Kinematic error correction method and system for explosion-proof robots

Abstract

The present application relates to a kinematic error correction method and system for an explosion-proof robot, which includes collecting joint information through a laser calibrator with a preset accuracy; extracting features from the collected information to obtain a joint feature set containing multiple features; inputting the information into an oblique compensation model to model and compensate for non-orthogonal joint errors, generating a compensation matrix, and generating control instructions based on the corrected model; the method comprehensively considers joint characteristics and dynamic characteristics, effectively solves the problem of actual kinematic characteristics deviating from the theoretical model due to mechanical manufacturing tolerances, assembly errors and wear, realizes precise correction of kinematic errors, improves the robot's spatial positioning and trajectory tracking capabilities when operating in hazardous environments, and ensures the safety and efficiency of task execution.

Description

Technical Field

[0001] The present application relates to the fields of intelligent control and robotics technology, and in particular to a kinematic error correction method and system for an explosion-proof robot. Background Art

[0002] In the field of explosion-proof robots, precise correction of kinematic errors is crucial for improving their operational accuracy and reliability. With the increasing demand for hazardous environments in industrial scenarios, explosion-proof robots are widely used in high-risk areas. The precision of their joint motion directly impacts the safety and efficiency of mission execution. However, due to factors such as mechanical structural manufacturing tolerances, assembly errors, and wear and tear from long-term operation, the actual kinematic characteristics of explosion-proof robots often deviate from theoretical models, affecting their spatial positioning and trajectory tracking capabilities.

[0003] In one existing technology, the kinematic error correction of explosion-proof robots mainly relies on traditional calibration methods, such as manual measurement and adjustment based on fixed reference points. Such methods usually manually record joint position information and perform offline corrections, lacking real-time and automation capabilities. At the same time, the existing technology rarely considers the non-orthogonal errors between joints and does not fully analyze the impact of dynamic characteristics such as velocity and acceleration on errors, resulting in limited compensation effects. It can be seen that the existing technology has room for improvement in error feature extraction, dynamic modeling, and compensation strategies. Therefore, there is an urgent need for a kinematic error correction method that can comprehensively consider joint characteristics and dynamic characteristics to improve the performance of explosion-proof robots. Summary of the Invention

[0004] The main purpose of this application is to provide a kinematic error correction method and system for an explosion-proof robot that can accurately correct the kinematic errors of the robot through joint characteristics and dynamic characteristics compensation models.

[0005] To achieve the above objectives, an embodiment of the present invention provides a kinematic error correction method for an explosion-proof robot, wherein the explosion-proof robot includes a plurality of movable joints, and the method includes:

[0006] A laser calibration instrument with preset accuracy collects the spatial position information and motion trajectory information of each joint of the explosion-proof robot;

[0007] Performing feature extraction on the spatial position information and motion trajectory information to obtain a joint feature set, wherein the joint feature set includes position features, posture features, and motion features of the joint, and the motion features include velocity and acceleration of the joint;

[0008] Inputting the joint feature set into a skew compensation model, modeling and compensating the non-orthogonal joint errors through the skew compensation model, and generating a joint error compensation matrix, wherein the joint error compensation matrix is ​​used to quantify the non-orthogonality deviations between the joints;

[0009] Correcting the kinematic model of the explosion-proof robot according to the joint error compensation matrix to obtain a corrected kinematic model, wherein the kinematic model describes the correlation between the position, velocity, and acceleration of each joint of the explosion-proof robot;

[0010] A corresponding control instruction is generated based on the modified kinematic model, and the control instruction is sent to the explosion-proof robot to perform motion operations.

[0011] Accordingly, an embodiment of the present application further provides a kinematic error correction system for an explosion-proof robot, wherein the explosion-proof robot includes a plurality of movable joints, and the system includes:

[0012] The acquisition module is used to collect the spatial position information and motion trajectory information of each joint of the explosion-proof robot based on a laser calibration instrument with preset accuracy;

[0013] A feature extraction module is used to extract features from the spatial position information and motion trajectory information to obtain a joint feature set, wherein the joint feature set includes position features, posture features, and motion features of the joint, and the motion features include velocity and acceleration of the joint;

[0014] an error compensation module, configured to input the joint feature set into a skew compensation model, model and compensate for non-orthogonal joint errors using the skew compensation model, and generate a joint error compensation matrix, wherein the joint error compensation matrix is ​​used to quantify the non-orthogonality deviations between the joints;

[0015] a correction module, configured to correct the kinematic model of the explosion-proof robot according to the joint error compensation matrix to obtain a corrected kinematic model, wherein the kinematic model describes the correlation between the position, velocity, and acceleration of each joint of the explosion-proof robot;

[0016] The instruction generation module is used to generate corresponding control instructions based on the modified kinematic model, and send the control instructions to the explosion-proof robot to perform motion operations.

[0017] In summary, by adopting the technical solution of the present application, joint information is collected through a laser calibration instrument with preset accuracy, and the spatial position and motion trajectory of the explosion-proof robot joint can be accurately obtained, overcoming the problem of lack of real-time and automation capabilities in manual measurement and adjustment of the existing technology; feature extraction is performed on the collected information to obtain a joint feature set containing multiple features, which comprehensively reflects the joint status; it is input into the skew compensation model to model and compensate for non-orthogonal joint errors to generate a compensation matrix, which makes up for the deficiency of the existing technology that does not fully consider non-orthogonal errors; the kinematic model is corrected according to the matrix, so that the model can more accurately describe the joint relationship; finally, control instructions are generated based on the corrected model, which improves the robot's operating accuracy and reliability. This whole set of processes comprehensively considers joint characteristics and dynamic characteristics, effectively solves the problem of actual kinematic characteristics deviating from the theoretical model due to mechanical manufacturing tolerances, assembly errors and wear, realizes accurate correction of kinematic errors, improves the robot's spatial positioning and trajectory tracking capabilities when operating in dangerous environments, and ensures the safety and efficiency of task execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 Schematic diagram of a kinematic error correction method for an explosion-proof robot according to an embodiment of the present application;

