Method, system and equipment for determining position and posture of space flexible mechanical arm and medium

Abstract

The invention discloses a space flexible mechanical arm pose determination method, system and device and a medium, and relates to the technical field of space robot control and flexible electronic perception.The method comprises the steps that after a flexible skin integrated with a distributed flexible sensing array is attached to the surface of a space flexible mechanical arm body, the flexible skin is attached to the surface of the space flexible mechanical arm body; multi-channel sensing electric signals caused by movement of the mechanical arm are collected in real time; performing synchronous acquisition, analog-to-digital conversion and preprocessing on the multi-channel sensing electric signal to obtain a preprocessed sensing electric signal; feature decoupling processing is conducted on the preprocessed sensing electric signals, and target feature vectors representing the motion state of the mechanical arm are obtained; inputting the target feature vector into a trained pose mapping model, and outputting a preliminary pose estimation value; and state filtering optimization is conducted on the initial pose estimation value, the final pose estimation result of the space flexible mechanical arm body is output, real-time high-precision determination of the pose of the space flexible mechanical arm body is achieved, and the space operation precision and safety are improved.

Description

Technical Field

[0001] This application relates to the field of space robot control and flexible electronic sensing technology, and in particular to a method, system, device and medium for determining the pose of a flexible space robotic arm. Background Technology

[0002] Space-based flexible robotic arms possess high specific stiffness, lightweight, and deployable characteristics during on-orbit operation. However, their high structural flexibility, significant vibration, and strong inertial coupling lead to error accumulation issues in traditional attitude calculations based on joint encoders and dynamic models. Environmental factors such as microgravity, thermal expansion and contraction, and structural springback further cause model mismatch, making it difficult to accurately estimate the body's motion state. Related attitude detection methods also exhibit poor accuracy. Therefore, a method for determining the pose of space-based flexible robotic arms with improved accuracy is lacking. Summary of the Invention

[0003] The purpose of this application is to provide a method, system, device and medium for determining the pose of a spatial flexible robotic arm, which can realize real-time high-precision determination of the pose of the spatial flexible robotic arm body, thereby improving the accuracy and safety of spatial operations.

[0004] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a method for determining the pose of a spatial flexible robotic arm, including: After attaching a flexible skin with an integrated distributed flexible sensor array to the surface of the space flexible robotic arm, multi-channel sensor electrical signals caused by the movement of the robotic arm are collected in real time. The multi-channel sensing electrical signals are synchronously acquired, converted from analog to digital and preprocessed to obtain preprocessed sensing electrical signals. The preprocessed sensor electrical signal is subjected to feature decoupling processing to obtain the target feature vector characterizing the motion state of the robotic arm. The target feature vector is input into the trained pose mapping model, and a preliminary pose estimate is output. The preliminary pose estimate includes the end position and the end pose quaternion. The trained pose mapping model is trained by using the sample feature vector of the sample sensing electrical signal of the spatial flexible manipulator body as input and the end pose of the spatial flexible manipulator body as the label. The initial pose estimation value is optimized by state filtering to output the final pose estimation result of the spatial flexible robotic arm body.

[0005] Optionally, the distributed flexible sensing array includes multiple flexible strain sensing units based on microcrack structures; the fabrication of the flexible strain sensing units includes: forming a microcrack array on a polymer substrate coated with single-walled carbon nanotubes, regularizing the microcracks through thermal shrinkage, encapsulating them with a flexible material, and removing the original substrate.

[0006] Optionally, the conductive nanomaterial used in the flexible skin is silver nanowire, polydimethylsiloxane composite film, or graphene composite film; the polymer substrate is polystyrene; and the encapsulation material is Ecoflex material.

[0007] Optionally, the preprocessed sensing electrical signal is subjected to feature decoupling processing to obtain a target feature vector characterizing the motion state of the robotic arm. Specifically, this includes using a principal component analysis algorithm to perform feature decoupling processing on the preprocessed sensing electrical signal to obtain a target feature vector characterizing the motion state of the robotic arm.

[0008] Optionally, the pose mapping model is a long short-term memory neural network model.

[0009] Optionally, the preliminary pose estimation value is optimized by state filtering to output the final pose estimation result of the spatial flexible robotic arm body, specifically including: Using the extended Kalman filter algorithm, the initial pose estimate is optimized by state filtering. The initial pose estimate is used as the observation value, and a state equation containing the smooth motion constraint of the robotic arm is established. Through recursive prediction and update steps, the final pose estimate of the spatial flexible robotic arm body is output.

