Asynchronous time twist state synchronization method in multi-user VR co-simulation
By integrating client-side prediction with server-side methods, the problem of inconsistent states caused by network latency in multi-user VR collaborative simulation was solved, achieving high-precision state synchronization and a smooth interactive experience. It adapts to complex network environments and improves the immersion of training and the accuracy of assessment.
Patent Information
- Application Number
- CN202610136628.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-30
- Publication Date
- 2026-05-15
AI Technical Summary
In multi-user VR collaborative simulation, network latency, differences in computing power, and fluctuations in local rendering frame rate lead to inconsistent scene states among different clients, resulting in time distortion and affecting the immersion of training and the accuracy of assessment.
The method of integrating client-side prediction and server-side authoritative state is adopted. Through the steps of local state prediction, state update sending, server integration and broadcasting, client correction and smooth presentation, asynchronous time-twisted state synchronization is achieved, and state correction is performed using preset state prediction and interpolation algorithms.
It significantly reduces state differences between clients caused by network latency, ensures consistency of operation nodes, improves the accuracy of collaborative operations and smooth interactive experience, enhances immersion and system stability, and adapts to complex network environments.
Smart Images
Figure CN122044504A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of simulation system technology, specifically to an asynchronous time-state synchronization method in multi-user VR collaborative simulation. Background Technology
[0002] In multi-user VR collaborative simulation training, such as virtual disassembly, assembly, maintenance, and operation training of mechanical equipment, multiple users are required to interact and collaborate in real time within the same virtual environment. Currently, general-purpose game engines (such as Unity3D and Unreal Engine) or dedicated simulation platforms are commonly used to build such applications.
[0003] However, due to differences in network latency, computing power, and local rendering frame rate fluctuations among different user clients, the scene states presented by each client (such as model position, pose, animation progress, and tool holding status) will differ in time, a phenomenon known as "time distortion." If these states are not synchronized, it will lead to inconsistent user perceptions, disrupted collaborative workflows, and even errors that violate physical rules, such as model penetration and operational failures, severely impacting the immersiveness, realism, and accuracy of training evaluation.
[0004] In existing technologies, simple state snapshot synchronization or lockstep synchronization mechanisms are often used. The former is prone to screen jumps due to latency, while the latter is heavily dependent on network conditions and is difficult to achieve good results in high-frequency, high-precision interactive scenarios required for complex mechanical simulation. Summary of the Invention
[0005] To address this issue, the present invention provides an asynchronous time-twist state synchronization method for multi-user VR collaborative simulation, which solves the problem that existing technologies often employ simple state snapshot synchronization or lockstep synchronization mechanisms. The former is prone to screen jumps due to latency, while the latter heavily relies on network conditions and is difficult to achieve good results in high-frequency, high-precision interactive scenarios required for complex mechanical simulations.
[0006] To achieve the above objectives, the present invention provides the following technical solution:
[0007] The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation includes the following steps:
[0008] Step S1: Local State Prediction and Update Preparation; In the current simulation frame, each VR client, based on its locally maintained authoritative scene state confirmed in the previous frame, and combined with the local user input operation commands collected in this frame, uses a preset state prediction algorithm to deduce a predicted current local scene state; wherein, the scene state includes at least the three-dimensional spatial pose of all dynamic models in the virtual scene, the disassembly and assembly state of repair parts, the holding state of repair tools, and the currently activated fault simulation state;
[0009] Step S2: State update transmission; The VR client encapsulates the predicted current local scene state, along with the local timestamp and the identifier of the specific object whose state has changed, into a state update data packet and sends it to the server;
[0010] Step S3: State update reception and integration; The server continuously listens for and receives state update data packets from each VR client; The server maintains a state update buffer for each client and sorts the received data packets according to the local timestamp they carry; The server integrates the state updates of each client within a preset time window based on the system time and calculates and generates a unified authoritative scene state for that moment.
[0011] Step S4: Authority Status Broadcast; The server multicasts the unified authority scene status generated in step S3, along with its corresponding server timestamp, to all online VR clients.
