Air conditioner control method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses an air conditioner control method and device, electronic equipment and a storage medium. The method comprises the steps that an environment image of a preset space and a movement track of a user in the preset space are obtained; performing visual semantic segmentation on the environment image to obtain a segmentation result; determining a semantic partition map of the preset space according to the segmentation result and the motion trail; the semantic partition map comprises at least one functional region; receiving a voice instruction of a user, analyzing a control intention and a target area in the voice instruction, and matching the target area with a functional area in the semantic partition map; and a control strategy of the air conditioner is generated according to the control intention and the matched functional area, and the air conditioner is controlled according to the control strategy. Different functional areas can be recognized by generating the semantic partition map, and an accurate control strategy is generated by analyzing the control intention and the target area in the instruction and combining the control intention and the target area, so that the air conditioner can be adjusted according to the different functional areas, and the control efficiency of the air conditioner can be improved.

Description

Technical Field

[0001] This invention relates to the field of air conditioning control technology, and in particular to an air conditioning control method, an air conditioning control device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] With the increasing popularity of smart homes and central air conditioning systems, indoor environmental control is gradually evolving from simple temperature regulation to refined services driven by contextual awareness and user intent. Traditional air conditioning control methods mainly rely on manual operation of remote controls or fixed temperature sensors, or on timers, average indoor temperature, or remote control via mobile apps.

[0003] While some air conditioners support voice control, their interaction modes are typically simple commands, such as "set to 26 degrees" or "turn on the air conditioner." When users issue complex commands involving specific areas, the system cannot accurately correlate the verbal expression with the actual location in the physical space. This prevents the system from accurately mapping user voice commands or behavioral intentions to spatial zones, limiting the air conditioner's control capabilities and resulting in low control efficiency and a poor user experience. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention are proposed to provide an air conditioning control method, an air conditioning control device, an electronic device, and a computer-readable storage medium that overcome or at least partially solve the above problems.

[0005] To address the above problems, a first aspect of the present invention provides an air conditioning control method, the method comprising: Acquire environmental images of a preset space and the user's movement trajectory within that preset space; The environmental image is subjected to visual semantic segmentation to obtain the segmentation result; Based on the segmentation results and the motion trajectory, a semantic partition map of the preset space is determined; the semantic partition map includes at least one functional region. Receive user voice commands, analyze the control intent and target region in the voice commands, and match the target region with the functional regions in the semantic partitioning map; A control strategy for the air conditioner is generated based on the control intent and the matched functional area, and the air conditioner is controlled according to the control strategy.

[0006] Optionally, the method further includes: If the target region is not obtained from the voice command, or if the target region does not match the functional region in the semantic partitioning map, then the control strategy for the air conditioner is generated based on the user's position reflected by the motion trajectory and the control intent.

[0007] Optionally, performing visual semantic segmentation on the environmental image to obtain the segmentation result includes: The environmental image is identified to obtain the layout information of the objects in the environmental image and the preset space, and semantic labels are marked for each object; Based on the semantic tags of the objects and the layout information of the preset space, the preset space is divided into regions to obtain the segmentation result; the segmentation result includes multiple regions.

[0008] Optionally, determining the semantic partitioning map of the preset space based on the segmentation result and the motion trajectory includes: Generate initial functional semantic labels and confidence scores for each region; the confidence scores are used to characterize the accuracy of the initial semantic labels. If the confidence score is higher than or equal to the preset threshold score, then the initial functional semantic label is determined as the functional semantic label corresponding to the region; If the confidence score is lower than the preset threshold score, the region is corrected according to the motion trajectory; Based on the determined functional semantic label region and the corrected region, a semantic partition map of the preset space is determined.

[0009] Optionally, correcting the region based on the motion trajectory includes: The motion trajectory is mapped onto the semantic partitioning map through coordinate projection to generate a trajectory partitioning map; the trajectory partitioning map is a coordinate map with the air conditioner as the origin and includes the motion trajectory and functional areas. The region is corrected based on the trajectory partitioning map.

[0010] Optionally, the step of correcting the region according to the trajectory partitioning map includes: At least one candidate label is generated for the region; The overlap of a user's trajectory in a given area is determined based on the trajectory partitioning map; the trajectory overlap is the proportion of the user's dwell time in that area to the preset time. The candidate labels, the trajectory overlap, and the layout information of the preset space are input into a preset functional semantic label inference model, and the functional semantic labels of the region are output.

[0011] Optionally, receiving the user's voice command, analyzing the control intent and target region in the voice command, and matching the target region with the functional regions in the semantic partitioning map includes: The voice command is converted into a text command, and the text command is analyzed to identify the semantic tags and control intent of the target region in the voice command; The semantic labels of the target region are semantically matched with the functional semantic labels of the functional regions in the semantic partitioning map.

[0012] Optionally, the method further includes: Receive user instructions to redefine a specified functional area; Replace the original functional semantic labels of the specified functional area with the redefined functional semantic labels; When the recovery conditions are met, the functional semantic labels of the specified functional area are restored to the original semantic labels; the recovery conditions include at least one of the following: the duration of replacing the functional semantic labels exceeds a preset duration, the user leaves the specified functional area, and the user's recovery instruction is received.

[0013] According to a second aspect of the present invention, an air conditioning control device is provided, the device comprising: The image trajectory acquisition module is used to acquire environmental images of a preset space and the user's movement trajectory in the preset space; The image semantic segmentation module is used to perform visual semantic segmentation on the environmental image to obtain the segmentation result; A semantic partitioning map determination module is used to determine a semantic partitioning map of the preset space based on the segmentation result and the motion trajectory; the semantic partitioning map includes at least one functional region; The voice command analysis module is used to receive the user's voice commands, analyze the control intent and target area in the voice commands, and match the target area with the functional area in the semantic partitioning map. The control strategy generation module is used to generate a control strategy for the air conditioner based on the control intent and the matched functional area, and to control the air conditioner according to the control strategy.