[0010] Optionally, before inputting the target feature vector into the trained pose mapping model, the spatial flexible robotic arm pose determination method further includes: Obtain a sample set; the sample set includes sample feature vectors of several sample sensing electrical signals of the spatial flexible manipulator body and the end pose of the spatial flexible manipulator body corresponding to each sample sensing electrical signal. The pose mapping model is trained using a sample set to obtain a trained pose mapping model.

[0011] Secondly, this application provides a spatial flexible robotic arm pose determination system, comprising: The flexible electronic skin sensing module is used to collect multi-channel sensing electrical signals caused by the movement of the robotic arm in real time after the flexible skin integrating a distributed flexible sensor array is attached to the surface of the flexible robotic arm body. The signal acquisition and synchronization module is used to synchronously acquire, convert analog to digital and preprocess the multi-channel sensing electrical signals to obtain preprocessed sensing electrical signals. The feature decoupling processing module is used to perform feature decoupling processing on the preprocessed sensing electrical signal to obtain the target feature vector characterizing the motion state of the robotic arm. The preliminary pose estimation module is used to input the target feature vector into the trained pose mapping model and output a preliminary pose estimate. The preliminary pose estimate includes the end position and end pose quaternion. The trained pose mapping model is trained by using the sample feature vector of the sample sensing electrical signal of the spatial flexible manipulator body as input and the end pose of the spatial flexible manipulator body as label. The final pose estimation result determination module is used to perform state filtering optimization on the preliminary pose estimation value and output the final pose estimation result of the spatial flexible robotic arm body.

[0012] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the steps of the above-described spatial flexible robotic arm pose determination method.

[0013] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described spatial flexible robotic arm pose determination method.

[0014] According to the specific embodiments provided in this application, this application has the following technical effects: The inventors discovered that relevant attitude detection methods include: encoder and gyroscope fusion cannot reflect the elastic deformation of the linkage; visual measurement methods are affected by lighting and occlusion, and are not suitable for shadows or confined spaces; structural strain gauges have high rigidity and are not suitable for flexible surfaces; model inversion estimation requires complex modeling and has poor real-time performance. Based on this, this application provides a method, system, device, and medium for determining the pose of a spatial flexible robotic arm. By deploying a flexible electronic skin sensor array (a flexible skin integrating a distributed flexible sensor array) on the surface of the spatial flexible robotic arm, the motion information of the spatial flexible robotic arm can be directly perceived from the structural surface. Combined with dynamic feature extraction and fusion algorithms, feature extraction and fusion are performed based on the sensing signals collected by the flexible electronic skin sensor array to improve the accuracy of data analysis and prediction models. Then, a trained pose mapping model is used to perform preliminary pose estimation, and state filtering optimization is applied to the preliminary pose estimate to further improve the estimation accuracy of the spatial flexible robotic arm's pose. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is an application environment diagram of a spatial flexible robotic arm pose determination method according to an embodiment of this application.

[0017] Figure 2 This is a flowchart illustrating a spatial flexible robotic arm pose determination method provided in an embodiment of this application.

[0018] Figure 3 This is a schematic diagram illustrating the specific process of a spatial flexible robotic arm pose determination method provided in an embodiment of this application.

[0019] Figure 4 This is a schematic diagram of the spatial flexible robotic arm structure and flexible skin layout provided in one embodiment of this application.

[0020] Figure 5 This is a schematic diagram of the flexible strain sensing unit structure and conductive path provided in an embodiment of this application.

[0021] Figure 6 This is a schematic diagram of a long short-term memory neural network structure provided in an embodiment of this application.

[0022] Figure 7 Provided for an embodiment of this application Error distribution curve between the orientation sensor position and the actual position.

[0023] Figure 8 This is a schematic diagram of a spatial flexible robotic arm pose determination system provided in an embodiment of this application.

[0024] Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] The spatial flexible robotic arm pose determination method provided in this application embodiment can be applied to, for example, Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on other servers. Terminal 102 can send the multi-channel sensor signals to be processed to server 104. After receiving the multi-channel sensor signals, server 104 performs synchronous acquisition, analog-to-digital conversion, and preprocessing on the multi-channel sensor signals to obtain preprocessed sensor signals. It then performs feature decoupling processing on the preprocessed sensor signals to obtain target feature vectors representing the motion state of the robotic arm. These target feature vectors are input into a trained pose mapping model, outputting a preliminary pose estimate. The preliminary pose estimate is then optimized by state filtering to output the final pose estimate result of the spatial flexible robotic arm. Server 104 can feed back the final pose estimate result for the multi-channel sensor signals to terminal 102. In addition, in some embodiments, the spatial flexible robotic arm pose determination method can also be implemented by the server 104 or the terminal 102 alone. For example, the terminal 102 can directly estimate the pose motion state of the robotic arm based on the multi-channel sensor electrical signals to be processed, or the server 104 can obtain the multi-channel sensor electrical signals to be processed from the data storage system and estimate the pose motion state of the robotic arm based on the multi-channel sensor electrical signals to be processed.