[0020] Figure 2 A flowchart of a kinematic error correction method for an explosion-proof robot is provided for an embodiment of the present application;

[0021] Figure 3 A schematic diagram of the information acquisition process provided in the embodiment of the present application;

[0022] Figure 4 A schematic diagram of the process of obtaining three-dimensional space coordinate data provided in an embodiment of the present application;

[0023] Figure 5 A schematic diagram of a process for generating an error compensation matrix according to an embodiment of the present application;

[0024] Figure 6 A schematic diagram of the model correction process provided in an embodiment of the present application;

[0025] Figure 7 A schematic diagram of the feature extraction process provided in the embodiment of the present application;

[0026] Figure 8A schematic diagram of the flow chart for generating instructions provided in an embodiment of the present application;

[0027] Figure 9 A schematic diagram of the structure of a kinematic error correction system for an explosion-proof robot provided in an embodiment of the present application;

[0028] Figure 10 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0030] The embodiments of the present application provide a kinematic error correction method and system for an explosion-proof robot, which will be described in detail below.

[0031] In this application, kinematic error correction for explosion-proof robots refers to the process of correcting the kinematic errors of explosion-proof robots used in hazardous environments (such as petrochemicals and warehouses storing flammable and explosive materials) through a series of technical means. Kinematic error refers to the deviation of the actual kinematic characteristics of an explosion-proof robot from a pre-defined theoretical motion model due to various factors, such as mechanical structural manufacturing tolerances, assembly errors, and wear caused by long-term operation. This deviation manifests as parameters such as the position, posture, velocity, and acceleration of the robot's joint motion not matching the theoretical values, thereby affecting the robot's spatial positioning accuracy and trajectory tracking ability. Kinematic error correction involves using high-precision measurement equipment (such as laser calibration equipment) to collect joint spatial position and motion trajectory information. Then, through a series of operations such as feature extraction, modeling (such as skew compensation models), and model correction, control commands are generated to drive the robot's motion. This improves the robot's operational accuracy and reliability, ensuring that the explosion-proof robot can accurately perform tasks such as inspection and material handling in hazardous environments, ensuring the safety and efficiency of task execution.

[0032] The method of this application can be applied to explosion-proof robots with robotic arms. The following describes the structure of one such explosion-proof robot. For example, the chassis of the explosion-proof robot is rectangular, with large circular wheels mounted at the four corners for mobility. A compartment in the center of the chassis houses the power system and control unit. A multi-degree-of-freedom robotic arm is mounted on one side of the chassis. The arm is composed of multiple circular joints connected by rectangular links, enabling rotation and bending. The end effector can be a claw (composed of two opposing triangles with a central connection point) or a suction cup (circular with radial lines inside to represent suction). The robot is encased in a thick, curved outer shell with small holes for heat dissipation or ventilation. In addition, a circular or square camera is mounted on the front of the shell or on the arm. Various sensors, such as distance sensors and temperature sensors, are distributed around the chassis and near the arm joints.

[0033] In practical applications, explosion-proof robots can also be robot dogs, etc.

[0034] As shown in Figure 1, a kinematic error correction method scenario for an explosion-proof robot is provided. The kinematic error correction scenario for an explosion-proof robot mainly includes an explosion-proof robot, a laser calibrator with a preset accuracy, and a control platform; wherein the explosion-proof robot, the laser calibrator, and the control platform are connected via a wireless network.

[0035] For example, in a petrochemical production workshop, where a large number of flammable and explosive chemicals are present, explosion-proof robots are responsible for important tasks such as equipment inspection and material handling. These robots are composed of multiple movable joints.

[0036] A laser calibrator with a preset accuracy is placed in a suitable location within the workshop to collect spatial position and motion trajectory information for each joint of the explosion-proof robot. The laser calibrator scans the mechanical surface of the robot's joints with a high-precision beam, acquiring relevant information. For example, as the robot's arm joints extend and flex during material handling, the laser calibrator accurately captures spatial coordinate point cloud data from the joint surfaces. Furthermore, as the robot moves along its inspection route and its joints continuously adjust their posture, the laser calibrator continuously records the changes in the joint's trajectory during this continuous motion. By filtering the collected spatial coordinate point cloud data to remove noise points and extract the geometric center coordinates of key nodes, the spatial position information of the joints can be calculated. Furthermore, time series analysis is used to analyze the joint's motion trajectory, extracting the joint's displacement curve and its time derivative, thus fully capturing the joint's motion trajectory information.

[0037] After acquiring the spatial position and motion trajectory information uploaded by the laser calibration instrument, the control platform performs feature extraction on this information to generate a joint feature set. For example, principal component analysis (PCA) of the spatial position information, such as when analyzing the robot's leg joints, accurately extracts the primary position features and determines the key position parameters of the joints in three-dimensional space. Frequency domain transformation of the motion trajectory information is performed to derive the joint's posture characteristics when the robot's joints perform complex rotations and swings. Combining the position and posture features, wavelet decomposition is used to further extract the joint's velocity and acceleration characteristics, essentially breaking down every detail of the joint's motion for analysis. These position, posture, velocity, and acceleration features are combined into a joint feature set that comprehensively describes the joint's static and dynamic characteristics in space.

[0038] Then, the control platform inputs the joint feature set into the skew compensation model. During this process, the skew compensation model analyzes the source of non-orthogonality errors between joints based on the joint feature set. For example, when there is a certain manufacturing tolerance in the joint connection structure of the robot, the model can identify how this non-orthogonality error is generated. By constructing the initial parameter matrix of the skew compensation model, the initial parameter matrix is ​​fitted using the least squares method to obtain preliminary parameters, and then the preliminary parameters are regularized to eliminate overfitting, and finally the optimal parameters are obtained, thereby generating a joint error compensation matrix. This matrix can accurately quantify the non-orthogonality deviations between the joints.

