A Smart Robot Scheduling System and Method Based on High-Speed Chip Computing
By constructing a home status time series set and budget curve through a home dispatch controller, and dynamically adjusting dispatch instructions, the problems of inaccurate dispatching and frequent start-stop of equipment in home automation control are solved. Effective constraints on power and thermal load are achieved, improving the accuracy of dispatching and the stability of continuous service.
Patent Information
- Application Number
- CN202610982997.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-02
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies lack a unified time reference for multi-source state data in home automation control, leading to inaccurate scheduling judgments, excessively high power peaks and chip heat accumulation. Task sequencing is difficult to adapt to continuous service scenarios, and changes in state can easily cause frequent task rewriting and scheduling oscillations.
The home status data is collected in real time by the home scheduling controller, a home status time series set is constructed, a home power budget curve and a chip thermal budget curve are generated, service availability level classification and timing arrangement are performed, a resource reservation table is formed, and scheduling instructions are dynamically adjusted in combination with execution commitment window and differential recalculation boundary to keep the total home power and chip thermal load within the budget curve.
It achieves the simultaneous consideration of total power and chip thermal load constraints in home continuous service scenarios, reduces excessive power peaks and chip heat accumulation, improves the accuracy of scheduling judgment and the stability of continuous service, and reduces task rewriting and frequent device start-up and shutdown.
Smart Images

Figure CN123067358A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of smart home technology, specifically to a smart robot scheduling system and method based on high-speed chip computing. Background Technology
[0002] With the development of smart home interconnection and control, edge computing chips, and home service robots, home automation control has gradually evolved from single-device timed execution to a comprehensive scheduling mode of multi-device, multi-task collaboration. Existing technologies typically connect service robots, air conditioners, lighting, kitchen appliances, and cleaning equipment through a home gateway, and combine status information such as power, temperature, space occupancy, and task triggering to achieve continuous service functions such as welcoming guests upon arrival, assisting with dinner, and post-dinner cleanup.
[0003] Currently, the lack of a unified time benchmark for multi-source state data leads to inaccurate scheduling judgments. The scheduling process usually only considers a single factor such as household power or task priority, lacking joint constraints on total household power and the thermal load of the main control chip, which can easily result in excessively high power peaks and chip heat accumulation. The lack of unified modeling for the pre-dependencies and continuous service relationships between tasks makes it difficult for task sorting to adapt to continuous service scenarios. After state changes, global reordering is often performed directly, which can easily cause frequently rewritten tasks and scheduling oscillations. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides an intelligent robot scheduling system and method based on high-speed chip computing, which solves the problems mentioned in the background technology.
[0005] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps:
[0006] S1, collect home status data of the smart robot in real time through the home scheduling controller. The home status data includes power distribution circuit power data, smart robot main control chip temperature data, smart robot battery status data, home device operation status data, home space occupancy status data, and service task data to be executed; construct a home status time series set based on the home status data, and determine a service task set to be executed based on the service task data to be executed.
[0007] S2, based on the household status time series set and the set of service tasks to be executed, determine the household power budget curve and the chip thermal budget curve; according to the household power budget curve and the chip thermal budget curve, classify the set of service tasks to be executed into service retention levels to obtain service retention levels;
[0008] S3, based on the service retention level, the set of service tasks to be executed is time-series orchestrated to obtain a service intent sequence; according to the service intent sequence, reservation time slots are allocated for intelligent robot motion resources and home device control resources on a unified time axis to form a resource reservation table; then, combined with the home power budget curve and the chip thermal budget curve, the resource reservation table is constrained and checked to determine the execution commitment window and differential recalculation boundary;
[0009] S4. Based on the execution commitment window, the differential recalculation boundary, and the resource reservation table, issue intelligent robot scheduling instructions and home device control instructions; when the home status data changes, only adjust the service intent sequence and resource reservation table corresponding to the differential recalculation boundary, keep the corresponding service in the execution commitment window unchanged, so that the total home power and the thermal load of the intelligent robot main control chip are limited by the home power budget curve and the chip thermal budget curve, respectively.
[0010] Preferably, in S1, the home dispatch controller collects power data of the power distribution circuits in the smart home, temperature data of the smart robot's main control chip, battery status data of the smart robot, operating status data of home devices, occupancy status data of home space, and data of service tasks to be executed, according to a unified sampling clock.
[0011] The collected data is time-aligned to obtain subsets of power distribution branch power status, chip thermal status, space occupancy status, and task status, and these subsets are then aggregated to form the household status time series set.
[0012] The set of service tasks to be executed is determined from the service tasks to be executed that are in the pending scheduling state and correspond to the same family continuous service scenario and the same service target identifier. Based on the execution order constraints, resource call constraints and result trigger constraints of each service task to be executed, the task pre-dependencies between the service tasks to be executed are determined.
[0013] Preferably, in S2, the home scheduling controller determines the allowable range of home power and the allowable range of chip thermal load for each time slice according to the home status time sequence set and the same preset time window.
[0014] Then, based on the preset mapping relationship between chip thermal load and equivalent power, the allowable range of chip thermal load for each time slice is converted into the equivalent power constraint range.
[0015] Intersection of the allowable power range and the equivalent power constraint range for each time slice is performed to obtain the coupling constraint range for each time slice.
[0016] Preferably, in S2, a continuous service chain is determined based on the task prerequisite dependencies in the set of service tasks to be executed; the continuous service chain consists of multiple service tasks to be executed, wherein the multiple service tasks to be executed correspond to the same service target identifier and have task prerequisite dependencies on each other;
[0017] Then, calculate the estimated energy consumption and incremental chip thermal load for each service task to be executed, and combine them with the interruption tolerance index of the corresponding continuous service chain, and calculate the weighted evaluation value according to the preset weights.
[0018] Based on the comparison results of the weighted evaluation value with the first division threshold and the second division threshold, as well as the matching of the execution energy consumption estimate and the chip thermal load increment estimate with the coupling constraint interval, the service retention level is divided for each service task to be executed, and the service retention level is obtained.
[0019] The service retention levels include retained service, downgraded service, and deferred service;
[0020] Wherein, the first division threshold is greater than the second division threshold;
[0021] When the weighted evaluation value is greater than or equal to the first division threshold, the corresponding service task to be executed will be classified as a reserved service;
[0022] When the weighted evaluation value is less than the first division threshold and greater than or equal to the second division threshold, the control parameter item corresponding to the service task to be executed is extracted. The control parameter item includes at least one of the following: intelligent robot movement speed, intelligent robot movement acceleration, home device operating power, and task execution time.
[0023] The control parameter items are then compressed according to a preset compression rule to obtain a compressed set of control parameters.
[0024] Based on the set of compressed control parameters, the estimated energy consumption after compression and the estimated incremental thermal load of the chip after compression are recalculated.
[0025] When both the compressed execution energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range of the corresponding time period, and the compressed task execution time does not exceed the maximum allowable delay corresponding to the continuous service chain, the corresponding service task to be executed will be classified as a degraded service; otherwise, the corresponding service task to be executed will be classified as a delayed service.
[0026] The maximum allowable delay for a continuous service chain is determined by the home scheduling controller based on the task prerequisites, service target completion time limits, and home device response time limits in the continuous service chain.
[0027] When the weighted evaluation value is less than the second division threshold, the corresponding service task to be executed will be classified as a delayed service.
[0028] Preferably, in S3, the set of service tasks to be executed is time-series arranged according to the service retention level and the continuous service chain to obtain the service intent sequence;
[0029] When generating the service intent sequence, the task prerequisites in the continuous service chain are kept unchanged, and reserved services are arranged first, followed by degraded services, and finally delayed services.
[0030] When the priority order corresponding to the service retention level conflicts with the task's prerequisite dependency, the home scheduling controller will only adjust the order of pending service tasks that do not have a direct prerequisite dependency, while keeping the task's prerequisite dependency unchanged.
[0031] Preferably, in S3, based on the service intent sequence, reservation time slots are allocated for intelligent robot motion resources and home device control resources on a unified time axis to form the resource reservation table;
[0032] The home scheduling controller establishes a time slot locking relationship between corresponding scheduled time slots for tasks that have a pre-dependent relationship in the same continuous service chain.
[0033] The time slot locking relationship is used to limit the minimum and maximum intervals between consecutive and related reservation time slots;
[0034] When multiple appointment time slots conflict, the home scheduling controller adjusts the conflicting appointment time slots in the order of reserved service, degraded service, and delayed service.
[0035] Preferably, in S3, the home scheduling controller filters reservation time slots based on the resource reservation table and the coupling constraint interval; the filtered reservation time slots meet the following conditions:
[0036] The reserved time period corresponds to a reserved service, or a degraded service that, after being compressed by control parameters, the compressed energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range.
[0037] The selected appointment time slots are designated as the execution commitment window, and the remaining appointment time slots are designated as the differential recalculation boundary.
[0038] The execution commitment window is used to keep the start time and resource usage relationship of the corresponding reservation period unchanged.
[0039] Preferably, in S4, when the household status data changes, the household scheduling controller updates the execution energy consumption estimate and the chip thermal load increment estimate corresponding to the affected reservation period, and compares the updated execution energy consumption estimate and the chip thermal load increment estimate with the coupling constraint interval of the corresponding time slice.
[0040] When there is a reservation period exceeding the coupling constraint interval within the execution commitment window, the home scheduling controller adjusts the resource consumption and power allocation within the corresponding time slice by migrating the reservation period corresponding to the delayed service within the differential recalculation boundary and switching the degraded service within the differential recalculation boundary to the execution state after control parameter compression, so as to release the coupling constraint interval margin of the corresponding time slice.
[0041] When there is a reservation period outside the coupling constraint interval within the differential recalculation boundary, the home scheduling controller performs local recalculation on the service intent sequence corresponding to the affected part and the resource reservation table;
[0042] The local recalculation maintains the start time, execution order, and resource usage relationship of the scheduled time slots within the execution commitment window unchanged.
[0043] Preferably, in S4, the home scheduling controller updates the service retention level, service intent sequence, and resource reservation table according to the result of the local recalculation, and reissues the intelligent robot scheduling instructions and home device control instructions according to the updated service retention level, service intent sequence, and resource reservation table;
[0044] When the updated reservation period still exceeds the coupling constraint interval of the corresponding time slice, the home scheduling controller will migrate the corresponding delayed service to a time slice with a coupling constraint interval margin, or switch the corresponding degraded service to the execution state after the control parameters are compressed.