[0028] The terminal 102 can be, but is not limited to, various desktop computers, laptops, tablets, and IoT devices. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers, or it can be a cloud server.

[0029] In one exemplary embodiment, such as Figure 2 As shown, a method for determining the pose of a spatial flexible robotic arm is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 205.

[0030] Step 201: After attaching the flexible skin with an integrated distributed flexible sensor array to the surface of the space flexible robotic arm, the multi-channel sensor electrical signals caused by the movement of the robotic arm are collected in real time.

[0031] Step 202: Synchronously acquire, convert analog to digital and preprocess the multi-channel sensing electrical signals to obtain preprocessed sensing electrical signals.

[0032] Step 203: Perform feature decoupling processing on the preprocessed sensing electrical signal to obtain the target feature vector characterizing the motion state of the robotic arm.

[0033] Step 204: Input the target feature vector into the trained pose mapping model and output a preliminary pose estimate; the preliminary pose estimate includes the end-effector position and end-effector pose quaternion; the trained pose mapping model is trained by using the sample feature vector of the sample sensing electrical signal of the spatial flexible manipulator body as input and the end-effector pose of the spatial flexible manipulator body as label.

[0034] Step 205: Perform state filtering optimization on the preliminary pose estimation value and output the final pose estimation result of the spatial flexible robotic arm body.

[0035] By implementing steps 201 to 205 above, a flexible electronic skin sensor array (a flexible skin integrating a distributed flexible sensor array) is deployed on the surface of the flexible spatial manipulator. This allows for direct perception of the manipulator's motion information from the structural surface. Combined with dynamic feature extraction and fusion algorithms, feature extraction and fusion are performed based on the sensor signals collected by the flexible electronic skin sensor array, improving the accuracy of the data analysis and prediction model. Then, a trained pose mapping model is used to perform a preliminary pose estimation, and the preliminary pose estimation value is further optimized by state filtering, which can further improve the estimation accuracy of the flexible spatial manipulator's pose.

[0036] The method provided in this embodiment can be divided into five parts: a flexible electronic sensing module, a signal acquisition and synchronization module, a feature extraction and data fusion module, a data training module, and a state determination and output module. Figure 3 As shown, the flexible electronic sensing module is attached to the surface of the flexible spatial manipulator and conformally bends with the movement of the manipulator, thereby generating multi-channel sensing electrical signals. The signal acquisition and synchronization module connects the flexible electronic skin sensing module to the computer platform and converts the multi-channel sensing electrical signals into computer-processable data. The feature extraction and data fusion module decouples the data that are related or mutually influential from different sensors in the preprocessed sensing electrical signals, improving the accuracy of subsequent data training. The data training module trains the sensing electrical signals and the corresponding pose of the flexible spatial manipulator using deep learning neural network methods, thereby establishing a nonlinear mapping relationship between the sensing electrical signals and the corresponding pose data of the flexible spatial manipulator. The state determination and output module outputs the pose sensing results of the flexible spatial manipulator and connects them with the application, and uses a filtering algorithm to further reduce noise. The final output pose estimation result (state parameters) includes the end position and end pose quaternion of the flexible spatial manipulator.

[0037] The flexible electronic skin sensing module consists of a flexible skin and a distributed flexible sensor array. The functions of the flexible electronic skin sensing module include: conforming to the surface of the spatial flexible robotic arm and conformally bending with the movement of the spatial flexible robotic arm, thereby generating multi-channel sensing electrical signals.

[0038] The distributed flexible sensing array comprises multiple flexible strain sensing units based on microcrack structures. The fabrication of these flexible strain sensing units includes: forming a microcrack array on a polymer substrate coated with single-walled carbon nanotubes; regularizing the microcracks through thermal shrinkage; encapsulating the microcracks with a flexible material; and removing the original substrate. The conductive nanomaterials used for the flexible skin are silver nanowires, polydimethylsiloxane composite films, or graphene composite films; the polymer substrate is polystyrene; and the encapsulation material is Ecoflex.

