Multi-objective dimension reduction execution compilation method and system based on four-dimensional spacetime intent container

Abstract

The application relates to the technical field of human-computer interaction, and provides a multi-target dimension reduction execution compiling method and system based on a four-dimensional space-time intention container. The application constructs a four-dimensional space-time container containing a three-dimensional space coordinate system and a time axis, the time axis carries path nodes and event nodes sharing a unified coordinate reference with the three-dimensional space coordinate system, users can express task intentions completely and unambiguously in a four-dimensional operation space, a unified three-dimensional space coordinate system in the four-dimensional space-time container is established, the precision of the space dimension is ensured, a complete four-dimensional intention description is formed, corresponding instruction formats are compiled and generated according to execution target types, and the repeated description cost of cross-field tasks is greatly reduced.

Description

Technical Field

[0001] This application relates to the field of human-computer interaction technology, and in particular to a multi-objective dimensionality reduction execution compilation method and system based on a four-dimensional spatiotemporal intent container. Background Technology

[0002] Human task intent encompasses seven semantic dimensions: object, action, sequence, condition, constraint, space, and time. Existing intent expression tools are all low-dimensional carriers, resulting in a dimensional gap. Specifically, natural language prompts are one-dimensional linear texts with high ambiguity; two-dimensional maps / GCS only support planar space, losing information such as height and time; three-dimensional modeling tools lack timelines and event logic, requiring a high level of expertise; and time-series scripts / code lack spatial dimensions, requiring programming skills.

[0003] High-dimensional manipulation and low-dimensional output is a mature paradigm in tool design, with Photoshop, CAD / BIM, and music DAWs all achieving industrial innovation based on this approach. However, the field of task intent expression still lacks a four-dimensional (three-dimensional space × time) manipulation container. Existing technologies such as digital twins, game engines, language-driven planning, traditional ground stations, and hand-drawn programming either lack sufficient dimensions, have too high a barrier to entry, offer limited output, or lack visual editing capabilities, making it difficult to fully construct and compile multi-target execution intents. Summary of the Invention

[0004] Therefore, it is necessary to provide a multi-objective dimensionality reduction execution compilation method and system based on a four-dimensional spatiotemporal intent container to address the above-mentioned technical problems.

[0005] A multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container, the method comprising the following steps: Construct a four-dimensional spatiotemporal container that includes a three-dimensional spatial coordinate system and a time axis. The three-dimensional spatial coordinate system takes the scene center as the origin and three orthogonal spatial planes as the container boundaries. The time axis loads and path nodes and event nodes that share a unified coordinate reference with the three-dimensional spatial coordinate system. Collect image data from at least two orthogonal perspectives of the target scene, map the image data to the corresponding spatial surface of the four-dimensional spatiotemporal container, and establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container; Acquire user intent annotations on the spatial planes and time axis of the four-dimensional spacetime container to form a complete four-dimensional intent description; Based on the type of execution target, the four-dimensional intent description is dimensionality-reduced according to the subset of dimensions required by the execution target, compiled to generate an instruction format that the execution target can directly execute, and the folded dimension information is marked during the dimensionality reduction process.

[0006] In one embodiment, establishing a unified three-dimensional spatial coordinate system within the four-dimensional spacetime container includes: When establishing a unified three-dimensional spatial coordinate system, alignment strategies are selected according to a preset priority order, and the confidence score of each alignment strategy is calculated. When the confidence score is lower than the threshold, the alignment strategy is automatically downgraded to the next priority. The alignment strategies, ranked from highest to lowest priority, are: GPS combined with compass, IMU inertial integration, visual feature point matching, and user-manual anchor point annotation.

[0007] In one embodiment, the user's intent annotation operations on the spatial plane and time axis of the four-dimensional spatiotemporal container are acquired to form a complete four-dimensional intent description, including: The intent annotation operations of the user on the spatial plane and time axis of the four-dimensional spatiotemporal container are obtained. The intent annotation operations include region annotation, path drawing, semantic input on the spatial plane, and path node timing setting and event node mounting on the time axis, forming a complete four-dimensional intent description. The intention annotation operations on the spatial plane and time axis will be stored together.

[0008] In one embodiment, the method for acquiring image data of the target scene includes mobile phone photography, drone acquisition, and map API.

[0009] In one embodiment, the execution objectives include movie rendering, game engines, drone flight control, robot motion control, industrial process control, and 4D printing.