[0045] If, after migration and switching, there is still no executable time slice that satisfies the coupling constraint interval, the home scheduling controller suspends delayed services that have no task prerequisites with the current execution commitment window, in order to maintain the continuous execution of the corresponding services within the execution commitment window.
[0046] A smart robot scheduling system based on high-speed chip computing includes: a home status acquisition and construction module, a budget evaluation and grading module, an intent orchestration and reservation module, and an execution control recalculation module;
[0047] The home status acquisition and construction module collects home status data of the intelligent robot in real time through the home scheduling controller. The home status data includes power distribution circuit power data, intelligent robot main control chip temperature data, intelligent robot battery status data, home device operation status data, home space occupancy status data, and pending service task data. Based on the home status data, a home status time series set is constructed, and based on the pending service task data, a pending service task set is determined.
[0048] The budget assessment and grading module determines the household power budget curve and the chip thermal budget curve based on the household status time series set and the set of service tasks to be executed; and classifies the set of service tasks to be executed into service retention levels according to the household power budget curve and the chip thermal budget curve to obtain the service retention level.
[0049] The intent orchestration and reservation module orchestrates the set of service tasks to be executed in a time sequence based on the service retention level to obtain a service intent sequence; it allocates reservation time slots for intelligent robot motion resources and home device control resources on a unified time axis according to the service intent sequence to form a resource reservation table; and then, in combination with the home power budget curve and the chip thermal budget curve, it performs constraint verification on the resource reservation table to determine the execution commitment window and differential recalculation boundary.
[0050] The execution control recalculation module issues intelligent robot scheduling instructions and home device control instructions based on the execution commitment window, the differential recalculation boundary, and the resource reservation table. When the home status data changes, it only adjusts the corresponding service intent sequence and resource reservation table within the differential recalculation boundary, keeping the corresponding services within the execution commitment window unchanged, so that the total home power and the thermal load of the intelligent robot main control chip are limited by the home power budget curve and the chip thermal budget curve, respectively.
[0051] This invention provides an intelligent robot scheduling system and method based on high-speed chip computing. It has the following beneficial effects:
[0052] (1) This method constructs a family state time series set, and generates a family power budget curve and a chip thermal budget curve on this basis. Then, it combines the continuous service chain and service maintenance level to perform layered processing of the service tasks to be executed. In this way, it can simultaneously take into account the total power constraint of the family and the thermal load constraint of the main control chip in the family continuous service scenario. Compared with the existing scheduling method that only schedules according to task priority, single power condition or device status, this solution can reduce the problems of high power peak, chip thermal accumulation and task arrangement imbalance in the evening peak scenario, and improve the accuracy of scheduling judgment and the stability of continuous service.
[0053] (2) This method, through the coordination of service intent sequences, resource reservation tables, execution commitment windows, and differential recalculation boundaries, distinguishes between the determined key reservation periods and the adjustable reservation periods. When the household status data changes, only the service intent sequences and resource reservation tables within the differential recalculation boundaries are locally adjusted, while keeping the corresponding services within the execution commitment window unchanged. This reduces the frequent rewriting of issued tasks, repeated device start-ups and shutdowns, and the spread of local disturbances into global reordering, thereby improving the controllability of the execution phase and the stability of system operation.
[0054] (3) This method combines the service retention level classification mechanism under dual budget constraints with the execution commitment window and differential recalculation boundary mechanism. It can not only reasonably distinguish between retained services, degraded services and delayed services before scheduling, but also make a boundary-based dynamic response to state changes during execution. This ensures that the critical continuous service chain is prioritized and that the adjustable task can complete the migration, degrade or delay execution without destroying the core service, thereby improving the scheduling continuity, resource utilization efficiency and anti-disturbance capability in the home continuous service scenario. Attached Figure Description
[0055] Figure 1 This is a flowchart of the method of the present invention;
[0056] Figure 2 This is a schematic diagram of the modules of the present invention;
[0057] Figure 3 This is a diagram showing the core content relationships of the present invention;
[0058] Figure 4 The service level classification logic diagram of this invention is maintained;
[0059] Figure 5 This is a diagram of the execution commitment window and differential recalculation boundary mechanism of the present invention. Detailed Implementation
[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0061] Example 1
[0062] Please see Figures 1-5 This invention provides an intelligent robot scheduling method based on high-speed chip computing. To achieve the above objectives, this invention is implemented through the following technical solution, including the following steps:
[0063] S1, collect home status data of the smart robot in real time through the home scheduling controller. The home status data includes power distribution circuit power data, smart robot main control chip temperature data, smart robot battery status data, home device operation status data, home space occupancy status data, and service task data to be executed; construct a home status time series set based on the home status data, and determine a service task set to be executed based on the service task data to be executed.
[0064] S2, based on the household status time series set and the set of service tasks to be executed, determine the household power budget curve and the chip thermal budget curve; according to the household power budget curve and the chip thermal budget curve, classify the set of service tasks to be executed into service retention levels to obtain service retention levels;
[0065] S3, based on the service retention level, the set of service tasks to be executed is time-series orchestrated to obtain a service intent sequence; according to the service intent sequence, reservation time slots are allocated for intelligent robot motion resources and home device control resources on a unified time axis to form a resource reservation table; then, combined with the home power budget curve and the chip thermal budget curve, the resource reservation table is constrained and checked to determine the execution commitment window and differential recalculation boundary;
[0066] S4. Based on the execution commitment window, the differential recalculation boundary, and the resource reservation table, issue intelligent robot scheduling instructions and home device control instructions; when the home status data changes, only adjust the service intent sequence and resource reservation table corresponding to the differential recalculation boundary, keep the corresponding service in the execution commitment window unchanged, so that the total home power and the thermal load of the intelligent robot main control chip are limited by the home power budget curve and the chip thermal budget curve, respectively.
[0067] In this embodiment, the application is applied to a continuous home service scenario during the evening peak hours. The application scenario is set as a three-bedroom house, where family members enter the house in a concentrated manner between 18:30 and 20:00 and trigger multiple service tasks at the same time.
[0068] The residence will be equipped with one service robot, one sweeping and mopping robot, one air conditioner, one fresh air system, one electric water heater, one dishwashing machine, several lighting fixtures, and one set of kitchen heating equipment.
[0069] The home scheduling controller is deployed in the edge computing host, which communicates with the service-type intelligent robot, the sweeping and mopping intelligent robot, and various home devices through the home local area network.
[0070] The unified sampling clock is set to 1 second, the scheduling time slice is set to 1 minute, and the rolling scheduling time domain is set to 15 minutes.
[0071] At 18:35, the home scheduling controller begins executing the scheduling method.
[0072] The home dispatch controller collects home status data in real time, including power distribution circuit power data, smart robot main control chip temperature data, smart robot battery status data, home device operating status data, home space occupancy status data, and pending service tasks data.
[0073] Power distribution branch power data is collected by power distribution monitoring nodes; temperature data of the intelligent robot main control chip and battery status data of the intelligent robot are collected by the monitoring units of the service-type intelligent robot and the sweeping and mopping-type intelligent robot; operating status data of home devices are fed back by the device control interface; occupancy status data of home space is jointly collected by door magnets, millimeter-wave presence sensors and indoor positioning nodes; and data of service tasks to be executed are jointly generated by preset home service strategies and current triggering events.
[0074] The home scheduling controller aligns home status data by time, constructs a home status time series set, and determines the set of service tasks to be executed from the service task data to be executed.
[0075] In this embodiment, the set of service tasks to be executed includes at least the following tasks: welcoming mobile task, entrance hall lighting turn-on task, living room air conditioning switching task, kitchen heating maintenance task, tableware transfer task, dishwashing appointment task, partial cleaning task, and hot water compensation task.
[0076] The home scheduling controller generates a home power budget curve and a chip thermal budget curve for the next 15 minutes based on the home status time sequence set and the set of service tasks to be executed.
[0077] The household power budget curve is obtained based on the rated upper limit of the residential power distribution circuit, the basic load, and the safety redundancy. The chip thermal budget curve is obtained based on the current temperature of the intelligent robot's main control chip, recent load changes, and the preset temperature rise boundary.
[0078] During the period from 18:35 to 18:50, the average available power margin given by the household power budget curve is set to 2.4kW, and the available thermal load margin of the main control chip corresponding to the chip thermal budget curve is converted to an equivalent power margin of 0.9kW.
[0079] The home dispatch controller classifies the service retention levels of the set of service tasks to be executed based on the home power budget curve and the chip thermal budget curve, thus obtaining the service retention levels.
[0080] The tasks of welcoming guests, turning on the entrance lighting, and switching on the living room air conditioner are classified as reserved services, while the task of partial cleaning is classified as downgraded services, and the tasks of scheduling dishwashing and replenishing hot water are classified as delayed services.
[0081] After obtaining the service hold level, the home dispatch controller orchestrates the set of service tasks to be executed in a time sequence to form a service intent sequence.
[0082] In this embodiment, the service intent sequence is organized in the order of "turning on the entrance lighting, welcoming guests, switching the living room air conditioning, maintaining the kitchen heating, transferring tableware, partial cleaning, scheduling dishwashing, and then replenishing hot water".
[0083] Subsequently, the home scheduling controller allocates reservation time slots for intelligent robot motion resources and home device control resources on a unified time axis according to the service intent sequence, forming a resource reservation table.
[0084] In the resource reservation schedule, 18:36 to 18:39 corresponds to the joint operation of welcoming guests and entrance lighting; 18:39 to 18:45 corresponds to the joint operation of living room air conditioning switching and kitchen heating maintenance; 18:45 to 18:52 corresponds to the tableware transfer and partial cleaning period; and after 18:52 corresponds to the dishwashing reservation and hot water replenishment period.
[0085] The home scheduling controller then combines the home power budget curve and the chip thermal budget curve to perform constraint verification on the resource reservation table and determine the execution commitment window and differential recalculation boundary.
[0086] The execution commitment window is set for the period from 18:36 to 18:45, and the differential recalculation boundary is set for the reservation period after 18:45.
[0087] During the execution phase, the home dispatch controller issues intelligent robot dispatch instructions and home device control instructions based on the execution commitment window, differential recalculation boundary, and resource reservation table.
[0088] At 18:42, the actual power of the kitchen heating equipment increased, causing a change in the household status data.
