A method, system, device, and medium for optimizing the wear resistance of a robot

Abstract

This invention provides a method, system, device, and medium for optimizing the wear resistance of a robot, comprising: collecting environmental perception data and initial wear information of the robot; predicting the wear amount of the robot using an extreme random tree algorithm based on the environmental perception data and the initial wear information, thereby obtaining wear amount prediction information of the robot; generating a dynamic control strategy for the robot using a control decision algorithm based on the wear amount prediction information of the robot; and adjusting the motion control parameters of the robot according to the dynamic control strategy. This application, by employing the extreme random tree algorithm, can process high-dimensional, nonlinear environmental perception data, thereby accurately predicting the wear amount of the robot; the dynamic control strategy generated by the control decision algorithm enables the robot to autonomously adapt to complex working condition changes, thereby improving the robot's wear resistance and energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of robotics, and more specifically to a method, system, device, and medium for optimizing the wear resistance of robots. Background Technology

[0002] In recent years, the widespread application of robots in industrial manufacturing, medical surgery, rescue and exploration has placed higher demands on their reliability and service life. Wear, as one of the main causes of robot performance degradation, not only affects motion accuracy and stability, but may also lead to premature failure of critical components (such as joints, gears, and bearings), increasing maintenance costs and downtime. Therefore, optimizing the wear resistance of robots is of great significance for improving their overall performance and reducing operating costs.

[0003] Currently, the main methods for combating robot wear include material improvement, surface treatment, and dynamic adjustment of joint torque. Among these, dynamic adjustment of joint torque, which mitigates wear by adjusting robot motion parameters (such as speed, acceleration, and contact force) in real time, is one of the most widely used technologies. However, existing dynamic adjustment methods do not fully integrate environmental perception data (such as friction, temperature, and humidity) for dynamic adjustment, leading to a mismatch between the control strategy and actual working conditions. For example, when service robots transition from smooth floors to carpets, traditional methods cannot adjust their gait in real time, potentially resulting in localized wear or unstable movement. Therefore, due to the inability to dynamically perceive and respond to environmental changes, traditional methods have limited effectiveness in complex scenarios. Furthermore, as robots become more prevalent in more complex scenarios (such as home service and field exploration), higher demands are placed on their adaptability.

[0004] In summary, traditional robot wear resistance methods lack environmental awareness and are unable to adapt to complex working conditions, resulting in low wear resistance and energy efficiency. Summary of the Invention

[0005] To address the problem that traditional robot wear resistance methods lack environmental awareness and struggle to adapt to complex working conditions, resulting in low wear resistance and energy efficiency, this invention proposes a method for optimizing robot wear resistance, comprising:

[0006] Collect environmental perception data and initial wear information of the robot;

[0007] Based on the environmental perception data and the initial wear information, the wear amount of the robot is predicted using an extreme random tree algorithm to obtain the wear amount prediction information of the robot.

[0008] Based on the wear prediction information of the robot, a dynamic control strategy for the robot is generated using a control decision algorithm;

[0009] The motion control parameters of the robot are adjusted according to the dynamic control strategy.

[0010] Optionally, the step of predicting the wear amount of the robot using an extreme random tree algorithm based on the environmental perception data and the initial wear information to obtain the robot's wear amount prediction information includes:

[0011] Feature extraction is performed based on the environmental perception data to obtain the robot's environmental feature data;

[0012] Based on the initial wear information, feature extraction is performed to obtain the wear feature data of the robot;

[0013] The environmental feature data and the wear feature data are cross-referenced to obtain the associated feature data of the robot;

[0014] Based on the associated feature data, the wear amount of the robot is predicted using an extreme random tree algorithm to obtain the wear amount prediction information of the robot.

[0015] Optionally, the step of performing feature cross-referencing on the environmental feature data and the wear feature data to obtain the associated feature data of the robot includes:

[0016] The environmental feature data and the wear feature data are time-aligned using a time alignment algorithm to obtain time-aligned environmental feature data and wear feature data.

[0017] A feature crossover algorithm is used to perform feature interaction between the time-aligned environmental feature data and the wear feature data to obtain interactive feature data.

[0018] The interaction feature data is filtered to obtain the robot's associated feature data.

[0019] Optionally, the step of predicting the wear amount of the robot using an extreme random tree algorithm based on the associated feature data to obtain the wear amount prediction information of the robot includes:

[0020] A subset of features and split points are randomly selected from the associated feature data, and a decision tree is constructed based on the randomly selected subset of features and split points.

[0021] Based on the decision tree, the wear probability distribution corresponding to the associated feature data is obtained;

[0022] Based on the wear probability distribution, generate corresponding time series prediction values;

[0023] An adaptive filtering algorithm is used to smooth the predicted time series values, and the wear prediction information of the robot is output.

