Artificial intelligence-based game rapid production method and game production cloud platform

Abstract

The application relates to a game rapid production method and a game production cloud platform, which comprises the following steps: binding received natural language requirement description data, sketch trajectory data and real-time voice stream into long-term storage associated data objects, and assigning a globally unique ID and a version update record identifier to the data objects; according to the response urgency of the instructions in the data objects, different data transmission channels are assigned; through a time-space synchronization engine, voice features, sketch structure features and text semantic features of the same data object are matched at a time point, and a unified instruction vector is output; a pre-trained game content generation model is called to convert the unified instruction vector into a game element which can be independently stored; modification operation of a player on the game element is monitored, modification difference data is extracted, a generation model optimization module is trained, and a version record is updated. The application realizes cross-session state inheritance of NPCs / scenes, dynamic preemptive scheduling of computing power resources, and reverse correction of AI model effects by player operation data.

Description

Technical Field

[0001] This application relates to the technical field of game production, and in particular to a rapid game production method based on artificial intelligence and a game production cloud platform. Background Technology

[0002] With the booming development of the video game industry and the increasingly diversified and frequent demands of users for game content, game development efficiency and cost control have become core competitive factors in the industry. In recent years, breakthroughs in artificial intelligence (AI) technology have provided automated tools for game development, and some tools and platforms aimed at accelerating game development have emerged in the market, such as template-based rapid development frameworks, visual script editing tools, and online game asset libraries.

[0003] However, while current AI game creation platforms can automatically generate basic content such as scenes and characters, they have fundamental bottlenecks. For example, NPC behaviors and scene parameters cannot be persisted across sessions, requiring re-initialization for each player interaction; they cannot dynamically schedule based on real-time operations, resulting in latency exceeding 2000ms during high concurrency; and player corrections only trigger partial recompilation without reverse-engineering the underlying model, leading to repeated errors.

[0004] Based on the aforementioned shortcomings, there is an urgent need to overcome the bottlenecks of traditional technologies and build a new generation of game production engine that supports state persistence, elastic resource allocation, and reverse model optimization. Summary of the Invention

[0005] To address at least one of the aforementioned technical problems, this application provides an artificial intelligence-based method for rapid game creation and a game creation cloud platform.

[0006] Firstly, this application provides a method for rapid game production based on artificial intelligence, employing the following technical solution: Perform the following steps on the cloud server: The received natural language requirement description data, sketch trajectory data, and real-time voice stream are bound into a long-term associated data object, and a globally unique ID and version update record identifier are assigned to the data object. Differentiated data transmission channels are allocated based on the urgency of the instructions in the data object; The spatiotemporal synchronization engine performs time-point matching of the speech features, sketch structure features, and text semantic features of the same data object, and outputs a unified instruction vector. A pre-trained game content generation model is invoked to transform a unified instruction vector into independently storable game elements, including NPC elements, scene structure elements, and task flow elements. Monitor player modifications to game elements, extract modification difference data, train and generate a model optimization module, and update version records.

[0007] By adopting the above technical solutions, cross-session inheritance of NPCs / scenes is achieved through associating data objects and version records. For example, NPCs remember players' tactical preferences from the previous day, thus solving the problem of state persistence. Simultaneously, differentiated transmission channels dynamically allocate bandwidth according to urgency, breaking through resource allocation rigidity. Furthermore, modifying differentiated data into a closed loop within the model optimization module significantly reduces similar errors, enabling reverse model optimization. In summary, this achieves the effects of cross-session state inheritance of NPCs / scenes, dynamic preemptive scheduling of computing resources, and reverse correction of AI models using player operation data.

[0008] In one possible implementation, the step of allocating differentiated data transmission channels based on the urgency of the instructions in the data object specifically includes: The speech stream is segmented into semantic segments, and key anchor points of the sketch trajectory are associated with the speech feature matching. Perform grammatical structure analysis on the text description, extract core action words, and map them to the central area of ​​the sketch structure.

[0009] By adopting the above technical solution and utilizing multimodal instruction correlation, the timing mismatch of multi-source signals is eliminated, and the fusion error rate is greatly reduced.

[0010] In one possible implementation, the step of invoking a pre-trained game content generation model to transform a uniform instruction vector into independently storable game elements includes the following NPC element construction: Extract player behavior preference features from historical interaction records; Set a behavior decay coefficient with an initial value of 1.0. The decay is 0.1 every 24 hours without interaction. When the coefficient is less than 0.5, NPC active interaction behavior is triggered.

