VR olfaction orientation method and system and VR equipment

Abstract

The invention relates to the technical field of virtual reality, and discloses a VR olfaction orientation method and system and a VR device.The method comprises the steps that firstly, the user pose, the target direction and environment medium flow data are collected, a particle filtering algorithm is adopted for odor diffusion simulation, and an odor distribution diagram of a virtual environment is generated; secondly, the concentration space gradient of the current position of the user is calculated, and whether the modulus length is lower than a threshold value or not is judged; if the particle filter parameter is lower than the threshold value, updating the particle filter parameter to strengthen the anisotropy of diffusion, and generating a corrected distribution diagram; secondly, with minimization of direction ambiguity as a target, space coordinates of smell release points are optimized through a gradient descent algorithm, and an optimal release point set is obtained; and finally, synthesizing a direction anchoring signal based on the corrected distribution diagram and the release point set, encoding the signal into a timing and strength-fixed instruction, and controlling the multi-channel smell release device to realize differentiated smell output, thereby providing stable and accurate smell direction guidance for a user.

Description

Technical Field

[0001] This invention relates to the field of virtual reality technology, and in particular to a VR olfactory orientation method, system, and VR device. Background Technology

[0002] The goal of virtual reality (VR) technology is to create a highly immersive perceptual environment for users through the synergy of multiple sensory channels. Currently, the interaction technologies of the visual and auditory channels are becoming increasingly mature, providing stable and accurate spatial information. However, the research and development of smell—an indispensable part of the human perceptual environment—in VR is still in a relatively early stage, especially in the area of ​​using smell for spatial guidance, where significant technical bottlenecks exist.

[0003] The core challenge of using the olfactory channel for spatial guidance stems from its inherent physical characteristics. Unlike the propagation of light and sound, the diffusion of odors in space is not uniform or isotropic, but highly dependent on the flow of environmental media (e.g., air velocity and direction), temperature gradients, and the layout of physical obstacles in the virtual environment. This anisotropic diffusion characteristic means that odors emitted from a single release point cannot form a stable and reliable directional vector field in three-dimensional space. The propagation path of odors is unpredictable and dynamically changing, making it difficult to provide a clear and constant directional signal like a visual arrow or stereo sound.

[0004] Existing VR olfactory cues mostly borrow the concept of left / right channels from audio processing, using simple on / off switching of odor emitters located on the user's left or right side to indicate direction. However, this simplified control mode completely ignores the complex diffusion dynamics described above. The direct consequence is that the concentration difference signal actually formed in the user's nasal cavity is extremely unstable, fluctuating in strength, and may even produce completely misleading results due to transient interference from environmental media (such as a gust of wind in the virtual environment), thus preventing the user from obtaining a clear, reliable, and robust sense of direction. Summary of the Invention

[0005] Therefore, the purpose of this invention is to overcome the problem of ambiguous and unreliable olfactory direction signals caused by the anisotropy of odor diffusion in the prior art, and to provide a VR olfactory orientation method, system and VR device. Instead of treating the anisotropic diffusion of odor as a technical barrier to be avoided, it treats it as an intrinsic variable and obtains a stable and reliable direction signal through virtual simulation, dynamic optimization and precise synthesis. This fundamentally solves the problem of unstable olfactory guidance and significantly enhances the user's navigation efficiency and immersive experience in the virtual environment.

[0006] Firstly, to solve the above-mentioned technical problems, the present invention provides a VR olfactory orientation method, comprising: Acquire data, including real-time pose data of the user in the virtual reality environment, target guidance direction, and real-time medium flow data of the physical environment; use a particle filtering algorithm to simulate odor diffusion based on the acquired data to generate an odor distribution map of the virtual environment; Based on the odor distribution map, calculate the spatial gradient of the concentration at the user's current location, and determine whether the mean magnitude of the spatial gradient of the concentration is lower than the magnitude threshold. In response to the mean modulus of the concentration spatial gradient being lower than the modulus threshold, the parameters of the particle filter algorithm are updated to enhance the anisotropic features and an updated odor distribution map is generated. Using minimizing the directional ambiguity at the user's location as the objective function, a gradient descent algorithm is employed to optimize the spatial coordinates of the odor release points, thereby obtaining a set of release points. Based on the updated odor distribution map and the set of release points, a directional anchoring signal is synthesized; wherein, the directional anchoring signal is encoded as a timing and intensity command for controlling at least two odor release channels to perform differentiated outputs.

[0007] Preferably, a particle filtering algorithm is used to simulate odor diffusion and generate an odor distribution map of the virtual environment, including: particle initialization, generating an initial particle set containing N particles from a preset odor release point; the initial state of each particle includes three-dimensional position coordinates, a three-dimensional velocity vector, age, and weight; at each time step, updating the state of each particle according to the particle state transition equation to obtain a predicted particle set; for each particle in the predicted particle set, calculating its Euclidean distance to the observation point, and updating the weight of each particle through the Euclidean distance to obtain a weighted particle set; calculating the effective number of particles in the weighted particle set and determining whether the effective number of particles is lower than a threshold number of particles; in response to the effective number of particles being lower than the threshold number of particles, performing resampling to obtain an effective particle set; and generating an odor distribution map of the virtual environment based on the effective particle set.

[0008] Preferably, generating an odor distribution map of a virtual environment based on the effective particle set includes: discretizing the virtual environment space into a three-dimensional grid; traversing the effective particle set and mapping each particle to a voxel of the three-dimensional grid according to its position; accumulating the weights of all particles within each voxel to obtain the concentration value of that voxel; normalizing the concentration values ​​of all voxels to generate the odor distribution map at the current moment.