[0014] Optionally, the device further includes: An air conditioning control strategy generation module is used to generate an air conditioning control strategy based on the user's position reflected by the motion trajectory and the control intent if the target area is not obtained from the voice command, or if the target area does not match the functional area in the semantic partitioning map.

[0015] Optionally, the image semantic segmentation module includes: The object labeling submodule is used to identify the environmental image, obtain the layout information of the objects in the environmental image and the preset space, and label each object with a semantic label. The region segmentation submodule is used to segment the preset space into regions based on the semantic tags of the objects and the layout information of the preset space, so as to obtain the segmentation result; the segmentation result includes multiple regions.

[0016] Optionally, the semantic partitioning graph determination module includes: The label score generation submodule is used to generate initial functional semantic labels and confidence scores for each region; the confidence scores are used to characterize the accuracy of the initial semantic labels. The region label determination submodule is used to determine the initial functional semantic label as the functional semantic label corresponding to the region if the confidence score is higher than or equal to a preset threshold score. The region correction submodule is used to correct the region based on the motion trajectory if the confidence score is lower than the preset threshold score. The partition map determination submodule is used to determine the semantic partition map of the preset space based on the region where the functional semantic label is determined and the corrected region.

[0017] Optionally, the region correction submodule includes: The trajectory mapping unit is used to map the motion trajectory onto the semantic partitioning map through coordinate projection to generate a trajectory partitioning map; the trajectory partitioning map is a coordinate map containing the motion trajectory and functional areas with the air conditioner as the origin. A region correction unit is used to correct the region according to the trajectory partition map.

[0018] Optionally, the region correction unit includes: A candidate label generation subunit is used to generate at least one candidate label for the region; The trajectory overlap determination subunit is used to determine the trajectory overlap of the user in the area based on the trajectory partitioning map; the trajectory overlap is the proportion of the user's stay time in the area to the preset time. The functional label output subunit is used to input the candidate labels, the trajectory overlap, and the layout information of the preset space into the preset functional semantic label inference model, and output the functional semantic label of the region.

[0019] Optionally, the voice command analysis module includes: The instruction recognition submodule is used to convert the voice instruction into a text instruction and analyze the text instruction to identify the semantic tags and control intent of the target region in the voice instruction; The tag matching submodule is used to perform semantic matching between the semantic tags of the target region and the functional semantic tags of the functional regions in the semantic partitioning map.

[0020] Optionally, the device further includes: The instruction receiving module is used to receive user instructions to redefine specified functional areas; The label replacement module is used to replace the original functional semantic labels of the specified functional area with the redefined functional semantic labels; The tag recovery module is used to restore the functional semantic tags of the specified functional area to the original semantic tags when recovery conditions are met; the recovery conditions include at least one of the following: the duration of replacing the functional semantic tags exceeds a preset duration, the user leaves the specified functional area, and the user's recovery instruction is received.

[0021] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of the air conditioning control method as described in any of the preceding embodiments.

[0022] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the air conditioning control method as described in any of the preceding embodiments.

[0023] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses an air conditioning control method, device, electronic device, and storage medium. The method includes: acquiring an environmental image of a preset space and a user's movement trajectory within the preset space; performing visual semantic segmentation on the environmental image to obtain a segmentation result; determining a semantic partition map of the preset space based on the segmentation result and the movement trajectory; the semantic partition map includes at least one functional region; receiving a user's voice command, analyzing the control intent and target region in the voice command, and matching the target region with the functional region in the semantic partition map; generating an air conditioning control strategy based on the control intent and the matched functional region, and controlling the air conditioning according to the control strategy. By performing visual semantic segmentation on the environmental image and generating a semantic partition map based on the segmentation result and the movement trajectory, different functional regions can be identified. By analyzing the control intent and target region in the command and combining the two, an accurate control strategy can be generated, allowing the air conditioning to adjust for different functional regions, thereby improving the control efficiency of the air conditioning. Attached Figure Description

[0024] Figure 1 This is a flowchart of the steps of an air conditioning control method provided in an embodiment of the present invention; Figure 2 This is a flowchart of another air conditioning control method provided in an embodiment of the present invention; Figure 3 This is a semantic partitioning diagram of an air conditioning control method provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of a visual interface for an air conditioning control method provided in an embodiment of the present invention; Figure 5 This is a logic block diagram of an air conditioning control method provided in an embodiment of the present invention; Figure 6 This is a structural block diagram of an air conditioning control device provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] Traditional air conditioning interaction modes typically involve simple commands. When users issue complex commands involving specific areas, the system cannot accurately correlate verbal descriptions with actual physical locations. This prevents the system from accurately mapping user voice commands or behavioral intentions to spatial zones, limiting the air conditioner's control capabilities and resulting in low control efficiency and a poor user experience.

[0027] One of the core concepts of this invention is that by generating a semantic partitioning map, different functional areas can be identified. By analyzing the control intent and target area in the instructions and combining the two, an accurate control strategy can be generated, enabling the air conditioner to adjust for different functional areas and improving the control efficiency of the air conditioner.

[0028] Reference Figure 1 The diagram illustrates a flowchart of an air conditioning control method according to an embodiment of the present invention. The method specifically includes the following steps: Step 101: Obtain an environmental image of the preset space and the user's movement trajectory in the preset space; The preset space refers to any area that requires refined air conditioning control, as served by the embodiments of the present invention. In the embodiments of the present invention, the preset space can be a single room or a specific location, and the embodiments of the present invention do not limit this.

[0029] The environmental image of the preset space is acquired visual data that reflects the layout, furnishings, and semantic information of objects within the preset space. The user's movement trajectory within the preset space is spatiotemporal path data formed by continuously acquiring and recording the user's physical position changes within the preset space using millimeter-wave radar sensors.

[0030] The air conditioner is equipped with a high-precision millimeter-wave radar sensor that communicates with the air conditioner controller via a serial port, transmitting millimeter-wave data to the main controller. The main controller analyzes and processes the millimeter-wave data to identify and determine the presence, location, and movement trajectory of a human body. The millimeter-wave radar sensor monitors the status of human targets in the room in real time, detecting their presence, location, distance, angle, and movement trajectory.