[0089] The home scheduling controller reacquires the updated home status data and adjusts only the service intent sequence and resource reservation table within the differential recalculation boundary. It switches the local cleaning task from standard mode to low power mode and postpones the dishwashing reservation task by 6 minutes. At the same time, it keeps the welcoming mobile task, the entrance lighting turn-on task, the living room air conditioner switching task, and the kitchen heating maintenance task within the execution commitment window unchanged.
[0090] After adjustments, the total household power is maintained within the allowable range corresponding to the household power budget curve, the thermal load of the smart robot's main control chip is maintained within the allowable range corresponding to the chip's thermal budget curve, and the core service chain in the continuous household service scenario continues to execute continuously.
[0091] As can be seen from this embodiment, in the scenario of continuous home service, this method can not only simultaneously constrain the home power load and the thermal load of the smart robot's main control chip, but also clearly distinguish between stable service actions and service actions that can be adjusted by executing commitment windows and differential recalculation boundaries, thereby reducing the phenomenon of repeated rescheduling and frequent start-stop of equipment in the evening peak scenario.
[0092] Example 2
[0093] Please refer to Figures 1-5 Specifically: In S1, the home dispatch controller collects power data of the power distribution circuits in the smart home, temperature data of the smart robot's main control chip, battery status data of the smart robot, operating status data of home devices, occupancy status data of home space, and data of service tasks to be executed, according to a unified sampling clock.
[0094] The collected data is time-aligned to obtain subsets of power distribution branch power status, chip thermal status, space occupancy status, and task status, and these subsets are then aggregated to form the household status time series set.
[0095] The set of service tasks to be executed is determined from the service tasks to be executed that are in the pending scheduling state and correspond to the same family continuous service scenario and the same service target identifier. Based on the execution order constraints, resource call constraints and result trigger constraints of each service task to be executed, the task pre-dependencies between the service tasks to be executed are determined.
[0096] In this embodiment, the same evening peak family continuous service scenario as in Embodiment 1 is implemented. The focus of this embodiment is on step S1, namely, family status data collection, time alignment, construction of family status time series set, generation of service task set to be executed, and determination of task prerequisite dependencies.
[0097] During the scheduling period from 18:20 to 19:20, the home scheduling controller collects power data of the power distribution circuits in the smart home, temperature data of the main control chip of the smart robot, battery status data of the smart robot, operating status data of home devices, occupancy status data of home space, and data of service tasks to be executed according to a unified sampling clock.
[0098] Among them, the power distribution branch power data is collected by the branch monitoring nodes in the residential distribution box. The collection objects include the living room air conditioner branch, the kitchen heating equipment branch, the dishwashing equipment branch, the water heater branch, and the robot charging branch. The sampling period is 1 second.
[0099] Temperature data from the main control chip of the intelligent robot is reported by the temperature monitoring unit of the intelligent home robot itself, with a sampling period of 2 seconds.
[0100] The battery status data of the intelligent robot is reported by the intelligent home robot battery management unit, with a sampling period of 1 second.
[0101] The operating status data of home appliances is returned by the home appliance control interface, including the appliance's on / off status, operating level, operating stage, and current power level;
[0102] The occupancy status data of the home space is jointly provided by door magnetic sensors, millimeter-wave presence sensors and indoor positioning nodes, which record the occupancy status of each room and main passage at the corresponding time.
[0103] The data for service tasks to be executed is generated by preset home service policies and real-time triggering events, and includes at least the task identifier, task type, target service identifier, current task status, triggering conditions, target device or target area.
[0104] Taking 18:35:20 as an example, the home dispatch controller collected data showing that the power of the kitchen heating equipment was 2.08kW, the power of the living room air conditioner was 0.76kW, the temperature of the main control chip of the service robot was 63.4℃, the remaining battery power of the service robot was 58%, and the entrance area was determined to be occupied by the door magnet and millimeter wave. At the same time, the system generated the task of turning on the entrance lighting and the task of welcoming guests.
[0105] To enable asynchronously sourced data to participate in unified scheduling, the home scheduling controller performs time alignment processing on the collected data.
[0106] Specifically, a discrete time sequence t0, t1, t2, ... is defined using a unified sampling clock, with an interval of 1 second between any two adjacent time points.
[0107] For power distribution branch power data and smart robot battery status data with a sampling period of 1 second, the home dispatch controller directly maps the original sampled values to the corresponding time.
[0108] For the temperature data of the main control chip of the smart robot with a sampling period of 2 seconds, the home scheduling controller uses linear interpolation between two adjacent temperature sampling points to supplement the temperature value at the intermediate time.
[0109] For event-triggered home space occupancy status data and pending service task data, the home scheduling controller maps the event occurrence time to the nearest and no earlier unified sampling time, and maintains this state until the next state change event arrives.
[0110] For a small number of missing samples caused by communication jitter, if the missing duration does not exceed 3 seconds, the most recent valid state will be used.
[0111] If the missing duration exceeds 3 seconds, the corresponding state will be marked as a state to be verified, and the weight of this state will be reduced in subsequent scheduling.
[0112] Through the above processing, data with different sampling frequencies, different triggering methods, and different sources are unified onto the same time axis, providing a time-consistent foundation for subsequent state modeling.
[0113] After time alignment is completed, the home dispatch controller obtains the power status subsets of the power distribution branches, the thermal status subsets of the chips, the space occupancy status subsets, and the task status subsets, and summarizes each subset to form the home status time sequence set.
[0114] The power status subset of the distribution branch is indexed by the distribution branch identifier. The instantaneous power value of the corresponding branch is recorded at each unified sampling time, and the local average power is calculated on 10 consecutive sampling points to characterize the load level of the branch in a short period of time.
[0115] The chip thermal status subset is indexed by the robot identifier, recording the temperature value of each robot's main control chip and the rate of temperature change in the past 30 seconds. The battery status data is also written into the corresponding status record of the robot as a field of the robot's sustainable operation capability.
[0116] The space occupancy status subset is indexed by rooms and passageways, recording the occupancy status and duration of the entrance hall, living room, kitchen, dining room and connecting passageways at each unified sampling time;
[0117] The task status subset is indexed by the task identifier and records the service target identifier to which the task belongs, the current task status, the target area, the triggering conditions, the required resources, and whether the task has entered the waiting-to-schedule state.
[0118] The above four state subsets are summarized on the same timeline to form the family state time series set.
[0119] Therefore, the family state time series not only preserves the original physical meaning of the multi-source states, but also ensures the time consistency of subsequent budget curve calculation and task selection.
[0120] In this embodiment, a total of 13 candidate tasks were identified from the data of service tasks to be executed, including entryway lighting turn-on task, living room air conditioner switching task, welcoming guest movement task, kitchen heating maintenance task, tableware transfer task, dishwashing appointment task, partial cleaning task, hot water compensation task, and other background tasks that are not directly related to the current scene.
[0121] The home scheduling controller filters out the service tasks to be executed that are in the scheduling state and correspond to the same continuous service scenario and the same service target identifier, thus obtaining a set of service tasks to be executed.
[0122] The home-based continuous service scenario here remains consistent with Example 1, which is a scenario of home delivery and dinner coordination during the evening peak hours; the service target identifier is uniformly set as "dinner home delivery coordination service".
[0123] After screening, the following tasks were identified as pending service tasks: entrance lighting activation, living room air conditioning switching, welcoming guests movement, kitchen heating maintenance, tableware transfer, dishwashing appointment, partial cleaning, and hot water replenishment. The remaining five background tasks were excluded from the pending service task set because they did not belong to the current service target identifier or had not yet met the scheduling status.
[0124] After the set of service tasks to be executed is generated, the family scheduling controller further determines the pre-dependencies between the service tasks to be executed based on the execution order constraints, resource call constraints, and result trigger constraints.
[0125] The execution sequence constraint is used to limit the start of another task only after the completion of one task or the entry into a specified stage; for example, in this embodiment, the task of turning on the entrance lighting is executed before the task of welcoming guests, in order to ensure that the welcoming guest movement path and the entrance area have sufficient lighting conditions.
[0126] Resource allocation constraints are used to limit the order in which tasks occupy the same resource; for example, both the tableware transfer task and the local cleaning task need to occupy the passage path and work area of the restaurant area, so the local cleaning task can only be executed after the tableware transfer task releases the corresponding space resources.
[0127] Result-triggered constraints are used to limit the execution of a task to a specified result before the next task can enter the pending state. For example, a dishwashing reservation task is only triggered after the dish transfer task is completed and the dishwashing equipment loading conditions are met.
[0128] The home scheduling controller analyzes the target area, required equipment, current status, and triggering conditions of each task, formalizes the above three types of constraints into task prerequisite dependencies, and establishes a task dependency table.
[0129] In this embodiment, a total of 8 valid task prerequisite dependencies were identified. Among them, there is a result-triggered dependency between the entryway lighting turn-on task and the welcoming movement task, an execution order dependency between the kitchen heating maintenance task and the tableware transfer task, a resource call dependency between the tableware transfer task and the local cleaning task, and the remaining dependencies are determined by the equipment occupancy and result release conditions.
[0130] Example 3
[0131] Please refer to Figures 1-5Specifically: In S2, the home scheduling controller determines the home power allowable range and chip thermal load allowable range corresponding to each time slice according to the home status time sequence set and the same preset time window.
[0132] Then, based on the preset mapping relationship between chip thermal load and equivalent power, the allowable range of chip thermal load for each time slice is converted into the equivalent power constraint range.
[0133] Intersection of the allowable power range and the equivalent power constraint range for each time slice is performed to obtain the coupling constraint range for each time slice.
[0134] In S2, a continuous service chain is determined based on the task prerequisite dependencies in the set of service tasks to be executed; the continuous service chain consists of multiple service tasks to be executed, wherein the multiple service tasks to be executed correspond to the same service target identifier and have task prerequisite dependencies on each other;
[0135] Then, calculate the estimated energy consumption and incremental chip thermal load for each service task to be executed, and combine them with the interruption tolerance index of the corresponding continuous service chain, and calculate the weighted evaluation value according to the preset weights.
[0136] Based on the comparison results of the weighted evaluation value with the first division threshold and the second division threshold, as well as the matching of the execution energy consumption estimate and the chip thermal load increment estimate with the coupling constraint interval, the service retention level is divided for each service task to be executed, and the service retention level is obtained.