[0012] Step S5: State Correction and Smooth Presentation; After each VR client receives the authoritative scene state broadcast by the server, it first compares the server timestamp in the received state with the local time to calculate the network transmission delay; then, the client compares and corrects the locally maintained predicted state with the received authoritative state; for spatial pose states, the client uses an interpolation algorithm in the following frames to smoothly transition the locally displayed model state from the predicted pose before correction to the authoritative pose; for discrete logic states, it is immediately updated to the authoritative state.
[0013] Preferably, in step S1, the predicted current local scene state is represented in the form of a multidimensional state vector, which includes at least: the world coordinate position components of all interactive model nodes, the Euler angle rotation components, the disassembled flag of the repair part component, the current owner role identifier of the repair tool component, and the fault activation code of the fault simulation component.
[0014] Preferably, the preset state prediction algorithm is an extrapolation algorithm based on a physical motion model; for position and rotation states, the state vector dimension used by the prediction algorithm also includes the instantaneous linear velocity, angular velocity, linear acceleration, and angular acceleration of the interactive model nodes; the prediction algorithm calculates the predicted position and rotation of the current frame based on the authoritative position, rotation, velocity, and acceleration of the previous frame, combined with the time difference between the current frame and the previous frame.
[0015] Preferably, in step S3, the depth of the state update buffer is configurable to accommodate out-of-order data packets caused by network jitter. The integration process includes: for each state attribute to be synchronized, the server checks all update values from different clients within a preset time window. If all values are consistent or within the tolerance range, the consistent value or average value is used as the authoritative value. If there are conflicts exceeding the tolerance range, they are handled according to a predefined conflict resolution strategy. The conflict resolution strategy includes using the earliest timestamp value, or using the value sent by the co-creator client, or specifying a priority client based on the object type.
[0016] Preferably, in step S5, the client maintains a local state history queue to store the local predicted state and corresponding local timestamps of the past several frames; when the authoritative state is received, the client finds the local predicted state record that is closest in time in the local state history queue according to the server timestamp and the calculated delay, and uses this as the benchmark for correction to calculate the state difference.
[0017] Preferably, in step S5, the interpolation algorithm is linear interpolation or spherical linear interpolation; the number of continuous frames for smooth transition is dynamically adjusted according to the calculated network transmission delay. The greater the delay, the more frames are used for smooth transition, but it does not exceed the preset maximum frame limit.
[0018] Preferably, the scene state also includes the real-time posture of the VR controller and the enabled / disabled state of the collider components attached to the repair parts and repair tool models; in the prediction of step S1, the controller posture is driven by data directly provided by the client's local VR runtime; the collider state is logically derived based on the disassembly / assembly state of its associated repair parts or the usage state of the repair tools, and is included in the state update.
[0019] Preferably, in step S1, when performing local state prediction, a collision prediction step is also included: based on the predicted model pose change, the local physics engine is called to perform fast collision detection pre-calculation; if an illegal penetrating collision is predicted to occur, the predicted motion path is constrained or marked, and this constraint information is attached as metadata to the state update data packet and sent to the server; during the integration process in step S3, the server will consider this constraint information. If multiple clients have consistent predictions for the same potential conflict, the generated authoritative state will include this motion constraint and will be broadcast in step S4 to force each client to correct.
[0020] Preferably, the synchronization of the fault simulation state is specifically implemented as follows: when a client triggers a fault simulation event, the fault simulation event and the corresponding fault status code are updated as key logical states and sent immediately in step S2; the server prioritizes processing such updates in step S3; after all clients receive the fault simulation event status in step S5, they immediately drive the local particle system, material changes, or model animations according to the fault status code to present the fault phenomenon in a consistent manner.
[0021] The present invention has the following advantages:
[0022] High-precision status synchronization: By combining client-side prediction with authoritative server-side status, the status differences between clients caused by network latency are significantly reduced, ensuring the consistency of status of all participants at key operational nodes (such as part disassembly, tool transfer, and assembly), and improving the accuracy of collaborative operations.
[0023] Smooth interactive experience: By utilizing local client prediction and smooth interpolation processing after receiving authoritative states, screen jumps or stutters during synchronization are avoided, providing users with continuous and natural operational feedback and enhancing the immersive experience of VR training.
