A master-slave robot control system and method for nuclear decommissioning operations

Abstract

This invention proposes a master-slave robotic arm control system and method for nuclear decommissioning operations. The system includes a data acquisition module that collects real-time posture data, corresponding communication delay data, and hot chamber environment data of the master and slave robotic arms; a data processing module that preprocesses the collected data, extracts standard feature information of the real-time posture of the master and slave robotic arms, standard feature information of the communication delay, and feature data of the hot chamber environment, and stores them in a motion information database; and a motion prediction module that constructs a trajectory prediction model for the slave robotic arm based on a neural network and outputs the predicted motion trajectory information of the slave robotic arm. In the nuclear decommissioning hot chamber environment, this system enhances the environmental adaptability of the system by collecting real-time posture data, communication delay data, and hot chamber environment data of the master and slave robotic arms, and combining them with a neural network to construct a high-precision trajectory prediction model for the slave robotic arm, thereby improving the accuracy and stability of the slave robotic arm in following the master arm's movements and ensuring precise synchronous operation of the master and slave robotic arms in complex environments.

Description

Technical Field

[0001] This invention relates to the field of master-slave robotic arms for nuclear decommissioning, and more particularly to a control system and method for a master-slave robotic arm for nuclear decommissioning operations. Background Technology

[0002] In the nuclear field, laboratories and factories engaged in nuclear scientific research and production rely heavily on specialized remote manipulation equipment, specifically nuclear robotic arms, for handling radioactive materials. The key difference between these devices and other ordinary manipulation equipment lies in their ability to isolate operators from hazardous environments through appropriate biological shielding. The purpose is to separate operators from the manipulated objects, ensuring the safety of both personnel and the operating environment while handling radioactive materials. Their applications span nuclear power, reprocessing, laboratory analysis, and the medical field, making them indispensable remote manipulation tools in the nuclear sector.

[0003] A robotic arm control system based on a master-slave teleoperated robot, disclosed in CN113084775A, includes a potentiometer group, an AD conversion module, a master controller, a drive module, a communication module, a slave controller, an DA conversion module, a servo control module, a force sensor group, and a control center. The potentiometer group, force sensor group, AD conversion module, master controller, and drive module constitute the master control system. This robotic arm control system based on a master-slave teleoperated robot collects various data from the master hand through the master control system, analyzes and processes the data through the control center, and drives the slave control system to control the slave hand to perform synchronous operation. The slave control system collects slave hand data in real time and transmits it to the control center to drive the master control system to adjust the master hand, so that the master and slave robotic arms are synchronized and precise control is achieved.

[0004] In existing robotic arm control systems based on master-slave teleoperation, although the master control system collects master hand data, the control center analyzes and processes the data, and drives the slave control system to achieve master-slave synchronous operation, and the slave control system can collect slave hand data in real time and feed it back to the control center to adjust the master hand, the impact of environmental factors and communication time errors on the performance of the slave robotic arm is not fully considered when facing the complex and ever-changing hot chamber environment. This results in the slave robotic arm's accuracy in following the master hand's movements in the hot chamber environment, reducing the slave robotic arm's accuracy and stability in following the master hand's movements. Summary of the Invention

[0005] In view of this, the present invention proposes a master-slave robot control system and method for nuclear decommissioning operations. In the nuclear decommissioning hot chamber environment, by collecting real-time posture data, communication delay data and hot chamber environment data of the master and slave robots, and combining them with neural networks to construct a high-precision trajectory prediction model of the slave robot, the system's environmental adaptability is enhanced, the accuracy and stability of the slave robot in following the master robot's movements are improved, and the precise synchronous operation of the master and slave robots in complex environments is ensured.

[0006] The technical solution of this invention is implemented as follows: This invention provides a master-slave robotic arm control system for nuclear decommissioning operations, the system comprising:

[0007] The acquisition module is used to collect real-time posture data of the master and slave robotic arms, as well as corresponding communication delay data and hot chamber environment data.

[0008] The data processing module is used to preprocess the collected data, extract the real-time posture standard feature information, communication delay standard feature information and hot chamber environment feature data of the master and slave robot arms, and store them into the motion information database.

[0009] The motion prediction module is used to build a robot arm trajectory prediction model based on a neural network. It takes historical data and real-time posture data, as well as corresponding communication delay data and hot chamber environment data from the motion information database as input, and trains the robot arm trajectory prediction model to output the motion trajectory prediction information of the robot arm.

[0010] The execution module is used to convert the motion trajectory prediction information of the current robotic arm into control signals and send them to the slave control module. The slave control module executes corresponding actions according to the received control signals to achieve accurate tracking of the master arm's movements.

[0011] Based on the above technical solutions, preferably, the preprocessing of the collected data to obtain real-time posture standard feature information of the master and slave robot arms, communication delay standard feature information, and hot chamber environment feature data includes:

[0012] The posture data of the master and slave robotic arms includes the angle vectors of each joint, and the hot chamber environment data includes radiation intensity, temperature and humidity;

[0013] Low-pass filtering was used to denoise the collected data and normalize it to obtain real-time posture standard information of the master and slave robot arms, communication delay standard information, and hot chamber environment data.

[0014] Standard feature information is extracted from the real-time posture standard information, communication delay standard information and hot chamber environment data of the master and slave robot arms to obtain the real-time posture standard feature information, communication delay standard feature information and hot chamber environment feature data of the master and slave robot arms;

[0015] A motion information database is constructed, which associates the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and stores them in the motion information database.

