Dynamic adaptive virtual reality dog training method and system, storage medium
Through virtual reality technology and machine learning algorithms, combined with convolutional neural networks and long-term memory networks, the dog training plan is dynamically adjusted, and the problems of data collection complexity and difficulty of emotional prediction in intelligent dog training technology are solved, achieving efficient and personalized The effect of dog training.
Patent Information
- Application Number
- CN202410545586.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-06
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-05-06
AI Technical Summary
The existing intelligent dog training technology has complexity and environmental interference problems in data collection and processing, making it difficult to accurately understand and predict dog mood changes. The limitations of algorithms and models lead to lack of adaptability and universality in training plans. It cannot meet the diverse needs of different dogs.
Combining virtual reality technology and machine learning algorithms, analyzing dog's physiological and behavioral data through convolutional neural networks and long-term memory networks, adjusting training plans in real time, simulate diversified training environments, dynamically adapting to dog's emotional state, and improving training accuracy and adaptability.
It achieves highly personalized and dynamic adaptable training effects, improves the accuracy and efficiency of training, solves the problem that traditional methods cannot cover, and makes up for the shortcomings of existing intelligent dog training technology.
Smart Images

Figure CN118435880B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent pet training, and in particular to a dynamic adaptive virtual reality dog training method and system, and a storage medium. Background Art
[0002] In modern society, as people's demand for behavioral training of pet dogs grows, traditional dog training methods face various limitations, such as low training efficiency, single methods, and difficulty in dealing with complex behavioral problems. Intelligent dog training, that is, using advanced technical means to train dog behavior, is regarded as an important means to solve these problems due to its high efficiency and precision. Intelligent dog training can provide personalized training programs by integrating sensors, data analysis, machine learning and other technologies to better meet the specific needs of dogs and dog owners.
[0003] Although intelligent dog training has significant advantages, it also faces some problems in practical applications. First, existing intelligent dog training solutions often rely on a large amount of data input, such as the dog's physiological and behavioral data. The collection and processing of this data is complex and easily affected by environmental interference, making it difficult to ensure data quality and reliability. Secondly, existing technologies often seem powerless when dealing with highly personalized and situational behaviors, and it is difficult to accurately understand and predict the emotional changes of dogs, which directly affects the effectiveness and efficiency of training. In addition, the universality and scalability of intelligent training methods are often limited by the limitations of algorithms and models, making it difficult to adapt to the diverse needs of different dogs.
[0004] In the face of these problems, existing intelligent dog training technologies (Chinese invention patent, publication number: CN117546796A) show some obvious deficiencies. For example, most existing solutions lack effective adaptability and flexibility, and cannot adjust the training plan in real time to cope with the dog's immediate reaction in different training environments. In addition, these technical solutions still have defects in the accurate identification and processing of emotional states, and often lead to misjudgment due to inaccurate algorithms or improper data processing. These problems need to be overcome through more advanced technical solutions. Summary of the invention
[0005] In view of the many problems existing in the above-mentioned prior art, the present invention provides a dynamically adaptive virtual reality dog training method and system, and a storage medium. The present invention combines virtual reality technology and machine learning algorithms to realize a personalized and dynamically adaptive dog training system; through virtual reality technology, the present invention can create a diverse training environment and simulate various situations in the real world to test and strengthen the dog's behavioral response ability; at the same time, the machine learning algorithms of convolutional neural networks (CNN) and long short-term memory networks (LSTM) are applied to effectively process and analyze the collected dog's behavioral action data and physiological characteristic data, predict and evaluate the dog's emotional state in real time, and adjust the training difficulty and method accordingly; this processing method not only improves the accuracy of training, but also enhances the adaptability of the training process, ensuring that the training plan responds to the dog's actual performance and needs at any time.
[0006] A dynamic adaptive virtual reality dog training method comprises the following steps:
[0007] The sensor components installed on the dog collect physiological characteristic data, behavioral action data and environmental condition data. After all the data are integrated and synchronized, noise and outliers are removed to obtain a cleaned status data set.
[0008] The emotional behavior model is trained based on the annotated historical training data. The emotional behavior model obtains emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment;
[0009] Adjust the training difficulty and method in real time based on the emotional state data, and use the adaptive algorithm to update the training plan for the first time;
[0010] Use virtual reality technology to simulate training scenarios to test the dog's ability to cope with the training plan. In the training scenario, collect the dog's behavioral action data and physiological characteristic data in real time, use the emotional behavior model to re-evaluate the dog's emotional state, adjust the complexity and type of virtual training according to the evaluation results, and update the training plan for the second time to ensure that the training plan matches the dog's current behavior and physiological state;
[0011] Collect training performance data of the training plan after the second update, evaluate the training effect, and feed the training results back to the emotional behavior model to fine-tune the model and training plan.
[0012] Preferably, the physiological characteristic data include: heart rate and body temperature; the behavioral action data include: dog's movement speed and action mode; the environmental condition data include: temperature and humidity;
[0013] Integrate and synchronize the collected physiological characteristic data, behavioral action data, and environmental condition data so that all data come from the same time window.
[0014] Preferably, the emotional behavior model adopts a combination of a convolutional neural network and a long short-term memory network;
[0015] The spatial features of each time point extracted from behavioral actions and environmental conditions are processed through convolutional neural networks. The changes of spatial features over time are tracked and analyzed through long-term short-term memory networks to capture long-term behavioral trends and emotional fluctuations.
[0016] Emotional behavioral models are used to predict a dog’s emotional state in real time, understand the emotional drivers behind a dog’s behavior, and support the development of personalized training programs based on each dog’s specific reactions and behavioral patterns.
[0017] Preferably, the emotional behavior model obtains emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment, including:
[0018] The emotional behavior model predicts the current emotional state based on the physiological characteristic data and the behavioral action data, and outputs the emotional state data;
[0019] Using environmental condition data as additional input features of the emotional behavior model so that the emotional behavior model understands potential external influences on the emotional state; retraining or adjusting the emotional behavior model based on new data containing environmental condition data;
[0020] Cross-validation is used to ensure that the emotional behavior model remains accurate on new data, and the emotional state data is updated after adjusting the model parameters based on feedback to adapt to the influence of different environments.
[0021] Preferably, the expression of the emotional behavior model is:
[0022]
[0023] in, Represents emotional state; function Represents the emotional behavior model, Represents physiological characteristic data; Represents behavioral action data; Indicates environmental condition data; Represents model parameters, which are learned during the training process; function This is achieved through a combination of convolutional neural networks and long short-term memory networks.
[0024] Preferably, an adaptive algorithm is used to update the training plan for the first time:
[0025] Collect real-time feedback from training and clearly define rewards or punishments based on training goals and the dog’s feedback;
[0026] Using historical and real-time data to train an adaptive algorithm to learn a response strategy that identifies the type of action instructions that maximize long-term rewards in a given situation and periodically evaluates and adjusts it;
[0027] Implement training plans based on coping strategies and adjust training difficulty and methods in real time.