[0137] The service retention levels include retained service, downgraded service, and deferred service;
[0138] Wherein, the first division threshold is greater than the second division threshold;
[0139] When the weighted evaluation value is greater than or equal to the first division threshold, the corresponding service task to be executed will be classified as a reserved service;
[0140] When the weighted evaluation value is less than the first division threshold and greater than or equal to the second division threshold, the control parameter item corresponding to the service task to be executed is extracted. The control parameter item includes at least one of the following: intelligent robot movement speed, intelligent robot movement acceleration, home device operating power, and task execution time.
[0141] The control parameter items are then compressed according to a preset compression rule to obtain a compressed set of control parameters.
[0142] Based on the set of compressed control parameters, the estimated energy consumption after compression and the estimated incremental thermal load of the chip after compression are recalculated.
[0143] When both the compressed execution energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range of the corresponding time period, and the compressed task execution time does not exceed the maximum allowable delay corresponding to the continuous service chain, the corresponding service task to be executed will be classified as a degraded service; otherwise, the corresponding service task to be executed will be classified as a delayed service.
[0144] The maximum allowable delay for a continuous service chain is determined by the home scheduling controller based on the task prerequisites, service target completion time limits, and home device response time limits in the continuous service chain.
[0145] When the weighted evaluation value is less than the second division threshold, the corresponding service task to be executed will be classified as a delayed service.
[0146] In this embodiment, the same evening peak-hour family continuous service scenario as in Embodiments 1 and 2 is implemented.
[0147] The application targets are still three-bedroom residences, and the residence is equipped with 1 service-type intelligent robot, 1 sweeping and mopping intelligent robot, 1 air conditioner, 1 fresh air system, 1 electric water heater, 1 dishwashing machine, 1 set of kitchen heating equipment, and several lighting devices.
[0148] The home scheduling controller is deployed in the edge gateway, with a unified sampling clock of 1 second, a scheduling time slice of 1 minute, a rolling scheduling time domain of 15 minutes, and the current scheduling start time set to 18:35.
[0149] In Example 2, the family status time series set, the set of service tasks to be executed, and the task prerequisite dependencies have been obtained. This example focuses on explaining the specific generation process of the family power allowable range, the chip thermal load allowable range, the coupling constraint range, the continuous service chain, the execution energy consumption estimate, the chip thermal load increment estimate, the weighted evaluation value, and the service retention level.
[0150] At 18:35, the home dispatch controller extracts the status data corresponding to 18:35 to 18:50 from the home status time series set. The total power distribution limit for the residence remains at 6.50kW, and the safety redundancy remains at 0.45kW.
[0151] The home dispatch controller performs sliding statistics on the power status of the distribution circuits over the past 60 seconds, and obtains the average power of the kitchen heating equipment circuit at 2.08kW, the average power of the living room air conditioning circuit at 0.76kW, the average power of the lighting circuit at 0.18kW, and the average power of other fixed base loads at 0.95kW.
[0152] Therefore, the home dispatch controller determines the allowable power range for each time slot by subtracting the basic load from the total power distribution limit and the safety redundancy.
[0153] Taking the five time slots from 18:36 to 18:40 as an example, the upper limits of the allowable power range for households are 3.53kW, 3.61kW, 3.58kW, 3.49kW and 3.55kW respectively. Therefore, the corresponding allowable power ranges for households are 0 to 3.53kW, 0 to 3.61kW, 0 to 3.58kW, 0 to 3.49kW and 0 to 3.55kW respectively.
[0154] Subsequently, the home scheduling controller determines the allowable range of chip thermal load.
[0155] At 18:35, the temperature of the main control chip of the service robot was 63.4℃, and the temperature of the main control chip of the sweeping and mopping robot was 60.8℃. The thermal protection temperature of the main control chip was set to 82℃, and the thermal safety margin was set to 4℃. Therefore, the acceptable temperature rise margin of the service robot was 14.6℃, and the acceptable temperature rise margin of the sweeping and mopping robot was 17.2℃.
[0156] The home scheduling controller invokes a preset mapping relationship between chip thermal load and equivalent power. This mapping relationship is established based on offline calibration: under an ambient temperature of 26°C, different computational loads are applied to the main control chip, 30 seconds of temperature rise data are recorded, and a correspondence table between chip power and temperature rise is fitted.
[0157] Based on this mapping relationship, the acceptable temperature rise margin is converted into the equivalent power constraint interval corresponding to each time slice.
[0158] Taking service robots as an example, the upper bounds of the equivalent power constraint intervals from 18:36 to 18:40 are 1.02kW, 0.98kW, 0.95kW, 0.92kW and 0.90kW respectively. Therefore, the corresponding equivalent power constraint intervals are 0 to 1.02kW, 0 to 0.98kW, 0 to 0.95kW, 0 to 0.92kW and 0 to 0.90kW respectively.
[0159] The home dispatch controller performs interval intersection on the home power allowable interval and the equivalent power constraint interval within each time slice. Specifically, it takes the larger of the two interval lower limits as the lower limit of the coupling constraint interval and the smaller of the two interval upper limits as the upper limit of the coupling constraint interval.
[0160] Since the lower limit of both intervals in this embodiment is 0, the coupling constraint intervals from 18:36 to 18:40 are 0 to 1.02kW, 0 to 0.98kW, 0 to 0.95kW, 0 to 0.92kW, and 0 to 0.90kW, respectively.
[0161] Thus, the home dispatch controller obtains a unified boundary that can simultaneously satisfy both home power distribution constraints and main control chip thermal constraints.
[0162] After obtaining the coupling constraint interval, the home scheduling controller determines the continuous service chain based on the task pre-dependencies already formed in Example 2.
[0163] The current set of service tasks to be executed includes tasks such as turning on the entrance lighting, welcoming guests, switching the living room air conditioner, maintaining the kitchen heating, transferring tableware, scheduling dishwashing, partial cleaning, and replenishing hot water. All of these tasks correspond to the same service target identifier: "Dinner Delivery Collaboration Service".
[0164] Among them, the task of turning on the entrance lighting precedes the task of welcoming guests. After the task of welcoming guests is completed, the task of switching the living room air conditioner enters the execution phase, thus forming the first continuous service chain. The task of transferring tableware can only be executed after the task of maintaining kitchen heating is completed. After the task of transferring tableware is completed, the task of scheduling dishwashing enters the pending execution state, thus forming the second continuous service chain. After the task of transferring tableware releases the resources of the dining area, the task of partial cleaning can begin, thus forming the third continuous service chain.
[0165] Continuous service chains are used to characterize the continuous execution relationship and interrupt sensitivity between tasks under the same service objective.
[0166] After the continuous service chain is determined, the home scheduling controller calculates the estimated energy consumption and the estimated incremental chip thermal load for each service task to be executed.
[0167] The execution energy consumption estimate is used to represent the total electrical energy consumption of a certain service task to be executed during the expected execution period, and the chip thermal load increment estimate is used to represent the predicted temperature rise of the intelligent robot's main control chip caused by the service task to be executed during the execution period.
[0168] To ensure the feasibility of maintaining service tiers, the home dispatch controller employs different forecasting methods for different task types.
[0169] For robot motion tasks, the home scheduling controller divides task energy consumption into three parts: drive energy consumption, computing energy consumption, and auxiliary energy consumption.
[0170] The driving energy consumption is determined by the task path length, motion speed, motion acceleration, load mass, and motion resistance of the robot chassis.
[0171] Computational energy consumption is determined by the visual processing frequency, path planning complexity, and control loop refresh frequency.
[0172] The auxiliary energy consumption consists of communication, sensor power supply and standby load of the main body.
[0173] The home scheduling controller first determines the expected execution time based on the task path length and action parameters. Then, it combines the drive power calculation formula and power calculation formula obtained from the calibration to calculate the average drive power, average calculation power, and average auxiliary power during the task execution period. Finally, it multiplies the sum of the three by the expected execution time to obtain the estimated energy consumption of the task.
[0174] Taking the welcoming mobile task as an example, the task path length is 11.6 m, the standard movement speed is 0.82 m / s, and the movement acceleration is 0.56 m / s². 2 The execution time is expected to be 15.1 seconds. After calibration, the average driving power is 420 W, the average computing power is 110 W, and the auxiliary power is 30 W. Therefore, the estimated energy consumption of this task is approximately 2.35 Wh.
[0175] For the tableware transfer task, due to the longer path and the load handling, the average drive power is increased to 510 W, the average computing power is 130 W, the auxiliary power is 40 W, the expected execution time is 17.4 s, and the corresponding execution energy consumption is estimated to be about 3.28 Wh.
[0176] For home device control tasks, the home scheduling controller directly calculates the estimated energy consumption based on the device's operating power and expected runtime.
[0177] If the equipment has multiple power phases during the task, the power is accumulated separately for each phase.
[0178] Taking a scheduled dishwashing task as an example, if the rated operating power of the dishwashing equipment is 0.94 kW and the expected running time is 32 minutes, then the estimated energy consumption for the task is approximately 0.50 kWh.
[0179] The rated operating power of the hot water compensation task is 1.46 kW, the expected operating time is 18 min, and the estimated energy consumption is approximately 0.44 kWh.
[0180] For mixed tasks such as localized cleaning, which involve both robot movements and equipment load, the home scheduling controller calculates the combined energy consumption of the robot during movement and the energy consumption of the cleaning module during operation.
[0181] In this embodiment, the cleaning power of the local cleaning task is 180 W, the average power of robot movement and calculation is about 280 W, and the expected execution time is 9 min. Therefore, the estimated energy consumption is about 69 Wh.
[0182] Corresponding to the estimated energy consumption, the home dispatch controller further calculates the estimated incremental thermal load of the chip.
[0183] The chip thermal load increment estimate is used to represent the temperature rise increment of the main control chip of the intelligent robot caused by a certain service task to be executed during the expected execution period. In this embodiment, the home scheduling controller pre-establishes a task characteristic chip temperature rise increment lookup table, and before the task is executed, it looks up the chip thermal load increment estimate from the task characteristic chip temperature rise increment lookup table according to the task characteristics of the service task to be executed.
[0184] The task feature chip temperature rise increment comparison table is obtained through pre-calibration.
[0185] Specifically, under the condition of maintaining an ambient temperature of 26 ℃, the intelligent robot was made to perform multiple sets of standard action tasks and mixed tasks, and the initial temperature of the main control chip before the task execution, the final temperature of the main control chip at the end of the task execution, and the duration of the task were recorded.
[0186] For each task group, at least the task path length, action speed, action acceleration, load quality, number of visual inference attempts, number of path planning nodes, control loop refresh frequency, and task execution time should be recorded.
