System manipulation method, device and equipment based on n+1-dimensional intention container

Abstract

The application relates to a system operation method, device and equipment based on an N+1-dimensional intention container. The method comprises the following steps: identifying a plurality of independent description dimensions of a target system, and determining the number N of the plurality of independent description dimensions. An N+1-dimensional intention container is constructed. The N+1-dimensional intention container is projected to each dimension in a dimension reduction manner to generate a plurality of slice interfaces. Intention labeling information of a system operator on the plurality of slice interfaces is acquired, the intention labeling information on the plurality of slice interfaces is comprehensively processed, and complete intention description in the N+1-dimensional intention container is restored. The complete intention description is compiled into an execution instruction of the target system. By using the method, the cognitive threshold of man-machine cooperation is reduced, and the fidelity and transferability of complex system operation are improved.

Description

Technical Field

[0001] This application relates to the field of human-computer interaction system control technology, and in particular to a system control method, apparatus and equipment based on an N+1 dimensional intent container. Background Technology

[0002] When operating complex systems (such as drone swarms, software architecture, creative choreography, and human intent perception), the operator needs to convey their intentions to the target system, which then generates corresponding execution commands. Traditional human-computer interaction methods typically employ a "dimensionality reduction approximation" strategy: projecting the high-dimensional system state space onto a low-dimensional interactive interface (e.g., using a two-dimensional map to control a drone in three-dimensional space, using a static storyboard to choreograph movie shots that include a time dimension, or using fragmented UML diagrams to describe multi-layered software architecture). The operator inputs commands on the low-dimensional interface, and the system then maps the commands back to the high-dimensional system space.

[0003] However, existing methods share a common technical flaw: the descriptive dimension of the operator's intent is always lower than the system's own dimension, leading to irreversible information loss during intent transmission. Specifically, for complex systems with N independent descriptive dimensions (N≥1), the interactive interfaces provided by existing control tools can only express information in N-1 or fewer dimensions. Missing dimensions (such as time dimension, semantic constraint dimension, sentiment annotation dimension, etc.) can only be implicitly supplemented by the operator's experience and cannot be explicitly understood by the system. For example, in drone formation control, the system requires complete spatiotemporal commands in four dimensions (x, y, z, t), but the operator can only draw flight paths on a two-dimensional map. The intent in the time dimension and spatial altitude dimension can only be approximated through discrete parameter inputs or post-event adjustments, making continuous and synchronous intent annotation impossible. Similarly, in software system architecture design, the system involves five independent dimensions: interface layer, business logic layer, data layer, timing layer, and boundary layer. However, traditional UML class diagrams or BPMN flowcharts can only express the static relationships of some dimensions and cannot describe cross-dimensional dynamic constraints and design intents within the same framework.

[0004] Furthermore, existing technologies lack universal methods for dimension recognition and frameworks for constructing intent containers. Control tools in different domains (such as flight control ground stations, storyboard software, and UML modeling tools) each employ dedicated dimensionality reduction strategies, making them incompatible. When operators need to manage multiple complex systems of different types simultaneously, they must learn and switch between multiple completely different interaction paradigms, resulting in high cognitive load and low efficiency. Two conventional approaches exist: one is to increase the dimensions of the interactive interface (such as 3D input devices), but the operator's cognitive load increases exponentially with dimensionality, making it impractical; the other is to use AI to infer missing dimensions, but the inference results are uncontrollable and ambiguous, unsuitable for safety-critical tasks. Therefore, a new mechanism is needed that can maintain low-dimensional interaction while fully preserving high-dimensional intent.

[0005] Therefore, how to provide a general method for manipulating complex systems that can fully preserve the operator's intentions, so that for any target system with N independent descriptive dimensions, the operator can input a high-dimensional complete intention in a low-dimensional interactive manner and compile the intention into the target system's execution instructions without loss, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] Therefore, it is necessary to provide a system control method, apparatus, and device based on an N+1 dimensional intent container that can improve the fidelity and portability of controlling complex human-computer interaction systems, in order to address the above-mentioned technical problems.

[0007] A system manipulation method based on an N+1 dimensional intent container, the method comprising: Identify multiple independent descriptive dimensions of the target system and determine the number N of these independent descriptive dimensions, where N is a positive integer greater than or equal to 1.

[0008] Construct an N+1 dimensional intent container, where the number of dimensions of the N+1 dimensional intent container is N+1. The first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension.

[0009] The N+1 dimensional intent container is projected into each dimension in a dimensionality reduction manner to generate multiple slice interfaces.

[0010] The system acquires the intent annotation information of the operator on multiple slice interfaces, integrates the intent annotation information from multiple slice interfaces, and restores it into a complete intent description in an N+1 dimensional intent container.

