Intelligent interactive system for AI glasses

Abstract

The present invention discloses an intelligent interaction system applied to AI glasses, which relates to the field of human-computer interaction technology and is used to solve the problems of increased resource utilization redundancy and reduced equipment power consumption efficiency. The system collects three-axis acceleration and angular velocity data through an inertial measurement unit integrated in the AI ​​glasses, calculates the rotational composite value and the acceleration change intensity, and fuses them to form the head posture change amount. The discrete degree is analyzed based on the MAD algorithm to judge the complexity of the interaction scene; according to the complexity, resource optimization or enhanced interaction response strategy is selected, such as improving the accuracy of speech recognition and adapting to complex sound fields; further, the proportional coefficient of light intensity, motion speed and noise intensity is introduced to construct a long short-term memory neural network model to predict the interaction complexity, and dynamic interaction mode adjustment is realized through a gating mechanism and regularization method. According to the predicted output, it is judged whether to improve the interaction mode, thereby ensuring the adaptive and high-reliability interaction capabilities of AI glasses in complex environments.

Description

Technical Field

[0001] The present invention relates to the field of human-computer interaction technology, and more specifically, to an intelligent interaction system applied to AI glasses. Background Art

[0002] With the rapid development of artificial intelligence, wearable devices and human-computer interaction technologies, AI glasses, as intelligent terminals integrating display, perception and computing functions, have gradually been widely deployed in application scenarios such as navigation assistance, real-time translation, and industrial remote guidance. Existing AI glasses usually integrate image sensors, voice recognition modules and inertial measurement units, and realize human-computer interaction through multimodal methods such as touch, voice, and gesture.

[0003] The existing technology has the following deficiencies:

[0004] At present, the human-computer interaction methods in AI glasses mostly rely on fixed gesture recognition, voice command triggering and preset interaction processes. They lack the ability to perceive natural movements such as head posture in real time, cannot dynamically identify fine-grained changes in user intentions, and the interaction strategy lacks scene adaptability. The system cannot dynamically allocate resources according to the complexity of the interaction, resulting in increased resource utilization redundancy and reduced device power consumption efficiency. Therefore, an intelligent interaction system for AI glasses is proposed.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an intelligent interaction system applied to AI glasses. By utilizing an intention recognition mechanism based on head posture changes, a dynamic evaluation algorithm for the complexity of interaction scenarios, a resource scheduling optimization strategy, and an adaptive control mechanism for interaction modes, the system solves the problems of delayed interaction response, low resource utilization, and inability to dynamically optimize modes raised in the above-mentioned background technology, thereby realizing intelligent response, dynamic resource allocation, and refined interaction experience of AI glasses during human-computer interaction.

[0007] To achieve the above objectives, the present invention provides the following technical solution, which is applied to the intelligent interactive system of AI glasses, including a data acquisition module, an intention recognition module, a strategy decision module, and a mode optimization module. The functions of each module are as follows:

[0008] The data acquisition module collects head posture data through the inertial measurement unit integrated into the AI ​​glasses and calculates the head posture change based on the head posture data;

[0009] The intention recognition module analyzes the changes in head posture and determines the complexity of the interaction scene;

[0010] The strategy decision module determines whether the interaction strategy needs to be adjusted based on the complexity of the interaction scenario. If the interaction scenario complexity is low, system resource optimization is performed; if the interaction scenario complexity is high, the interaction response capability is enhanced and the resource optimization results are evaluated based on the interaction response capability.

[0011] The mode optimization module determines whether the interaction mode needs to be improved based on the system resource optimization results.

[0012] In a preferred embodiment, an X, Y, and Z three-axis coordinate system is constructed with the geometric center of the inertial measurement unit integrated into the AI ​​glasses as the origin, and the head posture data is the linear acceleration and angular velocity along the X, Y, and Z three-axis coordinate system measured by the inertial measurement unit;

[0013] Starting from the current moment, a fixed time period is selected as the collection period. The angular velocity of the X, Y, and Z coordinate systems is collected, and the modulus length of the angular velocity is calculated to obtain the rotational composite value.