[0024] Optionally, the expression corresponding to the adaptive filtering algorithm is as follows:

[0025]

[0026] In the formula,

[0027]

[0028] Among them, y t This represents the predicted value of the time series at time t; This indicates the predicted wear amount at time t; This represents the wear prediction information at time t-1; α t Δy represents the smoothing factor at time t; β represents the smoothing factor adjustment parameter; Δy t exp represents the predicted change in the time series at time t; exp represents the exponential function.

[0029] Optionally, the step of generating a dynamic control strategy for the robot based on the robot's wear prediction information using a control decision algorithm includes:

[0030] The wear prediction information of the robot is divided into states to obtain the wear state category of the robot;

[0031] Based on the wear condition category, a corresponding risk level is generated;

[0032] Based on the risk level, a strategy mapping is performed to obtain the static control strategy of the robot;

[0033] Based on the wear prediction information, the risk level, and the static control strategy, fuzzy logic control is performed to output the robot's dynamic control strategy.

[0034] Optionally, the environmental perception data includes environmental data and robot state data;

[0035] The environmental data includes one or more of the following: contact surface friction data, temperature data, humidity data, obstacle data, and terrain feature data;

[0036] The robot state data includes one or more of the following: joint torque data, motion trajectory, motion posture, and energy consumption data.

[0037] Based on the same inventive concept, the present invention also provides a system for optimizing the wear resistance of a robot, comprising:

[0038] The information acquisition module is used to collect the robot's environmental perception data and initial wear information;

[0039] The data prediction module is used to predict the wear amount of the robot based on the environmental perception data and the initial wear information, using an extreme random tree algorithm, to obtain the wear amount prediction information of the robot.

[0040] The strategy generation module is used to generate a dynamic control strategy for the robot based on the wear prediction information of the robot and using a control decision algorithm.

[0041] The dynamic adjustment module is used to adjust the motion control parameters of the robot according to the dynamic control strategy.

[0042] Optionally, the data prediction module includes:

[0043] An environmental feature extraction submodule is used to extract features based on the environmental perception data to obtain the environmental feature data of the robot.

[0044] The wear feature extraction submodule is used to extract features based on the initial wear information to obtain the wear feature data of the robot;

[0045] The feature crossover submodule is used to perform feature crossover on the environmental feature data and the wear feature data to obtain the associated feature data of the robot;

[0046] The wear prediction submodule is used to predict the wear of the robot based on the associated feature data using an extreme random tree algorithm, thereby obtaining the wear prediction information of the robot.

[0047] Optionally, the feature crossing submodule includes:

[0048] The time alignment unit is used to perform time alignment on the environmental feature data and the wear feature data using a time alignment algorithm to obtain time-aligned environmental feature data and wear feature data.

[0049] The feature interaction unit is used to perform feature interaction on the time-aligned environmental feature data and the wear feature data using a feature cross algorithm to obtain interactive feature data.

[0050] The feature filtering unit is used to filter the interactive feature data to obtain the associated feature data of the robot.

[0051] Optionally, the wear prediction submodule includes:

[0052] The decision tree construction unit is used to randomly select a subset of features and split points from the associated feature data, and construct a decision tree based on the randomly selected subset of features and split points.

[0053] The probability distribution unit is used to obtain the wear probability distribution corresponding to the associated feature data based on the decision tree;

[0054] A time series prediction unit is used to generate corresponding time series prediction values ​​based on the wear probability distribution;

[0055] The smoothing unit is used to smooth the time series prediction values ​​using an adaptive filtering algorithm and output the wear prediction information of the robot.

[0056] Optionally, the expression corresponding to the adaptive filtering algorithm is as follows:

[0057]

[0058] In the formula,

[0059]

[0060] Among them, y t This represents the predicted value of the time series at time t; This indicates the predicted wear amount at time t; This represents the wear prediction information at time t-1; α t Δy represents the smoothing factor at time t; β represents the smoothing factor adjustment parameter; Δy t exp represents the predicted change in the time series at time t; exp represents the exponential function.

[0061] Optionally, the strategy generation module includes:

[0062] The state division submodule is used to divide the wear prediction information of the robot into states to obtain the wear state category of the robot;

[0063] The risk classification submodule is used to generate a corresponding risk level based on the wear condition category;

[0064] The strategy mapping submodule is used to perform strategy mapping based on the risk level to obtain the static control strategy of the robot;

[0065] The fuzzy control submodule is used to perform fuzzy logic control based on the wear prediction information, the risk level, and the static control strategy, and output the dynamic control strategy of the robot.

[0066] Optionally, the environmental perception data includes environmental data and robot state data;

[0067] The environmental data includes one or more of the following: contact surface friction data, temperature data, humidity data, obstacle data, and terrain feature data;

[0068] The robot state data includes one or more of the following: joint torque data, motion trajectory, motion posture, and energy consumption data.

[0069] In another aspect, the present invention also provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus;

[0070] The memory is used to store one or more programs;

[0071] When the one or more programs are executed by the at least one processor, a method for optimizing the wear resistance of a robot as described above is implemented.