[0028] Preferably, the complexity and type of virtual training are adjusted according to the evaluation results, including:
[0029] When dogs show excessive stress responses, reduce the difficulty of training, including: reducing environmental noise, reducing task complexity or providing more positive incentives;
[0030] When the dog appears relaxed and in control in certain training scenarios, increase the difficulty, introducing more complex tasks or small stressors to heighten the training challenge.
[0031] Preferably, the training performance data includes: the implementation effect of the training plan in virtual reality and real environment after the second update, and the evaluation of the training effect is completed based on the training performance data, and the implementation effect includes: task completion rate, error rate, and completion speed of training tasks.
[0032] A system for implementing the above-mentioned dynamic adaptive virtual reality dog training method, comprising:
[0033] A sensor component is arranged on the dog and used to collect physiological characteristic data, behavioral action data and environmental condition data of the dog;
[0034] A data processing unit connected to the sensor assembly, for integrating and synchronizing the collected data and removing noise and outliers to obtain a cleaned state data set;
[0035] An emotional behavior model unit, configured with a processor and a memory for training an emotional behavior model based on labeled historical training data, wherein the emotional behavior model is used to obtain emotional state data by analyzing physiological characteristic data and behavioral action data, and to improve the accuracy of emotional state judgment by combining environmental condition data;
[0036] A training adjustment unit, used to adjust the difficulty and method of training in real time according to the emotional state data output by the emotional behavior model unit, and to perform the first update of the training plan using an adaptive algorithm;
[0037] A virtual reality unit is used to simulate training scenarios to test the dog's ability to cope with the training plan. In the training scenario, the dog's behavioral action data and physiological characteristic data are collected in real time, and the emotional state of the dog is re-evaluated using the emotional behavior model. The complexity and type of virtual training are adjusted according to the evaluation results, and the training plan is updated for the second time accordingly to ensure that the training plan matches the dog's current behavior and physiological state;
[0038] The performance evaluation unit is used to collect and evaluate the training performance data of the training plan after the second update, evaluate the training effect, and feed back the training results to the emotional behavior model unit to fine-tune the model and training plan.
[0039] A storage medium stores a computer program, which, when executed by a processor, implements the steps of the dynamic adaptive virtual reality dog training method.
[0040] Compared with the prior art, the advantages and beneficial effects of the present invention are:
[0041] Combining virtual reality technology and machine learning algorithms, the present invention achieves highly personalized and dynamically adaptable training effects;
[0042] By using virtual reality technology, the present invention can simulate various real training environments and provide dogs with diverse training scenarios, thereby effectively improving the application scope and effect of training;
[0043] The machine learning algorithms used (especially the combination of convolutional neural networks and long short-term memory networks) can accurately analyze and predict the dog's behavior and emotional state, adjust the training plan based on real-time data, and ensure that the training program always adapts to the dog's current emotional and behavioral needs;
[0044] The present invention significantly improves the accuracy of the training effect and the efficiency of the training process, solves the problems that the traditional methods cannot cover, and also makes up for the shortcomings of the existing intelligent dog training technology. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 It is a schematic diagram of the process of the present invention;
[0046] Figure 2 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION
[0047] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0048] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0049] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.
[0050] In the case of using expressions such as "at least one of A, B, and C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.). In the case of using expressions such as "at least one of A, B, or C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, or C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).
[0051] Some block diagrams and / or flow charts are shown in the accompanying drawings. It should be understood that some boxes or combinations thereof in the block diagrams and / or flow charts can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that these instructions can create a device for implementing the functions / operations described in these block diagrams and / or flow charts when executed by the processor. The technology of the present disclosure can be implemented in the form of hardware and / or software (including firmware, microcode, etc.). In addition, the technology of the present disclosure can take the form of a computer program product on a computer-readable storage medium storing instructions, which can be used by an instruction execution system or in combination with an instruction execution system.
[0052] like Figure 1 As shown, a dynamic adaptive virtual reality dog training method comprises the following steps:
[0053] The sensor components installed on the dog collect physiological characteristic data, behavioral action data and environmental condition data. After all the data are integrated and synchronized, noise and outliers are removed to obtain a cleaned status data set.
[0054] The present invention can monitor and record the dog's physiological and behavioral data in real time by installing various sensor components on the dog, such as a heart rate monitor, a thermometer, and a motion capture device. The technical basis of these sensors is to use biofeedback and environmental sensing technology to capture subtle physiological changes and behavioral dynamics. For example, a heart rate monitor can continuously track heartbeat changes, and when the dog experiences different training scenarios, rapid changes in heart rate can indicate that it is emotionally excited or nervous.
[0055] The collected data needs to be integrated and synchronized within a unified time frame to ensure the temporal consistency and integrity of the data set when analyzed. This is achieved through advanced data processing algorithms that can calibrate data from different sources and different time points to form a continuous and comparable data stream. If the motion capture device records the dog's rapid movement behavior, and the heart rate data synchronously shows an increase in heart rate, this set of data can be integrated to analyze the dog's stress response under specific environmental stimuli.
[0056] The collected data often includes noise and outliers, which may be caused by sensor errors, environmental interference or atypical behavior. The principle of data cleaning is to identify and remove these abnormal data through algorithms, such as using statistical methods to identify data points that deviate from the average, to ensure the accuracy and reliability of subsequent analysis. If a body temperature sensor suddenly records an abnormally high temperature value due to a malfunction, the data cleaning process will identify this and remove it from the data set to avoid false health alerts.
[0057] By monitoring the dog's physiological and behavioral responses in real time and accurately, the training plan can be adjusted more flexibly to suit the specific needs of each dog. For example, if the data shows that the dog continues to show high tension in a certain training session, the trainer can adjust the difficulty or method of the session in time to reduce the dog's stress.
[0058] In addition, the cleaned data set in the present invention provides a reliable basis for the evaluation of training effects, making the quantification and optimization of training effects more scientific and specific. This not only improves the efficiency of training, but also ensures the sustainability and repeatability of training results.
[0059] Preferably, the physiological characteristic data include: heart rate and body temperature; the behavioral action data include: dog's movement speed and action mode; the environmental condition data include: temperature and humidity;
[0060] Integrate and synchronize the collected physiological characteristic data, behavioral action data, and environmental condition data so that all data come from the same time window.
[0061] Physiological characteristic data, such as heart rate and body temperature, provide basic information about the dog's physiological state, which reflects the dog's health and stress level. A rapid increase in heart rate may indicate that the dog is nervous or excited; changes in body temperature may indicate the dog's health or the suitability of the environment.
