VR data interaction method and system for large-space scene

Abstract

The invention relates to the technical field of VR data interaction, in particular to a VR data interaction method and system for a large-space scene, and the method comprises the steps: constructing a space mapping data model; acquiring real-time spatial attitude data through the head-mounted device; calculating a predicted displacement vector according to the angular velocity and acceleration data and determining a predicted position coordinate; a spatial consistency metric value is obtained through collaborative attitude consistency calculation, and global registration is performed when conditions are met to generate a global attitude matrix; and monitoring a user motion vector change rate, calculating logic time offset according to the state offset information, and correcting sampling reference time by performing time compensation on a local time counter. According to the method, the consistency of multi-terminal data on a logic timeline is ensured through a dynamic time compensation mechanism, high-precision multi-user VR interaction synchronization is realized in combination with multi-source data fusion prediction and collaborative consistency judgment, and the real-time performance and stability of a virtual interaction process are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of VR data interaction, and particularly relates to a VR data interaction method and system for a large space scene. BACKGROUND

[0002] Virtual reality (VR) technology has been widely applied in recent years, and gradually develops from single-person immersive experience to multi-person collaborative interaction scene. Multiple users need to complete tasks, pose alignment or synchronous triggering operations in the same virtual space.

[0003] In the multi-person collaborative interaction in a large space scene, the traditional VR data transmission and synchronization mechanism is still mainly based on an independent client clock. The sampling time of each terminal is controlled by a local hardware timer. When network delay or physical drift occurs, the sampling data of different terminals is difficult to maintain logical consistency, resulting in problems such as pose offset, position misplacement or interaction triggering delay in the virtual space. The existing technology usually relies on increasing the number of sensors or improving the sampling frequency to alleviate this problem, but cannot solve the offset accumulation from the fundamental level of time synchronization, and the system resource consumption also increases.

[0004] In addition, the existing VR system generally adopts a centralized state broadcast mechanism in real-time multi-person interaction. This kind of scheme lacks a dynamic time correction mechanism. When the motion state of a certain terminal changes dramatically, the drift of its local clock cannot be corrected immediately, causing the synchronization error to continue to amplify during the entire interaction period.

[0005] Therefore, how to realize dynamic self-correction of the pose sampling time of each terminal in a large space and multi-user collaborative VR scene, so that the pose data of multiple users remains consistent in the logical time domain, has become a problem to be solved. SUMMARY

[0006] (1) Technical problem to be solved The purpose of the present application is to provide a VR data interaction method and system for a large space scene, to solve the problem that in a large space and multi-user collaborative virtual reality scene, the terminal device is affected by network delay, hardware clock drift and user motion state change when sampling pose data, resulting in offset of sampling time of different users, and causing problems such as unsynchronized space position and unstable collaborative triggering.

[0007] (2) Technical scheme To achieve the above purpose, on the one hand, the present application provides a VR data interaction method for a large space scene, which comprises: Step S1: obtaining site boundary information and virtual task structure information; determining a coordinate boundary range of a virtual space according to the site boundary information; determining a number N of task interaction regions and task identifiers of the task interaction regions according to the virtual task structure information; and constructing a space mapping data model by locating the N task interaction regions in the coordinate boundary range.

[0008] Step S2: obtaining real-time space posture data of a user at a sampling reference time through a head-mounted device to obtain a posture cache sequence; the space posture data includes position coordinates, a current orientation, angular velocity data, and acceleration data; a predicted displacement vector is calculated according to the current orientation, the angular velocity data, and the acceleration data; a predicted position coordinate at a target time point is determined according to the predicted displacement vector and the position coordinates; a space consistency measure value is obtained through a cooperative consistency calculation according to the predicted position coordinate; an interaction synchronization signal is generated when the space consistency measure value is less than a preset measure value threshold; when the interaction synchronization signal is received, a global posture matrix is obtained by performing coordinate registration on predicted position coordinates of N users and interaction node positions in the space mapping data model; and position coordinates in posture cache sequences of the N users are subjected to coordinate transformation according to the global posture matrix to obtain a globally registered posture cache sequence.