[0187] The home scheduling controller determines the difference between the end temperature and the initial temperature as the chip temperature rise increment for that group of tasks, and writes it into the task characteristic chip temperature rise increment lookup table according to the task type and task characteristic range.
[0188] In the actual scheduling process, the home scheduling controller first determines the task characteristics of the service task to be executed.
[0189] For robot motion tasks such as welcoming mobile tasks and tableware transfer tasks, the task characteristics should include at least the task path length, motion speed, motion acceleration, load mass, number of visual inferences, number of path planning nodes, and task execution time.
[0190] For mixed tasks such as localized cleaning, the task characteristics also include cleaning power levels;
[0191] For equipment control tasks such as dishwashing scheduling and hot water replenishment, since the robot only performs low-intensity scheduling control, the task characteristics are mainly the task duration and the control loop refresh frequency.
[0192] Based on the above task characteristics, the home scheduling controller first determines the corresponding task type, and then searches for the calibration record that is closest to the current task characteristics in the task characteristic chip temperature rise increment comparison table.
[0193] If the current task characteristics are between two calibration intervals, then the temperature rise increments corresponding to the two calibration records are linearly interpolated to obtain the estimated value of the chip thermal load increment.
[0194] Taking the welcoming movement task as an example, the current task path length is 11.6 m, the movement speed is 0.82 m / s, and the movement acceleration is 0.56 m / s². 2 The visual reasoning iterations were performed 24 times, the path planning nodes numbered 82, and the task execution time was 15.1 seconds. The home scheduling controller, referring to the task characteristics—chip temperature rise increment comparison table, found that when the path length was between 10 m and 12 m, the movement speed between 0.75 m / s and 0.85 m / s, and the movement acceleration was between 0.50 m / s², the following conditions were met: 2 up to 0.60 m / s 2 When the number of visual reasoning iterations is between 20 and 25 and the number of path planning nodes is between 75 and 90, the corresponding chip temperature rise increment range is 1.9℃ to 2.2℃. Since the current task characteristics are in the middle of this range, the home scheduling controller takes the interpolated result of 2.1℃ as the estimated chip thermal load increment for the welcoming mobile task.
[0195] Taking the tableware transfer task as an example, the current task path length is 14.8 m, the load mass is 3.4 kg, the movement speed is 0.88 m / s, and the movement acceleration is 0.62 m / s². 2 The path planning involves a large number of nodes and requires replanning to avoid obstacles along the way. The home dispatch controller found in the task characteristic chip temperature rise increment comparison table that the chip temperature rise increment for the corresponding task characteristic range is 2.6 ℃ to 2.9 ℃. After interpolation, 2.8 ℃ is taken as the estimated chip heat load increment for the tableware transfer task.
[0196] Taking a local cleaning task as an example, the cleaning power is 180 W, the moving speed is 0.43 m / s, the task execution time is 9 min, and the chip temperature rise increment in the corresponding range in the table is 1.7 ℃ to 2.0 ℃. After interpolation, 1.9 ℃ is taken as the estimated value of the chip thermal load increment for the local cleaning task.
[0197] For the dishwashing reservation task and the hot water compensation task, since the robot's main control chip only performs low-intensity task scheduling and status monitoring, the estimated values of its chip heat load increment obtained by the home scheduling controller after looking up the table are 0.4 ℃ and 0.3 ℃, respectively.
[0198] The preset compression rule is a set of parameter compression rules pre-stored by the home scheduling controller. It is used to compress the control parameter items of the corresponding service task step by step when the weighted evaluation value of the service task to be executed is less than the first division threshold and greater than or equal to the second division threshold, so as to reduce the estimated execution energy consumption and the estimated incremental chip thermal load of the service task to be executed.
[0199] The preset compression rules include at least the control parameter selection rules, control parameter compression order rules, control parameter compression step size rules, control parameter compression lower limit rules, and compression stop determination rules.
[0200] The control parameter selection rules are used to determine the control parameters that can be compressed based on the task type. For robot motion tasks, the control parameters include at least one of motion speed, motion acceleration, and task execution time. For home appliance control tasks, the control parameters include at least one of device operating power, device output level, and task execution time. For mixed tasks, the control parameters include at least two of motion speed, motion acceleration, device operating power, and task execution time.
[0201] The control parameter compression order rule is used to limit the order in which each control parameter is adjusted. The home scheduling controller prioritizes compressing control parameters that have a small impact on continuous service results but a significant impact on the estimated energy consumption and the estimated incremental chip thermal load. For robot motion tasks, the acceleration is compressed first, then the speed, and finally the task execution time is extended. For home device control tasks, the device operating power or output level is reduced first, and then the task execution time is extended.
[0202] The compression step size rule in the control parameter item is used to limit the compression amplitude each time. The action speed decreases step by step according to the preset ratio of the rated action speed, the action acceleration decreases step by step according to the preset ratio of the rated action acceleration, and the equipment operating power decreases step by step according to the preset ratio of the rated operating power or adjacent power levels.
[0203] The control parameter compression lower limit rule is used to limit the boundaries of the compressed control parameters. The compressed action speed is not lower than the lower limit of the safe operating speed of the intelligent robot, the compressed action acceleration is not lower than the lower limit of the steady-state control of the intelligent robot, the compressed equipment operating power is not lower than the minimum stable operating power of the equipment, and the compressed task execution time does not exceed the maximum allowable delay corresponding to the continuous service chain.
[0204] The compression stop determination rule is used to limit the compression termination conditions. After each parameter compression, the home dispatch controller recalculates the post-compression execution energy consumption estimate and the post-compression chip thermal load increment estimate. When both the post-compression execution energy consumption estimate and the post-compression chip thermal load increment estimate fall within the coupling constraint range of the corresponding time period, compression stops. When all control parameter items reach the corresponding compression lower limit but still do not meet the coupling constraint range, compression also stops, and the corresponding pending service task is classified as a delayed service.
[0205] After obtaining the estimated energy consumption and the estimated incremental thermal load of the chip, the home scheduling controller uses both as quantitative inputs for task classification.
[0206] For service tasks to be executed that fall between the first and second division thresholds, the home scheduling controller will also extract the corresponding task's control parameter items and form a compressed control parameter set according to the preset compression rules. Then, based on the compressed action speed, action acceleration, device operating power, and task execution duration, it will recalculate the compressed execution energy consumption estimate and the compressed chip thermal load increment estimate.
[0207] The home scheduling controller will classify a task as a degraded service only if both the compressed execution energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range of the corresponding time slice, and the compressed task execution time does not exceed the maximum allowable delay corresponding to the continuous service chain; otherwise, the task will be classified as a delayed service.
[0208] Therefore, the estimated energy consumption and incremental chip thermal load are not only used for simple condition assessment, but also directly constitute the basis for determining service availability levels and subsequent resource reservations.
[0209] After obtaining the estimated energy consumption and incremental chip thermal load, the home scheduling controller comprehensively evaluates each service task to be executed by combining the interruption tolerance index of the continuous service chain. The interruption tolerance of the first continuous service chain is set to 0.91, the second continuous service chain to 0.78, and the third continuous service chain to 0.63. The home scheduling controller generates weighted evaluation values using interruption tolerance, task timeliness, and task dependency depth as inputs, with preset weights of 0.5, 0.3, and 0.2. The weighted evaluation values for the tasks of turning on the entryway lighting, welcoming guests, switching the living room air conditioner, maintaining the kitchen heating, transferring tableware, partial cleaning, scheduling dishwashing, and compensating for hot water are 0.83, 0.79, 0.74, 0.77, 0.66, 0.58, 0.41, and 0.36, respectively. The first division threshold is set to 0.72, and the second division threshold is set to 0.44, with the first division threshold being greater than the second division threshold. The home dispatch controller classifies the service availability level of each service task to be executed based on the matching of the weighted evaluation value, the estimated energy consumption, the estimated incremental chip thermal load, and the coupling constraint interval.
[0210] The weighted evaluation values of the tasks of turning on the entrance lighting, welcoming guests, switching the living room air conditioner, and maintaining the kitchen heating are all no less than 0.72, and they can meet the coupling constraint interval within the corresponding time slice. Therefore, they are classified as reserved services.
[0211] For tasks with weighted evaluation values between 0.44 and 0.72, the home dispatch controller enters a downgrade evaluation process. Taking the tableware transfer task as an example, the home dispatch controller extracts the action speed, action acceleration, and task execution time as control parameters, and compresses the action speed from 0.88 m / s to 0.58 m / s and the action acceleration from 0.62 m / s² to 0.37 m / s² according to preset compression rules. 2 After recalculation, the estimated energy consumption for the tableware transfer task decreased from 0.19 kWh to 0.15 kWh, the estimated incremental chip heat load decreased from 2.8℃ to 1.9℃, and the task execution time increased from 4.6 min to 6.3 min.
[0212] For localized cleaning tasks, the home dispatch controller reduces the cleaning power from 180W to 120W and the moving speed from 0.43m / s to 0.31m / s. After recalculation, the estimated energy consumption decreases from 0.16kWh to 0.12kWh, the estimated chip heat load increment decreases from 1.9℃ to 1.3℃, and the task execution time is extended from 9 minutes to 11 minutes. The maximum allowable delay for the second continuous service chain is determined by the home dispatch controller to be 8 minutes based on task prerequisites, service target completion time limits, and home device response time limits. The maximum allowable delay for the third continuous service chain is determined to be 6 minutes.
[0213] The tableware transfer task and the local cleaning task, after compression, did not exceed the maximum allowable delay of their respective continuous service chains, and their compressed execution energy consumption estimate and chip thermal load increment estimate could fall within the coupling constraint range of the corresponding time slice. Therefore, they were classified as degraded services.
[0214] The weighted evaluation values for the dishwashing reservation task and the hot water compensation task were 0.41 and 0.36, respectively, which were less than the second classification threshold, and were directly classified as delayed services.
[0215] After the above processing, the service retention level results obtained by the home dispatch controller in this embodiment are as follows: the tasks of turning on the entrance lighting, welcoming guests, switching the living room air conditioner, and maintaining the kitchen heating are retained services; the tasks of transferring tableware and partial cleaning are degraded services; and the tasks of scheduling dishwashing and compensating for hot water are postponed services.