[0014] The same fixed time period before the acquisition cycle is selected as the previous acquisition cycle. The linear accelerations of the previous sampling cycle and the acquisition cycle are collected and subtracted. The modulus of the difference is then calculated to obtain the acceleration change intensity.

[0015] In a preferred embodiment, a fixed number of identical fixed time periods before the acquisition period are selected;

[0016] Get the maximum rotation composite value and acceleration change intensity within the same fixed time period;

[0017] The rotation composite value and acceleration change intensity within the acquisition period are divided by the rotation composite value and acceleration change intensity with the maximum value within all the same fixed time periods to obtain normalized values ​​respectively;

[0018] The normalized values ​​are fused according to the preset weighting coefficient to obtain the head posture change.

[0019] In a preferred embodiment, the position coordinates of the head are obtained, and a spherical area with a fixed diameter extending around the space with the position coordinates of the head as the center is used as the sample range;

[0020] Obtain all head posture changes within the sample range and merge them into a change dataset;

[0021] Arrange the head posture variation data in the variation data set from small to large, and obtain the median of the head posture variation in the variation data set;

[0022] Calculate the absolute deviation of each head posture change from the median to form an absolute deviation set;

[0023] The median of the absolute deviation set is taken to obtain the MAD value.

[0024] In a preferred embodiment, the MAD value is compared with a threshold for determining the complexity of the interaction scene;

[0025] When the MAD value is greater than the interaction scene complexity judgment threshold, the interaction scene complexity is judged to be high;

[0026] When the MAD value is less than or equal to the interaction scene complexity judgment threshold, the interaction scene complexity is determined to be low.

[0027] In a preferred embodiment, if the complexity of the interaction scenario is low, system resource optimization is directly performed;

[0028] If the complexity of the interaction scenario is high, improve the interactive response capability before optimizing system resources.

[0029] When the complexity of the interaction scene is low, the angular velocity and acceleration of the head posture are obtained, and the response is adjusted in combination with the operating frequency of the system until the head posture change reaches the head posture change threshold to achieve resource optimization.

[0030] In a preferred embodiment, when the complexity of the interaction scene is high, the speech recognition accuracy, noise intensity and ambient light intensity within the sample range are obtained to comprehensively judge the interaction response capability;

[0031] If the predicted speech recognition accuracy is not within the preset speech recognition accuracy range, the speech recognition accuracy is improved by an adaptive filtering algorithm;

[0032] The ambient light intensity is merged into the light intensity dataset, and the median of the ambient light intensity is taken as the light intensity threshold;

[0033] The ambient light intensity is divided by the light intensity threshold in turn to obtain each light intensity coefficient.

[0034] In a preferred embodiment, when the illumination intensity coefficient is within a preset illumination intensity coefficient interval threshold, the corresponding ambient illumination intensity is retained and the average value is taken to obtain the average illumination intensity coefficient;

[0035] The ratio of the predicted light intensity of the interactive scene to the average light intensity coefficient is used as the light intensity proportional coefficient;

[0036] Obtain the motion speed data within the sample range, the angular velocity of the predicted interaction scene, and the angular velocity value of the sample range, and comprehensively calculate the motion speed proportional coefficient;

[0037] The ratio of the noise intensity to the preset noise intensity threshold is used as the noise intensity proportional coefficient.

[0038] In a preferred embodiment, a long short-term memory network model is constructed by using a light intensity proportional coefficient, a motion speed proportional coefficient, and a noise intensity proportional coefficient to obtain a predicted output value;

[0039] If the predicted output value is less than the preset output threshold, the improved interactive mode will not be performed;

[0040] If the predicted output value is greater than the preset output threshold, the interaction mode is improved.