[0039] The control platform modifies the explosion-proof robot's kinematic model based on the joint error compensation matrix, generating a revised kinematic model. By incorporating the joint error compensation matrix into the original kinematic model, the revised kinematic equations are formed. Finally, corresponding control instructions are generated based on the revised kinematic model and sent to the explosion-proof robot to execute motion operations. Based on the revised kinematic model, the target position, target velocity, and target acceleration of each joint of the explosion-proof robot are calculated. When the robot needs to conduct inspections in narrow equipment aisles, accurate target position, velocity, and acceleration calculations ensure the robot can safely avoid various obstacles. These target parameters are then converted into control signals for the servo motors and transmitted to the explosion-proof robot's servo control system via a communication interface. The servo control system drives the joints to execute the corresponding motion operations based on the control signals. During material handling, the robot collects real-time feedback on the actual joint states, compares them with the target states, and adjusts the control signals to minimize dynamic errors. For example, if the actual velocity of a joint deviates from the target velocity during handling, timely adjustment of the control signal ensures the robot's motion accuracy, preventing material from falling or colliding with surrounding equipment, thereby ensuring production safety and efficiency in the workshop.

[0040] refer to Figure 2 , Figure 2 This is a flow chart of a kinematic error correction method for an explosion-proof robot provided in an embodiment of the present application. The execution subject of the method may be a computer device, which may be a single computer device or a cluster of multiple computer devices. The computer device may be a terminal device or a server. The kinematic error correction method for an explosion-proof robot provided in an embodiment of the present application specifically includes:

[0041] In step S10, a laser calibration instrument with a preset accuracy collects spatial position information and motion trajectory information of each joint of the explosion-proof robot.

[0042] Explosion-proof robots are specially designed for use in environments with explosion hazards, such as petrochemical plants and underground coal mines. They feature special explosion-proof structures and materials to prevent explosions caused by sparks or high temperatures during operation.

[0043] Joints are the movable parts of explosion-proof robots, similar to the joints of the human body. Through the movement of the joints, the robot can achieve various postures and actions, such as rotation, bending, etc. The movement state of the joints directly affects the overall movement of the robot.

[0044] A laser calibrator is a device that uses laser technology for measurement and positioning. Preset accuracy refers to the measurement accuracy set before the laser calibrator is used. This accuracy is determined based on the specific application scenario and kinematic error correction requirements of the explosion-proof robot. For example, in scenarios requiring high kinematic accuracy for explosion-proof robots, such as handling hazardous materials in confined spaces, the preset accuracy may be set to the millimeter level.

[0045] Spatial position information refers to the coordinate positions of each joint of an explosion-proof robot in three-dimensional space. This information can be obtained by using a laser calibrator to emit a laser beam to the joint surface and then calculating it based on information such as the time and angle of the laser's reflection. For example, if a joint of an explosion-proof robot is located at the (x, y, z) position in a Cartesian coordinate system, the laser calibrator can accurately determine this coordinate value through its measurement system.

[0046] Motion trajectory information is the path of a joint's movement in space over time, reflecting its dynamic motion. This can be obtained by continuously collecting the joint's spatial position information multiple times and arranging this position information in chronological order. For example, within a time period, a joint moves from position A (x1, y1, z1) to position B (x2, y2, z2). All the position points along the way constitute the motion trajectory information.

[0047] The purpose of this step is to use a laser calibrator with a preset accuracy to obtain the spatial position and motion trajectory information of each joint of the explosion-proof robot. The laser calibrator emits a laser beam onto the joint surface and then calculates the joint's spatial position based on the reflected laser beam (such as time and angle). The motion trajectory information is obtained by continuously collecting the joint's spatial position at different time points and combining this chronological position data. This information serves as the basis for subsequent error correction.

[0048] In one embodiment, a laser calibrator uses a method of emitting multiple laser beams simultaneously to illuminate the joints of the explosion-proof robot from different angles. Each laser beam is equipped with a high-precision time measurement device and an angle sensor. When the laser beam irradiates the joint surface and reflects back, the distance from the joint to the laser calibrator is calculated based on the speed of laser propagation and the time difference. Combined with the laser beam emission angle measured by the angle sensor, the coordinate position of the joint in space is accurately calculated using mathematical methods such as trigonometric functions. For the collection of motion trajectory information, a fixed time interval is set, and the spatial position of the joint is collected within each time interval to obtain a series of position points. These position points are connected in chronological order to form motion trajectory information. This method of simultaneous measurement of multiple laser beams can improve the accuracy and reliability of measurement and reduce the measurement blind spots or errors that may exist due to a single laser beam. Its technical effect is that the spatial position and motion trajectory information of the explosion-proof robot joint can be obtained more comprehensively and accurately, providing a reliable data basis for subsequent error correction.

[0049] In step S20 , feature extraction is performed on the spatial position information and the motion trajectory information to obtain a joint feature set. The joint feature set includes position features, posture features, and motion features of the joints. The motion features include the velocity and acceleration of the joints.

[0050] Among them, feature extraction refers to extracting information that can represent joint characteristics from the collected original data (i.e., spatial position information and motion trajectory information).

[0051] A joint feature set is a collection of various joint features. Position features are specific attributes that describe the position of a joint in space, such as the coordinate distance of a joint relative to a reference point, the position component of a joint in a specific direction, etc.

[0052] Posture features are features that reflect the direction and angle state of the joint, such as the rotation angle and tilt angle of the joint, which can help determine the orientation of the joint in space.

[0053] Motion characteristics include joint velocity and acceleration. Velocity refers to the change in position of a joint per unit time, for example, how many millimeters a joint moves per second in the x-direction. Acceleration refers to the change in velocity per unit time, reflecting the speed of changes in the joint's motion state.