[0031] The environmental image of the preset space is a panoramic or multi-view RGB image uploaded by the user. To protect privacy, faces / identifiable information can be blurred, or only an indoor panoramic view can be uploaded to remove details outside the windows. If the room layout changes, the user needs to re-upload the panoramic image. If no image is available, millimeter-wave radar detection is used. Existing indoor image acquisition usually relies on real-time detection and image recognition by cameras. However, camera-based visual partitioning is affected by field of view and occlusion, raising privacy concerns. This embodiment of the invention, by having the user upload environmental images, better protects user privacy.

[0032] In this embodiment of the invention, during initialization, an environmental image of the preset space is obtained by using an indoor panoramic view or photo uploaded by the user, thereby obtaining the static semantic layout of the preset space. Simultaneously, a millimeter-wave radar installed on the air conditioner continuously detects the user's real-time location and movement path within the room, forming a dynamic position sequence, thereby obtaining the user's movement trajectory within the preset space.

[0033] When acquiring a user's movement trajectory, interference sources are filtered out to distinguish between human reflections and dynamic interference (such as swaying curtains). The algorithm can filter out interference sources of fixed frequencies, and users can also manually label the area as an interference source. Users can label a specific interference area in a preset space using their mobile phones, or they can label it via voice interaction, such as directly labeling the location of the curtains as an interference source.

[0034] Step 102: Perform visual semantic segmentation on the environmental image to obtain the segmentation result; Visual semantic segmentation involves pixel-level analysis of user-uploaded images of pre-defined spatial environments, identifying and delineating every important object (such as a sofa, dining table, or bed) and structure (such as walls, doors, and windows) in the image, and labeling each identified object region with its semantic tag. Through visual semantic segmentation, a raw RGB image can be transformed into a map that marks different functional areas such as "sofa area," "sleeping area," and "dining area."

[0035] A multimodal visual understanding model is used to identify images, outputting instance masks (sofa, dining table, bed, door, window, desk, chair, etc.) and confidence scores. Each instance includes a bounding box, center point, and semantic label. Room boundaries, wall lines, main passageways, etc., are identified using a layout estimation model, outputting room topology (such as wall and floor boundaries) and door and window positions.

[0036] In this embodiment of the invention, a visual model trained on large-scale data and deployed in the cloud is used to perform deep analysis on user-uploaded indoor environment images. The identified object instances and their precise outlines (instance segmentation) are used to infer the overall room layout (such as walls, doors, and windows). This result provides a basis for the next step of generating a semantic partitioning map with semantic labels such as "sofa area" and "dining area".

[0037] Step 103: Based on the segmentation result and the motion trajectory, determine the semantic partition map of the preset space; the semantic partition map includes at least one functional region; Semantic partitioning maps are maps that integrate visually recognized objects with spatial layouts and assign them clear functional attributes. The data source is the segmentation results, namely the identified object outlines and labels. Based on the spatial location and relationships of these objects, as well as the room's layout and structure (such as walls), spatial reasoning and common sense are used to aggregate scattered objects into meaningful "functional areas." For example, if the identified "sofa," "coffee table," and "TV cabinet" are spatially adjacent, the system will aggregate them and their surrounding area, labeling them as a "living room relaxation area."

[0038] In this embodiment of the invention, an initial functional area sketch with confidence assessment is generated based on the "segmentation result," i.e., the visually recognized objects and layout. Subsequently, verification and correction are performed based on "motion trajectory" data: the user's real-time and historical motion trajectories are superimposed onto the initial functional area sketch through coordinate mapping, and the distribution of the trajectories is analyzed. For example, a user's long-term stillness in a certain area indicates a rest area, while frequent passage indicates a passageway. The functional labels or boundaries are iteratively corrected based on the trajectory evidence.

[0039] The final generated semantic partition map is a dynamic map containing at least one functional area after multimodal information fusion optimization. It is the foundation for achieving precise air conditioning control of the area, ensuring that when a user says "Don't blow air on the sofa," the system accurately understands and controls the "sofa area" that the user is actually using.

[0040] Step 104: Receive the user's voice command, analyze the control intent and target region in the voice command, and match the target region with the functional region in the semantic partitioning map; The system receives user voice commands and converts the user's voice into text through automatic speech recognition. A multimodal large language model performs semantic parsing on the text to identify target objects / regions (such as "sofa" and "dining table") and action intentions ("don't blow" and "warm it up").

[0041] In this embodiment of the invention, the process involves converting the user's natural language instructions into executable spatial control commands. Specifically, upon receiving a user's voice instruction (such as "I'm a bit hot, turn up the airflow over the sofa"), it is first converted into text using Automatic Speech Recognition (ASR), and then deep semantic analysis is performed on the text using Large Language Modeling (LLM) technology. The process includes: identifying the control intent: parsing the operation the user wishes to perform, such as "turn up the airflow" (increase the wind speed) or "don't blow" (avoid the wind). Identifying the target area: extracting the spatial target involved in the instruction, such as "over the sofa".

[0042] The "target region" extracted from the instruction (such as "over the sofa") is semantically associated with a functional region in the semantic partitioning map (such as a polygonal region labeled "sofa area"). Through analysis and matching, a vague "over the sofa" is precisely mapped to a "sofa area" with a defined coordinate range in the semantic partitioning map.

[0043] Step 105: Generate a control strategy for the air conditioner based on the control intent and the matched functional area, and control the air conditioner according to the control strategy.

[0044] The air conditioner features a controllable fan for changing airflow direction, variable speed / volume drive, and multi-channel air duct control. Its top and bottom airflow ducts ensure that every area of ​​the room receives airflow.

[0045] In this embodiment of the invention, this step involves generating a refined control strategy that can drive the air conditioning hardware based on the parsed control intent and the matched functional areas, and ensuring its reliable execution.

