Environment adaptability self-optimization method and system of intelligent sensing robot

Abstract

The invention discloses an environment adaptability self-optimization method and system of an intelligent sensing robot, and relates to the technical field of robot control, and the method comprises the steps: obtaining a historical environment data set; acquiring a historical performance data set of the target robot; obtaining an optimal matching relationship; constructing a feature data set; constructing an environment-performance prediction model based on the feature data set; generating predicted performance data of the target robot; and optimizing and adjusting a strategy according to the predicted performance data. The technical problems that in the prior art, an intelligent sensing robot cannot dynamically and accurately predict self performance according to different complex and changeable environments, so that the environment adaptability of the intelligent sensing robot is insufficient, and diversified scene requirements are difficult to meet are solved, the performance of the intelligent sensing robot in the different environments is predicted, and the intelligent sensing robot is more intelligent. And optimization and adjustment can be carried out as required based on a prediction result, so that the method can better adapt to the environment, and the performance is improved.

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to an environmental adaptability self-optimization method and system for an intelligent perception robot. Background Art

[0002] In today's era of rapid technological development, intelligent perception robots are widely used in many fields, such as industrial production, logistics and transportation, medical services, and environmental monitoring. However, the intelligent perception robots in the existing technology face many severe challenges when dealing with complex and changeable actual working environments. On the one hand, traditional intelligent perception robots lack the ability to deeply perceive and analyze different environmental conditions, and it is difficult to accurately predict their performance in a specific environment. For example, when environmental factors such as temperature, humidity, and light intensity change, the robot's key performance indicators such as motion accuracy, response speed, sensor sensitivity, and energy consumption will be affected to varying degrees, but the existing technical means cannot predict these changes in advance, which may lead to problems such as low task execution efficiency, increased accuracy deviation, and even frequent failures in actual operation. On the other hand, even if the robot can obtain some environmental information, there is a lack of effective mechanisms to optimize and adjust its own performance based on this information.

[0003] The existing technology has the technical problem that intelligent perception robots are unable to dynamically and accurately predict their own performance according to complex and changing environments, resulting in insufficient environmental adaptability and difficulty in meeting the needs of diverse scenarios. Summary of the invention

[0004] The present application provides a method and system for self-optimization of environmental adaptability of an intelligent perception robot, which is used to solve the technical problem in the prior art that the intelligent perception robot is unable to dynamically and accurately predict its own performance according to complex and changeable environments, resulting in insufficient environmental adaptability and difficulty in meeting the needs of diverse scenarios.

[0005] In view of the above problems, the present application provides an environmental adaptability self-optimization method and system for an intelligent perception robot.

[0006] The first aspect of the present application provides an environmental adaptability self-optimization method for an intelligent perception robot, the method comprising:

[0007] Acquire a historical environment data set related to a target area; determine a target robot, interact with a robot control system, and acquire a historical performance data set of the target robot; perform slope verification on the historical performance data set, identify a fluctuation pattern of performance output, calculate a slope change rate of the historical performance data set, perform a correlation analysis on the slope change rate of the historical performance data set and a historical environment data set, and acquire an optimal matching relationship between the historical environment data set and the historical performance data set; construct a feature data set based on the matched historical performance data set and historical environment data set; construct an environment-performance prediction model based on the feature data set; input the real-time acquired environment data into the environment-performance prediction model to generate predicted performance data of the target robot;

[0008] The second aspect of the present application provides an environmental adaptability self-optimization system for an intelligent perception robot, the system comprising:

[0009] A historical environment data set acquisition module is used to acquire a historical environment data set related to a target area; a historical performance data set acquisition module is used to determine a target robot device, an interactive robot control system, and acquire a historical performance data set of the target robot device; an optimal matching relationship acquisition module is used to perform slope verification on the historical performance data set, identify the fluctuation pattern of the performance output, calculate the slope change rate of the historical performance data set, perform correlation analysis on the slope change rate of the historical performance data set and the historical environment data set, and acquire the optimal matching relationship between the historical environment data set and the historical performance data set; a feature data set construction module is used to construct a feature data set based on the matched historical performance data set and the historical environment data set; an environment-performance prediction model construction module is used to construct an environment-performance prediction model based on the feature data set; a predicted performance data generation module is used to input the real-time acquired environment data into the environment-performance prediction model to generate predicted performance data of the target robot; an optimization adjustment strategy module is used to perform optimization adjustment strategies based on the predicted performance data.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Acquire historical environment data sets related to the target area; determine the target robot, interact with the robot control system, and acquire the historical performance data set of the target robot; acquire the best matching relationship between the historical environment data set and the historical performance data set; construct a feature data set based on the matched historical performance data set and historical environment data set; construct an environment-performance prediction model; input the real-time environmental data into the environment-performance prediction model to generate predicted performance data of the target robot; optimize and adjust the strategy according to the predicted performance data. The technical effect of predicting the performance of the intelligent perception robot in different environments and optimizing and adjusting it as needed based on the prediction results is achieved, so that it can better adapt to the environment and improve performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 A schematic diagram of a process flow of a method for self-optimization of environmental adaptability of an intelligent perception robot provided in an embodiment of the present application;

