VR large-space collaborative management platform and dynamic path planning method thereof

Abstract

The invention provides a VR large-space collaborative management platform and a dynamic path planning method thereof, and belongs to the technical field of VR game management. The platform comprises a ticket checking end used for ticket business management, player account management and team setting; the field control terminal is used for automatically distributing VR equipment and managing game processes of players and teams; and the VR management background is used for managing VR equipment, setting game parameters, planning a playing path according to the game parameters and player playing data, carrying out player interaction identification, processing abnormities in the playing process and updating VR items. According to the dynamic path planning method, a playing path is dynamically planned through an improved A * algorithm, a social force model and a Q-Learning algorithm, and multi-dimensional optimization is carried out.

Description

Technical Field

[0001] The present invention relates to a VR large space collaborative management platform and a dynamic path planning method thereof, and belongs to the technical field of VR game management. Background Art

[0002] With the rise of VR technology, more and more VR game projects have appeared offline in recent years, and they are gradually developing in the direction of large space and multi-person interaction. However, most VR devices on the market and their matching platform functions are relatively simple, basically only have the basic functions of VR terminal equipment management, and do not have ticketing, project management, and venue control functions. It is difficult to meet the actual operation needs of VR game projects, which increases the operation costs of VR game projects. Moreover, most of the existing VR management platforms are limited to the operation plan of handle interaction in a small space, either only supporting a single-person in-situ experience, or only supporting a small number of players to play together. The utilization rate of large space is low. When there are many players, they need to queue up to enter, the venue operation efficiency is low, and the game experience is poor.

[0003] In addition, most VR game venues on the market have fixed play routes, low utilization of conventional space, and poor play flexibility, which cannot fully utilize the immersive experience of VR games. Therefore, VR game projects lack reasonable dynamic path planning methods, which affects the development and operation of VR play projects. Summary of the invention

[0004] The purpose of the present invention is to provide a VR large-space collaborative management platform and a dynamic path planning method thereof, by which VR project management, VR equipment management, ticketing management, venue control and other functions are simultaneously realized through the collaborative management platform, thereby improving the adaptability of the VR management platform. At the same time, the platform of the present invention supports multiple groups of players to play in the same venue space at the same time by dynamically planning the play path, thereby improving the venue operation efficiency and the player play experience.

[0005] In order to achieve the above purpose / solve the above technical problems, the present invention is implemented by adopting the following technical solutions:

[0006] On the one hand, the present invention provides a VR large space collaborative management platform, including:

[0007] The ticket checking terminal is used for ticket management, player account management and team setting;

[0008] On-site control terminal, used to automatically allocate VR equipment and manage the game progress of players and teams;

[0009] The VR management background is used to manage VR devices, set game parameters, plan the game path according to game parameters and player game data, perform interactive identification during the player's game, handle exceptions during the game, and update VR projects.

[0010] In combination with the first aspect, further, the VR management background includes a VR device management subsystem, a game setting subsystem, a play path planning subsystem, an interactive identification subsystem, an exception handling subsystem and a project update subsystem; the VR device management subsystem is used to enter the information of all VR devices in the venue and manage the VR device data; the game setting subsystem is used to import the venue map and VR game content, set the game boundaries, game effects and game rules; the play path planning subsystem is used to dynamically plan the play path according to the player's play data using the improved A* algorithm, social force model and Q-Learning algorithm; the interactive identification subsystem is used to use ArUco marking technology to identify the position and posture of the VR device in space, use 6DoF tracking technology to detect the change in the field of view angle caused by the movement of the player's head and body, and use gesture recognition technology to identify the changes in the player's hand bones to trigger preset gesture events; the project update subsystem is used to update VR game projects.

[0011] In combination with the first aspect, further, the project update subsystem updates the VR game project, including:

[0012] Obtain new VR game project packages through the project update subsystem and automatically review the data in the project packages;

[0013] If the review is passed, a version list will be generated through the project update subsystem, and the project package will be pushed to some VR devices in the venue;

[0014] In response to the project package installation status of the VR device, the project package is pushed to all VR devices in batches through the project update subsystem to complete the project hot update;

[0015] If the review fails, an error report is returned to the developer through the project update subsystem.