[0046] Generating control strategies requires integrating various information for decision-making. Its core instructions are the control intent (e.g., "avoid drafts") and the target functional area (e.g., "sofa area"). Environmental context includes whether other areas are occupied, current wind speed, and temperature settings. Equipment capabilities include which independently controllable air outlets the air conditioner has (e.g., top outlet, bottom outlet, left / right swing).

[0047] Based on the above factors, the system will generate a specific and quantifiable control strategy. For example, for the instruction "Do not blow air onto the sofa area," the strategy may not be to simply turn off all fans, but to close the air outlet directly facing the 'sofa area,' while raising the angle of another air outlet by 15 degrees and increasing the wind speed of the main air duct in the living room by 10% to compensate for the air volume.

[0048] The execution control strategy translates commands into physical actions, driving the corresponding stepper motors to adjust the angle of the air guide vanes, regulate the fan speed, etc., ultimately achieving directional air delivery or precise wind avoidance, thus completing a full closed-loop control from user voice to air conditioning feel.

[0049] Reference Figure 2 The diagram illustrates a flowchart of another air conditioning control method provided by an embodiment of the present invention, the method specifically including the following steps: Step 201: Obtain an environmental image of the preset space and the user's movement trajectory in the preset space; In this embodiment of the invention, during initialization, an environmental image of the preset space is obtained by using an indoor panoramic view or photo uploaded by the user, thereby obtaining the static semantic layout of the preset space. Simultaneously, a millimeter-wave radar installed on the air conditioner continuously detects the user's real-time location and movement path within the room, forming a dynamic position sequence, thereby obtaining the user's movement trajectory within the preset space.

[0050] When acquiring a user's movement trajectory, interference sources are filtered out to distinguish between human reflections and dynamic interference (such as swaying curtains). The algorithm can filter out interference sources of fixed frequencies, and users can also manually label the area as an interference source. Users can label a specific interference area in a preset space using their mobile phones, or they can label it via voice interaction, such as directly labeling the location of the curtains as an interference source.

[0051] Step 202: Perform visual semantic segmentation on the environmental image to obtain the segmentation result; In this embodiment of the invention, a visual model trained on large-scale data and deployed in the cloud is used to perform deep analysis on user-uploaded indoor environment images. The identified object instances and their precise outlines (instance segmentation) are used to infer the overall room layout (such as walls, doors, and windows). This result provides a basis for the next step of generating a semantic partitioning map with semantic labels such as "sofa area" and "dining area".

[0052] In some embodiments, step 202 may include the following sub-steps: Sub-step S11: Recognize the environment image to obtain the layout information of the objects in the environment image and the preset space, and label each object with a semantic tag; In this embodiment of the invention, a visual model is used to perform deep analysis on the environmental images uploaded by the user, transforming the pixel array into structured data containing semantics. The instance segmentation model not only identifies all the main, independent object instances in the image, such as "Sofa-1", "Dining Table-1", "Chair-1", and "Chair-2", but also accurately delineates the outline (boundary box or mask) of each object and assigns it an accurate semantic label (such as "Sofa", "Dining Table", and "Chair").

[0053] Simultaneously running a room layout estimation model, the system can infer the boundaries and spatial relationships of walls, doors, windows, ceilings, and floors from the image, thereby reconstructing the room's three-dimensional topology. Ultimately, it obtains the semantic labels, precise locations, and outlines of all identified objects, as well as the overall spatial layout skeleton.

[0054] Sub-step S12: Based on the semantic tags of the object and the layout information of the preset space, the preset space is divided into regions to obtain the segmentation result; the segmentation result includes multiple regions.

[0055] In this embodiment of the invention, based on identified objects and layouts, the entire preset space is aggregated and divided into several areas. By comprehensively analyzing the semantic tags of the objects and the layout information of the preset space, and based on this information, and according to common sense and spatial relationships, functionally related and geographically adjacent objects are grouped together to form a functional unit. For example, closely adjacent "sofa," "coffee table," and "TV cabinet" are aggregated and initially defined as a "living room rest area." The physical boundaries of each area are determined using layout information (such as walls) and the spatial relationships between objects. Walls are the most direct partition boundaries, while in open spaces (such as an integrated living and dining area), the system may define virtual boundaries between the "living room rest area" and the "dining area" based on the relative distances between object clusters and passageway spaces.

[0056] Finally, the output "segmentation result" is a preliminary spatial planning map consisting of multiple regions. Each region is a polygonal area with an initial functional label inferred based on its core object (such as "rest area", "dining area", "passage").

[0057] By identifying objects (such as "sofa") and understanding their attributes through semantic labeling, combined with layout information, functional areas such as "rest area" and "corridor" can be more accurately divided, rather than simply segmenting pixels. The fusion of object semantics and spatial layout segmentation results ensures that subsequent steps in generating semantic partition maps are both accurate and reasonable in dividing functional areas, greatly improving the accuracy and reliability of the entire system's decision-making.

[0058] Step 203: Based on the segmentation result and the motion trajectory, determine the semantic partition map of the preset space; the semantic partition map includes at least one functional region; In this embodiment of the invention, an initial functional area sketch with confidence assessment is generated based on the "segmentation result," i.e., the visually recognized objects and layout. Subsequently, verification and correction are performed based on "motion trajectory" data: the user's real-time and historical motion trajectories are superimposed onto the initial functional area sketch through coordinate mapping, and the distribution of the trajectories is analyzed. For example, a user's long-term stillness in a certain area indicates a rest area, while frequent passage indicates a passageway. The functional labels or boundaries are iteratively corrected based on the trajectory evidence.

[0059] The final generated semantic partition map is a dynamic map containing at least one functional area after multimodal information fusion optimization. It is the foundation for achieving precise air conditioning control of the area, ensuring that when a user says "Don't blow air on the sofa," the system accurately understands and controls the "sofa area" that the user is actually using.

