Augmented reality safety management support system and control method using a pre-generated three-dimensional reference space in indoor facilities.

Abstract

This invention improves event safety control capabilities in indoor facilities. [Solution] The facility is pre-scanned using LiDAR, etc., and a permanent 3D reference coordinate system DB is permanently stored. Staff terminals align their own position by matching camera feature points. Central control generates AR instructions based on congestion level and attributes using role-based mutual exclusion, and location updates, route recalculation, and synchronous distribution are triggered by abnormalities. Collision avoidance is autonomously determined via P2P between terminals, and instruction × coordinate × time × terminal ID × movement history is set and stored in a way that allows for tamper detection.

Description

[[Technical Field]]

[0001] The present invention relates to event operation and safety management support technology for indoor facilities such as concert venues, exhibition facilities, and sports facilities, and particularly to a system and a control method thereof that integrate congestion density information and staff position information based on pre-generated facility reference three-dimensional spatial data, and dynamically generate individually optimized augmented reality (AR) instructions for each staff terminal. [[Background Art]]

[0002] In conventional indoor event operation, instruction transmission via audio wireless communicators has been mainstream. However, mistransmission is likely to occur in noisy environments, and since position information and instruction information are not integrated, rapid and accurate control has been difficult.

[0003] Furthermore, when an abnormal event such as a threat occurs, organizers and facility managers need to quickly determine whether operation can be continued while ensuring the safety of visitors. However, with only conventional audio instructions and monitoring videos, the correspondence between instruction content and actual staff behavior is not systematically recorded, and there tends to be a lack of trails to satisfy accountability requirements (post-event verification, insurance handling, investigation cooperation).

[0004] The present invention is not directed to simple evacuation guidance for general visitors, but to advanced work control for on-site staff (security, guidance, first aid, reception, etc.) performing work in facilities. The technical problem addressed by the present invention is to achieve synchronous deployment of instructions during abnormal events, suppression of information overload, and ensuring evidentiality of the control (correspondence among instructions, coordinates, time, and behavior). Although various prior arts exist for indoor augmented reality technology, these require re-calibration for each event, and safety control technology based on the premise of a permanently installed facility spatial reference has not been established. In addition, since outdoor positioning technologies such as GPS cannot be used indoors, a separate means for stably maintaining high-precision position alignment is required.

[0005] Furthermore, conventional staff instruction systems were limited to one-way instruction distribution from a central control unit, resulting in inefficiencies such as multiple staff members having to move to the same area repeatedly, and failing to provide selective information tailored to staff attributes. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2025-128093 [Patent Document 2] Japanese Patent Publication No. 2016-210230 (Domestic publication corresponding to JPWO2016170767A1, crowd guidance / crowd indicators) [Patent Document 3] US9864878B2 (Tampering detection log) [Non-patent literature]

[0007] [Non-Patent Document 1] AAMAS2023: Asynchronous Execution Research on Distributed Multi-Agent Path Planning (MAPF) [Overview of the project] [Problems that the invention aims to solve]

[0008] The present invention aims to solve the following problems in order to improve event safety control capabilities in indoor facilities.

[0009] To generate a facility-specific, persistent three-dimensional reference coordinate system, eliminating the need for recalibration after each event.

[0010] To quantify real-time congestion risk using a unique congestion evaluation function that combines spatial density, flow velocity deviation, and dwell time.

[0011] To achieve autonomous route optimization where the final choice for collision avoidance is determined by the terminal through cooperative processing between staff terminals.

[0012] To implement exclusive filtering of augmented reality information according to the role type of staff member.

[0013] In response to an abnormal condition detection trigger, the system will update and acquire the location of staff within the target area, synchronously distribute the updated guidance instructions to staff terminals, and save the instructions and location history in a format that allows for tamper detection, thereby ensuring the possibility of post-incident verification.

[0014] By assigning the central control unit the responsibility for role-based exclusive display and synchronous distribution, while entrusting final mediation, such as collision avoidance at the site, to the autonomous distributed control on the terminal side, the system can avoid disrupting business control even if communication delays or congestion occur.

[0015] To enable post-event verification and use as legal evidence by preserving evidence that can be detected for tampering. [Means for solving the problem]

[0016] System Configuration Overview The system of the present invention comprises a pre-mapping device (10), a spatial reference generation unit (20), a facility reference database (30), a staff terminal (40), a central control unit (50), and an evidence storage unit (60).