[0011] The complete intent description is compiled into executable instructions for the target system.

[0012] A system control device based on an N+1 dimensional intent container, the device comprising: The identification module is used to identify multiple independent descriptive dimensions of the target system and determine the number N of the multiple independent descriptive dimensions, where the number N is a positive integer greater than or equal to 1.

[0013] The intent container building module is used to build an N+1 dimensional intent container. The N+1 dimensional intent container has N+1 dimensions. The first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension.

[0014] The dimension reduction projection module is used to project the N+1 dimensional intent container into various dimensions to generate multiple slice interfaces.

[0015] The intent description restoration module is used to obtain the intent annotation information of the system operator on multiple slice interfaces, and to comprehensively process the intent annotation information on multiple slice interfaces to restore it into a complete intent description in an N+1 dimensional intent container.

[0016] The instruction compilation module is used to compile the complete intent description into executable instructions for the target system.

[0017] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps: Identify multiple independent descriptive dimensions of the target system and determine the number N of these independent descriptive dimensions, where N is a positive integer greater than or equal to 1.

[0018] Construct an N+1 dimensional intent container, where the number of dimensions of the N+1 dimensional intent container is N+1. The first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension.

[0019] The N+1 dimensional intent container is projected into each dimension in a dimensionality reduction manner to generate multiple slice interfaces.

[0020] The system acquires the intent annotation information of the operator on multiple slice interfaces, integrates the intent annotation information from multiple slice interfaces, and restores it into a complete intent description in an N+1 dimensional intent container.

[0021] The complete intent description is compiled into executable instructions for the target system.

[0022] The aforementioned system control method, apparatus, and equipment based on an N+1 dimensional intent container fundamentally solves the technical problem of "insistence loss due to insufficient description dimensions" in traditional control methods by first identifying the number N of independent descriptive dimensions of the target system, and then constructing an intent container with N+1 dimensions (the first N dimensions correspond one-to-one with the system dimensions, and the N+1th dimension is the intent annotation dimension). Existing technologies (such as controlling a four-dimensional drone with a two-dimensional map, arranging four-dimensional creative ideas with a static storyboard, and describing five-dimensional software architecture with fragmented UML diagrams) all suffer from the lack of an intent annotation dimension, forcing operators to approximate high-dimensional intents with low-dimensional interfaces, resulting in information loss and ambiguity. This method generates multiple slice interfaces by dimensionality-reducing projection of the N+1 dimensional container onto each dimension, which not only reduces the cognitive burden on the operator (annotating intents on two-dimensional slices) but also restores the complete N+1 dimensional intent description through subsequent comprehensive processing, thus achieving a technological leap of "low-dimensional interaction and high-dimensional expression". Meanwhile, this method compiles the complete intent description into execution instructions for the target system, enabling the operator's intent to be losslessly transformed into precise control parameters executable by the system, significantly improving control accuracy, efficiency, and cross-scenario adaptability. Since this method does not limit the specific value of N or the system type, its technical effects are universal: for any complex system satisfying "N independent description dimensions" (from four-dimensional drones to six-dimensional human intent systems), the complete transmission and compilation of intent can be achieved by constructing an N+1-dimensional intent container, automatically identifying N, avoiding subjective bias when manually specifying dimensions, and avoiding the engineering waste of repeatedly developing dedicated control tools for different domains. In summary, this method achieves a breakthrough from "low-dimensional approximate control" to "high-dimensional precise intent transmission," lowering the cognitive threshold for human-machine collaboration and improving the fidelity and transferability of complex system control. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a system manipulation method based on an N+1 dimensional intent container in one embodiment. Figure 2 This is a schematic diagram of the interactive interface of six slice interfaces (six-sided boxes) in a drone formation control scenario in one embodiment; Figure 3 This is a structural block diagram of a system control device based on an N+1 dimensional intent container in one embodiment; Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0025] In one embodiment, such as Figure 1 As shown, a system manipulation method based on an N+1 dimensional intent container is provided. This method is also applicable to other types of target systems (such as creative content systems, software systems, manufacturing process systems, human intent systems, etc.), requiring only corresponding adjustments to the dimension recognition and compilation rule base. The method includes the following steps: Step 102: Identify multiple independent descriptive dimensions of the target system and determine the number N of these multiple independent descriptive dimensions.

[0026] The quantity N is a positive integer greater than or equal to 1.