[0011] By adopting the above technical solution and utilizing NPC behavior entity construction, the continuous evolution of NPC behavior can be achieved.

[0012] In one possible implementation, the steps of monitoring player modifications to game elements, extracting modification difference data, training and generating a model optimization module, and updating version records include: When players modify NPC dialogue logic, the semantic differences before and after the modification are recorded; By inputting differential data into a training process that incorporates adversarial mechanisms, the ability of the generative model to discriminate dialogue logic is enhanced.

[0013] By adopting the above technical solution and utilizing reverse optimization of dialogue logic, the problem of repeated dialogue errors can be overcome.

[0014] One possible implementation also includes: When a player logs in again, the latest version of the data object is loaded based on the global ID; Automatically replay key changes in the version update for players to confirm.

[0015] By adopting the above technical solution and utilizing cross-session state synchronization, it is ensured that players inherit the complete game state each time they log in.

[0016] Secondly, this application provides a game production cloud platform system, including: The instruction inference cluster includes parallel speech processors, sketch parsers, and text analyzers. The modules are connected through a dynamic bandwidth bus, which adjusts the data transmission bandwidth between modules in real time according to the data type. An element repository stores game elements and their version records. The game elements include a basic information area, a behavior record area, and a structural change area. The resource scheduler has a built-in real-time load detector that dynamically reallocates GPU resources based on the number of players and the type of operation. Its resource allocation priority strategy is: voice processing > real-time combat > scene generation. The feedback learning gateway includes a player action difference extractor and a model fine-tuning adapter, which convert player actions into model training data.

[0017] By adopting the above technical solutions, the signal blocking problem is solved by using an instruction processing cluster, parallel processors and dynamic bus; the voice processing latency is effectively reduced compared to static allocation by using a resource scheduler and real-time load detection; and the feedback learning gateway is used so that player operations directly drive model fine-tuning, reducing manual debugging costs.

[0018] In one possible implementation, the dynamic allocation strategy of the resource scheduler is: When the amount of voice interaction exceeds the threshold, the GPU resource weight of the scene generation module will be reduced to 10%, while the weight of the voice processing module will be increased to 70%. When combat test requests surge, enable degraded rendering mode and free up 20% of video memory for behavioral logic processing.

[0019] By adopting the above technical solutions and utilizing dynamic resource allocation, the resource utilization rate in high-concurrency scenarios is effectively improved.

[0020] In one possible implementation, the player action difference extractor of the feedback learning gateway performs: Extract movement trajectory optimization data from player dragging NPC paths; Extract the difficulty change gradient from the player's operation record of adjusting the level difficulty.

[0021] By adopting the above technical solutions and using operational difference extraction, player intentions can be accurately captured, and model-oriented optimization can be achieved. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a rapid game creation method based on artificial intelligence, provided as an embodiment of this application.

[0023] Figure 2 This is a schematic diagram of the structure of the game creation cloud platform provided in an embodiment of this application.

[0024] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0025] The technical solutions in this application will now be described with reference to all the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0026] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. "And / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Furthermore, in the description of the embodiments of this application, "plural" or "multiple" refers to two or more than two.

[0027] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0028] The terminology used in the following embodiments is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to also include expressions such as “one or more,” unless the context clearly indicates otherwise. It should also be understood that in the following embodiments of this application, “at least one” and “one or more” refer to one, two, or more than two.

[0029] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "one embodiment," "some embodiments," "another embodiment," "other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0030] This application provides a method for rapid game creation based on artificial intelligence, executed by an electronic device. This electronic device can be a standalone physical electronic device, a cluster of multiple physical electronic devices, a distributed system, or a cloud electronic device providing cloud computing services. This application does not impose limitations on this method. Figure 1 As shown, the method includes performing the following steps on a cloud server: S1. Bind the received natural language requirement description data, sketch trajectory data, and real-time voice stream into a long-term storage associated data object, and assign a globally unique ID and version update record identifier to the data object.

[0031] Specifically, the cloud server is equipped with three types of data receiving interfaces: natural language descriptions are transmitted to the text buffer via the network port; hand-drawn sketch trajectories are captured by the digitizer and stored in the trajectory buffer as coordinate sequences; and voice streams are converted into digital signal packets by the microphone array.