[0016] Based on the above technical solutions, preferably, the step of extracting standard feature information from the real-time posture standard information, communication delay standard information, and hot chamber environment data of the master and slave robot arms to obtain the real-time posture standard feature information, communication delay standard feature information, and hot chamber environment feature data of the master and slave robot arms includes:

[0017] Based on the joint angle vectors of the master and slave robotic arms, the position features of the master and slave robotic arms are calculated and obtained, where the position feature expression is:

[0018] p m (t)=f kin (q m (t))

[0019] p s (t)=f kin (q s (t))

[0020] In the formula, P m (t) represents the position vector of the main robotic arm at time t, q m (t) represents the joint angle vector of the main robotic arm at time t, P s q(t) represents the position vector of the robot arm at time t. s (t) represents the joint angle vector of the robot arm at time t, f kin It is the forward kinematics function;

[0021] Taking the first derivative of the position features of the master and slave robotic arms with respect to time, we obtain the velocity features v of the master robotic arm. m (t) and the velocity characteristics v from the robot arm s (t); Taking the second derivative of the position characteristics of the master and slave robotic arms with respect to time, we obtain the acceleration characteristics a of the master robotic arm. m (t) and the acceleration characteristics from the robotic arm a s (t);

[0022] A fixed time step is preset, and communication delay data is acquired at each time point within that fixed time step. The average delay and delay fluctuation within the fixed time step are calculated to obtain standard characteristic information of communication delay. The delay fluctuation expression is as follows:

[0023]

[0024] Where σΔt is the delayed fluctuation, Δt i The communication delay data at the current time point, Δt avg The average delay is N, and the number of time points with a fixed time step is N.

[0025] The radiation intensity, temperature, and humidity at each time point within a fixed time step are obtained. The average radiation intensity, average temperature, and average humidity within the fixed time step are calculated to obtain the environmental characteristic data of the hot chamber.

[0026] Based on the above technical solutions, preferably, the motion prediction module constructs a robot arm trajectory prediction model based on a neural network. The network structure of the robot arm trajectory prediction model includes an input layer, a multi-scale feature extraction module, a hot chamber environment feature enhancement module, a Transformer encoder, and a multi-task output module. The input layer receives real-time posture standard feature information, communication delay standard feature information, and hot chamber environment feature data from the master and slave robot arms. The output of the input layer is connected to the input of the multi-scale feature extraction module to extract the spatial and temporal features of the input data. The output of the multi-scale feature extraction module is connected to the input of the hot chamber environment feature enhancement module to dynamically adjust feature weights based on the hot chamber environment feature data. The output of the hot chamber environment feature enhancement module is connected to the input of the Transformer encoder to capture global dependencies through a multi-head self-attention mechanism. The output of the Transformer encoder is connected to the input of the multi-task output module to predict and output the position vector, velocity features, and acceleration features of the operator arm, respectively.

[0027] Based on the above technical solutions, preferably, the multi-scale feature extraction module includes a CNN network extraction unit, an LSTM network extraction unit, and a feature fusion unit, wherein,

[0028] The CNN network extraction unit is used to extract spatial features from the input data, and its expression is:

[0029] F cnn =CNN(X)

[0030] In the formula, F cnn X represents the spatial features extracted by the CNN network extraction unit, and X represents the input master-slave robot real-time pose standard feature information, communication delay standard feature information, and hot chamber environment feature data.

[0031] The LSTM network extraction unit is used to extract the temporal features of the input data, expressed as:

[0032] F lstm =LSTM(X)

[0033] In the formula, F lstm Temporal features extracted by the LSTM module;

[0034] The feature fusion unit is used to fuse the features output by the CNN network extraction unit and the features output by the LSTM network extraction unit to obtain fused feature data, expressed as:

[0035] In the formula, F fusion This is the result of splicing the spatial features extracted by the CNN network extraction unit and the temporal features extracted by the LSTM module.

[0036] Based on the above technical solutions, preferably, the hot chamber environment characteristic enhancement module includes a preset radiation resistance weighting matrix and a temperature and humidity weighting matrix, wherein,

[0037] The radiation resistance weight matrix is ​​obtained by dynamically adjusting the basic weight matrix based on the radiation intensity. The expression for this matrix is:

[0038]

[0039] In the formula, W rad (t) is the radiation resistance weight matrix. Here, α is the radiation resistance weighting matrix, γ(t) is the radiation intensity influence coefficient, and γ(t) is the radiation intensity at time point t.

[0040] The temperature and humidity weighting matrix is ​​obtained by dynamically adjusting the basic matrix based on temperature and humidity, and its expression is:

[0041]

[0042] In the formula, W dynamic (t) is the temperature and humidity weight matrix. Let β be the fundamental matrix for temperature and humidity weights. T β is the influence coefficient of temperature. H Let T(t) be the temperature at time t, and H(t) be the humidity at time t.

[0043] The fused feature data is enhanced based on the radiation resistance weight matrix and the temperature and humidity weight matrix, resulting in enhanced feature data, expressed as follows:

[0044] F dynamic =F fusion· W rad (t) · W dynamic (t)

[0045] In the formula, F dynamic To enhance feature data.

