Multi-source control instruction arbitration method and device

Abstract

The application relates to a multi-source control instruction arbitration method and device. Through first type conflict detection and second type conflict detection, a task execution conflict situation existing in an end-side execution mechanism is detected, an adjusted conflict-free time axis task sequence is generated based on the conflict situation through time axis task sequence dynamic planning adjustment, so that the end-side execution mechanism can smoothly execute multiple tasks in turn, and the occurrence of the task conflict situation is reduced.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to a multi-source control command arbitration method and apparatus. Background Technology

[0002] With the rapid development of the Internet of Things (IoT), artificial intelligence (AI), and edge computing technologies, smart homes, smart cockpits, and wearable health devices have gradually entered people's lives. Currently, there are various intelligent control systems on the market. Especially after the interconnection of people, vehicles, and homes, conflicts in task execution by edge-side actuators can easily occur in multi-agent environments. Summary of the Invention

[0003] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a multi-source control command arbitration method, which solves the technical problem that the end-side actuator of the human-vehicle-home control system is prone to task execution conflicts.

[0004] To achieve the above objectives, the main technical solutions adopted by the present invention include: In a first aspect, embodiments of the present invention provide a multi-source control command arbitration method.

[0005] The multi-source control command arbitration method proposed in this invention is applied to the edge subsystem of a human-vehicle-home control system. The human-vehicle-home control system includes a terminal subsystem, an edge subsystem, and a cloud subsystem. The terminal subsystem is deployed in home and vehicle scenarios and includes various terminal actuators. The edge subsystem is deployed at a home main control edge computing center, and the cloud subsystem is deployed on a cloud server. The method includes: Obtain a set of instructions to be arbitrated; wherein the set of instructions to be arbitrated contains at least two candidate instructions, and each candidate instruction is generated by an agent in the multi-Agent cluster of the edge subsystem; Obtain predefined declarative constraint rules from the constraint rule library; wherein, the declarative constraint rules include at least: exclusive use rules for each end-side actuator and the shortest state switching time for each end-side actuator between predetermined action type pairs; The first type of conflict is detected according to the exclusive use rule, and the second type of conflict is detected according to the shortest state switching time when no first type of resource conflict is detected. When the first type of conflict detection or the second type of resource conflict is detected, arbitration is performed according to a predefined priority matrix, and the timeline task sequence is dynamically planned and adjusted to generate an adjusted conflict-free timeline task sequence; wherein, the timeline task sequence is obtained by sorting the candidate instructions of the same end-side execution mechanism according to the execution start time; A final control command queue is generated based on the adjusted conflict-free timeline task sequence, and the final control command queue is sent to the corresponding end-side execution mechanism through the edge-side subsystem.

[0006] In some embodiments, each candidate instruction includes at least a target end-side actuator identifier and an action execution time period; The first type of conflict detection based on the exclusive use rule includes: Parse the target end-side actuator identifier and the action execution time period in each candidate instruction; The target end-side actuator identifier and the action execution time period in each of the parsed candidate instructions are compared pairwise. When it is detected that the same end-side actuator is assigned two or more different candidate instructions within the same action execution time period, it is determined that the first type of resource conflict exists.

[0007] In some embodiments, the step of performing second-type conflict detection based on the shortest state transition time includes: The timeline task sequence is obtained by sorting the candidate instructions involving the same end-side execution agency in the set of instructions to be arbitrated according to their execution start time; Iterate through the two adjacent candidate instructions in the timeline task sequence in turn, and calculate the actual interval between the execution end time of the preceding candidate instruction and the execution start time of the adjacent following candidate instruction. Compare the actual interval duration with the corresponding shortest state transition time; When the actual interval duration is less than the minimum time required for state switching, a second type of resource conflict is determined to exist.

[0008] In some embodiments, the arbitration process based on a predefined priority matrix and the dynamic programming adjustment of the timeline task sequence to generate an adjusted conflict-free timeline task sequence include: Based on the priority matrix, the priority levels of the multiple conflicting candidate instructions are obtained. The priority levels of the multiple candidate instructions are compared, and the candidate instruction with the highest priority level is determined as the baseline instruction. Using the execution time period of the baseline instruction as the anchor point, the execution time periods of the candidate instructions other than the baseline instruction in the time axis task sequence are sequentially shifted forward along the time axis, so that the actual interval between adjacent candidate instructions is greater than or equal to the shortest time consumption of the corresponding state transition, thereby generating the adjusted conflict-free time axis task sequence.

[0009] In some embodiments, the constraint rule base includes a device type mapping table; wherein the device type mapping table marks end-side actuators as motion-coupled and electrical-decoupled types; Before performing the first type of conflict detection according to the exclusive use rule, the method includes: In response to obtaining the set of instructions to be arbitrated, the device type mapping table is queried based on the target end-side actuator identifier in each candidate instruction; When the target-side actuator is electrically decoupled, the first type of conflict detection and the second type of conflict detection are bypassed, and the candidate instruction is directly marked as a conflict-free instruction. When the target-side actuator is of the motion-coupled type, the first type of conflict detection and the second type of conflict detection are performed.

[0010] In some embodiments, before obtaining predefined declarative constraint rules from the constraint rule base, the method includes: Determine the shortest time consumption for state transitions between predetermined action type pairs for each end-side actuator; wherein, the calibration method for determining the shortest time consumption for state transitions between predetermined action type pairs for each end-side actuator includes: Each action type of the actuator is encoded, and the action type is represented by a node. A directed edge represents the switch from the first action type to the second action type. The edge weight is the shortest time consumption of the state switch, and an action type state transition graph is constructed.

[0011] In some embodiments, generating the final control instruction queue based on the adjusted conflict-free timeline task sequence includes: Based on the adjusted conflict-free timeline task sequence, a new idle waiting period between two adjacent candidate instructions is determined, and an idle waiting delay instruction is automatically generated; wherein, the duration of the idle waiting delay instruction is greater than or equal to the shortest state switching time of the corresponding adjacent candidate instruction; The candidate instructions and the idle wait delay instructions are arranged in chronological order to form a serialized final control instruction queue.