[0017] The pre-mapping device (10) is equipped with a LiDAR sensor and an inertial measurement unit (IMU) to acquire three-dimensional point cloud data of indoor facilities. The spatial reference generation unit (20) generates a facility-specific three-dimensional reference coordinate system from the point cloud data and permanently stores it in the facility reference database (30). The staff terminal (40) aligns its own position by matching feature points of camera images and fusion processing of multiple positioning sources. The central control unit (50) is responsible for congestion evaluation, optimal route calculation, role-specific augmented reality instruction generation, and provision of supervisory information. The evidence storage unit (60) stores the displayed instruction information and location history in a format that allows for tamper detection.

[0018] (2) Pre-mapping process A pre-mapping apparatus (10) scans an indoor facility using a LiDAR sensor and an IMU to acquire three-dimensional point cloud data. A spatial reference generation unit (20) integrates the acquired point cloud data, extracts persistent feature points based on fixed structures (columns, wall surfaces, emergency exits, etc.), and generates a facility-specific three-dimensional reference coordinate system.

[0019] The generated reference coordinate system is stored in a facility reference database (30), and is operated as a persistent reference that does not require re-measurement for each event after being acquired once. This eliminates the calibration man-hours for each event required by conventional techniques.

[0020] (3) Position alignment processing during operation (anchor fusion) A staff terminal (40) collates feature points extracted from camera images against the facility reference database (30), and aligns its own position to the reference coordinate system. Furthermore, the staff terminal executes anchor fusion processing that integrates absolute coordinates based on distance measurement data acquired from fixed beacons (70) and relative coordinates based on matching processing of three-dimensional point clouds by a state estimation filter.

[0021] As the state estimation filter, methods based on Bayesian estimation such as a Kalman filter, an extended Kalman filter, and a particle filter can be applied, and as the point cloud matching processing, an ICP (Iterative Closest Point) algorithm or the like can be used.

[0022] In addition, pre-registration of stationary structures in the facility as spatial feature anchors is performed, and drift correction processing that resets cumulative errors each time the staff terminal (40) re-recognizes the anchors is executed. This stably maintains highly accurate position alignment even in large-scale indoor facilities.

[0023] (4) Congestion degree evaluation function A central control device (50) calculates a congestion degree score in real time by using a congestion degree evaluation function including at least three variables, namely a spatial density term, a flow velocity deviation term and a residence time term. It is preferable to calculate the congestion degree score as a weighted sum obtained by multiplying each of the three terms by a weighting factor that can be changed according to an operation mode, or as a composite function having a monotonic increase characteristic equivalent thereto, and the following mathematical formula can be used as one embodiment.

[0024] C = α(N / A) + β(Vavg - Vtarget) + γΔT

[0025] In the above mathematical formula, N / A is spatial density (the number of people N per unit area A), which is calculated from point cloud analysis by LiDAR scanning. Vavg - Vtarget is flow velocity deviation, and the more the measured average moving speed Vavg is lower than Vtarget (target speed), the more obvious the residence tendency is. ΔT is residence time, which is the cumulative time that a specific individual continues to stay at the same coordinate. α, β and γ are weighting factors that are dynamically changed according to the operation mode.

[0026] Dynamic switching of weighting factors and stationary individual detection mode In a normal operation state, the weighting factor α of the spatial density term is preferentially increased to realize early detection of a congested area. When transitioning to the stationary individual detection mode with at least one of congestion threshold excess detection, operator input, external sensor notification and establishment of a predetermined rule as a stationary individual detection trigger, the weighting factor γ of the residence time term is preferentially increased, and stationary individuals staying at the same coordinate for a long time are highlighted on an augmented reality display. This switching process supports the early detection of people with sudden illness, people who have fallen and other such people.

[0027] (5) Role-based visual exclusive control The central control unit (50) dynamically filters the augmented reality information to be displayed based on the role type pre-registered as attribute information of the staff terminals (40). Guidance staff are exclusively provided with guidance route information to crowded areas and visitor guidance arrows, rescue staff are exclusively provided with the shortest route to rescue facilities (AED / first-aid room), and monitoring staff are exclusively provided with stationary individual highlight information and a congestion score map.

[0028] Information corresponding to each role is distributed exclusively, preventing delays in decision-making due to information overload. Role types can be changed through pre-registration, allowing for flexible operation according to the event scale and venue characteristics.