[0216] Compared to methods that rely solely on household power limits for task retention and postponement, this embodiment incorporates household power distribution capacity and robot thermal carrying capacity into the scheduling boundary through coupled constraint intervals; it transforms isolated scheduling objects into chained objects within a continuous service chain through continuous service chains and interruption tolerance indicators; and it enables some tasks to remain in the service chain in a degraded manner rather than being directly postponed through a dual threshold and control parameter compression mechanism.
[0217] Therefore, the technical solution defined in this embodiment can not only more accurately identify critical tasks and compressible tasks in the evening peak family continuous service scenario, but also significantly reduce subsequent scheduling oscillations and continuous service chain breaks, thus demonstrating the core inventiveness of this application.
[0218] Example 4
[0219] Please refer to Figures 1-5 Specifically: In S3, based on the service retention level and the continuous service chain, the set of service tasks to be executed is time-series arranged to obtain the service intent sequence;
[0220] When generating the service intent sequence, the task prerequisites in the continuous service chain are kept unchanged, and reserved services are arranged first, followed by degraded services, and finally delayed services.
[0221] When the priority order corresponding to the service retention level conflicts with the task's prerequisite dependency, the home scheduling controller will only adjust the order of pending service tasks that do not have a direct prerequisite dependency, while keeping the task's prerequisite dependency unchanged.
[0222] In S3, based on the service intent sequence, reservation time slots are allocated for intelligent robot motion resources and home device control resources on a unified time axis to form the resource reservation table;
[0223] The home scheduling controller establishes a time slot locking relationship between corresponding scheduled time slots for tasks that have a pre-dependent relationship in the same continuous service chain.
[0224] The time slot locking relationship is used to limit the minimum and maximum intervals between consecutive and related reservation time slots;
[0225] When multiple appointment time slots conflict, the home scheduling controller adjusts the conflicting appointment time slots in the order of reserved service, degraded service, and delayed service.
[0226] In S3, the home scheduling controller filters reservation time slots based on the resource reservation table and the coupling constraint interval; the selected reservation time slots meet the following conditions:
[0227] The reserved time period corresponds to a reserved service, or a degraded service that, after being compressed by control parameters, the compressed energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range.
[0228] The selected appointment time slots are designated as the execution commitment window, and the remaining appointment time slots are designated as the differential recalculation boundary.
[0229] The execution commitment window is used to keep the start time and resource usage relationship of the corresponding reservation period unchanged.
[0230] In this embodiment, the same evening peak household continuous service scenario as in Embodiments 1 to 3 is still applied.
[0231] This embodiment focuses on the implementation process, namely the generation of service intent sequences, the formation of resource reservation tables, the establishment of time period locking relationships, and the filtering methods for execution commitment windows and differential recalculation boundaries.
[0232] At 18:35, the home scheduling controller first performs time-series orchestration of the set of service tasks to be executed based on service availability levels and continuous service chains, generating a service intent sequence.
[0233] Instead of directly sorting globally according to a single priority during orchestration, the system first keeps the pre-dependencies of tasks in the continuous service chain unchanged, and then prioritizes reserved services, followed by degraded services, and finally postponed services, provided that the dependencies are satisfied.
[0234] Based on the continuous service chain already determined in Example 3, the first continuous service chain is "entrance lighting turn-on task, welcoming guest movement task, living room air conditioner switching task", the second continuous service chain is "kitchen heating maintenance task, tableware transfer task, dishwashing appointment task", and the third continuous service chain is "tableware transfer task, local cleaning task".
[0235] The home dispatch controller first schedules the entryway lighting turn-on task in the first continuous service chain from 18:36 to 18:37, the welcoming guest movement task from 18:37 to 18:41, and the living room air conditioner switching task from 18:41 to 18:42; at the same time, it schedules the kitchen heating maintenance task in the second continuous service chain from 18:36 to 18:45, the tableware transfer task from 18:45 to 18:51, and the dishwashing reservation task from 18:52 to 18:58.
[0236] The local cleaning task in the third continuous service chain was initially scheduled after 18:51 because it was related to the tableware transfer task in terms of restaurant area resources.
[0237] Therefore, the sequence of service intentions generated by the home dispatch controller is as follows: entryway lighting turn-on task, welcoming guest movement task, living room air conditioning switching task, kitchen heating maintenance task, tableware transfer task, partial cleaning task, dishwashing reservation task, and hot water compensation task.
[0238] If a conflict arises between the priority of the service retention level and the pre-dependency relationship of the task during the sorting process, the pre-dependency relationship of the task will be kept unchanged, and only the order of tasks that do not have a direct pre-dependency relationship will be adjusted locally.
[0239] In this embodiment, although the partial cleaning task is a degraded service with a higher priority than the dishwashing reservation task and the hot water compensation task, the home scheduling controller keeps the dish transfer task in advance because there is a resource release task prerequisite relationship between the partial cleaning task and the dish transfer task. The partial cleaning task is not inserted before the dish transfer task.
[0240] After the service intent sequence is generated, the home scheduling controller allocates reservation time slots for intelligent robot action resources and home device control resources according to a unified timeline, forming a resource reservation table. When reserving resources, the home scheduling controller breaks down the task into a five-tuple format of "task identifier, start time slice, end time slice, occupied resource, occupied area".
[0241] Taking this embodiment as an example, the task of turning on the entrance lighting occupies lighting control resources, with a scheduled time period from 18:36 to 18:37; the task of welcoming guests occupies the movement resources of the service robot and the passageway resources from the entrance to the living room, with a scheduled time period from 18:37 to 18:41; the task of switching the living room air conditioner occupies the air conditioner control resources, with a scheduled time period from 18:41 to 18:42; the task of maintaining kitchen heating occupies the kitchen heating equipment control resources, with a scheduled time period from 18:36 to 18:45; the task of transferring tableware occupies the loading resources of the service robot, the dining area resources, and the passageway resources, with a scheduled time period from 18:45 to 18:51; the task of partial cleaning occupies the cleaning resources of the sweeping and mopping robot and the cleaning operation resources of the dining area, with an initial scheduled time period from 18:51 to 19:00; the task of scheduling dishwashing occupies the dishwashing equipment control resources, with a scheduled time period from 18:52 to 18:58; and the task of replenishing hot water occupies the water heater control resources, with a scheduled time period from 19:00 to 19:18.
[0242] The resource reservation table formed from the above reservation information not only includes the task order, but also clearly reflects the resource occupancy status of each task on a unified timeline, thus providing a basis for conflict identification and subsequent screening.
[0243] To prevent tasks in a continuous service chain from being pulled too close or too far apart in subsequent adjustments, the home scheduling controller further establishes time-slot locking relationships in the resource reservation table.
[0244] The time-slot locking relationship is established for tasks that have a pre-task dependency relationship in the same continuous service chain, and limits the minimum and maximum intervals between the corresponding scheduled time slots.
[0245] The minimum interval is used to ensure that there is sufficient time for the state to be released after the previous task is completed, while the maximum interval is used to limit the subsequent task from being delayed for too long and thus disrupting the continuous service chain.
[0246] In this embodiment, the minimum interval between the kitchen heating maintenance task and the tableware transfer task is set to 1 minute, and the maximum interval is set to 6 minutes, to ensure that the kitchen appliances switch states and the service robot completes loading preparation; the minimum interval between the tableware transfer task and the dishwashing reservation task is set to 0 minutes, and the maximum interval is set to 7 minutes, to ensure that the dishwashing equipment can resume operation in a timely manner after the tableware is delivered; the minimum interval between the tableware transfer task and the local cleaning task is set to 2 minutes, and the maximum interval is set to 8 minutes, to ensure that the restaurant area resources are released and cleaning operations are prepared. After establishing the time-slot locking relationship, the home scheduling controller performs a conflict check on the resource reservation table.
[0247] This embodiment identifies three types of conflicts: First, both tableware transfer tasks and local cleaning tasks occupy restaurant area resources, forming an area occupation conflict;
[0248] Second, if the dishwashing reservation task and the hot water compensation task are started at the same time, they will cause an additional conflict on the total power consumption of the household during the period from 18:58 to 19:00.
[0249] Third, if local cleaning tasks are not downgraded, they will squeeze the coupling constraint interval between 18:54 and 18:58 together with dishwashing appointment tasks.
[0250] In response to these conflicts, the home dispatch controller adjusts the conflict appointment time according to the order of reserved service, degraded service and delayed service. The partial cleaning task is postponed to start at 18:54 and the hot water compensation task is postponed to start at 19:04, thereby eliminating resource conflicts without disrupting the preceding relationship of the continuous service chain.
[0251] After the resource reservation table stabilizes, the home scheduling controller filters reservation time periods based on the resource reservation table and coupling constraint intervals to determine the execution commitment window and differential recalculation boundary.
[0252] The selection rules are as follows: reservation time slots for services that are to be retained are given priority to enter the candidate set; reservation time slots for services that are to be downgraded are only entered into the candidate set if, after the control parameters are compressed, both the compressed execution energy consumption estimate and the compressed chip thermal load increment estimate fall within the current time slice coupling constraint range; reservation time slots for services that are to be delayed are not entered into the candidate set.
[0253] In this embodiment, the tasks of turning on the entrance lighting, welcoming guests, and switching the living room air conditioner from 18:36 to 18:42 are all reserved services, as is the task of maintaining the kitchen heating from 18:36 to 18:45. Therefore, all these scheduled time slots are included in the candidate set.
[0254] The tableware transfer task is a downgraded service. After the control parameters in Example 3 are compressed, the compressed execution energy consumption estimate and the compressed chip heat load increment estimate both fall into the coupling constraint interval corresponding to each time slice from 18:45 to 18:51. Therefore, the reservation time slot from 18:45 to 18:51 also enters the candidate set.
[0255] Localized cleaning tasks are also considered degraded services. Although the power requirements can be met after compression in the two time slots after 18:59, they are close to the upper limit of the chip's thermal constraint. Therefore, the home scheduling controller only considers the localized cleaning appointment time slot from 18:54 to 18:58 as a candidate and does not include the appointment time slot after 18:59 in the scope of the execution commitment.
[0256] Ultimately, the home scheduling controller determines the reservation period from 18:36 to 18:51 as the execution commitment window, and the remaining reservation period after 18:52 as the differential recalculation boundary.
[0257] Once the reservation period in the execution commitment window is determined, its start time and resource occupancy relationship remain unchanged; the reservation period in the differential recalculation boundary allows for local adjustments when the state changes subsequently.
[0258] Through the above implementation process, the service intent sequence, resource reservation table, time period locking relationship, execution commitment window, and differential recalculation boundary form a continuously transmitted execution chain.