[0012] A second aspect of this invention provides a multi-source control command arbitration device applied to an edge subsystem of a human-vehicle-home control system. The human-vehicle-home control system includes a terminal subsystem, an edge subsystem, and a cloud subsystem. The terminal subsystem is deployed in home and vehicle scenarios and includes various terminal actuators. The edge subsystem is deployed at a home main control edge computing center, and the cloud subsystem is deployed on a cloud server. The device includes: The instruction acquisition unit is used to acquire a set of instructions to be arbitrated; wherein the set of instructions to be arbitrated contains at least two candidate instructions, and each candidate instruction is generated by an agent in the multi-Agent cluster of the edge subsystem; The rule acquisition unit is used to acquire predefined declarative constraint rules from the constraint rule library; wherein, the declarative constraint rules include at least: exclusive use rules for each end-side actuator and the shortest time consumption for state switching between predetermined action type pairs of each end-side actuator; The conflict detection unit is used to perform a first type of conflict detection according to the exclusive use rule, and to perform a second type of conflict detection according to the shortest state switching time when no first type of resource conflict is detected. The sequence generation unit is used to perform arbitration processing according to a predefined priority matrix when the first type of conflict detection or the second type of resource conflict is detected, and to dynamically plan and adjust the time axis task sequence to generate an adjusted conflict-free time axis task sequence; wherein, the time axis task sequence is obtained by sorting the candidate instructions of the same end-side execution mechanism according to the execution start time; The instruction issuing unit is used to generate a final control instruction queue based on the adjusted conflict-free timeline task sequence, and to issue the final control instruction queue to the corresponding end-side execution mechanism through the edge-side subsystem.

[0013] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of the method described in the first aspect above.

[0014] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described in the first aspect above.

[0015] This invention detects task execution conflicts in the end-side actuator through first-type and second-type conflict detection. Based on the conflict situation, it dynamically plans and adjusts the time-axis task sequence to generate an adjusted conflict-free time-axis task sequence, thereby enabling the end-side actuator to smoothly execute multiple tasks in sequence and reducing the occurrence of task conflicts. Attached Figure Description

[0016] Figure 1 A flowchart of a multi-source control command arbitration method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a multi-source control command arbitration device provided in an embodiment of the present invention. Detailed Implementation

[0017] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] The multi-source control command arbitration method proposed in this invention is used to solve the problem of task execution conflicts that easily occur in the end-side actuators of the human-vehicle-home control system in the prior art. By detecting the first type of conflict detection and the second type of conflict detection, the task execution conflict situation of the end-side actuator is detected. Based on the conflict situation, the time axis task sequence is dynamically planned and adjusted to generate an adjusted conflict-free time axis task sequence, thereby enabling the end-side actuator to smoothly execute multiple tasks in sequence, thereby reducing the occurrence of task conflicts.

[0019] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0020] Figure 1 This is a flowchart illustrating a multi-source control command arbitration method provided in an embodiment of the present invention. Figure 1 As shown, the multi-source control command arbitration method proposed in this embodiment of the invention is applied to the edge subsystem of a human-vehicle-home control system. The human-vehicle-home control system includes a terminal subsystem, an edge subsystem, and a cloud subsystem. The terminal subsystem is deployed in home and vehicle scenarios and includes various terminal actuators. The edge subsystem is deployed at the home main control edge computing center, and the cloud subsystem is deployed on a cloud server. The method includes: Step 100: Obtain the set of instructions to be arbitrated; wherein the set of instructions to be arbitrated contains at least two candidate instructions, and each candidate instruction is generated by an Agent in the multi-Agent cluster of the edge subsystem; Step 110: Obtain predefined declarative constraint rules from the constraint rule library; wherein, the declarative constraint rules include at least: exclusive use rules for each end-side actuator and the shortest state transition time for each end-side actuator between predetermined action type pairs; wherein, the constraint rule library is stored locally on the edge-side subsystem, the exclusive use rules define that each end-side actuator can only be assigned to one control task at a time; the shortest state transition time is pre-calibrated based on the physical and mechanical characteristics of the end-side actuator; Step 120: Perform first-type conflict detection according to the exclusive use rule, and if no first-type resource conflict is detected, perform second-type conflict detection according to the shortest state switching time. Step 130: When the first type of conflict detection or the second type of resource conflict is detected, arbitration is performed according to the predefined priority matrix, and the time axis task sequence is dynamically planned and adjusted to generate an adjusted conflict-free time axis task sequence; wherein, the time axis task sequence is obtained by sorting the candidate instructions of the same end-side execution mechanism according to the execution start time; Step 140: Generate a final control command queue based on the adjusted conflict-free timeline task sequence, and send the final control command queue to the corresponding end-side execution mechanism through the edge-side subsystem.

[0021] In this exemplary embodiment, the priority matrix refers to a multi-level task priority mapping table, used to assign a specific priority level to different types of events. The priority matrix is ​​preset in the following order: emergency events (P0) have a higher priority than security events (P1), security events have a higher priority than comfort control events (P2), and comfort control events have a higher priority than energy-saving scheduling events (P3). Furthermore, for the same type of event, different priorities are assigned to different objects. For example, in emergency events, the priority for emergency care of the elderly is higher than that for emergency care of children, and the priority for emergency care of children is higher than that of ordinary adults. Similarly, security event priorities include the priority of front door security being higher than that of interior security.

[0022] Edge-side actuators refer to specific hardware devices deployed in the edge-side subsystem that receive control commands and execute physical actions, including but not limited to: dual-arm wheeled service robots, smart air conditioners, smart refrigerators, rice cookers, automatic cooking machines, washing machines, smart switches / sockets, vehicle air conditioning compressors, window motors, steering wheel vibration motors, etc. Comfort control events have higher priority than energy-saving scheduling events; for example, in a vehicle scenario, controlling seat adjustments has higher priority than controlling the vehicle's air conditioning. The home main control edge computing center can be networked with the vehicle controller.

