An open-source platform voice-activated control method and system for multi-ecosystem device access

Abstract

This invention provides an open-source platform voice-activated control method and system for multi-ecosystem device access, relating to the field of intelligent voice control technology. The method includes: receiving and standardizing capability information from different ecosystem devices to construct a device capability index; generating a voice intent template library based on user voice expressions and standardizing the parsing of voice intent requests to obtain target voice intent instances; retrieving and filtering target device capability entries based on the parsing results, constructing cross-ecosystem linkage control logic and orchestrating tasks; performing conflict detection under resource and policy constraints to form a securely executable sequence of target linkage control tasks; mapping standardized device operations to native instructions of each ecosystem and issuing them to the target devices; updating the control status through device feedback; and finally generating the voice-activated execution result. This invention achieves reliable linkage control of multiple ecosystem devices under a unified voice entry point, improving the accuracy and scalability of cross-ecosystem collaboration.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent voice control, and particularly to an open-source platform voice linkage control method and system for multi-ecosystem device access. Background Art

[0002] With the development of home Internet of Things and building intelligence, many device manufacturers have built relatively closed device ecosystem around their respective cloud platforms. Devices such as lighting, air conditioning, security, household appliances, and environmental monitoring are usually connected to the control platforms of their respective brands, and voice control is realized through supporting mobile terminal applications or smart speakers. Existing voice control methods mostly rely on the account systems and protocol stacks of single manufacturers. Within this ecosystem, device status queries, single-device control, and a small number of same-ecosystem scenario linkages can be completed. Some platforms also provide limited voice scenarios such as "turn on the light" and "adjust the temperature". However, when users deploy multiple intelligent devices of different brands in the same physical environment, there is a lack of unified ability description and control interfaces between different ecosystems. Voice control often requires switching between multiple applications and accounts, and the linkage ability is limited within their respective ecosystems.

[0003] To improve the user experience and reduce the access cost, device access platforms and neutral control platforms implemented in an open-source manner have gradually emerged in the industry. By docking various communication protocols and cloud interfaces through adapters or gateways, attempts are made to achieve unified device management at the platform layer. At the same time, natural language processing technology has been continuously progressing, and the voice recognition and semantic understanding capabilities have been significantly improved, enabling users to hope to uniformly control multi-brand devices through more natural language under the same interface and trigger complex scenario linkages across spaces and categories. The future trend of multi-ecosystem collaborative control is: with an open-source platform as the base, a unified ability model as the middle layer, and at the upper layer, device-level, subsystem-level, and even building-level linkage control is achieved through voice-intention-driven automated orchestration. It is required that the platform can dynamically discover new devices, adapt to new ecosystems, and automatically generate safe and controllable linkage strategies according to voice commands.

[0004] Existing multi-ecosystem control solutions generally have several technical deficiencies. First, many platforms only stay at the level of protocol bridging and simple instruction forwarding, lacking a unified ability description and semantic abstraction across ecosystems. As a result, the binding between voice intentions and specific device functions can only be roughly achieved through fixed rules or manual scripts, making it difficult to maintain stably as the ecosystem expands. Second, the existing open-source platforms mostly support single-device control or simple sequential scenarios for voice linkage, lacking a mechanism for modeling and task orchestration of linkage control diagrams for multi-device and multi-ecosystem scenarios, and unable to describe complex dependency relationships and execution timings under a unified model. Third, the existing solutions lack systematic conflict detection and priority management when controlling multiple ecosystems simultaneously. When different voice commands or automation rules generate conflicting control targets on the same device, they often overwrite the earlier ones with the later ones or simply reject them, easily resulting in unpredictable execution results and even potential safety hazards. Therefore, there is an urgent need for an open-source platform voice linkage control method for multi-ecosystem device access, which forms an integrated technical solution in terms of unified ability modeling, voice-intention-driven linkage orchestration, and cross-ecosystem conflict resolution and instruction mapping, to solve the problems of fragmentation, poor maintainability, and insufficient security in multi-ecosystem voice linkage control in the existing technology. Summary of the Invention

[0005] In order to overcome the deficiencies of the existing technology, the purpose of the present invention is to provide an open-source platform voice linkage control method and system for multi-ecosystem device access. The present invention solves the problems of lack of unified ability abstraction, lack of cross-ecosystem linkage orchestration, and inability to effectively resolve conflicts in multi-ecosystem device voice control in the existing technology.

[0006] To achieve the above purpose, the present invention provides the following solutions: An open-source platform voice linkage control method for multi-ecosystem device access, comprising: Receiving device ability information and status information of multiple target intelligent devices by using an open-source platform, and performing standardized modeling on the device ability information and status information based on a preset unified ability description specification to obtain a standardized device ability index set; Inducing and forming voice intention templates from common user voice expressions, and establishing an association relationship between each voice intention template and the ability entries in the standardized device ability index set to obtain a voice intention template library; Obtaining a voice intention request, and performing standardized parsing on the voice intention request according to the voice intention template library to obtain a target voice intention instance; Based on the target voice intention instance, selecting target device ability entries that meet the target control conditions in the standardized device ability index set to construct a cross-ecosystem linkage control logic; The linkage control logic is arranged into tasks according to a preset execution order rule to obtain an initial linkage control task sequence; Based on the initial linkage control task sequence, cross-ecological linkage conflict detection is performed according to preset resource constraints and strategy constraints to obtain the target linkage control task sequence after conflict resolution. Based on the target linkage control task sequence, standardized equipment operations are mapped to native control commands corresponding to each equipment ecosystem, resulting in a cross-ecosystem native command set. The native instruction set is sent to the corresponding target device to update the control logic state based on device feedback, thereby obtaining the voice linkage execution result of multiple ecosystem devices.