[0009] Step S3: monitoring a difference vector of position coordinates of adjacent two frames in the globally registered posture cache sequence to obtain a motion vector; a motion vector change rate is calculated according to the motion vector; state offset information is recorded when the motion vector change rate is greater than a preset speed threshold; a logical time offset is calculated according to the state offset information; a local time counter is time-compensated according to the logical time offset to obtain a corrected sampling reference time; a sampling reference time of a next sampling period is updated according to the corrected sampling reference time; and subsequent real-time space posture data is collected according to the sampling reference time of the next sampling period.

[0010] Preferably, the method of constructing the space mapping data model by locating the N task interaction regions in the coordinate boundary range comprises: determining space center point coordinates and boundary contours of the N task interaction regions in the coordinate boundary range; generating N corresponding interaction nodes according to interaction types corresponding to the task interaction regions in the virtual task structure information; assigning unique task identifiers to the N corresponding interaction nodes, and recording space position parameters and orientation parameters of the interaction nodes; and constructing a mapping relationship between a virtual space and a real site according to the space position parameters, the orientation parameters, and the task identifiers of the interaction nodes to obtain the space mapping data model.

[0011] Preferably, the method of calculating the predicted displacement vector according to the current orientation, angular velocity data and acceleration data comprises: differential-integrating the quaternion representation of the current orientation according to the angular velocity data with a preset sampling interval time as the integral period to obtain a predicted orientation after the preset sampling interval time; eliminating the gravity acceleration component after coordinate transformation of the acceleration data according to the predicted orientation to obtain a linear acceleration; double-integrating the linear acceleration within the preset sampling interval time to obtain a first displacement vector; calculating a second displacement vector according to the last frame position coordinate and the current frame position coordinate provided by the optical positioning system; and performing weighted summation on the first displacement vector and the second displacement vector to obtain the predicted displacement vector.

[0012] Preferably, the method of performing weighted summation on the first displacement vector and the second displacement vector to obtain the predicted displacement vector comprises: calculating the predicted displacement vector according to the first displacement vector and the second displacement vector .

[0013] The calculation formula of the predicted displacement vector is: .

[0014] wherein, the first displacement vector is the second displacement vector is the fusion weight coefficient is, and .

[0015] The calculation formula of the fusion weight coefficient is: .

[0016] wherein, the angular velocity vector length formed by the angular velocity data is the preset angular velocity threshold is and the preset constant is, and .

[0017] Preferably, the method of calculating the spatial consistency measure value through cooperative consistency calculation according to the predicted position coordinates comprises: converting the predicted orientations of N users into corresponding predicted orientation vectors; taking a preset spatial target point as a common focus point and taking each of the predicted orientation vectors as a line-of-sight vector of the corresponding user at a target time point; for each user, calculating the included angle between the line-of-sight vector and a reference vector pointing from the predicted position coordinate of the user to the preset spatial target point wherein Take N of the aforementioned included angles The maximum value is used as the spatial consistency metric.

[0018] Preferably, the method for calculating the logical time offset based on the state offset information includes: Extract the rate of change of motion vector from the state offset information. ;when When setting the logical time offset ;when At that time, calculate the initial offset. The initial offset The calculation formula is: .

[0019] in, This is a preset proportional coefficient. The preset rate of change threshold is used; the initial offset is... After performing amplitude limiting and smoothing processing, the logical time offset is obtained. .

[0020] Preferably, the initial offset After performing amplitude limiting and smoothing processing, the logical time offset is obtained. The methods include: The initial offset Limited to a preset offset range Within, the limiting offset is obtained. According to the aforementioned limiting offset The logical time offset is calculated using a recursive smoothing strategy. The logical time offset The calculation formula is: .

[0021] in, For the preset smoothing coefficient, This is the logical time offset of the previous sampling period.

[0022] Preferably, the method for time compensation of the local time counter based on the logical time offset to obtain the corrected sampling reference time includes: Obtain the sampling reference time for the current sampling period. The nominal sampling period is read according to the local time counter. According to the sampling reference time Nominal sampling period and logical time offset The corrected sampling reference time was calculated. ; the corrected sampling reference time The calculation formula is: .

[0023] Based on the same inventive concept, in another aspect, the application also provides a VR data interaction system for a large space scene, the system comprising: a space modeling module, configured to acquire site boundary information and virtual task structure information; determine a coordinate boundary range of a virtual space according to the site boundary information; determine a number N of task interaction areas and a task identifier of each task interaction area according to the virtual task structure information; and construct a space mapping data model by locating the N task interaction areas within the coordinate boundary range.