[0046] Based on the above technical solutions, preferably, the multi-task output module includes outputting the position vector, velocity characteristics, and acceleration characteristics of the operator, wherein,

[0047] The expression for the predicted location vector is:

[0048] PYs (t+Δt)=W p h+b p

[0049] In the formula, PY s (t+Δt) is the position vector predicted from the operator, W p To predict the weights of the position vector, b p The bias for predicting the position vector;

[0050] The expression for the velocity characteristic is:

[0051] vY s (t+Δt)=W v h+b v

[0052] In the formula, vY s (t+Δt) represents the velocity characteristic predicted from the operator, W v For the weights of the predicted velocity feature, b v The bias used to predict velocity features;

[0053] The expression for the acceleration characteristic is:

[0054] aY s (t+Δt)=W a h+b a

[0055] In the formula, aY s (t+Δt) represents the acceleration characteristic predicted from the operator, W a To predict the weights of acceleration features, b a The bias used to predict acceleration characteristics.

[0056] Based on the above technical solution, preferably, the robot arm trajectory prediction model further includes setting a loss function, and using the Adam optimization algorithm to iteratively train the model according to the loss function to adjust the model parameters until convergence is achieved, at which point iterative training stops. The expression for the loss function is:

[0057] L=λ1||PY s (t+Δt)-P s (t)|| 2 +λ2||vY s (t+Δt)-v s (t)|| 2 +λ3||aY s (t+Δt)-a s (t)|| 2

[0058] In the formula, L is the multi-task loss function, λ1 is the weight coefficient for position vector prediction, λ2 is the weight coefficient for velocity feature prediction, and λ3 is the weight coefficient for acceleration feature prediction.

[0059] Based on the above technical solutions, preferably, the execution module calculates the driving torque of each joint of the robotic hand according to the predicted trajectory information using an inverse kinematics algorithm, converts it into analog control signals, and sends the control signals to the slave hand manipulation module. The slave hand manipulation module executes corresponding actions according to the received control signals, thereby achieving accurate tracking of the master hand's movements.

[0060] Secondly, the present invention also provides a master-slave robot control method for nuclear decommissioning operations, implemented using a master-slave robot control system for nuclear decommissioning operations, comprising the following steps:

[0061] S1 collects real-time posture data of the master and slave robotic arms, as well as corresponding communication delay data and hot chamber environment data;

[0062] S2 preprocesses the collected data, extracts the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and stores them in the motion information database;

[0063] S3, based on a neural network, constructs a master-slave robot trajectory prediction model. It takes historical data and real-time posture data, corresponding communication delay data and hot chamber environment data from the motion information database as inputs to train the slave robot trajectory prediction model to output the predicted motion trajectory information of the slave robot.

[0064] S4 converts the current motion trajectory prediction information of the robotic arm into a control signal and sends it to the slave control module. The slave control module executes the corresponding action according to the received control signal to achieve accurate tracking of the master arm's actions.

[0065] The master-slave robot control system and method for nuclear decommissioning operations of the present invention have the following advantages over the prior art:

[0066] Beneficial effects:

[0067] (1) By collecting real-time posture data, communication delay data and hot chamber environment data of master and slave robot hands, the slave robot hand can maintain stable operation performance in different hot chamber environments, which enhances the environmental adaptability of the system. Based on the neural network, a trajectory prediction model of the slave robot hand is constructed, which enables the system to predict the movement trajectory of the slave robot hand more accurately, thereby improving the tracking accuracy and stability of the slave robot hand to the master hand's movements.

[0068] (2) The multi-scale feature extraction module is set up to extract spatial and temporal features through CNN and LSTM respectively, which can capture information in the input data more comprehensively, improve the expressive power of features, and enable the model to capture spatial and temporal features in the input data more comprehensively, thereby improving the accuracy of trajectory prediction.

[0069] (3) The hot chamber environment feature enhancement module is responsible for dynamically adjusting the feature weights according to the specific features of the hot chamber environment, thereby enhancing the model’s adaptability and prediction accuracy under different environmental conditions. Attached Figure Description

[0070] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0071] Figure 1 This is a structural block diagram of the master-slave robot control system for nuclear decommissioning operations of the present invention;

[0072] Figure 2 This is a flowchart of the master-slave robot control method for nuclear decommissioning operations according to the present invention;

[0073] Figure 3 This is a network model structure diagram of the master-slave robot control system for nuclear decommissioning operations of the present invention. Detailed Implementation

[0074] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0075] like Figure 1 As shown, the present invention provides a master-slave robotic arm control system for nuclear decommissioning operations. The system includes:

[0076] The acquisition module is used to acquire real-time posture data of the master and slave robotic arms, corresponding communication delay data, and hot chamber environment data, respectively. In this embodiment, the posture data of the master and slave robotic arms includes the angle vectors of each joint, and the hot chamber environment data includes radiation intensity, temperature, and humidity.

[0077] It should be noted that high-precision encoders and inertial measurement units are installed at each joint of the master and slave robotic arms to acquire joint angle vectors in real time. Data is transmitted using a CANopen bus to ensure transmission reliability in an electromagnetically shielded environment. The round-trip time of master-slave commands is calculated using UDP timestamps, and a hardware-level time synchronization protocol is deployed to achieve microsecond-level accuracy. Communication delay data refers to the time delay of command transmission between the master and slave robotic arms. In the hot chamber environment, radiation sensors and temperature and humidity sensors are installed. The radiation sensor has a range of 0-100 Sv / h and an error of ±5%, while the temperature and humidity sensors are resistant to high temperatures and radiation. Data is transmitted via optical fiber to avoid cable aging. Edge computing nodes are used for initial data packaging, and the data is transmitted to the main control room through a dual redundant channel of 5G private network and wired optical fiber.

[0078] The data processing module is used to preprocess the collected data, extract the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and store them in the motion information database.