[0023] The edge subsystem is deployed at the home-based edge computing center, running an operating system with real-time RT-Linux patches, and featuring a multi-agent cluster and the Harness scheduling engine. The cloud subsystem is deployed on a cloud server, containing a large language model and a global knowledge graph. Task execution records from each edge actuator can be uploaded to cloud storage.

[0024] In this exemplary embodiment, this embodiment includes three sub-scenarios, which respectively demonstrate the complete process of three detection / processing paths.

[0025] Scenario 1: A Type I resource conflict occurs (arbitration and dynamic programming are used to adjust the path). The user drove home at 18:00 on Friday. The edge subsystem received the following three candidate instructions in a short period of time, as shown in Table 1. All of them involve the same dual-arm wheeled service robot ROBOT_01: Table 1 Instructions Table 1

[0026] The Harness scheduling engine obtains the above three candidate instructions to form a set of instructions to be arbitrated. Then it reads the constraint rule library to obtain the exclusive use rule of ROBOT_01 (only one task can be executed at the same time) and the shortest time consumption of state switching between each action type (valve closing → child protection: 4.0s; child protection → valve closing: 4.0s; valve closing → home preparation: 2.0s).

[0027] The Harness scheduling engine first performs a first-type conflict detection based on the exclusive use rule: it parses the target device identifier (all ROBOT_01) and the execution time period of the three instructions, and after comparing them pairwise, it finds that the three instructions overlap in the time period of 18:00:15-18:00:30, and determines that there is a first-type resource conflict.

[0028] Since the first type of resource conflict was detected, the Harness scheduling engine did not perform the second type of conflict detection and directly entered the arbitration and dynamic programming adjustment stage: taking the highest priority instruction 3 (P0 level) as the base instruction, instruction 2 was shifted to the end of instruction 3 (18:02:15+4.0s=18:02:19), and instruction 1 was shifted to the end of instruction 2 (18:03:19+2.0s=18:03:21), generating the adjusted conflict-free timeline task sequence: [instruction 3 (18:00:15-18:02:15), instruction 2 (18:02:19-18:03:19), instruction 1 (18:03:21-18:03:51)], generating the final control instruction queue and issuing it for execution.

[0029] Scenario 2: No Type I resource conflict occurred, but a Type II resource conflict occurred (arbitration and dynamic programming path adjustment were initiated); the robot vacuum cleaner ROBOT_CLEAN has pre-assigned two non-overlapping tasks, as shown in Table 2: Table 2 Instructions Table 2

[0030] The Harness scheduling engine first performs the first type of conflict detection: it analyzes the execution time periods of instructions 4 and 5, and finds no overlap between 18:05:00 and 18:05:03, thus determining that no first type of resource conflict has been detected.

[0031] Since no Type I resource conflict was detected, the Harness scheduling engine triggered Type II conflict detection: The two instructions were sorted to obtain a timeline task sequence, and the actual interval between adjacent instructions was calculated as 18:05:03 - 18:05:00 = 3.0 seconds. A query of the mode switching time table showed that the shortest time for switching from dry sweeping to wet mopping was 5.0 seconds (requiring replacement of the mop module and water tank refill). Since 3.0 seconds < 5.0 seconds, a Type II resource conflict was determined to exist.

[0032] Entering the arbitration and dynamic programming adjustment phase: Although priority P2>P3, the two instructions do not overlap, so there is no need to change the execution order. Based on instruction 4 (the preceding instruction), the start time of instruction 5 is adjusted to 18:05:00+5.0s=18:05:05. The adjusted conflict-free timeline task sequence is generated: [Instruction 4 (18:00:00-18:05:00), Instruction 5 (18:05:05-18:10:05)], and the final control instruction queue is generated and issued for execution.

[0033] Scenario 3: No Type I conflict or Type II conflict occurred (passes directly) If the actual interval between instruction 4 and instruction 5 is 6.0 seconds, and the shortest time for dry sweeping to wet mopping is 5.0 seconds, then the first type of detection passes (no overlap), and the second type of detection also passes (6.0 ≥ 5.0). The final control instruction queue is then generated directly according to the original timeline and sent out for execution.

[0034] By detecting the first type of conflict and the second type of conflict, the task execution conflicts of the end-side actuator are detected. Based on the conflict situation, the time-axis task sequence is dynamically planned and adjusted to generate an adjusted conflict-free time-axis task sequence, thereby enabling the end-side actuator to execute multiple tasks smoothly in sequence and reducing the occurrence of task conflicts.

[0035] In some embodiments, each candidate instruction includes at least a target-side actuator identifier, an action type, and an action execution time period; The first type of conflict detection based on the exclusive use rule includes: Parse the target end-side actuator identifier and the action execution time period in each candidate instruction; The target end-side actuator identifier and the action execution time period in each of the parsed candidate instructions are compared pairwise. When it is detected that the same end-side actuator is assigned two or more different candidate instructions within the same action execution time period, it is determined that the first type of resource conflict exists.

[0036] In this exemplary embodiment, after the Harness scheduling engine obtains the set of instructions to be arbitrated, the specific steps for performing the first type of conflict detection are as follows: The first step is to parse the target end-side actuator identifier and action execution time period in each candidate instruction, as shown in Table 3: Table 3 Instruction Table 3

[0037] The second step is to compare the target end-side actuator identifier and the action execution time period in each of the parsed candidate instructions pairwise: Command 1 and Command 2: The same device ROBOT_01, the time period 18:00:00-18:00:30 overlaps (overlap duration 30 seconds); Instructions 2 and 3: The same device ROBOT_01 overlaps in time period 18:00:15-18:01:00 (overlap duration 45 seconds). Command 1 and Command 3: The same device ROBOT_01, the time period 18:00:15-18:00:30 overlaps (overlap duration 15 seconds).