[0024] a pose prediction and interaction synchronization module, configured to acquire real-time spatial pose data of a user at a sampling reference time through a head-mounted device to obtain a pose cache sequence; the spatial pose data comprises position coordinates, a current orientation, angular velocity data and acceleration data; calculate a predicted displacement vector according to the current orientation, the angular velocity data and the acceleration data; determine predicted position coordinates of a target time point according to the predicted displacement vector and the position coordinates; obtain a spatial consistency measure value through a cooperative consistency calculation according to the predicted position coordinates; generate an interaction synchronization signal when the spatial consistency measure value is less than a preset measure value threshold; when receiving the interaction synchronization signal, perform coordinate registration on predicted position coordinates of N users and interaction node positions in the space mapping data model to obtain a global pose matrix; perform coordinate transformation on the position coordinates in the pose cache sequence of the N users according to the global pose matrix to obtain a globally registered pose cache sequence.

[0025] a time correction and cyclic update module, configured to monitor a difference vector of adjacent two frames of position coordinates in the globally registered pose cache sequence to obtain a motion vector; calculate a motion vector change rate according to the motion vector; record state offset information when the motion vector change rate is greater than a preset speed threshold; calculate a logical time offset according to the state offset information; perform time compensation on a local time counter according to the logical time offset to obtain a corrected sampling reference time; update a sampling reference time of a next sampling period according to the corrected sampling reference time; and collect subsequent real-time spatial pose data according to the sampling reference time of the next sampling period.

[0026] (3) Beneficial effects Compared with the prior art, the beneficial effects of the application are: 1. This invention monitors the transient changes in the user's motion state, dynamically calculates and compensates for the logical time offset of the terminal, and performs closed-loop correction on the local sampling time. From the time reference level, it ensures the consistency of data from multiple terminals on the logical timeline, effectively eliminates sampling errors caused by the accumulation of clock drift, and keeps the multi-user data synchronization process stable.

[0027] 2. This invention integrates inertial measurement and optical positioning data to predict user posture, and uses the coordinated consistency of all users' lines of sight to the target in the public space as the synchronization trigger condition, thereby achieving high-precision consistency determination of multi-user virtual posture in the coordinate registration stage, thus significantly improving the real-time performance and immersion of the virtual interaction process. Attached Figure Description

[0028] Figure 1 This is a flowchart of a VR data interaction method for a large-space scene according to Embodiment 1 of the present invention; Figure 2 This is a block diagram of a VR data interaction system for a large-space scene according to Embodiment 2 of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Before providing examples, it's necessary to describe the application scenarios of this invention, which are applicable to large-space VR interaction scenarios involving multiple users. In such large-space scenarios that support free movement of multiple users within a significant physical area, multiple users wearing head-mounted displays synchronously experience virtual tasks within the same physical space, changing their perspective in the virtual world through realistic actions such as moving, turning, and pointing. When multiple users need to simultaneously perform collaborative operations on specific spatial targets, the system needs to perceive the perspective direction and positional relationship of each user in real time and maintain global consistency of interactive objects in the virtual scene. The high timeliness and precision requirements of such collaborative operations mean that traditional single-terminal or independent rendering solutions cannot guarantee synchronization, resulting in problems such as perspective misalignment and action delays. The data interaction method proposed in this invention can achieve dual consistency of logical time and spatial state in such large-space, multi-terminal collaborative VR scenarios.

[0031] Example 1: As Figure 1 As shown, this embodiment provides a VR data interaction method for large-space scenes, the method including: Step S1: obtaining site boundary information and virtual task structure information; determining the coordinate boundary range of the virtual space according to the site boundary information; determining the number N of task interaction regions and the task identification of each task interaction region according to the virtual task structure information; and constructing a space mapping data model by locating the N task interaction regions within the coordinate boundary range.

[0032] Step S2: obtaining real-time spatial posture data of the user at a sampling reference time through a head-mounted device to obtain a posture cache sequence; the spatial posture data includes position coordinates, a current orientation, angular velocity data, and acceleration data; a predicted displacement vector is calculated according to the current orientation, angular velocity data, and acceleration data; a predicted position coordinate at a target time point is determined according to the predicted displacement vector and the position coordinates; a spatial consistency measure value is obtained through a cooperative consistency calculation according to the predicted position coordinate; an interaction synchronization signal is generated when the spatial consistency measure value is less than a preset measure value threshold; when the interaction synchronization signal is received, the predicted position coordinates of the N users are coordinate registered with the interaction node positions in the space mapping data model to obtain a global posture matrix; the position coordinates in the posture cache sequence of the N users are coordinate transformed according to the global posture matrix to obtain a globally registered posture cache sequence.