[0039] In this embodiment, the spatial flexible robotic arm structure and flexible skin layout are as follows: Figure 4 As shown in the diagram. The flexible skin is made of a flexible conductive composite material, commonly including silver nanowire / polydimethylsiloxane composite films or graphene composite films. These materials possess both good conductivity and flexibility, allowing them to bend synchronously with the deformation of the robotic arm. The fabrication process of the flexible strain sensing unit is as follows: First, a microcrack array is formed on a polystyrene film coated with single-walled carbon nanotubes using a computer-controlled laser device. This process must ensure that the polystyrene film substrate remains intact. Then, the laser-treated crack array undergoes thermally induced dimensional shrinkage. Due to the thermal responsiveness of the polystyrene film substrate, it shrinks biaxially when the temperature exceeds its glass transition temperature. Next, a 2mm thick layer of Ecoflex material is coated on top of the shrunken device, and the rigid polystyrene film substrate is removed, finally obtaining a flexible strain sensing unit coated with Ecoflex material. A schematic diagram in this embodiment is shown below. Figure 5 As shown. In practical applications, the prepared flexible skin needs to be attached to the surface of the flexible robotic arm, while the flexible strain sensing unit needs to be attached to the flexible skin to correspond to the deformation position of the flexible robotic arm, so as to ensure comprehensive coverage of the robotic arm's deformation and accurate monitoring of key areas.

[0040] The signal acquisition and synchronization module primarily converts sensor signals into processable data, including A / D conversion, power management, and data timestamp synchronization. The A / D conversion module has a sampling accuracy of at least 16 bits to ensure complete preservation of signal details; the power management module adopts a low-power design to meet the energy supply requirements of space applications; data transmission uses SPI and I / O. 2 C-mode multi-bus communication, with the SPI bus used for transmitting high-speed strain and acceleration data, I 2The C bus is used to transmit resistance and capacitance data, and the real-time performance and stability of data transmission are ensured through bus division of labor.

[0041] The preprocessed sensor electrical signal is subjected to feature decoupling processing to obtain the target feature vector representing the motion state of the robotic arm. Specifically, the preprocessed sensor electrical signal is subjected to feature decoupling processing using the principal component analysis algorithm to obtain the target feature vector representing the motion state of the robotic arm.

[0042] Before inputting the target feature vector into the trained pose mapping model, the spatial flexible manipulator pose determination method further includes: acquiring a sample set; the sample set includes sample feature vectors of several sample sensing electrical signals of the spatial flexible manipulator body and the end pose of the spatial flexible manipulator body corresponding to each sample sensing electrical signal; and using the sample set to train the pose mapping model to obtain the trained pose mapping model.

[0043] The acquired sample set specifically includes: synchronously acquiring the original sensing electrical signals of several flexible strain sensing units on the body of the spatial flexible robotic arm; preprocessing the original sensing signals to obtain sample sensing electrical signals; and decoupling the sample sensing electrical signals using principal component analysis algorithm through the feature extraction and data fusion module to obtain the sample feature vector of the sample sensing electrical signals.

[0044] The core objective of the feature extraction and data fusion module is to provide a clean and consistent data foundation for efficient model training. This requires a systematic data preprocessing workflow. First, data cleaning is necessary: ​​this includes detecting missing values ​​in the dataset and selecting imputation or deletion strategies based on data characteristics to ensure data integrity; simultaneously, outlier data points are identified through statistical analysis and processed accordingly to ensure data quality and consistency. Second, noise reduction is required, specifically through signal filtering or waveform analysis techniques to suppress noise in the acquired multi-channel sensor signals and improve the signal-to-noise ratio.

[0045] Given the structural characteristics of flexible robotic arms, signals collected by different sensors may exhibit correlation or mutual interference. Such coupling can mask or distort key signal features, thus affecting the performance of subsequent models. Therefore, it is necessary to decouple the preprocessed sensor signal data to reveal cleaner, more independent information sources, thereby improving the accuracy of data analysis and prediction models. In this embodiment, principal component analysis (PCA) is employed. Its core idea is to map the original feature space into a set of linearly independent principal components through orthogonal transformation, while preserving the direction of maximum variance in the data to maintain key information, ultimately achieving effective decoupling of multi-channel sensor data.