[0014] Figure 2 Schematic diagram of the structure of the environmental adaptability self-optimization system of the intelligent perception robot provided in the embodiment of the present application.

[0015] Explanation of the accompanying drawings: historical environment data set acquisition module 10, historical performance data set acquisition module 20, best matching relationship acquisition module 30, feature data set construction module 40, environment-performance prediction model construction module 50, prediction performance data generation module 60, optimization adjustment strategy module 70. DETAILED DESCRIPTION

[0016] The present application provides a method and system for self-optimization of environmental adaptability of an intelligent perception robot, aiming to solve the technical problem in the prior art that the intelligent perception robot is unable to dynamically and accurately predict its own performance according to complex and changeable environments, resulting in insufficient environmental adaptability and difficulty in meeting the needs of diverse scenarios.

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] Embodiment 1, as Figure 1 As shown, the present application provides an environmental adaptability self-optimization method for an intelligent perception robot, the method comprising:

[0019] Step S100: Acquire a historical environment data set related to a target area.

[0020] Specifically, we must first clarify the scope of the target area and related environmental factors, which may include but are not limited to temperature, humidity, light intensity, topography, wind speed and direction, electromagnetic environment, obstacle distribution, and air quality. There are many ways to obtain data. For example, various types of sensors can be deployed in the target area, including temperature sensors, humidity sensors, light sensors, anemometers, electromagnetic sensors, etc. These sensors can monitor and record the corresponding environmental data in real time. Over time, historical environmental data sets are accumulated. At the same time, existing environmental monitoring databases, geographic information system (GIS) data, and data released by relevant research institutions can also be used to extract historical environmental information related to the target area and integrate them into a unified data set. These historical environmental data will provide a key basis for the subsequent analysis of the relationship between the performance of the robot in the area and the environment.

[0021] Step S200: Determine a target robot, interact with the robot control system, and obtain a historical performance data set of the target robot.

[0022] Specifically, the target robot is determined from among many robots. This target robot is a specific individual that needs to be self-optimized for environmental adaptability. After being determined, it interacts with the control system of the robot through specific communication interfaces and protocols. The robot control system is the core part of managing and monitoring the operating status and various parameters of the robot. During the interaction process, the historical performance data set of the target robot is obtained from the control system. These historical performance data cover various operating indicators of the robot in different time periods in the past, such as the robot's movement speed, the accuracy of the movement trajectory, the success rate of executing tasks, energy consumption, the movement angle and strength of each joint, the detection accuracy of the sensor, and the response time to different environmental stimuli. These data comprehensively reflect the actual performance of the robot in the previous work process, and provide key basic data for the subsequent analysis of the relationship between robot performance and the environment.

[0023] Step S300: perform slope verification on the historical performance data set, identify the fluctuation pattern of the performance output, calculate the slope change rate of the historical performance data set, perform correlation analysis on the slope change rate of the historical performance data set and the historical environment data set, and obtain the best matching relationship between the historical environment data set and the historical performance data set.

[0024] Specifically, the slope of the historical performance data set is first verified. The data is first cleaned and standardized to remove outliers and irregular data points to ensure data quality. Then, the adjacent performance data points after preprocessing are extracted for difference calculation to obtain the slope value at each time point, so as to observe the change trend of the performance data at different time points and identify the fluctuation pattern of the performance output. Then, based on the slope value at each time point, the change amplitude of the slope between adjacent time points is calculated to obtain the slope change rate of the historical performance data set. After that, the slope change rate of the historical performance data set is correlated with the historical environment data set, and each environmental feature in the historical environment data set is traversed, and the correlation is calculated with the slope change rate respectively to obtain the correlation score between each pair of environmental features and the slope change rate. Finally, these correlation scores are compared with the pre-set correlation threshold, and the combination with the correlation score greater than or equal to the threshold is screened out, so as to obtain the best matching relationship between the historical environment data set and the historical performance data set, so as to determine the environmental factors that have a significant impact on the performance of the robot.

