A spatiotemporal coordinate linkage AI autonomous social scene generation system
By analyzing spatiotemporal coordinates and calculating topological impedance factors, and by adjusting the adjudication threshold using a deviation self-compensation feedback module, the problem of pseudo-proximity matching in commercial complexes is solved, enabling more accurate scene push and reducing server load, thus adapting to complex environmental changes.
Patent Information
- Application Number
- CN202611251062.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-18
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies in data processing systems in commercial complexes and other locations cannot effectively distinguish between pseudo-proximity matches caused by different floors and wall obstructions, resulting in invalid scene pushes and data retrievals, increasing the server's computing load, and lacking an adaptive adjustment mechanism, making it difficult to balance computing overhead and push accuracy in complex environments.
The spatiotemporal coordinate analysis module is used to perform gridded spatial indexing and dynamic state vector generation. The matching weight value is calculated by combining the vertical floor difference and the topological impedance factor of physical barrier characteristics. The adjudication threshold is dynamically adjusted through the deviation self-compensation feedback module to reduce false proximity matching and invalid push.
It effectively distinguishes between terminals that are geographically close but cannot actually pass each other and commercial scene nodes, reduces server computing overhead, improves the accuracy of matching results and system stability, adapts to changes in pedestrian density, and reduces invalid data retrieval and state switching.
Smart Images

Figure CN123114254A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to an AI-driven autonomous social scene generation system with spatiotemporal coordinate linkage, belonging to the field of data processing technology. Background Technology
[0002] In commercial complexes and similar locations, data processing systems typically employ a combination of planar geofencing and static tag matching. The system collects coordinate data from mobile terminals, calculates the planar straight-line distance between the mobile terminal and nodes in the commercial scene, and issues scene push commands when the distance reaches a preset condition. However, in actual commercial spaces, there are often physical barriers such as different floors and solid walls. Judging solely based on planar distance assumes that locations can be directly approached. Mobile terminals located on different floors or opposite sides of walls may be judged as close due to their planar proximity, even if they cannot actually pass through directly, leading to false matches. At the same time, the positioning signal may also experience random drift and jitter. When the coordinates change repeatedly near the judgment boundary, the system is prone to frequently changing the matching state and continuously initiating invalid data retrievals, increasing the server's computational load and memory usage.
[0003] Simply reducing the judgment radius or increasing the sampling frequency cannot effectively solve the above problems. The higher the sampling frequency, the more likely the coordinate jitter near the judgment boundary will cause repeated judgments. A large number of location update requests will also occupy the server's computing power. If the protection boundary is expanded to reduce state switching, the scene response will be slower and it will be difficult to reflect the changes in the location and state of the mobile terminal in a timely manner. In addition, existing systems usually lack a mechanism for adaptive adjustment based on historical conversion results and cannot adjust the adjudication threshold in real time based on the density of people on site. Therefore, it is still difficult to balance computational overhead and push accuracy in complex application environments.
[0004] Therefore, how to reduce the data processing load of the system, reduce cross-barrier mismatches, and make the matching threshold continuously adjusted and stabilized according to the actual effect in the presence of multiple floors, wall obstruction and high-frequency positioning noise has become a technical problem that needs to be solved. Summary of the Invention
[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A spatiotemporal coordinate-linked AI autonomous social scene generation system, the system comprising: The spatiotemporal coordinate parsing module is used to receive a continuous spatiotemporal coordinate sequence sent by the mobile terminal, which includes longitude, latitude, sampling timestamp and altitude. It maps the continuous spatiotemporal coordinate sequence to a gridded spatial index structure. Combined with the mobile terminal's real-time displacement velocity vector and direction offset, it generates a dynamic state vector that includes spatiotemporal compactness and motion trend. The spatial hysteresis buffer module, which adapts to the real-time crowd density in the commercial scene node, performs threshold filtering on coordinate displacement jitter. The scene graph matching module, connected to the spatiotemporal coordinate parsing module, is used to obtain the real-time activity index and semantic attributes of commercial scene nodes, receive dynamic state vectors, introduce the topological impedance factor determined by the vertical floor difference and physical barrier features to calculate the anisotropic spatiotemporal impedance measure, and calculate the matching weight value between the mobile terminal and the commercial scene node. The state adjudication control module, connected to the scene graph matching module, compares the matching weight value with the adaptive adjudication threshold and outputs a social scene generation command when the matching weight value is greater than or equal to the adaptive adjudication threshold; and The deviation self-compensation feedback module, connected to the state adjudication control module, is used to collect the interactive response signal after the scene is issued, calculate the deviation between the actual scene conversion rate and the target benchmark conversion rate within the sampling period, and dynamically fine-tune the adaptive adjudication threshold according to the deviation by a preset step size.
[0006] Preferably, the scene map matching module calculates the topological impedance factor. The calculation model used is as follows: ,in, This represents the vertical floor difference between the mobile terminal and the business scenario nodes. This is the floor impedance attenuation coefficient. To indicate the presence of discrete variables that act as physical barriers, The impedance coefficient of the wall medium.
[0007] Preferably, the matching weight value is directly proportional to the semantic interest similarity metric value in the semantic attributes and inversely proportional to the anisotropic spatiotemporal impedance measure; the anisotropic spatiotemporal impedance measure is jointly calculated and determined by the physical plane distance between the mobile terminal and the commercial scene node, the preset benchmark reference distance, and the topological impedance factor.
[0008] Preferably, when the actual conversion rate is lower than the target baseline conversion rate, the deviation self-compensation feedback module raises the adaptive decision threshold for the next sample by a preset gain step size; when the actual conversion rate is higher than the target baseline conversion rate, the adaptive decision threshold for the next sample is lowered by a preset gain step size.
[0009] Preferably, the spatial hysteresis buffer module is set with a dynamic spatial protection boundary radius; when the position change of the mobile terminal does not exceed the dynamic spatial protection boundary radius, the spatiotemporal coordinate analysis module keeps the current grid index state unchanged; when the position change of the mobile terminal continuously exceeds the dynamic spatial protection boundary radius for a preset time window, the spatiotemporal coordinate analysis module updates the grid index state and reconstructs the dynamic state vector.