[0010] In one embodiment, when marking a region on a spatial plane, the material type selected by the user is associated with the region, and a multi-dimensional physical attribute vector containing traversability, perceptible penetration, time-varying properties, hazard level, and applicable carrier type is automatically assigned; the attribute vector is used as a parameter of the constraint condition during dimensionality reduction compilation.

[0011] In one embodiment, image data from at least two orthogonal perspectives of the target scene are acquired, and the image data is mapped to the corresponding spatial surface of the four-dimensional spatiotemporal container to establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container. The method further includes: For spatial surfaces corresponding to viewpoints for which no image data has been acquired, the following strategies are used to complete the data: using a generative model to infer and generate the data based on the acquired spatial surfaces; replacing the data with a mirror image of the corresponding spatial surface; or having the user acquire additional data.

[0012] In one embodiment, when compiling instruction formats for at least two different execution targets, the dimensionality usage of each execution target is compared in the compilation summary, and the semantic equivalence relationships and information dimensionality reduction differences between the instruction formats of different execution targets are marked.

[0013] In one embodiment, when performing intention annotation operations such as region labeling or path drawing on a spatial plane, the orthogonal projection of the intention annotation operation on other spatial planes is calculated in real time, displayed synchronously in the form of semi-transparent auxiliary lines, and a time sequence mark corresponding to the current intention annotation operation is generated on the time axis.

[0014] A multi-objective dimensionality reduction execution compilation system based on a four-dimensional spatiotemporal intent container, the system comprising: The four-dimensional spatiotemporal container construction module is used to construct a four-dimensional spatiotemporal container that includes a three-dimensional spatial coordinate system and a time axis. The three-dimensional spatial coordinate system takes the scene center as the origin and three orthogonal spatial planes as the container boundaries. The time axis loads share path nodes and event nodes with the three-dimensional spatial coordinate system that have a unified coordinate reference. The unified coordinate system construction module is used to collect image data from at least two orthogonal perspectives of the target scene, map the image data to the corresponding spatial surface of the four-dimensional spatiotemporal container, and establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container. The four-dimensional intent description generation module is used to obtain the user's intent annotation operations on the spatial plane and time axis of the four-dimensional spatiotemporal container and form a complete four-dimensional intent description. The multi-target dimensionality reduction compilation module is used to reduce the dimensionality of the four-dimensional intent description according to the subset of dimensions required by the execution target based on the type of the execution target, compile it into an instruction format that the execution target can directly execute, and mark the folded dimension information during the dimensionality reduction process.

[0015] The aforementioned multi-objective dimensionality reduction execution compilation method and system based on a four-dimensional spatiotemporal intent container constructs a four-dimensional spatiotemporal container containing a three-dimensional spatial coordinate system and a time axis. The path nodes and event nodes that share a unified coordinate reference with the three-dimensional spatial coordinate system on the time axis enable users to express task intents completely and unambiguously in a four-dimensional operation space. By establishing a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container, the accuracy of spatial dimensions is guaranteed. By forming a complete four-dimensional intent description and compiling it into a corresponding instruction format according to the type of execution target, the cost of repetitive description of cross-domain tasks is significantly reduced. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container in one embodiment. Detailed Implementation

[0017] 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.

[0018] In one embodiment, a multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container is provided, the method comprising the following steps: A four-dimensional spatiotemporal container is constructed, which includes a three-dimensional spatial coordinate system and a time axis. The three-dimensional spatial coordinate system takes the scene center as the origin and three orthogonal spatial planes as the container boundaries. The time axis loads and path nodes and event nodes share a unified coordinate reference with the three-dimensional spatial coordinate system.

[0019] It should be noted that the complete four-dimensional structure of the four-dimensional spacetime container is spatial form × time change sequence × material properties × triggering conditions.

[0020] Collect image data from at least two orthogonal perspectives of the target scene, map the image data to the corresponding spatial surface of the four-dimensional spatiotemporal container, and establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container.

[0021] The system acquires the user's intent annotations on the spatial plane and time axis of the four-dimensional spacetime container, forming a complete four-dimensional intent description.

[0022] Based on the type of execution target, the four-dimensional intent description is dimensionality-reduced according to the subset of dimensions required by the execution target, compiled to generate an instruction format that the execution target can directly execute, and the folded dimension information is marked during the dimensionality reduction process.

[0023] It should be noted that, depending on the execution goal, the four-dimensional intent description will be folded or discarded, thereby reducing the dimensionality and converting the output format.