[0025] Step S400: constructing a feature data set based on the matched historical performance data set and historical environment data set.

[0026] Specifically, a feature dataset is constructed based on the matched historical performance dataset and historical environment dataset. First, the two datasets are time-aligned to ensure the consistency of the data in the time dimension so that the environmental features and performance data can be accurately associated in the future. Then, the time-aligned data is subjected to sliding window processing. By setting an appropriate window size, the environmental features and corresponding performance data in each period are extracted to form time series feature data. In order to reduce the complexity of the data and improve the efficiency of the subsequent model, the time series feature data is subjected to dimensionality reduction processing, and key features can be extracted using methods such as principal component analysis. Finally, the time series feature data after dimensionality reduction processing is subjected to feature combination, and different key features are combined according to certain rules to generate a feature dataset containing key information of the environment and performance, providing a suitable data basis for building an environment-performance prediction model.

[0027] Step S500: constructing an environment-performance prediction model based on the feature data set.

[0028] Specifically, an environment-performance prediction model is constructed based on the feature data set. First, the feature data set is divided into a training set and a validation set. Reasonable division ratios are conducive to effective training and evaluation of the model. Then, according to the characteristics of the feature data set and the complexity of the relationship between the robot's performance and the environment, a long short-term memory network (LSTM) is selected. The model is initialized and trained using the training set by setting appropriate hyperparameters, such as learning rate and number of hidden layer nodes. During the training process, the model continuously learns the relationship between environmental features and performance data in the feature data set, and adjusts its own parameters to minimize the prediction error. At the same time, the validation set is used to evaluate the model in the training process, and the loss value on the validation set, such as mean square error (MSE) and other indicators, is calculated to determine whether the model is overfitting or underfitting. The model is adjusted according to the evaluation results, such as adjusting hyperparameters, adding regularization terms, or changing the model structure, and finally an environment-performance prediction model that can accurately predict the performance of the robot in different environments is constructed.

[0029] Step S600: Input the real-time acquired environmental data into the environment-performance prediction model to generate predicted performance data of the target robot.

[0030] Specifically, ensure that the environmental data acquired in real time meets the input requirements of the environment-performance prediction model. This requires some preprocessing operations on the real-time environmental data, such as data cleaning and standardization, so that it is consistent with the feature data set used to build the model in terms of data format and feature distribution. Then the preprocessed real-time environmental data is input into the environment-performance prediction model. The model analyzes and processes the input real-time environmental data based on the relationship pattern between the environmental features and performance data that have been learned internally. Through a series of complex calculations and reasoning processes, the predicted performance data of the target robot is finally generated. These predicted performance data can reflect the possible working performance of the robot under the current real-time environmental conditions, and provide a key basis for subsequent optimization and adjustment strategies.

[0031] Step S700: Optimizing and adjusting strategies according to the predicted performance data.

[0032] Specifically, the predicted performance data is compared with the robot's predetermined target performance, and the deviation between the two is calculated. If the deviation is greater than the preset allowable error threshold, it means that the robot's current predicted performance does not meet expectations and an optimization adjustment strategy is required. At this time, the PID control algorithm can be used to adjust the performance deviation. The PID control algorithm generates an optimization adjustment signal based on the size of the deviation, the accumulation of the deviation, and the rate of change of the deviation through three links: proportional, integral, and differential. Based on these optimization adjustment signals, a specific optimization adjustment strategy is determined, such as adjusting the robot's movement speed, the sensitivity of the sensor, the operating force of the actuator, etc., so that the robot's performance in the current environment is closer to the predetermined target performance. If the deviation is less than the preset allowable error threshold, it means that the robot's performance basically meets the requirements and no optimization adjustment strategy is required.

[0033] In a possible implementation, step S300 further includes:

[0034] Step S310: performing data cleaning and standardization processing on the historical performance data to obtain a preprocessed historical performance data set.

[0035] Step S320: extract adjacent performance data points of the preprocessed historical performance data set, perform difference calculation, and obtain the slope value at each time point.