[0007] An open-source platform voice-activated control system for multi-ecosystem device access includes: The standardized device capability modeling module is used to receive device capability information and status information of multiple target intelligent devices using an open source platform, and to perform standardized modeling of the device capability information and status information based on a preset unified capability description specification to obtain a standardized device capability index set. The voice intent template construction module is used to summarize commonly used user voice expressions into voice intent templates, and to establish an association between each voice intent template and the capability entries in the standardized device capability index set to obtain a voice intent template library. The voice intent parsing module is used to obtain a voice intent request and perform standardized parsing of the voice intent request according to the voice intent template library to obtain a target voice intent instance; The cross-ecosystem linkage control logic construction module is used to select target device capability entries that meet the target control conditions from the standardized device capability index set based on the target voice intent instance, and construct cross-ecosystem linkage control logic. The linkage task orchestration module is used to orchestrate the linkage control logic according to a preset execution order rule to obtain an initial linkage control task sequence. The cross-ecological linkage conflict detection module is used to perform cross-ecological linkage conflict detection based on the initial linkage control task sequence and according to preset resource constraints and strategy constraints, so as to obtain the target linkage control task sequence after conflict resolution. The native instruction mapping module is used to map standardized device operations into native control instructions corresponding to each device ecosystem based on the target linkage control task sequence, thereby obtaining a cross-ecosystem native instruction set. The native command issuance and execution result generation module is used to issue the cross-ecosystem native command set to the corresponding target device, update the control logic state according to the device feedback, and obtain the voice linkage execution result of multiple ecosystem devices.

[0008] The present invention discloses the following technical effects: This invention provides an open-source platform voice-linked control method and system for multi-ecosystem device access. Through steps such as unified capability description modeling, cross-ecosystem voice intent template construction, standardized voice intent parsing, linked control logic generation, linked task orchestration, cross-ecosystem conflict detection, and native command mapping execution, this invention forms a complete chain-like control system spanning "semantics → capability → logic → command." It fundamentally solves the problems mentioned in the background art, such as the lack of unified capability abstraction among multi-ecosystem devices, difficulty in cross-ecosystem voice intent matching, the inability to systematically orchestrate linked control logic, and the susceptibility to conflict and failure in cross-ecosystem control. It achieves unified voice expression, unified capability mapping, unified linked logic, and a unified execution mechanism for multi-brand, multi-protocol, and multi-ecosystem smart devices on an open-source platform. This results in highly consistent, reliable, and scalable voice-linked control across ecosystem scenarios, significantly improving the stability and user experience of multi-ecosystem collaborative control, and significantly reducing system maintenance and ecosystem expansion costs. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart of an open-source platform voice linkage control method for multi-ecosystem device access, provided as an embodiment of the present invention. Detailed Implementation

[0011] 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.

[0012] 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.

[0013] like Figure 1 As shown, this invention provides an open-source platform voice-activated control method for multi-ecosystem device access, including: Step 100: Receive device capability information and status information of multiple target smart devices using an open-source platform, and perform standardized modeling of the device capability information and status information based on a preset unified capability description specification to obtain a standardized device capability index set; Step 200: Summarize commonly used user voice expressions to form voice intent templates, and establish an association between each voice intent template and the capability entries in the standardized device capability index set to obtain a voice intent template library; Step 300: Obtain the voice intent request and perform standardized parsing of the voice intent request according to the voice intent template library to obtain the target voice intent instance; Step 400: Based on the target voice intent instance, select target device capability entries that meet the target control conditions from the standardized device capability index set, and construct cross-ecosystem linkage control logic; Step 500: Arrange the linkage control logic into tasks according to the preset execution order rules to obtain an initial linkage control task sequence; Step 600: Based on the initial linkage control task sequence, perform cross-ecological linkage conflict detection according to preset resource constraints and strategy constraints to obtain the target linkage control task sequence after conflict resolution; Step 700: Based on the target linkage control task sequence, map the standardized equipment operation to the native control instructions corresponding to each equipment ecosystem to obtain a cross-ecosystem native instruction set; Step 800: Send the set of native instructions to the corresponding target device to update the control logic state based on the device feedback, and obtain the voice linkage execution result of multiple ecosystem devices.

[0014] Furthermore, the specific implementation process of step 100 is as follows: This embodiment first establishes data interaction channels with multiple target smart devices through an open-source platform. While ensuring the online status of the devices can be verified, it continuously collects device identification information, device capability information, and device status information to form structured raw device capability data. This embodiment maintains a set of device information access rules on the open-source platform side, ensuring that the raw data already has type labeling, field integrity marking, and capability item classification boundaries during the acquisition process, providing a consistent input basis for subsequent capability extraction and standardization processing.

[0015] After obtaining the original capability data of the equipment, this embodiment extracts capability attributes from the capability, control, and status fields of the original data based on a preset unified capability description specification. This embodiment employs an extraction method based on capability semantic units, decomposing the equipment's functional capabilities, controllable parameters, state variables, and constraints according to capability attributes, ensuring that each capability attribute has clear capability semantics, attribute type, and triggering requirements. By parsing the action types, parameter characteristics, and state dimensions of the equipment capabilities, this embodiment removes the dependence on equipment manufacturer protocols, enabling it to possess cross-ecosystem common expression capabilities, thereby forming a set of equipment capability attributes with unified structured semantics.

[0016] After obtaining the set of device capability attributes, this embodiment maps each capability attribute to a standardized capability description entry, independent of the device ecosystem, according to the capability mapping rules defined in the unified capability description specification. During the mapping process, this embodiment establishes a one-to-one correspondence between the type, control method, configurable parameter range, and state observability of the capability attributes, semantically aligning capabilities from different ecosystems to a unified capability category, forming a standardized set of capability description entries. Subsequently, this embodiment classifies and indexes this set, constructing a hierarchical capability index structure using functional capability type, parameter structure, and target state type as index keys, ultimately obtaining a standardized device capability index set that can be directly used for subsequent voice intent parsing and linkage logic construction.