[0027] Example 1 illustrates the specific implementation of a system control method based on an N+1 dimensional intent container, using a UAV swarm (a four-dimensional spatiotemporal physical system) as the target system. The hardware environment includes: a ground control station (equipped with a touchscreen display), five quadcopter UAVs (each equipped with a flight control module and a communication module), and a computing server (used to construct, project, restore, and compile the intent container). The ground control station communicates with the computing server, and the computing server communicates with the UAV swarm.

[0028] Example 2: For creative content systems (such as the shot arrangement of a film director), simply change the dimension recognition results to (camera trajectory, event sequence, scene material, character action), change the compilation target to AI video generation API instructions, and adopt a parameterized version management method (separating spatial structure parameters and style parameters) to achieve automatic generation from storyboard to multiple versions of video.

[0029] Example 3: For the human intention system, the dimensions include (physical structure, emotional state, neural intention, conscious intention, and time sequence), and the compilation targets include exoskeleton assistance commands, emotional response lights, BCI control signals, etc., which can unify the data flow of related physical perception, emotional computing, brain-computer interface and intention execution system.

[0030] Identify multiple independent descriptive dimensions of the target system and determine their number N.

[0031] Specifically, for a drone formation system, its complete state needs to be represented by three-dimensional spatial coordinates. Together with time t, it describes the state variables. Enumerate all state variables: the position coordinates of each drone. And timestamps. After orthogonally decomposing these variables and eliminating linearly correlated variables, the minimum set of independent dimensions is obtained as: x-axis, y-axis, z-axis, and time axis. Therefore, N=4.

[0032] Step 104: Construct an N+1 dimensional intent container.

[0033] The N+1 dimension intent container has N+1 dimensions. The first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension.

[0034] Specifically, based on the four identified independent descriptive dimensions We construct a four-dimensional coordinate system using axes. Then, we add a fifth dimension as the Intent Annotation Dimension (IAD), creating a five-dimensional intent space.

[0035] Furthermore, an intent description vector is defined in the five-dimensional intent space. ,in This represents the semantic annotation value of the operator's intention at the corresponding spatiotemporal location (such as the encoded value of commands like "hover", "fly around", "accelerate").

[0036] Step 106: Project the N+1 dimensional intent container into each dimension to generate multiple slice interfaces.

[0037] Specifically, determine the projection directions corresponding to each dimension of the five-dimensional intent container: directions perpendicular to the x-axis, y-axis, z-axis, t-axis, and intent annotation dimension.

[0038] Furthermore, for each projection direction, points in the five-dimensional container are projected onto a hyperplane spanned by the other four dimensions, generating a four-dimensional slice. A total of five four-dimensional slices are generated. Since the operator cannot directly understand the four-dimensional slices, each slice is further converted into a two-dimensional interactive interface (e.g., ...). Figure 2 (As shown).

[0039] In this embodiment, since the target system comprises a three-dimensional space (x, y, z), the projection of the first three dimensions is particularly crucial. Specifically: Projecting perpendicular to the z-axis yields a slice of the xy-plane (top view). Projecting perpendicular to the y-axis yields a slice of the xz plane (front view). Projecting perpendicular to the x-axis yields a slice of the yz plane (side view). Projecting perpendicular to the t-axis yields a three-dimensional spatial slice (static snapshot) at a given moment. Projecting perpendicular to the intention annotation dimension yields a pure four-dimensional spatiotemporal slice ignoring the intention annotation.

[0040] Furthermore, the operator sees six two-dimensional interactive interfaces on the touchscreen of the ground control station (i.e., the six sides of a six-sided box: top view, bottom view, front view, rear view, left view, and right view). Each interface can receive intention annotations such as finger or stylus trajectory drawing, selection, and parameter input.

[0041] Step 108: Obtain the intent annotation information of the system operator on multiple slice interfaces, and comprehensively process the intent annotation information on multiple slice interfaces to restore the complete intent description in the N+1 dimensional intent container.

[0042] Specifically, the operator can mark their intentions on six slice interfaces, such as: drawing a continuous flight path on the top view; marking the required climb height at a certain coordinate point on the front view; marking the area to avoid obstacles on the side view; and setting the time nodes for formation changes on the timeline view.

[0043] Furthermore, the ground control station acquires the two-dimensional annotation information on each slice interface. Based on the projection transformation matrix (pre-stored) corresponding to each slice interface, the two-dimensional annotation information is inversely projected back to a local intent vector in the five-dimensional intent space. For example, a point (u,v) on the top view is transformed into a straight line in three-dimensional space (x,y fixed, z free) through inverse projection, and the three-dimensional space point can be uniquely determined by combining it with the annotation of the front view.