[0024] The multi-objective dimensionality reduction execution compilation method based on the four-dimensional spatiotemporal intent container described above constructs a four-dimensional spatiotemporal container containing a three-dimensional spatial coordinate system and a time axis. The path nodes and event nodes that share a unified coordinate reference with the three-dimensional spatial coordinate system on the time axis enable users to express their task intent completely and unambiguously in the four-dimensional operation space. By establishing a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container, the accuracy of the spatial dimension is guaranteed. By forming a complete four-dimensional intent description and compiling it into the corresponding instruction format according to the type of execution target, the cost of repetitive description of cross-domain tasks is significantly reduced.

[0025] In one embodiment, establishing a unified three-dimensional spatial coordinate system within the four-dimensional spacetime container includes: When establishing a unified three-dimensional spatial coordinate system, alignment strategies are selected according to a preset priority order, and the confidence score of each alignment strategy is calculated. When the confidence score is lower than the threshold, the alignment strategy is automatically downgraded to the next priority. The alignment strategies, ranked from highest to lowest priority, are: GPS combined with compass, IMU inertial integration, visual feature point matching, and user-manual anchor point annotation.

[0026] In this embodiment, a unified three-dimensional spatial coordinate system is constructed by setting an adaptive alignment strategy selection mechanism, thereby improving scene adaptability and matching flexibility, making the established unified three-dimensional spatial coordinate system more in line with the requirements and more accurate.

[0027] It should be noted that the scene spatial information can be decomposed into three orthogonal views, corresponding to the three main observation directions of the six-sided container. Specifically, the top view (the horizontal spatial layout of the scene from above) corresponds to the +Y and -Y planes; the side view (the vertical cross-sectional information from the side) corresponds to the +X and -X planes, used to express the height relationship and the distribution of vertical obstacles; and the front view (the depth and width information directly in front) corresponds to the +Z and -Z planes.

[0028] In one embodiment, the user's intent annotation operations on the spatial plane and time axis of the four-dimensional spatiotemporal container are acquired to form a complete four-dimensional intent description, including: The intent annotation operations of the user on the spatial plane and time axis of the four-dimensional spatiotemporal container are obtained. The intent annotation operations include region annotation, path drawing, semantic input on the spatial plane, and path node timing setting and event node mounting on the time axis, forming a complete four-dimensional intent description. The intention annotation operations on the spatial plane and time axis will be stored together.

[0029] It should be noted that user intent annotation operations on the spatial plane and time axis of the four-dimensional spacetime container can be performed through point selection, drawing, path drawing, and voice or text input. Point selection involves clicking on the spatial plane, recording the (x, y, z, t) coordinates of the point in the four-dimensional container coordinate system, and defining a point of interest; drawing involves smearing an area on the spatial plane, defining a material attribute annotation area (solid / liquid / gas / biological / energy field), and assigning it a five-dimensional physical attribute vector (traversability / perceptual penetration / time-varying nature / hazard level / applicable carrier type); path drawing involves drawing a trajectory curve on the spatial plane, mapping the curve to a sequence of four-dimensional path nodes, and automatically generating the corresponding temporal sequence on the time axis; voice or text input associates with the selected area or path node on the spatial plane to form natural language semantic annotation, and records the source through the modal tracing field of the UIP protocol.

[0030] In one embodiment, when marking a region on a spatial plane, the material type selected by the user is associated with the region, and a multi-dimensional physical attribute vector containing traversability, perceptible penetration, time-varying properties, hazard level, and applicable carrier type is automatically assigned; the attribute vector is used as a parameter of the constraint condition during dimensionality reduction compilation.

[0031] In one embodiment, acquiring the user's intent annotation operations on the spatial plane and time axis of the four-dimensional spatiotemporal container to form a complete four-dimensional intent description further includes: The path drawing operation generates a sequence of path nodes carrying spatial coordinates and timestamps in a four-dimensional spatiotemporal container; event nodes can be anchored by time offsets, attached to path nodes or independently attached to a specific moment on the time axis; when the spatial position or timing of a path node is modified, the effective trigger time of the event node anchored to the current path node is automatically updated.

[0032] It should be noted that during compilation, the path node sequence and event node sequence are compiled into a timing instruction stream of the target execution format.

[0033] In one embodiment, the method for acquiring image data of the target scene includes mobile phone photography, drone acquisition, and map API.

[0034] It should be noted that when the map API is selected instead of the top view, satellite images and corresponding geographic coordinates of the specified geographic area are obtained from the map service and automatically registered and aligned with the side view and front view manually taken by the user using GPS coordinates to establish a four-dimensional container space dimensional benchmark containing real geographic coordinates.