[0036] Step S330: Based on the slope value at each time point, calculate the slope change amplitude between adjacent time points to obtain the slope change rate.

[0037] Specifically, data cleaning and standardization of historical performance data are performed to improve data quality and make it more suitable for subsequent analysis. The data cleaning process mainly identifies and removes outliers and erroneous data in the data. For example, if there is data that deviates significantly from other data points, it may be due to sensor failure or other abnormal conditions, and these data will be marked and deleted. Standardization is to convert data of different magnitudes and units into a unified standard form. This helps to eliminate analytical biases that may be caused by magnitude differences between data. The historical performance data is processed through appropriate standardization methods, such as Z-score standardization, to finally obtain a preprocessed historical performance data set.

[0038] Adjacent performance data points are extracted from the preprocessed historical performance data set for difference calculation to obtain the slope value at each time point. For two adjacent data points in the time series, their difference is calculated, and this difference represents the slope value at that time point. By performing such calculations on the entire preprocessed historical performance data set in chronological order, a series of slope values ​​reflecting the changing trend of performance data at different time points can be obtained.

[0039] Based on the slope value at each time point, the slope change between adjacent time points is calculated to obtain the slope change rate. Assume that the slope value at time point t is S t , the slope value at time point t+1 is S t+1 , then the slope change rate can be expressed by (S t+1 -S t ) calculation. By performing such calculations on the slope values ​​of the entire time series, a slope change rate sequence can be obtained that reflects the slope change trend of the historical performance data set at different time points. This slope change rate is of great significance for analyzing the fluctuation pattern of performance data and subsequent correlation analysis with environmental data sets.

[0040] In a possible implementation, step S300 further includes:

[0041] Step S340: traverse the environmental features in the historical environmental data set and perform correlation calculation with the slope change rate to obtain a correlation score between each pair of environmental features and the slope change rate.

[0042] Step S350: Compare the correlation score with a correlation threshold, and filter to obtain the best matching relationship between the historical environment data set and the historical performance data set.

[0043] Specifically, each environmental feature in the historical environmental dataset is traversed and the correlation with the slope change rate is calculated. Correlation calculation is a statistical analysis method used to measure the degree of association between two variables. Common correlation calculation methods such as the Pearson correlation coefficient are used here. For each environmental feature in the historical environmental dataset, the correlation score between it and the slope change rate is calculated. This correlation score can quantify the degree of linear relationship between the environmental feature and the slope change rate. If the correlation score is close to 1 or -1, it means that there is a strong correlation between the two; if the correlation score is close to 0, it means that there is almost no linear correlation between the two.

[0044] The calculated correlation score between each pair of environmental features and the slope change rate is compared with the pre-set correlation threshold. The correlation threshold is a predetermined standard used to determine whether the correlation between the environmental feature and the slope change rate is significant enough. If the correlation score is greater than or equal to this threshold, it is considered that there is a significant correlation between the environmental feature and the slope change rate, and the corresponding environmental feature and slope change rate combination is screened out. These screened combinations constitute the best match between the historical environmental data set and the historical performance data set. In this way, it is possible to determine which environmental features have the most significant impact on the robot's performance, providing an important basis for subsequent analysis and model building.

[0045] In a possible implementation, step S400 further includes:

[0046] Step S410: performing time alignment processing on the historical performance data set and the historical environment data set.

[0047] Step S420: Perform sliding window processing on the time-aligned data to extract environmental characteristics and corresponding performance data within each period of time to form time series feature data.

[0048] Step S430: performing dimensionality reduction processing on the time series feature data.

[0049] Step S440: performing feature combination on the time series feature data after the dimensionality reduction process to generate a feature data set.

[0050] Specifically, since the historical performance data set and the historical environment data set may be collected at different time points and at different frequencies, they need to be time-aligned first. This step is to ensure that the data in the two data sets can accurately correspond to each other in the time dimension, so as to facilitate subsequent effective analysis and processing. For example, if the historical performance data is recorded every 5 minutes, and the historical environment data is recorded every 10 minutes, then they need to be unified to the same time interval through appropriate methods (such as data interpolation or data aggregation) so that there are corresponding environment data and performance data at each time point, thus providing a basis for subsequent joint analysis.