[0038] The third step is to determine if two or more different candidate instructions are assigned to the same end-side actuator ROBOT_01 within the same action execution time period, thus determining that there is a type 1 resource conflict.

[0039] This embodiment achieves accurate identification of overlapping time windows of physical execution mechanisms in multi-agent concurrent scenarios by parsing the device identifier and time period in candidate instructions and performing precise pairwise comparisons. Compared with the existing technology that uniformly locks all resources indiscriminately, this solution provides an accurate conflict data foundation for subsequent arbitration through fine-grained analysis at the device level, avoiding device instruction conflicts caused by missed detections.

[0040] In some embodiments, the step of performing second-type conflict detection based on the shortest state transition time includes: The timeline task sequence is obtained by sorting the candidate instructions involving the same end-side execution agency in the set of instructions to be arbitrated according to their execution start time; Iterate through the two adjacent candidate instructions in the timeline task sequence in turn, and calculate the actual interval between the execution end time of the preceding candidate instruction and the execution start time of the adjacent following candidate instruction. Compare the actual interval duration with the corresponding shortest state transition time; When the actual interval duration is less than the minimum time required for state switching, a second type of resource conflict is determined to exist.

[0041] In this exemplary embodiment, the Harness scheduling engine performs the second type of conflict detection after failing to detect the first type of resource conflict, and the specific steps are as follows: The first step is to sort the candidate instructions involving the same end-side execution agency in the set of instructions to be arbitrated according to their execution start time, resulting in a timeline task sequence, as shown in Table 4: Table 4 Instructions Table 4

[0042] The second step is to iterate through the task sequence on the timeline and calculate the actual interval between the execution end time of the preceding candidate instruction and the execution start time of the adjacent following candidate instruction: Actual interval duration = Start time of instruction 5 18:05:03 - End time of instruction 4 18:05:00 = 3.0 seconds; The third step is to compare the actual interval duration with the corresponding shortest state transition time: Querying the ROBOT_CLEAN state transition time table in the constraint rule base, we found that the shortest state transition time from dry sweeping to wet mopping is 5.0 seconds.

[0043] If 3.0 seconds < 5.0 seconds, the physical handover constraint is not met.

[0044] The fourth step is to determine if a second type of resource conflict exists and record the rule number of the triggered state transition constraint rule: STATE_SWITCH_DRY_TO_WET.

[0045] Scenario Variation (Second Type Detection Passed): If the execution time of instruction 5 is 18:05:06-18:10:00, then the actual interval duration = 18:05:06-18:05:00 = 6.0 seconds ≥ 5.0 seconds, which is determined to be no second type resource conflict detected, and it passes directly.

[0046] This embodiment calculates the actual interval between adjacent candidate commands and compares it with the shortest physical state switching time, achieving accurate identification of implicit conflicts that do not overlap in time but whose physical switching is slower. Such conflicts are completely ignored in existing single-dimensional time window detection technologies—although two commands are queued on the timeline, switching from dry sweeping mode to wet mopping mode requires replacing the mop module and filling the water tank, a physical process that cannot be compressed. Ignoring this constraint and forcibly executing the command would cause the wet mopping module to start without proper preparation, resulting in motor idling or even damage.

[0047] In some embodiments, the arbitration process based on a predefined priority matrix and the dynamic programming adjustment of the timeline task sequence to generate an adjusted conflict-free timeline task sequence include: Based on the priority matrix, the priority levels of the multiple conflicting candidate instructions are obtained. The priority levels of the multiple candidate instructions are compared, and the candidate instruction with the highest priority level is determined as the baseline instruction. Using the execution time period of the baseline instruction as the anchor point, the execution time periods of the candidate instructions other than the baseline instruction in the time axis task sequence are sequentially shifted forward along the time axis, so that the actual interval between adjacent candidate instructions is greater than or equal to the shortest time consumption of the corresponding state transition, thereby generating the adjusted conflict-free time axis task sequence.

[0048] In this exemplary embodiment, the three candidate instructions in Table 1 (instruction 1 / P2, instruction 2 / P1, and instruction 3 / P0, all involving ROBOT_01, with overlapping time) are adopted.

[0049] The first step is to obtain the priority level of each of the multiple conflicting candidate instructions based on the priority matrix: Command 3: P0 (Emergency Event); Command 2: P1 (Security Event); Command 1: P2 (Comfort Control); The priority matrix is ​​set to the following order: emergency events (P0) > security events (P1) > comfort control (P2) > energy-saving scheduling (P3).

[0050] The second step is to compare the priority levels of multiple candidate instructions and determine the candidate instruction with the highest priority level as the baseline instruction: P0>P1>P2, therefore instruction 3 (P0 level, child protection) is the baseline instruction.

[0051] The third step is to use the execution time of the baseline instruction as the anchor point and shift the execution time of the candidate instructions (excluding the baseline instruction) in the timeline task sequence sequentially along the positive timeline: Baseline instruction 3 is fixed as: 18:00:15-18:02:15 (anchor point); Processing instruction 2 (P1, valve shut-off): Originally overlapping with instruction 3, it needs to be shifted until instruction 3 has finished. Instruction 3 end time 18:02:15 + valve shut-off → child safety switching time 4.0s = 18:02:19. Instruction 2 is adjusted to: 18:02:19-18:03:19; Processing instruction 1 (P2, Home Preparation): Originally overlapped with instructions 2 and 3, it needs to be further shifted until instruction 2 has finished. Instruction 2 end time 18:03:19 + valve closing → Home Preparation switching time 2.0s = 18:03:21. Instruction 1 is adjusted to: 18:03:21-18:03:51; The adjusted conflict-free timeline task sequence is generated as follows: [Instruction 3 (18:00:15-18:02:15), Instruction 2 (18:02:19-18:03:19), Instruction 1 (18:03:21-18:03:51)], with adjacent instruction intervals of 4.0s and 2.0s respectively, both satisfying the physical constraints.