[0054] In one embodiment, to extract features from spatial position information, the collected joint spatial position coordinate data is first normalized, mapping the coordinate values ​​to a specific interval for ease of subsequent calculation and comparison. Principal component analysis (PCA) is then used to decompose the eigenvalues ​​of the data covariance matrix to identify the primary directions of variation in the data. The coordinate components corresponding to these primary directions of variation are the primary positional features of the joint. For motion trajectory information, the trajectory data is first smoothed to eliminate small fluctuations caused by factors such as measurement noise. The trajectory data is then subjected to a discrete Fourier transform (DFT) to analyze the frequency components of the trajectory in the frequency domain. Low-frequency components can reflect the overall posture trend of the joint, and the parameters corresponding to these low-frequency components are the posture features. Next, the position coordinates of adjacent time points are subtracted to obtain joint velocity information. A similar subtraction operation is then performed on the velocity information to obtain acceleration information. This velocity and acceleration information is then combined with the previously obtained position and posture features to form a joint feature set. This technical implementation effectively extracts a joint feature set from the raw data through a series of mathematical operations, accurately reflecting the joint state. By using this technical implementation method, the various features of the joints can be accurately extracted while removing noise and redundant information in the original data, providing a more accurate data basis for subsequent error compensation and improving the accuracy of error correction.

[0055] In step S30 , the joint feature set is input into the skew compensation model, and the non-orthogonal joint errors are modeled and compensated by the skew compensation model to generate a joint error compensation matrix. The joint error compensation matrix is ​​used to quantify the non-orthogonality deviations between the joints.

[0056] The skew compensation model is a mathematical model specifically designed to address non-orthogonal joint errors. In explosion-proof robots, joints may be non-orthogonal, meaning their coordinate axes are not perpendicular to each other. This non-orthogonality can lead to errors in joint motion.

[0057] Non-orthogonal joint error refers to the deviation of the actual relative relationship between joints from the ideal orthogonal relationship due to deformation during the manufacturing, assembly, or use of explosion-proof robot joints. For example, in theory, the coordinate axes of two joints should be perpendicular (orthogonal) to each other, but in reality, there may be some deviation due to problems such as machining accuracy.

[0058] The joint error compensation matrix is ​​a data structure in matrix form, which is used to quantify the non-orthogonality deviations between joints. The elements in the matrix represent the specific numerical relationship between the non-orthogonality errors between different joints. Through this matrix, the joint errors can be accurately compensated.

[0059] In this embodiment, the joint feature set is input into the skew compensation model to model and compensate for non-orthogonal joint errors, thereby generating a joint error compensation matrix. This step, based on the previously obtained joint feature set, further analyzes the non-orthogonality errors between joints and calculates the compensation matrix through model calculation, providing a basis for subsequent corrections to the kinematic model of the explosion-proof robot.

[0060] In one embodiment, a skew compensation model can be used to analyze the input joint feature set. For positional features, the relative positional relationships of the joints in space are determined; for posture features, joint orientation deviations are analyzed; and for motion features, the effects of velocity and acceleration on non-orthogonality are considered. Within the model, the relationships between the joints are described by establishing spatial vector equations. For example, for two joints J1 and J2, let the vectors corresponding to their ideal orthogonal relationship be V1 and V2, respectively, while the vectors corresponding to the actual measured relationship are V1' and V2'. The deviation vectors are then calculated using a vector difference operation. A matrix is ​​then constructed based on these deviation vectors; this matrix is ​​the joint error compensation matrix. The elements in this matrix quantify the non-orthogonality deviations between the joints. This technical implementation effectively models and quantifies non-orthogonal joint errors through precise vector analysis and matrix construction. The technical effect is that it can accurately determine the non-orthogonality deviations between joints, providing critical compensation information for revising the kinematic model.

[0061] In one embodiment, a genetic algorithm is used to optimize the skew compensation model to generate a joint error compensation matrix. First, a joint feature set is used as input data for the genetic algorithm, and the parameters of the skew compensation model are encoded and represented as chromosomes. Next, a fitness function is defined, which evaluates the quality of chromosomes (i.e., parameter combinations) based on their effectiveness in compensating joint errors. Through genetic operations such as selection, crossover, and mutation, the population is continuously evolved to find the chromosome that maximizes the fitness function value. The parameters corresponding to this chromosome are the optimal parameters of the skew compensation model. Finally, the joint error compensation matrix is ​​constructed based on these optimal parameters. This genetic algorithm-based method has global search capabilities, avoiding being trapped in local optimal solutions. Technical Effect: Utilizing the global search capabilities of the genetic algorithm, the parameter space of the skew compensation model can be more comprehensively searched, resulting in a more optimal parameter combination, thereby generating a more accurate joint error compensation matrix. This improves the compensation effect for non-orthogonal joint errors and, consequently, enhances the overall performance of kinematic error correction in explosion-proof robots.

[0062] In step S40 , the kinematic model of the explosion-proof robot is corrected according to the joint error compensation matrix to obtain a corrected kinematic model, wherein the kinematic model describes the correlation between the position, velocity, and acceleration of each joint of the explosion-proof robot.

[0063] A kinematic model is a mathematical model that describes the relationship between the position, velocity, and acceleration of each joint in an explosion-proof robot. It is based on the robot's mechanical structure and motion principles, for example, using the kinematic equations of the joints to represent the motion transfer relationship between them.

[0064] The modified kinematic model is a model that is modified based on the original kinematic model according to the joint error compensation matrix. This model can more accurately reflect the actual motion relationship of the explosion-proof robot joints.

[0065] In this embodiment, modifying the kinematic model of the explosion-proof robot based on the joint error compensation matrix to obtain a corrected kinematic model is a crucial step. Because the original kinematic model may not accurately reflect the robot's actual motion due to errors such as joint non-orthogonality, this correction improves the accuracy of the kinematic model, making it more consistent with actual joint motion relationships.