[0032] Furthermore, three types of data input objects are bound to a processor, which includes: Unique ID generator: Generates a 128-bit identity code using a hardware random source; Version recorder: Creates a version chain start marker and change flag in the storage chip; Data association unit: Three types of data are merged and stored through a high-speed electronic switch, and the identity code and version number are written in the header of the storage area.

[0033] Furthermore, the bound data packet is stored in non-volatile memory (such as a flash memory array) via a direct-access controller, and the storage location is recorded in an address mapping table. When a new data packet is ready, an interrupt signal is triggered to notify the channel allocation module to prepare for reception.

[0034] S2. Allocate differentiated data transmission channels based on the urgency of the instructions in the data object.

[0035] Specifically, the data transmission channel allocation system includes three levels of control: Priority determination unit: Voice data triggers a red alarm signal and directly connects to the dedicated high-speed channel; sketch data triggers a yellow alarm signal and enters the buffer channel with flow control; text data goes through the green ordinary channel.

[0036] Channel controller: Voice channels are allocated independent physical lines to ensure transmission speed; sketch channels are marked with bandwidth protection; text channels use high-capacity transmission protocols.

[0037] Dynamic regulator: Monitors channel congestion in real time and automatically compresses text channel bandwidth when voice latency exceeds the limit.

[0038] Furthermore, the speech data is segmented into meaningful segments and matched to key positioning points in the sketch using a voiceprint recognition chip; the text data is processed by a syntax parser to extract core verbs and mapped to the central coordinate area of ​​the sketch.

[0039] S3. The spatiotemporal synchronization engine performs time-point matching of the speech features, sketch structure features, and text semantic features of the same data object, and outputs a unified instruction vector.

[0040] Specifically, the spatiotemporal synchronization engine includes: Clock reference source: A high-precision clock chip provides a synchronized time marker.

[0041] Feature processing unit: Speech feature extractor generates speaker graph; sketch analyzer outputs structural contour data; text processor generates word meaning code.

[0042] Alignment buffer: Three types of feature data with time stamps are stored in a dual-port memory, and the time window controller performs feature matching within a 50ms tolerance range.

[0043] Furthermore, the successfully matched feature data is input into the splicing circuit to generate a combined instruction data packet. When a timing misalignment occurs, the synchronization error counter triggers a re-matching instruction.

[0044] S4. Call the pre-trained game content generation model to transform the unified instruction vector into game elements that can be stored independently.

[0045] The game elements include NPC elements, scene structure elements, and task flow elements.

[0046] Specifically, the game content generation model includes: The combined instruction data packet is transmitted to the AI ​​processing chip via a high-speed bus.

[0047] The chip has three parallel working areas: a character generation area, which outputs character behavior parameters (including decay counters and preference records); a scene generation area, which generates scene coordinate data and collision detection templates; and a task generation area, which compiles task trigger markers and branch selection tables.

[0048] The generated data is written to the element database through the storage controller: the basic information area stores the creator's identity and generation time in the storage chip; the behavior record area stores the interaction history in the circular storage area; and the structure record area stores scene change records using a tree index.

[0049] Furthermore, the character behavior decay counter automatically decrements daily, and triggers the interaction event generation circuit when the value falls below a threshold.

[0050] S5. Monitor player modifications to game elements, extract modification difference data, train and generate a model optimization module, and update version records.

[0051] Specifically, the player feedback optimization module includes: Operation monitoring circuit: Player-modified operations are converted into electronic operation codes.

[0052] Difference extractor: Dialogue modification, calculates the difference value before and after text modification; Path adjustment, records the amount of change in movement trajectory.

[0053] Model updater: When the difference data is stored in the training cache, the model update process is to lock the original model parameter storage area, load the difference dataset, and start the parameter update state machine.

[0054] Furthermore, the version update employs "copy-on-write" technology, where new version data is stored in a new storage area, and the version chain pointer automatically jumps. When a player logs in again, their identity code directly locates the latest data, and the version comparator generates a change report which is sent to the display interface.