[0017] Furthermore, the specific implementation process of step 200 is as follows: This embodiment first collects commonly used voice expressions from users in a multi-ecosystem device control scenario using an open-source platform. Real-time commands issued by users through the voice entry point and text commands from historical interaction records are uniformly included in the voice expression collection scope. This data is then processed by combining time tags, usage scenarios, and triggering device information to form raw voice expression data. During the collection process, this embodiment also marks the spatial location of the voice expression, the type of terminal used, and whether cross-device interaction is involved. This ensures that the raw voice expression data not only contains the natural language text itself but also carries contextual information related to the device ecosystem and scenario, providing usable background features for subsequent semantic mining and intent induction.

[0018] After obtaining the original voice expression data, in this embodiment, semantic unit extraction and expression intention classification are performed on each voice expression. By splitting and analyzing the control object words, control action words, target value description words, and constraint condition words in the sentence, sentences with the same control purpose but different specific statements are classified into the same voice expression category, thus forming a voice expression category set. In this embodiment, by comparing the semantic unit combination method and the context scene label, sentences such as "turn on the light a little brighter" and "turn up the brightness of the living room light" are grouped into the same "increase brightness" category, and sentences such as "turn off the air conditioner" and "turn off the bedroom air conditioner" are grouped into the same "turn off the air conditioner" category, so as to remove the individual differences at the natural language expression level and gradually precipitate stable semantic categories corresponding to specific control intentions.

[0019] When constructing the voice intention template, in this embodiment, based on each voice expression category in the voice expression category set, a corresponding voice intention template is constructed for this category, and the semantic slots and their parameter structures for carrying the control object, control action, target parameter, and scene limitation are explicitly defined inside the voice intention template, so that different natural language expressions under the same category can be mapped to the unified template structure. In this embodiment, further according to the ability entries of different device types, ability types, and status quantity types in the standardized device ability index set, a one-to-one or many-to-many association relationship is established between the semantic slots in the voice intention template and the corresponding standardized ability identifiers, thus forming a cross-ecosystem voice intention template association structure, and writing this association structure into the voice intention template storage structure of the open-source platform in a retrievable data structure, and finally obtaining a voice intention template library that can directly drive the linkage control of multi-ecosystem devices.

[0020] Specifically, in this embodiment, the overall structure of the voice intent template is implemented as follows: First, after completing the classification of voice expression categories, this embodiment generates a template record for each voice expression category. This template record contains multiple fields, including at least a user identifier field to identify the user or user group to which the current template applies, a scene tag field to indicate the applicable scenario of the current template, a quantity field to record the number of semantic slots contained in the template, and a semantic slot list field to describe the specific content of each semantic slot. In the semantic slot list field, this embodiment records three elements for each semantic slot in sequence: a semantic tag to characterize the semantic role of that semantic slot. This includes a capability entry identifier that points to a specific capability entry in the standardized equipment capability index set, and target parameters that describe the target control value or target state value corresponding to the semantic slot. The semantic tag comes from the control object word, control action word, or constraint condition word obtained when extracting semantic units from the original speech expression data. The capability entry identifier comes from the capability entry identifier in the established standardized equipment capability index set that semantically matches the semantic tag. The target parameters are organized and summarized according to common control values ​​in the speech expression category, such as percentage values ​​in the brightness adjustment scenario, temperature values ​​in the air conditioning temperature adjustment scenario, or mode names in the scene switching command.

[0021] More specifically, in this embodiment, a concrete example of a voice intent template can be described as follows: the overall structure of the voice intent template corresponds to a template record for lighting brightness adjustment, and its associated user identifier is 1024, indicating that the template is applicable to user number 1024; the scene label of the template is 3, indicating that it is applicable to the living room lighting control scene; the template contains a total of 3 semantic slots, of which the semantic label of the first semantic slot is "object", its associated standardized device capability entry identifier is 2001, corresponding to the device object of the living room light, and its target parameter is 1, used to point to a specific device; the semantic label of the second semantic slot is "action", and its associated capability entry... The target identifier is 3105, corresponding to the device capability of brightness adjustment, and the target parameter is "increase", which is used to express the control intention of increasing the brightness; the semantic label of the third semantic slot is "parameter", and its associated capability entry identifier is 4102, corresponding to the capability entry of brightness target parameter, and the target parameter is 70, which is used to express that the light brightness should be adjusted to 70; through the above structured method, this embodiment can accurately map the natural language "turn up the living room light to 70" into a voice intent template with user identifier, scene label, number of semantic slots, and three parts including object, action and parameter, so that the template content can be directly used by the subsequent voice parsing and linkage control module.

[0022] Furthermore, the specific implementation process of step 300 is as follows: In this embodiment, the acquisition of voice intent requests is based on existing front-end speech recognition and semantic understanding services. When a user issues a control command through a voice input, this embodiment utilizes an open-source platform to receive the recognized text and corresponding semantic understanding results returned by the speech recognition service. These two data, along with a timestamp, input terminal identifier, user identifier, and session identifier, are then compiled into the raw data for the voice intent request. During the receiving phase, this embodiment performs integrity checks and basic cleaning on the raw data. Requests that cannot yield stable results through speech recognition or semantic understanding are marked for manual or rule-based rollback processing and excluded from subsequent template matching processes. This ensures that the raw data for voice intent requests entering the standardized parsing process has clear text content and usable semantic tag information.