[0016] In a second aspect, the present invention provides a dynamic path planning method for a VR large space collaborative management platform, comprising the following steps:

[0017] Obtain VR large-space venue information and players' real-time play data through the VR large-space collaborative management platform;

[0018] Discretize the venue into a grid map based on the VR large-space venue information;

[0019] In the grid map, the initial play path is generated using the improved A* algorithm based on the player's real-time play data;

[0020] Execute the game path and periodically check for path conflicts;

[0021] When path conflict is detected, the play path is fine-tuned based on the social force model;

[0022] If path conflicts still exist after fine-tuning, use the improved A* algorithm to incrementally optimize the play path.

[0023] Combined with the second aspect, further, the heuristic function of the improved A* algorithm is:

[0024] (3)

[0025] Among them, h(n) represents the heuristic function of the improved A* algorithm, represents the heuristic function of the traditional A* algorithm, Represents the player density influence factor around node n, represents the load penalty item of the VR device associated with node n, is the population density adjustment factor, Penalize weights for device load;

[0026] The calculation formula is:

[0027] (4)

[0028] (5)

[0029] in, represents the player density of the grid where node n is located, is the density threshold, q is the smoothing coefficient, Indicates the number of players in the grid where node n is located, Represents the area of ​​a single grid in the grid map;

[0030] The calculation formula is:

[0031] (6)

[0032] Among them, Load(n) represents the computational load cost of the VR device associated with the node. Indicates the GPU utilization of node n.

[0033] In combination with the second aspect, further, the improved A* algorithm is trained and optimized using a cost function, and the expression of the cost function is:

[0034] C(n) = α·D(n) + β·Crowd(n) + γ·Load(n) (8)

[0035] Among them, C(n) represents the cost function of the nth node, α is the path distance weight, D(n) represents the geometric distance cost from the current node n to the target point, β is the crowd density weight, Crowd(n) represents the player density cost around node n, and γ is the device load weight;

[0036] The calculation formula of D(n) is as follows:

[0037] (9)

[0038] in, represents the Euclidean distance from the current node n to the node G, D max Indicates the maximum diagonal distance in a large VR space venue;

[0039] The calculation formula of Crowd(n) is as follows:

[0040] (10)

[0041] Among them, N is the total number of players in the current venue, Indicates the current node n to the i-th player The Euclidean distance of represents the real-time coordinates of the i-th player, is the density influence radius;

[0042] The calculation formula of Load(n) is as follows:

[0043] (11)

[0044] Where M represents the number of VR devices associated with node n, GPUUtil m Indicates the GPU utilization of the mth VR device, NetLatency m represents the network delay of the mth VR device, L max Indicates the maximum allowed delay.

[0045] Combined with the second aspect, further, when a new obstacle is detected within 5m around the current path node or the crowd density change exceeds the preset change threshold, it is considered that there is a path conflict, and the play path is fine-tuned based on the social force model. The adjusted play path is:

[0046] (12)

[0047] in,( , ) represents the adjusted path, represents the obstacle avoidance force in the x direction, represents the obstacle avoidance force in the y direction, is the adjustment factor;

[0048] The calculation formula of obstacle avoidance force vector is:

[0049] (13)

[0050] in, represents the obstacle avoidance force vector, A is the force intensity coefficient, represents the expected safe distance between player i and player j, represents the actual distance between player i and player j, B is the decay speed parameter, is the unit direction vector.

[0051] Combined with the second aspect, further, if path conflicts still exist after fine-tuning, the improved A* algorithm is used to incrementally optimize the play path, including:

[0052] 90% of the game path is retained, and the improved A* algorithm is used to re-plan the 10% game path related to the conflicting position.

[0053] In combination with the second aspect, further, during the path execution process, cubic spline interpolation is used to perform path smoothing.