[0052] Sub-example B: Arbitration and Adjustment Triggered by Type II Resource Conflicts Following Table 2 (Instruction 4 / P3 Dry Sweeping 18:00-18:05, Instruction 5 / P2 Wet Mopping 18:05:03-18:10:00, interval 3.0s < switchover time 5.0s). The arbitration and adjustment steps are as follows: The first step is to obtain the priority level of the conflicting commands: command 4 is P3 (energy saving scheduling), and command 5 is P2 (comfort control).

[0053] The second step is to determine the baseline instruction: P2 > P3, with instruction 5 having higher priority. However, since the two instructions do not overlap in time, the priority comparison is only used to determine the adjustment strategy—the higher-priority instruction should maintain its original time point as much as possible.

[0054] The third step is to shift the execution time of the baseline instruction as the anchor point: Since instruction 4 (low priority) comes first and instruction 5 (high priority) comes later, and instruction 5 has a higher priority, using the original start time of instruction 5 (18:05:03) as the anchor point, adjusting the end time of instruction 4 forward would disrupt its execution integrity. Therefore, instruction 5 is shifted backward: the new start time of instruction 5 = the end time of instruction 4 (18:05:00) + the switching time (5.0s) = 18:05:05. Instruction 5 is adjusted to: 18:05:05-18:10:05.

[0055] Generate the adjusted conflict-free timeline task sequence: [Instruction 4 (18:00:00-18:05:00), Instruction 5 (18:05:05-18:10:05)], with an adjacent interval of 5.0s, satisfying the physical constraints.

[0056] This embodiment employs a dynamic programming adjustment strategy that uses the highest priority instruction as the anchor point and shifts lower priority instructions forward along the time axis. All candidate instructions are retained and executed, with no tasks discarded, ensuring the complete preservation of user intent. For example, in Scenario 1, the P2-level "Prepare to Go Home" instruction, although postponed to 18:03:21, is still executed, and the air conditioner and water heater remain on after the user arrives home. Adjacent instructions are forced to satisfy physical state switching time constraints, preventing mechanical damage at its source. Adjustments triggered by the second type of conflict also follow the shifting strategy, demonstrating that this mechanism has a unified framework for handling both types of conflicts. If new time overlaps occur during the shifting process (such as a subsequent task for instruction 4), the system will automatically iterate the shift until there are no global conflicts, ensuring convergence.

[0057] In some embodiments, the constraint rule base includes a device type mapping table; wherein the device type mapping table marks end-side actuators as motion-coupled and electrical-decoupled types; Before performing the first type of conflict detection according to the exclusive use rule, the method includes: In response to obtaining the set of instructions to be arbitrated, the device type mapping table is queried based on the target end-side actuator identifier in each candidate instruction; When the target-side actuator is electrically decoupled, the first type of conflict detection and the second type of conflict detection are bypassed, and the candidate instruction is directly marked as a conflict-free instruction. When the target-side actuator is of the motion-coupled type, the first type of conflict detection and the second type of conflict detection are performed.

[0058] In this exemplary embodiment, the user returns home at 18:00, and the edge subsystem simultaneously receives the following six candidate instructions within 1 second, as shown in Table 5: Table 5 Instructions Table 5

[0059] After the Harness scheduling engine obtains the set of instructions to be arbitrated, before performing conflict detection, it first queries the device type mapping table based on the target end-side actuator identifier in each candidate instruction, as shown in Table 6: Table 6 Mapping Table

[0060] For "electrically decoupled" devices (commands C, D, E, F): Bypass both Type I and Type II conflict detection, directly mark the four candidate commands as conflict-free commands, and issue them all in parallel. Four electrical devices (two lights, a rice cooker, and a water heater) start simultaneously without interfering with each other.

[0061] For "motion-coupled" devices (commands A and B): Perform a complete Type I conflict detection. If command A and command B are detected to overlap during the time period of 18:00:30-18:02:00, an arbitration mechanism is triggered. Using command A (or arrival timestamp if priorities are the same) as the reference, command B is shifted until command A ends and the physical switching time is inserted, generating a serialized control command queue for the dual-arm robot.

[0062] This embodiment implements a differentiated conflict detection strategy through a device type mapping table. If existing technology were used to perform conflict detection on all devices equally, four electrical commands (two lights, a rice cooker, and a water heater) would be misjudged as "resource conflicts," causing some commands to be suspended. Users would then experience unpleasant experiences such as "delayed light activation" or "late rice cooker startup." This solution bypasses conflict detection for electrically decoupled devices, enabling electrical commands to be issued in parallel at millisecond levels. Simultaneously, it performs strict serial arbitration only on motion-coupled devices (robots) with physical safety risks, achieving a balance between system response efficiency and physical safety. Actual testing shows that in high-concurrency scenarios (more than 50 commands simultaneously), this differentiated detection mechanism can significantly reduce the average command response latency.

[0063] In some embodiments, before obtaining predefined declarative constraint rules from the constraint rule base, the method includes: Determine the shortest time consumption for state transitions between predetermined action type pairs for each end-side actuator; wherein, the calibration method for determining the shortest time consumption for state transitions between predetermined action type pairs for each end-side actuator includes: Each action type of the actuator is encoded, and each action type is represented by a node. Directed edges represent the transition from the first action type to the second action type, with the edge weight being the shortest time required for the state transition. An action type state transition graph is then constructed. The shortest state transition time is pre-acquired and stored in the constraint rule base using at least one of the following methods: The shortest time required for the state switching is determined based on the mechanical reset time, motor inertia buffer time, or compressor pressure balancing time recorded in the equipment manual of the end-side actuator. The shortest time for state switching is determined by taking the average value of multiple actual measurements on the end-side actuator. The shortest time required for state switching is monitored and dynamically updated in real time by sensors built into the end-side actuator.

[0064] In this exemplary embodiment, this embodiment describes the pre-calibration and storage process for the shortest state transition time. This process is completed during the system deployment phase and is a preparatory step for obtaining predefined declarative constraint rules from the constraint rule base.