[0024] Robust asynchronous adaptability: The method does not rely on strict network timing and can tolerate certain network jitter and differentiated client processing latency, adapting to the complex network environment in real-world applications and ensuring the stability and reliability of the collaborative training system.
[0025] Deep integration with business logic: Special consideration has been given to the unique state elements in mechanical simulation (such as the disassembly and assembly state of maintenance parts, the holding state of tools, the switching of collision bodies, the state of fault simulation, etc.), realizing the unified synchronization of business logic state and spatial transformation state, and meeting the needs of professional training. Attached Figure Description
[0026] To more intuitively illustrate the prior art and this application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing this application; for example, based on the technical concept disclosed in this application and the exemplary drawings, those skilled in the art are able to easily make conventional adjustments or further optimizations to the addition / reduction / classification, specific shapes, positional relationships, connection methods, size ratios, etc. of certain units (components).
[0027] Figure 1 A flowchart of an asynchronous time-twist state synchronization method in multi-user VR collaborative simulation provided in this application embodiment. Detailed Implementation
[0028] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these embodiments are merely for further explanation of the present invention and should not be construed as limiting the scope of protection of the present invention. Technical engineers in the field can make some non-essential improvements and adjustments to the present invention based on the above-described content. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0029] Please see Figure 1 The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation includes the following steps:
[0030] Step S1: Local State Prediction and Update Preparation; In the current simulation frame, each VR client, based on its locally maintained authoritative scene state confirmed in the previous frame, and combined with the local user input operation commands collected in this frame, uses a preset state prediction algorithm to deduce a predicted current local scene state; wherein, the scene state includes at least the three-dimensional spatial pose of all dynamic models in the virtual scene, the disassembly and assembly state of repair parts, the holding state of repair tools, and the currently activated fault simulation state;
[0031] Step S2: State update transmission; The VR client encapsulates the predicted current local scene state, along with the local timestamp and the identifier of the specific object whose state has changed, into a state update data packet and sends it to the server;
[0032] Step S3: Receiving and integrating status updates;
[0033] The server continuously listens for and receives status update data packets from each VR client; the server maintains a status update buffer for each client and sorts the received data packets according to the local timestamp they carry; the server integrates the status updates of each client within a preset time window based on the system time and calculates and generates a unified authoritative scene state for that moment.
[0034] Step S4: Broadcast authoritative status;
[0035] The server will multicast the unified authoritative scene state generated in step S3, along with its corresponding server timestamp, to all online VR clients.
[0036] Step S5: State correction and smooth rendering;
[0037] After receiving the authoritative scene state broadcast by the server, each VR client first compares the server timestamp in the received state with the local time to calculate the network transmission delay. Then, the client compares and corrects the locally maintained predicted state with the received authoritative state. For spatial pose states, the client uses an interpolation algorithm in the following frames to smoothly transition the locally displayed model state from the predicted pose before correction to the authoritative pose. For discrete logic states, the client immediately updates to the authoritative state.
[0038] This method addresses the issue of inconsistent states (i.e., "time distortion") caused by latency and computational differences among multiple VR clients in complex network environments. It forms a complete closed loop through five steps: Step S1 (local prediction) enables each client to perform "advanced" calculations based on the latest authoritative state and local operations, ensuring immediate responsiveness for user actions and effectively hiding network latency. Steps S2 and S4 (sending and broadcasting) establish an efficient state communication mechanism between the client and the server. Step S3 (server consolidation) performs time alignment and conflict resolution on the predicted states of all clients, generating a single, correct "source of fact" and fundamentally eliminating state discrepancies. The final step, S5 (client correction), uses smooth interpolation technology to gracefully converge the local display state to the authoritative state, avoiding abrupt changes in the screen and ensuring visual continuity. This complete process organically combines prediction, arbitration, and correction, ensuring consistency in collaborative operation logic while maximizing the smoothness and immersion of local interactions among clients, forming the foundation for a highly available multi-user VR collaborative training system.