[0072] In another aspect, the present invention also provides a computer device readable storage medium having an executable program stored thereon, wherein when the executable program is executed, it implements the aforementioned method for optimizing the wear resistance of a robot.

[0073] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0074] This invention provides a method, system, device, and medium for optimizing the wear resistance of a robot, comprising: collecting environmental perception data and initial wear information of the robot; predicting the wear amount of the robot using an extreme random tree algorithm based on the environmental perception data and the initial wear information, thereby obtaining wear amount prediction information of the robot; generating a dynamic control strategy for the robot using a control decision algorithm based on the wear amount prediction information of the robot; and adjusting the motion control parameters of the robot according to the dynamic control strategy. This application, by employing the extreme random tree algorithm, can process high-dimensional, nonlinear environmental perception data, thereby accurately predicting the wear amount of the robot; the dynamic control strategy generated by the control decision algorithm enables the robot to autonomously adapt to complex working condition changes, thereby improving the robot's wear resistance and energy efficiency. Therefore, the method of this invention, through a closed-loop design of environmental perception, wear prediction, and dynamic control, realizes the intelligent operation of the robot under complex working conditions, improving both wear resistance and energy utilization efficiency, providing strong technical support for the widespread application of robots in industry, service, and other fields. Attached Figure Description

[0075] Figure 1 A flowchart illustrating a method for optimizing the wear resistance of a robot provided by the present invention;

[0076] Figure 2 A schematic diagram of the framework for predicting wear amount in a method for optimizing the wear resistance of a robot provided by the present invention;

[0077] Figure 3A schematic diagram illustrating the execution flow of the extreme random tree algorithm in the wear resistance optimization method for a robot provided by the present invention;

[0078] Figure 4 A schematic diagram of the structural composition of a wear resistance optimization system for a robot provided by the present invention;

[0079] Figure 5 This is a schematic diagram of the structure of an electronic device provided by the present invention. Detailed Implementation

[0080] This invention proposes a method, system, device, and medium for optimizing the wear resistance of robots. The specific embodiments of this invention will be further described in detail below with reference to the accompanying drawings.

[0081] Example 1:

[0082] This invention provides a method for optimizing the wear resistance of a robot, as illustrated in the flowchart below. Figure 1 As shown, it includes:

[0083] Step 1: Collect environmental perception data and initial wear information of the robot;

[0084] Step 2: Based on environmental perception data and initial wear information, the extreme random tree algorithm is used to predict the wear amount of the robot, and the wear amount prediction information of the robot is obtained.

[0085] Step 3: Based on the robot's wear prediction information, generate a dynamic control strategy for the robot using a control decision algorithm;

[0086] Step 4: Adjust the robot's motion control parameters according to the dynamic control strategy.

[0087] Generally, anti-wear methods for robots typically focus on researching material structures and surface treatment technologies to improve their durability and wear resistance. However, these methods often only alleviate wear problems to a certain extent and cannot predict and control wear conditions in real time and accurately. This invention collects detailed environmental perception data and initial wear information of the robot through pre-set electronic devices (which can be independent of the robot's collaborative devices or dedicated acquisition units embedded in the robot). It then uses an extreme random tree algorithm to predict wear amount and supplements this with a dynamic control strategy generated by a control decision algorithm. This allows for intelligent adjustment of the robot's motion control parameters, achieving good adaptation to complex working conditions.

[0088] For example, the environmental perception data in step 1 above may include environmental data and robot state data;

[0089] The environmental data mentioned above may include one or more of the following: contact surface friction data, temperature data, humidity data, obstacle data, and terrain feature data;

[0090] The robot state data mentioned above may include one or more of the following: joint torque data, motion trajectory, motion posture, and energy consumption data;

[0091] In this example, by combining environmental data such as friction, temperature, and terrain features with the robot's own state data such as joint torques and motion trajectories to form a rich perception network, the robot can gain a comprehensive understanding of its operating environment and its own dynamic state. This in-depth analysis and collaborative processing of multi-dimensional data creates conditions for achieving a highly accurate decision support system. This enables the robot not only to effectively cope with various external changes and uncertainties in the internal model, but also to optimize energy consumption and improve work efficiency through real-time self-adjustment. The robot can then autonomously complete tasks in complex environments, significantly improving the reliability and endurance of operations, thereby supporting more challenging and complex application scenarios.