[0023] In the semantic structuring stage, this embodiment extracts key semantic elements related to device control from the recognized text and semantic understanding results in the original voice intent request data. These elements include descriptions of the controlled object, control actions, target parameters, and scene constraints. These semantic elements are then categorized according to preset semantic role types to construct a semantic element set. During the construction of the semantic element set, this embodiment also standardizes synonymous and near-synonymous expressions. Different natural language expressions, such as "turn on the light" and "turn on the light," or "adjust the brightness" and "brighten it a little," are grouped into the same semantic action category. Similarly, "living room light" and "living room main light" are grouped into the same controlled object category. This ensures that each element in the semantic element set contains three parts: semantic role type, original text fragment, and standardized semantic label, providing a unified input format for subsequent voice intent template matching.

[0024] In the template matching and instance generation stage, this embodiment uses a semantic element set as a foundation. First, it filters candidate voice intent template sets from the voice intent template library based on user identifiers and scene tags. Then, it compares the semantic tags in the semantic element set with the semantic slots defined in the candidate voice intent templates one by one. By matching information such as semantic role type, controlled object category, and control action category, it determines the target voice intent template identifier that best matches the current voice expression. After determining the target template, this embodiment fills the corresponding slots with specific semantic elements from the semantic element set according to the type requirements of each semantic slot in the target template. It also performs numerical or normalized processing on the target control parameters; for example, adjusting the temperature to 25 degrees is uniformly converted to the value 25, and adjusting the brightness to 70 is uniformly converted to the value 70, thus obtaining the instantiated voice intent structure. Subsequently, this embodiment writes the instantiated voice intent structure into the intent parsing structure maintained by the open-source platform, recording the template identifier, slot filling results, and associated user and scene information, ultimately forming the target voice intent instance.

[0025] Furthermore, the specific implementation process of step 400 is as follows: After obtaining the target voice intent instance, this embodiment first parses the descriptions of the control object, control action, target parameters, and scene constraints contained therein, transforming these descriptions into a structured set of target control conditions. To ensure accurate location of the device the user wishes to control and its specific action, this embodiment combines the user identifier and scene tag in the target voice intent instance during parsing to disambiguate the original description. For example, when the user's statement is "turn on the bedroom air conditioner," this embodiment will determine the space to which the control object belongs based on the two semantic units "bedroom" and "air conditioner" in the target control object description, and will also determine whether the user triggered the command within that space based on the current scene constraints. Through the above method, this embodiment forms a set of target control conditions containing four key control elements: control object, control action, target parameters, and scene constraints, providing clear matching conditions for subsequent capability entry matching in the standardized device capability index set.

[0026] In constructing the candidate capability item set, this embodiment performs an item-by-item matching search in the standardized device capability index set based on the object category, action category, and parameter requirements extracted from the target control condition set. This embodiment limits the range of candidate device types by matching the control object category, then filters out device capability items with corresponding control capabilities by matching the control action category, and further filters out capability items that support the target parameter range based on the target parameter value, thus forming the candidate device capability item set. Subsequently, this embodiment combines device online status information, device spatial location, and user preference information to perform availability filtering on the candidate device capability item set. For example, when multiple devices support the same control action, this embodiment prioritizes selecting the device in the current space or the preferred device set in user preferences; when a device is offline or in a restricted state, this embodiment automatically removes it from the candidate set. Finally, a target device capability item set that meets the actual control requirements of voice intent is formed.

[0027] After obtaining the set of target device capability entries, this embodiment abstracts each device capability entry in the set into an independent linkage control node, and binds its corresponding control action and target parameters to each linkage control node, giving the control node executable action semantics. For example, when the user's voice command is more complex, such as turning off all lights and adjusting the air conditioner to 25°C, this embodiment will construct all relevant device capability entries into multiple linkage control nodes, with each node bound to actions such as turning off and adjusting the temperature. Subsequently, based on the execution order requirements and scenario constraints in the target control condition set, this embodiment establishes sequential, parallel, or conditionally triggered connection relationships between these linkage control nodes, thereby forming a cross-ecosystem linkage control logic. This linkage control logic not only preserves the sequential relationship in the user's voice semantics but also reflects the functional dependencies and spatial collaboration relationships between devices, providing a clear logical basis and structured data support for subsequent task orchestration and conflict detection.

[0028] Furthermore, the specific implementation process of step 500 is as follows: After obtaining the cross-ecosystem linkage control logic, this embodiment first analyzes the logical relationships between the linkage control nodes to extract their sequential dependencies, conditional triggering relationships, and parallel execution capabilities. This embodiment analyzes the control actions, target parameters, and target device spatial locations bound to each linkage control node to identify whether there are execution order requirements between nodes. For example, the temperature can only be set after the air conditioner is turned on, or some lights need to be turned off before switching lighting scenes. Simultaneously, for nodes without explicit dependencies, this embodiment further determines whether their execution will affect each other or cause resource conflicts to determine the feasibility of parallel execution. Through the above methods, this embodiment forms a linkage control node dependency structure to describe node dependencies, sequence requirements, and parallel conditions, providing a structured foundation for subsequent task sorting and batch partitioning.

[0029] After establishing the dependency structure of the linkage control nodes, this embodiment sorts all linkage control nodes according to a preset execution order rule. This execution order rule is jointly formulated based on the implicit order requirements in the user's voice intent, the operational order limitations of the device's capabilities, and scene constraints. For example, when adjusting the brightness of lights, if there is a prerequisite of turning off the lights, the switching operation must precede the brightness adjustment operation. Furthermore, when a user issues multiple action commands at once, this embodiment will prioritize actions that have a greater impact on the overall situation, such as scene mode switching or environmental control actions. After the above sorting, this embodiment obtains the linkage control node sorting result. Based on this, according to the dependencies between nodes, nodes that do not depend on each other are grouped into the same batch as tasks that can be executed in parallel, thus obtaining the task batching result.