[0066] In one embodiment, elements in the joint error compensation matrix can be integrated with relevant parameters in the kinematic model. For example, the kinematic model contains matrix equation A, which describes the position relationship between joints, matrix equation B, which describes the velocity relationship, and matrix equation C, which describes the acceleration relationship. The position deviation elements in the joint error compensation matrix are added or multiplied (depending on the specific model structure and correction algorithm) with the corresponding elements in matrix equation A. A similar operation is performed for the matrix equations describing the velocity and acceleration relationships, integrating the velocity and acceleration deviation elements in the joint error compensation matrix with the corresponding matrix equations. In this way, the correlation between position, velocity, and acceleration in the kinematic model is corrected, resulting in a corrected kinematic model. This technical implementation directly integrates error compensation information into the kinematic model, ensuring that the model accurately reflects the actual joint motion relationships. The technical effect is to improve the accuracy of the kinematic model, making it more consistent with the actual motion state of the explosion-proof robot.

[0067] In step S50 , corresponding control instructions are generated based on the modified kinematic model, and the control instructions are sent to the explosion-proof robot to perform motion operations.

[0068] Control instructions are commands used to instruct the explosion-proof robot to perform various actions. They are calculated based on the modified kinematic model. For example, control instructions can include information such as the rotation angle, rotation speed, and movement direction of each joint.

[0069] In this embodiment, the final step in the kinematic error correction method is to generate corresponding control instructions based on the corrected kinematic model and send them to the explosion-proof robot to execute the motion. Using the accurate corrected kinematic model obtained in the previous steps, appropriate control instructions are generated, achieving precise control of the explosion-proof robot's motion.

[0070] In one embodiment, a modified kinematic model can be used to calculate parameters such as the target position and velocity required for each joint based on the current task requirements (such as reaching a specific location or performing a specific action). These parameters are then converted into a control instruction format recognizable by the explosion-proof robot. For example, if the explosion-proof robot utilizes a specific communication protocol and instruction encoding scheme, the calculated joint parameters are encoded accordingly to form control instructions. Finally, the control instructions are transmitted to the explosion-proof robot's controller via a wireless communication module (such as Wi-Fi, Bluetooth, or a specialized industrial wireless communication protocol module). Upon receiving the control instructions, the explosion-proof robot's controller drives the motors and other actuators of each joint to perform motion operations according to the instructions. This technical implementation ensures that the explosion-proof robot accurately moves according to the error-corrected motion requirements. The technical effect is to improve the accuracy and stability of the explosion-proof robot's motion, enabling it to better complete explosion-proof tasks, such as accurately transporting dangerous goods in hazardous environments or conducting hazardous area detection operations.

[0071] In one embodiment, reference Figure 3 , step S10 may include:

[0072] Step S101: using a laser calibration instrument with a preset accuracy to collect the three-dimensional spatial coordinate data of each joint of the explosion-proof robot in static and dynamic states.

[0073] Three-dimensional spatial coordinate data refers to the coordinate values ​​that describe the position of an object in three-dimensional space (usually using the Cartesian coordinate system, i.e., the space formed by the x, y, and z axes). For the joints of an explosion-proof robot, these coordinate values ​​determine the specific position of the joint in space.

[0074] The static state refers to the state in which the joints of the explosion-proof robot are stationary. At this time, the positions of the joints are relatively fixed, and no displacement, rotation or other movements occur.

[0075] The dynamic state is the opposite of the static state. It refers to the state in which the joints of the explosion-proof robot are in motion and their positions, postures, etc. change over time.

[0076] The embodiment of the present application can use a laser calibrator with preset accuracy to collect three-dimensional spatial coordinate data of each joint of the explosion-proof robot in static and dynamic states. First, for the collection in the static state, this is to obtain the initial position information of the joint. In many application scenarios, it is very important to know the initial position of the joint. For example, in some precise explosion-proof operation tasks, such as the seal inspection of hazardous chemical containers, the accuracy of the initial position of the robot joint may affect the accuracy of the entire task. When the joint is in a static state, the laser calibrator can stably measure its coordinate data, which can be used as a benchmark for subsequent dynamic measurement and error analysis.

[0077] The purpose of dynamic data collection is to capture the positional changes of joints during motion. When performing tasks such as handling explosives, explosion-proof robots experience constant joint movement. The movement trajectory and positional changes of joints can significantly impact operational safety and accuracy. By collecting dynamic coordinate data, we can analyze joint positional deviations during different phases of motion, providing comprehensive data support for kinematic error correction.

[0078] Collecting 3D coordinate data in these two states provides a more comprehensive description of the joint's spatial positional characteristics. Static data reflects the joint's basic positional relationships, while dynamic data reflects the joint's changes during actual motion. Combining the two allows for a more precise analysis of the joint's kinematic characteristics, providing a richer data base for subsequent error correction.

[0079] In one embodiment, reference Figure 4 , step S101 can be implemented as follows:

[0080] Step S1011: using a laser calibration instrument with a preset accuracy to collect the mechanical structure surface of each joint of the explosion-proof robot in static and dynamic states to obtain spatial coordinate point cloud data of the joint.

[0081] Spatial coordinate point cloud data is a collection of vectors in a three-dimensional coordinate system. For explosion-proof robot joints, spatial coordinate point cloud data is the 3D coordinate information of numerous discrete points on the joint's mechanical surface, collected by a laser calibrator. These points collectively describe the joint's surface shape and position.

[0082] In this step, a laser calibration instrument with a preset accuracy is used to capture the mechanical surface of each joint of the explosion-proof robot in both static and dynamic states, aiming to obtain spatial coordinate point cloud data of the joints. In the static state, the collected point cloud data can reflect the surface characteristics and positional relationships of the joints at rest. For example, for a complex joint, the point cloud data can accurately depict its surface irregularities and overall spatial layout.

[0083] When a joint moves, the point cloud data on its mechanical surface changes over time. These changes include information about the joint's position, posture, and possible deformation during motion. For example, when an explosion-proof robot performs high-load motion, the joint may undergo subtle elastic deformation, which is reflected in the dynamically collected point cloud data. Analyzing this data provides a more comprehensive understanding of the joint's characteristics under different operating conditions.