[0009] Preferably, calculating the concentration spatial gradient of the user's current location based on the odor distribution map includes: mapping the user's current location to the three-dimensional grid to obtain a center location; determining a local spatial neighborhood based on the center location; extracting the concentration values ​​of each voxel within the local spatial neighborhood from the odor distribution map; and calculating the concentration spatial gradient of the center location based on the central difference method according to the concentration values ​​of each voxel within the local spatial neighborhood.

[0010] Preferably, before particle initialization, the method further includes: determining an initial particle release strategy based on the target guidance direction, wherein the initial release strategy includes: dynamically allocating the particle generation ratio of the two odor release points on both sides according to the target guidance direction relative to the user's current orientation, and / or adjusting the spatial position of the release points.

[0011] Preferably, the particle state transition equation is set as follows: ; Δt represents the simulation time step; i represents the particle number; Represents the three-dimensional spatial coordinates of particle i at time t+1; Represents the three-dimensional position coordinates of particle i at time t; Let represent the three-dimensional velocity vector of particle i at time t; Let represent the three-dimensional velocity vector of particle i at time t+1; D represents the unit vector of the flow direction of the medium in the virtual reality environment. The velocity represents the flow rate of the medium in the virtual reality environment; k represents the wind speed influence coefficient, used to adjust the intensity of the wind's influence on particle motion, with a value ranging from 0.1 to 1.5S. -1 ; This represents the survival time of particle i at time t+1; This represents the survival time of particle i at time t.

[0012] Preferably, in response to the mean modulus of the concentration spatial gradient being lower than the modulus threshold, the parameters of the particle filter algorithm are updated to enhance anisotropic characteristics, including: increasing the wind speed influence coefficient and / or the flow direction of the covering medium; the flow direction of the covering medium includes: setting the flow direction of the medium in the state transition equation representing the virtual reality environment as the target guidance direction; or using a weighted fusion of the target guidance direction and the original flow direction of the medium as the updated flow direction.

[0013] Preferably, the spatial coordinates of the odor release point are optimized using a gradient descent algorithm, with the objective function being minimizing the directional ambiguity at the user's location. This includes: constructing the objective function: Where m represents the odor release point number; P mThe three-dimensional coordinates of the odor release point m; This represents the spatial magnitude of the concentration gradient at the odor release point m; The angle between the spatial gradient vector of the odor release point m and the target guidance direction vector is represented. Using the three-dimensional coordinates of the odor release point as optimization variables, the three-dimensional coordinates of the odor release point are iteratively adjusted by calculating the gradient of the objective function with respect to the optimization variables until the objective function converges to the minimum value or reaches the maximum number of iterations.

[0014] Preferably, the direction anchoring signal is synthesized based on the updated odor distribution map and the set of release points, including: calculating the consistency coefficient between the medium flow direction and the target guidance direction based on the updated odor distribution map and the set of release points; and adjusting the mode of the differentiated output based on the consistency coefficient: when the consistency coefficient is greater than or equal to a threshold coefficient, an intensity modulation mode that is coordinated with the medium flow direction is adopted; when the consistency coefficient is less than the threshold coefficient, a frequency or pulse width modulation mode that is opposed to the medium flow direction is adopted.

[0015] Secondly, to solve the above-mentioned technical problems, the present invention provides a VR olfactory orientation system, comprising: The data acquisition module is used to acquire and output acquired data, including real-time pose data of the user in the virtual reality environment, target guidance direction, and real-time medium flow data of the physical environment. The odor diffusion simulation module is used to simulate odor diffusion based on the collected data using a particle filtering algorithm, and generate an odor distribution map of a virtual environment. The analysis module is used to calculate the concentration spatial gradient of the user's current location based on the odor distribution map, and determine whether the magnitude of the concentration spatial gradient is lower than a preset magnitude threshold; in response to the magnitude of the concentration spatial gradient being lower than the preset magnitude threshold, the parameters of the particle filtering algorithm in the odor diffusion simulation module are updated to enhance the anisotropic characteristics, and the odor diffusion simulation module is triggered to generate an updated odor distribution map. The release point optimization module is used to optimize the spatial coordinates of the odor release points by using a gradient descent algorithm with the objective function of minimizing the directional ambiguity at the user's location, thereby obtaining a set of release points. The output module is used to synthesize a directional anchoring signal based on the updated odor distribution map and the set of release points; wherein the directional anchoring signal is encoded as a timing and intensity command for controlling at least two odor release channels to perform differentiated outputs.

[0016] Thirdly, to solve the above-mentioned technical problems, the present invention provides a VR device including at least two odor release channels, wherein the VR device integrates the VR olfactory orientation system and executes the VR olfactory orientation method.

[0017] Compared with the prior art, the above-described technical solution of the present invention has the following advantages: The VR olfactory orientation method, system, and VR device described in this invention no longer regard the anisotropic diffusion of odors as a technical barrier that needs to be avoided. Instead, it treats it as an intrinsic variable and obtains a stable and reliable directional signal through virtual simulation, dynamic optimization, and precise synthesis, thereby fundamentally solving the problem of unstable olfactory guidance.