[0044] Furthermore, the six local intent vectors are fused and calculated using a weighted summation formula: ; in, To provide a complete intent description vector, For the first Preset confidence weights for each slice of the interface. For the first Local intent vectors for each slice of the interface. Preset confidence weights for each slice. The confidence weight of each slice interface can be dynamically adjusted based on the operator's historical operational accuracy. Dynamic calculation based on the operator's historical operational accuracy: =(1-error rate_j) / Σ(1-error rate_k), where error rate This represents the average deviation between the operator's annotations on the sliced ​​interface and the actual execution results. During initialization, the weight of spatial dimension slices (e.g., top view) is set to 0.3, the weight of time axis slices to 0.2, the weight of height annotations in the front view to 0.2, and the weight of intent annotation dimension slices to 0.1, etc. The complete intent description vector is then calculated. It contains complete flight path, speed, and formation changes of the UAV formation throughout the entire spatiotemporal domain. The errors of the independent annotation of each slice are mutually corrected through confidence weights, resulting in an overall reconstruction accuracy higher than any single slice.

[0045] Step 110: Compile the complete intent description into executable instructions for the target system.

[0046] Specifically, the target system type is identified as a "physical space system," and the corresponding UAV flight control compilation rule library is selected. The complete intent description vector is then... Input the compilation rule base to generate a sequence of flight control commands for each UAV (e.g., MAVLink format SET_POSITION_TARGET_LOCAL_NED messages).

[0047] Furthermore, the computing server sends the instruction sequence to each UAV flight control module via a communication link, and the UAV formation executes the flight mission according to the operator's intention.

[0048] The aforementioned system control method based on an N+1 dimensional intent container fundamentally solves the technical problem of "insistence loss due to insufficient description dimensions" in traditional control methods by first identifying the number N of independent descriptive dimensions of the target system and then constructing an intent container of N+1 dimensions (the first N dimensions correspond one-to-one with the system dimensions, and the N+1th dimension is the intent annotation dimension). Existing technologies (such as controlling a four-dimensional drone with a two-dimensional map, arranging four-dimensional creative ideas with a static storyboard, and describing five-dimensional software architecture with fragmented UML diagrams) all lack an intent annotation dimension, causing operators to only approximate high-dimensional intents with low-dimensional interfaces, resulting in information loss and ambiguity. This method generates multiple slice interfaces by dimensionality-reducing projection of the N+1 dimensional container onto each dimension, which reduces the operator's cognitive burden (annotating intents on two-dimensional slices) and restores the complete N+1 dimensional intent description through subsequent comprehensive processing, thus achieving a technological leap of "low-dimensional interaction and high-dimensional expression". At the same time, this method compiles the complete intent description into execution instructions for the target system, enabling the operator's intent to be losslessly converted into precise control parameters that the system can execute, significantly improving control accuracy, efficiency, and cross-scene adaptability. Because this method does not limit the specific value of N or the system type, its technical effect is universal: for any complex system that satisfies "N independent descriptive dimensions" (from four-dimensional drones to six-dimensional human intention systems), the complete transmission and compilation of intentions can be achieved by constructing an N+1-dimensional intention container, automatically identifying N, avoiding subjective bias when manually specifying dimensions, and avoiding the engineering waste of repeatedly developing dedicated control tools for different domains. In summary, this method achieves a breakthrough from "low-dimensional approximate control" to "high-dimensional precise intention transmission," lowers the cognitive threshold for human-machine collaboration, and improves the fidelity and transferability of complex system control.

[0049] In one embodiment, all state variables of the target system are enumerated to construct a set of state variables. Orthogonal decomposition is performed on the variables in the set of state variables to eliminate linearly correlated variables, resulting in an orthogonal basis set. The set of minimum independent dimensions is extracted from the orthogonal basis set; the number of dimensions in the minimum independent dimension set is the number N of the multiple independent descriptive dimensions.

[0050] In one embodiment, an N-dimensional spatial coordinate system is constructed using the identified N independent description dimensions as coordinate axes. An (N+1)th dimension is added to the N-dimensional spatial coordinate system as the intent annotation dimension, constructing an N+1-dimensional intent space. Intent description vectors are defined in the N+1-dimensional intent space: ; in, to These are the coordinates of the first N dimensions. These are semantic annotation values ​​used to characterize the intent of the system operator at the corresponding N-dimensional spatial location. The coordinate values ​​of the dimension to be labeled.

[0051] In one embodiment, projection directions corresponding to each dimension of the N+1 dimensional intent container are determined, with each projection direction perpendicular to a coordinate axis. For each projection direction, points in the N+1 dimensional intent container are projected onto a hyperplane spanned by the dimensions other than the perpendicular coordinate axis, generating an N-dimensional slice. Each N-dimensional slice is then converted into a two-dimensional interactive interface for receiving intent annotations from the system operator.