[0046] For ease of explanation, the multi-channel sensor electrical signal data acquired in the flexible electronic skin sensing module is defined as follows: This data should be a The matrix, where This refers to the number of flexible strain sensing units in the flexible electronic skin sensing module. They are the 1st, 2nd, and 3rd respectively. Sensor signals collected by a flexible strain sensing unit For time steps. The pose data of the end effector of the spatial flexible robotic arm is as follows: ,in The coordinates of the end effector position of the flexible spatial robotic arm. Let be the quaternion representing the end effector posture of the flexible spatial robotic arm. The algorithm for principal component analysis is shown below: The first step is to perform data preprocessing, specifically data centralization: (1); in, The first digit of the centered matrix Column data; for The Middle Column data, ; This is the sample mean vector.

[0047] Then, a covariance matrix is ​​defined to quantify the linear correlation between features: (2); in, Here is the covariance matrix; superscript Indicates transpose; for The centralized data matrix; The number of samples; The original feature dimension; for The transpose of .

[0048] Data decoupling and dimensionality reduction through eigenvalue decomposition: solving the eigenvalue problem of the covariance matrix. ,in The eigenvalues ​​represent the variance contributions of the corresponding principal components. This represents the corresponding eigenvector (i.e., the principal component direction), and the eigenvalues ​​are arranged according to... Arrange in order (retain the principal component with the largest variance); select the first The eigenvectors constitute the projection matrix. Then the target feature vector or sample feature vector (the data after dimensionality reduction) Represented as: (3).

[0049] By inputting the principal component features obtained from the above decoupling into the pose mapping model (machine learning model), the independent information sources in the data can be effectively revealed, thereby improving the prediction performance of the subsequent model on the dynamic behavior of the flexible robotic arm.

[0050] The data training module employs a deep learning neural network algorithm to model and predict the nonlinear mapping relationship between real-time pose and sensor electrical signals. In this embodiment, the pose mapping model is a Long Short-Term Memory (LSTM) neural network model. The LSTM algorithm is detailed below: At some point A single-layer network consists of two information flows, one above the other. arrive The information flow represents the transmission of cell states, and the entire line interacts linearly with the information flow below through three gating structures. The gating structures allow information to selectively pass through, flowing upwards... arrive The information flow deletes or adds information about the cell state. In the gating structure... Activation function layers and tanh activation function layers can respectively transform the input to Between and Between these parameters, weights are generated for the input data, thereby filtering the input data. The LSTM network structure is as follows: Figure 6 As shown, each layer of the LSTM network has three gate structures to control the cell state.

[0051] (1) Forget Gate: The expression for the forget gate is as follows: (4); in, The activation function layer uses the hidden state from the previous time step. and Input at any time A value between 0 and 1 is obtained as the probability that the cell state of the previous layer is forgotten, and is considered as... The decay coefficient of memory, denoted as ; and These are the weights and bias matrices of the forget gate, respectively.

[0052] (2) Input gate: Part of the input gate will Input at any time Hidden state from the previous moment After linear combination, through Activation function layer activation results This part determines which information needs to be updated; this information is what is selected to be forgotten through the forgetting gate. The other part hides the previous state. and Input at any time A vector is generated through a tanh layer. This refers to the alternative content to be updated. Then, the two parts will be combined to update the state. Updated to The formula for the input gate update process is: (5); (6); (7); in, For input gate time Data after activation of the activation function layer; and In the input gates The weights and bias matrices of the activation function layer; For input gate The output of the tanh activation function layer at time t; and These are the weights and bias matrices of the tanh activation function layer in the input gate, respectively; for The output of the input gate is always being input; This represents convolution.

[0053] (3) Output gate: (8); (9); in, and These represent the weights and bias matrices of the output gate, respectively, which are also parameters that need to be learned during training; For the output gate The output of the activation function layer; For output gate The output at each moment. Because each loop uses information from the previous loop. and Each output state is influenced by the previous state, so LSTM has the ability to remember long-term historical information.

[0054] The initial pose estimation value is optimized by state filtering to output the final pose estimation result of the spatial flexible manipulator body. Specifically, the optimization is performed by using the Extended Kalman Filter (EKF) algorithm to optimize the initial pose estimation value by state filtering, taking the initial pose estimation value as the observation value, and establishing a state equation containing the smooth motion constraint of the manipulator body. Through recursive prediction and update steps, the final pose estimation result of the spatial flexible manipulator body is output.