[0041] The technical effects and advantages of the present invention are as follows:

[0042] The present invention collects three-axis acceleration and angular velocity data through the inertial measurement unit integrated in the AI ​​glasses, calculates the rotational synthesis value and the acceleration change intensity, and fuses them into the head posture change. The discrete degree of the change is analyzed based on the MAD algorithm, thereby judging the complexity of the interaction scene. According to the complexity of the scene, the execution resource optimization or enhanced interaction response strategy is selected, such as improving the accuracy of speech recognition and adapting to complex sound fields. The proportional coefficient of light intensity, movement speed and noise intensity is further introduced to construct a long short-term memory neural network model to predict the interaction complexity. The gating mechanism and regularization method are used to realize efficient and dynamic interaction mode adjustment. The predicted output value is used to judge whether the interaction mode needs to be improved, so as to ensure that the AI ​​glasses have adaptive and highly reliable intelligent interaction capabilities in different complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 Schematic diagram of the steps of the intelligent interaction system applied to AI glasses of the present invention.

[0044] Figure 2 The figure is a flowchart of the implementation of the intelligent interaction system of the present invention applied to AI glasses. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0046] Example 1, an intelligent interactive system applied to AI glasses, such as Figure 1 and Figure 2 As shown, it includes a data acquisition module, an intention recognition module, a strategy decision module, and a pattern optimization module. The electrical signal connections of each module have the following functions:

[0047] The data acquisition module collects head posture data through the inertial measurement unit integrated into the AI ​​glasses and calculates the head posture change based on the head posture data;

[0048] The intention recognition module uses the MAD algorithm to analyze the changes in head posture and judge the complexity of the interaction scene accordingly;

[0049] The strategy decision module determines whether the interaction strategy needs to be adjusted based on the complexity of the interaction scenario. If the interaction scenario complexity is low, system resource optimization is performed; if the interaction scenario complexity is high, the interaction response capability is enhanced and the resource optimization results are evaluated based on the interaction response capability.

[0050] The mode optimization module determines whether the interaction mode needs to be improved based on the system resource optimization results.

[0051] The specific implementation is as follows:

[0052] In the data acquisition module, the head posture data generated by the user when using AI glasses is collected. The AI ​​glasses have an integrated inertial measurement unit, which consists of a three-axis accelerometer and a three-axis gyroscope. The head posture data is the linear acceleration and angular velocity along the X, Y, and Z three-axis coordinate system measured by the three-axis accelerometer and the three-axis gyroscope respectively. The origin of the X, Y, and Z three-axis coordinate system is located at the geometric center of the inertial measurement unit, which is fixed to the upper edge of the nose pad of the glasses frame and maintains a rigid connection with the user's head movement.

[0053] Starting from the current moment, a fixed time period is selected as the acquisition period. The angular velocity of the X, Y, and Z coordinate systems is collected. The modulus length of the angular velocity is calculated to obtain the rotational composite value, which reflects the rotational motion intensity of the head during the acquisition period. The specific calculation formula is: ;

[0054] in, are the angular velocities of the X, Y, and Z axes, It is the rotation composite value, which is used to describe the comprehensive degree of change in head posture around any axis. The larger the value, the more significant the rotation amplitude.

[0055] The same fixed time period before the acquisition cycle is selected as the previous acquisition cycle. The linear acceleration of the previous sampling cycle and the acquisition cycle are recorded respectively. The linear acceleration of the acquisition cycle is subtracted from the linear acceleration of the previous sampling cycle, and the modulus of the difference is calculated to obtain the acceleration change intensity, which reflects whether there is a sudden movement of the head. The specific calculation formula is as follows:

[0056] ;

[0057] in, is the linear acceleration of the X, Y, and Z axes during the acquisition period, is the linear acceleration of the X, Y, and Z axes in the previous acquisition cycle, The acceleration change intensity is larger, and the larger the value is, the more significant the acceleration change is.