[0051] After completing the time alignment, the data is processed by sliding windows. By setting a fixed-size time window, sliding along the time series in sequence, and extracting the environmental characteristics and corresponding performance data within the time period at each window position. For example, a sliding window with a length of 30 minutes is set, starting from the starting time, and moving for 10 minutes each time. In each window, environmental characteristic data such as temperature, humidity, and light intensity, as well as performance data such as the robot's movement speed and operation accuracy are collected. Combining these data together forms a data set with time series characteristics, which can reflect the relationship between the environment and performance in different time periods.

[0052] The time series feature data obtained after sliding window processing may have high dimensions and contain a lot of redundant information, which will not only increase the computational burden, but also may affect the training efficiency and accuracy of subsequent models. Therefore, it is necessary to reduce the dimensionality of the data. Principal component analysis (PCA) is used to map the high-dimensional time series feature data into a low-dimensional space, retain the main feature information of the data, and remove those minor features that have little impact on the robot performance, thereby simplifying the data structure, improving the efficiency and effect of data processing, and reducing the risk of overfitting, making the data more suitable for subsequent model construction.

[0053] After the dimensionality reduction process, the obtained low-dimensional time series feature data is subjected to feature combination. This step is to recombine different environmental features and performance features according to certain rules and logic to form a more representative and comprehensive feature vector. For example, the temperature and humidity features after dimensionality reduction can be linearly or nonlinearly combined with the robot's movement speed and operation accuracy features to construct new feature variables. These new feature variables can more comprehensively reflect the complex relationship between environment and performance. Through such feature combination operations, a feature data set that can effectively describe the performance of the robot under different environmental conditions is finally generated, providing high-quality data input for the subsequent construction of the environment-performance prediction model.

[0054] In a possible implementation, step S500 further includes:

[0055] Step S510: Divide the feature data set into a training set and a validation set.

[0056] Step S520: Using the training set and a long short-term memory network, construct the environment-performance prediction model.

[0057] Step S530: Use the validation set to evaluate and optimize the environment-performance prediction model.

[0058] Specifically, the acquired feature data set is divided into a training set and a validation set according to a certain ratio. The purpose of the division is to use the training set to train the model during the model construction process so that it can learn the characteristics and rules in the data, and to use the validation set to independently evaluate the performance of the model to avoid problems such as overfitting. The common division ratio is 70% to 80% of the data as the training set, and the remaining 20% ​​to 30% as the validation set. For example, if the feature data set contains 1,000 data records, 700 may be selected as the training set for model parameter learning and fitting, and the remaining 300 as the validation set to test the generalization ability and accuracy of the model on unseen data.

[0059] The environment-performance prediction model is constructed by using the divided training set and the long short-term memory network (LSTM). LSTM is a special recurrent neural network structure with memory units and gating mechanisms, which can effectively process long-term dependencies in time series data. When building the model, it is necessary to first determine the structure of the LSTM network, including the number of nodes in the input layer, hidden layer, and output layer, as well as hyperparameters such as the number of hidden layers. Then, the feature data in the training set is input into the LSTM network, the output results are calculated through forward propagation, and compared with the actual performance data to calculate the loss function (such as mean square error loss). Then, the back-propagation algorithm is used to update the weight parameters of the network according to the gradient of the loss function. This process is continuously iterated so that the model can gradually learn the complex mapping relationship between environmental features and robot performance, thereby accurately predicting the performance of the robot in different environments.

[0060] Use the validation set to evaluate and optimize the built environment-performance prediction model. Input the environmental feature data in the validation set into the model to obtain the predicted performance data, which is then compared with the actual performance data in the validation set to calculate evaluation indicators, such as root mean square error (RMSE), mean absolute error (MAE), etc. These indicators can intuitively reflect the degree of deviation between the model prediction value and the actual value. If the evaluation indicators show that the performance of the model is not ideal, such as the RMSE value is too large, it means that the model may have problems such as overfitting or underfitting, and the model needs to be optimized. The optimization method includes adjusting the hyperparameters of the LSTM network, such as increasing or decreasing the number of hidden layer nodes, changing the learning rate, adding regularization terms, etc., and then re-using the training set for training and evaluating it again with the validation set. Repeat this process until the performance of the model on the validation set reaches a satisfactory level, thereby obtaining an effective model that can accurately predict the robot environment-performance relationship.

[0061] In a possible implementation, step S700 further includes:

[0062] Step S710: Compare the predicted performance data with the preset target performance data of the robot, calculate the deviation between the two, and if the deviation is greater than the preset allowable error threshold, perform an optimization adjustment strategy; if the deviation is less than the preset allowable error threshold, do not perform an optimization adjustment strategy.