[0055] The state determination and output module outputs the pose sensing results of the spatial flexible robotic arm and connects them with the application, and uses a filtering algorithm to further reduce noise. Although deep learning neural network algorithms have achieved modeling and prediction of the nonlinear mapping relationship between real-time pose and sensor electrical signals, the model output still has certain errors. To further suppress noise interference and reduce prediction errors, this embodiment uses the EKF algorithm for secondary noise processing. The EKF algorithm is a recursive state estimation method for nonlinear systems. Its core idea is to achieve local linearization by performing a first-order Taylor expansion on the nonlinear system model, thereby solving for the optimal state estimate during the recursive process, ultimately achieving the goal of suppressing noise and improving pose prediction accuracy. The EKF algorithm is shown below: Consider a discrete-time nonlinear dynamic system, whose state-space model is as follows: (10); in, yes The system state vector at any given time corresponds to the pose of the spatial flexible robotic arm. It is a nonlinear state transition function; yes The system state vector at time t; It is a known control input, corresponding to the sensor electrical signal; It is process noise, assumed to follow a zero-mean Gaussian distribution, i.e. ; The process noise covariance matrix represents the uncertainty of the system model.

[0056] The observation equation (measurement model) is as follows: (11); in, yes The observation vector at time; It is a nonlinear observation function; It is observation noise, assumed to follow a zero-mean Gaussian distribution, i.e. ; The observation noise covariance matrix characterizes the uncertainty of sensor measurements. It is assumed that process noise and observation noise are independent.

[0057] The key to the EKF algorithm lies in finding the best estimation point at the current state (prior state estimate). or posterior state estimation At point ), for nonlinear functions and nonlinear observation function Perform a first-order Taylor expansion and ignore higher-order terms.

[0058] exist Prior state estimation at time 1 and process noise Expanding the nonlinear function : (12); in, It is a nonlinear function For the state vector The Jacobian matrix, in Calculation at point: (13); in, This represents the partial derivative.

[0059] It is a nonlinear function For process noise vector Jacobian matrix: (14).

[0060] Posterior state estimation and observation noise Expand the nonlinear observation function at the location : (15); in, It is a nonlinear observation function For the state vector The Jacobian matrix, in Calculation at point: (16).

[0061] It is a nonlinear observation function For the observation noise vector Jacobian matrix: (17).

[0062] Based on the linearization model described above, EKF recursively performs the following two steps: Step 1: Prediction (Time Update) Prior state prediction: (18); This is the constant term in the linearized expansion.

[0063] Prior error covariance prediction: According to the linearized model, the covariance propagation of the state prediction error is as follows: (19); in, yes k The prior error covariance matrix at time t; yes The posterior error covariance matrix at time t; for The transpose of the matrix; for The transpose of .

[0064] Step 2: Update (Measurement Update) Kalman gain calculation: Based on the linearized observation model, the optimal Kalman gain is: (20); in, The optimal Kalman gain is determined by the tradeoff between prediction covariance and observation noise, which in turn determines the reliability of the observation information.

[0065] Posterior state update: using actual observation vectors Correct the residuals between the observed predictions and the predicted state: (twenty one); in, This is called the observation residual.

[0066] Posterior error covariance update: updates the uncertainty measure of the state estimate, as shown in the following formula: (twenty two); in, yes The posterior error covariance matrix at time t; It is an identity matrix.

[0067] To maintain numerical stability and symmetry, the uncertainty measure for updating the state estimate is often simplified: (twenty three).

[0068] Finally, based on the above modules, the final pose estimation result of the end effector of the spatial flexible robotic arm can be obtained through the flexible electronic skin intelligent sensing method. In this embodiment, Orientation sensing position (the position of the end effector of the spatial flexible robotic arm in the final pose estimation result) The error distribution curve between the coordinates and the actual position is shown in the figure below. Figure 7 As shown.

[0069] This application aims to address the technical problems of poor fit, lag in dynamic deformation perception, and insufficient accuracy in motion state estimation caused by traditional sensing methods in flexible space manipulators due to their flexible structure and complex operating environment. The method first integrates a multimodal flexible electronic sensor array that adaptively fits the curved contour of the flexible space manipulator along key deformation regions and joints. Through distributed flexible strain sensors within the array, real-time multi-dimensional sensing data, including real-time pose and sensor electrical signals, are collected during the manipulator's operation. Subsequently, a multi-source data preprocessing module is constructed to suppress noise and extract features from the original sensing signals. Deep learning neural network algorithms are then used to fuse and analyze the real-time pose and sensor electrical signals from the processed data. Finally, through collaborative decision-making in the data fusion layer, an extended Kalman filter algorithm is introduced to further reduce noise in the predicted pose and estimate the motion state of the flexible space manipulator, outputting the real-time end-effector position and end-effector quaternion of the manipulator body, thus achieving high-precision determination of the motion state.