[0010] Preferably, the spatiotemporal coordinate analysis module adaptively scales the radius of the dynamic space protection boundary based on the real-time crowd density within the commercial scene node; when the real-time crowd density increases, the radius of the dynamic space protection boundary is reduced by a preset ratio; when the real-time crowd density decreases, the radius of the dynamic space protection boundary is expanded by a preset ratio.
[0011] Preferably, the state adjudication control module includes a rule reconstruction module. The rule reconstruction module is used to solve the Pareto optimal equilibrium point based on the dynamic state vector of each mobile terminal when multiple mobile terminals are concurrently matched to the same business scenario node, and generate collaborative social rules that include multi-party participation thresholds and tiered preferential rights.
[0012] Preferably, the spatiotemporal coordinate parsing module maps the continuous spatiotemporal coordinate sequence to a three-dimensional meshed spatial index bucket, and combines the instantaneous displacement velocity vector and direction offset of the mobile terminal to generate a dynamic state vector containing spatiotemporal compactness and motion trend.
[0013] Preferably, the state adjudication control module synchronously sends the social scene generation instructions to the mobile terminal and the on-site control module of the commercial scene node; the social scene generation instructions include the effective time window for interaction, spatial physical access constraints, and commercial concession rules.
[0014] Preferably, the spatial hysteresis buffer module performs numerical threshold truncation and state locking on discrete positioning drift in a continuous spatiotemporal coordinate sequence, and blocks the graph structure traversal calculation when the mobile terminal is in a state of slight displacement jitter.
[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the AI-driven autonomous social scene generation with spatiotemporal coordinate linkage, by considering the vertical floor difference and physical barrier features in the distance judgment, it is possible to distinguish between mobile terminals and commercial scene nodes that are close in planar location but cannot actually pass directly. This reduces false proximity matching in cases of cross-floor or wall separation, thereby avoiding invalid scene pushes and corresponding data retrieval and state switching. This makes the matching results of social scenes more consistent with the actual passage conditions of commercial spaces and reduces the computational overhead caused by the server processing invalid data.
[0016] 2. By using a gridded spatial index in conjunction with a spatial hysteresis buffer module, the impact of positioning drift and minor displacement jitter on the matching state can be reduced. The dynamic spatial protection boundary radius can be adjusted randomly according to the crowd density, maintaining a high position resolution capability when the crowd is dense and reducing unnecessary state updates when the crowd is sparse. The dynamic state vector is only reconstructed when the position change continuously meets the update conditions, thereby reducing frequent graph structure traversal and grid index updates, and enabling the system to maintain a relatively stable operating state when a large number of mobile terminals access the system concurrently.
[0017] 3. By adjusting the adaptive adjudication threshold based on the deviation between the actual scenario conversion rate and the target benchmark conversion rate through the deviation self-compensation feedback module, the scenario delivery threshold can be continuously corrected according to the actual interaction results. When the actual conversion effect is low, the adjudication requirements are increased to reduce the delivery of low-intent scenarios; when the actual conversion effect is high, the adjudication requirements are appropriately reduced to maintain the scenario coverage. This can reduce the impact of changes in traffic and fluctuations in user intent on the matching results, avoid push mismatch after long-term use of fixed adjudication thresholds, and make the scenario delivery status of the system gradually stabilize. Attached Figure Description
[0018] Figure 1 This is a diagram showing the data flow structure of the internal modules of the social scene generation system of this invention. Figure 2 This is an operational diagram of the continuous spatiotemporal coordinate mapping and scene matching adjudication control of the present invention.
[0019] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0020] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.
[0021] A spatiotemporal coordinate-linked AI autonomous social scene generation system, the system includes: The spatiotemporal coordinate parsing module is used to receive a continuous spatiotemporal coordinate sequence sent by the mobile terminal, which includes longitude, latitude, sampling timestamp and altitude. It maps the continuous spatiotemporal coordinate sequence to a gridded spatial index structure. Combined with the mobile terminal's real-time displacement velocity vector and direction offset, it generates a dynamic state vector that includes spatiotemporal compactness and motion trend. The spatial hysteresis buffer module, which adapts to the real-time crowd density in the commercial scene node, performs threshold filtering on coordinate displacement jitter. The scene graph matching module, connected to the spatiotemporal coordinate parsing module, is used to obtain the real-time activity index and semantic attributes of commercial scene nodes, receive dynamic state vectors, introduce the topological impedance factor determined by the vertical floor difference and physical barrier features to calculate the anisotropic spatiotemporal impedance measure, and calculate the matching weight value between the mobile terminal and the commercial scene node. The state adjudication control module, connected to the scene graph matching module, compares the matching weight value with the adaptive adjudication threshold and outputs a social scene generation command when the matching weight value is greater than or equal to the adaptive adjudication threshold; and The deviation self-compensation feedback module, connected to the state adjudication control module, is used to collect the interactive response signal after the scene is issued, calculate the deviation between the actual scene conversion rate and the target benchmark conversion rate within the sampling period, and dynamically fine-tune the adaptive adjudication threshold according to the deviation by a preset step size.
[0022] Preferably, the scene map matching module calculates the topological impedance factor. The calculation model used is as follows: ,in, This represents the vertical floor difference between the mobile terminal and the business scenario nodes. This is the floor impedance attenuation coefficient. To indicate the presence of discrete variables that act as physical barriers, The impedance coefficient of the wall medium.
[0023] Preferably, the matching weight value is directly proportional to the semantic interest similarity metric value in the semantic attributes and inversely proportional to the anisotropic spatiotemporal impedance measure; the anisotropic spatiotemporal impedance measure is jointly calculated and determined by the physical plane distance between the mobile terminal and the commercial scene node, the preset benchmark reference distance, and the topological impedance factor.
[0024] Preferably, when the actual conversion rate is lower than the target baseline conversion rate, the deviation self-compensation feedback module raises the adaptive decision threshold for the next sample by a preset gain step size; when the actual conversion rate is higher than the target baseline conversion rate, the adaptive decision threshold for the next sample is lowered by a preset gain step size.