[0063] Specifically, first, obtain the predicted performance data generated by the environment-performance prediction model, which represents the expected performance of the robot under the current environmental conditions based on historical data and model learning. At the same time, clarify the robot's pre-set target performance, which is the performance indicator that the robot should achieve under ideal conditions, covering multiple key aspects such as motion accuracy, speed stability, task completion success rate, energy consumption level, etc., and each aspect has a corresponding quantitative standard.

[0064] Next, through a specific calculation method, the predicted performance data and the target performance data are compared one by one on each key performance indicator to calculate the deviation between the two. For example, if the target performance requires the robot's motion accuracy to be within a certain range, and the predicted performance data shows that the actual possible motion accuracy deviates from this range, the specific degree of deviation can be calculated through a mathematical formula. Then, the calculated deviation is compared with the preset allowable error threshold. This threshold is a critical value pre-set based on the actual application requirements of the robot, historical experience, and the acceptable range of performance fluctuations.

[0065] If the deviation is greater than the preset allowable error threshold, this indicates that the robot's current predicted performance is significantly different from the expected target performance and may not meet the requirements of the actual work. At this time, it is necessary to start the optimization adjustment strategy to improve the robot's performance and make it closer to the predetermined goal. On the contrary, if the deviation is less than the preset allowable error threshold, it means that the robot's predicted performance is within an acceptable range and can basically meet the requirements of the current task. Therefore, no additional optimization adjustment operations are required, thereby avoiding unnecessary resource consumption and increased system complexity, allowing the robot to continue to operate efficiently.

[0066] In a possible implementation, step S710 further includes:

[0067] Step S711: Use the PID control algorithm to adjust the performance deviation and generate an optimization adjustment signal.

[0068] Step S712: generating the optimization adjustment strategy based on the optimization adjustment signal.

[0069] Specifically, when it is detected that there is a deviation between the predicted performance data and the robot's predetermined target performance and the deviation is greater than the preset allowable error threshold, the PID control algorithm is activated for adjustment. First, the proportional link in the PID algorithm generates a control signal component in proportion to the size of the deviation. The larger the deviation, the stronger the effect of this component, which can quickly respond to large deviations and adjust the robot in the direction of reducing the deviation. Then, the integral link integrates the deviation. As the influence of the accumulated deviation over time, it can eliminate the steady-state error caused by factors such as system inertia, ensuring that the robot can stably approach the target performance in long-term operation, and can continue to fine-tune even if the deviation is small. At the same time, the differential link predicts the development trend of the deviation in advance by calculating the rate of change of the deviation. When the deviation changes rapidly, a larger control signal change is generated, thereby suppressing the possible overshoot phenomenon and making the adjustment process smoother and more accurate. Through the coordinated operation of these three links, the PID algorithm comprehensively generates a series of continuous optimization adjustment signals, which will serve as the key basis for the subsequent generation of specific optimization adjustment strategies, guiding the robot to gradually correct the performance deviation to better adapt to the environment and achieve the expected performance goals.

[0070] After obtaining the optimization adjustment signal generated by the PID control algorithm, specific optimization adjustment strategies will be formulated based on these signals. If the optimization adjustment signal indicates that the robot's movement speed needs to be changed, the adjustment value of the motor drive voltage or current will be accurately calculated according to the strength and direction of the signal, thereby achieving precise control of the robot's movement speed and making it more in line with the target performance requirements. For the sensor system, if the signal indicates that the perception accuracy needs to be improved, the sensitivity parameters of the sensor will be optimized accordingly, such as increasing the gain value of the light sensor or adjusting the threshold range of the pressure sensor to ensure that it can more keenly capture subtle changes in the environment and provide more accurate data support for the robot's decision-making. In terms of the actuator, the joint motion parameters of the robot arm may be reset based on the optimization adjustment signal, such as accurately adjusting the rotation angle of the joint, the size of the control force, and the time series of the movement, so as to improve the accuracy and stability of the execution of the task. At the same time, we will also consider combining these adjustment measures with the robot's overall task planning and path planning. For example, according to environmental conditions and performance deviations, a more efficient and stable motion path can be re-planned to avoid complex areas that may affect performance, or the time and resources for task execution can be reasonably allocated to ensure that the robot can make full use of these optimization and adjustment strategies during the execution of the task, comprehensively improve its own performance, achieve a high degree of fit with the predetermined target performance, and better adapt to the complex and changing working environment.