[0030] After completing the task batch division, this embodiment writes the target device capability entries and their corresponding control actions within each task batch into the executable task structure, forming a preliminary set of linkage control tasks. Subsequently, this embodiment further organizes this preliminary task set according to execution order rules, ensuring that actions within each task batch are parallel and executable, while actions between batches strictly follow a sequential relationship. For example, in scene switching instructions, the light-off batch must precede the air conditioning adjustment batch, while the ambient light brightness adjustment batch can run concurrently with the air conditioning operation. This embodiment ultimately combines the organized executable task structure into an initial linkage control task sequence, giving the entire task chain a clear execution batch structure and sequence constraints. This provides a clear scheduling basis for subsequent conflict detection and instruction generation, thereby ensuring the feasibility and stability of multi-ecosystem linkage processes.

[0031] Furthermore, the specific implementation process of step 600 is as follows: After obtaining the initial linkage control task sequence, this embodiment first performs attribute parsing on each linkage control task in the sequence. The target device identifier, planned start and end times, and target control parameters required for the task are extracted from the task record. This information is then organized into a linkage control task attribute set. This embodiment uses this attribute set to achieve a structured description of the entire task sequence. This allows each linkage control task to be quickly located and compared in subsequent constraint verification and conflict detection processes using three key attributes: target device, execution time interval, and target control parameters. This provides complete and usable basic data for cross-ecosystem linkage conflict detection.

[0032] After obtaining the set of linkage control task attributes, this embodiment further combines the device's exclusive resource characteristics, safety status restrictions, and user-preset preference strategies to construct a set of cross-ecosystem linkage constraint rules based on preset resource and strategy constraints. The device's exclusive resource characteristics primarily characterize the capability boundary of certain devices to accept only one control command within a given time period; for example, motor-type devices cannot frequently switch operating states simultaneously during start-up and shutdown. Safety status restrictions constrain devices from repeatedly switching between contradictory states within a short period; for example, air conditioner temperatures cannot rapidly fluctuate between extreme high and low values. User preference strategies describe the preferred device or scenario selection when multiple feasible control paths exist. This embodiment uses the set of linkage control task attributes and the set of cross-ecosystem linkage constraint rules as input, iterates and compares all linkage control tasks, and detects task combinations that overlap in execution time intervals on the same target device, have contradictory target control parameters within the same time period, or have a state switching frequency exceeding a preset threshold within a short period. Tasks meeting any of the above conflict determination conditions are merged and organized into a set of linkage conflict tasks.

[0033] After obtaining the set of conflicting linkage tasks, this embodiment adjusts each group of conflicting linkage control tasks according to the pre-set priority strategy in the cross-ecosystem linkage constraint rule set. Specifically, this includes retaining high-priority tasks, postponing the execution of lower-priority tasks that still need to be executed, merging tasks with the same control objectives and close execution times, and canceling tasks that are obviously redundant or inconsistent with user preferences, thereby forming an adjusted set of linkage control tasks. Subsequently, this embodiment re-serializes and reorganizes the adjusted set of linkage control tasks according to the execution order rules adopted in the initial task generation stage. This ensures that the adjusted tasks maintain consistency with the original design in terms of time sequence and batch structure, while eliminating resource competition, parameter conflicts, and high-frequency state jitter issues in the cross-ecosystem linkage process. Finally, a target linkage control task sequence with conflict resolution is obtained, providing safe, reliable, and predictable task input for subsequent native instruction mapping and execution.

[0034] Furthermore, the specific implementation process of step 700 is as follows: After obtaining the target linkage control task sequence, this embodiment first performs a detailed analysis of each linkage control task in the sequence, extracting the target device capability item, target control action, and corresponding target control parameters from the task record. This embodiment combines these three types of information to form a standardized set of device operations, enabling each device operation to be expressed in a unified structure as "device capability item plus control action plus control parameters," thereby abstracting cross-brand and cross-protocol device control requirements into a unified data description. Through this structured and standardized representation, this embodiment ensures that devices from different ecosystems have consistent semantic expression before entering the instruction mapping process, providing a directly processable, standardized input foundation for the subsequent native instruction generation process, thus improving the stability and maintainability of cross-ecosystem adaptation.

[0035] After forming a standardized set of device operations, this embodiment retrieves the matching native command mapping rule from the device ecosystem command mapping rule library maintained by the open-source platform, based on the device ecosystem identifier carried in each standardized device operation. The device ecosystem command mapping rule library records information such as the native control format, command field structure, necessary parameters, and optional parameters supported by various device ecosystems. For example, some ecosystems require commands to be issued in key-value pair format, while others require array structure or binary message format; furthermore, for the same brightness adjustment action, different ecosystems may have different requirements for the field name, value retrieval method, and precision of the target value. This embodiment obtains the native command mapping rule set by reading the corresponding rules in the mapping rule library, ensuring that each standardized device operation can obtain the native command format definition of its respective ecosystem, thus ensuring that the subsequent command conversion process conforms to the command specifications of the device ecosystem.

[0036] After obtaining the set of native instruction mapping rules, this embodiment converts standardized device operations into native control instructions that conform to the device ecosystem requirements, one by one, according to each mapping rule. The conversion process includes determining the format of the native instruction fields according to the mapping rules, writing the target control action into the corresponding action field, writing the target control parameters into the corresponding parameter field, and supplementing necessary header fields or check fields according to the device ecosystem requirements, ultimately forming a set of native control instructions for the device ecosystem. This embodiment then classifies and organizes these native control instructions according to the target device type and device ecosystem, so that instructions under the same ecosystem can be issued in batches, and instructions of the same device type can undergo additional sequence optimization processing when necessary, thereby obtaining a cross-ecosystem set of native instructions.