[0054] In combination with the second aspect, further, after each play path is executed, the play path related data is saved, and the Q-Learning algorithm is used to optimize the data and learn the data features.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention proposes a VR large space collaborative management platform and a dynamic path planning method thereof, which can simultaneously realize VR project management, VR equipment management, ticket management, venue control and other functions, improve the adaptability of the VR management platform, and at the same time, manage the player's play process through automatic path planning, gesture recognition and other technologies to improve the player's play experience. The dynamic path planning method proposed in the present invention can realize play path planning and multi-dimensional optimization, while reducing the amount of path planning calculations and shortening the time-consuming path planning, it improves the rationality and intelligence of path planning, avoids the problem of path conflict when multiple groups of players play at the same time, effectively improves the utilization rate of large spaces, and improves the efficiency of venue operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 The figure shows a structural schematic diagram of a VR large space collaborative management platform provided by the present invention;

[0058] Figure 2 The figure shows a schematic diagram of the play path planning of the VR management background in an embodiment of the present invention;

[0059] Figure 3 The figure is a schematic diagram of the process of hot updating of the project update subsystem in an embodiment of the present invention;

[0060] Figure 4 The figure shows a schematic diagram of the global overview page of the VR management backend in an embodiment of the present invention;

[0061] Figure 5 The figure is a schematic diagram of the steps of a dynamic path planning method for a VR large space collaborative management platform provided by the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0063] Example 1

[0064] This embodiment introduces a VR large space collaborative management platform, such as Figure 1 As shown, it mainly includes the ticket checking terminal, on-site control terminal and VR management background.

[0065] The ticket checking terminal is mainly used to connect to the third-party ticketing system for ticket management. At the same time, the ticket checking terminal is also used for player account management and team settings.

[0066] After the ticket checking is completed, the ticket checking end automatically generates a player account corresponding to the ticketing information. The player can modify the information in the player account through the ticket checking end, customize the player ID, virtual image, etc. It supports customized face, clothing, and the use of photos to generate virtual images, etc., so that players have a unique virtual image in the game.

[0067] In the present invention, players can play alone or in a team, and establish a team composed of multiple player accounts through the ticket checking terminal, supporting customized team name, team color, team leader and other information. In the subsequent game process, both the on-site control terminal and the VR management background can view the player account information and team information.

[0068] The on-site control terminal is used by on-site staff to manage the game progress of players and teams. After receiving the player account information and team information from the ticket checking terminal, the on-site control terminal automatically allocates VR equipment to the player. After the player wears the VR equipment and is ready, the game officially starts through the on-site control terminal. When the player's VR equipment fails or other abnormalities occur, the on-site control terminal displays the specific faulty equipment and warning information, and the on-site staff assists the player to quickly replace the normal equipment or continue the game after handling the abnormality.

[0069] The VR management backend is used for global management and control, including but not limited to the VR device management subsystem, game setting subsystem, play path planning subsystem, interactive recognition subsystem, exception handling subsystem, project update subsystem, operation analysis subsystem and global overview page.

[0070] The VR device management subsystem is used to uniformly enter the information of all VR devices in the venue into the VR management background, manage VR device data, and support the rapid access of new devices.

[0071] The game settings subsystem is used to import venue maps and VR game content, and to set game-related parameters such as game boundaries, game effects, and game rules.

[0072] The play path planning subsystem is used to dynamically plan the play path according to the player's play data. The play path planning subsystem reads the data from the ticket checking terminal and the on-site control terminal in real time, combines the pre-set game parameters, automatically plans the play path for the player, and dynamically predicts and generates the next level area in real time to avoid overlapping of the play paths of different players in the routine, resulting in blockage or conflict, such as Figure 2 As shown. Through dynamic path management, the venue space utilization rate is improved, and the venue is operated with high floor space efficiency. In addition, after each group of players enters the venue, the next group of players can enter immediately. Through dynamic path planning, the distance between different players is adjusted, so that the experience content of each group of players is independent of each other.

[0073] In the embodiment of the present invention, the game path planning subsystem adopts the improved A* algorithm, social force model, Q-Learning algorithm and other technologies to perform multi-dimensional path planning and path optimization.