[0079] In this embodiment, low-pass filtering is used to denoise the collected data and normalize it to obtain real-time posture standard information of the master and slave robot arms, communication delay standard information, and hot chamber environment data.

[0080] Standard feature information is extracted from the real-time posture standard information, communication delay standard information and hot chamber environment data of the master and slave robot arms to obtain the real-time posture standard feature information, communication delay standard feature information and hot chamber environment feature data of the master and slave robot arms;

[0081] A motion information database is constructed, which associates the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and stores them in the motion information database.

[0082] It should be noted that RC low-pass filters are integrated at the output of each sensor to suppress high-frequency noise, and adaptive Kalman filtering is used to process the attitude data. The process noise covariance is dynamically adjusted to cope with mechanical vibration interference. Sliding window mean filtering is used for communication delay data to eliminate the impact of sudden network jitter, with a window length of 50ms. Z-score normalization is used to normalize the attitude data to retain anomaly detection capability. The hot chamber environment data is normalized to the [0,1] interval to adapt to the neural network input layer. After feature extraction, the real-time attitude standard feature information, communication delay standard feature information, and hot chamber environment feature data of the master and slave robot are obtained. The extracted real-time attitude standard feature information, communication delay standard feature information, and hot chamber environment feature data of the master and slave robot are correlated to form a complete data record. The correlated data is stored in the motion information database to provide training data for the subsequent trajectory prediction model.

[0083] Specifically, standard feature information is extracted from the real-time posture standard information, communication delay standard information, and hot chamber environment data of the master and slave robot arms to obtain the real-time posture standard feature information, communication delay standard feature information, and hot chamber environment feature data of the master and slave robot arms, including:

[0084] Based on the joint angle vectors of the master and slave robotic arms, the position features of the master and slave robotic arms are calculated and obtained, where the position feature expression is:

[0085] p m (t)=f kin (q m (t))

[0086] p s (t)=f kin (q s (t))

[0087] In the formula, P m (t) represents the position vector of the main robotic arm at time t, q m (t) represents the joint angle vector of the main robotic arm at time t, P s q(t) represents the position vector of the robot arm at time t. s (t) represents the joint angle vector of the robot arm at time t, f kin It is the forward kinematics function;

[0088] Taking the first derivative of the position features of the master and slave robotic arms with respect to time, we obtain the velocity features v of the master robotic arm. m (t) and the velocity characteristics v from the robot arm s (t); Taking the second derivative of the position characteristics of the master and slave robotic arms with respect to time, we obtain the acceleration characteristics a of the master robotic arm. m (t) and the acceleration characteristics from the robotic arm a s (t);

[0089] Specifically, by taking the first derivative of the position features of the master and slave robotic arms with respect to time, the velocity features v of the master robotic arm are obtained. m (t) and the velocity characteristics v from the robot arm s (t), the expression is:

[0090]

[0091]

[0092] In the formula, v m (t) represents the velocity characteristic of the main robot arm at time t, v s (t) represents the velocity characteristic of the robotic arm at time t;

[0093] Specifically, by taking the second derivative of the position features of the master and slave robotic arms with respect to time, the acceleration feature 'a' of the master robotic arm is obtained. m (t) and the acceleration characteristics from the robotic arm a s (t), the expression is:

[0094]

[0095] In the formula, a m (t) represents the acceleration characteristic of the main robotic arm at time t, a s (t) represents the acceleration characteristic of the robot arm at time t;

[0096] A fixed time step is preset, and communication delay data is acquired at each time point within that fixed time step. The average delay and delay fluctuation within the fixed time step are calculated to obtain standard characteristic information of communication delay. The delay fluctuation expression is as follows:

[0097]

[0098] Where σΔt is the delayed fluctuation, Δt i The communication delay data at the current time point, Δt avg The average delay is N, and the number of time points with a fixed time step is N.

[0099] The radiation intensity, temperature, and humidity at each time point within a fixed time step are obtained. The average radiation intensity, average temperature, and average humidity within the fixed time step are calculated to obtain the environmental characteristic data of the hot chamber.

[0100] It should be noted that by calculating the positional features of the master and slave robots using the forward kinematics function and then taking the first and second derivatives of these features to obtain velocity and acceleration characteristics, a more comprehensive description of the robot's motion state can be achieved. These characteristics provide rich information for the trajectory prediction model, enabling the model to more accurately capture the robot's motion trends and dynamic changes, thereby improving the accuracy of trajectory prediction. Furthermore, extracting hot chamber environment feature data can reflect the physical characteristics of the robot's operating environment. Adding these environmental features to the trajectory prediction model allows the model to better adapt to robot motion under different environmental conditions, further improving prediction accuracy. Simultaneously, calculating the average delay and delay fluctuation within a fixed time step yields standard feature information on communication delay, which helps the model understand the distribution and range of changes in communication delay. This allows the model to consider the impact of delay when predicting the trajectory, enhancing the system's robustness to communication delay. By pre-setting a fixed time step to acquire communication delay data and hot chamber environment data, the periodicity and regularity of data acquisition can be ensured, facilitating more stable system operation and reducing delays and fluctuations caused by irregular data acquisition.

[0101] The motion prediction module is used to build a robot arm trajectory prediction model based on a neural network. It takes historical data and real-time posture data, as well as corresponding communication delay data and hot chamber environment data from the motion information database as input, and trains the robot arm trajectory prediction model to output the motion trajectory prediction information of the robot arm.