[0025] Preferably, the spatial hysteresis buffer module is set with a dynamic spatial protection boundary radius; when the position change of the mobile terminal does not exceed the dynamic spatial protection boundary radius, the spatiotemporal coordinate analysis module keeps the current grid index state unchanged; when the position change of the mobile terminal continuously exceeds the dynamic spatial protection boundary radius for a preset time window, the spatiotemporal coordinate analysis module updates the grid index state and reconstructs the dynamic state vector.
[0026] Preferably, the spatiotemporal coordinate analysis module adaptively scales the radius of the dynamic space protection boundary based on the real-time crowd density within the commercial scene node; when the real-time crowd density increases, the radius of the dynamic space protection boundary is reduced by a preset ratio; when the real-time crowd density decreases, the radius of the dynamic space protection boundary is expanded by a preset ratio.
[0027] Preferably, the state adjudication control module includes a rule reconstruction module. The rule reconstruction module is used to solve the Pareto optimal equilibrium point based on the dynamic state vector of each mobile terminal when multiple mobile terminals are concurrently matched to the same business scenario node, and generate collaborative social rules that include multi-party participation thresholds and tiered preferential rights.
[0028] Preferably, the spatiotemporal coordinate parsing module maps the continuous spatiotemporal coordinate sequence to a three-dimensional meshed spatial index bucket, and combines the instantaneous displacement velocity vector and direction offset of the mobile terminal to generate a dynamic state vector containing spatiotemporal compactness and motion trend.
[0029] Preferably, the state adjudication control module synchronously sends the social scene generation instructions to the mobile terminal and the on-site control module of the commercial scene node; the social scene generation instructions include the effective time window for interaction, spatial physical access constraints, and commercial concession rules.
[0030] Preferably, the spatial hysteresis buffer module performs numerical threshold truncation and state locking on discrete positioning drift in a continuous spatiotemporal coordinate sequence, and blocks the graph structure traversal calculation when the mobile terminal is in a state of slight displacement jitter.
[0031] Example 1: The current mobile terminal is deployed in a commercial complex with a high population density and many walls, and transmits data including latitude at a sampling frequency of 10Hz. ,longitude Sampling timestamp With altitude Continuous spatiotemporal coordinate sequence The spatiotemporal coordinate parsing module receives a continuous spatiotemporal coordinate sequence. This is mapped to a 150m × 150m gridded spatial index structure, and combined with the mobile terminal's instantaneous displacement velocity vector and direction offset to generate a dynamic state vector. subscript To represent the spatiotemporal state, in specific processing, the spatiotemporal coordinate analysis module extracts the changes in longitude, latitude, and altitude between adjacent sampling times, and calculates the instantaneous displacement velocity vector of the mobile terminal in three-dimensional space; it also extracts the change in the angle between the motion direction angle of the mobile terminal on the horizontal coordinate plane and the preset main channel axis to obtain the directional offset. The ratio of the cumulative displacement distance within the sampling time interval to the volume of the spatial bounding box is determined as the spatiotemporal compactness, and the product of the magnitude of the instantaneous displacement velocity vector and the cosine of the directional offset is determined as the motion trend degree. The spatiotemporal compactness is used to characterize the degree of spatial aggregation of the mobile terminal in a specific area, and the motion trend degree is used to characterize the tendency of the mobile terminal to approach the target commercial scene node. Both are scalars and are arranged in a preset order to form a two-dimensional dynamic state vector. This is provided for use by the scene map matching module.
[0032] Mobile terminal A is located in the instantaneous crowd density Maximum capacity density: 60 people per 100 square meters For 100 people per 100 square meters, the baseline protection radius is... 15 meters and density scaling factor Within a commercial scenario node with a value of 0.50, the spatial hysteresis buffer module follows... Meter calculation of dynamic space protection boundary radius , where subscript Indicates real-time crowd density, subscript Indicates the maximum carrying capacity, subscript Indicates the baseline value, subscript The spatial hysteresis buffer module represents the buffer boundary. As crowd density increases, the dynamic spatial protection boundary radius is proportionally reduced to improve location discrimination accuracy in dense crowds. Simultaneously, a time window locking mechanism is employed. The grid index state is only updated when the location change continuously exceeds the dynamic spatial protection boundary radius and reaches a preset time window. Even if the location change is merely an instantaneous coordinate fluctuation, the spatial hysteresis buffer module still applies a numerical threshold to truncate the change, maintaining the grid index state locked. When the location change of mobile terminal A does not exceed 10.5 meters within a 3-second time window, the spatiotemporal coordinate resolution module maintains the current grid index state and dynamic state vector. If the positional change remains unchanged and does not trigger graph structure traversal calculations, then when the positional change exceeds 10.5 meters continuously and reaches a 3-second time window, update the grid index state and reconstruct the dynamic state vector. The reconstruction results are then transmitted to the scene map matching module.
[0033] The scene graph matching module targets the second layer and the real-time activity index. The semantic interest similarity metric is 0.60 and is the same as that of mobile terminal A. To match a business scenario node B with a value of 0.80, the physical plane distance between mobile terminal A and business scenario node B is... The preset reference distance is 25 meters. The vertical floor difference is 100 meters. =0, wall barrier indicator variable Under the condition of zero for no partitions on the same floor, according to Calculate the topological impedance factor , where subscript Indicates real-time activity level, subscript Indicates semantic interest similarity, subscript Indicates physical planar distance, subscript Indicates the reference distance, subscript Indicates the vertical floor difference, subscript Indicates the state of wall obstruction, subscript Indicates topological impedance; This is the floor impedance attenuation coefficient. The impedance coefficient of the wall medium.
[0034] Scene graph matching module according to Calculation of anisotropic spacetime impedance measure subscript Represents spatiotemporal impedance, with preset weighting coefficients. The weighting coefficient is 0.60. It is 0.40 and satisfies The scene graph matching module uses a matching weight model. Calculate the matching weight value between the mobile terminal and the business scenario nodes, where the subscript... Indicates mobile terminal, subscript Representing the business scenario node, after substituting the above parameters, we obtain... Semantic interest similarity metric Anisotropic spacetime impedance measure, normalized to a dimensionless scalar in the 0-1 interval. After normalization to the preset benchmark reference distance, it is also a dimensionless scalar; Real-time activity index The ratio of real-time foot traffic to the historical maximum foot traffic at commercial scene nodes is mapped to a range of 0 to 1. Both terms in the matching weight model are dimensionless values, and the sum of the weight coefficients is 1, making the matching weight value... It remains in the range of 0 to 1.