[0033] Step S3: monitoring the difference vector of the position coordinates of two adjacent frames in the globally registered posture cache sequence to obtain a motion vector; a motion vector change rate is calculated according to the motion vector; when the motion vector change rate is greater than a preset speed threshold, state offset information is recorded; a logical time offset is calculated according to the state offset information; a local time counter is time compensated according to the logical time offset to obtain a corrected sampling reference time; the sampling reference time of the next sampling period is updated according to the corrected sampling reference time; and subsequent real-time spatial posture data is collected according to the sampling reference time of the next sampling period.

[0034] For example, the embodiment is implemented in a VR team challenge scene named "virtual resonance array". First, the site boundary information and the virtual task structure information that three players need to cooperatively activate a magic array at three specific positions are obtained. According to this information, a circular temple model is constructed as a space mapping data model in the virtual space, and three asymmetrically distributed task interaction regions are mapped to the north coordinate , the southeast corner coordinate , and the southwest corner coordinate position of the physical site respectively. Three players (denoted as P1, P2, and P3) wear VR head-mounted devices and enter the site. The system starts running with an initial sampling reference time (for example, system timestamp their real-time spatial pose data through the head-mounted device. For example, at t = 0 ms, the data of player P1 is obtained as: position coordinates , current orientation , angular velocity , acceleration . These data are continuously stored in a pose buffer sequence with a capacity of 30 frames. Based on the current data, a predicted displacement vector is calculated. Taking the data of P1 as an example, according to its orientation, angular velocity and acceleration, through integration and other operations, it is predicted that at a future time (i.e. the target time point ), the predicted position coordinates of P1 are . Similarly, the predicted position coordinates of P2 and P3 at are calculated as and respectively. Taking the core point of the virtual BOSS: coordinates as the preset spatial target point, the included angles between the reference vectors of the three players from their respective predicted position coordinates to the BOSS and the line-of-sight vectors defined by their predicted orientations are calculated as , and . The maximum value is taken as the current spatial consistency measure value. If the spatial consistency measure value is less than the preset measure value threshold (determined by statistical analysis of the line-of-sight convergence tolerance of multiple testers under multiple cooperative tasks), it is determined that the cooperative condition is met, and an interaction synchronization signal is immediately generated. The set of predicted position coordinates of the three players is optimally aligned with the set of three interaction node positions defined in the spatial mapping data model, and a global pose matrix (for example, a transformation matrix containing rotation and translation amounts) is calculated. The global pose matrix is applied to coordinate transform all historical position coordinates in the pose buffer sequence. For example, the original coordinates of player P1 at are updated to in the globally aligned pose buffer sequence after transformation. This process unifies the motion trajectories of all players to the global coordinate system with the virtual temple center as the origin.

[0035] In subsequent links, the globally aligned pose buffer sequence is monitored. For example, it is found that the position of player P2 changes from to in two consecutive frames, and the motion vector obtained by difference is , and its module length is . The module length of the next frame motion vector becomes , and the motion vector change rate instantaneous value is calculated as greater than a preset speed threshold , record this state offset information. According to this state offset information, the system calculates the logical time offset as . After limiting and smoothing processing, the finally adopted logical time offset is . According to the logical time offset , the local time counter is time compensated, and finally the sampling reference time of the next sampling period is updated to 111.4ms. It needs to be explained that this compensation means that the system actively waits for the player P2 whose motion state changes dramatically, so that the data streams of all players are aligned on the logical time line.

[0036] The method comprises the following steps: determining the spatial center point coordinates and boundary contour of the N task interaction areas within the coordinate boundary range; generating N corresponding interaction nodes according to the interaction types corresponding to each task interaction area in the virtual task structure information; assigning unique task identifiers to the N corresponding interaction nodes, and recording the spatial position parameters and orientation parameters of the interaction nodes; constructing the mapping relationship between the virtual space and the real site according to the spatial position parameters, orientation parameters and task identifiers of the interaction nodes to obtain the spatial mapping data model.