[0074] The exception handling subsystem is used to alarm and handle exceptions during the player's play, including VR device failure alarm, play path conflict alarm, player abnormality alarm, etc. When the player leaves the safe area, the exception handling subsystem generates a warning message and feeds it back to the on-site control end and VR device end.

[0075] The interactive recognition subsystem is used to monitor the player's gaming process in real time. It uses ArUco marking technology as a spatial anchor to determine the position and posture of the VR device in space, and uses 6DoF tracking technology to detect changes in field of view angle caused by the player's head rotation. It can also detect changes in up, down, front, back, left, and right displacements caused by body movement. It interacts with players through gesture recognition and somatosensory interaction, providing players with a better immersive and interactive experience.

[0076] In the embodiment of the present invention, the development environment is PICO 4 all-in-one development (Android), and PICO-related UE5-SDK is used. According to the Open XR standard, the hand bones are divided, and the Transform of each bone of the hands is obtained. When the hand gesture is close to a specific gesture, the gesture event is triggered. In the VR project, when zooming in or out of an object, the pinch gesture is recognized, and the distance between the two hands is judged after the two hands are pinched. If the distance becomes longer, it is defined as zooming in, and if the distance becomes shorter, it is defined as zooming out.

[0077] The project update subsystem supports hot updates of VR projects. When existing game projects in a VR game venue need to be updated or new game projects need to be connected to the system, hot updates are performed through the project update subsystem. Figure 3 As shown in the figure, the developer uploads the new project package to the project update subsystem, which automatically reviews the data in the project package, generates a version list if the review is passed, and pushes the pre-release package to some VR devices in the venue. In response to the project package installation status of the VR device, the project update subsystem pushes the full package in batches to complete the project hot update. If the project update subsystem fails to review the project package, an error report is returned to the developer.

[0078] The global overview page organizes and displays the player's play process data, providing a one-screen overview of the play status of each player and team in the venue, such as Figure 4 As shown, the global overview page includes the team board and the venue board. The team board can display dynamic data such as team progress, game market, equipment stations, abnormal alarms, etc. The venue board visualizes the venue map, team game path, player location, abnormal alarms and other dynamic data.

[0079] The operation analysis subsystem is used to analyze ticket checking data, player play data, etc. to obtain the operation analysis report of the current venue. The operation analysis report includes multi-dimensional revenue statistics, receivables trend analysis, traffic density analysis, traffic structure analysis, etc.

[0080] The platform architecture of the present invention adopts an independent team server. Each player's VR device is regarded as a terminal. The terminals of a team are uniformly connected to an independent team server. Multiple independent team servers are connected to a server cluster. The server cluster interacts with the ticket checking terminal, the on-site control terminal and the VR management background for data.

[0081] The collaborative management platform of the present invention adopts Lic authorization, supports online and offline control, and supports customized function permissions and authorization validity periods.

[0082] Example 2

[0083] Based on the management platform introduced in Example 1, this embodiment introduces a dynamic path planning method for a VR large space collaborative management platform, such as Figure 5 As shown, the following steps are included:

[0084] Step A: Obtain VR large space venue information and players' real-time play data through the VR large space collaborative management platform. Players' real-time play data includes players' real-time coordinates, players' game progress, etc.

[0085] Step B: discretize the venue into a grid map based on the VR large-space venue information.

[0086] The formula for grid discretization is as follows:

[0087] (1)

[0088] Formula (1) can convert continuous physical space into a discrete grid map, which helps to reduce computational complexity while retaining sufficient accuracy.

[0089] The grid resolution can be changed according to project requirements. In the embodiment of the present invention, the grid resolution is 0.5m.

[0090] In the grid map, the levels or key coordinate points on the game path are used as nodes on the grid map. The node is defined as n=(x,y,t)∈Z3, where x and y represent the coordinates of node n in the discretized grid map (unit: meter), and t is the timestamp (unit: second), which is used to model the time dimension of dynamic scenes.

[0091] Step C: Generate an initial play path using the improved A* algorithm based on the player's real-time play data.