[0065] Taking the dual-arm wheeled service robot ROBOT_MAIN as an example, the types of actions it supports include: home preparation (H), gas valve closing (G), food retrieval (T), food delivery (D), child protection (C), and mobile navigation (M).

[0066] The first step is to encode each action type of the actuator, as shown in Table 7: Table 7 Coding Table

[0067] The second step, based on the information in Table 8, uses nodes to represent action types, directed edges to represent the transition from the first action type to the second action type, and edge weights to represent the shortest state transition time, to construct an action type state transition graph: Table 8 Information Table

[0068] The value of the shortest state switching time mentioned above comes from three aspects: (1) the mechanical joint reset time and motor inertia buffer time recorded in the equipment manual; (2) the average value obtained by multiple actual measurements before leaving the factory; (3) after deployment, the motor temperature is monitored in real time by the joint angle sensor built into the actuator. When the motor temperature is detected to exceed the warning threshold, the weight of the corresponding side is automatically increased by 0.5s to achieve dynamic update.

[0069] After the system is deployed, the Harness scheduling engine reads the state transition graph directly from the constraint rule base when executing the arbitration method to obtain the shortest time for state switching between any action type pair.

[0070] This embodiment decouples physical characteristic parameters from arbitration logic code by constructing an action type state transition diagram and storing it in a constraint rule base using declarative rules. On the one hand, maintenance personnel can add, delete, or modify device action types or adjust switching time parameters without modifying the code, significantly improving the maintainability and scalability of the system. On the other hand, the dynamic update mechanism of edge weights enables physical parameters to adaptively adjust with factors such as equipment aging and changes in ambient temperature, ensuring that arbitration decisions are always based on accurate physical constraints and avoiding security risks caused by fixed parameters.

[0071] In some embodiments, generating the final control instruction queue based on the adjusted conflict-free timeline task sequence includes: Based on the adjusted conflict-free timeline task sequence, a new idle waiting period between two adjacent candidate instructions is determined, and an idle waiting delay instruction is automatically generated; wherein, the duration of the idle waiting delay instruction is greater than or equal to the shortest state switching time of the corresponding adjacent candidate instruction; The candidate instructions and the idle wait delay instructions are arranged in chronological order to form a serialized final control instruction queue.

[0072] In this exemplary embodiment, the conflict-free timeline task sequence adjusted from Table 1 is adopted, as shown in Table 9 below: Table 9 Instructions Table 9

[0073] The Harness scheduling engine generates the final control instruction queue based on the adjusted conflict-free timeline task sequence as follows: The first step is to determine the newly generated idle waiting period between two adjacent candidate instructions based on the adjusted conflict-free timeline task sequence: Instruction 3 end time: 18:02:15, Instruction 2 start time: 18:02:19; Idle waiting period = 18:02:15 to 18:02:19, duration = 4.0 seconds; The mechanism behind this idle waiting period is as follows: Instruction 2 is shifted from its original position at 18:00:00 to the end of Instruction 3 (18:02:15). The shift amount is determined by two constraints: eliminating time overlap and satisfying the minimum time of 4.0 seconds for switching from valve closure to child safety protection. Here, 4.0 seconds is a natural mathematical gap between the end time of Instruction 3 and the start time of Instruction 2 after the shift operation, rather than a fixed safety margin preset by the user.

[0074] Instruction 2 end time: 18:03:19, Instruction 1 start time: 18:03:21; Idle waiting period = 18:03:19 to 18:03:21, duration = 2.0 seconds; The mechanism for generating this idle waiting period is as follows: Instruction 1 is further shifted from its original position of 18:00:00 to the end of Instruction 2 (18:03:19). The shift amount is determined by eliminating the overlap with Instruction 2 and satisfying the minimum time of 2.0 seconds for switching from valve closing to home preparation.

[0075] The second step is to automatically generate an idle wait delay instruction: Waiting for delay instruction 1: Duration 4.0 seconds (inserted between instruction 3 and instruction 2) Waiting for delay instruction 2: duration 2.0 seconds (inserted between instruction 2 and instruction 1) The third step is to arrange the candidate instructions and idle wait delay instructions in chronological order to form a serialized final control instruction queue, as shown in Table 10 below: Table 10 Instruction Table 10

[0076] The Harness scheduling engine sends the serialized final control command queue sequentially to the dual-arm robot ROBOT_01. After executing command 3, the robot automatically waits for 4.0 seconds (the robotic arm resets from the children's room door lock posture to the valve-closed posture) before executing command 2; after executing command 2, it automatically waits for 2.0 seconds (the robotic arm resets from the valve-closed posture to the standby position) before executing command 1.

[0077] This embodiment achieves automated enforcement of physical constraints by automatically generating idle waiting delay instructions based on the adjusted conflict-free timeline task sequence. As shown in the embodiment, the 4.0-second and 2.0-second idle periods are not fixed safety margins reserved by the scheduler actively cutting the timeline, but rather numerical differences naturally formed between the end and start times of adjacent tasks during the process of shifting conflicting instructions backward to meet the physical switching time constraints, using the baseline instruction as an anchor point. This property proves that the arbitration and adjustment logic itself inherently satisfies the physical constraints, and the idle waiting instruction merely explicitly manifests this inherent constraint as an executable delay instruction.

[0078] By arranging the action instructions and idle waiting delay instructions in chronological order, the edge execution mechanism only needs to execute them sequentially according to the queue order. There is no need to calculate physical constraints or arbitration logic online, which reduces the computational burden of the edge execution mechanism and is especially suitable for edge devices with limited computing power.

[0079] In some embodiments, the dynamic planning and adjustment of the timeline task sequence further includes: When the translation causes a new time overlap between the execution time of any candidate instruction and the third candidate instruction, the translation operation is iteratively executed until the actual interval between all adjacent candidate instructions in the timeline task sequence is not less than the corresponding shortest state switching time. Record the candidate instruction identifier, translation amount, and translation reason involved in each translation operation. The translation reason includes the rule number of the declarative constraint rule that was triggered.