[0037] For example, within the physical coordinate boundary range, the spatial center point coordinates and boundary contour of three task interaction areas are accurately determined. Specifically, the center point of the north region is set to , and the boundary contour is defined as a circular region with a radius of 1.0 meters with the point as the center; the center point of the southeast corner region is , and the boundary contour is a circular region with a radius of 1.2 meters; the center point of the southwest corner region is , and the boundary contour is also a circular region with a radius of 1.2 meters. According to the virtual task structure information, three corresponding interaction nodes are generated for the three regions. These nodes are respectively assigned unique task identifiers, for example, identified using the strings "Mission_North", "Mission_SouthEast" and "Mission_SouthWest". The spatial position parameters, i.e. the center point coordinates, and the orientation parameters of each interaction node are recorded. For example, the orientation of the "Mission_North" node is set to face the center of the virtual temple, i.e. the yaw angle Y is ; the orientation of the "Mission_SouthEast" node is set to face the northwest direction, i.e. Y is ; the orientation of the "Mission_SouthWest" node is set to face the northeast direction, i.e. Y is The spatial mapping data model completely defines the mapping relationship between each key point of the physical site and the virtual temple.

[0038] The method for calculating the predicted displacement vector according to the current orientation, angular velocity data and acceleration data comprises: Differential integration is performed on the quaternion representation of the current orientation with a preset sampling interval time as an integration period to obtain a predicted orientation after the preset sampling interval time; the acceleration data is subjected to coordinate transformation to eliminate the gravity acceleration component according to the predicted orientation, and linear acceleration is obtained; double integration is performed on the linear acceleration within the preset sampling interval time to obtain a first displacement vector; a second displacement vector is calculated according to the last frame position coordinate and the current frame position coordinate provided by the optical positioning system; the first displacement vector and the second displacement vector are subjected to weighted summation to obtain the predicted displacement vector.

[0039] For example, the data of the player P1 at the system time is taken as an example. The current orientation is represented by a quaternion as , which corresponds to a yaw angle of ; the angular velocity data is ; and the acceleration data is . The preset sampling interval time is . The system first performs differential integration on the quaternion of the current orientation according to the angular velocity data with as the integration period, and the quaternion representation of the predicted orientation is , which corresponds to a yaw angle of about . The original acceleration data is subjected to coordinate transformation to rotate it from the sensor coordinate system to the world coordinate system and eliminate the gravity acceleration component . Assuming that the gravity direction is the negative direction of the Z axis of the world coordinate system, the linear acceleration is calculated to be . Double integration is performed on the linear acceleration within the preset sampling interval time of . The first integration obtains a velocity increment vector , and the second integration obtains a first displacement vector . According to the data provided by the optical positioning system, the last frame position coordinate of P1 is , the current frame position coordinate is , and the second displacement vector is calculated to be .

[0040] It should be noted that the first displacement vector is mainly based on IMU integration, which is accurate in a short time but prone to drift; the second displacement vector is based on optical absolute positioning, which is stable in accuracy but may have a delay due to sampling rate or occlusion. The first displacement vector and the second displacement vector are weighted and summed to obtain a final predicted displacement vector .

[0041] The method of weighting and summing the first displacement vector and the second displacement vector to obtain the predicted displacement vector comprises: According to the first displacement vector and the second displacement vector, a predicted displacement vector is calculated.

[0042] The calculation formula of the predicted displacement vector is: .

[0043] wherein, is the first displacement vector, is the second displacement vector, is a fusion weight coefficient, and .

[0044] The calculation formula of the fusion weight coefficient is: .

[0045] wherein, is the length of the angular velocity vector composed of angular velocity data, is a preset angular velocity threshold, and are preset constants, and .

[0046] For example, the first displacement vector of the player P1 is , and the second displacement vector is . The length of the angular velocity vector is calculated according to the angular velocity data , and the length is . The preset angular velocity threshold is , the preset constant is 0.7, is 0.3, and this set of parameters is optimized through a large number of comparative experiments. Since the current angular velocity length is greater than the threshold , the fusion weight coefficient takes a value of Weighted sum The components of the predicted displacement vector are calculated as follows: X component: , Y component: , Z component: .

[0047] The predicted displacement vector is meters (calculated to the nearest millimeter (mm)). It is noted that this weighting strategy makes the system more dependent on the inertial measurement unit data when the user angular velocity is high. Because fast rotation can cause the optical positioning data to be temporarily unreliable due to motion blur, increasing the value, i.e. increasing the weight of the inertial measurement unit data, helps to maintain stability in the short-term prediction in such dynamic scenarios.