[0058] To unify the dimensions and ensure data comparability, a fixed number of identical fixed time periods before the acquisition period were selected, and the rotational composite values ​​and acceleration change intensities within all fixed time periods were calculated. The rotational composite values ​​and acceleration change intensities within the acquisition period were divided by the maximum rotational composite values ​​and acceleration change intensities within all identical fixed time periods to obtain normalized values. The range of normalized values ​​was limited to The normalized rotation change intensity and acceleration change intensity are fused according to the preset weighting coefficient to obtain the head posture change. The specific calculation formula is:

[0059] ;

[0060] in, is the head posture change, is the preset weight parameter, satisfying , They are the maximum values ​​of the rotation synthesis value and acceleration change intensity in all fixed time periods.

[0061] It should be noted that the inertial measurement unit is a sensor unit integrated inside the AI ​​glasses. It is mainly used to collect the position coordinates and dynamic motion information of the head in real time. It consists of a three-axis accelerometer and a three-axis gyroscope, which are used to measure the linear acceleration and angular velocity of the head along three orthogonal axes in space respectively; its modulus is the square root of the sum of the squares of the components of any three-dimensional vector, which is used to measure the size of the vector and will not be elaborated here.

[0062] In the intent recognition module, the head position coordinates are obtained through the inertial measurement unit. A spherical area with a fixed diameter, centered around the head position coordinates, is used as the sample range. All head posture changes within the sample range are obtained and combined into a change data set. The MAD algorithm is used to analyze the change data set to determine the complexity of the current user's interaction scenario.

[0063] Arrange the head posture variation data in the variation data set from small to large, obtain the median of the head posture variation in the variation data set, and calculate the absolute deviation of each head posture variation from the median to form an absolute deviation set;

[0064] The median of the absolute deviation set is taken to obtain the MAD value. The MAD value is used to measure the overall discreteness of the head posture change, that is, the fluctuation amplitude of the user's head movement. The larger the MAD value, the more frequent the head movement changes and the less concentrated the posture changes, indicating that the user is in a more active and frequent interaction state at the current stage, and the complexity of the interaction scene is higher. Conversely, if the MAD value is smaller, it means that the user's movement changes are stable or weaker, the interaction intention is not obvious, and the complexity of the interaction scene is less.

[0065] Compare the MAD value with the interaction scene complexity judgment threshold. When the MAD value is greater than the interaction scene complexity judgment threshold, the interaction scene complexity is judged to be high.

[0066] When the MAD value is less than or equal to the interaction scene complexity judgment threshold, the interaction scene complexity is judged to be low;

[0067] It should be noted that the MAD algorithm is a robust statistical method used to measure the degree of discreteness of a set of data; the threshold for judging the complexity of the interaction scenario is a preset value, which is derived by professionals based on a large number of historical sample experiments. It represents the system's baseline for judging the activity level of interactive behavior and will not be elaborated here.

[0068] In the policy decision module, if the complexity of the interaction scenario is low, system resource optimization is performed directly; if the complexity of the interaction scenario is high, the interaction response capability is enhanced before system resource optimization.

[0069] When the complexity of the interactive scene is low, the angular velocity and acceleration of the head posture are further obtained through the inertial measurement unit, and the response is adjusted in combination with the operating frequency of the system;

[0070] When the running frequency is dynamically adjusted so that the head posture change reaches the head posture change threshold, it can be confirmed that the current interaction effect has met the resource optimization conditions, and the resource optimization operation is completed.

[0071] Among them, the system's operating frequency is the execution rate at which the core processing module responsible for interactive response tasks in the AI ​​glasses completes data processing tasks per unit time.

[0072] When the complexity of the interaction scene is high, the voice recognition accuracy and noise intensity are detected through the voice signal processing module, and the ambient light intensity is collected through the built-in photosensor of the AI ​​glasses.

[0073] Ambient light intensity usually does not directly affect speech recognition accuracy, but it can determine the stability of the user's visual focus, thereby indirectly affecting interactive responsiveness.

[0074] Among them, the speech signal processing module performs speech recognition through an array microphone, and takes the ratio of the speech recognition result to the voice command data as the speech recognition accuracy; after the microphone array collects the environmental background sound, the sound pressure level of the calculated background sound is obtained through fast Fourier transform as the noise intensity.