[0035] For the physical planar distance located on the third floor and from node B in the second-floor commercial scene... For the same 25-meter mobile terminal C, the vertical floor difference The value is 1, and due to the existence of wall partitions, the wall obstruction indicator variable is 1. The impedance attenuation coefficient is 1 at each floor level. The wall dielectric impedance coefficient is 0.80. With a value of 0.50, the topological impedance factor is Anisotropic spacetime impedance measure is Maintain semantic interest similarity metric and real-time activity index Unchanged, calculated Vertical floor slabs and solid walls can cause wireless communication signal attenuation and increase the time people spend traveling around stairs, elevators or corridors. The topological impedance factor, as a multiplicative coefficient of physical plane distance, converts the above signal attenuation and travel detour effects into anisotropic spatiotemporal impedance measures, thereby reducing the matching weight value corresponding to cross-floor and partition wall positions.
[0036] The status adjudication control module receives the matching weight value. and the initial value of the adaptive decision threshold. Compare the subscripts. Indicates the threshold for adjudication, indicated by superscript. This represents the initial value. For mobile terminal A, since... The status adjudication control module generates a social scene generation instruction and simultaneously sends this instruction to the on-site control modules of mobile terminal A and commercial scene node B. The social scene generation instruction includes the effective interaction time window, spatial physical access constraints, and commercial concession rules. The spatial physical access constraints are based on the vertical floor difference. and wall barrier indicator variable Generate; when there are no walls on the same floor, mark it as directly passable; when there are floors or walls, record the access conditions to reach the commercial scene node via stairs, elevators or corridors.
[0037] When multiple mobile terminals concurrently match to the same business scenario node B, the state adjudication control module initiates the rule reconstruction module. The rule reconstruction module, based on the dynamic state vectors of each mobile terminal, performs the rule reconstruction. The system determines the benefits of each mobile terminal participating independently and collaboratively, with constraints that the final benefit of any mobile terminal is not less than its benefit when participating independently, and the total benefit of all mobile terminals does not exceed the maximum total amount of concessions for the commercial scenario node. The Pareto optimal equilibrium point is solved with the goal of maximizing the product of benefits for all parties. The comprehensive matching weight corresponding to this equilibrium point is determined as the threshold for multi-party participation. Then, based on the benefit distribution result corresponding to the equilibrium point, the preferential level of each participant is determined, forming a tiered preferential right and generating collaborative social rules. For mobile terminal C, in subsequent sampling periods, when the deviation self-compensation feedback module adjusts the adaptive adjudication threshold to be higher than its matching weight value of 0.5447, the state adjudication control module intercepts the social scenario generation instructions under conditions of cross-floor and wall separation.
[0038] After the social scene generation command is issued, the deviation self-compensation feedback module collects the interaction response signal and counts the number of times the user accepts the scene command within a sampling period of 100 samples. There were 20 responses received within this sampling period, and the actual scene conversion rate was... The target baseline conversion rate is The preset gain step size is , where subscript Indicates the actual conversion rate in the scenario, subscript Indicates the target baseline conversion rate, and the deviation self-compensation feedback module follows... Calculate the adaptive decision threshold for the next sampling period, where the superscript... Indicates the current sampling period, superscript This indicates the next sampling period. The first correction result after the initial sampling period is... superscript This indicates the first sampling period after the initial period.
[0039] During feedback control, the preset gain step size determines the adjustment speed of the adaptive decision threshold. When the preset gain step size is greater than 0.20, the closed-loop poles will exceed the unit circle, and the adaptive decision threshold will oscillate in adjacent sampling periods. When the preset gain step size is less than 0.01, the system's response speed to changes in pedestrian flow and conversion rate deviations is insufficient. When the preset gain step size is 0.05, the adaptive decision threshold can converge to near the target benchmark conversion rate within 3 to 5 sampling periods. The target benchmark conversion rate is determined based on the historical long-term operational data of commercial scenario nodes, taking the expected median of historical actual scenario conversion rates. When the actual scenario conversion rate is lower than the target benchmark conversion rate, the deviation self-compensation feedback module raises the adaptive decision threshold for the next sampling period to reduce the issuance of low-intent scenarios. When the actual scenario conversion rate is higher than the target benchmark conversion rate, the adaptive decision threshold for the next sampling period is lowered to increase the coverage of high-intent scenarios.
[0040] Example 2: In a building area of 120,000 m² 2 A system test environment was set up inside a multi-story commercial building. The server-side communication server used a 3.2GHz processor and 64GB of RAM. It received continuous spatiotemporal coordinate sequences transmitted by 500 mobile terminals at a sampling frequency of 10Hz through wireless network nodes distributed on the first to third floors. The sequence contains latitude. ,longitude Sampling timestamp and altitude To simulate positioning noise caused by indoor wireless multipath and electromagnetic interference, in a continuous spatiotemporal coordinate sequence Discrete Gaussian noise with an amplitude of 0.5m to 3.2m and base station drift signals with a frequency of 0.1Hz to 0.5Hz are superimposed on the spatiotemporal coordinate analysis module, scene map matching module, state adjudication control module and deviation self-compensation feedback module to record the operation data of the spatiotemporal coordinate analysis module, scene map matching module, state adjudication control module and deviation self-compensation feedback module under three-dimensional physical impedance and high-frequency positioning noise conditions.
[0041] Spatial hysteresis buffer module according to the relation Scaling the dynamic space protection boundary radius During the test, the real-time crowd density within the grid was measured. In every 100m 2 The number of people varies between 20 and 80, with a baseline protection radius. Take 15m, density scaling factor Take 0.50, maximum carrying density Take every 100m 2 100 people; floor impedance attenuation coefficient Take 0.80, the wall dielectric impedance coefficient Set the value to 0.50, and set the preset reference distance. Take 100m, weighting coefficient Set the weighting factor to 0.60. Take 0.40, and satisfy... Target baseline conversion rate Set the gain step size to 0.30. Set the value to 0.05, the initial value for the adaptive decision threshold. Take 0.50.