[0039] In step S1, the predicted current local scene state is represented in the form of a multi-dimensional state vector, which includes at least: the world coordinate position components of all interactive model nodes, the Euler angle rotation components, the disassembled flag of the repair part component, the current owner role identifier of the repair tool component, and the fault activation code of the fault simulation component.
[0040] The multidimensional state vector not only includes the model's spatial pose (position, rotation), but also specifically covers key logical states unique to mechanical simulation operations, such as the "disassembled" flag for parts, the "holder" identifier for tools, and "fault activation codes." This definition method unifies and synchronizes geometric states with business logic states, ensuring "what you see is what you get," and thus enabling synchronization data packets to possess complete semantic information. The server and client can accurately understand and process the business meaning behind each state change (e.g., a part changing from "not disassembled" to "disassembled"), thereby supporting the accurate synchronization of complex, step-based maintenance training processes and avoiding logical ambiguities that may result from only synchronizing position information.
[0041] The preset state prediction algorithm is an extrapolation algorithm based on a physical motion model. For position and rotation states, the state vector dimension used by the prediction algorithm further includes the instantaneous linear velocity, angular velocity, linear acceleration, and angular acceleration of the model node. The prediction algorithm calculates the predicted position and rotation of the current frame based on the authoritative position, rotation, velocity, and acceleration of the previous frame, combined with the time difference between the current frame and the previous frame.
[0042] When implemented, the above scheme ensures that client-side predictions no longer involve simple linear extrapolation of position, but rather follow fundamental physical laws of motion (such as inertia). For example, for a screw being unscrewed, its rotation angle and angular velocity will be predicted simultaneously. This method significantly improves the accuracy of short-term predictions, making locally presented motion more natural and physically intuitive. Even with network latency, the trajectory of objects seen by the user is closer to their actual physical behavior, thereby reducing the correction range between the predicted state and the final authoritative state, further enhancing the smoothness and realism of synchronization.
[0043] In step S3, the depth of the state update buffer is configurable to accommodate out-of-order data packets caused by network jitter. The integration process includes: for each state attribute to be synchronized, the server checks all update values from different clients within the preset time window. If all values are consistent or within the tolerance range, the consistent value or average value is used as the authoritative value. If there are conflicts exceeding the tolerance range, they are handled according to a predefined conflict resolution strategy, which includes using the earliest timestamp value, or using the value sent by the co-creator client, or specifying a priority client based on the object type.
[0044] By setting configurable buffers to tolerate network out-of-order behavior and designing an integrated process that includes consistency checks, tolerance judgments, and multiple conflict resolution strategies (such as timestamp priority and creator priority), the robustness and fairness of the system are enhanced. When faced with unavoidable network jitter or conflicting operations by different clients on the same object (e.g., two people simultaneously trying to grab the same tool), this mechanism can generate deterministic and authoritative results based on clear rules, preventing state splitting. This ensures the authoritative determination of the operation order during collaborative training, which is crucial for subsequent evaluation (judging the correctness of operations).
[0045] In step S5, the client maintains a local state history queue to store the local predicted state and corresponding local timestamps of the past several frames. When the authoritative state is received, the client finds the local predicted state record that is closest in time in the local state history queue based on the server timestamp and the calculated delay, and uses this as the benchmark for correction to calculate the state difference.
[0046] When the client receives a delayed authoritative state, it doesn't simply compare it to the current frame state. Instead, it goes back in time to find the local prediction record corresponding to the authoritative state at that moment and calculates the difference, thus achieving more accurate state rollback and correction. This solves the problem of "correction benchmark misalignment" caused by uncertain network latency, more realistically restoring the amount of error generated by local prediction during the latency period. This makes subsequent smooth interpolation correction more accurate and reliable, effectively avoiding image jitter or logical errors caused by over-correction or under-correction.
[0047] In step S5, the interpolation algorithm is linear interpolation or spherical linear interpolation; the number of continuous frames for smooth transition is dynamically adjusted according to the calculated network transmission delay. The greater the delay, the more frames are used for smooth transition, but it does not exceed the preset maximum frame limit to ensure the timeliness of synchronization.