[0060] In some embodiments, step 203 may include the following sub-steps: Sub-step S21: Generate initial functional semantic labels and confidence scores for each region; the confidence scores are used to characterize the accuracy of the initial semantic labels. Initial functional semantic labels refer to the preliminary, untested category labels assigned when a region is functionally interpreted solely based on visual information for the first time. By analyzing all identified objects within a region and identifying the most functionally indicative core objects, the primary purpose of that region can be inferred.

[0061] The confidence score is a quantified probability value used to characterize the credibility or probability of correctness of the initial functional semantic labels assigned by the system to a certain region.

[0062] In this embodiment of the invention, based on the semantic labels of core objects aggregated within the area (such as an area containing "bed" and "bedside table"), and combined with common sense about spatial layout, a generalized initial functional semantic label (such as "sleeping area") is assigned to the area. Simultaneously, a confidence score is calculated for each initial label. This score is a quantified probability value used to characterize the likelihood that the functional label is correct. The confidence assessment is based on multiple factors: visual confidence in the identification of core objects within the area; contextual plausibility; for example, if a "toilet" is also identified in an area labeled "office area," its confidence will decrease significantly. This confidence score will clearly indicate which areas are definitively reliable and which are ambiguous and questionable, requiring further verification and correction based on user movement trajectories.

[0063] Sub-step S22: If the confidence score is higher than or equal to a preset threshold score, then the initial functional semantic label is determined as the functional semantic label corresponding to the region; The preset threshold score is a pre-set baseline score that serves as the decision boundary for determining whether to trust the initial functional semantic labels generated by the visual model.

[0064] When the confidence score of a region is greater than or equal to a preset threshold score, it indicates that the visual recognition result for that region is of high quality and highly reliable. It will be considered a "sure region," and its initial functional semantic label will be directly determined as the final label without subsequent correction. When the confidence score of a region is less than the preset threshold score, it indicates that the recognition result for that region has ambiguity, uncertainty, or conflict. It will be marked as a "region to be verified," and iterative correction will be performed using dynamic data such as the user's motion trajectory.

[0065] In this embodiment of the invention, after the confidence score of a certain region is calculated, it is compared with a preset threshold score.

[0066] If the confidence score is higher than or equal to this threshold, it means that the objects in the region are clearly and with high probability identified, and these objects are highly consistent in function and spatial relationship, conforming to common sense. No additional verification is needed; visual information alone is sufficient to confidently identify the function of the region, and no correction is required.

[0067] Therefore, the previous inference will be directly adopted, and this initial functional semantic label (such as "sleep zone") will be formally determined as the final "functional semantic label" for that area. This determined label will directly enter the final "semantic partitioning map" and become a reliable basis for responding to user commands.

[0068] Sub-step S23: If the confidence score is lower than the preset threshold score, then the region is corrected according to the motion trajectory; In this embodiment of the invention, when the system determines that the confidence score of a certain area is lower than a preset threshold, it means that relying solely on static visual information is no longer sufficient to reliably determine the function of that area, and multimodal verification and correction of that area are required. For example, if a visually identified area is a "desk area" with low confidence, but trajectory data shows that the user remains stationary there for extended periods each night, the system will correct its function label to "sleep area".

[0069] In some embodiments, step S23 may include the following sub-steps: Sub-step S231: The motion trajectory is mapped onto the semantic partitioning map through coordinate projection to generate a trajectory partitioning map; the trajectory partitioning map is a coordinate map with the air conditioner as the origin and includes the motion trajectory and functional areas; Sub-step S232: Correct the region according to the trajectory partition map.

[0070] In this embodiment of the invention, a coordinate transformation model is established to convert the user trajectory detected by millimeter-wave radar into corresponding coordinates, and the coordinates of the motion trajectory are mapped onto the semantic partitioning map in a coordinate system with the air conditioner as the origin. After the transformation is completed, each physical location point of the user is mapped to a pixel position in the semantic partitioning map, resulting in a trajectory partitioning map that includes the user's motion trajectory and functional areas. This not only clearly shows "where the sofa area is", but also intuitively shows "where the user is actually active".

[0071] By analyzing all trajectory data mapped to the interior and boundaries of low-confidence regions, and using actual user trajectory data, the visual inference was validated. This allows for direct quantitative correlation between user location information and spatial functional area information, significantly improving the accuracy of the semantic partitioning map. Based on the collected user trajectory data, the partitioning map can be continuously fine-tuned and optimized to adapt to user habits.

[0072] In some embodiments, step S232 may include the following sub-steps: Sub-step S2321: Generate at least one candidate label for the region; Sub-step S2322: Determine the overlap of the user's trajectory in the area based on the trajectory partition map; the trajectory overlap is the proportion of the user's stay time in the area to the preset time. Sub-step S2323: Input the candidate label, the trajectory overlap degree and the layout information of the preset space into the preset functional semantic label inference model, and output the functional semantic label of the region.

[0073] In this embodiment of the invention, one or more possible functional labels are generated for the area based on various information sources. These information sources include: when the visual model identifies the object, it may provide other possibilities second only to the initial label (for example, if the confidence level for identifying an object as a "sofa" is 0.5 and the confidence level for identifying it as a "bench" is 0.4, then both will enter the candidate list). The relationship between this area and surrounding high-confidence areas is analyzed; for example, an undefined area adjacent to a "dining table" is likely to be nominated as a "dining chair area" or "passage." If the system has records of users previously making temporary definitions in this area or similar areas, these historical labels will also become strong candidates.

[0074] On the trajectory partition map, for the area to be corrected, within a preset observation time window (such as the past 30 minutes), the total time that the user's movement speed within the boundary of that area is below a certain threshold (determined as a stop) is accumulated. The stop time / total observation time is calculated to obtain the trajectory overlap. For example, a trajectory overlap of 0.7 for an area strongly indicates that the area is a major activity or rest point; while an area with an overlap of only 0.05 is likely an occasional passage.