[0092] For example, the initial wear information mentioned above can include: component material properties (such as hardness, wear resistance, coefficient of friction, etc.), geometric feature information (such as component size, shape, and surface roughness, etc.), and wear patterns (such as initial wear marks and wear characteristics). In this example, by collecting the robot's initial wear information, the initial wear condition can be monitored, helping to identify early signs of failure in robot components and detect potential problems early, enabling preventative maintenance and reducing the risk of sudden failures. After a comprehensive understanding and analysis of the robot's working environment and its state, rich environmental perception data and state information can be collected. This information is crucial for accurately predicting the robot's wear condition and formulating effective control strategies. By building an intelligent predictive model, not only can the robot's wear condition be monitored in real time, but it can also provide dynamic support for its motion control, thereby extending the equipment's service life and improving operational efficiency. Specifically:

[0093] In one implementation, such as Figure 2 As shown, step 2 above, which uses an extreme random tree algorithm to predict the wear amount of the robot based on environmental perception data and initial wear information, can include the following steps:

[0094] Feature extraction is performed based on environmental perception data to obtain the robot's environmental feature data;

[0095] Feature extraction is performed based on the initial wear information to obtain the robot's wear feature data;

[0096] By performing feature cross-referencing on environmental feature data and wear feature data, we can obtain the robot's associated feature data;

[0097] Based on the associated feature data, the extreme random tree algorithm is used to predict the wear amount of the robot, and the wear amount prediction information of the robot is obtained.

[0098] In this implementation, environmental perception data and robot state data are collected, and combined with feature extraction and cross-feature processing, environmental feature data and wear feature data of the robot are obtained. These feature data are then used to perform feature cross-correlation to construct a comprehensive feature description. Finally, an extreme random tree algorithm is used to predict the wear amount of the robot to obtain wear prediction information. This process can not only predict the wear situation in real time and accurately, but also dynamically adjust the robot's operating parameters according to different wear states, thereby significantly improving the robot's service life and operating efficiency. Through this intelligent management, unnecessary maintenance work can be reduced, operating costs can be lowered, and the stable and efficient operation of the robot can be ensured in complex and ever-changing environments, greatly improving the reliability and overall performance of the system.

[0099] In predicting robot wear based on environmental perception data and initial wear information, feature extraction and feature cross-interaction are key steps. Feature cross-interaction effectively fuses environmental and wear feature data to generate correlated feature data that comprehensively reflects the robot's state, providing complete and accurate input for wear prediction. Specifically:

[0100] In one implementation, the process of performing feature cross-referencing on environmental feature data and wear feature data to obtain the robot's associated feature data may include:

[0101] A time alignment algorithm is used to align environmental feature data and wear feature data in time, resulting in time-aligned environmental feature data and wear feature data.

[0102] A feature crossover algorithm is used to perform feature interaction on time-aligned environmental feature data and wear feature data to obtain interactive feature data.

[0103] Feature filtering is performed on the interaction feature data to obtain the robot's associated feature data;

[0104] In this implementation, a time alignment algorithm is used to accurately align environmental feature data and wear feature data, ensuring data synchronization in the time dimension. Then, a feature cross-validation algorithm is used to achieve deep interaction between environmental feature data and wear feature data, generating rich interactive feature data. By filtering these interactive feature data, the robot's associated feature data is finally obtained. This method can not only significantly improve the accuracy and reliability of wear prediction, but also dynamically adjust the robot's operating parameters according to different wear states, thereby significantly improving the robot's service life and operating efficiency. While time alignment and feature cross-referencing algorithms are relatively common methods in data processing, they demonstrate significant advantages in specific applications of robot wear management and optimization. For example, time alignment ensures that environmental and wear features are analyzed on the same time frame, which is crucial in dynamic and rapidly changing robot working environments, helping to improve the robot's reaction speed and adaptability during operation. Secondly, feature cross-referencing reveals the potential impact of environmental conditions on wear behavior, enabling a deeper understanding of complex wear mechanisms. This approach to processing and analyzing high-dimensional data brings new insights. Furthermore, the effective implementation of feature selection strategies helps identify correlated features that significantly influence wear prediction, thereby enhancing model performance. Therefore, although these algorithms are not inherently novel in the field of robotics, combining, customizing, and optimizing them to meet specific needs makes their application in particular scenarios especially meaningful.

[0105] During the feature intersection stage, the associated feature data, after time alignment, interaction, and feature filtering, can comprehensively integrate environmental and wear features, providing high-quality input data for subsequent wear prediction. To further accurately assess the robot's wear condition and predict its future wear trend, the generated associated feature data can be used to accurately predict the robot's wear. Specifically:

[0106] In one implementation, such as Figure 3 As shown, the process described above, which uses an extreme random tree algorithm to predict robot wear based on associated feature data, and obtains robot wear prediction information, may include:

[0107] A subset of features and split points are randomly selected from the associated feature data, and a decision tree is constructed based on the randomly selected subset of features and split points.

[0108] Based on the decision tree, the wear probability distribution corresponding to the associated feature data is obtained;

[0109] Based on the wear probability distribution, generate corresponding time series prediction values;

[0110] An adaptive filtering algorithm is used to smooth the time series prediction values ​​and output the robot's wear prediction information.

[0111] Specifically, in this implementation, the process of randomly selecting a subset of features and split points from the associated feature data, and constructing a decision tree based on the randomly selected subset of features and split points, may include:

[0112] S1: Randomly select a subset of features from the associated feature data;

[0113] S2: For each feature in the selected feature subset, randomly select a split point within a preset value range;

[0114] S3: Based on the randomly selected feature subset and split point, divide the data of the current node into left and right subtrees, and recursively execute steps S1-S2 until any of the following termination conditions are met, the recursion is completed, and the decision tree is output.