[0035] In one embodiment, the execution objectives include movie rendering, game engines, drone flight control, robot motion control, industrial process control, and 4D printing.

[0036] Understandably, the corresponding instruction formats are movie / animation rendering sequences (3D frames), game engine scene scripts, drone flight control instructions, robot motion sequences, industrial PLC instructions, and 4D printing path files.

[0037] In one embodiment, when the target is 4D printing, the dimensionality reduction compilation process uses all four dimensions of a four-dimensional spacetime container. Specifically, the geometric information of the three-dimensional spatial dimension (i.e., the three-dimensional spatial coordinate system) is converted into a layered printing path, the deformation sequence of the time dimension (i.e., the time axis) is converted into the material response coefficient distribution of each printing area, and the trigger condition parameters (temperature threshold, humidity threshold, stress threshold, etc.) of the material properties are converted into the trigger response parameters of the printing material. The compilation output includes a layered printing path file and a material ratio parameter file for each area, so that the printed object undergoes a predetermined spatial deformation in a time sequence when specific environmental trigger conditions are met.

[0038] In one embodiment, image data from at least two orthogonal perspectives of the target scene are acquired, and the image data is mapped to the corresponding spatial surface of the four-dimensional spatiotemporal container to establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container. The method further includes: For the spatial planes (bottom, back, left) corresponding to the viewpoints for which no image data has been acquired, the following strategies are used to complete the data: use a generative model to infer and generate the data based on the acquired spatial planes; replace the data with a mirror image of the corresponding spatial plane; or have the user acquire additional data.

[0039] In this embodiment, the spatial plane corresponding to the viewpoint of the uncollected image data is completed by the completion strategy, which reduces the image data acquisition requirements and can construct an accurate four-dimensional spatiotemporal container based on less image data.

[0040] In one embodiment, when compiling instruction formats for at least two different execution targets, the dimensionality usage of each execution target is compared in the compilation summary, and the semantic equivalence relationships and information dimensionality reduction differences between the instruction formats of different execution targets are marked.

[0041] In this embodiment, by marking the semantic equivalence relationships and information dimensionality reduction differences between the instruction formats of different execution targets, users can be aware of the degree of information dimensionality reduction.

[0042] In one embodiment, when performing intention annotation operations such as region labeling or path drawing on a spatial plane, the orthogonal projection of the intention annotation operation on other spatial planes is calculated in real time, displayed synchronously in the form of semi-transparent auxiliary lines, and a time sequence mark corresponding to the current intention annotation operation is generated on the time axis.

[0043] In this embodiment, corresponding intention annotation operations on other spatial planes are obtained through orthogonal projection, and corresponding temporal markers are generated on the time axis, thereby realizing three-way linkage visualization of space-space and space-time.

[0044] To verify the effectiveness of the present invention, practical applications in different scenarios are described.

[0045] Scenario 1: A performance company needs to choreograph a light show with 50 drones, requiring them to complete 15 formation changes within 5 minutes. A FourD-Container (four-dimensional spacetime container) is established in the center of the performance venue, covering a 200m × 200m × 100m area (performance space), with a timeline length of 300 seconds (5 minutes of performance). On the +Y plane (top view), the planar position of each formation of the 50 drones is defined by point selection (point selection generates 50 PathNode planar coordinates). The height (z-coordinate) of each drone is defined on the +X plane (side view), forming a complete 3D position set for each formation. 15 formation change moments are marked on the timeline, with each change point attached to the timeline as an EventNode. The trigger condition is "reaching this moment", and the action is "all drones switch to the next formation". The shortest collision-free path is automatically calculated for each drone between adjacent formation positions, generating its own PathNode sequence and allocating flight time slots on the timeline. The PathNode sequence (3D coordinates + timestamp) and EventNode (light color switching moment) of each drone are compiled into the corresponding MAVLink task file and distributed to each drone. Before translation, the spatiotemporal conflicts of 50 flight paths are automatically detected (safety distance check), and the timing of conflicting paths is automatically adjusted (path node timestamp offset) to ensure the safety of the performance.