[0080] In some embodiments, the arbitration process further includes an award step of equal priority: When multiple conflicting candidate instructions have the same priority level, compare the arrival timestamps of each conflicting candidate instruction; The candidate instruction with the earliest arrival timestamp is retained as the reference instruction; When the arrival timestamps are the same, the baseline instruction is determined according to the preset arbitration priority order of each agent in the multi-agent cluster and the cloud subsystem.

[0081] In some embodiments, an arbitration tracing data encapsulation step is also included: When determining the existence of a Type I or Type II resource conflict, extract the unique rule number of the declarative constraint rule triggered by the conflict; The unique rule number, the identifier of the conflicting end-side execution agency, the identifier and priority level of the conflict candidate instruction, and the time axis shift used are encapsulated as arbitration metadata. When returning a task suspension notification to the source sender of the candidate instruction, the arbitration metadata is also returned.

[0082] This invention provides a multi-source control command arbitration device applied to the edge subsystem of a human-vehicle-home control system. The human-vehicle-home control system includes a terminal subsystem, an edge subsystem, and a cloud subsystem. The terminal subsystem is deployed in home and vehicle scenarios and includes various terminal actuators. The edge subsystem is deployed in the home main control edge computing center, and the cloud subsystem is deployed on a cloud server. Figure 2 This is a schematic diagram of a multi-source control command arbitration device provided in an embodiment of the present invention. Figure 2 As shown, the device includes: The instruction acquisition unit 20 is used to acquire a set of instructions to be arbitrated; wherein the set of instructions to be arbitrated contains at least two candidate instructions, wherein each candidate instruction is generated by an agent in the multi-Agent cluster of the edge subsystem; Rule acquisition unit 21 is used to acquire predefined declarative constraint rules from the constraint rule library; wherein, the declarative constraint rules include at least: exclusive use rules for each end-side actuator and the shortest state transition time for each end-side actuator between predetermined action type pairs; wherein, the constraint rule library is stored locally on the edge-side subsystem, the exclusive use rules define that each end-side actuator can only be assigned to one control task at a time; the shortest state transition time is pre-calibrated based on the physical and mechanical characteristics of the end-side actuator; The conflict detection unit 22 is used to perform a first type of conflict detection according to the exclusive use rule, and when no first type of resource conflict is detected, to perform a second type of conflict detection according to the shortest state switching time. The sequence generation unit 23 is used to perform arbitration processing according to a predefined priority matrix when the first type of conflict detection or the second type of resource conflict is detected, and to dynamically plan and adjust the time axis task sequence to generate an adjusted conflict-free time axis task sequence; wherein, the time axis task sequence is obtained by sorting the candidate instructions of the same end-side execution mechanism according to the execution start time; The instruction issuing unit 24 is used to generate a final control instruction queue based on the adjusted conflict-free timeline task sequence, and to issue the final control instruction queue to the corresponding end-side execution mechanism through the edge-side subsystem.

[0083] A priority matrix is ​​a multi-level task priority mapping table used to assign specific priority levels to different types of events. The priority matrix is ​​preset according to the following order: emergency events (P0) have a higher priority than security events (P1), security events have a higher priority than comfort control events (P2), and comfort control events have a higher priority than energy-saving scheduling events (P3). Furthermore, for the same type of event, different priorities are assigned to different objects. For example, in emergency events, the priority for rescuing the elderly is higher than that for rescuing children, and the priority for rescuing children is higher than that for ordinary adults. Similarly, security event priorities include the priority of front door security being higher than that of interior security.

[0084] Edge-side actuators refer to specific hardware devices deployed in the edge-side subsystem that receive control commands and execute physical actions, including but not limited to: dual-arm wheeled service robots, smart air conditioners, smart refrigerators, rice cookers, automatic cooking machines, washing machines, smart switches / sockets, vehicle air conditioning compressors, window motors, steering wheel vibration motors, etc. Comfort control events have higher priority than energy-saving scheduling events; for example, in a vehicle scenario, controlling seat adjustments has higher priority than controlling the vehicle's air conditioning. The home main control edge computing center can be networked with the vehicle controller.

[0085] The edge subsystem is deployed at the home-based edge computing center, running an operating system with real-time RT-Linux patches, and featuring a multi-agent cluster and the Harness scheduling engine. The cloud subsystem is deployed on a cloud server, containing a large language model and a global knowledge graph. Task execution records from each edge actuator can be uploaded to cloud storage.

[0086] By detecting the first type of conflict and the second type of conflict, the task execution conflicts of the end-side actuator are detected. Based on the conflict situation, the time-axis task sequence is dynamically planned and adjusted to generate an adjusted conflict-free time-axis task sequence, thereby enabling the end-side actuator to execute multiple tasks smoothly in sequence and reducing the occurrence of task conflicts.

[0087] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0088] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the methods described in the above embodiments.

[0089] This invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the methods described in the above embodiments.

[0090] 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.

[0091] 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, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0092] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0093] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0094] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0095] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A multi-source control command arbitration method, characterized in that, An edge-side subsystem is applied in a human-vehicle-home control system, the human-vehicle-home control system including a terminal subsystem, an edge-side subsystem, and a cloud subsystem; the terminal subsystem is deployed in home and vehicle scenarios, and the terminal subsystem includes various terminal actuators; the edge-side subsystem is deployed in a home main control edge computing center; and the cloud subsystem is deployed on a cloud server; the method includes: Obtain a set of instructions to be arbitrated; wherein the set of instructions to be arbitrated contains at least two candidate instructions, and each candidate instruction is generated by an agent in the multi-Agent cluster of the edge subsystem; Obtain predefined declarative constraint rules from the constraint rule library; wherein, the declarative constraint rules include at least: exclusive use rules for each end-side actuator and the shortest state switching time for each end-side actuator between predetermined action type pairs; The first type of conflict is detected according to the exclusive use rule, and the second type of conflict is detected according to the shortest state switching time when no first type of resource conflict is detected. When the first type of conflict detection or the second type of resource conflict is detected, arbitration is performed according to a predefined priority matrix, and the timeline task sequence is dynamically planned and adjusted to generate an adjusted conflict-free timeline task sequence; wherein, the timeline task sequence is obtained by sorting the candidate instructions of the same end-side execution mechanism according to the execution start time; The final control command queue is generated based on the adjusted conflict-free timeline task sequence, and the final control command queue is sent to the corresponding end-side execution mechanism through the edge-side subsystem.