[0042] The test was divided into the present invention sample group, control group 1 (without topological impedance factor), control group 2 (without spatial hysteresis buffer module), control group 3 (without deviation self-compensation feedback module), and an out-of-range control group (density scaling factor exceeding the set value). Each group ran continuously for 6 hours, with physical plane distance... The semantic interest similarity metric is uniformly set to 25m. Set to 0.80, Real-time Activity Index Set to 0.60, during the test, 50 mobile terminals were selected from 500 and repeatedly moved along a preset trajectory within a designated area. Each mobile terminal continuously collected no less than 1000 location data points. The server used a fixed observation window of 10 seconds and accumulated the number of times the grid index was updated, which was used as the number of state reconstructions. On-site testers verified whether there were any unidentified vertical floors or walls between the mobile terminal corresponding to each social scene generation command and the commercial scene node. The proportion of commands issued across barriers to the total number of commands issued was used as the pseudo-proximity issuance rate. Actual scene conversion rate. The ratio of the number of times a user clicks to receive social scene-generated instructions within the sampling period to the total number of instructions issued is taken, and the average value of each discrete statistical data is taken from multiple consecutive test periods.
[0043] In the same-layer transparent working condition of the sample group of the present invention, the instantaneous population density For every 100m 2 After substituting 60 people into the formula for the radius of the dynamic space protection boundary, The value is 10.5m; neither the vertical floor difference nor the wall barrier indicator variable generates additional impedance, and the topological impedance factor is [value missing]. 1.0, anisotropic spacetime impedance measure The value is 0.25, and the scene graph matching module calculates the matching weight value accordingly. The value was 0.624. This group performed a state reconstruction once every 10 seconds, with a pseudo-close delivery rate of 1.2%. The actual scenario conversion rate was... The value is 0.32, and the deviation self-compensation feedback module follows... The adaptive decision threshold for the next cycle is adjusted to 0.499.
[0044] After changing the spatial configuration to a cross-floor partition wall, the instantaneous crowd density of the sample group in this invention was... Still per 100m 2 60 people, dynamic space protection boundary radius Still 10.5m, vertical floor difference The value is 1, indicating the wall barrier variable. The topological impedance factor is 1, according to Increased to 2.3; the anisotropic spacetime impedance measure correspondingly becomes The matching weight value was reduced from 0.624 under the same-layer transparent condition to The group performed a state reconstruction once every 10 seconds, with a pseudo-near-deployment rate of 0.8% and an actual scenario conversion rate of The threshold is 0.31, and the adaptive decision threshold for the next cycle is 0.4995.
[0045] Control group 1 did not introduce a topological impedance factor. This group was in a cross-floor partition wall condition, with immediate population density. For every 100m 2 60 people, dynamic space protection boundary radius It is 10.5m, but the topological impedance factor is... Still 1.0, matching weight value The value remains at 0.624. Because cross-floor and wall barriers are not reflected in the matching weight value, this group performs a state reconstruction once every 10 seconds, resulting in a pseudo-close delivery rate of 38.6%. The actual scenario conversion rate... The value is 0.12; the deviation self-compensation feedback module adjusts the adaptive decision threshold for the next cycle to 0.509 based on this conversion rate.
[0046] Control group 2 retained the topological impedance calculation but did not configure the spatial hysteresis buffer module. In the case of cross-floor partition walls, the instantaneous crowd density... For every 100m 2 60 people, the protection boundary is in an unbuffered state, topological impedance factor The matching weight value is 2.3. The value is 0.5447. Location noise in the continuous spatiotemporal coordinate sequence directly triggers grid index updates, causing the group to perform 18 state reconstructions within 10 seconds; its pseudo-proximity delivery rate is 0.9%, and the actual scene conversion rate is... The threshold is 0.28, and the adaptive decision threshold for the next cycle is 0.501.
[0047] Control group 3 retained the space hysteresis buffer module and topological impedance calculation, but did not configure the deviation self-compensation feedback module. In the case of cross-floor partition walls, the instantaneous crowd density... For every 100m 2 60 people, dynamic space protection boundary radius The topological impedance factor is 10.5m. The matching weight value is 2.3. The value was 0.5447. This group performed a state reconstruction once within 10 seconds, with a pseudo-near-deployment rate of 0.8%, and an actual scenario conversion rate of... The threshold is 0.18; since no feedback correction is performed, the adaptive decision threshold for the next cycle remains at 0.500.
[0048] During the closed-loop operation of 10 consecutive sampling cycles, when the actual scenario conversion rate... When the conversion rate consistently falls below the target baseline of 0.30, the deviation self-compensation feedback module raises the adaptive adjudication threshold periodically, based on the actual scenario conversion rate. After continuous correction of the low conversion rate of 0.18, the adjudication threshold is raised to above 0.550. At this time, the matching weight value of 0.5447 is lower than 0.550, and the status adjudication control module no longer issues the corresponding cross-floor partition social scene generation instruction. When the actual scene conversion rate is higher than the target benchmark conversion rate of 0.30, the deviation self-compensation feedback module lowers the adaptive adjudication threshold for the next cycle. After continuous operation, the actual scene conversion rate of the sample group of this invention is stable between 0.31 and 0.32.
[0049] The control group beyond the specified range used a same-floor, transparent working condition, with real-time population density. All per 100m 2 60 people, topological impedance factor All are 1.0, matching weight value Both are 0.624, when the density scaling factor is... When the value is 0.10, the radius of the dynamic space protection boundary Expanding to 14.1m, no state reconstruction was performed within 10 seconds, and the recorded number of state reconstructions was 0. Because the change in the mobile terminal's position after the protection boundary expansion could not trigger timely grid index updates, the pseudo-proximity delivery rate was 14.5%, and the actual scenario conversion rate... The adaptive decision threshold for the next cycle is 0.5045, which is 0.21, when the density scaling factor... When the value is 0.90, the radius of the dynamic space protection boundary Reduced to 3.0m, the positioning drift frequently crossed the protection boundary, and 24 state reconstructions were performed within 10 seconds; the false proximity delivery rate for this group was 1.5%, and the actual scene conversion rate was... The value is 0.29, the adaptive decision threshold for the next cycle is 0.5005, and the density scaling factor is... When the value is 0.50, the radius of the dynamic space protection boundary is 10.5m. The number of state reconstructions of the sample group of this invention within 10s is 1 under both the same-floor transparency and cross-floor partition wall conditions.