[0055] In summary, hardware-level data binding and non-volatile storage principles are used to achieve permanent preservation of game state (cross-session inheritance); a three-level alarm channel and dynamic bandwidth adjustment principle are used to achieve rapid response to voice commands (latency ≤200ms); clock synchronization and dual-port storage matching principles are used to achieve accurate fusion of multi-source commands and reduce error rate; three-way parallel generation and dedicated storage principles are used to achieve adaptive evolution of NPCs; and differential electronic computing and copy-on-write principles are used to achieve intelligent model optimization and reduce error rate.

[0056] Based on this, it is possible to achieve cross-session state inheritance of NPCs / scenes, dynamic preemptive scheduling of computing resources, and reverse correction of AI models using player operation data.

[0057] In some embodiments, in order to eliminate timing mismatch of multi-source signals and significantly reduce the fusion error rate, S2 specifically includes: S201. Segment the speech stream into semantic segments and associate them with key anchor points of the sketch trajectory through sound feature matching.

[0058] Specifically, the speech stream is input to the speech segmentation module via an audio interface, and this module includes: Endpoint detection circuit: An energy threshold comparator is used to monitor the voice amplitude in real time. When the amplitude is below the silence threshold for 5 consecutive ms, a segmented pulse signal is output.

[0059] Intent segment generator: The segmented pulse triggers the storage controller to store the speech stream in segments into independent storage blocks of dual-port RAM.

[0060] Voiceprint matcher: A lookup table (LUT) pre-stores user voiceprint feature codes. It receives speech segments, outputs similarity values ​​to the comparator, and activates the association signal when a match is successful.

[0061] Furthermore, the associated signal triggers the coordinate binding unit, which writes the storage block address of the current speech segment into the key anchor point coordinate register of the sketch trajectory.

[0062] Furthermore, anchor points are marked by the sketch preprocessing module, including trajectory inflection point coordinates (identified by curvature calculation circuitry) and user-marked points (captured by click event detectors). The binding results are stored in an association mapping table.

[0063] S202. Perform grammatical structure analysis on the text description, extract core action words, and map them to the central area of ​​the sketch structure.

[0064] Specifically, the text data input syntax parsing unit, hardware execution flow: Part-of-speech tagger: Identifies verbs / nouns using a finite state machine, and tags verbs as candidate action words.

[0065] The backbone extractor filters core verbs based on a dependency rule base (stored in ROM). For example, when receiving "build [object] at [location]", it extracts "build"; when receiving "move [character] to [location]", it extracts "move".

[0066] Coordinate mapper: Inputs the core verb encoding into the address decoder and outputs the corresponding sketch area coordinate range.

[0067] Furthermore, the sketch structure is divided using a region meshing processor, which divides the bounding box of the sketch trajectory into an 8×8 grid. The high-density region is marked as the central region by the edge density detection circuit, and the matching result of the coordinate range of the core verb and the central region grid is stored in the central mapping register.

[0068] In summary, hardware endpoint detection and voiceprint LUT matching are used to accurately associate voice commands with spatial locations; finite state machine syntax parsing and grid partitioning are used to dynamically bind text commands to core regions.

[0069] In some embodiments, to enable the continuous evolution of NPC behavior, the construction of NPC elements in S4 includes: S401. Extract player behavior preference features from historical interaction records.

[0070] Specifically, the historical interaction record memory (such as a DDR4 memory module) outputs the raw operation code stream to the behavior feature extraction circuit, which includes... Operation Classifier: Identifies player operation type (attack / defense / cooperation) through instruction decoder and outputs type label pulse.

[0071] Frequency counter: Accumulates and counts the pulses of each type of operation over 24 hours.

[0072] Priority encoder: compares the operation count values ​​of various types and encodes the most frequent type into a 3-bit preference feature code.

[0073] Furthermore, the preferred feature code is written into the feature vector register group, such as 001 for attack type, 010 for defense type, and 100 for cooperative type. The register output is connected to the mode selection pin of the NPC behavior generator, controlling the subsequent behavior logic generation path.

[0074] S402. Set the behavior decay coefficient, with an initial value of 1.0. The decay is 0.1 every 24 hours without interaction. When the coefficient is <0.5, the NPC will actively interact.

[0075] Specifically, the behavior decay system consists of the following hardware: Real-time clock module: Uses DS3231 chip to generate 24-hour timing pulses.

[0076] Attenuation control unit: includes an initial value register, which writes an 8-bit binary value "11111111" (corresponding to an attenuation coefficient of 1.0) upon power-on reset; a decrementer that performs an arithmetic right shift (equivalent to ×0.9) for each received timing pulse; and a threshold comparator that uses an LM393 chip to compare the attenuation register output with the threshold "01111111" (set to 0.5).