[0046] Scenario 2: An engineer designs a 4D printed part. Its initial form is a flat plate, which automatically bends into an arc-shaped support when the temperature exceeds 60°C. Create a FourD-Container, defining the physical dimensions of the part (100mm × 50mm × 3mm) in the spatial dimension and the deformation process (0-30 seconds) in the time axis. In the initial form (t=0), create the initial flat geometry of the part within the six-sided container using top and side view photos; the side view shows the thickness distribution. Insert an EventNode at t=30s on the time axis, with the trigger condition: temperature > 60°C, and execute the action: transform the shape into the target arc shape. Define the target arc shape within the six-sided container (the side view shows the curvature and radius). In the top view, use a painting operation to divide the part into three regions: label the middle region as "High-response material (deformation coefficient 0.8)" and the two end regions as "Low-response material (deformation coefficient 0.2)", thus defining the deformation gradient distribution. Then, navigate to Initial Form → Spatial Geometry → Print Path; Time Series × Material Properties → Material Gradient Distribution (using high-response SMP material in the middle and low-response matrix material at both ends); and then apply the trigger condition → Material Parameter Labeling (response temperature 60°C). Outputs layered path files and material proportioning parameters that the printer can directly execute.

[0047] Scenario 3: A game designer (not a programmer) designs a guard level for an action game. The level has three NPC guards, and their patrol paths and response logic to player actions need to be defined. The designer uses a mobile phone to photograph the three-view model of the level prototype. The system generates a FourD-Container, displaying the level's geometry within its six sides. On the top view, the designer marks the alert zone (red, for NPCs to chase after the player), patrol zone (blue, for regular NPC patrols), and restricted area (gray, physical obstacles). A patrol path (closed polygon path) is drawn for NPC-A on the top view, generating a corresponding cyclic PathNode sequence on the timeline. NPC-A's patrol cycle is set to 20 seconds per lap. An EventNode is inserted: "Trigger condition: Player enters the alert zone," executing the action: "NPC-A, B, C switch to the pursuit path, target = player position"; another EventNode is inserted: "Trigger condition: Player leaves the alert zone for more than 10 seconds," executing the action: "NPC resumes patrol path." The FourD-Container is compiled into a Unity behavior tree script (patrol / pursuit state machine) and a level configuration file, which the game engine directly reads and executes.

[0048] It should be understood that, although Figure 1 The 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.

[0049] In one embodiment, a multi-objective dimensionality reduction execution compilation system based on a four-dimensional spatiotemporal intent container is provided, the system comprising: The four-dimensional spatiotemporal container construction module is used to construct a four-dimensional spatiotemporal container that includes a three-dimensional spatial coordinate system and a time axis. The three-dimensional spatial coordinate system takes the scene center as the origin and three orthogonal spatial planes as the container boundaries. The time axis load shares path nodes and event nodes with a unified coordinate reference with the three-dimensional spatial coordinate system.

[0050] The unified coordinate system construction module is used to collect image data from at least two orthogonal perspectives of the target scene, map the image data to the corresponding spatial surface of the four-dimensional spatiotemporal container, and establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container.

[0051] The four-dimensional intent description generation module is used to obtain the user's intent annotation operations on the spatial plane and time axis of the four-dimensional spatiotemporal container, and form a complete four-dimensional intent description.

[0052] The multi-target dimensionality reduction compilation module is used to reduce the dimensionality of the four-dimensional intent description according to the subset of dimensions required by the execution target based on the type of the execution target, compile it into an instruction format that the execution target can directly execute, and mark the folded dimension information during the dimensionality reduction process.

[0053] Specific limitations regarding the multi-objective dimensionality reduction execution compilation system based on a four-dimensional spatiotemporal intent container can be found in the limitations of the multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container mentioned above, and will not be repeated here. Each module in the aforementioned multi-objective dimensionality reduction execution compilation system based on a four-dimensional spatiotemporal intent container 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.

[0054] 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.

[0055] 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 multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container, characterized in that, The method includes the following steps: Construct a four-dimensional spatiotemporal container that includes a three-dimensional spatial coordinate system and a time axis. The three-dimensional spatial coordinate system takes the scene center as the origin and three orthogonal spatial planes as the container boundaries. The time axis loads and path nodes and event nodes that share a unified coordinate reference with the three-dimensional spatial coordinate system. Collect image data from at least two orthogonal perspectives of the target scene, map the image data to the corresponding spatial surface of the four-dimensional spatiotemporal container, and establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container; Obtain the user's intent annotations on the spatial planes and time axis of the four-dimensional spacetime container to form a complete four-dimensional intent description, including: The intent annotation operations of the user on the spatial plane and time axis of the four-dimensional spatiotemporal container are obtained. The intent annotation operations include region annotation, path drawing, semantic input on the spatial plane, and path node timing setting and event node mounting on the time axis, forming a complete four-dimensional intent description. And the intent annotation operations on the spatial plane and time axis will be stored together; When marking regions on a spatial plane, the material type selected by the user is associated with the region, and a multi-dimensional physical attribute vector containing traversability, perceptible penetration, time-varying properties, hazard level, and applicable carrier type is automatically assigned; the attribute vector is used as a parameter of the constraint condition during dimensionality reduction compilation. Based on the type of execution target, the four-dimensional intent description is dimensionality-reduced according to the subset of dimensions required by the execution target, compiled to generate an instruction format that the execution target can directly execute, and the folded dimension information is marked during the dimensionality reduction process.