[0102] like Figure 3 As shown, in this embodiment, the motion prediction module constructs a robot arm trajectory prediction model based on a neural network. The network structure of the robot arm trajectory prediction model includes an input layer, a multi-scale feature extraction module, a hot chamber environment feature enhancement module, a Transformer encoder, and a multi-task output module. The input layer is used to receive real-time posture standard feature information, communication delay standard feature information, and hot chamber environment feature data of the master and slave robot arms. The output of the input layer is connected to the input of the multi-scale feature extraction module to extract the spatial and temporal features of the input data. The output of the multi-scale feature extraction module is connected to the input of the hot chamber environment feature enhancement module to dynamically adjust the feature weights according to the hot chamber environment feature data. The output of the hot chamber environment feature enhancement module is connected to the input of the Transformer encoder to capture global dependencies through a multi-head self-attention mechanism. The output of the Transformer encoder is connected to the input of the multi-task output module to predict and output the position vector, velocity features, and acceleration features of the operator arm, respectively.

[0103] It should be noted that the multi-scale feature extraction module enables the model to capture spatial and temporal features in the input data more comprehensively, thereby improving the accuracy of trajectory prediction. The hot chamber environment feature enhancement module can dynamically adjust feature weights according to environmental features, enabling the model to maintain high prediction accuracy under different environmental conditions. The Transformer encoder captures global dependencies through a multi-head self-attention mechanism, making the model more robust to noise and outliers in the input sequence. The multi-task output module can provide more comprehensive and stable trajectory prediction information by predicting position, velocity, and acceleration separately, further enhancing the robustness of the model. This model structure can handle feature data from different sources and at different scales, has strong generalization ability, and can efficiently process input data and output prediction results.

[0104] In this embodiment, the multi-scale feature extraction module includes a CNN network extraction unit, an LSTM network extraction unit, and a feature fusion unit.

[0105] The CNN network extraction unit is used to extract spatial features from the input data, and its expression is:

[0106] F cnn =CNN(X)

[0107] In the formula, F cnn X represents the spatial features extracted by the CNN network extraction unit, and X represents the input master-slave robot real-time pose standard feature information, communication delay standard feature information, and hot chamber environment feature data.

[0108] The LSTM network extraction unit is used to extract the temporal features of the input data, expressed as:

[0109] F lstm =LSTM(X)

[0110] In the formula, F lstm Temporal features extracted by the LSTM module;

[0111] The feature fusion unit is used to fuse the features output by the CNN network extraction unit and the features output by the LSTM network extraction unit to obtain fused feature data, expressed as:

[0112]

[0113] In the formula, F fusion This is the result of splicing the spatial features extracted by the CNN network extraction unit and the temporal features extracted by the LSTM module.

[0114] Understandably, CNN (Convolutional Neural Network) can automatically learn spatial hierarchical features in input data through structures such as convolutional layers and pooling layers. In trajectory prediction, CNN can capture the standard features of the real-time posture of the master and slave robot and the spatial patterns in the hot chamber environment feature data. LSTM (Long Short-Term Memory) network can handle long-term dependencies in sequence data. In trajectory prediction, LSTM can capture the temporal dynamics of the master and slave robot motion sequence, such as the changing trends of speed and acceleration, as well as the temporal characteristics of communication delay. Typically, a concatenation method is used to fuse the spatial features extracted by the CNN network and the temporal features extracted by the LSTM network to obtain fused feature data. Then, by extracting spatial and temporal features through CNN and LSTM respectively, the information in the input data can be captured more comprehensively, improving the expressive power of the features.

[0115] In this embodiment, the hot chamber environment feature enhancement module includes a preset radiation resistance weighting matrix and a temperature and humidity weighting matrix.

[0116] The radiation resistance weight matrix is ​​obtained by dynamically adjusting the basic weight matrix based on the radiation intensity. The expression for this matrix is:

[0117]

[0118] In the formula, W rad (t) is the radiation resistance weight matrix. Here, α is the radiation resistance weighting matrix, γ(t) is the radiation intensity influence coefficient, and γ(t) is the radiation intensity at time point t.

[0119] The temperature and humidity weighting matrix is ​​obtained by dynamically adjusting the basic matrix based on temperature and humidity, and its expression is:

[0120]

[0121] In the formula, W dynamic (t) is the temperature and humidity weight matrix. Let β be the fundamental matrix for temperature and humidity weights. T β is the influence coefficient of temperature. H Let T(t) be the temperature at time t, and H(t) be the humidity at time t.

[0122] The fused feature data is enhanced based on the radiation resistance weight matrix and the temperature and humidity weight matrix, resulting in enhanced feature data, expressed as follows:

[0123] F dynamic =F fusion· W rad (t) · W dynamic (t)

[0124] In the formula, F dynamic To enhance feature data.

[0125] It should be noted that the hot chamber environment feature enhancement module is responsible for dynamically adjusting feature weights based on the specific characteristics of the hot chamber environment, thereby enhancing the model's adaptability and prediction accuracy under different environmental conditions. Among them, the preset radiation resistance weight matrix and temperature and humidity weight matrix are used. The radiation resistance weight matrix represents the degree of contribution of each feature to the radiation resistance capability under the basic conditions, i.e., no radiation or low radiation intensity. The temperature and humidity weight matrix represents the degree of influence of temperature and humidity on each feature under the basic conditions.