[0048] By optimizing the interpolation algorithm and transition strategy in the smoothing correction process, the interpolation duration can be dynamically adjusted based on the real-time calculated network latency (longer smoothing time for higher latency), achieving an adaptive synchronization experience. Under good network conditions, the system synchronizes quickly and responds sensitively; under high-latency network conditions, the system extends the smoothing time, distributing state corrections across more frames, thus completely avoiding abrupt "teleportation" or "jumping" phenomena. This flexible processing mechanism, while ensuring the consistency of the final state, consistently provides users with the smoothest possible visual feedback, significantly improving the consistency of user experience across different network qualities.
[0049] The scene state also includes the real-time posture of the VR controller and the enabled / disabled state of the collider components attached to the repair parts and tool models. In the prediction of step S1, the controller posture is driven by high-frequency data directly provided by the client's local VR runtime, without cross-network prediction synchronization, but the state of scene objects affected by its operation results participates in the prediction. The collider state is logically derived based on the disassembly / assembly state of its associated repair parts or the usage state of the repair tools, and is included in the state update, thus balancing the realism of the interaction with the synchronization efficiency, allowing users to experience a lag-free controller control feel. The key results triggered by the controller operation (such as successful grabbing, collider on / off) are strictly synchronized through a prediction-correction process, ensuring the correctness of the interaction logic under multiple perspectives. For example, after a user grabs a tool, their local controller animation responds immediately, and after synchronization, the tool's "held" state will also disappear or become uninterrupted in the perspectives of other users.
[0050] In step S1, the local state prediction process also includes a collision prediction step: based on the predicted model pose changes, a local physics engine is invoked for rapid collision detection pre-calculation; if an illegal penetrating collision is predicted, the predicted motion path is constrained or marked, and this constraint information is attached as metadata to the state update data packet and sent to the server; during the integration process in step S3, the server considers this constraint information. If multiple clients have consistent predictions for the same potential conflict, the generated authoritative state will include this motion constraint and be broadcast in step S4, forcing each client to correct to avoid collision errors. Thus, when this solution is implemented, it can proactively prevent and resolve potential physical conflict problems, elevating synchronization from "post-correction" to "pre-avoidance." In complex mechanical assembly and disassembly, it can predict interference in part assembly paths or tool collisions in advance and constrain motion paths in advance through server arbitration. This avoids the situation where clients actually experience errors such as penetration due to asynchronous predictions, requiring laborious rollback corrections, greatly improving the synchronization quality and system stability in complex physical interaction scenarios.
[0051] The virtual maintenance training simulation applied to mechanical equipment is implemented as follows: when a client triggers a fault simulation event, the event and the corresponding fault status code are updated as key logical statuses and sent immediately in step S2; the server prioritizes such updates in step S3 to ensure their rapid integration and broadcast; after all clients receive this authoritative fault status in step S5, they immediately drive their local particle systems, material changes, or model animations according to the status code to present the fault phenomenon in a consistent manner.
[0052] For the specific scenario of fault simulation in mechanical maintenance training, this solution optimizes its state synchronization mechanism. It defines fault events as critical logical states requiring "immediate" and "strongly consistent" synchronization, assigning them high priority in the server processing flow. This ensures the training effectiveness and fairness of the assessment. All collaborative training participants will observe identical fault phenomena (such as smoke from the same location or flashing of the same indicator light) within a very short time difference, and begin troubleshooting operations based on this consistent state. This avoids coordination confusion caused by differences in the timing or manifestation of fault phenomena, ensuring the rigor of the training course and the accuracy of the assessment.