[0029] (6) Cooperative path optimization (distributed MAPF) Staff terminals (40) exchange predicted travel paths (estimated paths) with other staff terminals using a P2P communication protocol such as gRPC or WebRTC. If spatial overlap is detected between the predicted path of one terminal and the predicted path of another terminal, the central control unit (50) only provides supervisory information and performs cooperative path optimization processing (distributed multi-agent pathfinding, MAPF) to autonomously determine the final selection of an alternative path to avoid collisions through terminal-local processing.

[0030] By placing the decision-making authority on the terminal side, the computational load on the central control unit (50) is reduced while automatically avoiding duplicate movements (matchmaking) of multiple staff members.

[0031] (7) Abnormal condition handling process When an abnormal state is detected, triggered by the detection of exceeding a congestion threshold, operator input, external sensor notification, or the fulfillment of a predetermined rule, the system updates the location of staff within the target area, performs route recalculation considering the congestion concentration area and entry restriction area, and synchronously distributes the updated augmented reality instructions to each staff terminal (40). At the same time, the system increases the weight coefficient γ as it transitions to stationary individual detection mode, prioritizing the early detection of lingering individuals.

[0032] (8) Evidence preservation process The evidence storage unit (60) stores the displayed augmented reality instructions, display time, terminal identifier, staff location, and movement history as a set in a tamper-proof format. The terminal identifier is identification information that uniquely identifies each staff terminal (40), and by reliably recording the correspondence between the instructions and the terminal / staff, it enhances accountability after the event and the reliability of the evidence as legal proof. As a tamper-proof format, encryption, hash value addition, digital signature, and storage on a WORM (Write Once Read Many) recording medium can be employed. The stored data is used for post-event security verification, staff behavior analysis, and as legal evidence as needed. [Effects of the Invention]

[0033] According to the present invention, the facility standard spatial database reduces the need for reconfiguration for each event, reduces misidentification through role-based exclusive display in the event of anomalies, and maintains final on-site mediation through autonomous terminal avoidance. Furthermore, by setting and storing augmented reality instructions, location, time, and terminal identifier in a way that allows for tamper detection, it is possible to technically guarantee accountability for post-incident verification, insurance claims, and cooperation with investigations.

[0034] By persisting a facility-specific three-dimensional reference coordinate system, recalibration after each event becomes unnecessary, reducing operational costs.

[0035] The congestion level evaluation function, which uses a weighted sum or a composite function with equivalent monotonically increasing characteristics, allows for dynamic switching of the weighting of the three variables according to the operating conditions, enabling both normal operation and stationary individual detection modes to be implemented in a single system.

[0036] Anchor fusion and drift correction, which integrate multiple positioning sources with a state estimation filter, enable high-precision position alignment even in large indoor facilities.

[0037] Role-based visual mutual exclusion control can improve staff decision-making speed and prevent errors caused by information overload.

[0038] By using distributed MAPF, where the terminal side makes the final decision, the load on the central control unit (50) can be reduced while automatically avoiding duplicate movements of multiple staff members.

[0039] Preserving evidence that can be detected for tampering allows for post-event verification and its use as legal evidence. [Brief explanation of the drawing]

[0040] [Figure 1] This is a diagram showing the overall system configuration of the present invention.

[0041] [Figure 2] This is a flowchart illustrating the control method of the present invention, including processing branching for normal operation and abnormal condition response.

[0042] [Figure 3] This is a conceptual diagram illustrating the dynamic switching of the congestion evaluation function and weight coefficients.

[0043] [Figure 4] This is a conceptual diagram illustrating the operation of role-based visual mutual exclusion, showing the differences in augmented reality information delivered for each role type.

[0044] [Figure 5] This is a conceptual diagram of anchor fusion processing and drift correction, illustrating the integrated processing of multiple positioning sources. [Examples]

[0045] Example 1: Indoor concert venue (capacity of approximately 5,000 people)

[0046] (a) Venue specifications This embodiment is an example of applying the present invention in an indoor concert venue with a capacity of 5,000 people (floor area: approximately 3,000 m2, ceiling height: approximately 12 m). The main floor area of ​​the venue is 2,000 m2, and the corridors and surrounding areas are 1,000 m2. The staff consisted of 23 people in total: 15 for guidance, 3 for medical assistance, and 5 for monitoring. 32 UWB beacons (70) were placed on pillars and walls as fixed beacons (average coverage distance 15 m). AR glasses (field of view 52°, weight 85 g) were used as staff terminals (40), and the central control unit (50) was installed on-premise in the venue's server room.