[0062] Behavioral motion data, such as movement speed and behavior patterns, reveal the dog’s activity level and behavior habits, which is critical for judging its behavioral state and emotional adjustment. For example, frequent running may indicate that the dog is excited or anxious in the virtual environment.
[0063] Environmental condition data, such as temperature and humidity, affect the dog's comfort and behavior. The suitability of the ambient temperature directly affects the dog's body temperature and heart rate, while changes in humidity may also affect its behavioral responses.
[0064] Merge data from different sources and types (physiological, behavioral, environmental) into a unified data set. This requires advanced data processing techniques to ensure consistency in format and scale. All data needs to be synchronized within the same time window to ensure temporal consistency between the data. This is usually achieved through timestamp and data alignment techniques to ensure the accuracy of analysis and model training. Time window synchronization, for example, if the timestamps of heart rate data and movement speed data are inconsistent, it may lead to errors in the analysis results. Through time window synchronization, it can be ensured that each data point corresponds accurately during analysis, thereby accurately reflecting the true state of the dog at a specific time.
[0065] By integrating and synchronizing data, the emotional behavior model of the present invention can more accurately analyze and predict the emotional state of the dog. The integrated data provides a comprehensive perspective, helping the model capture the correlation between different data, such as the relationship between increased heart rate and increased exercise speed.
[0066] The synchronized data set allows the training program to be more flexibly adjusted based on real-time data. For example, if the synchronized data shows that the dog's heart rate is abnormally high in a hot and humid environment, the training program can be automatically adjusted to reduce external pressure and reduce the difficulty of training, thereby avoiding the impact of excessive stress on the dog.
[0067] Extensive data recording and analysis allows each dog's training program to be individually adjusted based on its unique physiological and behavioral data. This adjustment is based on a deep understanding of each dog's reactions, thereby providing a training method that best suits its specific needs.
[0068] The emotional behavior model is trained based on the annotated historical training data. The emotional behavior model obtains emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment;
[0069] Emotional behavior models are trained using machine learning techniques, specifically supervised learning. Such models use historical training data that includes labeled emotional states (such as calm, nervous, excited, etc.), as well as physiological characteristic data and behavioral action data associated with these emotional states. The goal of model training is to learn to recognize patterns in the data that can predict emotional states in similar situations in the future. For example: if historical data shows that dogs show nervous or excited emotions when heart rate and movement speed increase, the model will learn this association and use it to predict emotional responses under similar conditions.
[0070] The emotional behavior model identifies emotional states by analyzing input physiological feature data (such as heart rate and body temperature) and behavioral action data (such as movement speed and behavior patterns). This analysis involves complex algorithms, such as convolutional neural networks (CNN) for extracting features from time series data, and long short-term memory networks (LSTM) for capturing the dynamic changes of these features over time. For example: During training, the model identified a specific combination of heart rate patterns and movement behavior patterns, which are often associated with a "stressed" state. In actual applications, when the model detects this pattern, it can predict that the dog is in a state of tension.
[0071] To improve the accuracy of emotional state judgment, the model also needs to consider environmental condition data, such as temperature and humidity. These environmental factors may directly affect the dog's physiological reactions and behavioral performance, thereby indirectly affecting its emotional state. For example, high temperatures may make dogs appear more irritable or restless, and this physiological reaction may be mistakenly interpreted as emotional excitement. By incorporating environmental temperature factors into the analysis, the model can more accurately distinguish whether physiological and behavioral changes are caused by the environment or the dog's internal emotional state.
[0072] The present invention uses a comprehensive analysis of physiological data, behavioral data, and environmental data, and the emotional behavior model can more accurately judge the emotional state of a dog. This comprehensive analysis helps the model distinguish between various influencing factors and improve the reliability and accuracy of emotional judgment. The model is able to predict the emotional state of each dog based on its specific response, providing trainers with data-backed insights to make training and intervention more personalized. This personalized approach can better meet the specific needs of each dog and improve the effectiveness and efficiency of training. The adaptability of the emotional behavior model ensures that the emotional state of the dog can be accurately assessed even in a dynamically changing training environment. This is particularly important for virtual reality training, because the VR environment can simulate a variety of different training scenarios, and the flexibility and adaptability of the model directly affect the effectiveness of training.
[0073] The emotional behavior model plays a core role in the dynamic adaptation of virtual reality dog training programs. It not only provides in-depth insights into the dog's emotional state, but also supports the optimization and personalization of training methods, greatly improving the scientificity and effectiveness of the training program.
[0074] Preferably, the emotional behavior model adopts a combination of a convolutional neural network and a long short-term memory network;
[0075] The spatial features of each time point extracted from behavioral actions and environmental conditions are processed through convolutional neural networks. The changes of spatial features over time are tracked and analyzed through long-term short-term memory networks to capture long-term behavioral trends and emotional fluctuations.
[0076] Emotional behavioral models are used to predict a dog’s emotional state in real time, understand the emotional drivers behind a dog’s behavior, and support the development of personalized training programs based on each dog’s specific reactions and behavioral patterns.
[0077] A combination of convolutional neural networks (CNNs) and long short-term memory networks (LSTMs) is designed to optimize the ability to process time series data and spatial features. CNNs are used to extract spatial features from behavioral actions and environmental conditions, for example, extracting a dog's posture or movement pattern from a video frame. LSTMs are used to process how these features change over time, capturing dynamic and long-term trends in behavior and emotions. For example, in a training scenario, if a dog's movement speed suddenly increases, CNN can identify this change and extract relevant features, and LSTM then tracks the subsequent development of these features to determine whether this is a short-term state of excitement or a persistent state of tension.
[0078] CNN layer and node configuration,In order to effectively extract spatial features, CNN usually includes multiple convolutional layers and pooling layers. Each convolutional layer uses multiple filters to extract different features, such as edges, textures, etc., and the subsequent pooling layer is used to reduce the spatial size of the features and enhance the generalization ability of the model. When processing behavioral action data, such as the movement pattern of a dog, the configured filter size, step size, and number need to be optimized according to the characteristics of the action data to capture key action features.
[0079] LSTM layer and node configuration,The LSTM layer is used to process time series data,capturing the changes of features over time.,In this setting, the number of units in the LSTM layer is usually proportional to the,complexity of the time series data. For complex behavior patterns and,emotional changes, more LSTM units are required to capture,long-term dependencies.
[0080] The flow of data from CNN to LSTM is sequential. First, the features extracted from the behavioral actions and environmental conditions data are processed by the CNN layer and converted into a set of feature maps that summarize the key spatial properties of the input data. These feature maps are then fed into the LSTM layer, which processes the dynamic changes of these features over time and outputs a feature vector for each time step, which helps capture the time-dependent characteristics of behavior and emotions.