[0050] Example 3: In commercial exhibition areas where pedestrian traffic is concentrated and location interference is frequent, mobile terminals transmit data including latitude and longitude at a sampling frequency of 10Hz. ,longitude Sampling timestamp With altitude Continuous spatiotemporal coordinate sequence The spatiotemporal coordinate parsing module receives a continuous spatiotemporal coordinate sequence. This is mapped to a 150m × 150m gridded spatial index structure, and combined with the mobile terminal's instantaneous displacement velocity vector and direction offset, to generate a dynamic state vector that includes spatiotemporal compactness and motion trend. .
[0051] Real-time crowd density within the business scenario node where mobile terminal A is located Maximum occupancy density: 80 people per 100 square meters The baseline protection radius is 100 people per 100 square meters. 15 meters, density scaling factor The value is 0.50, and the spatial hysteresis buffer module is configured according to... Meter calculation of dynamic space protection boundary radius When the position change of mobile terminal A does not exceed 9.0 meters within a 3-second time window, the spatiotemporal coordinate resolution module maintains the current grid index state and dynamic state vector. The indoor wireless signal multipath drift remains unchanged and does not trigger graph structure traversal calculations; when the position change continuously exceeds 9.0 meters and reaches a 3-second time window, the grid index state is updated and the dynamic state vector is reconstructed. The reconstruction results are then transmitted to the scene map matching module.
[0052] Mobile terminal A and mobile terminal B are concurrently matched to the second-layer business scene node C within the same spatial grid. The scene graph matching module receives the dynamic state vectors of the two mobile terminals. By performing matching calculations corresponding to node C in the business scenario, the obtained matching weight values are used as projection values of each dynamic state vector toward that node, resulting in... and Both matching weight values are greater than the current adaptive adjudication threshold. The status adjudication control module then initiates the rule reconstruction module based on this.
[0053] The rule reconstruction module extracts the revenue functions of mobile terminal A and mobile terminal B. and ,in, and These represent the game payoffs for the two mobile terminals, respectively. and These are the interaction response time windows for the two mobile terminals in the current interaction process. The interaction response time window starts counting when the instruction generated in the social scene reaches the mobile terminal and ends when the mobile terminal submits and accepts the response. The time obtained is used to calculate the corresponding game payoff value.
[0054] During the rule reconstruction process, the initial revenue of a mobile terminal participating independently is used as the initial revenue. The difference in revenue after it transitions to collaborative participation is calculated and used as the collaborative gain. The final allocated revenue of any mobile terminal must not be lower than its initial revenue when participating independently, and the total revenue of the two mobile terminals must not exceed the maximum total amount of concessions preset by the business scenario node C. Under the above constraints, with the goal of maximizing the product of multi-party revenues, the Lagrange multiplier method is used to perform numerical iteration on the boundary of the convex feasible region determined by the revenue function.
[0055] The numerical iteration uses the current adaptive adjudication threshold of 0.505 as the lower bound of the candidate threshold and the smaller of the two matching weight values of 0.580 as the upper bound of the candidate threshold. For each candidate threshold, the rule reconstruction module will... and Each mobile terminal is compared to the threshold, and those that meet the candidate threshold are retained. Then, the corresponding interaction response time window is used to calculate... and Simultaneously, the two constraints of ensuring the final return is not less than the return of independent participants and that the total return does not exceed the maximum total concession amount are verified. The return products under different candidate thresholds are compared, and after numerical iteration convergence, the comprehensive matching weight threshold corresponding to when the return product reaches its maximum value is determined as the multi-party participation threshold. .
[0056] The rule reconstruction module solves for the following conditions: and The Pareto optimal equilibrium point yields the threshold for multi-party participation. ,because and Both mobile terminals A and B simultaneously enter the collaborative participation state once the multi-party participation threshold is met. The rule reconstruction module uses the 10% discount rate for independent participation as the basic tier. Under the condition that it does not exceed the maximum total discount amount of the business scenario node C, the collaborative gain corresponding to the equilibrium point is included in the tier for joint participation by two people, forming a tiered preferential benefit with a 15% discount rate for two people and a 10% discount rate for one person, and generating a collaborative social rule with an effective interaction time window of 300 seconds.
[0057] The status adjudication control module synchronously sends the social scene generation instruction containing the collaborative social rules to the on-site control module of mobile terminal A, mobile terminal B, and business scene node C. After the scene is sent, the deviation self-compensation feedback module collects the interaction response signal and calculates the actual scene conversion rate within a sampling period of 100 samples. The actual conversion rate in this scenario is higher than the target baseline conversion rate. The deviation self-compensation feedback module follows Calculate the adaptive decision threshold for the next cycle to obtain After the adaptive adjudication threshold is reduced from 0.505 to 0.5025, the state adjudication control module processes the concurrent matching results of the next sampling period according to the updated threshold, and continues to adjust the matching threshold and collaborative social rules based on subsequent interaction responses.
[0058] Example 4: This example combines Figures 1 to 2 This paper describes an AI-driven autonomous social scene generation system that uses spatiotemporal coordinate linkage, such as... Figure 1As shown, a continuous spatiotemporal coordinate sequence is transmitted to a spatiotemporal coordinate parsing module. The spatiotemporal coordinate parsing module includes a spatial hysteresis buffer module to filter coordinate jitter and generate a dynamic state vector. The dynamic state vector is output to a scene graph matching module. The scene graph matching module performs the operation of calculating the matching weight value and transmits the matching weight value to a state adjudication control module. The state adjudication control module performs the operation of comparing the adaptive threshold and outputs a scene generation command to issue a social scene generation command. At the same time, the terminal interaction feedback generates an interaction response signal and inputs it to a deviation self-compensation feedback module. The deviation self-compensation feedback module performs the operation of fine-tuning the adaptive adjudication threshold and feeds back the fine-tuned adaptive adjudication threshold to the state adjudication control module.