[0092] The estimated minimum cost from the current node n to the target node G in the traditional A* algorithm, The calculation formula is as follows:

[0093] (2)

[0094] Among them, wd Indicates the distance weight coefficient, the value range is 0.8-1.2, and the default value is generally 1.0; Represents the Euclidean distance from the current node n to the target node G; is the horizontal coordinate of node n, is the ordinate of node n, is the horizontal coordinate of the target point G, is the ordinate of the target point G.

[0095] When there are fewer obstacles in the site, the traditional A* algorithm uses Euclidean distance calculation, which has a small amount of calculation. When the site terrain is complex, the traditional A* algorithm uses Manhattan distance calculation, which has a higher path feasibility.

[0096] The present invention improves the traditional A* algorithm to obtain a new incremental PE-A* algorithm, and the calculation formula of its heuristic function is:

[0097] (3)

[0098] Among them, h(n) represents the heuristic function of the improved A* algorithm, Represents the heuristic function of the traditional A* algorithm, also known as the basic heuristic function, Represents the player density influence factor around node n, represents the load penalty item of the VR device associated with node n, is the population density adjustment factor, is the device load penalty weight. The area around node n is defined as the grid area where node n is located or the area with node n as the center and a preset distance as the radius. In the embodiment of the present invention, the grid area where node n is located is used.

[0099] The normalized influence coefficient reflecting the density of people around node n is calculated as follows:

[0100] (4)

[0101] (5)

[0102] in, Represents the player density of the grid where node n is located; is the density threshold, i.e. the critical value that triggers avoidance; q is the smoothing coefficient, which is used to control the steepness of the function; Indicates the number of players in the grid where node n is located, Represents the area of ​​a single grid cell in a grid map.

[0103] In general =1.2 people / ㎡When sudden gathering is detected, the number of people will be automatically lowered To 0.8.

[0104] The calculation formula is:

[0105] (6)

[0106] Among them, Load(n) represents the computational load (such as GPU utilization, network latency) cost of the node-associated VR device. Indicates the GPU utilization of node n.

[0107] When the device load exceeds 70%, the cost is increased nonlinearly to force balanced resource allocation.

[0108] The parameter σ in the improved A* algorithm of the present invention is adjusted in an hourly cycle, and the crowd influence weight is increased during the daytime peak hours to improve the time sensitivity of the model, thereby planning a more reasonable play path. At the same time, when the node GPU utilization rate is >70%, the load penalty term is activated to better balance node resources when planning the path.

[0109] In the same test scenario, the comparison of the path planning effects of the traditional A* algorithm and the improved A* algorithm of the present invention is shown in Table 1:

[0110] Table 1 Comparison of the effects of traditional A* algorithm and improved A* algorithm

[0111] ;

[0112] The initial play path planned by the improved A* algorithm is a discrete grid path. In order to improve the naturalness of the VR mobile experience, the present invention smoothes the discrete grid path through a Bezier curve, thereby converting discrete path points into a continuous smooth curve. The formula of the Bezier curve is:

[0113] (7)

[0114] Among them, B(t) represents a smooth playing path, represents the kth intermediate quantity, The value range of is [0,1], Represents the coordinates of the kth node in the initial game path planned by the improved A* algorithm.

[0115] The present invention expands the path planning problem from single distance optimization to multi-objective collaborative optimization, uses the three key factors of balanced path, crowd density, and equipment load as optimization targets, and constructs the cost function of the dynamic path planning model. The formula is as follows:

[0116] C(n) = α·D(n) + β·Crowd(n) + γ·Load(n) (8)

[0117] Among them, C(n) represents the cost function of the nth node, α is the path distance weight, D(n) represents the geometric distance cost from the current node n to the next node, β is the crowd density weight, Crowd(n) represents the player density cost around the node calculated based on the real-time heat map, γ is the device load weight, and Load(n) represents the computing load (such as GPU utilization, network latency) cost of the VR device associated with the node.

[0118] The present invention trains and optimizes the improved A* algorithm through formula (8), comprehensively considering the path length, path congestion level and device resource configuration, so as to minimize the path length of the play path predicted by the model, avoid path congestion and balance resource allocation.