[0047] (b) Implementation of the pre-mapping process Three days before the event, a pre-mapping device (10) was used to scan the venue and acquire approximately 120 million three-dimensional point cloud data points. The spatial reference generation unit (20) extracted 1,240 persistent feature points such as columns, emergency exits, and fixed equipment from the acquired data and generated a facility-specific three-dimensional reference coordinate system. The generated reference coordinate system is stored in the facility reference database (30), eliminating the need for remapping for subsequent events.

[0048] (c) Example of calculating congestion score The following is an example of calculating the congestion score in the central area (100m2) of the main floor 30 minutes before the performance starts (during peak entry time). LiDAR point cloud analysis measured N=420 people (A=100m2), average movement speed Vavg=0.4m / s (Vtarget=1.0m / s), and dwell time ΔT=8 minutes. Using the weighting coefficients for normal operation mode (α=0.5, β=0.3, γ=0.2),

[0049] C = 0.5×4.2 + 0.3×(-0.6) + 0.2×8 = 2.10 - 0.18 + 1.60 = 3.52

[0050] The calculated congestion score of 3.52 exceeded the pre-set threshold of 3.0, so the central control unit (50) activated an abnormal condition detection trigger and distributed AR instructions for congestion dispersion guidance to the guidance staff terminal (40).

[0051] (d) Examples of AR instruction distribution by role The terminals of the guidance staff (15 people) displayed arrows and detours to guide visitors to the congested area, and they began guiding visitors to aisle B. The terminals of the first aid staff (3 people) displayed the shortest route to the AED location, and they moved their waiting position to near the congested area. The terminals of the monitoring staff (5 people) highlighted two stationary individuals whose dwell time ΔT exceeded 5 minutes, and they conducted verbal checks on these individuals.

[0052] (e) Processing flow for emergency recalculation Ten minutes before the show began, a monitoring staff member checked for stationary individual highlights and discovered a person who was suddenly ill (unconscious). The following emergency response was carried out according to the following procedure.

[0053] Step 1 (0 seconds): The monitoring staff member presses the emergency alarm button on the terminal. The abnormal condition detection trigger is activated.

[0054] Step 2 (0.3 seconds): The central control unit (50) updates and acquires the positions of all 23 staff members. The coordinates of the medical emergency are set as a danger area.

[0055] Step 3 (0.8 seconds): Switch to stationary individual detection mode. Switch the weight coefficient γ from 0.2 to 0.8. Highlight nearby stationary individuals on all terminals.

[0056] Step 4 (1.2 seconds): The shortest route AR to the sick person's coordinates is distributed to the terminals (40) of the three medical personnel. Instructions for dispersing personnel in the surrounding area are distributed to the guidance personnel.

[0057] Step 5 (1.5 seconds): Distributed MAPF detects route overlaps for the three rescue personnel. Local terminal processing automatically adjusts arrival times to determine the first person to arrive.

[0058] Step 6 (2.1 seconds): Update instructions are successfully distributed to all staff terminals (40). The time, location, terminal identifier, and AR instructions for each step are saved in the evidence storage unit (60) along with their hash values.

[0059] It took approximately 2.1 seconds from the time the alarm was issued until the instructions were updated for all staff.

[0060] The present invention is not limited to the above embodiments, but includes the following modifications.

[0061] Modification 1) Modification of positioning source: In the above embodiment, a UWB beacon was used, but the positioning source is not limited to this. The same can be implemented as long as a configuration is used in which at least two of multiple positioning sources, such as BLE (Bluetooth Low Energy) beacons, Wi-Fi RTT (Round-Trip Time), visible light communication (LiCOM), and geomagnetic mapping, are integrated with a state estimation filter.

[0062] Modification 2) Modification of staff terminal: In the above embodiment, an AR glasses-type terminal was used, but the staff terminal (40) is not limited to AR glasses. It can be implemented similarly with any mobile terminal equipped with a camera and display, such as a smartphone, tablet, head-mounted display (HMD), or smartwatch. When using a smartphone, the camera image is displayed as an AR overlay, and the feature point matching process with the facility standards database (30) is performed in the same way.

[0063] Modification 3) Configuration variation of the central control unit: In the above embodiment, the central control unit (50) was installed on-premise in the server room within the venue, but it is also possible to configure it on a cloud server. In the case of a cloud configuration, it is preferable to have a hybrid configuration in which edge nodes are placed within the venue to ensure low-latency communication with staff terminals (40), while heavy-load processing such as congestion evaluation and route calculation is performed on the cloud side.