[0052] In one embodiment, the two-dimensional annotation information input by the system operator on each slice interface is obtained. Based on the projection transformation matrix corresponding to each slice interface, the two-dimensional annotation information is inversely projected back to a local intent vector in the N+1-dimensional intent space. The local intent vectors corresponding to all slice interfaces are then fused to obtain the complete intent description vector. ; in, To provide a complete intent description vector, For the first Preset confidence weights for each slice of the interface. For the first Local intent vectors of each slice interface.

[0053] In one embodiment, the type of the target system is identified, and based on the type of the target system, a corresponding compilation rule base is selected. The complete intent description vector is input into the compilation rule base to generate an executable instruction sequence for the target system. Types include physical space systems, creative content systems, software systems, manufacturing process systems, human intent systems, organizational management systems, medical clinical systems, legal compliance systems, music creation systems, educational learning systems, and supply chain logistics systems.

[0054] In one embodiment, when the type is a creative content system, the spatial structure parameters and style parameters in the complete intent description vector are separated, and multiple versions of executable instruction sequences are generated by replacing the style parameters. The spatial structure parameters remain unchanged.

[0055] In one embodiment, a creative content system (a film director's shot arrangement) is used as the target system to illustrate the application of a system manipulation method based on an N+1 dimensional intent container in the creative industry. Spatial structure parameters and style parameters are separated, and multiple versions of executable instruction sequences are generated by replacing the style parameters.

[0056] The hardware environment used includes: an interactive storyboard tablet (supporting a stylus), a creative intent server (for constructing intent containers and compiling them), and multiple output targets (AI video generation engine, game engine, and stage lighting control system). The director's annotations on the tablet are transmitted to the server in real time.

[0057] Specifically, a complete description of the creative arrangement of a film scene should consist of the following four independent dimensions: Camera trajectory dimension: The movement path of the camera in three-dimensional space (push, pull, pan, tilt, follow).

[0058] Event sequence dimension: Events that occur in the scene (character entry, dialogue, explosion, etc.) and their temporal relationships.

[0059] Scene material dimension: Static elements in the scene such as objects, props, lighting, and textures.

[0060] Character movement dimension: The character's physical movements, facial expressions, movement, and other dynamic behaviors.

[0061] Enumerate all creative variables (camera position sequence, event trigger time point, object material ID, character skeletal animation parameters), and after orthogonal decomposition, we get N=4.

[0062] Furthermore, a four-dimensional coordinate system is constructed using camera trajectory, event sequence, scene materials, and character actions as four axes. A fifth dimension is added as a creative intent annotation dimension, constructing a five-dimensional intent space.

[0063] Define the intent description vector Where: c represents the camera trajectory parameters (such as Bézier curve control points); e represents the event sequence (the sequence of event IDs on the timeline). 'a' represents the scene material parameters (material sphere parameter set); 'a' represents the character motion parameters (motion fragment ID and blend weight). This indicates the director's intentional meaning attached to the corresponding creative elements (such as "emphasizing the protagonist's emotions" or "creating tension").

[0064] Furthermore, the five-dimensional intent container is projected onto various dimensions to generate five four-dimensional slices, each of which is further converted into a two-dimensional interactive interface: Director's View (fixed events, materials, actions, projected camera trajectory): A 2D plan view of the set from above, where the director can draw camera movement paths.

[0065] Top-down view of movement (fixed camera, materials, actions, and sequence of projected events): Similar to the top-down view of a stage set, the director can drag character icons to define the order of entry.

[0066] Background depth (fixed camera, event, action, projected scene material): used to replace materials, lighting, and background images in the scene.

[0067] Ground constraint surface (fixed camera, events, materials, projected character actions): used to bind specific action segments to the character.

[0068] Timeline plane (a slice of time at a fixed moment in the first four dimensions, projected time): used to arrange the trigger times of events and actions.

[0069] The director made annotations on each slice screen: Draw an "arc-shaped track from the doorway to the window" from the director's perspective; On the top-down view of the movement, mark "0s-5s: the main character moves from point A to point B; 5s: the supporting character enters from point C"; Select the "Forest Texture Pack" on the background surface with depth; Bind the "run-stop-turn" action to the character on the ground constraint surface; Set the precise time offset for each event on the timeline plane.

[0070] The system acquires the two-dimensional annotation information on each slice of the interface, and inversely projects it into a local intent vector in five-dimensional space based on the projection transformation matrix. Then, the weighted summation formula is used: ; Merge into a complete intent description vector .