[0018] Specifically, a stable and reliable direction vector field is generated by simulating dynamic odor diffusion, which solves the problem of signal instability; the system's robustness is significantly improved by actively compensating for environmental interference through gradient magnitude judgment and parameter adaptive update mechanism; the navigation accuracy is greatly improved by optimizing the spatial coordinates of the release point to achieve precise spatial anchoring of the guidance signal; and finally, by synthesizing multi-dimensional direction anchoring signals, the sense of smell is upgraded from a simple ambient cues to a precise interactive channel that can coordinate with vision and hearing, which significantly enhances the user's navigation efficiency and immersive experience in the virtual environment. Attached Figure Description

[0019] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart of the VR olfactory orientation method in a preferred embodiment of the present invention; Figure 2 This is a flowchart illustrating the generation of an odor distribution map of a virtual environment in a preferred embodiment of the present invention; Figure 3 This is a structural block diagram of the VR olfactory orientation system in a preferred embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of the VR device in a preferred embodiment of the present invention; Explanation of the markings on the attached diagrams in the instruction manual: 10 - Odor release channel one; 20 - Odor release channel two. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0021] The purpose of this invention is to overcome the problem of ambiguous and unreliable olfactory direction signals caused by the anisotropic diffusion of odors in the prior art, and to provide a VR olfactory orientation method, system and VR device. Instead of treating the anisotropic diffusion of odors as a technical barrier to be avoided, it treats it as an intrinsic variable and obtains a stable and reliable direction signal through virtual simulation, dynamic optimization and precise synthesis. This fundamentally solves the problem of unstable olfactory guidance and significantly enhances the user's navigation efficiency and immersive experience in the virtual environment.

[0022] Example 1: Refer to Figure 1 As shown, this embodiment of the invention discloses a VR olfactory orientation method, including: S100. Acquire the collected data, which includes the user's real-time pose data in the virtual reality environment, the target guidance direction, and the real-time medium flow data of the physical environment; use a particle filtering algorithm to simulate odor diffusion based on the collected data to generate an odor distribution map of the virtual environment; S200. Based on the odor distribution map, calculate the spatial gradient of the concentration at the user's current location, and determine whether the average magnitude of the spatial gradient is lower than the magnitude threshold. S300. In response to the mean modulus of the concentration spatial gradient being lower than the modulus threshold, the parameters of the particle filter algorithm are updated to enhance the anisotropic characteristics, and an updated odor distribution map is generated. S400. Using the objective function of minimizing the directional ambiguity at the user's location, the gradient descent algorithm is used to optimize the spatial coordinates of the odor release points to obtain a set of release points. S500. Based on the updated odor distribution map and the set of release points, a directional anchoring signal is synthesized; wherein, the directional anchoring signal is encoded as a timing and intensity command for controlling at least two odor release channels to perform differentiated outputs.

[0023] In specific application scenarios, real-time pose data is obtained through the VR headset's inside-out tracking system, which acquires the user's head's three-dimensional coordinates (x, y, z) and orientation angles (θ, φ, ψ) in virtual space. The target guidance direction is provided by the VR application logic; for example, the system calculates the target direction vector based on the coordinates of the target point to which the user needs to be guided. Real-time media flow data is obtained through a miniature wind speed and direction sensor integrated into the VR headset, or from the virtual environment physics engine, to acquire the current airflow speed. And the flow direction is the unit vector D.

[0024] A particle filtering algorithm is employed, running a simplified fluid simulation in the background. N virtual particles are generated from the current release point, each possessing position, velocity, age, and weight attributes. At each time step, the particle velocity is updated, thereby updating its position. Weights are calculated based on the distance between the particles and the user's position. Periodic resampling is performed, discarding low-weight particles and replicating high-weight particles. The particle weights are mapped onto a 3D mesh, and after normalization, an odor distribution map is generated, reflecting the spatiotemporal probability distribution of odor concentration. Through simulation, the actual propagation path of odors in complex environments can be predicted, making the guidance strategy more consistent with real physical laws.

[0025] In the obtained odor distribution map, locate the voxel where the user's current location is located, and use the central difference method to calculate the partial derivatives of the concentration in the X, Y, and Z directions (∂C / ∂x, ∂C / ∂y, ∂C / ∂z) at that location, and synthesize the spatial gradient vector of the concentration. Its direction points in the direction of the fastest increase in concentration, modulus Indicates the degree of drastic change. The gradient magnitude... It is compared with a preset magnitude threshold; if the gradient magnitude is... If the gradient magnitude is less than the modulus threshold, the current odor field is deemed to have insufficient directionality, requiring intervention. By using the objective and calculable metric of gradient modulus, the subjective feeling of unclear direction is transformed into a numerical value that can be used for logical judgment, enabling the system to understand the quality of its current guidance effect.

[0026] When intervention is deemed necessary, the system dynamically updates the parameters of the particle filter algorithm in S100. Using the updated parameters, it continues to run the particle filter simulation for several iterations, and based on the new example distribution, regenerates an updated odor distribution map. This achieves adaptive compensation for environmental disturbances; that is, when natural diffusion cannot provide effective guidance, the system actively intervenes in the virtual wind field, forcibly amplifying the directional differences in odor concentration to overcome the effects of adverse environments.

[0027] An objective function is constructed to evaluate the directional clarity at the user's location when the odor is released at the release point. Starting from a hypothetical release point (e.g., the left side of the headset), steps S100-S200 are simulated and calculated using the current release point coordinates to obtain the objective function value. By perturbing the release point coordinates, the gradient of the objective function relative to the coordinates is numerically estimated. The release point position is updated along the gradient descent direction, and this process is repeated until convergence, yielding optimized release point coordinates, which are then added to the release point set. By using minimizing the directional ambiguity at the user's location as the objective function, precise spatial anchoring of the guidance source is achieved, i.e., finding the optimal release position that generates the clearest directional signal in the current environment. By optimizing the release point, the contrast of the directional signal formed at the user's location is maximized, making it easier for the user to determine the correct direction and reducing the possibility of getting lost.

[0028] Finally, by comprehensively analyzing the updated odor distribution map and release point set, the appropriate signal mode is determined, including cooperative enhancement mode and counter-compensation mode. If the flow direction of the ambient medium is roughly consistent with the target direction, intensity modulation is used to ensure continuous high-intensity release from the channel on the target direction side. If the environment is unfavorable, timing / frequency modulation is used to ensure the channel on the target direction side operates with a higher frequency and longer pulse train. The strategy is encoded into a specific set of hardware control instructions, including channel (left / right), intensity (PWM duty cycle), mode (continuous / pulse), frequency, and duration. This structured direction anchoring signal is then sent to the main controller.