[0059] like Figure 2 As shown, the system receives a continuous spatiotemporal coordinate sequence and maps it to a gridded spatial index structure. It then generates a dynamic state vector by combining the instantaneous displacement velocity vector and the direction offset. By scaling the radius of the dynamic spatial protection boundary, it performs threshold filtering on coordinate displacement jitter. A topological impedance factor is introduced to calculate the anisotropic spatiotemporal impedance measure. The system combines the processing results of obtaining the real-time activity index and the semantic attributes to calculate the matching weight value. The data is then passed to the matching weight value for comparison with the adaptive adjudication threshold. When the data is greater than or equal to the adaptive adjudication threshold, a social scene generation command is output. The system collects terminal interaction response signals and dynamically fine-tunes the adaptive adjudication threshold. Finally, the processed results are cyclically fed back to the processing step of comparing the matching weight value with the adaptive adjudication threshold.
[0060] Example 5: Before the commercial scene node accesses the system, the mobile terminal collects indoor wireless signal strength attenuation data along a preset inspection path within the building. It measures the signal attenuation and travel time under conditions of no obstruction on the same floor, across floors, and through solid walls. The scene map matching module fits the impact of floor span and wall medium on signal attenuation and travel time based on the collected results, and calculates the floor impedance attenuation coefficient. The calibrated value is 0.80, which represents the dielectric impedance coefficient of the wall. The calibration is set to 0.50, and the calibration result is stored in memory.
[0061] During calibration, testers used a mobile terminal equipped with a standard wireless signal receiving module to sample along a preset inspection path in unobstructed areas on the same floor, areas spanning single floors, and areas separated by walls. Under unobstructed conditions on the same floor, the average signal attenuation corresponding to different physical plane distances was recorded as baseline data. Under conditions spanning multiple floors, while maintaining a consistent physical plane distance, the additional signal attenuation caused by the floor slabs and the travel time difference were measured. The additional signal attenuation and the increase in travel time were then fitted with the baseline data using a least-squares curve to obtain the floor impedance attenuation coefficient. Under the condition of a partition wall, the signal attenuation difference and passage obstruction on both sides of a solid brick wall or reinforced concrete wall are measured, and the wall dielectric impedance coefficient is obtained by fitting using the same method. .
[0062] When the measured impedance coefficient is below 0.10, the correction amplitude of the floor or wall for the anisotropic spatiotemporal impedance measurement is small, and a high matching weight value may still be obtained at the barrier-crossing position. When the impedance coefficient is above 2.00, the matching weight values corresponding to adjacent positions on both sides of the barrier decrease significantly, and normal proximity matching is also suppressed. Therefore, the scene map matching module completes the impedance coefficient calibration within the range of 0.10 to 2.00, and adopts... Calculate the topological impedance factor When the floor structure or wall position inside the building changes, the mobile terminal re-collects the signal attenuation and travel time of the corresponding area, and the scene map matching module updates the floor impedance attenuation coefficient in the memory accordingly. and wall dielectric impedance coefficient The topological impedance factor was recalculated using the updated coefficients. .
[0063] During system operation, the spatiotemporal coordinate analysis module extracts historical interaction data of business scenario nodes at the beginning of each sampling period, calculates the average actual scenario conversion rate within the sliding window, and determines this average value as the target benchmark conversion rate for the current sampling period. To adapt to changes in pedestrian flow between weekdays and holidays.
[0064] When the instantaneous population density within the grid When the number of people reaches 100 per 100 square meters, the baseline protection radius is... Take 15 meters, density scaling factor Take 0.50, maximum carrying density Taking 100 people per 100 square meters, the spatial hysteresis buffer module is designed according to... Calculate the radius of the dynamic space protection boundary The resulting dynamic space protection boundary radius is 7.5 meters. When the position change of the mobile terminal does not exceed 7.5 meters continuously, the spatiotemporal coordinate analysis module maintains the current grid index state; when the position change exceeds 7.5 meters continuously and reaches the preset time window, the grid index state is updated and the dynamic state vector is reconstructed.
[0065] The deviation self-compensation feedback module continuously compares the conversion rate in the actual scenario. Conversion rate relative to target baseline When the absolute value of the deviation between the two exceeds 0.05 for three consecutive sampling periods, according to Adjusting the adaptive decision threshold for the next sampling period, actual scenario conversion rate. Conversion rate below target baseline At the same time, increase the adaptive decision threshold for the next sampling period; actual scenario conversion rate Higher than the target baseline conversion rate When the adaptive decision threshold for the next sampling period is lowered, the adaptive decision threshold is adjusted periodically according to the actual scene conversion rate changes on different dates and at different times. The state decision control module controls the issuance of social scene generation instructions according to the updated threshold.
[0066] Example 6: When wireless communication signals in a commercial setting are affected by building obstruction or multipath reflection, the mobile terminal continuously transmits signals containing latitude at a sampling frequency of 10Hz. ,longitude Sampling timestamp With altitude Continuous spatiotemporal coordinate sequence The spatiotemporal coordinate parsing module receives a continuous spatiotemporal coordinate sequence. Before mapping it to a gridded spatial index structure, the coordinate data is de-identified and hash-mapped for dimensionality reduction, isolating sensitive personal identity attributes and forming a data stream without identity identifiers.
[0067] When the positioning signal is lost discretely, or when coordinate data is missing for three consecutive sampling points, the spatiotemporal coordinate analysis module starts a spatiotemporal smoothing compensation algorithm based on Kalman filtering. It calculates the position of the mobile terminal by combining the instantaneous displacement velocity vector and direction offset before the signal interruption, and keeps the current grid index state locked. During the compensation process, the spatiotemporal coordinate analysis module sets the acceleration term in the state transition matrix to zero, establishes a prediction equation according to the uniform linear motion state, and uses the instantaneous displacement velocity vector and direction offset at the last valid sampling moment before the signal interruption to recursively deduce the predicted position of the mobile terminal for each sampling period.