[0126] Understandably, radiation intensity is a crucial factor affecting the safety of the hot chamber environment. As radiation intensity increases, the contribution of the robot arm to radiation resistance may change; therefore, it is necessary to dynamically adjust the radiation resistance weight matrix based on radiation intensity. Temperature and humidity are also important factors affecting the hot chamber environment, influencing robot arm performance, material stability, and sensor accuracy. Thus, it is necessary to dynamically adjust the temperature and humidity weight matrix based on temperature and humidity. By using the dynamically adjusted radiation resistance and temperature / humidity weight matrices, the fused feature data is weighted, allowing the model to better adapt to hot chamber environments under different radiation intensities, temperatures, and humidity conditions. This facilitates more accurate prediction of the robot arm's trajectory, significantly improving the model's prediction accuracy and stability.

[0127] In this embodiment, the multi-task output module includes outputting the position vector, velocity characteristics, and acceleration characteristics of the operator.

[0128] The expression for the predicted location vector is:

[0129] PY s (t+Δt)=W p h+b p

[0130] In the formula, PY s (t+Δt) is the position vector predicted from the operator, W p To predict the weights of the position vector, b p The bias for predicting the position vector;

[0131] The expression for the velocity characteristic is:

[0132] vY s (t+Δt)=W v h+b v

[0133] In the formula, vY s (t+Δt) represents the velocity characteristic predicted from the operator, W v For the weights of the predicted velocity feature, b v The bias used to predict velocity features;

[0134] The expression for the acceleration characteristic is:

[0135] aY s (t+Δt)=W a h+b a

[0136] In the formula, aY s (t+Δt) represents the acceleration characteristic predicted from the operator, W a To predict the weights of acceleration features, b a The bias used to predict acceleration characteristics.

[0137] It should be noted that the position vector is key information describing the operator's position in space. The predicted position vector is obtained by linearly transforming the augmented feature data through a fully connected layer. The velocity feature is key information describing the speed and direction of the operator's movement. The predicted velocity feature is obtained by linearly transforming the augmented feature data through a fully connected layer. The acceleration feature is key information describing the rate of change of the operator's velocity. The predicted acceleration feature is obtained by linearly transforming the augmented feature data through a fully connected layer. This module can simultaneously predict the operator's position vector, velocity feature, and acceleration feature, providing comprehensive information for the control of the operator.

[0138] In this embodiment, the robot arm trajectory prediction model further includes setting a loss function, and using the Adam optimization algorithm to iteratively train the model based on the loss function to adjust the model parameters until convergence is achieved, at which point iterative training stops. The expression for the loss function is:

[0139] L=λ1||PY s (t+Δt)-P s (t)|| 2 +λ2||vY s (t+Δt)-v s (t)|| 2 +λ3||aY s (t+Δt)-a s (t)|| 2

[0140] In the formula, L is the multi-task loss function, λ1 is the weight coefficient for position vector prediction, λ2 is the weight coefficient for velocity feature prediction, and λ3 is the weight coefficient for acceleration feature prediction.

[0141] It should be noted that λ1||PY s (t+Δt)-P s (t)|| 2 This is the loss for predicting the location vector, and this term measures the predicted location vector PY. s (t+Δt) and the true position vector P s The mean square error between (t) and λ1 is the weighting coefficient of the position vector prediction, used to adjust the proportion of this term in the total loss; λ2||vY s (t+Δt)-v s (t)|| 2 This is the loss for velocity feature prediction; this term measures the predicted velocity feature vY. s (t+Δt) and the true velocity characteristic v s The mean square error between (t); λ3||aY s (t+Δt)-a s (t)|| 2 For the acceleration feature prediction loss, this term measures the predicted acceleration feature aY. s (t+Δt) and the true acceleration characteristic a s The mean square error between (t) and the multi-task loss function simultaneously considers the prediction errors of position, velocity and acceleration, enabling the model to learn more comprehensively from the motion law of the robot hand. By adjusting the weight coefficients λ1, λ2 and λ3, the proportion of different prediction tasks in the total loss can be flexibly controlled, thereby adapting to different application needs.

[0142] The execution module converts the predicted motion trajectory information of the current robotic arm into control signals and sends them to the slave control module. The slave control module executes corresponding actions according to the received control signals to accurately follow the movements of the master arm.

[0143] In this embodiment, the execution module calculates the driving torque of each joint of the robotic hand based on the predicted trajectory information using an inverse kinematics algorithm, converts it into analog control signals, and sends the control signals to the slave hand manipulation module. The slave hand manipulation module executes corresponding actions according to the received control signals, thereby achieving accurate tracking of the master hand's movements.

[0144] When it's necessary to explain, the execution module first receives the current-moment trajectory prediction information from the robot's trajectory prediction model, including position vector, velocity characteristics, and acceleration characteristics. Based on the predicted trajectory information, the execution module calculates the driving torque of each joint of the robot using an inverse kinematics algorithm. The execution module converts the calculated joint driving torque into analog control signals to drive each joint of the robot. The execution module then sends the converted analog control signals to the slave manipulator module, so that the slave manipulator module can execute corresponding actions based on these signals. This gives the system a high degree of flexibility. By adjusting the trajectory prediction model and inverse kinematics algorithm, it can adapt to different master-slave robot configurations and application scenarios, enabling the slave robot to flexibly follow the master robot's movements.