[0029] The VR olfactory orientation method described in this invention no longer regards the anisotropic diffusion of odors as a technical barrier that needs to be avoided, but treats it as an intrinsic variable. Through virtual simulation, dynamic optimization, and precise synthesis, a stable and reliable directional signal is obtained, fundamentally solving the problem of unstable olfactory guidance.

[0030] Specifically, a stable and reliable direction vector field is generated by simulating dynamic odor diffusion, which solves the problem of signal instability; the system's robustness is significantly improved by actively compensating for environmental interference through gradient magnitude judgment and parameter adaptive update mechanism; the navigation accuracy is greatly improved by optimizing the spatial coordinates of the release point to achieve precise spatial anchoring of the guidance signal; and finally, by synthesizing multi-dimensional direction anchoring signals, the sense of smell is upgraded from a simple ambient cues to a precise interactive channel that can coordinate with vision and hearing, which significantly enhances the user's navigation efficiency and immersive experience in the virtual environment.

[0031] Based on the above embodiments, refer to Figure 2 As shown, a particle filtering algorithm is used to simulate odor diffusion and generate an odor distribution map of a virtual environment. The process includes: particle initialization, generating an initial particle set containing N particles from a preset odor release point; the initial state of each particle includes three-dimensional position coordinates, a three-dimensional velocity vector, age, and weight; at each time step, updating the state of each particle according to the particle state transition equation to obtain a predicted particle set; for each particle in the predicted particle set, calculating its Euclidean distance to the observation point, and updating the weight of each particle using the Euclidean distance to obtain a weighted particle set; calculating the effective number of particles in the weighted particle set and determining whether the effective number of particles is lower than a threshold number of particles; resampling is performed in response to the effective number of particles being lower than the threshold number of particles to obtain an effective particle set; and generating an odor distribution map of the virtual environment based on the effective particle set.

[0032] In specific application scenarios, one or more preset odor release point 3D coordinates are determined. At time t=0, an initial particle set containing N particles (e.g., N=5000) is generated from this release point. Each particle i is assigned an initial state: the 3D position coordinates are initialized to the release point coordinates (x0, y0, z0), the 3D velocity vector is initialized to zero, the age is initialized to 0, and the weight is initialized to 1. All particles start from the same position and with the same weight, simulating the physical fact that the odor concentrates at the release point at the initial moment, providing a correct starting point for subsequent diffusion simulations.

[0033] At each simulation time step Δt, the current medium flow data is acquired. For each particle in the particle set, the state is updated according to the particle state transition equation. All particles whose survival time exceeds the maximum value are removed. The natural evaporation and decay of odor are simulated to obtain the predicted particle set after physical law prediction. The particle state transition equation is set as follows:

[0034] ;

[0035] Δt represents the simulation time step; i represents the particle number; Represents the three-dimensional spatial coordinates of particle i at time t+1; Represents the three-dimensional position coordinates of particle i at time t; Let represent the three-dimensional velocity vector of particle i at time t; Let represent the three-dimensional velocity vector of particle i at time t+1; D represents the unit vector of the flow direction of the medium in the virtual reality environment. The velocity represents the flow rate of the medium in the virtual reality environment; k represents the wind speed influence coefficient, used to adjust the intensity of the wind's influence on particle motion, with a value ranging from 0.1 to 1.5S. -1 ; This represents the survival time of particle i at time t+1; This represents the survival time of particle i at time t.

[0036] The discrete form of Newton's second law implicitly assumes that the particle's mass is unit mass; therefore, the acceleration is numerically equal to the net force acting on it. This simulation describes the drag force exerted on odor particles by the flow of a medium (wind). In fluid mechanics, tiny particles move along with the fluid in a flow field, and their acceleration is directly related to the velocity and direction of the fluid. D represents the flow direction unit vector, which determines the direction of the force and is the key to generating anisotropy. The air velocity represents the airflow speed; the higher the velocity, the stronger the drag force on the particles. The wind speed influence coefficient k is a calibration parameter used to adjust the proportional relationship between the virtual simulation and the real physical world or the desired effect. Because our particles represent odor molecule clusters, their efficiency in being affected by wind may differ from the ideal point; k is used to scale and adjust for this. (The last sentence appears to be incomplete and possibly refers to airflow speed.) Anisotropy is introduced by the flow direction unit vector D, which directly injects the physical conditions of the virtual or real environment into the simulation, making the odor diffusion dynamically related to the external environment.

[0037] Survival time records the total simulation time of each particle since its generation. Odors do not exist in space indefinitely. By setting a maximum lifetime, particles are removed when their age exceeds this value. This directly simulates the process of odor molecules settling, decomposing, or diluting to below the perception threshold.

[0038] Subsequently, the real-time observation point position of the user in the virtual environment is obtained, namely the three-dimensional coordinates of their nostrils. For each particle i in the prediction particle set, the Euclidean distance between its current position and the observation point is calculated. Based on this Euclidean distance, the particle weights are updated through a preset perception model. The higher the weight of a particle, the greater its contribution to the event that the user smells the odor at their nostrils, thus obtaining a weighted particle set. The perception model typically uses a Gaussian model, which simulates the perceptual characteristic that odor concentration decreases with increasing distance, giving high weights to particles closer to the user and low weights to particles farther away.

[0039] The effective particle number is calculated as follows: :

[0040] ;

[0041] The weight of particle i; the number of effective particles. This reflects the effectiveness of the weighted particle set when all particle weights are equal. The maximum value is equal to N; when a few particles have extremely high weights while the majority have weights close to 0. The minimum value is 1.