[0081] Processing of spatial features: CNN effectively processes and extracts spatial features in input data, such as shapes, edges, etc., through its multi-layer structure. This is particularly useful in analyzing dog behavior, such as parsing dog body language or behavior patterns. Time series analysis: LSTM's long-term memory capacity makes it very suitable for processing time series data, and it can remember and analyze the impact of past information on future states. This is particularly important for understanding and predicting the emotional changes of dogs, because emotional states are usually triggered by a series of events.
[0082] The model's ability to predict in real time allows training to be adjusted based on the dog's immediate emotional state. For example, if the model predicts that the dog is about to enter a high-stress state, the trainer can immediately take steps to reduce the difficulty of training or change the training method. The model supports the development of a personalized training plan for each dog by analyzing each dog's specific reactions and behavior patterns. This training plan takes into account the uniqueness of each dog and promotes its learning and adaptation more effectively.
[0083] Processing real-time data streams requires models to be able to respond quickly to new input data. In real-time systems, data is usually continuously input in the form of streams, and the model needs to process and predict new data immediately after receiving it. This can be achieved by implementing data stream pipelines (such as using messaging systems such as Apache Kafka or RabbitMQ) and optimizing the computational efficiency of the model (such as using GPU accelerated computing).
[0084] Based on the emotional state data output by the model, the training plan can be dynamically adjusted. For example, if the model predicts that the dog is currently in a high-stress state, the training parameters can be adjusted to reduce the intensity of the training and increase positive reinforcement. Specific adjustment parameters include the frequency and duration of training, the complexity of the training tasks, and the type of incentives used.
[0085] The model can accurately monitor the emotional changes of dogs, helping trainers to understand and respond to the needs of dogs in a timely manner, thereby effectively managing the emotional problems that may arise in dogs during training and reducing stress and anxiety during training; through real-time prediction and personalized training, the training program can be more accurately adjusted to the actual performance and needs of the dog, thereby improving the efficiency and effectiveness of training. This not only improves the quality of training, but also enhances the learning motivation and satisfaction of the dog; the dynamic adaptability of the model ensures that the training can flexibly respond to various emergencies, while strengthening the feedback mechanism of the training process. This means that training can be continuously optimized based on real-time data, achieving truly dynamic adaptive training.
[0086] Preferably, the emotional behavior model obtains emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment, including:
[0087] The emotional behavior model predicts the current emotional state based on the physiological characteristic data and the behavioral action data, and outputs the emotional state data;
[0088] Using environmental condition data as additional input features of the emotional behavior model so that the emotional behavior model understands potential external influences on the emotional state; retraining or adjusting the emotional behavior model based on new data containing environmental condition data;
[0089] Cross-validation is used to ensure that the emotional behavior model remains accurate on new data, and the emotional state data is updated after adjusting the model parameters based on feedback to adapt to the influence of different environments.
[0090] The emotional behavior model first predicts the current emotional state based on physiological characteristic data (such as heart rate, body temperature) and behavioral action data (such as movement speed, behavior pattern). This analysis involves complex pattern recognition and machine learning algorithms, such as neural networks, which can identify changes in emotions from these physiological and behavioral signals.
[0091] The model incorporates environmental condition data (such as temperature and humidity) as additional input features. This is done to allow the model to understand and explain that changes in emotional state are not only caused by internal physiological or behavioral factors, but may also be significantly affected by external environmental conditions. For example: in a hot and humid environment, a dog may show more restless or irritable behavior. By analyzing this environmental data, the model can more accurately distinguish that this behavior is caused by an uncomfortable climate rather than an internal emotional problem of the dog.
[0092] Parameter adjustment and optimization process of emotional behavior model The adjustment of model parameters is the key to optimizing model performance, especially retraining under new environmental conditions. It is possible to further describe which model parameters (such as learning rate, number of layers, number of nodes, etc.) usually need to be adjusted, including:
[0093] Learning rate: The learning rate is an important parameter that determines the magnitude of the model's parameter updates in each iteration. Setting the learning rate too high may cause oscillations and instability during model training, while setting the learning rate too low may cause the training speed to be too slow or fall into a local minimum. Depending on the data characteristics and model structure, the learning rate may need to be adjusted accordingly.
[0094] Number of layers and nodes: For convolutional neural networks and long short-term memory networks, the number of layers and nodes (or neurons) directly affects the complexity and capabilities of the model. Increasing the number of layers or nodes can improve the learning ability of the model, but it may also increase the risk of overfitting. Therefore, it is necessary to adjust these parameters based on the performance of the model on the training and validation data.
[0095] Regularization parameters: Regularization techniques (such as L1 and L2 regularization) are used to control the complexity of the model and prevent overfitting. By adjusting the regularization parameters, the model can be kept from overfitting the training data while maintaining good performance.
[0096] Using cross-validation results to guide parameter adjustments includes:
[0097] By dividing the dataset into multiple small subsets, cross-validation methods such as K-fold cross-validation can evaluate the performance of the model on unseen data. In each round, one subset is used as the test set and the rest as the training set, so that each data point has a chance to be used as test data.
[0098] Evaluate performance metrics after each round of cross-validation, such as accuracy, recall, F1 score, etc. Analyzing changes in these metrics can help determine whether the model is overfitting (for example, low training error but high test error) or underfitting (both training error and test error are high).
[0099] Adjust the model parameters based on the results of cross-validation. If the model appears to be overfitting, try reducing the number of layers or nodes or increasing the regularization strength. If the model appears to be underfitting, consider increasing the number of layers or nodes, or adjusting the learning rate to speed up the learning process.
[0100] Cross-validation is performed repeatedly until the optimal parameter combination is found to ensure that the model maintains stable high performance on various data subsets.
[0101] As environmental conditions change, the model needs to be retrained or adjusted to accommodate these changes. This is usually done by collecting new datasets containing environmental data and applying machine learning techniques to readjust the model's weights and parameters. Use cross-validation techniques to ensure that the model remains accurate on the new dataset. Cross-validation helps identify overfitting or other potential problems and optimizes the model's generalization ability by splitting the dataset and training and testing the model separately.
[0102] After the present invention integrates environmental condition data, the model can more accurately identify behavioral changes caused by changes in the external environment when judging emotional states. This makes the judgment of emotional states more accurate and reduces the possibility of misjudgment. Through continuous retraining and adjustment, the emotional behavior model can adapt to various dynamically changing environmental conditions. This adaptability allows the model to maintain its predictive validity and accuracy in different training environments, which is extremely important in virtual reality training. The model can provide personalized training plans based on the specific reactions and behavior patterns of each dog, as well as the specific conditions of its environment. This personalized training is more in line with the actual needs of each dog and improves the training effect.
[0103] Preferably, the expression of the emotional behavior model is:
[0104]
[0105] in, Represents emotional state; function Represents the emotional behavior model, Represents physiological characteristic data; Represents behavioral action data; Indicates environmental condition data; Represents model parameters, which are learned during the training process; function This is achieved through a combination of convolutional neural networks and long short-term memory networks.