[0075] If the predicted speech recognition accuracy is not within the preset speech recognition accuracy range, the speech recognition accuracy is improved by an adaptive filtering algorithm;

[0076] The adaptive filtering algorithm can adjust the filter weight coefficient or filter structure parameters in real time according to the current input voice command data and noise intensity, so that the filter can better fit the current sound environment characteristics, thereby maximally suppressing interference noise without losing effective voice components.

[0077] The adaptive filtering algorithm identifies frequency bands with low signal-to-noise ratios in speech signals and performs signal energy enhancement processing on these frequency bands, or performs energy attenuation on noise-dominated frequency bands, thereby optimizing the spectral structure of the speech signal and increasing the proportion of effective speech components in the overall signal.

[0078] The filtering output result is fed back to the speech recognition module in real time, and the parameters of the adaptive filter are iteratively updated according to the current speech recognition accuracy to match the filtering strategy with the speech recognition performance until the output recognition accuracy stabilizes within the preset speech recognition accuracy range.

[0079] Through the above processing method, the voice input quality is automatically optimized in complex noise environments, the dynamic adjustment of voice recognition accuracy is achieved, and the interactive system is guaranteed to have good recognition performance and interactive reliability under different sound field conditions.

[0080] The AI ​​glasses' built-in photosensor collects ambient light intensity. The ambient light intensities within the sample range are arranged from smallest to largest and combined into a light intensity dataset. The median of the light intensity dataset is taken as the light intensity threshold. The light intensities in the light intensity dataset are divided by the light intensity threshold to obtain the light intensity coefficient. This light intensity coefficient is then compared with the preset light intensity coefficient interval threshold.

[0081] When the illumination intensity coefficient is within the preset illumination intensity coefficient interval threshold, the corresponding ambient light intensity is retained; otherwise, the corresponding ambient light intensity is removed.

[0082] The retained illumination intensities are averaged to obtain the average illumination intensity coefficient, and the ratio of the illumination intensity of the predicted interaction scene to the obtained average illumination intensity coefficient is used as the illumination intensity proportional coefficient.

[0083] The faster the user moves, the more types of noise there are in the interactive scene, and the lower the self-adjustment ability of the AI ​​glasses system, resulting in a faster decline in interactive response capabilities.

[0084] By accessing a motion monitoring website in the sample area, the system obtains motion speed data within the sample area as a reference for the interactive environment. The system uses an inertial measurement unit to collect the angular velocity of the predicted interactive scene and the angular velocity value of the sample area. The motion speed value within the sample area is divided by the corresponding angular velocity value to obtain the motion speed density of the sample area.

[0085] The ratio of the angular velocity of the predicted interaction scene to the angular velocity value of the sample range is used as the angular velocity proportional coefficient. The motion velocity density is multiplied by the angular velocity proportional coefficient to obtain the motion velocity density of the predicted interaction scene;

[0086] The motion speed density is compared with a preset interactive scene motion speed density threshold to obtain a corresponding motion speed proportional coefficient.

[0087] The system collects and predicts the noise intensity in the interactive scenario through the voice signal processing module, and calculates the ratio of the noise intensity to the preset noise intensity threshold to obtain the noise intensity proportional coefficient.

[0088] The larger the motion speed proportional coefficient, the greater the motion speed density of the predicted interactive scene, the more types of noise in the predicted interactive scene, and the higher the interactive response ability; when the noise intensity proportional coefficient is larger, the greater the noise intensity of the predicted interactive scene, and the higher the interactive response ability.

[0089] A long short-term memory network model is constructed through the light intensity proportional coefficient, motion speed proportional coefficient and noise intensity proportional coefficient to obtain the predicted output value.

[0090] The hidden layer weight parameters in the long short-term memory network model are trained. During the training process, a gated recurrent unit mechanism is introduced to dynamically update the memory unit state. At the same time, a regularization method is used to compress the parameters of the long short-term memory network model output and calculate the predicted output value.