[0053] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. An asynchronous time-twist state synchronization method in multi-user VR collaborative simulation, characterized in that, Includes the following steps: Step S1: Local state prediction and update preparation; In the current simulation frame, each VR client, based on the authoritative scene state confirmed in the previous frame and the local user input operation commands collected in this frame, uses a preset state prediction algorithm to deduce a predicted current local scene state; wherein, the scene state includes at least the three-dimensional spatial pose of all dynamic models in the virtual scene, the disassembly and assembly state of repair parts, the holding state of repair tools, and the currently activated fault simulation state. Step S2: State update transmission; The VR client encapsulates the predicted current local scene state, along with the local timestamp and the identifier of the specific object whose state has changed, into a state update data packet and sends it to the server; Step S3: State update reception and integration; The server continuously listens for and receives state update data packets from each VR client; The server maintains a state update buffer for each client and sorts the received data packets according to the local timestamp they carry; The server integrates the state updates of each client within a preset time window based on the system time and calculates and generates a unified authoritative scene state for that moment. Step S4: Authority Status Broadcast; The server multicasts the unified authority scene status generated in step S3, along with its corresponding server timestamp, to all online VR clients. Step S5: State Correction and Smooth Presentation; After each VR client receives the authoritative scene state broadcast by the server, it first compares the server timestamp in the received state with the local time to calculate the network transmission delay; then, the client compares and corrects the locally maintained predicted state with the received authoritative state; for spatial pose states, the client uses an interpolation algorithm in the following frames to smoothly transition the locally displayed model state from the predicted pose before correction to the authoritative pose; for discrete logic states, it is immediately updated to the authoritative state.
2. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, In step S1, the predicted current local scene state is represented in the form of a multidimensional state vector, which includes at least: the world coordinate position components of all interactive model nodes, the Euler angle rotation components, the disassembled flag of the repair part component, the current owner role identifier of the repair tool component, and the fault activation code of the fault simulation component.
3. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 2, characterized in that, The preset state prediction algorithm is an extrapolation algorithm based on a physical motion model. For position and rotation states, the state vector dimension used by the prediction algorithm also includes the instantaneous linear velocity, angular velocity, linear acceleration, and angular acceleration of the interactive model nodes. The prediction algorithm calculates the predicted position and rotation of the current frame based on the authoritative position, rotation, velocity, and acceleration of the previous frame, combined with the time difference between the current frame and the previous frame.
4. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, In step S3, the depth of the state update buffer is configurable to accommodate out-of-order data packets caused by network jitter. The integration process includes: for each state attribute to be synchronized, the server checks all update values from different clients within a preset time window. If all values are consistent or within the tolerance range, the consistent value or average value is used as the authoritative value. If there are conflicts that exceed the tolerance range, they are handled according to a predefined conflict resolution strategy. The conflict resolution strategy includes using the earliest timestamp value, or using the value sent by the co-creator client, or specifying the priority client according to the object type.
5. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, In step S5, the client maintains a local state history queue to store the local predicted state and corresponding local timestamps of the past several frames. When the authoritative state is received, the client finds the local predicted state record that is closest in time in the local state history queue based on the server timestamp and the calculated delay, and uses this as the benchmark for correction to calculate the state difference.
6. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, In step S5, the interpolation algorithm is linear interpolation or spherical linear interpolation; the number of continuous frames for smooth transition is dynamically adjusted according to the calculated network transmission delay. The greater the delay, the more frames are used for smooth transition, but it does not exceed the preset maximum frame limit.
7. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, The scene state also includes the real-time posture of the VR controller and the enabled / disabled status of the collider components attached to the repair parts and repair tool models; in the prediction of step S1, the controller posture is driven by data directly provided by the client's local VR runtime; the collider status is logically derived based on the disassembly / assembly status of its associated repair parts or the usage status of the repair tools, and is included in the status update.
8. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, In step S1, the local state prediction process also includes a collision prediction step: based on the predicted model pose change, the local physics engine is invoked to perform a fast collision detection pre-calculation; if an illegal penetrating collision is predicted, the predicted motion path is constrained or marked, and this constraint information is attached as metadata to the state update data packet and sent to the server; during the integration process in step S3, the server will consider this constraint information. If multiple clients have consistent predictions for the same potential conflict, the generated authoritative state will include this motion constraint and will be broadcast in step S4 to force each client to correct.
9. The asynchronous time-twist state synchronization method in multi-user VR collaborative simulation according to claim 1, characterized in that, The synchronization of the fault simulation state is specifically implemented as follows: when a client triggers a fault simulation event, the fault simulation event and the corresponding fault status code are updated as key logical states and sent immediately in step S2. In step S3, the server prioritizes processing such updates; after all clients receive the fault simulation event status in step S5, they immediately drive the local particle system, material changes, or model animations according to the fault status code to present the fault phenomenon in a consistent manner.