[0106] The model that combines CNN and LSTM can predict emotional states very accurately because it can not only identify complex patterns in physiological and environmental data, but also understand how these patterns develop over time. Due to the addition of LSTM components, the model is able to dynamically understand and respond to changes in emotions, which is very important for training in dynamic and changing environments. The model is able to personalize the training plan based on the specific reactions and behavior patterns of each dog. For example, if the model recognizes that a dog has greater emotional fluctuations in a hot environment, the training plan can be adjusted to be conducted in a cooler environment.
[0107] In one example, a dog suddenly exhibited an anxious emotional state during training. By analyzing the sharp increase in its heart rate and the increase in temperature in the environment, combined with historical data, the model determined that this was due to the overheating of the environment rather than the training itself. Therefore, the training plan can be adjusted, such as changing the training time or environment, to adapt to the dog's emotional and physical needs and avoid similar emotional occurrences in the future.
[0108] In this way, the emotional behavior model not only enhances the understanding and prediction of dog emotions, but also provides a practical tool to dynamically adjust and optimize training strategies, ensuring the dog's well-being and efficiency in training.
[0109] Adjust the training difficulty and method in real time based on the emotional state data, and use the adaptive algorithm to update the training plan for the first time;
[0110] Based on real-time emotional state data collected, such as stress or excitement levels indicated by the dog's behavior and physiological indicators. The system analyzes this data to determine whether the difficulty of training needs to be adjusted, such as reducing stress or increasing stimulation. Based on emotional data, the training system automatically adjusts the complexity, duration or frequency of tasks. For example, if the data shows that the dog is in a highly tense state, the system will reduce the intensity of training or provide more rest time.
[0111] The adaptive algorithm optimizes the training plan by learning how the dog responds to different training tasks. It takes into account the deviation between the training effect and the expected goal and adjusts future training strategies based on the feedback. The algorithm analyzes historical and real-time data to identify which training methods are effective and which need improvement. Thereafter, the algorithm makes modification suggestions or automatically implements changes to the training plan.
[0112] Real-time adjustments to the training method allow the training program to be flexibly adapted to the immediate needs of each dog, increasing the level of personalization and effectiveness of the training. This approach ensures the well-being of the dog and optimal performance development during the training process.
[0113] The present invention continuously updates the training plan through an adaptive algorithm, which can continuously improve the training method and ensure the achievement of the training goal. This continuous optimization process helps to improve the training efficiency and the learning speed of the dog.
[0114] In one example, during a training session, a dog's heart rate is monitored to be consistently higher than normal, indicating that it may be anxious or overexcited. Based on this real-time data, the training system automatically switches to a less intense task or adds more positive incentives, such as food rewards or toys, to help the dog relax and stay focused.
[0115] After a series of training tasks, if the data shows that a specific task has not been effective in improving the dog's behavior, the adaptive algorithm will analyze the reasons for the failure, perhaps because the task is too complex or not motivating enough for the dog. The algorithm will then adjust the difficulty or variety of these tasks and implement new strategies in the next cycle of training.
[0116] Through this in-depth dynamic adjustment and adaptive strategy, the training plan can continue to evolve, better meet the development needs of dogs, and ultimately achieve a higher training success rate and animal welfare. This strategy is not only scientific and efficient, but also reflects respect and understanding for individual differences of animals.
[0117] Preferably, an adaptive algorithm is used to update the training plan for the first time:
[0118] Collect real-time feedback from training and clearly define rewards or punishments based on training goals and the dog’s feedback;
[0119] Using historical and real-time data to train an adaptive algorithm to learn a response strategy that identifies the type of action instructions that maximize long-term rewards in a given situation and periodically evaluates and adjusts it;
[0120] Implement training plans based on coping strategies and adjust training difficulty and methods in real time.
[0121] The present invention collects historical data and real-time feedback data. The historical data provides the dog's performance and reaction in previous training, and the real-time data captures the dog's behavior and physiological reaction in the current training.
[0122] Data preprocessing includes cleaning data, handling missing values, standardizing numerical values, and converting non-numerical data (such as label encoding). This ensures data quality and consistency, laying the foundation for effective training algorithms. Relevant features are selected from the collected data, including heart rate changes, behavioral patterns, and environmental factors such as temperature and noise levels. The purpose of feature selection is to identify the variables that have the greatest influence on predicting the dog's emotional state.
[0123] Depending on the nature of the problem and the characteristics of the data, different machine learning models can be chosen. Common choices include:
[0124] Decision tree: Suitable for classification problems, easy to understand and implement, and can clearly show the decision-making process.
[0125] Neural Networks: Good for complex pattern recognition, especially when there is a lot of data. Neural networks are able to learn and model non-linear relationships.
[0126] Reinforcement learning: Particularly suitable for situations where strategies need to be adjusted dynamically, such as real-time updates to training plans. Reinforcement learning optimizes the decision-making process through trial and error, learning how to achieve the best results in a specific environment.
[0127] Choose appropriate libraries and tools to implement the selected model, such as using Python's Scikit-learn library for decision tree training or TensorFlow, Keras for neural network development.
[0128] The adaptive algorithm first collects real-time feedback from training, which includes direct observation of the dog's behavioral response, such as the dog's compliance with commands, anxiety or excitement. These feedbacks help determine real-time rewards or punishments, which directly affect the dog's learning motivation and behavioral adjustments. For example, if a dog successfully completes a complex instruction, it will immediately receive a reward (such as food or toys) to reinforce the behavior; if it fails to respond, the trainer adopts a mild punishment method, such as repeating the voice command or reducing the reward.
[0129] Response strategies are defined based on training goals and the dog's behavioral feedback. For example, if the goal is to reduce the dog's anxious behavior, strategies include reducing the intensity of training or changing the training environment when signs of anxiety are observed. Strategies are specific to action instructions, how to apply them in specific situations. For example, if data shows that the dog is showing stress during a specific training action, the system can automatically adjust the instructions to reduce the frequency or difficulty of that action.
[0130] Use both quantitative and qualitative methods to evaluate the effectiveness of the strategy. Quantitative evaluation includes analyzing changes in the dog's behavioral and physiological data, and qualitative evaluation includes observation and feedback from the trainer. Regularly evaluate the results of the training to determine if further adjustments to the strategy are needed. This can be achieved by comparing the expected goals with the actual results to ensure the continuity and effectiveness of the training direction and methods.
[0131] Algorithms use historical and real-time data to train and optimize coping strategies. This includes analyzing which training methods are most effective in which situations and how to adjust strategies to maximize long-term rewards. This strategy is implemented through machine learning models, specifically reinforcement learning, which test different courses of action in various situations and learn which ones best achieve the training goals. For example, the adaptive algorithm recognizes that in a specific stressful situation, using positive reinforcement (such as play or extra rest time) is more effective than standard training and adjusts the training plan to include these elements.