[0259] The service intent sequence transforms the service hold-up level into a specific execution order; the resource reservation table places the execution order onto specific resource usage on a unified timeline; the time slot locking relationship further fixes the task pre-dependencies in the continuous service chain into time constraints; the execution commitment window filters out reservation time slots from the resource reservation table that need to be prioritized and cannot be easily rewritten; and the differential recalculation boundary clarifies the scope of subsequent local recalculation.
[0260] Compared to the method of directly issuing tasks based solely on priority, this embodiment can significantly reduce the probability of scheduled tasks being repeatedly written back before execution, and reduce queuing and service chain breaks during the execution phase caused by shared resource conflicts.
[0261] Especially in the scenario of continuous home service during the evening peak, this embodiment enables the core service chain to maintain high stability under the conditions of fluctuations in home power and changes in chip thermal load by dividing the execution commitment window and differential recalculation boundary.
[0262] Example 5
[0263] Please refer to Figures 1-5 Specifically: In S4, when the household status data changes, the household scheduling controller updates the execution energy consumption estimate and chip thermal load increment estimate corresponding to the affected reservation period, and compares the updated execution energy consumption estimate and chip thermal load increment estimate with the coupling constraint interval of the corresponding time slice;
[0264] When there is a reservation period exceeding the coupling constraint interval within the execution commitment window, the home scheduling controller adjusts the resource consumption and power allocation within the corresponding time slice by migrating the reservation period corresponding to the delayed service within the differential recalculation boundary and switching the degraded service within the differential recalculation boundary to the execution state after control parameter compression, so as to release the coupling constraint interval margin of the corresponding time slice.
[0265] When there is a reservation period outside the coupling constraint interval within the differential recalculation boundary, the home scheduling controller performs local recalculation on the service intent sequence corresponding to the affected part and the resource reservation table;
[0266] The local recalculation maintains the start time, execution order, and resource usage relationship of the scheduled time slots within the execution commitment window unchanged.
[0267] In S4, the home scheduling controller updates the service retention level, service intent sequence, and resource reservation table according to the results of the local recalculation, and reissues the intelligent robot scheduling instructions and home device control instructions according to the updated service retention level, service intent sequence, and resource reservation table.
[0268] When the updated reservation period still exceeds the coupling constraint interval of the corresponding time slice, the home scheduling controller will migrate the corresponding delayed service to a time slice with a coupling constraint interval margin, or switch the corresponding degraded service to the execution state after the control parameters are compressed.
[0269] If, after migration and switching, there is still no executable time slice that satisfies the coupling constraint interval, the home scheduling controller suspends delayed services that have no task prerequisites with the current execution commitment window, in order to maintain the continuous execution of the corresponding services within the execution commitment window.
[0270] In this embodiment, the same evening peak family continuous service scenario as in Embodiments 1 to 4 is still used.
[0271] This embodiment focuses on the update method after changes in household status data, the partial recalculation method, and the method of re-issuing intelligent robot scheduling instructions and household device control instructions after partial recalculation.
[0272] At 18:55, the home dispatch controller received a new home status data change event.
[0273] Specifically, the bathroom sends an immediate hot water demand, causing the hot water compensation task to change from a delayed state to a high-demand state within the current scheduling time domain; at the same time, the service robot encounters a temporary obstacle at the restaurant entrance, increasing the number of path planning nodes from 82 to 119, and the expected number of visual reasoning operations from 24 to 31, thereby increasing the estimated incremental value of the chip heat load corresponding to the tableware transfer task.
[0274] The home scheduling controller first identifies the affected booking period, which is from 18:56 to 19:08. Then, the home scheduling controller recalculates the estimated energy consumption and incremental chip thermal load for this affected booking period.
[0275] During the recalculation, the hot water compensation task needs to be executed in advance, and the equipment operating power in its corresponding time slice is calculated as 1.46 kW; due to the increased path complexity, the estimated value of the compressed chip heat load increment for the tableware transfer task increases from 1.9℃ to 2.4℃; the local cleaning task remains in the compressed execution state, and the estimated value of its execution energy consumption in its corresponding time slice remains 0.12 kWh, and the estimated value of the chip heat load increment remains 1.3℃.
[0276] The home scheduling controller compares the above update results with the corresponding time-slice coupling constraint interval on a time-by-time basis.
[0277] The comparison results show that the comprehensive equivalent power occupancy in the three time slots of 18:56, 18:57 and 18:58 increased to 0.98 kW, 1.03 kW and 0.96 kW respectively, while the upper limit of the coupling constraint interval of the corresponding time slots was 0.88 kW, 0.85 kW and 0.83 kW respectively, indicating that the current scheduling result has the risk of exceeding the limit in the period from 18:56 to 18:58.
[0278] After identifying the risk of exceeding limits, the home dispatch controller first determines whether the risk of exceeding limits will affect the reservation time slot within the commitment window.
[0279] Since the dishwashing reservation task has entered the execution commitment window from 18:52 to 18:58, and the three time slices from 18:56 to 18:58 fall within this execution commitment window, the home scheduling controller does not directly rewrite the start time, execution order, and resource usage relationship of the dishwashing reservation task. Instead, it prioritizes releasing the coupling constraint interval margin of the corresponding time slice by adjusting the reservation period within the differential recalculation boundary.
[0280] The specific processing method includes two steps.
[0281] The first step is to move the hot water compensation task within the differential recalculation boundary from 18:56 to 19:14 to 19:12 to 19:30, thereby releasing the equipment operating power occupancy during the period from 18:56 to 19:08.
[0282] The second step is to keep the local cleaning tasks within the differential recalculation boundary in the execution state after the control parameters are compressed, and further shift the two scheduled time periods after 19:00 in the tail segment to the back by 4 minutes to reduce the cleaning resource usage and computing load usage during the period from 18:58 to 19:02.
[0283] After completing the above processing, the home scheduling controller re-compares the comprehensive equivalent power occupancy of the three time slots of 18:56, 18:57 and 18:58. The results are reduced to 0.82 kW, 0.84 kW and 0.80 kW respectively, which have fallen back into the coupling constraint interval of the corresponding time slots. Therefore, the dishwashing reservation task in the execution commitment window remains unchanged.
[0284] After completing the execution commitment window protection, the home scheduling controller continues to check whether there are still reservation periods outside the coupling constraint interval within the differential recalculation boundary.
[0285] The inspection results showed that the time slices at 19:13 and 19:14 were still close to the upper limit of the coupling constraint interval due to the instantaneous upgrade of the air conditioner and the postponement of the hot water compensation task.
[0286] In response, the home scheduling controller performs a partial recalculation of the service intent sequence and resource reservation table corresponding to the affected portion. During the partial recalculation, the home scheduling controller only rearranges the reservation time periods corresponding to the local cleaning task and hot water compensation task within the differential recalculation boundary, without changing the start time, execution order, and resource usage relationship of the dishwashing reservation task within the execution commitment window.
[0287] After local recalculation, the last 4 minutes of the local cleaning task were moved from 19:00 to 19:04 to 19:24 to 19:28, and the start time of the hot water compensation task was adjusted from 19:12 to 19:18, so that the two time slices of 19:13 and 19:14 would again have coupling constraint interval margins of 0.09kW and 0.12kW.
[0288] After the partial recalculation is completed, the home scheduling controller updates the service retention level, service intent sequence, and resource reservation table based on the partial recalculation results, and reissues the smart robot scheduling instructions and home device control instructions based on the updated results.
[0289] Specifically, local cleaning tasks will remain downgraded services, and hot water compensation tasks will remain deferred services, but their execution time slots will be shifted to later.
[0290] In the service intent sequence, the dishwashing appointment task remains in its original position, while the end of the partial cleaning task and the hot water compensation task are reordered.
[0291] The corresponding reservation time slots and resource occupancy relationships are updated synchronously in the resource reservation table.
[0292] The home dispatch controller then sends a dispatch instruction to the service robot to remain on standby after the current tableware transfer is completed, a dispatch instruction to the sweeping and mopping robot to postpone the final stage of partial cleaning, and a control instruction to the water heater control interface to postpone the start of the hot water compensation task.
[0293] To further illustrate the execution flow of this embodiment, a set of high-perturbation sub-scenarios is also set up in this embodiment.
[0294] In this sub-scenario, at 19:09, the kitchen heating equipment generates an additional 0.37 kW instantaneous power fluctuation, and the living room air conditioner automatically switches from comfort mode to enhanced mode, further narrowing the coupling constraint range between 19:12 and 19:18.
[0295] In this situation, the home scheduling controller re-executes the above update and partial recalculation process and finds that even if the hot water compensation task is further postponed and the local cleaning task is kept compressed, there is still no continuous executable time slice that satisfies the coupling constraint interval from 19:18 to 19:22.
[0296] Since there is no prerequisite relationship between the hot water compensation task and the dishwashing appointment task in the current execution commitment window, the home dispatch controller will temporarily suspend the hot water compensation task until 19:24 when it will resume execution.
[0297] Through this process, the corresponding services within the execution commitment window are continuously executed, and there are no instances where the committed appointment time slots have been rewritten.
[0298] As can be seen from this embodiment, the technical solution of this embodiment does not simply reorder all tasks after the state changes. Instead, it first updates the execution energy consumption estimate and chip thermal load increment estimate corresponding to the affected reservation period, then compares the updated result with the coupling constraint interval of the corresponding time slice time slice by time slice, and prioritizes releasing the coupling constraint interval margin of the corresponding time slice by migrating the delayed service reservation period within the differential recalculation boundary and switching the degraded service execution state within the differential recalculation boundary, thereby protecting the reservation period within the execution commitment window from being rewritten.
[0299] Meanwhile, if local recalculation still cannot find an executable time slice that satisfies the coupling constraint interval for certain scheduled time periods, the home scheduling controller can suspend delayed services that have no task prerequisites with the current execution commitment window to ensure the continuous execution of the currently committed service chain.
[0300] Thus, this embodiment forms a dynamic response chain of "updating, comparing, releasing margin, partial recalculation, re-issuing, and pausing when necessary," which enables the home continuous service scenario to maintain core services without interruption when affected by sudden power disturbances, thermal load fluctuations, and changes in real-time user demand, and significantly reduces the probability of global reordering and frequent device start-ups and shutdowns.