[0084] Collecting spatial coordinate point cloud data from the surface of a joint's mechanical structure is a comprehensive way to obtain spatial information about the joint. Compared to only collecting the coordinates of specific joint points, point cloud data can more accurately depict the joint's shape and position, providing rich foundational data for subsequent processing and analysis.

[0085] In one embodiment: multi-view laser scanning technology is used. Multiple laser calibrators are set around the joint, and each calibrator scans the mechanical structure surface of the joint from a different perspective. In a static state, multiple calibrators emit laser beams in sequence according to a predetermined order to perform a comprehensive scan of the joint surface. For a dynamic state, a synchronous scanning method can be used, that is, at the same moment, multiple calibrators scan the moving joint at the same time. The point cloud data collected by each laser calibrator contains the coordinate information of a partial area of ​​the joint surface. Finally, these data from different perspectives are fused to obtain the complete spatial coordinate point cloud data of the joint in static and dynamic states.

[0086] Step S1012: filtering the spatial coordinate point cloud data and extracting the geometric center coordinates of key nodes.

[0087] Filtering is a process that processes data to remove noise, outliers, and other interfering factors. In spatial coordinate point cloud data, there may be some inaccurate data points due to measurement errors, environmental interference, and other factors. Filtering can improve data quality.

[0088] Key nodes are representative points selected from the spatial coordinate point cloud data. These points may be key points of joint surface shape changes or points of special significance in joint motion analysis.

[0089] For a group of points (such as key nodes), the geometric center coordinates are calculated through specific mathematical calculations to represent the overall position of these points. They are obtained by taking a weighted average of the coordinate values ​​of these points.

[0090] In this step, the spatial coordinate point cloud data is filtered to improve the accuracy and reliability of the data. During the acquisition process, it may be interfered with by various factors, such as laser scattering, dust in the environment, etc. These factors may cause noise points in the collected point cloud data. For example, some outliers may make the shape of the joint surface appear discontinuous or have erroneous protrusions or depressions. Through filtering, such as Gaussian filtering or median filtering, these noise points can be removed, so that the point cloud data can more realistically reflect the surface characteristics of the joint.

[0091] Extracting the geometric center coordinates of key nodes is intended to simplify data processing and highlight the key features of the joint. Joint spatial coordinate point cloud data can contain a large number of points, and directly processing these points increases computational complexity and time costs. By identifying key nodes, attention can be focused on points that are more important for joint motion and structural analysis. For example, points near joint connections and rotation axes may be defined as key nodes. The geometric center coordinates of these key nodes can, to a certain extent, represent the overall position of the joint.

[0092] In one embodiment, a statistical filtering method can be used for filtering. First, the average distance from each point in the spatial coordinate point cloud data to its adjacent points is calculated. A distance threshold is set, and points with a distance greater than the threshold are regarded as noise points and removed. For the extraction of key nodes, some rules can be pre-set according to the mechanical structure model of the joint. For example, for a joint with a cylindrical structure, the points on the contour line of the cylindrical surface and the points near the center of the two end faces of the cylinder are defined as key nodes. Then, for these key nodes, their geometric center coordinates are calculated. For key nodes on a plane, the center coordinates can be obtained by simply averaging their x and y coordinates; for key nodes in three-dimensional space, the geometric center coordinates can be calculated according to the weighted average method of spatial coordinates.

[0093] Step S1013: Calculate the three-dimensional space coordinate data of the joint according to the geometric center coordinates of the key node.

[0094] In this step, the three-dimensional spatial coordinate data of the joint is calculated based on the geometric center coordinates of the key nodes. The geometric center coordinates of the key nodes are the position information of the important feature points of the joint in space. Since these key nodes can represent the overall characteristics of the joint to a certain extent, it is reasonable to obtain the three-dimensional spatial coordinate data of the joint based on these geometric center coordinates through a specific calculation method. For example, for a simple ball joint, its key nodes may be the center of the ball and several contact points between the ball and the connection part. The geometric center coordinates of these points can be converted into the three-dimensional spatial coordinate data of the entire ball joint through a certain transformation relationship.

[0095] The advantage of this calculation method is that it simplifies the data processing process. The process of gradually simplifying and refining the information from a large amount of spatial coordinate point cloud data to the geometric center coordinates of key nodes and then to the three-dimensional spatial coordinate data of the joints is a process. This reduces the amount of calculation and also highlights the key spatial characteristics of the joints, facilitating subsequent operations such as kinematic error correction.

[0096] Furthermore, calculating the 3D joint coordinates using the geometric center coordinates of key nodes can improve data stability. Because key nodes are representative points that have been screened and processed, the joint coordinate data calculated based on them is relatively stable and less susceptible to factors such as local noise or minor deformations.

[0097] In one embodiment: a geometric model is established according to the mechanical structure of the joint. For example, if the joint is a rectangular parallelepiped structure, it is assumed that the key nodes are located at the centers of the eight vertices and six faces of the rectangular parallelepiped. According to the geometric center coordinates of these key nodes, calculations are performed using the geometric relationship of the rectangular parallelepiped. The center coordinates of the rectangular parallelepiped (that is, an approximation of the three-dimensional spatial coordinate data of the joint) can be obtained by calculating the average value of the coordinates of the vertex key nodes. If the structure of the joint is more complex, such as a compound joint with multiple substructures, the compound joint can be decomposed into several simple geometric structures, and the geometric center coordinates of the key nodes corresponding to each simple structure are calculated separately. Then, based on the connection relationship and relative position relationship between these substructures, the three-dimensional spatial coordinate data of the entire compound joint can be calculated through mathematical methods such as coordinate transformation.

[0098] Step S102: detecting the trajectory changes of the joint during the continuous motion process, and performing time series analysis on the detected trajectory change information to extract motion trajectory information, wherein the motion trajectory information includes the displacement curve of the joint and its time derivative.

[0099] Trajectory change refers to the change in the spatial position of the explosion-proof robot joints over time during continuous motion. This change can be a translation in one direction, a rotation around an axis, or a position change caused by complex spatial motion.