[0132] Based on the output of the adaptive algorithm, the training program adjusts the difficulty and method of training in real time. This adjustment is based on a comprehensive assessment of the dog's current performance and historical performance, ensuring that the training difficulty is both challenging enough and beyond the dog's processing ability. If the algorithm analysis shows that a dog shows persistent difficulty with a specific task over consecutive days of training, the training difficulty may be reduced, or the teaching method may be changed, such as switching from physical guidance to more visual and sound cues.
[0133] The present invention adjusts and optimizes the training strategy in real time, and the adaptive algorithm ensures that the training method always adapts to the current needs and reactions of the dog, thereby accelerating the learning process and improving the overall training efficiency. The adaptive algorithm allows for personalized adjustments to the training plan for each dog, ensuring that each training is customized according to its unique needs and behavior patterns, greatly increasing the effectiveness of the training. Real-time adjustment of training strategies not only improves training results, but also helps maintain and improve the welfare of dogs, prevents excessive stress and discomfort, and makes the training process more humane and scientific.
[0134] Through in-depth analysis of these principles and effects, we can see the importance and advantages of using adaptive algorithms to update training plans in dynamically adapting virtual reality dog training programs. This method not only increases the flexibility and responsiveness of training, but also improves the personalization and precision of training plans, making training more scientific and effective.
[0135] Use virtual reality technology to simulate training scenarios to test the dog's ability to cope with the training plan. In the training scenario, collect the dog's behavioral action data and physiological characteristic data in real time, use the emotional behavior model to re-evaluate the dog's emotional state, adjust the complexity and type of virtual training according to the evaluation results, and update the training plan for the second time to ensure that the training plan matches the dog's current behavior and physiological state;
[0136] Through virtual reality (VR) technology, various training environments and situations are simulated to provide a controlled and variable environment for testing and strengthening dog behavior and coping abilities. VR environments can accurately reproduce specific situations, such as noisy streets, quiet parks, etc. These environments are designed to simulate various situations that may be encountered in the real world. VR technology provides a safe and repeatable way to evaluate and train dogs' reactions in different environments without actually being in a potentially dangerous or uncontrollable environment.
[0137] During VR training, the dog's behavioral data and physiological characteristics, such as heart rate, body temperature, and behavioral responses, are collected in real time. These data provide an empirical basis for assessing the dog's emotional state and coping ability. The collected data is analyzed in real time using an emotional behavior model to reassess the dog's emotional state. The model predicts the dog's emotional response, such as anxiety, excitement, or fear, based on previous training experience and real-time data.
[0138] Based on the evaluation results obtained from the emotional behavior model, the complexity and type of virtual training are adjusted. For example, if the model indicates that the dog is highly anxious in a specific scenario, the complexity of the training scenario can be reduced or replaced with a more soothing environment. Based on these adjustments and evaluation results, a second update of the training plan is performed to ensure that the training plan better matches the dog's current behavior and physiological state.
[0139] The present invention can optimize the training plan more accurately for the individual differences of dogs by evaluating and adjusting in real time in a controlled virtual environment, improving the training effect while reducing unnecessary stress and discomfort; VR training allows dogs to test their behavioral responses in a variety of environments and enhance their ability to cope with various situations. This not only helps to improve the dog's behavioral performance, but also enhances its ability to adapt to different living environments; by accurately controlling the training environment and adjusting the training plan in real time, VR technology makes the training process more efficient, reduces resource waste, and improves the return on training investment.
[0140] In one embodiment, a busy market environment is simulated to test the dog's reaction. If the dog shows excessive anxiety in this environment, the emotional behavior model will capture this emotion in real time and notify the trainer to lower the noise level of the environment or change the environment settings. Subsequently, the model re-evaluates the dog's status based on the changed behavior and physiological data to further adjust the training plan.
[0141] Through the above methods, VR training not only improves the quality of dog training, but also provides a safe and controlled environment to fine-tune and optimize training strategies.
[0142] Preferably, the complexity and type of virtual training are adjusted according to the evaluation results, including:
[0143] When dogs show excessive stress responses, reduce the difficulty of training, including: reducing environmental noise, reducing task complexity or providing more positive incentives;
[0144] When the dog appears relaxed and in control in certain training scenarios, increase the difficulty, introducing more complex tasks or small stressors to heighten the training challenge.
[0145] Based on the behavioral and physiological response data collected in real time, the virtual training system can determine the dog's stress level and adaptation in the current training environment. When the dog shows excessive stress response, the system will automatically take measures to reduce these stresses, otherwise, it will increase the challenge of training. By adjusting the training difficulty and environmental settings, each dog can receive training within its comfort level and ability. This personalized training method can effectively improve the training effect while avoiding unnecessary mental or physical burden.
[0146] For dogs that display high levels of stress, training difficulty can be adjusted by reducing environmental noise, reducing the complexity of the task, or adding positive incentives such as food rewards, toys, extra petting, and praise. These measures can help reduce the dog's anxiety and allow them to focus more on learning. When dogs appear to be more relaxed in certain training scenarios, the training can be made more challenging by introducing more complex tasks or adding small amounts of stressors such as slight environmental changes or shorter time limits, which can help to continuously improve their skills and adaptability.
[0147] The present invention can not only make training more effective, but also ensure the well-being of the dog by adjusting the training difficulty in real time. This method helps to avoid excessive stress and potential negative behaviors during training, ensuring that the dog learns in a safe and positive environment. Precise adjustment of training difficulty can help dogs better cope with various environments and situations, enhancing their behavioral adaptability and emotional management ability. This is an extremely important skill for pet dogs or working dogs.
[0148] In one embodiment, a virtual training scenario in which a dog performs a search task in a high-noise environment. If the dog's heart rate is monitored to rise significantly, indicating an obvious stress response, the training system can immediately reduce the background noise and provide the dog with some positive incentives, such as food rewards and encouraging words. At the same time, the complexity of the task can be appropriately reduced, such as reducing the number of targets to be found, until the dog's mood is re-evaluated and shown to be stable.
[0149] This dynamic and responsive training method not only ensures the effectiveness of the training, but also significantly improves the dog's training experience, avoiding unnecessary stress and potential behavioral problems. This method highlights the advantages of modern training technology in dynamic adaptation and personalized training.
[0150] Collect training performance data of the training plan after the second update, evaluate the training effect, and feed the training results back to the emotional behavior model to fine-tune the model and training plan.
[0151] Performance data includes the dog's performance in training, such as speed, accuracy, and reaction time of completing tasks. In addition, it also includes behavioral movement data and physiological characteristic data of the dog, such as heart rate and body temperature, which can provide direct evidence about the dog's emotional state and stress level. Data is collected in real time using various sensors and tracking technologies, while the built-in recording function of the virtual reality training system is used to record all relevant activities during the training process.