[0037] Furthermore, the specific implementation process of step 800 is as follows: After obtaining the target linkage control task sequence and the corresponding cross-ecosystem native instruction set, this embodiment first groups and sorts all native control instructions based on the target device identifier and task execution order. Specifically, according to the target device identifier recorded in each linkage control task, native control instructions belonging to the same target device are grouped into the same group. Simultaneously, according to the existing batch order and sequence in the linkage control task sequence, the native control instructions within each group are arranged, thus forming a native instruction sending queue that is both device-based and follows the task execution order. During this process, this embodiment can also appropriately merge control actions that are adjusted multiple times within a short period on the same device, for example, retaining only the last target value for multiple brightness adjustments, to reduce invalid instruction issuance and ensure that the native instruction sending queue is reasonable and executable in terms of both quantity and execution order.

[0038] In this embodiment, after forming a native command sending queue, the native control commands in the queue are sent one by one to the corresponding target devices through the access adaptation channels already established between the open-source platform and the ecosystem of various devices. During the sending process, the sending time, target device identifier, adaptation channel identifier, and sending result marker of each native control command are recorded, thereby obtaining a set of native command sending records. After the native command is successfully delivered or a sending error occurs, this embodiment further utilizes the open-source platform to collect execution confirmation information, status reporting information, and error information returned by each native control command from the target device side. For example, whether the device successfully executed the control action, whether the working state after execution reached the target parameters, whether a timeout or execution refusal occurred, etc. The above information is organized according to the unique identifier of the native control command to form a set of device feedback information, providing a complete data foundation for subsequent linkage execution status statistics and control logic status updates.

[0039] After obtaining the set of device feedback information, this embodiment associates and matches each feedback record with the corresponding linkage control task in the target linkage control task sequence, marking the execution result, execution time, final device status, and whether any anomalies occurred for each linkage control task, forming a linkage execution status set. Subsequently, this embodiment updates the control status in the previously constructed cross-ecosystem linkage control logic based on the linkage execution status set, marking successfully executed control nodes as completed, and marking failed or canceled control nodes as abnormal or incomplete, and recording node information that needs to be retried or alarmed when necessary. On this basis, this embodiment generates a voice linkage result data structure to characterize the execution status of each target device. This data structure includes at least the execution conclusion of each linkage control task, the final working status of related devices, and whether the overall scene linkage has been achieved, thereby obtaining the voice linkage execution results of multi-ecosystem devices, providing directly usable result data support for upper-layer interface display, voice prompt feedback, and subsequent strategy optimization.

[0040] Specifically, in this embodiment, the calculation of the voice-linked execution result is based on the target linkage control task sequence and the corresponding device feedback information and device status change information. The specific implementation is as follows: After multi-ecosystem linkage is completed, this embodiment first counts the number of linkage control tasks actually executed in the target linkage control task sequence, using this number as the total number of tasks in subsequent calculations. Simultaneously, it extracts the execution confirmation result, error information, and status reporting result corresponding to each linkage control task from the aforementioned device feedback information set, and extracts the device status values ​​before and after the execution of each linkage control task from the status change set to evaluate whether the status change achieves the expected goal. Based on this, this embodiment generates a task validity indicator for each linkage control task. This indicator indicates whether the task was actually executed and not prematurely canceled. The value is 1 when the task is normally scheduled and completed, and 0 when the task is canceled due to conflict or does not enter the execution stage. Simultaneously, this embodiment converts the feedback result corresponding to each linkage control task into a feedback availability value based on whether the execution confirmation is complete, whether there are error codes, and whether the status reporting is timely in the device feedback information. This value is close to 1 when the device returns success as expected and the status reporting is complete; otherwise, it is not. When an error occurs, the value decreases from 0 to 1 according to a rule. For example, if the execution times out but ultimately succeeds, it can be 0.7; if only an error message is returned and the execution fails, it can be 0.2. Furthermore, this embodiment compares the target control parameters of each linkage control task with the actual parameters in the device state after execution. A state update consistency value is calculated based on the degree of deviation between the two, used to measure the degree of fit between the final device state and the target state. When the measured value is completely consistent with the target value, the value is 1; when there is a deviation, it decreases from 0 to 1 according to the magnitude of the deviation. For example, if the target brightness is 70 and the actual brightness is 65, it can be 0.9. When the brightness is only 30, a value of 0.4 can be used. In this embodiment, the single contribution value of each linkage control task to the overall linkage quality is obtained by multiplying the task effectiveness indication value, feedback availability value, and state update consistency value. Then, the contribution values ​​of all actually executed linkage control tasks are averaged, and the average result is used as a quantitative score of the voice linkage execution result of multi-ecosystem devices. This enables a comprehensive evaluation of the entire linkage process in three dimensions: execution integrity, feedback reliability, and state achievement. It can serve as an intuitive indicator to show the linkage effect to users, and also as a basis for subsequent strategy optimization and anomaly analysis.

[0041] This embodiment also provides an open-source platform voice-activated control system for multi-ecosystem device access, including: The standardized device capability modeling module is used to receive device capability information and status information of multiple target intelligent devices using an open source platform, and to perform standardized modeling of the device capability information and status information based on a preset unified capability description specification to obtain a standardized device capability index set. The voice intent template construction module is used to summarize commonly used user voice expressions into voice intent templates, and to establish an association between each voice intent template and the capability entries in the standardized device capability index set to obtain a voice intent template library. The voice intent parsing module is used to obtain a voice intent request and perform standardized parsing of the voice intent request according to the voice intent template library to obtain a target voice intent instance; The cross-ecosystem linkage control logic construction module is used to select target device capability entries that meet the target control conditions from the standardized device capability index set based on the target voice intent instance, and construct cross-ecosystem linkage control logic. The linkage task orchestration module is used to orchestrate the linkage control logic according to a preset execution order rule to obtain an initial linkage control task sequence. The cross-ecological linkage conflict detection module is used to perform cross-ecological linkage conflict detection based on the initial linkage control task sequence and according to preset resource constraints and strategy constraints, so as to obtain the target linkage control task sequence after conflict resolution. The native instruction mapping module is used to map standardized device operations into native control instructions corresponding to each device ecosystem based on the target linkage control task sequence, thereby obtaining a cross-ecosystem native instruction set. The native command issuance and execution result generation module is used to issue the cross-ecosystem native command set to the corresponding target device, update the control logic state according to the device feedback, and obtain the voice linkage execution result of multiple ecosystem devices.