[0119] The calculation formula of D(n) is as follows:

[0120] (9)

[0121] Among them, G represents the target point, represents the Euclidean distance from the current node n to the node G, D max Indicates the maximum diagonal distance (normalization factor) in a large VR space venue.

[0122] The calculation formula of Crowd(n) is as follows:

[0123] (10)

[0124] Among them, N is the total number of players in the current venue, Indicates the current node n to the i-th player The Euclidean distance of represents the real-time coordinates of the i-th player, is the density influence radius. In the embodiment of the present invention, =3 meters.

[0125] The calculation formula of Load(n) is as follows:

[0126] (11)

[0127] Where M represents the number of VR devices associated with node n, GPUUtil m Indicates the GPU utilization of the mth device, NetLatency m represents the network delay of the mth device, L max Indicates the maximum allowed delay (normalization factor).

[0128] In the present invention, the weight coefficient α+β+γ=1 (α, β, γ∈[0,1]), and the weight system configuration is different in different usage scenarios. High-density offline entertainment scenario: α+β+γ->0.4+0.5+0.1 low-density industrial simulation scenario: α+β+γ->0.5+0.3+0.2.

[0129] Step D: Execute the play path. When a path conflict is detected, the play path is dynamically adjusted in real time based on the social force model.

[0130] When a new obstacle or a sudden change in crowd density (>15% change) is detected within 5m around the current path node, the present invention uses the social force model to calculate the obstacle avoidance force, and then adjusts the local path according to the obstacle avoidance force. Its core function is to guide the path planning algorithm to actively avoid potential conflicts and avoid collisions between different players' play paths through mathematical modeling of crowd interaction behavior.

[0131] The social force model dynamically adjusts the play path in real time. The adjusted path is:

[0132] (12)

[0133] in,( , ) represents the adjusted path, represents the obstacle avoidance force in the x direction, represents the obstacle avoidance force in the y direction, is the adjustment factor, =0.1.

[0134] The calculation formula of obstacle avoidance force of social force model is as follows:

[0135] (13)

[0136] in, represents the obstacle avoidance force vector, which is the avoidance force generated by the individual due to the surrounding pedestrians or obstacles, and its direction is away from the threat; A is the force intensity coefficient, and the larger the value, the more sensitive it is to close-range threats; represents the expected safe distance between player i and player j; Indicates the actual distance between player i and player j; B is the attenuation speed parameter, the smaller the value, the more rapid the force increases at close distance; is the unit direction vector.

[0137] When calculating , in formula (11) , The distance in the x direction is calculated. When formula (11) , The distance in the y direction is calculated.

[0138] According to formula (7), when < When , the exponential term exp(⋅) produces a nonlinearly growing avoidance force, which can simulate human emergency avoidance behavior.

[0139] Through experimental calibration, when the parameters A=2.0 and B=0.3, the sensitivity and stability can be well balanced.

[0140] Step D fine-tunes the path points in the direction of the obstacle avoidance force with small steps to achieve smooth avoidance and avoid VR motion sickness caused by path mutations. Compared with using the improved A* algorithm to modify the play path after discovering an abnormality, using the social force model to fine-tune the path can greatly reduce the amount of calculation. According to experiments, only 3%-5% of the calculation amount of the improved A* algorithm is required.

[0141] Step E: Determine whether there is still a conflict in the play path. If there is still a conflict in the play path, use the improved A* algorithm to perform incremental optimization on the play path.

[0142] 90% of the gameplay path is retained, and the remaining 10% of the gameplay path related to conflicts is replanned to solve the path conflict problem while reducing the amount of calculation.

[0143] In an embodiment of the present invention, the crowd density around the play path (within a preset distance) can be calculated through a heat map, and combined with the VR device load to determine whether the player will conflict with obstacles or other players in the venue during the play process, so as to determine whether to adjust the path.

[0144] In the embodiment of the present invention, when the cumulative number of fine-tuning (step D) exceeds 3 times and the conflict is still not eliminated, incremental optimization can be performed through step E.