[0145] like Figure 2 As shown, in a second aspect, the present invention also provides a master-slave robot control method for nuclear decommissioning operations, implemented using a master-slave robot control system for nuclear decommissioning operations, comprising the following steps:

[0146] S1 collects real-time posture data of the master and slave robotic arms, as well as corresponding communication delay data and hot chamber environment data;

[0147] S2 preprocesses the collected data, extracts the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and stores them in the motion information database;

[0148] S3, based on a neural network, constructs a master-slave robot trajectory prediction model. It takes historical data and real-time posture data, corresponding communication delay data and hot chamber environment data from the motion information database as inputs to train the slave robot trajectory prediction model to output the predicted motion trajectory information of the slave robot.

[0149] S4 converts the current motion trajectory prediction information of the robotic arm into a control signal and sends it to the slave control module. The slave control module executes the corresponding action according to the received control signal to achieve accurate tracking of the master arm's actions.

[0150] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0151] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the system and modules described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0152] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0153] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0154] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0155] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0156] Furthermore, it should be noted that in the system and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof. This is something that those skilled in the art can achieve by using their basic programming skills after reading the description of the present invention.

[0157] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing system. The computing system can be a known general-purpose system. Therefore, the object of the present invention can also be achieved simply by providing a program product containing program code for implementing the method or apparatus. That is, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future. It should also be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent to the present invention. Furthermore, the steps for performing the above series of processes can naturally be performed in the order described, but are not necessarily required to be performed in chronological order. Some steps can be performed in parallel or independently of each other.

[0158] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A master-slave robotic arm control system for nuclear decommissioning operations, characterized in that, The system includes: The acquisition module is used to collect real-time posture data of the master and slave robotic arms, as well as corresponding communication delay data and hot chamber environment data. The data processing module is used to preprocess the collected data, extract the real-time posture standard feature information, communication delay standard feature information and hot chamber environment feature data of the master and slave robot arms, and store them into the motion information database. The radiation intensity, temperature and humidity at each time point of a fixed time step are obtained, and the average radiation intensity, average temperature and average humidity in the fixed time step are calculated to obtain the thermal chamber environmental characteristic data. The motion prediction module is used to build a robot arm trajectory prediction model based on a neural network. It takes historical data and real-time posture data, as well as corresponding communication delay data and hot chamber environment data from the motion information database as inputs to train the robot arm trajectory prediction model and output the motion trajectory prediction information of the robot arm. The network structure of the robotic arm trajectory prediction model includes an input layer, a multi-scale feature extraction module, a hot chamber environment feature enhancement module, a Transformer encoder, and a multi-task output module. The multi-scale feature extraction module includes CNN network extraction units, LSTM network extraction units, and feature fusion units. The hot chamber environment feature enhancement module includes a preset radiation resistance weight matrix and a temperature and humidity weight matrix. The radiation resistance weight matrix is ​​obtained by dynamically adjusting the basic weight matrix based on the radiation intensity, and its expression is as follows: In the formula, W rad (t) is the radiation resistance weight matrix. Here, α is the radiation resistance weighting matrix, γ(t) is the radiation intensity influence coefficient, and γ(t) is the radiation intensity at time point t. The temperature and humidity weighting matrix is ​​obtained by dynamically adjusting the basic matrix based on temperature and humidity, and its expression is: In the formula, W dynamic (t) is the temperature and humidity weight matrix. The fundamental matrix for temperature and humidity weights, β T β is the influence coefficient of temperature. H Let T(t) be the temperature at time t, and H(t) be the humidity at time t. The fused feature data is enhanced based on the radiation resistance weight matrix and the temperature and humidity weight matrix, resulting in enhanced feature data, expressed as: F dynamic =F fusion· W rad (t) · W dynamic (t) In the formula, F fusion F is the concatenation result of spatial features extracted by the CNN network extraction unit and temporal features extracted by the LSTM module. dynamic To enhance feature data; The execution module is used to convert the predicted motion trajectory information of the current robotic arm into control signals and send them to the slave control module. The slave control module executes corresponding actions according to the received control signals to achieve accurate tracking of the master arm's movements.

2. The master-slave robot control system for nuclear decommissioning operations as described in claim 1, characterized in that: The preprocessing of the collected data yields real-time posture standard feature information of the master and slave robotic arms, communication delay standard feature information, and hot chamber environment feature data, including: The posture data of the master and slave robotic arms includes the angle vectors of each joint, and the hot chamber environment data includes radiation intensity, temperature and humidity; The collected data was denoised and normalized using a low-pass filter to obtain real-time posture standard information of the master and slave robot arms, communication delay standard information, and hot chamber environment data. Standard feature information is extracted from the real-time posture standard information, communication delay standard information and hot chamber environment data of the master and slave robot arms to obtain the real-time posture standard feature information, communication delay standard feature information and hot chamber environment feature data of the master and slave robot arms; A motion information database is constructed, which associates the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and stores them in the motion information database.

3. The master-slave robot control system for nuclear decommissioning operations as described in claim 2, characterized in that: The extraction of standard feature information from the real-time posture standard information, communication delay standard information, and hot chamber environment data of the master and slave robot arms yields the real-time posture standard feature information, communication delay standard feature information, and hot chamber environment feature data of the master and slave robot arms, including: Based on the joint angle vectors of the master and slave robotic arms, the position features of the master and slave robotic arms are calculated and obtained, where the position feature expression is: p m (t)=f kin (q m (t)) p s (t)=f kin (q s (t)) In the formula, P m (t) represents the position vector of the main robotic arm at time t, q m (t) represents the joint angle vector of the main robotic arm at time t, P s q(t) represents the position vector of the robot arm at time t. s (t) represents the joint angle vector of the robot arm at time t, f kin It is the forward kinematics function; Taking the first derivative of the position features of the master and slave robotic arms with respect to time, we obtain the velocity features v of the master robotic arm. m (t) and the velocity characteristics v from the robot arm s (t); Taking the second derivative of the position characteristics of the master and slave robotic arms with respect to time, we obtain the acceleration characteristics a of the master robotic arm. m (t) and the acceleration characteristics from the robotic arm a s (t); A fixed time step is preset, and communication delay data is acquired at each time point within that fixed time step. The average delay and delay fluctuation within the fixed time step are calculated to obtain standard characteristic information of communication delay. The delay fluctuation expression is as follows: Where σΔt is the delayed fluctuation, Δt i The communication delay data at the current time point, Δt avg The average delay is N, and N is the number of time points with a fixed time step.