[0042] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0043] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A voice-activated control method for an open-source platform oriented towards multi-ecosystem device access, characterized in that, include: The system utilizes an open-source platform to receive device capability information and status information from multiple target smart devices, and performs standardized modeling of the device capability information and status information based on a preset unified capability description specification to obtain a standardized device capability index set. Commonly used user voice expressions are summarized into voice intent templates, and each voice intent template is associated with a capability entry in the standardized device capability index set to obtain a voice intent template library. Obtain a voice intent request, and perform standardized parsing of the voice intent request according to the voice intent template library to obtain a target voice intent instance; Based on the target voice intent instance, target device capability entries that meet the target control conditions are selected from the standardized device capability index set to construct cross-ecosystem linkage control logic; The linkage control logic is arranged into tasks according to a preset execution order rule to obtain an initial linkage control task sequence; Based on the initial linkage control task sequence, cross-ecological linkage conflict detection is performed according to preset resource constraints and strategy constraints to obtain the target linkage control task sequence after conflict resolution. Based on the target linkage control task sequence, standardized equipment operations are mapped to native control commands corresponding to each equipment ecosystem, resulting in a cross-ecosystem native command set. The native instruction set is sent to the corresponding target device to update the control logic state based on device feedback, thereby obtaining the voice linkage execution result of multiple ecosystem devices.

2. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The method involves receiving device capability and status information from multiple target smart devices using an open-source platform, and then standardizing and modeling the device capability and status information based on a pre-defined unified capability description specification to obtain a standardized device capability index set, including: By using an open-source platform, we can obtain the device identification information, device capability information, and device status information of each target smart device to obtain the original device capability data. Based on a preset unified capability description standard, capability attributes are extracted from the original capability data of the device to obtain a set of device capability attributes. According to the unified capability description specification, the capability attributes in the set of device capability attributes are mapped to standardized capability description entries that are independent of the device ecosystem, thus obtaining a set of standardized capability description entries. The standardized capability description item set is classified and indexed to obtain the standardized equipment capability index set.

3. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The process involves summarizing commonly used user speech expressions into speech intent templates, and establishing an association between each speech intent template and capability entries in the standardized device capability index set, resulting in a speech intent template library, including: By using an open-source platform to collect the text of common voice expressions used by users in multi-ecosystem device control scenarios, the raw voice expression data is obtained. Semantic units are extracted and expressive intents are categorized from the raw speech data to obtain a set of speech expression categories. Based on the set of speech expression categories, a corresponding speech intent template is constructed for each speech expression category, and semantic slots and parameter structures are defined in each speech intent template to obtain a set of speech intent templates; Based on the capability entries of different device types, capability types and status quantities in the standardized device capability index set, an association relationship is established between each voice intent template in the voice intent template set and the corresponding standardized capability entry to obtain a cross-ecosystem voice intent template association structure. The voice intent template association structure is written into the voice intent template storage structure of the open source platform to obtain the voice intent template library; The expression for the voice intent template is: ； in, The overall structure of the voice intent template; The user identifier or user group identifier associated with the current voice intent template; These are scene labels obtained based on historical interaction records and the current environmental context; The number of semantic slots in the speech intent template; For the first The semantic tags of each semantic slot are used to describe the controlled object, control action or constraint in the speech expression; In order to be with the first Standardized device capability entry identifiers bound to semantic slots; For the first The target control parameters corresponding to each semantic slot.

4. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The step of obtaining a voice intent request and standardizing and parsing the voice intent request according to the voice intent template library to obtain a target voice intent instance includes: The system utilizes an open-source platform to receive speech recognition text and corresponding semantic understanding results, thereby obtaining the raw data for the speech intent request. The raw data of the voice intent request is subjected to semantic structuring to obtain a set of semantic elements; Based on the speech intent template library, template matching is performed on the semantic element set to obtain the target speech intent template identifier; Based on the target speech intent template identifier, the semantic elements in the semantic element set are filled with parameters and bound to slots to obtain the parameter-instantiated speech intent structure; The instantiated voice intent structure is written into the intent parsing structure of the open-source platform to obtain the target voice intent instance.

5. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, Based on the target voice intent instance, the system selects target device capability entries that meet the target control conditions from the standardized device capability index set, and constructs cross-ecosystem linkage control logic, including: The target control object, target control action, target control parameters, and scene constraints of the target voice intent instance are analyzed to obtain a set of target control conditions. In the standardized equipment capability index set, standardized capability entries that match the target control object, target control action, and target control parameters are retrieved according to the target control condition set to obtain a candidate equipment capability entry set; Based on the device online status information, device spatial location information and user preference information in the current environment, the availability of the candidate device capability item set is filtered to obtain the target device capability item set. Each target device capability item in the target device capability item set is abstracted into a linkage control node, and each linkage control node is bound with a corresponding target control action and target control parameters to obtain a linkage control node set. Establish connections between the set of linkage control nodes to represent sequential execution, parallel execution, or conditional triggering, thereby obtaining cross-ecosystem linkage control logic.

6. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The process of arranging tasks for the linkage control logic according to a preset execution order rule to obtain an initial linkage control task sequence includes: Based on the aforementioned linkage control logic, the sequential relationship, dependency conditions, and parallel possibilities between linkage control nodes are analyzed to obtain the linkage control node dependency structure. Based on the dependency structure of the linkage control nodes, the linkage control nodes are sorted by executability according to a preset execution order rule to obtain the linkage control node sorting result; Based on the sorting results of the linkage control nodes, the independent linkage control nodes are divided into task batches that can be executed in parallel, and the task batch division results are obtained. According to the task batch division results, the target equipment capability entries and corresponding control actions in each task batch are written into the executable task structure to obtain a preliminary linkage control task set. The initial linkage control task set is structured and organized according to the execution order rules to obtain the initial linkage control task sequence.

7. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The process involves performing cross-ecosystem linkage conflict detection based on the initial linkage control task sequence, according to preset resource and policy constraints, to obtain a target linkage control task sequence after conflict resolution, including: Based on the initial linkage control task sequence, the target device, execution time interval and target control parameters corresponding to each linkage control task are analyzed to obtain the linkage control task attribute set. Based on device-exclusive resources, security status restrictions, and user preference policies, a set of cross-ecosystem linkage constraint rules is constructed according to preset resource constraints and policy constraints. Using the set of linkage control task attributes and the set of cross-ecological linkage constraint rules, linkage control tasks that overlap in execution time intervals of the same target device, contradict each other in target control parameters, and have abnormal state switching frequency are detected to obtain a set of linkage conflict tasks. Based on the priority strategy in the set of cross-ecological linkage constraint rules, the linkage control tasks in the set of linkage conflict tasks are adjusted by retaining, postponing, merging, and canceling, resulting in an adjusted set of linkage control tasks. The adjusted set of linkage control tasks is serialized and organized according to the original execution order rules to obtain the target linkage control task sequence after conflict resolution.

8. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The standardized device operations are mapped to native control commands corresponding to each device ecosystem based on the target linkage control task sequence, resulting in a cross-ecosystem native command set, including: Based on the target linkage control task sequence, the target equipment capability entries, target control actions and target control parameters in each linkage control task are analyzed to obtain a standardized set of equipment operations. Based on the device ecosystem identifier corresponding to each standardized device operation in the standardized device operation set, the native instruction mapping rule matching each standardized device operation is retrieved from the device ecosystem instruction mapping rule library of the open source platform to obtain the native instruction mapping rule set; Based on the set of native instruction mapping rules, each standardized device operation is converted into a native control instruction in the format specified by the device ecosystem, and the corresponding target control action and target control parameters are filled into each native control instruction to obtain the set of native control instructions for the device ecosystem. The set of native control commands for the device ecosystem is classified and organized according to the target device type and device ecosystem to obtain the cross-ecosystem native command set.

9. The voice-activated control method for an open-source platform oriented towards multi-ecosystem device access as described in claim 1, characterized in that, The step of sending the native instruction set to the corresponding target device to update the control logic state based on device feedback and obtain the voice linkage execution result of multiple ecosystem devices includes: Based on the target linkage control task sequence and the cross-ecosystem native instruction set, the native control instructions are grouped and sorted according to the target device identifier and task execution order to obtain the native instruction sending queue; Through the access and adaptation channels of the open source platform and various device ecosystems, the native control commands in the native command sending queue are sent one by one to the corresponding target devices to obtain a set of native command sending records; Based on the native command issuance record set, the execution confirmation information, status reporting information and error information returned by the target device in response to the native control command are collected using an open source platform to obtain the device feedback information set; The set of device feedback information is associated with the target linkage control task sequence to obtain a set of linkage execution states; The control state of the cross-ecosystem linkage control logic is updated based on the linkage execution state set, and a voice linkage result data structure representing the execution status of each target device is generated to obtain the voice linkage execution result of the multi-ecosystem devices. The expression for the result of the voice-activated execution is: ； in, The number of linkage control tasks actually executed in the target linkage control task sequence. This represents the set of device feedback information corresponding to the aforementioned linkage control task. This represents the set of state changes that occur to the target device during the execution of each linkage control task. For the first The indicator quantity for the effectiveness of each linkage control task. For the first The availability value of the device feedback results corresponding to each linkage control task. For the first Consistency of state updates for all linked control tasks This is the final result of voice-activated interaction among multiple ecosystem devices.

10. An open-source platform voice-linked control system for multi-ecosystem device access, characterized in that, include: The standardized device capability modeling module is used to receive device capability information and status information of multiple target intelligent devices using an open source platform, and to perform standardized modeling of the device capability information and status information based on a preset unified capability description specification to obtain a standardized device capability index set. The voice intent template construction module is used to summarize commonly used user voice expressions into voice intent templates, and to establish an association between each voice intent template and the capability entries in the standardized device capability index set to obtain a voice intent template library. The voice intent parsing module is used to obtain a voice intent request and perform standardized parsing of the voice intent request according to the voice intent template library to obtain a target voice intent instance; The cross-ecosystem linkage control logic construction module is used to select target device capability entries that meet the target control conditions from the standardized device capability index set based on the target voice intent instance, and construct cross-ecosystem linkage control logic. The linkage task orchestration module is used to orchestrate the linkage control logic according to a preset execution order rule to obtain an initial linkage control task sequence. The cross-ecological linkage conflict detection module is used to perform cross-ecological linkage conflict detection based on the initial linkage control task sequence and according to preset resource constraints and strategy constraints, so as to obtain the target linkage control task sequence after conflict resolution. The native instruction mapping module is used to map standardized device operations into native control instructions corresponding to each device ecosystem based on the target linkage control task sequence, thereby obtaining a cross-ecosystem native instruction set. The native command issuance and execution result generation module is used to issue the cross-ecosystem native command set to the corresponding target device, update the control logic state according to the device feedback, and obtain the voice linkage execution result of multiple ecosystem devices.

Images