[0064] Modification 4) Expansion of Role Types: In the above embodiment, three roles—guidance, rescue, and monitoring—were used as examples, but the role types are not limited to these. Any role type can be pre-registered according to the actual operation of the facility, such as firefighting, security, equipment management, and VIP support. In addition, if one staff member holds multiple roles, the system may be configured to display AR information based on the priority role.

[0065] Modification 5) Modification of the congestion evaluation function: In the above embodiment, a linear weighted sum was used, but it is also possible to use a composite function having equivalent monotonically increasing characteristics. For example, configurations in which each term is normalized by a logistic function and then integrated, and configurations in which a threshold function is applied to ΔT to achieve a rapid increase in weights are also included in the scope of the present invention. The weight coefficients α, β, and γ may be configured to be automatically optimized from past event data using machine learning.

[0066] Modification 6) Variation of evidence preservation: In the above embodiment, tamper detection was performed by adding a hash value, but it can be similarly implemented in any format that allows for tamper detection, such as encryption (AES-256, etc.), digital signature, storage on a WORM recording medium, or recording on a distributed ledger using blockchain.

[0067] Modification 7) Variation of the applicable facility: Although the present invention has shown an example of application to a concert venue, it can be similarly implemented in any indoor facility with a fixed structure, such as exhibition facilities, sports facilities, airport terminals, railway station premises, large commercial facilities, theme parks, etc. [Explanation of Symbols]

[0068] 10 Pre-mapping device

[0069] 20 Spatial reference generator

[0070] 30 Facility Standards Database

[0071] 40 Staff terminals

[0072] 50 Central Control Unit

[0073] 60 Evidence Preservation Department

[0074] 70 Fixed beacons (UWB, etc.)

[0075] C Congestion score

[0076] N / A Spatial density term

[0077] Vavg Average Movement Speed

[0078] Vtarget target movement speed

[0079] ΔT Residence time term

[0080] α, β, γ coefficients

Claims

1. A safety management support system that assists in the control of on-site staff's work and the preservation of legal evidence in indoor facilities, A pre-mapping device that scans indoor facilities using a LiDAR sensor and inertial measurement device to acquire three-dimensional point cloud data, A spatial reference generation unit extracts persistent feature points from fixed structures such as walls, columns, and ceilings based on the aforementioned point cloud data, generates a facility-specific three-dimensional reference coordinate system, and stores it in a facility reference spatial database along with spatial feature anchors. A staff terminal that compares feature points extracted from camera footage with the facility reference spatial database to align its own position with the three-dimensional reference coordinate system, A central control device that takes the location information, congestion density information, and staff attribute information of the staff terminal as input, calculates a congestion score using a congestion evaluation function that includes at least three variables: a spatial density term, a flow velocity deviation term, and a dwell time term, generates augmented reality instructions based on the congestion score and the staff attribute information, and exclusively controls the display priority and display target of the augmented reality instructions according to the staff attribute information, The aforementioned staff terminals exchange predicted route or planned occupied area information via inter-terminal communication, and a distributed avoidance control unit autonomously determines the final collision avoidance option on the terminal side. The evidence storage unit stores the displayed augmented reality instruction information, display time, terminal identifier, staff location, and movement history in a format that allows for tamper detection. Equipped with, A safety management support system characterized in that the central control unit uses at least one of the following as an abnormal state detection trigger: detection of exceeding a congestion threshold, operator input, notification from an external sensor, and the fulfillment of a predetermined rule, to update and acquire the location of staff in the target area, perform route recalculation considering the congestion concentration area and the entry restriction area, synchronously distribute the updated augmented reality instructions to the staff terminal, and save the contents of the synchronous distribution in the evidence storage unit.

2. In the system described in claim 1, The aforementioned congestion evaluation function includes a spatial density term calculated from the ratio of the number of people per unit area to the area of ​​the target area, a flow velocity deviation term calculated from the deviation between the measured average movement speed and a preset target speed, and a dwell time term calculated from the cumulative time that a specific individual remains at the same coordinate. A system characterized by calculating a congestion score as a weighted sum obtained by multiplying each of the three terms mentioned above by a weight coefficient that can be changed according to the operating mode.

3. In the system described in claim 2, The aforementioned congestion level evaluation function is, C = α(N / A) + β(Vavg ― Vtarget) + γΔT A system characterized by calculating a congestion score C using the following formula: (N: number of people, A: area, Vavg: average movement speed, Vtarget: target speed, ΔT: dwell time, α, β, γ: weighting coefficients).