[0115] The termination conditions may include: a predetermined tree depth, a number of samples in the feature subset being lower than a set threshold, and the sample categories in the feature subset.

[0116] In this implementation, by randomly selecting feature subsets and split points from associated feature data, the dynamic construction of the decision tree can directly generate a model based on the current associated feature data (such as environmental perception data and wear data). Compared with a pre-trained static model, dynamic construction can capture changes in working conditions in real time (such as sudden changes in friction and temperature fluctuations), determine complex patterns and relationships in the data, and reduce dependence on historical data. Randomness is re-injected in each prediction, enabling the model to more flexibly mine potential correlations in high-dimensional and nonlinear data, enhance the generalization ability to unknown data, and ultimately generate a more accurate wear probability distribution. Furthermore, by employing an adaptive filtering algorithm to smooth the time series prediction values, short-term fluctuations and noise are eliminated, thereby constructing a stable and reliable wear prediction model. Therefore, this implementation, which combines dynamic decision tree construction with adaptive filtering algorithms, not only reduces the risk of model overfitting but also enhances the model's anti-interference ability, maintaining high-precision prediction and evaluation even in the face of unstable environmental conditions.

[0117] For example, the expression corresponding to the above adaptive filtering algorithm can be as follows:

[0118]

[0119] In the formula,

[0120]

[0121] Among them, y tThis represents the predicted value of the time series at time t; This indicates the predicted wear amount at time t; This represents the wear prediction information at time t-1; α t Δy represents the smoothing factor at time t; β represents the smoothing factor adjustment parameter; Δy t Let represent the predicted change in the time series at time t; exp represent the exponential function. In this example, the smoothing factor is dynamically adjusted based on the current time series prediction value and historical prediction values. This allows the prediction results to retain the original prediction information while possessing stronger smoothness and stability. Compared to the traditional method with a fixed smoothing factor, this adaptive mechanism can better adapt to changes in the predicted values, improving the accuracy and reliability of the prediction information. Through this method, not only is the wear prediction accuracy of the robot significantly improved, but the robot's operating parameters can also be dynamically adjusted according to different wear states, thereby significantly improving the robot's service life and operating efficiency. Therefore, this example helps reduce unnecessary maintenance work, lowers operating costs, and ensures the stable and efficient operation of the robot in complex and changing environments, greatly improving the system's reliability and overall performance.

[0122] An intelligent wear prediction model is constructed using the extreme random tree algorithm, which can obtain reliable wear prediction information. Based on this wear prediction information, a control decision algorithm can be further used to generate control strategies according to real-time working conditions, environmental changes, and wear levels, so as to dynamically optimize the robot's motion parameters. This allows the robot to continuously learn and optimize during operation. Specifically:

[0123] In one implementation, step 3 above, which involves generating a dynamic control strategy for the robot based on its wear prediction information using a control decision algorithm, may include:

[0124] The wear prediction information of the robot is divided into states to obtain the wear state category of the robot;

[0125] Based on the wear condition category, a corresponding risk level is generated;

[0126] Based on the risk level, a strategy mapping is performed to obtain the robot's static control strategy;

[0127] Based on wear prediction information, risk level, and static control strategy, fuzzy logic control is performed to output the robot's dynamic control strategy.

[0128] In this implementation, by comprehensively utilizing wear prediction information and control decision algorithms, an efficient dynamic control strategy is generated for the robot, significantly improving its operational efficiency and reliability. Accurate prediction of robot wear and state classification based on the prediction results (e.g., low wear, medium wear, and high wear) helps identify potential risk factors early. The generated risk levels (e.g., low, medium, and high risk) and corresponding static control strategies provide strong support for subsequent control. Furthermore, by integrating wear prediction, risk assessment, and static strategies using fuzzy logic control, automatic and precise adjustment of the dynamic control strategy is achieved. This process not only reduces the need for human intervention but also ensures that the control strategy can respond to changes in wear conditions in real time, thereby extending the robot's service life and reducing maintenance costs. Although the strategy mapping and fuzzy logic control used in this implementation are relatively common techniques, their combination can demonstrate new roles and effects in specific application scenarios, especially in robot wear management and dynamic control scenarios. For example, risk level generation is not merely a simple mapping of wear status; it involves a comprehensive assessment of multiple factors, including the impact of wear on robot operational safety, efficiency, and maintenance costs. This approach dynamically generates optimal control strategies based on real-time wear prediction information, risk levels, and static control strategies. This targeted strategy mapping not only improves the effectiveness of risk management but also enhances the robot's intelligence level in specific applications. For instance, in the application of fuzzy logic control, by considering dynamic changes in environmental conditions, load fluctuations, and material properties, uncertainty and fuzziness can be effectively integrated into the decision-making process. Furthermore, it can quickly respond to real-time wear prediction information and dynamically adjust control strategies based on new data and risk assessments, ensuring the robot maintains optimal performance under different time periods and conditions. This allows the robot to respond more flexibly to changing situations and achieve more sensitive and personalized dynamic control.