[0048] The method for obtaining a spatial consistency measure value according to the predicted position coordinates by cooperative consistency calculation comprises: converting the predicted orientations of N users into corresponding predicted orientation vectors; taking a preset spatial target point as a common focus point, and taking each of the predicted orientation vectors as a line-of-sight vector of the corresponding user at a target time point; for each user, calculating an included angle between the line-of-sight vector and a reference vector pointing from the predicted position coordinate of the user to the preset spatial target point , wherein ; taking the maximum value of N of the included angles as the spatial consistency measure value.

[0049] For example, taking the core point of a virtual BOSS as the preset spatial target point, the coordinates of which are . The predicted position coordinates of three players at the target time point are respectively , , After conversion, the predicted orientation quaternions of the three players become the corresponding predicted orientation vectors (i.e. line-of-sight vectors) respectively as , , .

[0050] The reference vector pointing from the predicted position coordinate to the preset spatial target point is calculated for each player. For example, the reference vector of player P1 is , which is normalized to be approximately The line-of-sight vector of player P1 and the reference vector the angle between calculated by dot product formula, The initial calculation result shows that the line of sight deviates too much from the reference direction. After optimization calculation, the angles of the three players are respectively: , , The maximum value of 4.1 degrees among the three angles is finally taken as the spatial consistency measurement value of this calculation.

[0051] The method for calculating the logical time offset according to the state offset information comprises: extracting a motion vector change rate from the state offset information ; when , setting a logical time offset ; when , calculating an initial offset ; the calculation formula of the initial offset .

[0052] wherein is a preset proportion coefficient, is a preset change rate threshold; the initial offset is subjected to amplitude limiting and smoothing processing to obtain a logical time offset .

[0053] For example, the motion vector change rate of the player P2 extracted from the recorded state offset information is . The preset change rate threshold is , and the preset proportion coefficient is . The preset change rate threshold is set according to the observation of the acceleration change rate of common user motion modes (walking, running, sudden turning). The preset proportion coefficient is used to map the mutation degree of the motion state to the time compensation amount, and the specific value can be configured according to the system delay tolerance. Since , the initial offset is calculated. The initial offset is subjected to amplitude limiting and smoothing processing to obtain a logical time offset .

[0054] The method for subjecting the initial offset to amplitude limiting and smoothing processing to obtain a logical time offset comprises:​​ The initial offset is limited in a preset offset range to obtain a limited offset ; the logical time offset is calculated by using a recursive smoothing strategy according to the limited offset ; and a calculation formula of the logical time offset is as follows: .

[0055] wherein, is a preset smoothing coefficient, is a logical time offset of a previous sampling period.

[0056] Exemplarily, the initial offset is . The initial offset is limited in the preset offset range and remains after the limiting processing. The preset smoothing coefficient is 0.5. The logical time offset adopted finally in the previous sampling period is .

[0057] The method for performing time compensation on the local time counter according to the logical time offset to obtain a corrected sampling reference time comprises: obtaining a sampling reference time of a current sampling period; reading a nominal sampling period from the local time counter; and calculating a corrected sampling reference time according to the sampling reference time , the nominal sampling period and the logical time offset ; and a calculation formula of the corrected sampling reference time is as follows: .

[0058] Exemplarily, the sampling reference time of the current sampling period is . The nominal sampling period read from the local time counter is (corresponding to a sampling frequency of 90 Hz). The corrected sampling reference time .

[0059] Embodiment 2: Based on the same inventive concept, as Figure 2 As shown in the embodiment, the VR data interaction system of a large space scene also comprises: a space modeling module, configured to acquire site boundary information and virtual task structure information; determine a coordinate boundary range of a virtual space according to the site boundary information; determine a number N of task interaction regions and task identifiers of the task interaction regions according to the virtual task structure information; and construct a space mapping data model by locating the N task interaction regions within the coordinate boundary range.