[0071] Embodiment 2, based on the same inventive concept as the environmental adaptability self-optimization method of the intelligent perception robot in the above embodiment, Figure 2As shown, the present application provides an environmental adaptability self-optimization system for an intelligent perception robot, and the system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0072] The historical environment data set acquisition module 10 is used to acquire the historical environment data set related to the target area.

[0073] The historical performance data set acquisition module 20 is used to determine a target robot device, interact with the robot control system, and acquire a historical performance data set of the target robot device.

[0074] The best matching relationship acquisition module 30 is used to perform slope verification on the historical performance data set, identify the fluctuation pattern of the performance output, calculate the slope change rate of the historical performance data set, perform correlation analysis on the slope change rate of the historical performance data set and the historical environment data set, and obtain the best matching relationship between the historical environment data set and the historical performance data set.

[0075] The feature data set construction module 40 constructs a feature data set based on the matched historical performance data set and historical environment data set.

[0076] The environment-performance prediction model building module 50 builds an environment-performance prediction model based on the feature data set.

[0077] The predicted performance data generating module 60 is used to input the real-time acquired environmental data into the environment-performance prediction model to generate the predicted performance data of the target robot.

[0078] The optimization and adjustment strategy module 70 is used to perform optimization and adjustment strategies according to the predicted performance data.

[0079] Furthermore, the best matching relationship acquisition module 30 also includes:

[0080] The preprocessing historical performance data set acquisition unit is used to perform data cleaning and standardization processing on the historical performance data to obtain a preprocessing historical performance data set.

[0081] A slope value acquisition unit is used to extract adjacent performance data points of the preprocessed historical performance data set to perform difference calculation and obtain the slope value at each time point.

[0082] A change amplitude calculation unit is used to calculate the change amplitude of the slope between adjacent time points based on the slope value at each time point, and obtain the slope change rate.

[0083] Furthermore, the best matching relationship acquisition module 30 also includes:

[0084] A correlation score acquisition unit is used to traverse the environmental features in the historical environmental data set and perform correlation calculations with the slope change rate to obtain a correlation score between each pair of environmental features and the slope change rate.

[0085] A correlation threshold comparison unit is used to compare the correlation score with a correlation threshold, and screen and obtain the best matching relationship between the historical environment data set and the historical performance data set.

[0086] Furthermore, the feature data set construction module 40 also includes:

[0087] A time alignment processing unit is used to perform time alignment processing on the historical performance data set and the historical environment data set.

[0088] The time series feature data acquisition unit is used to perform sliding window processing on the time-aligned data, extract the environmental characteristics and corresponding performance data in each period of time, and form time series feature data.

[0089] A dimensionality reduction processing unit is used to perform dimensionality reduction processing on the time series feature data.

[0090] The feature data set generating unit is used to perform feature combination on the time series feature data after the dimensionality reduction process to generate a feature data set.

[0091] Furthermore, the environment-performance prediction model building module 50 also includes:

[0092] The feature data set division unit is used to divide the feature data set into a training set and a validation set.

[0093] An environment-performance prediction model construction unit is used to adopt the training set and use a long short-term memory network to construct the environment-performance prediction model.

[0094] An environment-performance prediction model optimization unit is used to evaluate and optimize the environment-performance prediction model using the validation set.

[0095] Furthermore, the optimization and adjustment strategy module 70 also includes:

[0096] The performance data comparison unit is used to compare the predicted performance data with the preset target performance data of the robot, calculate the deviation between the two, and if the deviation is greater than the preset allowable error threshold, perform an optimization adjustment strategy; if the deviation is less than the preset allowable error threshold, do not perform an optimization adjustment strategy.

[0097] Furthermore, the performance data comparison unit also includes:

[0098] The optimization adjustment signal generating unit is used to adjust the performance deviation by using a PID control algorithm to generate an optimization adjustment signal.

[0099] An optimization adjustment strategy generating unit is configured to generate the optimization adjustment strategy based on the optimization adjustment signal.