[0129] After completing the process of predicting robot wear using control decision algorithms and generating dynamic control strategies, the next key step is to effectively apply these dynamic control strategies to the robot's motion control to reduce wear and improve system performance. Generating dynamic control strategies is only one link in the entire optimization chain; its ultimate goal is to ensure the robot can operate efficiently and stably under complex working conditions through real-time adjustment of robot motion control parameters. Specifically:

[0130] In one implementation, step 4 above, which involves adjusting the robot's motion control parameters according to a dynamic control strategy, may include:

[0131] First, the dynamic control strategy is analyzed. The control strategy is usually given in the form of structured data, such as the adjustment amount of joint torque, the optimized value of motion speed, and the start frequency of the lubrication system. These strategies need to be parsed into specific control instructions by the algorithm to ensure that they can be recognized and executed by the robot's control system. In the parsing process, it may be necessary to combine the robot's current state (such as joint position and load) and environmental information (such as ground material and friction) to generate adjustment instructions that adapt to the actual working conditions. Next, the adjustment of motion control parameters is the core link. According to the parsed control instructions, the robot control system needs to modify the relevant parameters in real time. For example, adjusting joint torque can be achieved through a motor controller, optimizing motion trajectory can be accomplished through path planning algorithms, and activating the lubrication system can be achieved by actuators controlling the frequency and amount of lubricant release. This process requires high-precision sensor data to ensure the accuracy and real-time nature of the adjustments. For instance, force sensors can monitor changes in joint torque, encoders can provide feedback on the execution of the motion trajectory, and temperature sensors can assess the lubrication effect. Finally, execution and feedback are crucial for ensuring the effectiveness of the control. The adjusted parameters need to be applied to the robot's motion control system in real time, and the control effect needs to be verified through sensor feedback. For example, if the control objective is to reduce joint wear, changes in wear can be monitored through force sensors and encoders; if the objective is to optimize energy consumption, energy consumption can be assessed through current sensors and power meters. Based on the feedback data, the control system can further optimize the dynamic control strategy to form a closed-loop control, thereby continuously improving the robot's wear resistance and overall performance.

[0132] Furthermore, considering that the environment changes as the robot moves during operation (e.g., temperature, humidity, obstacle position, ground friction coefficient, etc.), it is necessary to dynamically adjust the data collection and optimization strategies. For example, real-time data collection can be achieved by configuring sensors (such as temperature sensors, humidity sensors, and LiDAR), rather than collecting environmental data only once at the start of the operation. This allows for real-time updates of environmental perception information during task execution. After acquiring the real-time environmental perception data, a data fusion algorithm can be used to combine the real-time environmental perception data with initial wear information to form a comprehensive environmental model. By dynamically updating the data, changes in the environment can be reflected in a timely manner.

[0133] If the environment changes during operation, the wear and tear may also change. Therefore, dynamic adjustments are needed to the wear prediction. For example, incremental learning methods can be used to update the existing wear prediction model based on new environmental perception data and wear information. Whenever an environmental change is detected, the model automatically adjusts and updates the wear trend prediction in real time. Furthermore, an adaptive control strategy can be adopted to adjust the robot's motion parameters based on real-time environmental data. For instance, when the robot travels on different surfaces (such as smooth and rough), its speed and acceleration are immediately adjusted to reduce wear. During task execution, the robot continuously collects wear feedback and task performance data. If an abnormal wear rate is detected, a maintenance prompt is automatically triggered, and the working strategy is adjusted to ensure that the robot continues to operate efficiently under new environmental conditions.

[0134] In summary, this invention addresses the problem of traditional robot wear resistance methods lacking environmental awareness and struggling to adapt to complex working conditions, resulting in low wear resistance and energy efficiency. It proposes an optimization method for robot wear resistance. Starting with environmental perception data and initial robot wear information, it comprehensively analyzes the impact of the robot's operating environment on wear. Utilizing an extreme random tree algorithm, it achieves highly accurate wear prediction. This algorithm can quickly process high-dimensional data and effectively avoid overfitting, making wear prediction more reliable and accurate. After obtaining the wear prediction information, a dynamic control strategy is generated through a control decision algorithm. This means the robot can adjust its motion control parameters in real time according to the current wear state and environmental changes to ensure continuous and efficient operation. Through this dynamic control, the robot can maintain optimal operating conditions when wear conditions change, reducing failure rates and extending the effective working time of the equipment. Therefore, this invention, by integrating advanced prediction algorithms and intelligent control strategies, not only improves the accuracy of wear prediction but also enhances the robot's autonomous adjustment capabilities. This method enables robots to better adapt to complex operating environments, reduce maintenance intervention, and bring higher economic benefits and lower operational risks, representing a significant technological advancement in the field of robotics.

