy. The KL divergence between the old and new strategies.

6. A collaborative working system for multiple robots and Internet of Things (IoT) devices, characterized in that, include: The raw data acquisition module is used to acquire the tasks to be processed and the real-time load status of multiple robots; The device group status acquisition module is used to generate a capability map of the IoT device group based on real-time environmental data of multiple IoT devices collected by the distributed sensor network. The task priority matrix generation module is used to generate a three-dimensional task priority matrix for the task to be processed based on a pre-established dynamic task evaluation model, according to the task to be processed, the real-time load status of multiple robots, and the capability map of the IoT device group. The robot-device allocation module is used to allocate a target robot-IoT device combination to the task to be processed based on the three-dimensional task priority matrix of the task to be processed using a resource matching algorithm based on reinforcement learning. The expression for the three-dimensional task priority matrix is ​​as follows: in, This is a three-dimensional task priority matrix. For time-dimension priority functions, For spatial dimension priority function, For resource-level priority functions, This refers to the priority weight coefficient for the time dimension. This refers to the priority weight coefficient for the spatial dimension. This refers to the priority weight coefficient for the resource dimension. The time-dimensional priority function The expression is as follows: in, The deadline for pending tasks. The current system time. To prevent the minimum value from being divided by zero, As the attenuation factor, The remaining workload for tasks to be processed; The time dimension priority weight coefficient The expression is as follows: in, These are the initial time dimension weighting coefficients. The deadline for pending tasks. The current system time. The preset critical time threshold; The spatial dimension priority function The expression is as follows: in, For dynamic path cost, For the robot's position coordinates, The coordinates of the mission target location. The spatial attenuation radius; The sum of all edges on the path. For the robot's moving speed, The obstacle penalty coefficient. Path segment Obstacle density in the area For time cost; The resource dimension priority function The expression is as follows: Where n is the total number of resource types, Let i be the weight of the i-th resource type. This is the excess capacity reward coefficient; Let i be the normalized task requirement value for the i-th type of resource. Let be the normalized device capability value of device j under the i-th type of resource. Let $\frac{j}{i}$ be the total margin between the normalized equipment capacity of equipment j under the $i$-th resource type and the normalized task requirement value under the $i$-th resource type. Map the total surplus to the interval (0,1); Let i be the resource requirement value for the task. Task requirement vector The maximum value in, For the demand values ​​of all resource types, , Let j be the capability value of IoT device j for the i-th type of resource. Capability vectors for IoT devices The maximum value, For device j, the capability values ​​for all resource types. .

7. An electronic device, characterized in that, include: At least one processor and memory; The memory and processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a collaborative working method for multiple robots and Internet of Things devices as described in any one of claims 1 to 5 is implemented.

8. A readable storage medium, characterized in that, It contains an execution program, which, when executed, implements the collaborative working method of multiple robots and Internet of Things devices as described in any one of claims 1 to 5.

Images