[0042] Compare the effective particle count with a preset threshold particle count, such as N / 2; if the effective particle count is less than the threshold particle count, perform the following operations to obtain the effective particle set:

[0043] Elimination: Randomly remove particles with low weight based on their weight ratio;

[0044] Copying: Randomly copy particles with high weights according to their weight ratios;

[0045] Reset: Set the weight of all new particles to 1.0.

[0046] By eliminating the weak and selecting the strong, the particle swarm is kept concentrated in the area with the highest odor concentration around the user, thus enabling a more detailed simulation of the diffusion in that area and improving the resolution and accuracy of the simulation results.

[0047] Specifically, generating an odor distribution map of a virtual environment based on the effective particle set includes: discretizing the virtual environment space into a three-dimensional grid; traversing the effective particle set and mapping each particle to a voxel of the three-dimensional grid according to its position; accumulating the weights of all particles within each voxel to obtain the concentration value of that voxel; normalizing the concentration values ​​of all voxels to generate the odor distribution map at the current moment.

[0048] In specific application scenarios, the entire virtual environment space is divided into a regular three-dimensional grid, namely a voxel grid. The effective particle set is traversed, and each particle is assigned to its corresponding voxel based on its position. For each voxel, the weights of all particles within it are accumulated to obtain the original concentration value of that voxel. The original concentration values ​​of all voxels are divided by the global maximum concentration value to normalize them to the [0, 1] interval, generating the odor distribution map for the current moment. This is a three-dimensional data field, where each grid point stores the normalized relative odor concentration at that location. This transforms a discrete set of particles representing possibilities into a continuous, structured, computer-queryable and analyzable probability field (concentration field).

[0049] Based on the above embodiments, the concentration spatial gradient of the user's current location is calculated according to the odor distribution map, including: mapping the user's current location to the three-dimensional grid to obtain the center location; determining a local spatial neighborhood based on the center location, and extracting the concentration values ​​of each voxel within the local spatial neighborhood from the odor distribution map; and calculating the concentration spatial gradient of the center location based on the central difference method according to the concentration values ​​of each voxel within the local spatial neighborhood.

[0050] In specific application scenarios, the system acquires the user's precise 3D coordinates and odor distribution map within a virtual environment. Using the physical parameters of the distribution map grid (origin coordinates (Xmin, Ymin, Zmin) and voxel dimensions (Δx, Δy, Δz)), the user's world coordinates are converted into grid index coordinates. The output index (ic, jc, kc) represents the central voxel, signifying the corresponding cell in the discrete grid where the user's current location is located. This precisely anchors the dynamically changing user position onto the static odor distribution data grid, ensuring that all subsequent calculations are based on the user's current location, guaranteeing the real-time nature and relevance of the analysis.

[0051] A predetermined-size region centered on a central voxel (ic, jc, kc) is used as the local spatial neighborhood. The most commonly used and effective is the 3x3x3 voxel neighborhood, which is the smallest and most efficient unit for calculating gradients in all directions. This neighborhood includes the central voxel and its directly adjacent voxels in the six directions of front, back, left, right, up, and down. It contains enough information to estimate the partial derivatives in three directions while avoiding the computational overhead and sensitivity to unrelated information from distant locations caused by an excessively large neighborhood. The concentration values ​​stored in all 27 voxels within this 3x3x3 neighborhood are read from the odor distribution map. By acquiring the concentration data around the center point, the system can perceive the spatial distribution of odors near the user's location, which is the basis for calculating the rate of change (gradient). A single concentration value can only indicate how concentrated the odor is, while the data from a neighborhood can indicate which direction the odor becomes more concentrated.

[0052] The central difference method is a numerical differentiation method that uses the value of a function to the left and right (or front and back, up and down) of a point to estimate the derivative at that point, with higher accuracy than forward or backward difference. It calculates the partial derivatives in the X (left and right), Y (up and down), and Z (front and back) directions, combining these three partial derivatives into a three-dimensional vector. This transforms a scalar concentration map, which only represents concentration, into a gradient signal at the user's location that includes both the intensity of the change (magnitude) and the direction of the change (vector direction). The direction of this vector points to the direction of the fastest increase in odor concentration at the user's current location. The gradient magnitude represents the drasticness of the concentration change, providing the system with an objective and calculable indicator of the strength of the directional signal. A long gradient magnitude indicates strong directionality; a short gradient magnitude indicates a flat concentration field and ambiguous directionality.

[0053] Based on the above embodiments, before particle initialization, the method further includes: determining an initial release strategy for the particles according to the target guidance direction, wherein the initial release strategy includes: dynamically allocating the particle generation ratio of the two odor release points according to the target guidance direction relative to the user's current orientation, and / or adjusting the spatial position of the release points.

[0054] In specific application scenarios, strategy one: dynamically allocate the proportion of particle generation.

[0055] The system acquires the target guidance direction unit vector and the user's current orientation unit vector, and calculates the left and right components defined by these two unit vectors, which can be achieved through dot product and cross product. A simplified method is to calculate the angle between the target direction and the user's left / right direction axis. If the target direction is directly in front of the user and slightly to the right, the system allocates a higher particle generation ratio to the right release point (e.g., 70% on the right and 30% on the left); conversely, if the target is on the left, the proportion of the left release point is increased. If the target is directly behind, a special strategy can be adopted, such as equal release on both sides but using subsequent timing signals to indicate the rear. In the subsequent particle initialization step, the total number of generated particles N will be distributed between the left and right release points according to this preset ratio. This strengthens the direction signal from the source, improves simulation efficiency and convergence speed, and provides a rough direction pre-caution. Even in the absence of complex simulations or in windless environments, this simple left-more or right-more-less release strategy can provide the user with a basic and intuitive direction cues.