[0077] Furthermore, when the comparator outputs a low level (behavior attenuation coefficient < 0.5), the NPC interaction enable circuit is triggered. The enable signal turns on the power gate of the NPC dialogue generator, and the behavior selector automatically loads the preset interaction script (stored in ROM) and sends the interaction request data packet to the player terminal through the UART interface.

[0078] In summary, this achieves the effects of quantifying player behavior preferences, avoiding software counting delays, enabling NPC behavior to evolve autonomously over time, and realizing an NPC proactive interaction mechanism.

[0079] In some embodiments, to address the problem of recurring dialogue errors, S5 specifically includes: S501. When a player modifies the NPC dialogue logic, record the semantic differences before and after the modification.

[0080] Specifically, player-modified actions are generated via input devices (such as keyboards / gamepads) and input into a dialogue difference recording module, which includes: The capture unit is modified so that the original dialogue text is stored in storage area A of the dual-port RAM, and the modified text is written to storage area B in real time.

[0081] The difference calculation circuit compares the contents of the A / B area byte by byte with the word comparison unit, and the difference position marker writes the different byte addresses into the FIFO queue.

[0082] The semantic vector generator takes a word vector lookup table as input for the differential text fragments and outputs a 32-bit differential feature vector, which is then stored in the differential data register.

[0083] Furthermore, the recording process triggers a version marking circuit, creating a new node in the version chain memory. The node header stores a difference vector address pointer and a timestamp counter value.

[0084] S502. Input the differential data into the training process that includes adversarial mechanisms to enhance the dialogue logic discrimination ability of the generative model.

[0085] Specifically, the hardware for training the adversarial mechanism includes: The training data selector has two channels: channel 0, which is the original generated dialogue vector (read from the NPC element storage area); and channel 1, which is the player-modified difference vector (read from the difference register). The selection signal is output alternately by the adversarial controller at a 1:1 ratio.

[0086] It also includes a discrimination capability enhancer, a dual-input comparator for comparing the similarity of channel 0 / 1 vectors, and a weight update state machine for triggering parameter adjustment when the similarity is greater than a threshold, such as freezing the original model weight storage area, activating the negative sample training path, and adjusting the discrimination threshold voltage comparator.

[0087] It also includes a feedback loop to reinforce the writing of results into the model discrimination flag register and update the dialog generation logic gate array via the system bus.

[0088] Furthermore, the training process is accelerated by hardware, with the difference vectors being directly transmitted to the training buffer via the DMA channel, and the adversarial controller clock frequency being increased to three times that of the basic mode.

[0089] In summary, this achieves the effects of accurately capturing dialogue modifications, avoiding real-time calculation delays, enhancing error recognition capabilities, and implementing an adversarial reinforcement mechanism.

[0090] In some embodiments, to ensure that players inherit the complete game state each time they log in, this method further includes: S6. When a player logs in again, the latest version of the data object is loaded based on the global ID.

[0091] Specifically, when logging in, players enter an identification code (such as a QR code / fingerprint), which is then matched against a global ID database via an ID verification circuit. This verification circuit includes: Address locator: Inputs a 128-bit global ID into the computing chip (such as the SHA-256 hardware accelerator), and outputs the value as a storage address index.

[0092] Version loading unit: The index address is located by the address decoder to the version chain head pointer in the non-volatile memory, and the latest version data package (including NPC status + scene coordinates + task progress) is loaded into the cache through the DMA controller.

[0093] State recovery circuit: Cache data is distributed to each processing module via the system bus, historical decay coefficients are written into the NPC behavior register group, and the scene coordinate latch is loaded with the latest position data.

[0094] Furthermore, the loading process enables a data validator, and a cyclic redundancy check circuit verifies the integrity of the version data. Once the verification is successful, a ready signal is sent to the game engine.

[0095] S7: Automatically replays key changes in version updates for players to confirm.

[0096] Specifically, the version replay system includes: Change point extractor: The version difference tree memory outputs a node tag stream (using a B+ tree hardware accelerator), and the key node filter identifies nodes with a change magnitude > 20% (such as NPC behavior mutation points).