[0070] The method proposed in this application relies on the excellent fit of flexible electronic skin to achieve full-area sensing coverage. The multimodal data fusion strategy reduces the position positioning percentage error to within 2%, controls the attitude angle measurement error within ±1°, and achieves a data update frequency of over 100Hz. It has the technical advantages of strong flexibility, fast real-time response, and high estimation accuracy. It can be directly applied to on-orbit robotic arms in tasks such as spacecraft cabin maintenance and space payload grasping. It is also suitable for self-sensing systems in fields such as flexible robots and deformable space structures. It is applicable to motion monitoring and attitude control scenarios of space service robots, on-orbit maintenance robotic arms, space collaborative assembly platforms, and foldable mechanisms.

[0071] This application does not rely on external sensors or vision systems; it can achieve motion state perception of the body solely through the flexible electronic skin on the surface of the spatial flexible robotic arm. Specifically, the system has a sampling frequency of over 100Hz, which can meet the application requirements of dynamic attitude tracking. At the same time, the skin is flexibly attached to the surface of the spatial flexible robotic arm and can conformally bend with the movement of the spatial flexible robotic arm, thereby adapting to complex control scenarios in the space environment.

[0072] This application also provides an application scenario in which the above-mentioned spatial flexible robotic arm pose determination method is applied. Specifically, the spatial flexible robotic arm pose determination method provided in this embodiment can be applied in a robotic arm pose estimation scenario. The robotic arm pose estimation scenario includes a content production stage, a content processing link, and a content distribution stage; multi-channel sensor electrical signals enter the content processing link from the content production stage, obtain the corresponding final pose estimation result through human-machine collaboration, and then enter the downstream content distribution stage. The spatial flexible robotic arm pose determination method provided in this embodiment belongs to the content processing link. Specifically, in the content processing link process for multi-channel sensor electrical signals, the multi-channel sensor electrical signals can be synchronously acquired, converted from analog to digital, and preprocessed to obtain preprocessed sensor electrical signals. Feature decoupling processing is performed on the preprocessed sensor electrical signals to obtain a target feature vector representing the motion state of the robotic arm. The target feature vector is input into a trained pose mapping model to output a preliminary pose estimation value. State filtering optimization is performed on the preliminary pose estimation value to output the final pose estimation result of the spatial flexible robotic arm body.

[0073] Based on the same inventive concept, this application also provides a spatial flexible robotic arm pose determination system for implementing the aforementioned spatial flexible robotic arm pose determination method. The solution provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more spatial flexible robotic arm pose determination system embodiments provided below can be found in the limitations of the spatial flexible robotic arm pose determination method described above, and will not be repeated here.

[0074] In one exemplary embodiment, such as Figure 8 As shown, a spatial flexible robotic arm pose determination system is provided, comprising the following modules: The flexible electronic skin sensing module T1 is used to collect multi-channel sensing electrical signals caused by the movement of the robotic arm in real time after the flexible skin integrating a distributed flexible sensor array is attached to the surface of the flexible robotic arm body. The signal acquisition and synchronization module T2 is used to synchronously acquire, convert analog to digital and preprocess the multi-channel sensing electrical signals to obtain the preprocessed sensing electrical signals. The feature decoupling processing module T3 is used to perform feature decoupling processing on the preprocessed sensing electrical signal to obtain the target feature vector characterizing the motion state of the robotic arm. The preliminary pose estimation module T4 is used to input the target feature vector into the trained pose mapping model and output a preliminary pose estimate. The preliminary pose estimate includes the end position and the end pose quaternion. The trained pose mapping model is trained by using the sample feature vector of the sample sensing electrical signal of the spatial flexible manipulator body as input and the end pose of the spatial flexible manipulator body as the label. The final pose estimation result determination module T5 is used to perform state filtering optimization on the preliminary pose estimation value and output the final pose estimation result of the spatial flexible robotic arm body.

[0075] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 9 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores the robot arm's pose and motion state estimation data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a spatial flexible robot arm pose determination method.

[0076] Those skilled in the art will understand that Figure 9 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0077] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0078] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Moreover, the collection, use and processing of the relevant data are carried out in compliance with the relevant data protection laws and policies of the country where the location is located, and with the authorization granted by the owner of the corresponding device.