[0100] Time series analysis is a method for analyzing data in chronological order. For joint trajectory information, time series analysis can reveal the trajectory's changing patterns, trends, and periodicity at different time points.

[0101] A displacement curve describes the change in joint displacement in space over time. The curve is plotted with time on the horizontal axis and the joint displacement (either in a specific direction or as a whole) on the vertical axis.

[0102] Time derivative represents the rate of change of displacement over time. In physical terms, it reflects the speed of joint movement.

[0103] The embodiments of the present application can detect changes in the trajectory of a joint during continuous motion and perform time series analysis on the detected trajectory change information to extract motion trajectory information. First, detecting trajectory changes is to understand the actual motion path of the joint during motion. For example, when an explosion-proof robot performs complex operating tasks, the joint may move along a specific trajectory, but due to various factors (such as joint wear, external interference, etc.), the actual trajectory may deviate from the ideal trajectory. By detecting trajectory changes, the existence of these deviations can be discovered in a timely manner.

[0104] Because joint motion is a continuous process that changes over time, time series analysis can extract useful information from large amounts of trajectory data. For example, analysis can reveal whether there are periodic fluctuations in joint motion, or whether there are abnormal accelerations or decelerations within a certain time period. This information is essential for accurately assessing the joint's motion state.

[0105] The displacement curve and its time derivative in the motion trajectory information are important indicators of the joint's kinematic characteristics. The displacement curve intuitively shows the joint's movement trajectory in space, while the time derivative further reveals the changes in the joint's velocity. For example, if the time derivative of the displacement curve suddenly increases within a certain period of time, it may mean that the joint experienced a sudden acceleration during this period, which may be due to a sudden change in the control command or an external impact on the joint. By analyzing this information, we can gain a deeper understanding of the joint's kinematic characteristics, thus providing a basis for kinematic error correction.

[0106] In one embodiment, special reflective markers can be installed on the joints, and a laser calibrator emits a laser beam to continuously track these markers, thereby accurately acquiring information about the joint's trajectory changes. For data processing, specialized time series analysis software (such as the pandas and statsmodels libraries in Python) is utilized. First, the collected trajectory change data is organized into a time series data format in chronological order. Then, the displacement data is fitted to obtain a displacement curve. To calculate the time derivative, numerical differencing methods can be used, such as using the central difference formula to calculate the first-order time derivative of the displacement curve. This method accurately extracts motion trajectory information, such as the joint displacement curve and its time derivative, while also enabling in-depth time series analysis of trajectory changes.

[0107] In one embodiment, reference Figure 5 , step S30 may specifically include:

[0108] S301 , analyzing sources of non-orthogonality errors between joints according to the joint feature set to construct an initial parameter matrix of a skew compensation model.

[0109] Sources of non-orthogonality errors: In the joint system of explosion-proof robots, orthogonality represents an ideal vertical relationship between joints. However, due to the influence of various factors, the actual joint relationship will deviate from this ideal state. These factors are the sources of non-orthogonality errors. These factors cover many aspects, including processing errors in the manufacturing process. For example, the precision limitations of the processing equipment may cause dimensional deviations of joint parts, thereby affecting the assembly relationship of the joints and generating non-orthogonality errors; human factors or limitations of the assembly process during assembly, such as inaccurate positioning during assembly or incorrect assembly sequence; joint wear is also a significant factor during the long-term operation of the robot. For example, friction at the joint connection will cause the relative position and posture of the joint to change; external environmental factors cannot be ignored. For example, temperature changes may cause thermal expansion and contraction of the joint material, thereby changing the relative relationship between the joints.

[0110] The initial parameter matrix of the skew compensation model is a mathematical matrix structure specifically used for constructing the skew compensation model. It is a collection of parameters derived from an analysis of the sources of joint non-orthogonality errors. These parameters are preliminary and unoptimized, and they quantitatively describe the factors associated with potential non-orthogonality errors between joints. For example, elements in the matrix may represent estimated joint position deviations along different coordinate axes, estimated angular deviations of joint poses relative to the ideal orthogonal pose, and coefficients related to joint motion velocity and acceleration that may affect non-orthogonality errors. These parameters provide a starting point for subsequent model fitting and optimization.

[0111] The embodiment of the present application can analyze the sources of non-orthogonality errors between joints based on the joint feature set to construct the initial parameter matrix of the skew compensation model. The joint feature set contains key information of many aspects of the joint, which is an important basis for analyzing the sources of non-orthogonality errors. In terms of position characteristics, if the actual measured joint coordinate position deviates from the coordinate position under the theoretical orthogonal relationship, this may indicate insufficient machining accuracy or inaccurate assembly during the manufacturing process. For example, the coordinate position of a joint in the x-axis direction is offset by a certain value compared to the position under the ideal orthogonal relationship. This may be due to a deviation in the dimensional control in the x-axis direction when machining the joint, or the joint was not installed in place in the x-axis direction during assembly.

[0112] In the embodiment of the present application, the posture characteristics can also provide a basis for the analysis of the source of non-orthogonality errors. When the actual posture angle of a joint is inconsistent with the ideal orthogonal posture angle, this may be due to the wear of the joint transmission components or inaccurate posture adjustment during assembly. For example, the rotation angle of a joint deviates from the angle it should have in the ideal orthogonal state. This may be because the gears inside the joint wear after long-term use, resulting in inaccurate rotation, thereby generating non-orthogonality errors. The changes in speed and acceleration in the motion characteristics can also reflect the problem. If the speed is uneven or the acceleration changes abnormally during joint movement, it may mean that there are defects in the coordination between the joints. This may be due to uneven assembly clearance or unstable power transmission inside the joint. These situations may be factors that lead to non-orthogonality errors.

[0113] The construction of the initial parameter matrix in this embodiment of the application is a process of quantifying and organizing the factors related to non-orthogonality error obtained from these analyses. Each element in this matrix corresponds to a factor that may affect joint non-orthogonality error. In this way, the foundation is laid for the further construction and optimization of the skew compensation model.