[0056] Strategy 2: Dynamically adjust the spatial position of the release point

[0057] Define the adjustable range: On a VR headset, a small three-dimensional spatial range is preset for the left and right release points, allowing them to shift their virtual coordinates near their physical locations. For example, the release points can move within a cube with sides of 20 centimeters centered on the default position.

[0058] Based on the target direction and the user's orientation, an offset vector is calculated. For example, if the target is to the right front, the right release point can be offset forward and to the right by a certain distance; simultaneously, the left release point can be offset backward and to the left by a certain distance. Essentially, this increases the projected distance between the two release points in the target direction. The calculated new coordinates are used as the initial positions for the particles. By spatially separating the release points, the paths that odor particles from different release points need to travel through become more distinct. This significantly enhances the distinguishability of the odor signals from the left and right sides when they reach the user's nasal cavity, making the concentration gradient calculation more stable and significant.

[0059] Based on the above embodiments, in response to the mean modulus of the concentration spatial gradient being lower than the modulus threshold, the parameters of the particle filter algorithm are updated to enhance anisotropic characteristics, including: increasing the wind speed influence coefficient and / or the flow direction of the covering medium; the flow direction of the covering medium includes: setting the flow direction of the medium in the state transition equation used to represent the virtual reality environment as the target guidance direction; or using a weighted fusion of the target guidance direction and the original flow direction of the medium as the updated flow direction.

[0060] In specific application scenarios, Option 1: Increase the wind speed influence coefficient

[0061] The system calls the stored current wind speed influence coefficient k and applies the following increase strategies: doubling strategy, fixed step size strategy, and adaptive strategy. The doubling strategy sets the new coefficient to a multiple of the old coefficient; the fixed step size strategy increases it by a fixed value; and the adaptive strategy links the increase to the scarcity of gradient modulus, that is, the larger the value of (threshold - current modulus) is, the larger the increase in wind speed influence coefficient k is.

[0062] Subsequently, the wind speed influence coefficient k in the particle filter algorithm's prediction step is updated to the increased wind speed influence coefficient. This allows particles to be blown towards the current wind direction more quickly, thereby establishing a significant concentration gradient in space more rapidly and shortening the time for the effective direction signal to form. Finally, the magnitude of the concentration gradient calculated at the user's location is... It will increase significantly.

[0063] Option 2: Cover the direction of medium flow, sub-strategy 1:

[0064] The original direction of the environmental medium flow used in the particle filtering prediction step is directly replaced with a normalized target guidance direction vector. This strategy is almost always used in conjunction with increasing the wind speed influence coefficient to simulate a strong wind blowing directly towards the target direction. This forces the particles to be blown towards the target direction regardless of the actual wind direction, which ensures that the gradient direction of the generated odor distribution map after correction is consistent with the height target guidance direction.

[0065] Sub-strategy 2:

[0066] Presetting a high fusion weight (e.g., 0.8) indicates the proportion of the target direction in the fusion. Performing weighted fusion to calculate the new direction vector is a smoother and more natural correction method. The change in direction is not instantaneous or abrupt, but a continuous transition, avoiding abrupt changes in simulation results and potentially bringing a more realistic user experience.

[0067] Based on the above embodiments, with minimizing the directional ambiguity at the user's location as the objective function, a gradient descent algorithm is used to optimize the spatial coordinates of the odor release point, including: constructing the objective function: Where m represents the odor release point number; P m The three-dimensional coordinates of the odor release point m; This represents the spatial magnitude of the concentration gradient at the odor release point m; The angle between the spatial gradient vector of the odor release point m and the target guidance direction vector is represented.

[0068] Using the three-dimensional coordinates of the odor release point as optimization variables, the three-dimensional coordinates of the odor release point are iteratively adjusted by calculating the gradient of the objective function with respect to the optimization variables until the objective function converges to the minimum value or reaches the maximum number of iterations.

[0069] The goal of this solution is to transform the odor release point from a fixed location into a decision variable that can be intelligently optimized in a virtual space, automatically finding the optimal release location that is most clearly and accurately perceived by the user. In specific application scenarios… This indicates when the odor is released from point P. m When the odor is released, the magnitude of the spatial gradient of the concentration at the user's current location is calculated through particle filtering simulation, which represents the intensity of the directional signal. This indicates the correctness of the direction signal; when the concentration gradient direction is consistent with the target direction, When perpendicular Conversely, it is less than zero. A comprehensive assessment at the odor release point P m The overall quality of the downward-directed signal; the larger this value, the clearer and more accurate the direction. This function considers both signal strength and accuracy, ensuring that the optimization process does not find a suboptimal solution where the signal is very strong but the direction is incorrect, or the direction is correct but the signal is weak.

[0070] Iterative optimization of release point location: Select an initial release point coordinate P m (For example, the default left or right position of the headset), set the learning rate η (e.g., 0.01) and the maximum number of iterations T.

[0071] For iterations from 0 to T-1, perform the following procedure:

[0072] a) Assume the release point is located at P m Re-execute the complete particle filter simulation S100, generate the odor distribution map, calculate the concentration gradient magnitude at the user's location, and calculate the current objective function value;

[0073] b) Disturb the release point P respectively m The x, y, z coordinates are given a tiny amount (e.g., 0.001 meters). For each perturbed coordinate, the simulation and evaluation in step a) are re-executed to calculate the corresponding objective function value.

[0074] c) Calculate the gradient based on the objective function values ​​calculated in step a) and step b, and update the release point position along the inverse direction of the gradient (the direction of the fastest descent of the function). When the function value converges in several consecutive iterations or reaches the maximum number of iterations, stop the optimization. The coordinates of the release point at this time are the coordinates of the locally optimal release point, and output them to the optimized release point set.