4. The master-slave robot control system for nuclear decommissioning operations as described in claim 3, characterized in that: The motion prediction module constructs a trajectory prediction model for the robotic arm based on a neural network. The input layer receives real-time posture standard feature information, communication delay standard feature information, and hot chamber environment feature data from the master and slave robotic arms. The output of the input layer is connected to the input of the multi-scale feature extraction module to extract the spatial and temporal features of the input data. The output of the multi-scale feature extraction module is connected to the input of the hot chamber environment feature enhancement module to dynamically adjust feature weights based on the hot chamber environment feature data. The output of the hot chamber environment feature enhancement module is connected to the input of the Transformer encoder to capture global dependencies through a multi-head self-attention mechanism. The output of the Transformer encoder is connected to the input of the multi-task output module to predict the position vector, velocity features, and acceleration features of the slave robotic arm, respectively.

5. The master-slave robot control system for nuclear decommissioning operations as described in claim 4, characterized in that: The CNN network extraction unit is used to extract spatial features from the input data, expressed as: F cnn =CNN(X) In the formula, F cnn X represents the spatial features extracted by the CNN network extraction unit, and X represents the input master-slave robot real-time pose standard feature information, communication delay standard feature information, and hot chamber environment feature data. The LSTM network extraction unit is used to extract the temporal features of the input data, expressed as: F lstm =LSTM(X) In the formula, F lstm Temporal features extracted by the LSTM module; The feature fusion unit is used to fuse the features output by the CNN network extraction unit and the features output by the LSTM network extraction unit to obtain fused feature data, expressed as: In the formula, F fusion This is the result of splicing the spatial features extracted by the CNN network extraction unit and the temporal features extracted by the LSTM module.

6. The master-slave robot control system for nuclear decommissioning operations as described in claim 5, characterized in that: The multi-task output module includes outputting the operator's position vector, velocity characteristics, and acceleration characteristics, wherein... The expression for the predicted location vector is: PY s (t+Δt)=W p h+b p In the formula, PY s (t+Δt) is the position vector predicted from the operator, W p To predict the weights of the position vector, b p The bias for predicting the position vector; The expression for the velocity characteristic is: vY s (t+Δt)=W v h+b v In the formula, vY s (t+Δt) represents the velocity characteristic predicted from the operator, W v For the weights of the predicted velocity feature, b v The bias used to predict velocity features; The expression for the acceleration characteristic is: aY s (t+Δt)=W a h+b a In the formula, aY s (t+Δt) represents the acceleration characteristic predicted from the operator, W a To predict the weights of acceleration features, b a The bias used to predict acceleration characteristics.

7. The master-slave robot control system for nuclear decommissioning operations as described in claim 6, characterized in that: The robot arm trajectory prediction model also includes setting a loss function, and using the Adam optimization algorithm to iteratively train the model based on the loss function to adjust the model parameters until convergence is achieved, at which point iterative training stops. The expression for the loss function is: L=λ1||PY s (t+Δt)-P s (t)|| 2 +λ2||vY s (t+Δt)-v s (t)|| 2 +λ3||aY s (t+Δt)-a s (t)|| 2 In the formula, L is the multi-task loss function, λ1 is the weight coefficient for position vector prediction, λ2 is the weight coefficient for velocity feature prediction, and λ3 is the weight coefficient for acceleration feature prediction.

8. The master-slave robot control system for nuclear decommissioning operations as described in claim 1, characterized in that: The execution module calculates the driving torque of each joint of the robotic arm based on the predicted trajectory information using an inverse kinematics algorithm, converts it into analog control signals, and sends the control signals to the slave hand manipulation module. The slave hand manipulation module executes corresponding actions according to the received control signals, thereby achieving accurate tracking of the master hand's movements.

9. A method for controlling a master-slave robotic arm for nuclear decommissioning operations, implemented using the master-slave robotic arm control system for nuclear decommissioning operations as described in any one of claims 1-8, characterized in that, Includes the following steps: S1 collects real-time posture data of the master and slave robotic arms, as well as corresponding communication delay data and hot chamber environment data; S2 preprocesses the collected data, extracts the real-time posture standard feature information of the master and slave robot arms, the communication delay standard feature information, and the hot chamber environment feature data, and stores them in the motion information database; S3, based on a neural network, constructs a master-slave robot trajectory prediction model. It takes historical data and real-time posture data, corresponding communication delay data and hot chamber environment data from the motion information database as inputs to train the slave robot trajectory prediction model to output the predicted motion trajectory information of the slave robot. S4 converts the current motion trajectory prediction information of the robotic arm into a control signal and sends it to the slave control module. The slave control module executes the corresponding action according to the received control signal to achieve accurate tracking of the master arm's actions.

Images