[0135] Example 2:

[0136] Based on the same inventive concept, this invention also provides a system for optimizing the wear resistance of a robot, the structural composition of which is shown in the schematic diagram below. Figure 4 As shown, it includes:

[0137] The information acquisition module is used to collect the robot's environmental perception data and initial wear information;

[0138] The data prediction module is used to predict the wear amount of the robot based on environmental perception data and initial wear information, using an extreme random tree algorithm to obtain the robot's wear amount prediction information.

[0139] The strategy generation module is used to generate dynamic control strategies for the robot based on the robot's wear prediction information and using a control decision algorithm.

[0140] The dynamic adjustment module is used to adjust the robot's motion control parameters according to the dynamic control strategy.

[0141] For example, environmental perception data includes environmental data and robot state data;

[0142] Environmental data includes one or more of the following: contact surface friction data, temperature data, humidity data, obstacle data, and terrain feature data;

[0143] Robot state data includes one or more of the following: joint torque data, motion trajectory, motion posture, and energy consumption data.

[0144] In one implementation, the data prediction module described above may include:

[0145] The environmental feature extraction submodule is used to extract features from environmental perception data to obtain the robot's environmental feature data.

[0146] The wear feature extraction submodule is used to extract features based on the initial wear information to obtain the robot's wear feature data;

[0147] The feature crossover submodule is used to perform feature crossover on environmental feature data and wear feature data to obtain the robot's associated feature data;

[0148] The wear prediction submodule is used to predict the wear of the robot based on the associated feature data and the extreme random tree algorithm, so as to obtain the wear prediction information of the robot.

[0149] In this implementation, the aforementioned feature crossover submodule may include:

[0150] The time alignment unit is used to perform time alignment on environmental feature data and wear feature data using a time alignment algorithm to obtain time-aligned environmental feature data and wear feature data;

[0151] The feature interaction unit is used to perform feature interaction on time-aligned environmental feature data and wear feature data using a feature cross algorithm to obtain interactive feature data.

[0152] The feature filtering unit is used to filter the interactive feature data to obtain the robot's associated feature data.

[0153] In this implementation, the wear prediction submodule described above may include:

[0154] The decision tree building unit is used to randomly select a subset of features and split points from the associated feature data, and to build a decision tree based on the randomly selected subset of features and split points.

[0155] The probability distribution unit is used to obtain the wear probability distribution corresponding to the associated feature data based on the decision tree;

[0156] The time series prediction unit is used to generate corresponding time series prediction values ​​based on the wear probability distribution;

[0157] The smoothing unit is used to smooth the time series prediction values ​​using an adaptive filtering algorithm and output the robot's wear prediction information.

[0158] For example, the expression corresponding to the above adaptive filtering algorithm can be as follows:

[0159]

[0160] In the formula,

[0161]

[0162] Among them, y t This represents the predicted value of the time series at time t; This indicates the predicted wear amount at time t; This represents the wear prediction information at time t-1; α t Δy represents the smoothing factor at time t; β represents the smoothing factor adjustment parameter; Δy t exp represents the predicted change in the time series at time t; exp represents the exponential function.

[0163] In one implementation, the strategy generation module described above may include:

[0164] The state division submodule is used to divide the robot's wear prediction information into states to obtain the robot's wear state categories.

[0165] The risk classification submodule is used to generate corresponding risk levels based on the wear condition category;

[0166] The strategy mapping submodule is used to map strategies according to risk levels to obtain the robot's static control strategy;

[0167] The fuzzy control submodule is used to perform fuzzy logic control based on wear prediction information, risk level, and static control strategy, and output the robot's dynamic control strategy.

[0168] Example 3:

[0169] like Figure 5As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0170] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the wear resistance optimization method for a robot in the above embodiments.

[0171] Example 4:

[0172] Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). An electronic device readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the robot wear resistance optimization method in the above embodiments.

[0173] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0174] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0175] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0176] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the application, but these changes, modifications or equivalent substitutions are all within the scope of protection of the claims pending approval.