[0075] The candidate labels, trajectory overlap, and layout information of the preset space are all input into the functional semantic label reasoning model for multi-round comprehensive reasoning. For example, "This area may visually be a 'desk' or 'cabinet' (candidate label), but the user has a trajectory overlap of up to 80% and is stationary here (trajectory overlap), and it is located in the corner of the bedroom (layout information). Based on common sense, a person who is stationary in the corner of the bedroom for a long time is most likely sleeping. Therefore, it is more reasonable to finally determine this area as a 'sleeping area'." Finally, the functional semantic label is output to complete the correction of the area.

[0076] Through cross-validation of multi-dimensional information, the final generated functional semantic tags improve the reliability and practicality of functional partitions, providing a solid foundation for subsequent voice control matching and ensuring that commands are correctly understood and executed.

[0077] Sub-step S24: Based on the region where the functional semantic label is determined and the corrected region, determine the semantic partition map of the preset space.

[0078] In this embodiment of the invention, the aggregated regional information includes regions directly adopted due to their high visual confidence and regions initially with low confidence but which have been verified and corrected using trajectory data and assigned new functional semantic labels. All regions with final functional semantic labels are combined with the room's layout information (walls, doors, and windows) to determine a semantic zoning map of the preset space. Using this determined semantic zoning map as a spatial reference for responding to subsequent user voice commands and executing refined air conditioning control improves the control efficiency of the air conditioning. Determining the semantic zoning map based on the regions with the determined functional semantic labels and the corrected regions enhances adaptability to complex and non-standard layouts.

[0079] Reference Figure 3 This illustrates a semantic partitioning diagram of an air conditioning control method provided by an embodiment of the present invention, based on... Figure 3 The semantic partitioning map shown is generated by identifying objects in an environmental image and generating corresponding semantic labels. Visual semantic segmentation is then performed on the environmental image. After region segmentation, a coordinate map is generated with the air conditioner as the origin. The motion trajectory detected by millimeter-wave radar is output as coordinates and mapped onto the coordinate map. The coordinates are then iteratively corrected to obtain the final semantic partitioning map.

[0080] Step 204: Receive the user's voice command, analyze the control intent and target region in the voice command, and match the target region with the functional region in the semantic partitioning map; In this embodiment of the invention, the process involves converting the user's natural language instructions into executable spatial control commands. Specifically, upon receiving a user's voice instruction (such as "I'm a bit hot, turn up the airflow over the sofa"), it is first converted into text using Automatic Speech Recognition (ASR), and then deep semantic analysis is performed on the text using Large Language Modeling (LLM) technology. The process includes: identifying the control intent: parsing the operation the user wishes to perform, such as "turn up the airflow" (increase the wind speed) or "don't blow" (avoid the wind). Identifying the target area: extracting the spatial target involved in the instruction, such as "over the sofa".

[0081] The "target region" extracted from the instruction (such as "over the sofa") is semantically associated with a functional region in the semantic partitioning map (such as a polygonal region labeled "sofa area"). Through analysis and matching, a vague "over the sofa" is precisely mapped to a "sofa area" with a defined coordinate range in the semantic partitioning map.

[0082] In some embodiments, step 204 may include the following sub-steps: Sub-step S31: Convert the voice command into a text command and analyze the text command to identify the semantic tags and control intent of the target region in the voice command; In this embodiment of the invention, automatic speech recognition (ASR) technology is used to convert the audio signal emitted by the user into accurate text information. The converted text instructions are then deeply analyzed to extract keywords representing spatial entities and the desired user action. For example, the noun "sofa" is extracted from the instruction, and the intention to "avoid the wind" is extracted from "don't...blow in the wind." The final output is a clear, structured instruction: {Control Intent: Avoid the wind, Target Area Semantic Label: Sofa}.

[0083] Sub-step S32 involves semantically matching the semantic labels of the target region with the functional semantic labels of the functional regions in the semantic partitioning map.

[0084] In this embodiment of the invention, the semantic label of the target region parsed from the voice command is compared with the functional semantic labels of all defined functional regions in the semantic partitioning map. For example, the word "sofa" in the command directly matches the "sofa area" in the map, ultimately outputting a clear functional region object. Traditional voice control is limited to executing simple, predefined commands, while this embodiment of the invention, by understanding the "target region" and "control intent," achieves context-based environmental interaction, ensuring the accurate execution of control commands.

[0085] Reference Figure 4 The diagram shows a visual interface schematic of an air conditioning control method provided by an embodiment of the present invention. Figure 4 The display shows the user-operable visual interface of the app, which includes textures of multiple objects and schematic diagrams of preset spaces. Users can annotate any area and drag textures to combine functional areas.

[0086] Step 205: Generate a control strategy for the air conditioner based on the control intent and the matched functional area, and control the air conditioner according to the control strategy.

[0087] In this embodiment of the invention, this step involves generating a refined control strategy that can drive the air conditioning hardware based on the parsed control intent and the matched functional areas, and ensuring its reliable execution.

[0088] Generating control strategies requires integrating various information for decision-making. Its core instructions are the control intent (e.g., "avoid drafts") and the target functional area (e.g., "sofa area"). Environmental context includes whether other areas are occupied, current wind speed, and temperature settings. Equipment capabilities include which independently controllable air outlets the air conditioner has (e.g., top outlet, bottom outlet, left / right swing).

[0089] Based on the above factors, the system will generate a specific and quantifiable control strategy. For example, for the instruction "Do not blow air onto the sofa area," the strategy may not be to simply turn off all fans, but to close the air outlet directly facing the 'sofa area,' while raising the angle of another air outlet by 15 degrees and increasing the wind speed of the main air duct in the living room by 10% to compensate for the air volume.

[0090] The execution control strategy translates commands into physical actions, driving the corresponding stepper motors to adjust the angle of the air guide vanes, regulate the fan speed, etc., ultimately achieving directional air delivery or precise wind avoidance, thus completing a full closed-loop control from user voice to air conditioning feel.

[0091] Step 206: If the target region is not obtained from the voice command, or the target region does not match the functional region in the semantic partitioning map, then the control strategy of the air conditioner is generated based on the user position reflected by the motion trajectory and the control intention.