[0060] a pose prediction and interaction synchronization module, configured to acquire real-time space pose data of a user at a sampling reference time through a head-mounted device to obtain a pose cache sequence; the space pose data comprises position coordinates, a current orientation, angular velocity data and acceleration data; calculate a predicted displacement vector according to the current orientation, the angular velocity data and the acceleration data; determine predicted position coordinates of a target time point according to the predicted displacement vector and the position coordinates; obtain a space consistency measure value through a cooperative consistency calculation according to the predicted position coordinates; generate an interaction synchronization signal when the space consistency measure value is less than a preset measure value threshold; when the interaction synchronization signal is received, perform coordinate registration on predicted position coordinates of N users and interaction node positions in the space mapping data model to obtain a global pose matrix; perform coordinate transformation on the position coordinates in the pose cache sequence of the N users according to the global pose matrix to obtain a globally registered pose cache sequence.

[0061] a time correction and cyclic update module, configured to monitor a difference vector of adjacent two frames of position coordinates in the globally registered pose cache sequence to obtain a motion vector; calculate a motion vector change rate according to the motion vector; record state offset information when the motion vector change rate is greater than a preset speed threshold; calculate a logical time offset according to the state offset information; perform time compensation on a local time counter according to the logical time offset to obtain a corrected sampling reference time; update a sampling reference time of a next sampling period according to the corrected sampling reference time; and collect subsequent real-time space pose data according to the sampling reference time of the next sampling period.

[0062] It should be noted that, as for the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0063] Finally, it should be noted that although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still make modifications to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to part of the technical features, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A VR data interaction method for large-space scenes, characterized in that, The method includes: Acquire site boundary information and virtual task structure information; determine the coordinate boundary range of the virtual space based on the site boundary information; determine the number N of task interaction areas and the task identifier of each task interaction area based on the virtual task structure information; construct a spatial mapping data model by locating the N task interaction areas within the coordinate boundary range. Real-time spatial attitude data of users is acquired using a head-mounted device at a sampling reference time to obtain an attitude cache sequence. The spatial attitude data includes position coordinates, current orientation, angular velocity data, and acceleration data. A predicted displacement vector is calculated based on the current orientation, angular velocity data, and acceleration data. The predicted position coordinates at the target time point are determined based on the predicted displacement vector and position coordinates. A spatial consistency metric is obtained through collaborative consistency calculation based on the predicted position coordinates. When the spatial consistency metric is less than a preset threshold, an interactive synchronization signal is generated. Upon receiving the interactive synchronization signal, coordinate registration is performed between the predicted position coordinates of N users and the interactive node positions in the spatial mapping data model to obtain a global attitude matrix. The position coordinates in the attitude cache sequence of the N users are transformed based on the global attitude matrix to obtain a globally registered attitude cache sequence. The motion vector is obtained by monitoring the difference vector of the position coordinates of two adjacent frames in the globally registered attitude buffer sequence; the motion vector change rate is calculated based on the motion vector; when the motion vector change rate is greater than a preset velocity threshold, state offset information is recorded; logical time offset is calculated based on the state offset information; time compensation is performed on the local time counter based on the logical time offset to obtain the corrected sampling reference time; the sampling reference time of the next sampling period is updated based on the corrected sampling reference time; subsequent real-time spatial attitude data is collected based on the sampling reference time of the next sampling period.

2. The VR data interaction method for a large-space scene according to claim 1, characterized in that, The method for constructing a spatial mapping data model by locating the N task interaction regions within the coordinate boundary range includes: Within the coordinate boundary range, determine the spatial center point coordinates and boundary contours of N task interaction areas; generate N corresponding interaction nodes according to the interaction type corresponding to each task interaction area in the virtual task structure information; assign a unique task identifier to the N corresponding interaction nodes, and record the spatial position parameters and orientation parameters of the interaction nodes; construct the mapping relationship between the virtual space and the real site based on the spatial position parameters, orientation parameters, and task identifiers of the interaction nodes to obtain a spatial mapping data model.

3. The VR data interaction method for a large-space scene according to claim 1, characterized in that, The method for calculating the predicted displacement vector based on the current orientation, angular velocity data, and acceleration data includes: Based on the angular velocity data, with a preset sampling interval as the integration period, differential and integral operations are performed on the quaternion representation of the current orientation to obtain the predicted orientation after the preset sampling interval. Based on the predicted orientation, the acceleration data is subjected to coordinate transformation to eliminate the gravitational acceleration component, resulting in linear acceleration. The linear acceleration is double-integrated within the preset sampling interval to obtain a first displacement vector. Based on the position coordinates of the previous frame and the current frame provided by the optical positioning system, a second displacement vector is calculated. The first and second displacement vectors are weighted and summed to obtain the predicted displacement vector.