[0100] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0101] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0102] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for self-optimization of environmental adaptability of an intelligent perception robot, characterized in that: The method comprises: Acquire historical environmental datasets relevant to the target area; Determine a target robot, interact with the robot control system, and obtain a historical performance data set of the target robot; Performing slope verification on the historical performance data set, identifying the fluctuation pattern of the performance output, calculating the slope change rate of the historical performance data set, performing correlation analysis on the slope change rate of the historical performance data set and the historical environment data set, and obtaining the best matching relationship between the historical environment data set and the historical performance data set; Based on the matched historical performance dataset and historical environment dataset, a feature dataset is constructed; Based on the feature data set, construct an environment-performance prediction model; Inputting the real-time acquired environmental data into the environment-performance prediction model to generate predicted performance data of the target robot; An optimization adjustment strategy is performed based on the predicted performance data.

2. The environmental adaptability self-optimization method of the intelligent perception robot according to claim 1, characterized in that: Performing slope verification on the historical performance data, identifying the fluctuation pattern of the performance output, and calculating the slope change rate of the historical performance data, including: Performing data cleaning and standardization processing on the historical performance data to obtain a preprocessed historical performance data set; Extract adjacent performance data points of the preprocessed historical performance data set to perform difference calculation to obtain the slope value at each time point; Based on the slope value at each time point, the slope change amplitude between adjacent time points is calculated to obtain the slope change rate.

3. The environmental adaptability self-optimization method of the intelligent perception robot according to claim 2, characterized in that: Performing a correlation analysis on the slope change rate of the historical performance data set and the historical environment data set to obtain the best matching relationship between the historical environment data set and the historical performance data set includes: Traversing the environmental features in the historical environmental data set and performing correlation calculation with the slope change rate, and obtaining a correlation score between each pair of environmental features and the slope change rate; The correlation score is compared with a correlation threshold to filter and obtain the best matching relationship between the historical environment data set and the historical performance data set.

4. The environmental adaptability self-optimization method of the intelligent perception robot according to claim 1, characterized in that: Based on the matched historical performance dataset and historical environment dataset, a feature dataset is constructed, including: Performing time alignment processing on the historical performance data set and the historical environment data set; Perform sliding window processing on the time-aligned data to extract the environmental characteristics and corresponding performance data within each period of time to form time series feature data; Performing dimensionality reduction processing on the time series feature data; The time series feature data after dimensionality reduction is combined to generate a feature data set.

5. The environmental adaptability self-optimization method of the intelligent perception robot according to claim 1, characterized in that: Based on the feature data set, an environment-performance prediction model is constructed, including: Dividing the feature data set into a training set and a validation set; Using the training set and a long short-term memory network, construct the environment-performance prediction model; The environment-performance prediction model is evaluated and optimized using the validation set.

6. The method for self-optimizing environmental adaptability of an intelligent perception robot according to claim 1, characterized in that: Optimizing and adjusting strategies based on the predicted performance data include: The predicted performance data is compared with the preset target performance data of the robot, and the deviation between the two is calculated. If the deviation is greater than the preset allowable error threshold, an optimization adjustment strategy is performed; if the deviation is less than the preset allowable error threshold, no optimization adjustment strategy is performed.

7. The environmental adaptability self-optimization method of the intelligent perception robot according to claim 6, characterized in that: If the deviation is greater than the preset allowable error threshold, an optimization adjustment strategy is performed, including: Use PID control algorithm to adjust performance deviation and generate optimization adjustment signal; The optimization adjustment strategy is generated based on the optimization adjustment signal.

8. The environmental adaptability self-optimization system of the intelligent perception robot is characterized by: The system is used to implement the environmental adaptability self-optimization method of the intelligent perception robot according to any one of claims 1 to 7, and the system comprises: A historical environment data set acquisition module, used to acquire a historical environment data set related to a target area; A historical performance data set acquisition module is used to determine a target robot device, interact with the robot control system, and acquire a historical performance data set of the target robot device; The best matching relationship acquisition module is used to perform slope verification on the historical performance data set, identify the fluctuation mode of the performance output, calculate the slope change rate of the historical performance data set, perform correlation analysis on the slope change rate of the historical performance data set and the historical environment data set, and obtain the best matching relationship between the historical environment data set and the historical performance data set; A feature data set construction module constructs a feature data set based on the matched historical performance data set and historical environment data set; An environment-performance prediction model building module, which builds an environment-performance prediction model based on the feature data set; A predicted performance data generation module, used to input the real-time acquired environmental data into the environment-performance prediction model to generate predicted performance data of the target robot; The optimization and adjustment strategy module is used to perform optimization and adjustment strategies according to the predicted performance data.

Images