[0068] The maximum estimation time for the predicted location does not exceed 5 sampling periods. The covariance matrix of the predicted state accumulates exponentially with the estimation time. When the estimation time reaches the maximum estimation time, or the accumulated covariance exceeds the preset safety threshold, the spatiotemporal coordinate analysis module terminates the location estimation, marks the location state of the mobile terminal as unknown, and no longer uses the predicted location to trigger grid index updates. Within the maximum estimation time, uniform state recursion is used to compensate for transient signal interruptions within 1 second, keeping the current grid index state unchanged, and avoiding erroneous state reconstruction and invalid graph structure traversal calculations triggered by instantaneous packet loss.
[0069] After the mobile terminal recovers the continuous positioning signal and retransmits the coordinate data, the spatiotemporal coordinate analysis module calculates the displacement deviation between the received coordinates and the predicted coordinates, and then compares this displacement deviation with the radius of the dynamic space protection boundary corresponding to the current grid. For comparison, the radius of the dynamic space protection boundary is calculated according to the following formula: When the displacement deviation does not exceed the radius of the dynamic space protection boundary At the same time, the spatiotemporal coordinate analysis module maintains the current grid index state and continues to stably output the dynamic state vector. When the displacement deviation continuously exceeds the radius of the dynamic space protection boundary Upon reaching the 3-second time window, update the grid index state and reconstruct the dynamic state vector. The updated dynamic state vector is then transmitted to the scene graph matching module.
[0070] The scene graph matching module matches the updated dynamic state vector. Recalculate the matching weight values between the mobile terminal and the business scenario nodes. The status adjudication control module will update the matching weight value. With adaptive decision threshold The system performs a comparison and controls the output of social scene generation instructions according to the comparison results, so that the system can maintain the stability of scene delivery under high concurrency and location signal disturbance conditions.
[0071] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A spatiotemporal coordinate-linked AI autonomous social scene generation system, characterized in that, The system includes: The spatiotemporal coordinate parsing module is used to receive a continuous spatiotemporal coordinate sequence sent by the mobile terminal, which includes longitude, latitude, sampling timestamp and altitude. It maps the continuous spatiotemporal coordinate sequence to a gridded spatial index structure. Combined with the mobile terminal's real-time displacement velocity vector and direction offset, it generates a dynamic state vector that includes spatiotemporal compactness and motion trend. The spatial hysteresis buffer module, which adapts to the real-time crowd density in the commercial scene node, performs threshold filtering on coordinate displacement jitter. The scene graph matching module, connected to the spatiotemporal coordinate parsing module, is used to obtain the real-time activity index and semantic attributes of commercial scene nodes, receive dynamic state vectors, introduce the topological impedance factor determined by the vertical floor difference and physical barrier features to calculate the anisotropic spatiotemporal impedance measure, and calculate the matching weight value between the mobile terminal and the commercial scene node. The state adjudication control module is connected to the scene graph matching module. It is used to compare the matching weight value with the adaptive adjudication threshold and output the social scene generation command when the matching weight value is greater than or equal to the adaptive adjudication threshold. as well as The deviation self-compensation feedback module, connected to the state adjudication control module, is used to collect the interactive response signal after the scene is issued, calculate the deviation between the actual scene conversion rate and the target benchmark conversion rate within the sampling period, and dynamically fine-tune the adaptive adjudication threshold according to the deviation by a preset step size.
2. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 1, characterized in that, The scene graph matching module calculates the topological impedance factor. The calculation model used is as follows: ,in, This represents the vertical floor difference between the mobile terminal and the business scenario nodes. This is the floor impedance attenuation coefficient. To indicate the presence of discrete variables that act as physical barriers, The impedance coefficient of the wall medium.
3. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 2, characterized in that, The matching weight value is directly proportional to the semantic interest similarity metric in the semantic attributes and inversely proportional to the anisotropic spatiotemporal impedance measure. The anisotropic spatiotemporal impedance measure is jointly calculated and determined by the physical plane distance between the mobile terminal and the commercial scene node, the preset benchmark reference distance, and the topological impedance factor.
4. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 1, characterized in that, When the actual conversion rate is lower than the target baseline conversion rate, the deviation self-compensation feedback module raises the adaptive decision threshold for the next sample by a preset gain step size; when the actual conversion rate is higher than the target baseline conversion rate, it lowers the adaptive decision threshold for the next sample by a preset gain step size.
5. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 1, characterized in that, The spatial hysteresis buffer module is set with a dynamic spatial protection boundary radius. When the position change of the mobile terminal does not exceed the dynamic spatial protection boundary radius, the spatiotemporal coordinate analysis module keeps the current grid index state unchanged. When the position change of the mobile terminal continuously exceeds the dynamic spatial protection boundary radius for a preset time window, the spatiotemporal coordinate analysis module updates the grid index state and reconstructs the dynamic state vector.
6. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 5, characterized in that, The spatiotemporal coordinate analysis module adaptively scales the radius of the dynamic space protection boundary based on the real-time crowd density within the business scene node; when the real-time crowd density increases, the radius of the dynamic space protection boundary is reduced by a preset ratio; when the real-time crowd density decreases, the radius of the dynamic space protection boundary is expanded by a preset ratio.
7. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 1, characterized in that, The state adjudication control module includes a rule reconstruction module. The rule reconstruction module is used to solve the Pareto optimal equilibrium point based on the dynamic state vector of each mobile terminal when multiple mobile terminals are concurrently matched to the same business scenario node, and generate collaborative social rules that include multi-party participation thresholds and tiered preferential rights.
8. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 1, characterized in that, The spatiotemporal coordinate parsing module maps the continuous spatiotemporal coordinate sequence to a three-dimensional meshed spatial index bucket, and combines the mobile terminal's instantaneous displacement velocity vector and direction offset to generate a dynamic state vector that includes spatiotemporal compactness and motion trend.
9. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 1, characterized in that, The status adjudication control module synchronously sends social scene generation instructions to mobile terminals and on-site control modules of commercial scene nodes; the social scene generation instructions include the effective time window for interaction, spatial physical access constraints, and commercial concession rules.
10. The AI autonomous social scene generation system with spatiotemporal coordinate linkage according to claim 5, characterized in that, The spatial hysteresis buffer module performs numerical threshold truncation and state locking for discrete positioning drift in a continuous spatiotemporal coordinate sequence, and blocks the graph structure traversal calculation when the mobile terminal is in a state of slight displacement jitter.