Claims

1. A method for optimizing the wear resistance of a robot, characterized in that, include: Collect environmental perception data and initial wear information of the robot; Feature extraction is performed based on the environmental perception data to obtain the robot's environmental feature data; Based on the initial wear information, feature extraction is performed to obtain the wear feature data of the robot; The environmental feature data and the wear feature data are cross-referenced to obtain the associated feature data of the robot; A subset of features and split points are randomly selected from the associated feature data, and a decision tree is constructed based on the randomly selected subset of features and split points. Based on the decision tree, the wear probability distribution corresponding to the associated feature data is obtained; Based on the wear probability distribution, generate corresponding time series prediction values; An adaptive filtering algorithm is used to smooth the predicted time series values, and the wear prediction information of the robot is output. Based on the wear prediction information of the robot, a dynamic control strategy for the robot is generated using a control decision algorithm; The motion control parameters of the robot are adjusted according to the dynamic control strategy. The step of randomly selecting a feature subset and split points from the associated feature data, and constructing a decision tree based on the randomly selected feature subset and split points, includes: S1: Randomly select a subset of features from the associated feature data; S2: For each feature in the selected feature subset, randomly select a split point within a preset value range; S3: Based on the randomly selected feature subset and split point, divide the data of the current node into left and right subtrees, and recursively execute steps S1-S2 until any of the following termination conditions are met, the recursion is completed, and the decision tree is output. The termination conditions include: the predetermined tree depth, the number of samples in the feature subset being lower than a set threshold, and the sample categories in the feature subset; The environmental perception data includes environmental data and robot status data; The environmental data includes one or more of the following: contact surface friction data, temperature data, humidity data, obstacle data, and terrain feature data; The robot state data includes one or more of the following: joint torque data, motion trajectory, motion posture, and energy consumption data.

2. The method as described in claim 1, characterized in that, The step of performing feature cross-referencing on the environmental feature data and the wear feature data to obtain the associated feature data of the robot includes: The environmental feature data and the wear feature data are time-aligned using a time alignment algorithm to obtain time-aligned environmental feature data and wear feature data. A feature crossover algorithm is used to perform feature interaction between the time-aligned environmental feature data and the wear feature data to obtain interactive feature data. The interaction feature data is filtered to obtain the robot's associated feature data.

3. The method as described in claim 1, characterized in that, The expression corresponding to the adaptive filtering algorithm is as follows: ； In the formula, ； in, express Time series predicted value at time point; express Wear and tear prediction information at any given time; express Wear and tear prediction information at any given time; express The smoothing factor at any given time; This represents the smoothing factor adjustment parameter; express Predicting changes in time series data at any given moment; This represents an exponential function.

4. The method as described in claim 1, characterized in that, The step of generating a dynamic control strategy for the robot based on the wear prediction information of the robot and using a control decision algorithm includes: The wear prediction information of the robot is divided into states to obtain the wear state category of the robot; Based on the wear condition category, a corresponding risk level is generated; Based on the risk level, a strategy mapping is performed to obtain the static control strategy of the robot; Based on the wear prediction information, the risk level, and the static control strategy, fuzzy logic control is performed to output the robot's dynamic control strategy.

5. A system for optimizing the wear resistance of a robot, characterized in that, include: The information acquisition module is used to collect the robot's environmental perception data and initial wear information; The data prediction module is used to predict the wear amount of the robot based on the environmental perception data and the initial wear information, using an extreme random tree algorithm, to obtain the wear amount prediction information of the robot. The strategy generation module is used to generate a dynamic control strategy for the robot based on the wear prediction information of the robot and using a control decision algorithm. The dynamic adjustment module is used to adjust the motion control parameters of the robot according to the dynamic control strategy. The data prediction module includes: An environmental feature extraction submodule is used to extract features based on the environmental perception data to obtain the environmental feature data of the robot. The wear feature extraction submodule is used to extract features based on the initial wear information to obtain the wear feature data of the robot; The feature crossover submodule is used to perform feature crossover on the environmental feature data and the wear feature data to obtain the associated feature data of the robot; The wear prediction submodule is used to predict the wear of the robot based on the associated feature data using an extreme random tree algorithm, thereby obtaining the wear prediction information of the robot. The wear prediction submodule includes: The decision tree construction unit is used to randomly select a subset of features and split points from the associated feature data, and construct a decision tree based on the randomly selected subset of features and split points. The probability distribution unit is used to obtain the wear probability distribution corresponding to the associated feature data based on the decision tree; A time series prediction unit is used to generate corresponding time series prediction values ​​based on the wear probability distribution; The smoothing unit is used to smooth the time series prediction values ​​using an adaptive filtering algorithm and output the wear prediction information of the robot. The decision tree construction unit is specifically used for: S1: Randomly select a subset of features from the associated feature data; S2: For each feature in the selected feature subset, randomly select a split point within a preset value range; S3: Based on the randomly selected feature subset and split point, divide the data of the current node into left and right subtrees, and recursively execute steps S1-S2 until any of the following termination conditions are met, the recursion is completed, and the decision tree is output. The termination conditions include: the predetermined tree depth, the number of samples in the feature subset being lower than a set threshold, and the sample categories in the feature subset; The environmental perception data includes environmental data and robot status data; The environmental data includes one or more of the following: contact surface friction data, temperature data, humidity data, obstacle data, and terrain feature data; The robot state data includes one or more of the following: joint torque data, motion trajectory, motion posture, and energy consumption data.

6. An electronic device, characterized in that, include: At least one processor and memory; The memory and processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a method for optimizing the wear resistance of a robot as described in any one of claims 1 to 4 is implemented.

7. A computing device readable storage medium, characterized in that, It contains an execution program, which, when executed, implements a method for optimizing the wear resistance of a robot as described in any one of claims 1 to 4.

Images