[0091] Compare the complexity prediction value with the preset output value to evaluate the complexity of the current interaction scenario, and determine whether the interaction mode needs to be optimized and adjusted based on the complexity evaluation result. The specific implementation is as follows:

[0092] Network state update: Updated by input data and the hidden state of the previous time step. This process can be expressed as follows: ,in, is the hidden state at time t, is the input data, is the input weight matrix, is the circulant weight matrix, is the activation function.

[0093] Output layer calculation: calculated by hidden state and output weight matrix. The output calculation formula is: ,in, is the output value at time t, is the output weight matrix.

[0094] Training: Gradient descent can be used on the output layer to optimize the output weight matrix. The light intensity scale factor, motion speed scale factor, and noise intensity scale factor are used as input data. The sum of the three outputs obtained through the above operations is used as the predicted output value.

[0095] Among them, the selection of the t moment, the input weight matrix, the weight circulation matrix and the output weight matrix involved above are set by professionals in this field according to actual conditions and are not analyzed here.

[0096] It should be noted that the LSTM model is a variant of a neural network. Randomly selected weights in the weight matrix are called memory states. The LSTM model structure consists of three main components: an input gate, a forget gate, and an output gate. The input gate controls the inflow of information, the forget gate controls the forgetting of information, and the output gate controls the output of information. The gated recurrent unit (GRU) is a simplified version of the LSTM model, used to describe the dynamic changes of a memory unit. The memory unit is assumed to be a dynamic system that accumulates input signals and releases output signals when certain conditions are met. Regularization is a technique in machine learning used to prevent model overfitting, which occurs when a model performs well on training data but poorly on untested data. A photosensor is an electronic component that converts ambient light intensity into an electrical signal. In this case, it is used to collect light intensity in real time. An adaptive filtering algorithm is a filtering technology with real-time self-adjustment capabilities that dynamically adjusts the filter structure or parameters based on changes in the statistical characteristics of the input signal. A local motion monitoring website refers to a motion status data service platform for the user's surrounding area, which can obtain real-time or statistical motion speed data within a specific area. The various thresholds involved in the above steps are set by professionals in this field according to actual conditions and will not be analyzed here.

[0097] In the mode optimization module, if the predicted output value is less than the preset output threshold, the interactive mode is not performed;

[0098] If the predicted output value is greater than the preset output threshold, the interaction mode is improved.

[0099] The larger the predicted output value, the larger the light intensity proportional coefficient, motion speed proportional coefficient, and noise intensity proportional coefficient, and the higher the interactive response capability; the smaller the predicted output value, the smaller the light intensity proportional coefficient, motion speed proportional coefficient, and noise intensity proportional coefficient, and the lower the interactive response capability.

[0100] The method for improving the interaction mode is to improve the speech recognition accuracy to reach the preset speech recognition accuracy range, and to improve the speech recognition accuracy through the adaptive filtering algorithm in the strategy decision module.

[0101] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0102] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0103] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0104] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0105] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0106] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0107] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0108] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0109] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0110] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An intelligent interactive system for AI glasses, characterized by: It includes data acquisition module, intent recognition module, strategy decision module and pattern optimization module. The functions of each module are as follows: The data acquisition module collects head posture data through the inertial measurement unit integrated into the AI ​​glasses and calculates the head posture change based on the head posture data; The intention recognition module analyzes the changes in head posture and determines the complexity of the interaction scene; The strategy decision module determines whether the interaction strategy needs to be adjusted based on the complexity of the interaction scenario. If the interaction scenario complexity is low, system resource optimization is performed; if the interaction scenario complexity is high, the interaction response capability is enhanced and the resource optimization results are evaluated based on the interaction response capability. The mode optimization module determines whether the interaction mode needs to be improved based on the system resource optimization results.