[0097] Playback data generation unit: During scene changes, coordinate difference data is input into the 3D rendering pipeline to generate scene evolution animation frames; during behavior changes, state machine transition records drive the NPC action generator to output behavior demonstration scripts.

[0098] Confirmation Interface: The animation frame sequence is stored in the video frame buffer and output to the player's monitor through the HDMI controller. The state of the confirmation button is read into the confirmation flag register through the interface.

[0099] Furthermore, when an acknowledgment signal is detected, the version synchronization circuit performs an atomic write operation, writing the playback node marker into the version acknowledgment record area.

[0100] The game production cloud platform provided in the embodiments of this application is described below. The game production cloud platform described below can be referred to in correspondence with the artificial intelligence-based rapid game production method described above.

[0101] refer to Figure 2 The game development cloud platform includes: Instruction inference cluster 1 contains a parallel speech processor (including ADC sampling circuit), a sketch parser (with coordinate trajectory buffer), and a text analyzer (with built-in word vector lookup table). The modules are connected through a dynamic bandwidth bus, which adjusts the data transmission bandwidth between modules in real time according to the data type.

[0102] Specifically, the modules are interconnected via a dynamic bandwidth control bus, as follows: The bus controller monitors data type identification codes in real time, such as voice data triggering a high-priority flag (red signal light), sketch priority (yellow signal light), and text low-priority flag (green signal light).

[0103] The electronic switch array switches physical paths according to the status of the signal lights. For example, in the red state, 16 data channels (total bandwidth 80Gbps) are allocated, in the yellow state, 8 data channels (40Gbps) are allocated, and in the green state, 4 data channels (20Gbps) are allocated.

[0104] The traffic monitor tracks the backlog of data in each module and triggers a bandwidth doubling signal when the cache exceeds the threshold.

[0105] Furthermore, when the speech processor outputs voiceprint feature packets, it automatically elevates the bus priority and preempts the text processor channel resources via an interrupt signal. The measured speech processing latency is consistently ≤200ms.

[0106] Element Repository 2 is used to store game elements and their version records. Game elements include a basic information area, a behavior record area, and a structural change area.

[0107] Specifically, in the basic information area, such as the creator ID and timestamp, the data is stored in a high-speed read / write chip; in the behavior record area, such as the NPC interaction summary, a circular storage technology is used, where the write pointer automatically overwrites the earliest data in a circular manner; in the structural change area, such as scene coordinate changes, a version tree storage is used.

[0108] Resource Scheduler 3 has a built-in real-time load detector that dynamically reallocates GPU resources based on the number of players and the type of operation. Its resource allocation priority strategy is: voice processing (default 60%) > real-time combat (default 25%) > scene generation (default 15%).

[0109] Feedback Learning Gateway 4 includes a player action difference extractor and a model fine-tuning adapter, which convert player action modifications into model training data.

[0110] The system employs an instruction processing cluster 1, which combines parallel processors with a dynamic bus to solve the signal blocking problem; a resource scheduler 3 is used, and real-time load detection effectively reduces voice processing latency compared to static allocation; a feedback learning gateway 4 is used, where player operations directly drive model fine-tuning, reducing manual debugging costs.

[0111] In some embodiments, to effectively improve resource utilization in high-concurrency scenarios, the dynamic allocation strategy of resource scheduler 3 is as follows: When the amount of voice interaction exceeds the threshold, the GPU resource weight of the scene generation module is reduced to 10%, while the weight of the voice processing module is increased to 70%.

[0112] When combat test requests surge, enable degraded rendering mode and free up 20% of video memory for behavioral logic processing.

[0113] In some embodiments, to accurately capture player intent and achieve model-oriented optimization, the player action difference extractor of the feedback learning gateway 4 is executed as follows: Extract movement trajectory optimization data from the player's dragging of the NPC's path.

[0114] Extract the difficulty change gradient from the player's operation record of adjusting the level difficulty.

[0115] This application provides an electronic device, such as... Figure 3 As shown, Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 The illustrated electronic device 300 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of this electronic device 300 does not constitute a limitation on the embodiments of this application.

[0116] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in connection with the embodiments of this application. Processor 301 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0117] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0118] The memory 303 may be a ROM (Read-Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM (Electrically Erasable Programmable Read-Only Memory), a CD-ROM (Compact Disc Read-Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0119] The memory 303 is used to store application code that executes the scheme of the embodiments of this application, and its execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0120] Among them, electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments described in this application.