4. In the system according to claim 2 or 3, The central control unit preferentially increases the weighting coefficient of the spatial density term in normal operating conditions. A system characterized by transitioning to a stationary individual detection mode in which the weight coefficient of the dwell time term is preferentially increased and stationary individuals are highlighted on an augmented reality display, using at least one of the following as a stationary individual detection trigger: detection of exceeding a congestion threshold, operator input, notification from an external sensor, and fulfillment of a predetermined rule.

5. In the system described in claim 1, The staff terminal performs anchor fusion processing to correct its own position by integrating absolute coordinates based on distance measurement data acquired from a fixed beacon and relative coordinates based on matching processing of a three-dimensional point cloud using a state estimation filter. Furthermore, the system is characterized by pre-registering stationary structures within the facility as spatial feature anchors and performing drift correction processing to reset the accumulated error when the anchors are re-recognized.

6. In the system described in claim 5, The aforementioned point cloud matching process is the ICP (Iterative Closest Point) algorithm, A system characterized in that the state estimation filter is a Kalman filter.

7. In the system described in claim 1, The staff terminal exchanges predicted travel paths with other staff terminals using a P2P communication protocol. A system characterized in that, when spatial overlap between the predicted path of one terminal and the predicted path of another terminal is detected, the central control unit only provides supervisory information and performs cooperative path optimization processing to autonomously determine the final selection of an alternative path for collision avoidance through terminal local processing.

8. In the system described in claim 1, The central control unit, based on the role type pre-registered as attribute information of the staff terminal, This system is characterized by its role-based visual mutual exclusion control, which exclusively filters and displays information such as guidance routes to crowded areas for guidance staff, the shortest routes to rescue facilities for rescue staff, and stationary individual highlight information for monitoring staff.

9. In the safety management support system described in claim 1, A safety management support system characterized in that the evidence storage unit stores at least terminal identifiers, staff attribute information, abnormal state detection trigger types, and identification information of augmented reality instructions synchronously distributed in response to said triggers, in association with the augmented reality instruction information, display time, staff location, and movement history.

10. In the system described in claim 9, The system is characterized in that the tamper-detectable format includes at least one of the following: encryption, hash value addition, digital signature, and storage on a WORM (Write Once Read Many) recording medium.

11. A safety management support method that supports the control of on-site staff's work and the preservation of legal evidence in indoor facilities, which is performed by a safety management support system comprising a pre-mapping device, a spatial reference generation unit, a facility reference spatial database, staff terminals, a central control unit, and an evidence storage unit, (a) The pre-mapping device scans the indoor facility using a LiDAR sensor and an inertial measurement device and acquires three-dimensional point cloud data, (b) The spatial reference generation unit extracts persistent feature points of fixed structures based on the point cloud data to generate a facility-specific three-dimensional reference coordinate system and stores it in the facility reference spatial database along with spatial feature anchors. (c) The staff terminal performs the step of matching the feature points extracted from the camera image with the facility reference spatial database to align its own position with the reference coordinate system, (d) The central control unit takes the location information of the staff terminal, congestion density information, and staff attribute information as input and calculates a congestion score using a congestion evaluation function that includes at least three variables: a spatial density term, a flow velocity deviation term, and a dwell time term. (e) The central control unit generates an augmented reality instruction based on the congestion score and the staff attribute information, and exclusively controls the display priority and display target of the augmented reality instruction according to the staff attribute information, (f) The central control unit updates the location of staff within the target area, performs route recalculation considering the congestion concentration area and the entry restriction area, and synchronously distributes the updated augmented reality instructions to the staff terminal, using at least one of the following as an abnormal state detection trigger: congestion threshold exceedance detection, operator input, external sensor notification, and fulfillment of a predetermined rule. (g) A process in which multiple staff terminals exchange predicted route or planned occupied area information via inter-terminal communication, and each staff terminal autonomously decides on the final choice for collision avoidance, (h) The evidence storage unit sets and stores the displayed augmented reality instruction information, display time, terminal identifier, staff location and movement history, and the contents of the synchronized distribution in a format that allows for detection of tampering, A safety management support method characterized by including the following.

12. In the safety management support method described in claim 11, A safety management support method characterized in that the route re-search in step (f) includes distributed control in which each staff terminal refers to the supervisory information generated by the central control unit, the staff terminals exchange predicted route or planned occupied area information with each other, and each staff terminal autonomously decides on the final choice for collision avoidance.

Images