2. The multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container according to claim 1, characterized in that, Establish a unified three-dimensional spatial coordinate system within the four-dimensional spacetime container, including: When establishing a unified three-dimensional spatial coordinate system, alignment strategies are selected according to a preset priority order, and the confidence score of each alignment strategy is calculated. When the confidence score is lower than the threshold, the alignment strategy is automatically downgraded to the next priority. The alignment strategies, ranked from highest to lowest priority, are: GPS combined with compass, IMU inertial integration, visual feature point matching, and user-manual anchor point annotation.

3. The multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container according to claim 1, characterized in that, Methods for acquiring image data of the target scene include mobile phone photography, drone data acquisition, and map APIs.

4. The multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container according to claim 1, characterized in that, The execution objectives include film rendering, game engines, drone flight control, robot motion control, industrial process control, and 4D printing.

5. The multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container according to claim 1, characterized in that, Acquire image data from at least two orthogonal viewpoints of the target scene, map the image data to the corresponding spatial surface of a four-dimensional spatiotemporal container, establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container, and also include: For spatial surfaces corresponding to viewpoints for which no image data has been acquired, the following strategies are used to complete the data: using a generative model to infer and generate the data based on the acquired spatial surfaces; replacing the data with a mirror image of the corresponding spatial surface; or having the user acquire additional data.

6. The multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container according to claim 1, characterized in that, When compiling instruction formats for at least two different execution targets, the compilation summary compares the dimensionality usage of each execution target and marks the semantic equivalence relationships and information dimensionality reduction differences between the instruction formats of different execution targets.

7. The multi-objective dimensionality reduction execution compilation method based on a four-dimensional spatiotemporal intent container according to claim 1, characterized in that, When performing intention annotation operations such as region labeling or path drawing on a spatial plane, the orthogonal projection of the intention annotation operation on other spatial planes is calculated in real time, displayed synchronously in the form of semi-transparent auxiliary lines, and a time sequence mark corresponding to the current intention annotation operation is generated on the time axis.

8. A multi-objective dimensionality reduction execution compilation system based on a four-dimensional spatiotemporal intent container, characterized in that, The system includes: The four-dimensional spatiotemporal container construction module is used to construct a four-dimensional spatiotemporal container that includes a three-dimensional spatial coordinate system and a time axis. The three-dimensional spatial coordinate system takes the scene center as the origin and three orthogonal spatial planes as the container boundaries. The time axis loads share path nodes and event nodes with the three-dimensional spatial coordinate system that have a unified coordinate reference. The unified coordinate system construction module is used to collect image data from at least two orthogonal perspectives of the target scene, map the image data to the corresponding spatial surface of the four-dimensional spatiotemporal container, and establish a unified three-dimensional spatial coordinate system within the four-dimensional spatiotemporal container. The four-dimensional intent description generation module is used to acquire the user's intent annotation operations on the spatial plane and time axis of the four-dimensional spatiotemporal container, forming a complete four-dimensional intent description, including: The intent annotation operations of the user on the spatial plane and time axis of the four-dimensional spatiotemporal container are obtained. The intent annotation operations include region annotation, path drawing, semantic input on the spatial plane, and path node timing setting and event node mounting on the time axis, forming a complete four-dimensional intent description. And the intent annotation operations on the spatial plane and time axis will be stored together; When marking regions on a spatial plane, the material type selected by the user is associated with the region, and a multi-dimensional physical attribute vector containing traversability, perceptible penetration, time-varying properties, hazard level, and applicable carrier type is automatically assigned; the attribute vector is used as a parameter of the constraint condition during dimensionality reduction compilation. The multi-target dimensionality reduction compilation module is used to reduce the dimensionality of the four-dimensional intent description according to the subset of dimensions required by the execution target based on the type of the execution target, compile it into an instruction format that the execution target can directly execute, and mark the folded dimension information during the dimensionality reduction process.

Images