[0152] The training effect is evaluated using quantitative methods (such as statistical analysis) and qualitative observations (such as feedback from trainers). The evaluation will consider multiple dimensions, including task completion, behavioral improvement, and emotional stability. The evaluation of training effect is usually carried out after a certain period of training program completion to ensure that there is sufficient data to support the evaluation conclusion.
[0153] Feed the collected training performance data back to the emotional behavior model. This process includes recalibrating the model's existing parameters and prediction logic to ensure the accuracy and practicality of the model's predictions. Make necessary fine-tuning to the emotional behavior model based on the results of the training performance evaluation. This involves adjusting the data processing flow, algorithm parameters, or model structure.
[0154] By continuously collecting performance data and periodically evaluating and adjusting the training plan, the present invention can more accurately adapt the training plan to the specific needs of each dog and improve the overall effectiveness of the training. Continuous data feedback and model adjustment help improve the predictive accuracy of the emotional behavior model, enabling it to more effectively identify and respond to the dog's behavioral and emotional changes under different training conditions. Fine-tuning the training plan helps optimize resource utilization and ensure that training resources (such as time, equipment, and manpower) are effectively allocated to where they are most needed.
[0155] In one example, during a VR training session, a dog showed a significant stress response in a complex maze task, with performance data showing a prolonged task completion time and increased heart rate. These data were evaluated and fed back to the emotional behavior model, which then adjusted the difficulty level of the task and increased the frequency of positive incentives to reduce the dog's stress and improve its performance.
[0156] In this way, VR training programs are not only adapted to the individual needs of the dog, but also continue to improve the science and effectiveness of training through a precise data-driven approach.
[0157] Preferably, the training performance data includes: the implementation effect of the training plan in virtual reality and real environment after the second update, and the evaluation of the training effect is completed based on the training performance data, and the implementation effect includes: task completion rate, error rate, and completion speed of training tasks.
[0158] Training performance data includes specific performance indicators obtained from virtual reality and real environments. These data are key to measuring the effectiveness of training program adjustments and are the basis for further optimization.
[0159] Task Completion Rate: Measures the frequency and consistency with which a dog completes a given training task and is a direct indicator of training effectiveness.
[0160] Error Rate: Recording how often the dog makes errors while performing the training task can help identify problem areas in training.
[0161] Completion speed of training tasks: measures the time required for dogs to complete training tasks, reflecting the training efficiency and the dog's reaction speed.
[0162] Based on the data collected above, systematically evaluate the training effect, analyze the data change trend, identify successful training strategies and areas for improvement. Use statistical methods and data analysis tools to process and analyze performance data, such as calculating the average completion time, standard deviation of error rate, etc., to obtain clear performance indicators.
[0163] Based on the evaluation results, the emotional behavior model and training plan are adjusted. This involves modifying the complexity of the training task, adjusting the reward strategy, or changing the training frequency. Continuous data feedback and model fine-tuning form a dynamic training improvement loop, so that the training plan always adapts to the dog's latest performance and learning speed.
[0164] By accurately measuring and evaluating indicators, the present invention allows training plans to be more finely adjusted to match the specific needs and abilities of each dog, thereby improving the overall training effect and the dog's learning efficiency. The evaluation and feedback mechanism ensures that the training method can flexibly adapt to the performance and reaction of different dogs, and timely adjust the method to meet various challenges, thereby enhancing the dog's behavioral adaptability and emotional stability. Systematic performance evaluation helps to more effectively allocate training resources, such as time, personnel and technical inputs, to ensure that the use of each resource can produce the greatest benefit.
[0165] In one embodiment, a virtual reality training scenario, a dog is trained to perform a search task in a high noise environment. Initial performance data showed that the dog's task completion rate in the noisy environment was lower than expected and the completion speed was slow. By analyzing this data, the trainer decided to reduce the environmental noise or increase the dog's motivation by increasing positive incentives, while continuing to monitor the changed performance data to evaluate the effect of the adjustment. Not only does it ensure the scientific and systematic nature of the training, but also through continuous monitoring and adjustment, it effectively improves the quality of the training program and the welfare of the dog.
[0166] like Figure 2 As shown, a system for implementing the above-mentioned dynamic adaptive virtual reality dog training method comprises:
[0167] The sensor component is installed on the dog to collect the dog's physiological characteristics data, behavioral data and environmental conditions data; it is installed on the dog to collect physiological characteristics data (such as heart rate, body temperature), behavioral data (such as gait, jump) and environmental conditions data (such as temperature, humidity) in real time. Through the synchronous operation of multiple sensors, it is ensured that the collected data fully reflects the dog's behavior and external environmental conditions.
[0168] The data processing unit is connected to the sensor component and is used to integrate and synchronize the collected data and remove noise and outliers to obtain a cleaned state data set; it receives the data from the sensor component and is responsible for data integration, synchronization and cleaning, including removing noise and outliers to form an accurate and clean data set. Advanced data processing algorithms are used to ensure the quality of the data set and provide a reliable basis for subsequent analysis.
[0169] The emotional behavior model unit is configured with a processor and a memory for training an emotional behavior model based on annotated historical training data, wherein the emotional behavior model is used to obtain emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment; the unit is configured with a processor and a memory, and uses the emotional behavior model trained based on the annotated historical training data to analyze physiological and behavioral data and output emotional state data. The model uses machine learning techniques such as CNN and LSTM to comprehensively analyze behavioral and physiological data and accurately predict the emotional state of the dog.
[0170] The training adjustment unit is used to adjust the training difficulty and method in real time according to the emotional state data output by the emotional behavior model unit, and use the adaptive algorithm to update the training plan for the first time; the training difficulty and method are adjusted in real time according to the emotional state data, and the training plan is updated using the adaptive algorithm. This unit dynamically adjusts the training parameters to ensure that the training plan responds to the dog's emotional and behavioral feedback at any time and optimizes the training effect.
[0171] The virtual reality unit is used to simulate training scenarios to test the dog's ability to cope with the training plan. In the training scenario, the dog's behavioral action data and physiological characteristic data are collected in real time, and the emotional state of the dog is re-evaluated using the emotional behavior model. The complexity and type of virtual training are adjusted according to the evaluation results, and the training plan is updated for the second time to ensure that the training plan matches the dog's current behavior and physiological state; various training scenarios are simulated to test the dog's ability to cope with the training plan, and the complexity and type of training are adjusted according to the evaluation of the emotional behavior model. By creating a training environment close to reality, the dog's ability to cope can be greatly improved, and data can be collected for further analysis.