[0075] Through iteration, the release point is moved to the optimal location that produces the clearest directional signal for the current user's position and environment. Because the release point is optimized to the optimal position, the olfactory directional signal perceived by the user is enhanced, with the highest signal-to-noise ratio, thus making the user's direction judgment faster and more accurate, and greatly reducing the possibility of getting lost.

[0076] Preferably, the direction anchoring signal is synthesized based on the updated odor distribution map and the set of release points, including: calculating the consistency coefficient between the medium flow direction and the target guidance direction based on the updated odor distribution map and the set of release points; and adjusting the mode of the differentiated output based on the consistency coefficient: when the consistency coefficient is greater than or equal to a threshold coefficient, an intensity modulation mode that is coordinated with the medium flow direction is adopted; when the consistency coefficient is less than the threshold coefficient, a frequency or pulse width modulation mode that is opposed to the medium flow direction is adopted.

[0077] In specific application scenarios, from the optimized set of release points, the point with the highest guiding efficiency for the current user's location is selected as the main release point, typically the point closest to the user or most aligned with the target direction. The vector Q pointing from the main release point to the current user's location is calculated and normalized; this vector represents the shortest and most direct ideal path direction for the odor from the release point to the user. From the simulation data of the updated odor distribution map, the unit vector D of the medium flow direction currently influencing the main release point region is obtained, and the cosine of the angle between vectors Q and D, i.e., their dot product, is calculated; because both are unit vectors, the consistency coefficient is... The range of values ​​for is [-1, 1]; This indicates that the wind direction is exactly the same as the ideal path (tailwind); This indicates that the wind direction is perpendicular to the path; Indicates a complete headwind.

[0078] The threshold coefficient ranges from 0.6 to 0.8. When the consistency coefficient is greater than or equal to the threshold coefficient, the system enters a collaborative enhancement mode: the target direction channel is set to high intensity and continuous output; the non-target direction channel is set to low intensity, continuous output, or completely shut down. This fully utilizes natural wind power to achieve maximum directional contrast with the simplest intensity and minimum total release.

[0079] When the consistency coefficient is less than the threshold coefficient, an adversarial compensation mode is entered: the target direction channel emits high-frequency, long pulses; the non-target direction channel emits low-frequency, short pulses. The human brain is very sensitive to temporal patterns, and this alternating, rhythmic pulse contrast can produce a stronger sense of direction than a simple difference in intensity, especially effective in complex environments.

[0080] Example 2: Refer to Figure 3 As shown, an embodiment of the present invention discloses a VR olfactory orientation system, comprising:

[0081] The data acquisition module is used to acquire and output acquired data, including real-time pose data of the user in the virtual reality environment, target guidance direction, and real-time medium flow data of the physical environment.

[0082] The odor diffusion simulation module is used to simulate odor diffusion based on the collected data using a particle filtering algorithm, and generate an odor distribution map of a virtual environment.

[0083] The analysis module is used to calculate the concentration spatial gradient of the user's current location based on the odor distribution map, and determine whether the magnitude of the concentration spatial gradient is lower than a preset magnitude threshold; in response to the magnitude of the concentration spatial gradient being lower than the preset magnitude threshold, the parameters of the particle filtering algorithm in the odor diffusion simulation module are updated to enhance the anisotropic characteristics, and the odor diffusion simulation module is triggered to generate an updated odor distribution map.

[0084] The release point optimization module is used to optimize the spatial coordinates of the odor release points by using a gradient descent algorithm with the objective function of minimizing the directional ambiguity at the user's location, thereby obtaining a set of release points.

[0085] The output module is used to synthesize a directional anchoring signal based on the updated odor distribution map and the set of release points; wherein the directional anchoring signal is encoded as a timing and intensity command for controlling at least two odor release channels to perform differentiated outputs.

[0086] The embodiments of the present invention are used to execute the VR olfactory orientation method in Embodiment 1. Both are based on the same inventive concept and have the same technical effect, which will not be repeated here.

[0087] Example 3: Refer to Figure 4 As shown, an embodiment of the present invention discloses a VR device, including at least two odor release channels, namely odor release channel one 10 and odor release channel two 20. The VR device integrates the VR olfactory orientation system and executes the VR olfactory orientation method.

[0088] The VR device described in this invention integrates the VR olfactory orientation system and executes the VR olfactory orientation method, providing highly reliable and accurate olfactory spatial guidance capabilities, fundamentally enhancing the immersion and interaction depth of the VR experience.

[0089] In summary, the VR olfactory orientation method, system, and VR device described in this invention no longer regard the anisotropic diffusion of odors as a technical barrier that needs to be avoided, but rather treats it as an intrinsic variable. Through virtual simulation, dynamic optimization, and precise synthesis, a stable and reliable directional signal is obtained, fundamentally solving the problem of unstable olfactory guidance.

[0090] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0091] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0092] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0094] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A VR olfactory orientation method, characterized in that, include: Acquire data, including real-time pose data of the user in the virtual reality environment, target guidance direction, and real-time medium flow data of the physical environment; use a particle filtering algorithm to simulate odor diffusion based on the acquired data to generate an odor distribution map of the virtual environment; Based on the odor distribution map, calculate the spatial gradient of the concentration at the user's current location, and determine whether the mean magnitude of the spatial gradient of the concentration is lower than the magnitude threshold. In response to the mean modulus of the concentration spatial gradient being lower than the modulus threshold, the parameters of the particle filter algorithm are updated to enhance the anisotropic features and an updated odor distribution map is generated. Using minimizing the directional ambiguity at the user's location as the objective function, a gradient descent algorithm is employed to optimize the spatial coordinates of the odor release points, thereby obtaining a set of release points. Based on the updated odor distribution map and the set of release points, a directional anchoring signal is synthesized; wherein, the directional anchoring signal is encoded as a timing and intensity command for controlling at least two odor release channels to perform differentiated outputs.