[0301] Example 6
[0302] Please see Figure 2 A smart robot scheduling system based on high-speed chip computing includes: a home status acquisition and construction module, a budget evaluation and grading module, an intent orchestration and reservation module, and an execution control recalculation module;
[0303] The home status acquisition and construction module collects home status data of the intelligent robot in real time through the home scheduling controller. The home status data includes power distribution circuit power data, intelligent robot main control chip temperature data, intelligent robot battery status data, home device operation status data, home space occupancy status data, and pending service task data. Based on the home status data, a home status time series set is constructed, and based on the pending service task data, a pending service task set is determined.
[0304] The budget assessment and grading module determines the household power budget curve and the chip thermal budget curve based on the household status time series set and the set of service tasks to be executed; and classifies the set of service tasks to be executed into service retention levels according to the household power budget curve and the chip thermal budget curve to obtain the service retention level.
[0305] The intent orchestration and reservation module orchestrates the set of service tasks to be executed in a time sequence based on the service retention level to obtain a service intent sequence; it allocates reservation time slots for intelligent robot motion resources and home device control resources on a unified time axis according to the service intent sequence to form a resource reservation table; and then, in combination with the home power budget curve and the chip thermal budget curve, it performs constraint verification on the resource reservation table to determine the execution commitment window and differential recalculation boundary.
[0306] The execution control recalculation module issues intelligent robot scheduling instructions and home device control instructions based on the execution commitment window, the differential recalculation boundary, and the resource reservation table. When the home status data changes, it only adjusts the corresponding service intent sequence and resource reservation table within the differential recalculation boundary, keeping the corresponding services within the execution commitment window unchanged, so that the total home power and the thermal load of the intelligent robot main control chip are limited by the home power budget curve and the chip thermal budget curve, respectively.
[0307] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.
Claims
1. A method for scheduling intelligent robots based on high-speed chip computing, characterized in that: Includes the following steps: S1, collect home status data of the smart robot in real time through the home scheduling controller. The home status data includes power distribution circuit power data, smart robot main control chip temperature data, smart robot battery status data, home device operation status data, home space occupancy status data, and service task data to be executed; construct a home status time series set based on the home status data, and determine a service task set to be executed based on the service task data to be executed. S2, based on the household status time series set and the set of service tasks to be executed, determine the household power budget curve and the chip thermal budget curve; according to the household power budget curve and the chip thermal budget curve, classify the set of service tasks to be executed into service retention levels to obtain service retention levels; S3, based on the service retention level, the set of service tasks to be executed is time-series orchestrated to obtain a service intent sequence; according to the service intent sequence, reservation time slots are allocated for intelligent robot motion resources and home device control resources on a unified time axis to form a resource reservation table; then, combined with the home power budget curve and the chip thermal budget curve, the resource reservation table is constrained and checked to determine the execution commitment window and differential recalculation boundary; S4. Based on the execution commitment window, the differential recalculation boundary, and the resource reservation table, issue intelligent robot scheduling instructions and home device control instructions; when the home status data changes, only adjust the service intent sequence and resource reservation table corresponding to the differential recalculation boundary, keep the corresponding service in the execution commitment window unchanged, so that the total home power and the thermal load of the intelligent robot main control chip are limited by the home power budget curve and the chip thermal budget curve, respectively.
2. The intelligent robot scheduling method based on high-speed chip computing according to claim 1, characterized in that: In S1, the home dispatch controller collects power data of the power distribution circuits in the smart home, temperature data of the main control chip of the smart robot, battery status data of the smart robot, operating status data of home devices, occupancy status data of home space, and data of service tasks to be executed, according to a unified sampling clock. The collected data is time-aligned to obtain the power status subsets of the power distribution branch, the thermal status subsets of the chip, the space occupancy status subsets, and the task status subsets, and the subsets are aggregated to form the home status time series set. The set of service tasks to be executed is determined from the service tasks to be executed that are in the pending scheduling state and correspond to the same family continuous service scenario and the same service target identifier. Based on the execution order constraints, resource call constraints and result trigger constraints of each service task to be executed, the task pre-dependencies between the service tasks to be executed are determined.
3. The intelligent robot scheduling method based on high-speed chip computing according to claim 2, characterized in that: In S2, the home scheduling controller determines the home power allowable range and chip thermal load allowable range corresponding to each time slice according to the home status time sequence set and the same preset time window. Then, based on the preset mapping relationship between chip thermal load and equivalent power, the allowable range of chip thermal load for each time slice is converted into the equivalent power constraint range. Intersection of the allowable power range and the equivalent power constraint range for each time slice is performed to obtain the coupling constraint range for each time slice.
4. The intelligent robot scheduling method based on high-speed chip computing according to claim 3, characterized in that: In S2, a continuous service chain is determined based on the task prerequisite dependencies in the set of service tasks to be executed; the continuous service chain consists of multiple service tasks to be executed, wherein the multiple service tasks to be executed correspond to the same service target identifier and have task prerequisite dependencies on each other; Then, calculate the estimated energy consumption and incremental chip thermal load for each service task to be executed, and combine them with the interruption tolerance index of the corresponding continuous service chain, and calculate the weighted evaluation value according to the preset weights. Based on the comparison results of the weighted evaluation value with the first division threshold and the second division threshold, as well as the matching of the execution energy consumption estimate and the chip thermal load increment estimate with the coupling constraint interval, the service retention level is divided for each service task to be executed, and the service retention level is obtained. The service retention levels include retained service, downgraded service, and deferred service; Wherein, the first division threshold is greater than the second division threshold; When the weighted evaluation value is greater than or equal to the first division threshold, the corresponding service task to be executed will be classified as a reserved service; When the weighted evaluation value is less than the first division threshold and greater than or equal to the second division threshold, the control parameter item corresponding to the service task to be executed is extracted. The control parameter item includes at least one of the following: intelligent robot movement speed, intelligent robot movement acceleration, home device operating power, and task execution time. The control parameter items are then compressed according to a preset compression rule to obtain a compressed set of control parameters. Based on the set of compressed control parameters, the estimated energy consumption after compression and the estimated incremental thermal load of the chip after compression are recalculated. When both the compressed execution energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range of the corresponding time period, and the compressed task execution time does not exceed the maximum allowable delay corresponding to the continuous service chain, the corresponding service task to be executed will be classified as a degraded service; otherwise, the corresponding service task to be executed will be classified as a delayed service. The maximum allowable delay for a continuous service chain is determined by the home scheduling controller based on the task prerequisites, service target completion time limits, and home device response time limits in the continuous service chain. When the weighted evaluation value is less than the second division threshold, the corresponding service task to be executed will be classified as a delayed service.
5. The intelligent robot scheduling method based on high-speed chip computing according to claim 4, characterized in that: In S3, the set of service tasks to be executed is time-series arranged according to the service retention level and the continuous service chain to obtain the service intent sequence. When generating the service intent sequence, the task prerequisites in the continuous service chain are kept unchanged, and reserved services are arranged first, followed by degraded services, and finally delayed services. When the priority order corresponding to the service retention level conflicts with the task's prerequisite dependency, the home scheduling controller will only adjust the order of pending service tasks that do not have a direct prerequisite dependency, while keeping the task's prerequisite dependency unchanged.
6. The intelligent robot scheduling method based on high-speed chip computing according to claim 5, characterized in that: In S3, based on the service intent sequence, reservation time slots are allocated for intelligent robot motion resources and home device control resources on a unified time axis to form the resource reservation table; The home scheduling controller establishes a time slot locking relationship between corresponding scheduled time slots for tasks that have a pre-dependent relationship in the same continuous service chain. The time slot locking relationship is used to limit the minimum and maximum intervals between consecutive and related reservation time slots; When multiple appointment time slots conflict, the home scheduling controller adjusts the conflicting appointment time slots in the order of reserved service, degraded service, and delayed service.
7. The intelligent robot scheduling method based on high-speed chip computing according to claim 6, characterized in that: In S3, the home scheduling controller filters reservation time slots based on the resource reservation table and the coupling constraint interval; the selected reservation time slots meet the following conditions: The reserved time period corresponds to a reserved service, or a degraded service that, after being compressed by control parameters, the compressed energy consumption estimate and the compressed chip thermal load increment estimate fall within the coupling constraint range. The selected appointment time slots are designated as the execution commitment window, and the remaining appointment time slots are designated as the differential recalculation boundary. The execution commitment window is used to keep the start time and resource usage relationship of the corresponding reservation period unchanged.
8. The intelligent robot scheduling method based on high-speed chip computing according to claim 7, characterized in that: In S4, when the home status data changes, the home scheduling controller updates the execution energy consumption estimate and chip thermal load increment estimate corresponding to the affected reservation period, and compares the updated execution energy consumption estimate and chip thermal load increment estimate with the coupling constraint interval of the corresponding time slice. When there is a reservation period exceeding the coupling constraint interval within the execution commitment window, the home scheduling controller adjusts the resource consumption and power allocation within the corresponding time slice by migrating the reservation period corresponding to the delayed service within the differential recalculation boundary and switching the degraded service within the differential recalculation boundary to the execution state after control parameter compression, so as to release the coupling constraint interval margin of the corresponding time slice. When there is a reservation period outside the coupling constraint interval within the differential recalculation boundary, the home scheduling controller performs local recalculation on the service intent sequence corresponding to the affected part and the resource reservation table; The local recalculation maintains the start time, execution order, and resource usage relationship of the scheduled time slots within the execution commitment window unchanged.
9. The intelligent robot scheduling method based on high-speed chip computing according to claim 8, characterized in that: In S4, the home scheduling controller updates the service retention level, service intent sequence, and resource reservation table according to the results of the local recalculation, and reissues the intelligent robot scheduling instructions and home device control instructions according to the updated service retention level, service intent sequence, and resource reservation table. When the updated reservation period still exceeds the coupling constraint interval of the corresponding time slice, the home scheduling controller will migrate the corresponding delayed service to a time slice with a coupling constraint interval margin, or switch the corresponding degraded service to the execution state after the control parameters are compressed. If, after migration and switching, there is still no executable time slice that satisfies the coupling constraint interval, the home scheduling controller suspends delayed services that have no task prerequisites with the current execution commitment window, in order to maintain the continuous execution of the corresponding services within the execution commitment window.
10. A smart robot scheduling system based on high-speed chip computing, applied to the smart robot scheduling method based on high-speed chip computing as described in any one of claims 1-9, characterized in that: include: The module includes a family status collection and construction module, a budget assessment and grading module, an intent orchestration and scheduling module, and an execution control and recalculation module.