[0071] Furthermore, the target system is identified as a "creative content system," and the corresponding multi-objective compilation rule library is selected. The spatial structure parameters (camera trajectory geometry, event sequence logic, character movement path) and style parameters (material texture, lighting color, motion style) in the complete intent description vector are separated.

[0072] After separation: The spatial structure parameters remain unchanged, serving as the "core creative framework"; style parameters can be replaced independently.

[0073] Then, by replacing the style parameters, multiple versions of the executable instruction sequence are generated: Version 1 (Realistic Style): Compiled into a scene file (.umap) for the UE5 game engine, using high-precision PBR textures for materials; Version 2 (Cartoon Style): Replaces material parameters with cartoon shader parameters and compiles into a Blender animation script; Version 3 (Stage Play Style): Replaces motion parameters with exaggerated stage movements and compiles them into timing instructions for DMX lighting and stage machinery; Version 4 (AI Video Generation): Compiled into JSON parameters (prompt words + control network) for the Runway Gen-2 API.

[0074] Directors can quickly generate multiple versions that adapt to different output media and artistic styles without changing the core creative idea (spatial structure), greatly improving the reuse efficiency of creative content.

[0075] It is worth noting that by formalizing the director's complete creative intent (shots, events, materials, actions, and their interactive semantics) through an N+1 dimensional intent container, and then separating spatial structure parameters from style parameters, "one-time annotation, multiple versions output" is achieved. Compared to traditional storyboards (which can only express static images and lack a timeline and cross-media compilation), this method significantly improves the efficiency of creative arrangement and cross-media adaptability. This embodiment can serve as a core technology for creative content arrangement-related products.

[0076] In one embodiment, using a human intention system as the target system, the specific application of the system manipulation method based on an N+1 dimensional intention container in "human-machine collaboration" and "intention perception and response" scenarios is illustrated. This embodiment can unify and coordinate the data flow of the physical perception system, emotion computing system, brain-computer interface system, and intention execution system to achieve multi-dimensional perception and response to the complete state of the human body.

[0077] The hardware environment employed includes: a multimodal sensor array (containing an RGB camera, millimeter-wave radar, WiFi CSI receiver, and a consumer-grade EEG headband), an edge computing device (for constructing, projecting, reproducing, and compiling intent containers), and an output device (such as a crab-like exoskeleton, a smart lighting system, and a digital human-driven screen). The sensor array continuously collects human body data, and the edge computing device communicates with the output device.

[0078] Specifically, the execution steps are: identify multiple independent descriptive dimensions of the target system (human body in the context of intent-perception-execution) and determine its number N.

[0079] Furthermore, the complete state of the human body needs to be described by the following five independent dimensions, and no single dimension can be fully derived from the other dimensions: The first dimension: physical structure dimension—skeletal posture, joint angles, movement trajectory, and muscle tension.

[0080] The second dimension: the emotional state dimension—the three-dimensional emotional coordinates of VAD (valence, arousal, dominance) and their changing trajectory.

[0081] The third dimension: the neural intent dimension—preconscious neural signals such as P300 event-related potentials, EEG rhythm power, and skin conductance responses.

[0082] The fourth dimension: conscious intention dimension—the user's explicit instructions, decisions, and expression of preferences.

[0083] The fifth dimension: the time series dimension—the trajectory of all the above states over time.

[0084] We enumerate all state variables (skeletal point coordinates, emotion score, EEG channel amplitude, voice command, timestamp) and perform orthogonal decomposition. After eliminating linearly correlated variables, the smallest independent dimension set has a size of 5. Therefore, N=5.

[0085] Furthermore, a five-dimensional spatial coordinate system is constructed using the five identified independent descriptive dimensions as coordinate axes. A sixth dimension is then added as the Intent Annotation Dimension (IAD), constructing a six-dimensional intent space.

[0086] Furthermore, an intent description vector is defined in the six-dimensional intent space. ,in: Represents the physical structure state (such as a skeleton coordinate vector); Represents emotional state (VAD 3D coordinates); Indicates neural intent characteristics (such as P300 amplitude). It indicates a conscious intention (such as "I want to pick up the water glass"). Represents a timestamp; This indicates the semantic annotation value of intent attached by the system or external operator (such as a therapist or AI assistant) at the corresponding human body state location (such as "needs more assistance" or "anxiety detected, reduce difficulty").