[0145] Step F: During the path execution process, the present invention uses cubic spline interpolation to smooth the path and generate a smooth moving trajectory to ensure that the VR device moves continuously and avoids sudden changes in acceleration.

[0146] In addition, the present invention can also issue graded warnings for situations where players deviate from the play path. According to the player's real-time coordinates and the planned play path, the distance the player deviates from the play path is calculated. When the preset deviation threshold is exceeded, a warning is issued on the VR device side, the on-site control side, and the VR management background. There can be multiple preset deviation thresholds, and different warning contents are set according to different deviation thresholds. For example, if the deviation exceeds 0.5m, the player is warned to deviate from the path, and if the deviation exceeds 1m, the player is warned to pause the game, etc.

[0147] Step G: After each play path is executed, save the play path related data, use the Q-Learning algorithm to optimize the data, fully learn the data features, and further optimize the path planning effect.

[0148] The update rule of the Q-Learning algorithm is:

[0149] (14)

[0150] in, It represents the Q value corresponding to the state s and action a. s represents the state, that is, the information set of the current environment of the system, and a represents the action, that is, the operation that the agent can perform in the current state during the deep learning process. is the learning factor, R is the reward function, which is used to quantify the feedback signal of the quality of the action. is the discount factor, Indicates the Q value corresponding to the new state s' and the new action a'.

[0151] In this embodiment of the present invention, the reward function is used to guide the algorithm to balance conflict avoidance (+10), load balancing (+5) and path efficiency (-0.3×length).

[0152] Based on the priority, experience samples with high prediction errors are replayed first to accelerate the learning of key scenarios (such as sudden aggregation and equipment failure). The expression of priority is:

[0153] (15)

[0154] In summary, the collaborative management platform of the present invention adapts the same game content to different areas, shapes, and obstacles without additional development; provides the ability to play on a come-as-you-go basis to maximize operational efficiency, so that users do not have to wait for sessions and equipment is not idle during peak hours; realizes dynamic planning of players' play paths to ensure that different content can be operated with high floor space efficiency in the current venue; promptly issues alarms for abnormal situations during play, which is conducive to improving operational stability; and improves the interactive gaming experience through gesture recognition, somatosensory interaction and other designs.

[0155] The dynamic path planning method proposed in the present invention can achieve multi-dimensional optimization, reduce the amount of path planning calculations, shorten the time spent on path planning, while improving the rationality and intelligence of path planning, avoiding path conflicts when multiple groups of players play at the same time, effectively improving the utilization rate of large spaces, and improving the efficiency of venue operations.

[0156] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A VR large space collaborative management platform, characterized in that: include: The ticket checking terminal is used for ticket management, player account management and team setting; On-site control terminal, used to automatically allocate VR equipment and manage the game progress of players and teams; The VR management background is used to manage VR devices, set game parameters, plan the game path according to game parameters and player game data, perform interactive identification during the player's game, handle exceptions during the game, and update VR projects.

2. The VR large space collaborative management platform according to claim 1, characterized in that: The VR management backend includes a VR device management subsystem, a game setting subsystem, a play path planning subsystem, an interactive identification subsystem, an exception handling subsystem, and a project update subsystem; the VR device management subsystem is used to input information of all VR devices in the venue and manage VR device data; the game setting subsystem is used to import venue maps and VR game content, set game boundaries, game effects, and game rules; the play path planning subsystem is used to dynamically plan the play path based on the player's play data using the improved A* algorithm, social force model, and Q-Learning algorithm; The interactive recognition subsystem is used to use ArUco marking technology to identify the position and posture of the VR device in space, use 6DoF tracking technology to detect the change in the field of view angle caused by the movement of the player's head and body, and use gesture recognition technology to identify the change in the player's hand bones to trigger a preset gesture event; The project update subsystem is used to update VR game projects.