4. The VR data interaction method for a large-space scene according to claim 3, characterized in that, The method for obtaining the predicted displacement vector by weighted summation of the first displacement vector and the second displacement vector includes: The predicted displacement vector is calculated based on the first displacement vector and the second displacement vector. ; The predicted displacement vector The calculation formula is: ； in, Let be the first displacement vector. This is the second displacement vector. To integrate the weighting coefficients, and ; The fusion weight coefficient The calculation formula is: ； in, The magnitude of the angular velocity vector composed of angular velocity data. The preset angular velocity threshold, and It is a preset constant, and .

5. The VR data interaction method for a large-space scene according to claim 3, characterized in that, The method for obtaining a spatial consistency metric value based on predicted location coordinates through collaborative consistency calculation includes: The predicted orientations of N users are converted into corresponding predicted orientation vectors; with a preset spatial target point as the common focus, each predicted orientation vector is used as the corresponding user's line-of-sight vector at the target time point; for each user, the angle between the line-of-sight vector and the reference vector pointing from the user's predicted position coordinates to the preset spatial target point is calculated. ,in Take N of the aforementioned included angles The maximum value is used as the spatial consistency metric.

6. The VR data interaction method for a large-space scene according to claim 1, characterized in that, The method for calculating the logical time offset based on the state offset information includes: Extract the rate of change of motion vector from the state offset information. ;when When setting the logical time offset ;when At that time, calculate the initial offset. The initial offset The calculation formula is: ； in, This is a preset proportional coefficient. The preset rate of change threshold is used; the initial offset is... After performing amplitude limiting and smoothing processing, the logical time offset is obtained. .

7. The VR data interaction method for a large-space scene according to claim 6, characterized in that, The initial offset After performing amplitude limiting and smoothing processing, the logical time offset is obtained. The methods include: The initial offset Limited to a preset offset range Within, the limiting offset is obtained. According to the aforementioned limiting offset The logical time offset is calculated using a recursive smoothing strategy. The logical time offset The calculation formula is: ； in, For the preset smoothing coefficient, This is the logical time offset of the previous sampling period.

8. A VR data interaction method for a large-space scene according to claim 6, characterized in that, The method for time compensation of the local time counter based on the logical time offset to obtain the corrected sampling reference time includes: Obtain the sampling reference time for the current sampling period. The nominal sampling period is read according to the local time counter. According to the sampling reference time Nominal sampling period and logical time offset The corrected sampling reference time was calculated. The corrected sampling reference time The calculation formula is: 。 9. A VR data interaction system for large-space scenes, characterized in that, The system includes: The spatial modeling module is used to acquire site boundary information and virtual task structure information; determine the coordinate boundary range of the virtual space based on the site boundary information; determine the number N of task interaction areas and the task identifier of each task interaction area based on the virtual task structure information; and construct a spatial mapping data model by locating the N task interaction areas within the coordinate boundary range. The attitude prediction and interaction synchronization module is used to acquire real-time spatial attitude data of users through a head-mounted device at a sampling reference time, and obtain an attitude cache sequence. The spatial attitude data includes position coordinates, current orientation, angular velocity data, and acceleration data. A predicted displacement vector is calculated based on the current orientation, angular velocity data, and acceleration data. The predicted position coordinates of the target time point are determined based on the predicted displacement vector and position coordinates. A spatial consistency metric is obtained through collaborative consistency calculation based on the predicted position coordinates. When the spatial consistency metric is less than a preset metric threshold, an interaction synchronization signal is generated. When the interaction synchronization signal is received, the predicted position coordinates of N users are registered with the interaction node positions in the spatial mapping data model to obtain a global attitude matrix. The position coordinates in the attitude cache sequence of N users are transformed based on the global attitude matrix to obtain a globally registered attitude cache sequence. The time correction and cyclic update module is used to monitor the difference vector of position coordinates between two adjacent frames in the globally registered attitude buffer sequence to obtain a motion vector; calculate the motion vector change rate based on the motion vector; record state offset information when the motion vector change rate is greater than a preset velocity threshold; calculate the logical time offset based on the state offset information; perform time compensation on the local time counter based on the logical time offset to obtain the corrected sampling reference time; update the sampling reference time of the next sampling period based on the corrected sampling reference time; and collect subsequent real-time spatial attitude data based on the sampling reference time of the next sampling period.

Images