[0087] Determine the projection directions corresponding to each dimension of the six-dimensional intent container. Generate six five-dimensional slices, each of which is further converted into a two-dimensional interactive interface (or a two-dimensional visualization view). In this embodiment, the six slice interfaces are specifically shown in Table 1: Table 1 Six Slice Interfaces

[0088] These slice interfaces are displayed on the therapist's control terminal or the user's AR glasses, receiving annotation input. The system acquires the annotation information input on each slice interface. For example: On the skeletal projection plane, the therapist selected the "right arm" area and marked it "requires assistance to improve by 30%"; In terms of emotional state, the system automatically detected that the user's VAD coordinates fell into the "high anxiety" area, and the therapist marked "start relaxation mode"; On the neural response surface, the EEG detected the P300 signal corresponding to the "water cup" icon, and the system automatically labeled it as "the user intended to select a water cup"; In terms of intent description, users can directly input "I want to drink water" via voice.

[0089] Furthermore, based on the projection transformation matrix corresponding to each slice interface, the two-dimensional annotation information is inversely projected and mapped back to the local intent vector in the six-dimensional intent space. Then, a weighted summation formula is used for fusion: ; The pre-set confidence weights for each slice can be dynamically adjusted: for example, the confidence of the neural intent slice decreases when the user is fatigued, while the confidence of the conscious intent slice increases when the user is awake. The fusion process yields a complete intent description vector. It contains the complete state of the human body at the current moment and in the short future time domain, as well as the annotations of the external response strategy.

[0090] Furthermore, the target system is identified as a "human intent system," and the corresponding multimodal compilation rule library is selected. Input the compilation rule base and generate execution instructions for multiple output devices: Exoskeleton (Mech Crab): Generate assist curve instructions (e.g., "Right arm shoulder joint assist 30%, elbow joint assist 10%)"); For smart environments: Generate commands for adjusting light color and music (e.g., "Switch to blue cool light and play soothing music"). Driven by digital humans: Generate facial expressions and motion commands for virtual avatars (such as "virtual human nods and hands over a water glass animation").

[0091] It's worth noting that by unifying the five dimensions of the human body—physical, emotional, neural, conscious, and temporal—into a single six-dimensional intention container, cross-dimensional collaborative responses are achieved. For example, when anxiety is detected in the emotional state dimension, the system can automatically lower the confidence threshold of the neural intention dimension (to avoid misjudgment) and simultaneously adjust the exoskeleton assistance strategy to a "gentle mode." This cross-dimensional linkage is impossible to achieve in traditional single-dimensional perception systems.

[0092] It should be understood that, although Figure 1The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0093] In one embodiment, such as Figure 3 As shown, a system control device based on an N+1 dimensional intent container is provided, including: a recognition module 302, an intent container construction module 304, a dimensionality reduction projection module 306, an intent description restoration module 308, and an instruction compilation module 310, wherein: The identification module 302 is used to identify multiple independent description dimensions of the target system and determine the number N of the multiple independent description dimensions, where the number N is a positive integer greater than or equal to 1.

[0094] The intent container construction module 304 is used to construct an N+1 dimensional intent container. The N+1 dimensional intent container has N+1 dimensions. The first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension.

[0095] The dimension reduction projection module 306 is used to project the N+1 dimensional intent container into various dimensions to generate multiple slice interfaces.

[0096] The intent description restoration module 308 is used to obtain the intent annotation information of the system operator on multiple slice interfaces, and to comprehensively process the intent annotation information on multiple slice interfaces to restore the complete intent description in the N+1 dimensional intent container.

[0097] Instruction compilation module 310 is used to compile a complete intent description into executable instructions for the target system.

[0098] Specific limitations regarding the system control device based on N+1 dimensional intent containers can be found in the limitations of the system control method based on N+1 dimensional intent containers described above, and will not be repeated here. Each module in the aforementioned system control device based on N+1 dimensional intent containers can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0099] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a system control method based on an N+1 dimensional intent container. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0100] Those skilled in the art will understand that Figures 3-4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0101] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to perform the following steps: Identify multiple independent descriptive dimensions of the target system and determine the number N of these independent descriptive dimensions, where N is a positive integer greater than or equal to 1.

[0102] Construct an N+1 dimensional intent container, where the number of dimensions of the N+1 dimensional intent container is N+1. The first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension.

[0103] The N+1 dimensional intent container is projected into each dimension in a dimensionality reduction manner to generate multiple slice interfaces.

[0104] The system acquires the intent annotation information of the operator on multiple slice interfaces, integrates the intent annotation information from multiple slice interfaces, and restores it into a complete intent description in an N+1 dimensional intent container.

[0105] The complete intent description is compiled into executable instructions for the target system.