3. The VR large space collaborative management platform according to claim 2 is characterized in that: The project update subsystem updates the VR game project, including: Obtain new VR game project packages through the project update subsystem and automatically review the data in the project packages; If the review is passed, a version list will be generated through the project update subsystem, and the project package will be pushed to some VR devices in the venue; In response to the project package installation status of the VR device, the project package is pushed to all VR devices in batches through the project update subsystem to complete the project hot update; If the review fails, an error report is returned to the developer through the project update subsystem.

4. A dynamic path planning method based on the VR large space collaborative management platform according to claim 2, characterized in that: The steps include: Obtain VR large-space venue information and players' real-time play data through the VR large-space collaborative management platform; Discretize the venue into a grid map based on the VR large-space venue information; In the grid map, the initial play path is generated using the improved A* algorithm based on the player's real-time play data; Execute the game path and periodically check for path conflicts; When path conflict is detected, the play path is fine-tuned based on the social force model; If path conflicts still exist after fine-tuning, use the improved A* algorithm to incrementally optimize the play path.

5. The dynamic path planning method of the VR large space collaborative management platform according to claim 4 is characterized in that: The heuristic function of the improved A* algorithm is: ； Among them, h(n) represents the heuristic function of the improved A* algorithm, represents the heuristic function of the traditional A* algorithm, Represents the player density influence factor around node n, represents the load penalty item of the VR device associated with node n, is the population density adjustment factor, Penalize weights for device load; The calculation formula is: ； ； in, represents the player density of the grid where node n is located, is the density threshold, q is the smoothing coefficient, Indicates the number of players in the grid where node n is located, Represents the area of ​​a single grid in the grid map; The calculation formula is: ； Among them, Load(n) represents the computational load cost of the VR device associated with the node. Indicates the GPU utilization of node n.

6. The dynamic path planning method of the VR large space collaborative management platform according to claim 4 is characterized in that: The improved A* algorithm is trained and optimized using a cost function, and the expression of the cost function is: C(n) = α·D(n) + β·Crowd(n) + γ·Load(n); Among them, C(n) represents the cost function of the nth node, α is the path distance weight, D(n) represents the geometric distance cost from the current node n to the target point, β is the crowd density weight, Crowd(n) represents the player density cost around node n, and γ is the device load weight; The calculation formula of D(n) is as follows: ； in, represents the Euclidean distance from the current node n to the node G, D max Indicates the maximum diagonal distance in a large VR space venue; The calculation formula of Crowd(n) is as follows: ； Among them, N is the total number of players in the current venue, Indicates the current node n to the i-th player The Euclidean distance of represents the real-time coordinates of the i-th player, is the density influence radius; The calculation formula of Load(n) is as follows: ； Where M represents the number of VR devices associated with node n, GPUUtil m Indicates the GPU utilization of the mth VR device, NetLatency m represents the network delay of the mth VR device, L max Indicates the maximum allowed delay.

7. The dynamic path planning method of the VR large space collaborative management platform according to claim 4 is characterized in that: When a new obstacle is detected within 5m around the current path node or the crowd density changes beyond the preset change threshold, it is considered that there is a path conflict. The play path is fine-tuned based on the social force model. The adjusted play path is: ； in,( , ) represents the adjusted path, represents the obstacle avoidance force in the x direction, represents the obstacle avoidance force in the y direction, is the adjustment factor; The calculation formula of obstacle avoidance force vector is: ； in, represents the obstacle avoidance force vector, A is the force intensity coefficient, represents the expected safe distance between player i and player j, represents the actual distance between player i and player j, B is the decay speed parameter, is the unit direction vector.

8. The dynamic path planning method of the VR large space collaborative management platform according to claim 4 is characterized in that: If path conflicts still exist after fine-tuning, use the improved A* algorithm to incrementally optimize the game path, including: 90% of the game path is retained, and the improved A* algorithm is used to re-plan the 10% game path related to the conflicting position.

9. The dynamic path planning method of the VR large space collaborative management platform according to claim 4 is characterized in that: During the path execution process, cubic spline interpolation is used for path smoothing.

10. The dynamic path planning method of the VR large space collaborative management platform according to claim 4 is characterized in that: After each game path is executed, the game path related data is saved and the Q-Learning algorithm is used to optimize the data.

Images