2. The multi-source control command arbitration method according to claim 1, characterized in that, Each candidate instruction includes at least the target-side actuator identifier and the action execution time period; The first type of conflict detection based on the exclusive use rule includes: Parse the target end-side actuator identifier and the action execution time period in each candidate instruction; The target end-side actuator identifier and the action execution time period in each of the parsed candidate instructions are compared pairwise. When it is detected that the same end-side actuator is assigned two or more different candidate instructions within the same action execution time period, it is determined that the first type of resource conflict exists.

3. The multi-source control command arbitration method according to claim 1 or 2, characterized in that, The second type of conflict detection based on the shortest state transition time includes: The timeline task sequence is obtained by sorting the candidate instructions involving the same end-side execution agency in the set of instructions to be arbitrated according to their execution start time; Iterate through the two adjacent candidate instructions in the timeline task sequence in turn, and calculate the actual interval between the execution end time of the preceding candidate instruction and the execution start time of the adjacent following candidate instruction. Compare the actual interval duration with the corresponding shortest state transition time; When the actual interval duration is less than the minimum time required for state switching, a second type of resource conflict is determined to exist.

4. The multi-source control command arbitration method according to claim 3, characterized in that, The step of arbitrating according to a predefined priority matrix and dynamically adjusting the timeline task sequence to generate an adjusted conflict-free timeline task sequence includes: Based on the priority matrix, the priority levels of the multiple conflicting candidate instructions are obtained. The priority levels of the multiple candidate instructions are compared, and the candidate instruction with the highest priority level is determined as the baseline instruction. Using the execution time period of the baseline instruction as the anchor point, the execution time periods of the candidate instructions other than the baseline instruction in the time axis task sequence are sequentially shifted forward along the time axis, so that the actual interval between adjacent candidate instructions is greater than or equal to the shortest time consumption of the corresponding state transition, thereby generating the adjusted conflict-free time axis task sequence.

5. The multi-source control command arbitration method according to claim 3, characterized in that, The constraint rule base includes a device type mapping table; wherein, the device type mapping table marks the end-side actuator as motion-coupled type and electrical-decoupled type; Before performing the first type of conflict detection according to the exclusive use rule, the method includes: In response to obtaining the set of instructions to be arbitrated, the device type mapping table is queried based on the target end-side actuator identifier in each candidate instruction; When the target-side actuator is electrically decoupled, the first type of conflict detection and the second type of conflict detection are bypassed, and the candidate instruction is directly marked as a conflict-free instruction. When the target-side actuator is of the motion-coupled type, the first type of conflict detection and the second type of conflict detection are performed.

6. The multi-source control command arbitration method according to claim 1, characterized in that, Before obtaining predefined declarative constraint rules from the constraint rule base, the method includes: Determine the shortest time consumption for state transitions between predetermined action type pairs for each end-side actuator; wherein, the calibration method for determining the shortest time consumption for state transitions between predetermined action type pairs for each end-side actuator includes: Each action type of the actuator is encoded, and the action type is represented by a node. A directed edge represents the switch from the first action type to the second action type. The edge weight is the shortest time consumption of the state switch, and an action type state transition graph is constructed.

7. The multi-source control command arbitration method according to claim 1, characterized in that, The step of generating the final control instruction queue based on the adjusted conflict-free timeline task sequence includes: Based on the adjusted conflict-free timeline task sequence, a new idle waiting period between two adjacent candidate instructions is determined, and an idle waiting delay instruction is automatically generated; wherein, the duration of the idle waiting delay instruction is greater than or equal to the shortest state switching time of the corresponding adjacent candidate instruction; The candidate instructions and the idle wait delay instructions are arranged in chronological order to form a serialized final control instruction queue.

8. A multi-source control command arbitration device, characterized in that, An edge-side subsystem is applied in a human-vehicle-home control system, which includes a terminal subsystem, an edge-side subsystem, and a cloud subsystem. The terminal subsystem is deployed in home and vehicle scenarios and includes various terminal actuators. The edge-side subsystem is deployed at a home main control edge computing center, and the cloud subsystem is deployed on a cloud server. The device includes: The instruction acquisition unit is used to acquire a set of instructions to be arbitrated; wherein the set of instructions to be arbitrated contains at least two candidate instructions, and each candidate instruction is generated by an agent in the multi-Agent cluster of the edge subsystem; The rule acquisition unit is used to acquire predefined declarative constraint rules from the constraint rule library; wherein, the declarative constraint rules include at least: exclusive use rules for each end-side actuator and the shortest time consumption for state switching between predetermined action type pairs of each end-side actuator; The conflict detection unit is used to perform a first type of conflict detection according to the exclusive use rule, and to perform a second type of conflict detection according to the shortest state switching time when no first type of resource conflict is detected. The sequence generation unit is used to perform arbitration processing according to a predefined priority matrix when the first type of conflict detection or the second type of resource conflict is detected, and to dynamically plan and adjust the time axis task sequence to generate an adjusted conflict-free time axis task sequence; wherein, the time axis task sequence is obtained by sorting the candidate instructions of the same end-side execution mechanism according to the execution start time; The instruction issuing unit is used to generate a final control instruction queue based on the adjusted conflict-free timeline task sequence, and to issue the final control instruction queue to the corresponding end-side execution mechanism through the edge-side subsystem.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 7.

Images