[0106] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchlink, DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0107] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0108] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A system manipulation method based on an N+1 dimensional intent container, characterized in that, The method includes: Identify multiple independent descriptive dimensions of the target system and determine the number N of the multiple independent descriptive dimensions, wherein the number N is a positive integer greater than or equal to 1; Construct an N+1 dimensional intent container, wherein the number of dimensions of the N+1 dimensional intent container is N+1, the first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension; The N+1 dimensional intent container is projected into each dimension in a dimensionality reduction manner to generate multiple slice interfaces; The system operator obtains the intent annotation information on the multiple slice interfaces, and performs comprehensive processing on the intent annotation information on the multiple slice interfaces to restore the complete intent description in the N+1 dimensional intent container. The complete intent description is compiled into executable instructions for the target system.

2. The method according to claim 1, characterized in that, Identifying multiple independent descriptive dimensions of the target system and determining the number N of these multiple independent descriptive dimensions includes: Enumerate all state variables of the target system and construct a set of state variables; Perform orthogonal decomposition on the variables in the state variable set to eliminate linearly correlated variables and obtain an orthogonal basis set; Extract the minimum independent dimension set from the orthogonal basis set. The number of dimensions in the minimum independent dimension set is the number N of the multiple independent descriptive dimensions.

3. The method according to claim 1, characterized in that, Construct an N+1 dimensional intent container, including: An N-dimensional spatial coordinate system is constructed using the identified N independent descriptive dimensions as coordinate axes; Based on the N-dimensional spatial coordinate system, add the (N+1)th dimension as the intent annotation dimension to construct an N+1-dimensional intent space; In the N+1 dimensional intent space, an intent description vector is defined as follows: in, to These are the coordinates of the first N dimensions. These are semantic annotation values ​​used to characterize the intent of the system operator at the corresponding N-dimensional spatial location. The coordinate values ​​of the dimension to be labeled.

4. The method according to any one of claims 1 to 3, characterized in that, The N+1 dimensional intent container is projected in a dimension-reduced manner across various dimensions to generate multiple sliced ​​interfaces, including: Determine the projection direction corresponding to each dimension of the N+1 dimensional intent container, with each projection direction perpendicular to a coordinate axis; For each projection direction, the points in the N+1 dimensional intention container are projected onto a hyperplane spanned by the dimensions other than the vertical coordinate axis, generating an N-dimensional slice. Each of the N-dimensional slices is converted into a two-dimensional interactive interface, which is used to receive the intention annotations of the system operator.

5. The method according to claim 4, characterized in that, The system operator obtains intent annotation information from the multiple slice interfaces, processes this intent annotation information comprehensively, and restores it to a complete intent description in the N+1-dimensional intent container, including: Obtain the two-dimensional annotation information input by the system operator on each slice interface; Based on the projection transformation matrix corresponding to each slice interface, the two-dimensional annotation information is inversely projected and mapped back to the local intent vector in the N+1-dimensional intent space; By fusing and calculating the local intent vectors corresponding to all the sliced ​​interfaces, a complete intent description vector is obtained: in, To provide a complete intent description vector, For the first Preset confidence weights for each slice of the interface. For the first Local intent vectors of each slice interface.

6. The method according to claim 5, characterized in that, The complete intent description is compiled into executable instructions for the target system, including: Identify the type of the target system, select the corresponding compilation rule base according to the type of the target system, input the complete intent description vector into the compilation rule base, and generate an executable instruction sequence for the target system; The types include physical space systems, creative content systems, software systems, manufacturing process systems, human intention systems, organizational management systems, medical clinical systems, legal compliance systems, music creation systems, education and learning systems, and supply chain logistics systems.

7. The method according to claim 6, characterized in that, When the type is a creative content system, the spatial structure parameters and style parameters in the complete intent description vector are separated, and multiple versions of executable instruction sequences are generated by replacing the style parameters. The spatial structure parameters remain unchanged.

8. A system control device based on an N+1 dimensional intent container, characterized in that, The device includes: The identification module is used to identify multiple independent descriptive dimensions of the target system and determine the number N of the multiple independent descriptive dimensions, wherein the number N is a positive integer greater than or equal to 1; The intent container construction module is used to construct an N+1 dimensional intent container, wherein the number of dimensions of the N+1 dimensional intent container is N+1, the first N dimensions correspond one-to-one with the N independent description dimensions of the target system, and the N+1th dimension is the intent annotation dimension; The dimension reduction projection module is used to project the N+1 dimensional intent container into various dimensions to generate multiple slice interfaces. The intent description restoration module is used to obtain the intent annotation information of the system operator on the multiple slice interfaces, and to comprehensively process the intent annotation information of the multiple slice interfaces to restore it into a complete intent description in the N+1 dimensional intent container. An instruction compilation module is used to compile the complete intent description into executable instructions for the target system.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

Images