[0079] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0080] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0081] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0082] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for determining the pose of a spatial flexible robotic arm, characterized in that, The method for determining the pose of the spatial flexible robotic arm includes: After attaching a flexible skin with an integrated distributed flexible sensor array to the surface of the space flexible robotic arm, multi-channel sensor electrical signals caused by the movement of the robotic arm are collected in real time. The multi-channel sensing electrical signals are synchronously acquired, converted from analog to digital and preprocessed to obtain preprocessed sensing electrical signals. The preprocessed sensor electrical signal is subjected to feature decoupling processing to obtain the target feature vector characterizing the motion state of the robotic arm. The target feature vector is input into the trained pose mapping model, and a preliminary pose estimate is output. The preliminary pose estimate includes the end position and the end pose quaternion. The trained pose mapping model is trained by using the sample feature vector of the sample sensing electrical signal of the spatial flexible manipulator body as input and the end pose of the spatial flexible manipulator body as the label. The initial pose estimation value is optimized by state filtering to output the final pose estimation result of the spatial flexible robotic arm body.

2. The spatial flexible robotic arm pose determination method according to claim 1, characterized in that, The distributed flexible sensing array includes multiple flexible strain sensing units based on microcrack structures. The fabrication of the flexible strain sensing unit includes: forming a microcrack array on a polymer substrate coated with single-walled carbon nanotubes, regularizing the microcracks through thermal shrinkage, encapsulating it with a flexible material, and removing the original substrate.

3. The spatial flexible robotic arm pose determination method according to claim 2, characterized in that, The flexible skin uses conductive nanomaterials such as silver nanowires, polydimethylsiloxane composite films, or graphene composite films; the polymer substrate is polystyrene; and the encapsulation material is Ecoflex material.

4. The method for determining the pose of a spatial flexible robotic arm according to claim 1, characterized in that, The preprocessed sensor electrical signal is subjected to feature decoupling processing to obtain the target feature vector representing the motion state of the robotic arm. Specifically, the preprocessed sensor electrical signal is subjected to feature decoupling processing using the principal component analysis algorithm to obtain the target feature vector representing the motion state of the robotic arm.

5. The spatial flexible robotic arm pose determination method according to claim 1, characterized in that, The pose mapping model is a long short-term memory neural network model.

6. The spatial flexible robotic arm pose determination method according to claim 1, characterized in that, The initial pose estimate is optimized by state filtering to output the final pose estimate of the spatial flexible robotic arm, specifically including: Using the extended Kalman filter algorithm, the initial pose estimate is optimized by state filtering. The initial pose estimate is used as the observation value, and a state equation containing the smooth motion constraint of the robotic arm is established. Through recursive prediction and update steps, the final pose estimate of the spatial flexible robotic arm body is output.

7. The spatial flexible robotic arm pose determination method according to claim 1, characterized in that, Before inputting the target feature vector into the trained pose mapping model, the spatial flexible robotic arm pose determination method further includes: Obtain a sample set; the sample set includes sample feature vectors of several sample sensing electrical signals of the spatial flexible manipulator body and the end pose of the spatial flexible manipulator body corresponding to each sample sensing electrical signal. The pose mapping model is trained using a sample set to obtain a trained pose mapping model.

8. A spatial flexible robotic arm pose determination system, characterized in that, The spatial flexible robotic arm pose determination system includes: The flexible electronic skin sensing module is used to collect multi-channel sensing electrical signals caused by the movement of the robotic arm in real time after the flexible skin integrating a distributed flexible sensor array is attached to the surface of the flexible robotic arm body. The signal acquisition and synchronization module is used to synchronously acquire, convert analog to digital and preprocess the multi-channel sensing electrical signals to obtain preprocessed sensing electrical signals. The feature decoupling processing module is used to perform feature decoupling processing on the preprocessed sensing electrical signal to obtain the target feature vector characterizing the motion state of the robotic arm. The preliminary pose estimation module is used to input the target feature vector into the trained pose mapping model and output a preliminary pose estimate. The preliminary pose estimate includes the end position and the end pose quaternion. The trained pose mapping model is trained by using the sample feature vector of the sample sensing electrical signal of the spatial flexible manipulator body as input and the end pose of the spatial flexible manipulator body as the label. The final pose estimation result determination module is used to perform state filtering optimization on the preliminary pose estimation value and output the final pose estimation result of the spatial flexible robotic arm body.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the spatial flexible robotic arm pose determination method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the spatial flexible robotic arm pose determination method according to any one of claims 1-7.

Images