[0121] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the steps of the above-described AI-based rapid game creation method.

[0122] Since the embodiments of the computer-readable storage medium portion correspond to the embodiments of the method portion, please refer to the description of the embodiments of the method portion for the embodiments of the computer-readable storage medium portion.

[0123] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures 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, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0124] The above are only some embodiments of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for rapid game production based on artificial intelligence, characterized in that, Perform the following steps on the cloud server: The received natural language requirement description data, sketch trajectory data, and real-time voice stream are bound into a long-term storage associated data object, and a globally unique ID and version update record identifier are assigned to the data object. The cloud server is equipped with three types of data receiving interfaces and three types of data input object binding processors. The processor includes a unique ID generator, a version recorder, and a data association unit. The bound data packets are stored in non-volatile memory through a direct storage controller. Differentiated data transmission channels are allocated based on the urgency of the instructions in the data object; The data transmission channel allocation system comprises three levels of control: Priority Judgment Unit: Real-time voice streams triggering a red alarm signal directly connect to the dedicated high-speed channel; sketch trajectory data triggering a yellow alarm signal enters the buffer channel with flow control; natural language demand description data goes through the green ordinary channel; Channel Controller: The dedicated high-speed channel is allocated an independent physical line to ensure transmission speed; the buffer channel is set with a bandwidth guarantee mark; the green ordinary channel adopts a high-capacity transmission protocol; Dynamic Adjuster: Real-time monitoring of channel congestion; when voice latency exceeds the standard, the bandwidth of the green ordinary channel is automatically compressed. The spatiotemporal synchronization engine performs time point matching on the real-time voice stream, sketch trajectory data, and natural language requirement description data of the same data object, and outputs a unified instruction vector. The spatiotemporal synchronization engine includes a clock reference source, a feature processing unit, and an alignment buffer. The three types of feature data are stored in a dual-port memory with time stamps, and the time window controller performs feature matching within a 50ms tolerance range. A pre-trained game content generation model is invoked to transform a unified instruction vector into independently storable game elements, including NPC elements, scene structure elements, and task flow elements. The construction of NPC elements includes extracting player behavior preference features from historical interaction records; Set a behavior decay coefficient with an initial value of 1.

0. The decay is 0.1 every 24 hours without interaction. When the coefficient is less than 0.5, trigger NPC active interaction behavior. Monitor player modifications to game elements, extract modification difference data, train and generate a model optimization module, and update version records; when players modify NPC dialogue logic, record semantic difference data before and after the modification; input the semantic difference data into a training process that includes adversarial mechanisms.

2. The method according to claim 1, characterized in that, Also includes: The speech stream is segmented into semantic segments, and key anchor points of the sketch trajectory are associated with the speech feature matching. Perform grammatical structure analysis on natural language demand description data, extract core action words, and map them to the central region of the sketch structure.

3. The method according to claim 1, characterized in that, Also includes: When a player logs in again, the latest version of the data object is loaded based on the global ID; Automatically replay key changes in the version update for players to confirm.

4. A game development cloud platform system, characterized in that, include: The instruction inference cluster includes parallel speech processors, sketch parsers, and text analyzers. The modules are connected through a dynamic bandwidth bus, which adjusts the data transmission bandwidth between modules in real time according to the data type. An element repository stores game elements and their version records. The game elements include a basic information area, a behavior record area, and a structural change area. The resource scheduler has a built-in real-time load detector that dynamically reallocates GPU resources based on the number of players and the type of operation. Its resource allocation priority strategy is: voice processing is greater than real-time combat, and real-time combat is greater than scene generation. The feedback learning gateway includes a player action difference extractor and a model fine-tuning adapter, which convert player actions into model training data.

5. The system according to claim 4, characterized in that, The dynamic allocation strategy of the resource scheduler is as follows: When the amount of voice interaction exceeds the threshold, the GPU resource weight of the scene generation module is reduced to 10%, while the weight of the voice processing module is increased to 70%. When combat test requests surge, enable degraded rendering mode and free up 20% of video memory for behavioral logic processing.

6. The system according to claim 5, characterized in that, The player action difference extractor of the feedback learning gateway executes: Extract movement trajectory optimization data from player dragging NPC paths; Extract the difficulty change gradient from the player's operation record of adjusting the level difficulty.

Images