2. The VR olfactory orientation method according to claim 1, characterized in that, Odor diffusion simulation is performed using a particle filter algorithm to generate an odor distribution map of a virtual environment, including: Particle initialization generates an initial particle set containing N particles from a preset odor release point; the initial state of each particle includes three-dimensional position coordinates, three-dimensional velocity vector, age, and weight. At each time step, the state of each particle is updated according to the particle state transition equation to obtain the predicted particle set. For each particle in the predicted particle set, calculate its Euclidean distance to the observation point, and update the weight of each particle using the Euclidean distance to obtain a weighted particle set. Calculate the number of effective particles in the weighted particle set and determine whether the number of effective particles is lower than the threshold number of particles; in response to the number of effective particles being lower than the threshold number of particles, perform resampling to obtain the effective particle set; An odor distribution map of the virtual environment is generated based on the effective particle set.

3. The VR olfactory orientation method according to claim 2, characterized in that, Based on the effective particle set, an odor distribution map of the virtual environment is generated, including: Discretize the virtual environment space into a three-dimensional mesh; Traverse the effective particle set and map each particle to a voxel of the three-dimensional mesh according to its position; The concentration value of a voxel is obtained by summing the weights of all particles within each voxel. The concentration values ​​of all voxels are normalized, and the odor distribution map is obtained at the current biochemical moment.

4. The VR olfactory orientation method according to claim 3, characterized in that, Based on the odor distribution map, the spatial gradient of concentration at the user's current location is calculated, including: Map the user's current location onto the 3D grid to obtain the center location; A local spatial neighborhood is determined based on the central location, and the concentration values ​​of each voxel within the local spatial neighborhood are extracted from the odor distribution map. Based on the concentration values ​​of each voxel in the local spatial neighborhood, the spatial gradient of concentration at the center position is calculated using the central difference method.

5. The VR olfactory orientation method according to claim 2, characterized in that, Before particle initialization, the process also includes: determining an initial particle release strategy based on the target guidance direction, wherein the initial release strategy includes: dynamically allocating the particle generation ratio of the two odor release points according to the target guidance direction relative to the user's current orientation, and / or adjusting the spatial position of the release points.

6. The VR olfactory orientation method according to claim 2, characterized in that, The particle state transition equation is set as follows: ； Δt represents the simulation time step; i represents the particle number; Represents the three-dimensional spatial coordinates of particle i at time t+1; Represents the three-dimensional position coordinates of particle i at time t; Let represent the three-dimensional velocity vector of particle i at time t; Let represent the three-dimensional velocity vector of particle i at time t+1; D represents the unit vector of the flow direction of the medium in the virtual reality environment. The velocity represents the flow rate of the medium in the virtual reality environment; k represents the wind speed influence coefficient, used to adjust the intensity of the wind's influence on particle motion, with a value ranging from 0.1 to 1.5S. -1 ; This represents the survival time of particle i at time t+1; This represents the survival time of particle i at time t.

7. The VR olfactory orientation method according to claim 6, characterized in that, In response to the mean magnitude of the concentration spatial gradient being lower than the magnitude threshold, the parameters of the particle filter algorithm are updated to enhance anisotropic characteristics, including: increasing the wind speed influence coefficient and / or the flow direction of the covering medium; the flow direction of the covering medium includes: setting the flow direction of the medium in the state transition equation used to represent the virtual reality environment as the target guidance direction; or using a weighted fusion of the target guidance direction and the original flow direction of the medium as the updated flow direction.

8. The VR olfactory orientation method according to claim 1, characterized in that, Using minimizing the directional ambiguity at the user's location as the objective function, a gradient descent algorithm is employed to optimize the spatial coordinates of the odor release point, including: Construct the objective function: Where m represents the odor release point number; P m The three-dimensional coordinates of the odor release point m; This represents the spatial magnitude of the concentration gradient at the odor release point m; The angle between the spatial gradient vector of the odor release point m and the target guidance direction vector is represented. Using the three-dimensional coordinates of the odor release point as optimization variables, the three-dimensional coordinates of the odor release point are iteratively adjusted by calculating the gradient of the objective function with respect to the optimization variables until the objective function converges to the minimum value or reaches the maximum number of iterations.

9. A VR olfactory orientation system, characterized in that, include: The data acquisition module is used to acquire and output acquired data, including real-time pose data of the user in the virtual reality environment, target guidance direction, and real-time medium flow data of the physical environment. The odor diffusion simulation module is used to simulate odor diffusion based on the collected data using a particle filtering algorithm, and generate an odor distribution map of a virtual environment. The analysis module is used to calculate the concentration spatial gradient of the user's current location based on the odor distribution map, and determine whether the magnitude of the concentration spatial gradient is lower than a preset magnitude threshold; in response to the magnitude of the concentration spatial gradient being lower than the preset magnitude threshold, the parameters of the particle filtering algorithm in the odor diffusion simulation module are updated to enhance the anisotropic characteristics, and the odor diffusion simulation module is triggered to generate an updated odor distribution map. The release point optimization module is used to optimize the spatial coordinates of the odor release points by using a gradient descent algorithm with the objective function of minimizing the directional ambiguity at the user's location, thereby obtaining a set of release points. The output module is used to synthesize a directional anchoring signal based on the updated odor distribution map and the set of release points; wherein the directional anchoring signal is encoded as a timing and intensity command for controlling at least two odor release channels to perform differentiated outputs.

10. A VR device, comprising at least two odor release channels, characterized in that, The VR device integrates the VR olfactory orientation system as described in claim 9, and performs the VR olfactory orientation method as described in any one of claims 1-8.

Images