[0092] In this embodiment of the invention, if a user command only contains a control intent but does not specify a specific area, and the target area mentioned in the command cannot be found in the existing semantic partitioning graph, the user's current position reflected by the motion trajectory will be queried, and the functional area where that position is located will be defaulted as the "implicit target area" of this command. The control intent is bound to this implicit target area to generate a control strategy. The user's operation intent is specific to their own environment, thereby transforming ambiguous global commands into precise control of specific areas, greatly improving the naturalness of the interaction and the practicality of the system.

[0093] In some embodiments, the method may further include the following steps: Receive a user's instruction to redefine a specified functional area; replace the original functional semantic label of the specified functional area with the redefined functional semantic label; when a recovery condition is met, restore the functional semantic label of the specified functional area to the original semantic label; the recovery condition includes at least one of the following: the duration of replacing the functional semantic label exceeds a preset duration, the user leaves the specified functional area, and the user's recovery instruction is received.

[0094] In this embodiment of the invention, users are allowed to redefine spatial functions according to temporary needs, and automatically restore them when conditions are met. Users can issue redefinition commands for any specified functional area via voice or an app. Upon receiving the command, the underlying semantic partitioning map is not modified; instead, a temporary overlay layer is created, replacing the original functional semantic labels of that area with the redefined functional semantic labels.

[0095] Multiple recovery conditions are preset. Once any condition is met, the system will automatically remove the temporary overlay and restore the functional semantic labels of the area to their original semantic labels. These conditions include: the replacement exceeding a preset time; the user leaving the designated functional area; and receiving a recovery command from the user. This function greatly enhances the system's usability and user experience, enabling air conditioning control to flexibly adapt to short-term scenario changes (such as temporary office work, gym sessions, or parties), while ensuring the accuracy and stability of long-term control through the automatic recovery mechanism, avoiding long-term chaos caused by temporary modifications.

[0096] Reference Figure 5 This diagram illustrates a logic block diagram of an air conditioning control method provided by an embodiment of the present invention, based on... Figure 5The diagram shown is a logical block diagram of an embodiment of the present invention. A semantic partition map is generated by the environmental image, and the user's location and movement trajectory are continuously acquired. After the user trajectory and semantic partition are fused by coordinate mapping, the system can parse the control intent and target area in the voice command, generate a precise air conditioning control strategy, and finally convert it into specific wind direction, wind speed and air duct control commands to drive the execution unit to complete personalized air supply or wind avoidance operations.

[0097] This invention discloses an air conditioning control method, device, electronic device, and storage medium. The method includes: acquiring an environmental image of a preset space and a user's movement trajectory within the preset space; performing visual semantic segmentation on the environmental image to obtain a segmentation result; determining a semantic partition map of the preset space based on the segmentation result and the movement trajectory; the semantic partition map includes at least one functional region; receiving a user's voice command, analyzing the control intent and target region in the voice command, and matching the target region with the functional region in the semantic partition map; generating an air conditioning control strategy based on the control intent and the matched functional region, and controlling the air conditioning according to the control strategy. By performing visual semantic segmentation on the environmental image and generating a semantic partition map based on the segmentation result and the movement trajectory, different functional regions can be identified. By analyzing the control intent and target region in the command and combining the two, an accurate control strategy can be generated, allowing the air conditioning to adjust for different functional regions, thereby improving the control efficiency of the air conditioning.

[0098] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0099] Reference Figure 6 The diagram shows a structural block diagram of an air conditioning control device according to an embodiment of the present invention. The device specifically includes the following modules: The image trajectory acquisition module 301 is used to acquire an environmental image of a preset space and the user's motion trajectory in the preset space. Image semantic segmentation module 302 is used to perform visual semantic segmentation on the environmental image to obtain segmentation results; The semantic partitioning map determination module 303 is used to determine the semantic partitioning map of the preset space based on the segmentation result and the motion trajectory; the semantic partitioning map includes at least one functional region; The voice command analysis module 304 is used to receive the user's voice command, analyze the control intent and target area in the voice command, and match the target area with the functional area in the semantic partitioning map. The control strategy generation module 305 is used to generate a control strategy for the air conditioner based on the control intent and the matched functional area, and to control the air conditioner according to the control strategy.

[0100] In some embodiments, the apparatus further includes: An air conditioning control strategy generation module is used to generate an air conditioning control strategy based on the user's position reflected by the motion trajectory and the control intent if the target area is not obtained from the voice command, or if the target area does not match the functional area in the semantic partitioning map.

[0101] In some embodiments, the image semantic segmentation module 302 includes: The object labeling submodule is used to identify the environmental image, obtain the layout information of the objects in the environmental image and the preset space, and label each object with a semantic label. The region segmentation submodule is used to segment the preset space into regions based on the semantic tags of the objects and the layout information of the preset space, so as to obtain the segmentation result; the segmentation result includes multiple regions.

[0102] In some embodiments, the semantic partitioning graph determination module 303 includes: The label score generation submodule is used to generate initial functional semantic labels and confidence scores for each region; the confidence scores are used to characterize the accuracy of the initial semantic labels. The region label determination submodule is used to determine the initial functional semantic label as the functional semantic label corresponding to the region if the confidence score is higher than or equal to a preset threshold score. The region correction submodule is used to correct the region based on the motion trajectory if the confidence score is lower than the preset threshold score. The partition map determination submodule is used to determine the semantic partition map of the preset space based on the region where the functional semantic label is determined and the corrected region.

[0103] In some embodiments, the region correction submodule includes: The trajectory mapping unit is used to map the motion trajectory onto the semantic partitioning map through coordinate projection to generate a trajectory partitioning map; the trajectory partitioning map is a coordinate map containing the motion trajectory and functional areas with the air conditioner as the origin. A region correction unit is used to correct the region according to the trajectory partition map.