2. The intelligent interactive system for AI glasses according to claim 1, characterized in that: An X, Y, and Z coordinate system is constructed with the geometric center of the inertial measurement unit integrated into the AI ​​glasses as the origin. The head posture data is the linear acceleration and angular velocity along the X, Y, and Z coordinate system measured by the inertial measurement unit. Starting from the current moment, a fixed time period is selected as the collection period. The angular velocity of the X, Y, and Z coordinate systems is collected, and the modulus length of the angular velocity is calculated to obtain the rotational composite value. The same fixed time period before the acquisition cycle is selected as the previous acquisition cycle. The linear accelerations of the previous sampling cycle and the acquisition cycle are collected and subtracted. The modulus of the difference is then calculated to obtain the acceleration change intensity.

3. The intelligent interactive system for AI glasses according to claim 2, characterized in that: Select a fixed number of identical fixed time periods before the acquisition period; Get the maximum rotation composite value and acceleration change intensity within the same fixed time period; The rotation composite value and acceleration change intensity within the acquisition period are divided by the rotation composite value and acceleration change intensity with the maximum value within all the same fixed time periods to obtain normalized values ​​respectively; The normalized values ​​are fused according to the preset weighting coefficient to obtain the head posture change.

4. The intelligent interactive system for AI glasses according to claim 3, characterized in that: Get the position coordinates of the head, and use the position coordinates of the head as the center to expand a spherical area with a fixed diameter around the space as the sample range; Obtain all head posture changes within the sample range and merge them into a change dataset; Arrange the head posture variation data in the variation data set from small to large, and obtain the median of the head posture variation in the variation data set; Calculate the absolute deviation of each head posture change from the median to form an absolute deviation set; The median of the absolute deviation set is taken to obtain the MAD value.

5. The intelligent interactive system for AI glasses according to claim 4, characterized in that: Compare the MAD value with the threshold for judging the complexity of the interaction scene; When the MAD value is greater than the interaction scene complexity judgment threshold, the interaction scene complexity is judged to be high; When the MAD value is less than or equal to the interaction scene complexity judgment threshold, the interaction scene complexity is determined to be low.

6. The intelligent interactive system for AI glasses according to claim 5, characterized in that: If the complexity of the interaction scenario is low, system resource optimization is performed directly; If the complexity of the interaction scenario is high, improve the interactive response capability before optimizing system resources. When the complexity of the interaction scene is low, the angular velocity and acceleration of the head posture are obtained, and the response is adjusted in combination with the operating frequency of the system until the head posture change reaches the head posture change threshold to achieve resource optimization.

7. The intelligent interactive system for AI glasses according to claim 5, characterized in that: When the complexity of the interaction scene is high, the speech recognition accuracy, noise intensity, and ambient light intensity within the sample range are obtained to comprehensively judge the interaction response capability; If the predicted speech recognition accuracy is not within the preset speech recognition accuracy range, the speech recognition accuracy is improved by an adaptive filtering algorithm; The ambient light intensity is merged into the light intensity dataset, and the median of the ambient light intensity is taken as the light intensity threshold; The ambient light intensity is divided by the light intensity threshold in turn to obtain each light intensity coefficient.

8. The intelligent interactive system for AI glasses according to claim 7, characterized in that: When the illumination intensity coefficient is within the preset illumination intensity coefficient interval threshold, the corresponding ambient light intensity is retained and the average value is taken to obtain the average illumination intensity coefficient; The ratio of the predicted light intensity of the interactive scene to the average light intensity coefficient is used as the light intensity proportional coefficient; Obtain the motion speed data within the sample range, the angular velocity of the predicted interaction scene, and the angular velocity value of the sample range, and comprehensively calculate the motion speed proportional coefficient; The ratio of the noise intensity to the preset noise intensity threshold is used as the noise intensity proportional coefficient.

9. The intelligent interactive system for AI glasses according to claim 8, characterized in that: The long short-term memory network model is constructed by using the light intensity proportional coefficient, motion speed proportional coefficient and noise intensity proportional coefficient to obtain the predicted output value; If the predicted output value is less than the preset output threshold, the improved interactive mode will not be performed; If the predicted output value is greater than the preset output threshold, the interaction mode is improved.

Images