[0172] The performance evaluation unit is used to collect and evaluate the training performance data of the training plan after the second update, evaluate the training effect, and feed back the training results to the emotional behavior model unit to fine-tune the model and training plan. The performance data of the training plan after the second update is collected, the training effect is evaluated, and the training results are fed back to the emotional behavior model unit for fine-tuning. By continuously monitoring and evaluating the training results, necessary feedback is provided for the continuous optimization of the training plan and emotional model.
[0173] The system of the present invention can adjust the training method and difficulty according to the specific situation of each dog, so as to maximize the training effect. The training plan can respond to and adapt to the dog's behavior and emotional changes in real time, ensuring the comfort and effectiveness of the training process. Through the high-precision emotional behavior model, the system can accurately judge the dog's emotional state and optimize the training strategy.
[0174] For example, if the system detects that the dog is anxious due to a difficult task during training, it will automatically reduce the difficulty of the task or change the training method, such as increasing positive motivation or adjusting the environment settings, to reduce the dog's stress. Such dynamic adjustments help the dog better adapt to training and improve learning efficiency.
[0175] A storage medium stores a computer program, which, when executed by a processor, implements the steps of the dynamic adaptive virtual reality dog training method.
[0176] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented 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.
[0177] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0178] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0179] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0180] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0181] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0182] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0183] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0184] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims
1. A dynamic adaptive virtual reality dog training method, characterized in that: The following steps are involved: The sensor components installed on the dog collect physiological characteristic data, behavioral action data and environmental condition data, integrate and synchronize all the data, remove noise and outliers, and obtain a cleaned state data set; The emotional behavior model is trained based on the annotated historical training data. The emotional behavior model obtains emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment. The emotional behavior model adopts a combination of a convolutional neural network and a long short-term memory network. The spatial features of each time point extracted from the behavioral action and environmental conditions are processed by the convolutional neural network. The long short-term memory network tracks and analyzes changes in spatial features over time to capture long-term behavioral trends and emotional fluctuations; the emotional behavior model is used to predict the dog's emotional state in real time, understand the emotional drivers behind the dog's behavior, and support the development of personalized training plans based on each dog's specific reactions and behavioral patterns; Adjust the training difficulty and method in real time based on the emotional state data, and use the adaptive algorithm to update the training plan for the first time; Use virtual reality technology to simulate training scenarios to test the dog's ability to cope with the training plan. In the training scenario, collect the dog's behavioral action data and physiological characteristic data in real time, use the emotional behavior model to re-evaluate the dog's emotional state, adjust the complexity and type of virtual training according to the evaluation results, and update the training plan for the second time to ensure that the training plan matches the dog's current behavior and physiological state; Collect training performance data of the training plan after the second update, evaluate the training effect, and feed the training results back to the emotional behavior model to fine-tune the model and training plan.
2. The dynamic adaptive virtual reality dog training method according to claim 1, characterized in that: The physiological characteristic data include: heart rate and body temperature; the behavioral action data include: dog's movement speed and action mode; environmental condition data include: temperature and humidity; Integrate and synchronize the collected physiological characteristic data, behavioral action data, and environmental condition data so that all data come from the same time window.
3. The dynamic adaptive virtual reality dog training method according to claim 1, characterized in that: The emotional behavior model obtains emotional state data by analyzing physiological characteristic data and behavioral action data, and combines environmental condition data to improve the accuracy of emotional state judgment, including: The emotional behavior model predicts the current emotional state based on the physiological characteristic data and the behavioral action data, and outputs the emotional state data; Using environmental condition data as additional input features of the emotional behavior model so that the emotional behavior model understands potential external influences on the emotional state; retraining or adjusting the emotional behavior model based on new data containing environmental condition data; Cross-validation is used to ensure that the emotional behavior model remains accurate on new data, and the emotional state data is updated after adjusting the model parameters based on feedback to adapt to the influence of different environments.
4. The dynamic adaptive virtual reality dog training method according to claim 1, characterized in that: The expression of the emotional behavior model is: in, Represents emotional state; function Represents the emotional behavior model, Represents physiological characteristic data; Represents behavioral action data; Indicates environmental condition data; Represents model parameters, which are learned during the training process; function This is achieved through a combination of convolutional neural networks and long short-term memory networks.
5. The dynamic adaptive virtual reality dog training method according to claim 1, characterized in that: Use the adaptive algorithm to update the training plan for the first time: Collect real-time feedback from training and clearly define rewards or punishments based on training goals and the dog’s feedback; Using historical and real-time data to train an adaptive algorithm to learn a response strategy that identifies the type of action instructions that maximize long-term rewards in a given situation and periodically evaluates and adjusts it; Implement training plans based on coping strategies and adjust training difficulty and methods in real time.
6. The dynamic adaptive virtual reality dog training method according to claim 1, characterized in that: Adjust the complexity and type of virtual training based on the assessment results, including: When dogs show excessive stress responses, reduce the difficulty of training, including: reducing environmental noise, reducing task complexity or providing more positive incentives; When the dog appears relaxed and in control in certain training scenarios, increase the difficulty, introducing more complex tasks or small stressors to heighten the training challenge.
7. The dynamic adaptive virtual reality dog training method according to claim 1, characterized in that: The training performance data includes: the implementation effect of the training plan in the virtual reality and real environment after the second update, and the evaluation of the training effect is completed based on the training performance data. The implementation effect includes: task completion rate, error rate, and completion speed of the training task.
8. A system for implementing the dynamic adaptive virtual reality dog training method according to any one of claims 1 to 7, characterized in that: include: A sensor component is arranged on the dog and used to collect physiological characteristic data, behavioral action data and environmental condition data of the dog; A data processing unit connected to the sensor assembly, for integrating and synchronizing the collected data and removing noise and outliers to obtain a cleaned state data set; An emotional behavior model unit, configured with a processor and a memory for training an emotional behavior model based on labeled historical training data, wherein the emotional behavior model is used to obtain emotional state data by analyzing physiological characteristic data and behavioral action data, and to improve the accuracy of emotional state judgment by combining environmental condition data; A training adjustment unit, used to adjust the difficulty and method of training in real time according to the emotional state data output by the emotional behavior model unit, and to perform the first update of the training plan using an adaptive algorithm; A virtual reality unit is used to simulate training scenarios to test the dog's ability to cope with the training plan. In the training scenario, the dog's behavioral action data and physiological characteristic data are collected in real time, and the emotional state of the dog is re-evaluated using the emotional behavior model. The complexity and type of virtual training are adjusted according to the evaluation results, and the training plan is updated for the second time accordingly to ensure that the training plan matches the dog's current behavior and physiological state; The performance evaluation unit is used to collect and evaluate the training performance data of the training plan after the second update, evaluate the training effect, and feed back the training results to the emotional behavior model unit to fine-tune the model and training plan.
9. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the dynamic adaptive virtual reality dog training method described in any one of claims 1 to 7 are implemented.