[0104] In some embodiments, the region correction unit includes: A candidate label generation subunit is used to generate at least one candidate label for the region; The trajectory overlap determination subunit is used to determine the trajectory overlap of the user in the area based on the trajectory partitioning map; the trajectory overlap is the proportion of the user's stay time in the area to the preset time. The functional label output subunit is used to input the candidate labels, the trajectory overlap, and the layout information of the preset space into the preset functional semantic label inference model, and output the functional semantic label of the region.

[0105] In some embodiments, the voice command analysis module 304 includes: The instruction recognition submodule is used to convert the voice instruction into a text instruction and analyze the text instruction to identify the semantic tags and control intent of the target region in the voice instruction; The tag matching submodule is used to perform semantic matching between the semantic tags of the target region and the functional semantic tags of the functional regions in the semantic partitioning map.

[0106] In some embodiments, the apparatus further includes: The instruction receiving module is used to receive user instructions to redefine specified functional areas; The label replacement module is used to replace the original functional semantic labels of the specified functional area with the redefined functional semantic labels; The tag recovery module is used to restore the functional semantic tags of the specified functional area to the original semantic tags when recovery conditions are met; the recovery conditions include at least one of the following: the duration of replacing the functional semantic tags exceeds a preset duration, the user leaves the specified functional area, and the user's recovery instruction is received.

[0107] As the apparatus embodiment is basically similar to the method embodiment, it is described in a relatively simple manner. For relevant details, please refer to the description of the method embodiment.

[0108] This invention also provides an electronic device, including: a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the various processes of the above-described air conditioning control method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here.

[0109] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described air conditioning control method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here.

[0110] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0111] Furthermore, it should be noted that the scope of the methods and apparatus in the embodiments of the present invention is not limited to performing functions in the order shown or discussed. It may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0112] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0113] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. An air conditioning control method, characterized in that, The method includes: Acquire environmental images of a preset space and the user's movement trajectory within that preset space; The environmental image is subjected to visual semantic segmentation to obtain the segmentation result; Based on the segmentation results and the motion trajectory, a semantic partition map of the preset space is determined; the semantic partition map includes at least one functional region. Receive user voice commands, analyze the control intent and target region in the voice commands, and match the target region with the functional regions in the semantic partitioning map; A control strategy for the air conditioner is generated based on the control intent and the matched functional area, and the air conditioner is controlled according to the control strategy.

2. The air conditioning control method according to claim 1, characterized in that, The method further includes: If the target region is not obtained from the voice command, or if the target region does not match the functional region in the semantic partitioning map, then the control strategy for the air conditioner is generated based on the user's position reflected by the motion trajectory and the control intent.

3. The air conditioning control method according to claim 1, characterized in that, The step of performing visual semantic segmentation on the environmental image to obtain the segmentation result includes: The environmental image is identified to obtain the layout information of the objects in the environmental image and the preset space, and semantic labels are marked for each object; Based on the semantic tags of the objects and the layout information of the preset space, the preset space is divided into regions to obtain the segmentation result; the segmentation result includes multiple regions.

4. The air conditioning control method according to claim 1, characterized in that, Determining the semantic partition map of the preset space based on the segmentation result and the motion trajectory includes: Generate initial functional semantic labels and confidence scores for each region; the confidence scores are used to characterize the accuracy of the initial semantic labels. If the confidence score is higher than or equal to the preset threshold score, then the initial functional semantic label is determined as the functional semantic label corresponding to the region; If the confidence score is lower than the preset threshold score, the region is corrected according to the motion trajectory; Based on the determined functional semantic label region and the corrected region, a semantic partition map of the preset space is determined.

5. The air conditioning control method according to claim 4, characterized in that, The step of correcting the region based on the motion trajectory includes: The motion trajectory is mapped onto the semantic partitioning map through coordinate projection to generate a trajectory partitioning map; the trajectory partitioning map is a coordinate map containing the motion trajectory and functional areas with the air conditioner as the origin. The region is corrected based on the trajectory partitioning map.

6. The air conditioning control method according to claim 5, characterized in that, The step of correcting the region based on the trajectory partitioning map includes: At least one candidate label is generated for the region; The overlap of a user's trajectory in a given area is determined based on the trajectory partitioning map; the trajectory overlap is the proportion of the user's dwell time in that area to the preset time. The candidate labels, the trajectory overlap, and the layout information of the preset space are input into a preset functional semantic label inference model, and the functional semantic labels of the region are output.

7. The air conditioning control method according to claim 1, characterized in that, The process of receiving user voice commands, analyzing the control intent and target region in the voice commands, and matching the target region with functional regions in the semantic partitioning graph includes: The voice command is converted into a text command, and the text command is analyzed to identify the semantic tags and control intent of the target region in the voice command; The semantic labels of the target region are semantically matched with the functional semantic labels of the functional regions in the semantic partitioning map.

8. The air conditioning control method according to claim 1, characterized in that, The method further includes: Receive user instructions to redefine a specified functional area; Replace the original functional semantic labels of the specified functional area with the redefined functional semantic labels; When the recovery conditions are met, the functional semantic labels of the specified functional area are restored to the original semantic labels; the recovery conditions include at least one of the following: the duration of replacing the functional semantic labels exceeds a preset duration, the user leaves the specified functional area, and the user's recovery instruction is received.

9. An air conditioning control device, characterized in that, The device includes: The image trajectory acquisition module is used to acquire environmental images of a preset space and the user's movement trajectory in the preset space; The image semantic segmentation module is used to perform visual semantic segmentation on the environmental image to obtain the segmentation result; A semantic partitioning map determination module is used to determine a semantic partitioning map of the preset space based on the segmentation result and the motion trajectory; the semantic partitioning map includes at least one functional region; The voice command analysis module is used to receive the user's voice commands, analyze the control intent and target area in the voice commands, and match the target area with the functional area in the semantic partitioning map. The control strategy generation module is used to generate a control strategy for the air conditioner based on the control intent and the matched functional area, and to control the air conditioner according to the control strategy.

10. An electronic device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the air conditioning control method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, A program is stored on the computer-readable storage medium, which, when executed by a processor, implements the steps of the air conditioning control method as described in any one of claims 1-8.

Images