[0114] S302 : Fitting the initial parameter matrix using the least square method to obtain preliminary parameters of the skew compensation model.

[0115] The least squares method is a mathematical optimization method whose core idea is to find the best function matching the data by minimizing the sum of squared errors. In this scenario, the goal is to find a set of parameters (i.e., the parameters of the skew compensation model) that minimizes the sum of squared errors between the non-orthogonality errors calculated based on these parameters and the actual observed non-orthogonality errors.

[0116] The preliminary parameters of the skew compensation model are obtained by fitting the initial parameter matrix using the least squares method. Compared to the initial parameters in the initial parameter matrix, these parameters better reflect the actual non-orthogonality error relationship between joints. They are intermediate results in the process of building an accurate skew compensation model and will serve as the basis for further optimization.

[0117] In step S302, the least squares method is used to fit the initial parameter matrix to obtain preliminary parameters for the skew compensation model. The principle of the least squares method is based on minimizing the sum of squared errors between the predicted values ​​and the actual observed values. In this specific application scenario, the parameters in the initial parameter matrix are preliminary settings and may not accurately reflect the non-orthogonality error relationship between the joints. By using the least squares method for fitting, the initial parameters can be adjusted and optimized based on the existing joint feature data and the actual observed non-orthogonality error conditions.

[0118] When applying the least squares method to the initial parameter matrix, it comprehensively considers all joint feature data points. For example, based on the relationship between joint position, posture, and motion feature data collected at multiple different times and the initial parameter matrix, the least squares method calculates a new set of parameters. This new set of parameters ensures that the calculated non-orthogonality error is as close as possible to the actual measured error. This calculated set of parameters serves as the preliminary parameters for the skew compensation model.

[0119] In one embodiment, the parameters in the initial parameter matrix are used as variables to establish an error function. This error function takes the joint feature data as input and outputs the sum of the squares of the difference between the calculated non-orthogonality error and the actual measured non-orthogonality error. Specifically:

[0120] Assume that the joint feature data is matrix X and the initial parameter matrix is , the actual measured non-orthogonality error is y, then the calculated non-orthogonality error can be expressed as , the error function is , where n is the number of data points. Then, the least squares algorithm is used to find the minimum value of this error function. In actual calculations, matrix operations can be used. For example, by taking the derivative of the error function and setting the derivative to zero, we get: , thereby solving the preliminary parameters of the skew compensation model.

[0121] S303: Regularize the preliminary parameters to obtain optimal parameters of the skew compensation model.

[0122] Regularization is a technical method to prevent model overfitting. During model building, overfitting occurs when a model overlearns noise and special cases in the training data, resulting in poor performance on new data. Regularization adds a regularization term to the optimization objective to constrain the range of model parameters, thereby preventing overfitting caused by overly complex models.

[0123] The optimal parameters of the skew compensation model: These are the parameters after regularization. These parameters comprehensively consider the needs of data fitting and preventing overfitting. They can more accurately model and compensate for the non-orthogonality errors between joints. They are the ideal parameters for the skew compensation model to ultimately generate the joint error compensation matrix.

[0124] In step S303, the preliminary parameters are regularized to obtain the optimal parameters for the skew compensation model. Regularization is necessary because, when obtaining the preliminary parameters through least squares analysis, data noise or overfitting may lead to unreasonable parameter values. For example, some parameters may become too large or too small, resulting in poor performance of the skew compensation model when applied to new data.

[0125] Regularization constrains the range of parameter values ​​by adding a regularization term to the optimization objective. For skew compensation models, this regularization term can be selected based on the model's characteristics and the data distribution. For example, L1 regularization or L2 regularization can be used. Regularization adjusts the initial parameters to a more reasonable range, making the skew compensation model more stable and capable of better generalization.

[0126] In one embodiment, L2 regularization can be selected to add a regularization term to the error function. This regularization term can be expressed as the square of the second norm of the parameter vector multiplied by a regularization coefficient . That is, the new objective function is the error function in is the initial parameter vector. Then, the optimization algorithm (such as gradient descent) is used again to solve the minimum value of this new objective function. In each iteration, the parameter vector is updated according to the gradient information. , until convergence to obtain the optimal parameters of the skew compensation model. Specifically, for the objective function , calculate its Gradient , and then follow the update formula (where is the learning rate) is updated iteratively until Convergence is achieved to obtain the optimal parameters. In this way, while ensuring data fitting, the size of the parameters is effectively controlled to prevent overfitting.

[0127] S304: Generate a joint error compensation matrix according to the optimal parameters.

[0128] The joint error compensation matrix is ​​used to quantify the non-orthogonality between joints. The elements in the matrix reflect the non-orthogonality between joints in different directions (such as x, y, and z). This matrix clearly defines the non-orthogonality between each joint and the others, providing a precise basis for subsequent corrections to the explosion-proof robot's kinematic model.

[0129] In step S304, a joint error compensation matrix is ​​generated based on the optimal parameters. These optimal parameters are obtained through analysis, fitting, and regularization in the previous steps. These parameters accurately reflect the non-orthogonality errors between joints. The primary function of the joint error compensation matrix is ​​to quantify the non-orthogonality deviations between joints, facilitating subsequent corrections to the kinematic model of the explosion-proof robot.

[0130] The optimal parameters are arranged according to specific rules into a joint error compensation matrix. For example, the matrix dimension may be determined by the number of joints. If an explosion-proof robot has n joints, then the joint error compensation matrix may be an n×n square matrix. The elements in the matrix are filled according to the corresponding relationships between the joints. For example, for a non-orthogonal deviation between two joints i and j in a certain direction, the corresponding optimal parameter is placed at position (i, j) in the matrix (or (j, i) if the deviation is symmetric). In this way, a complete joint error compensation matrix is ​​constructed.

[0131] Generating the joint error compensation matrix is ​​the final result of the entire skew compensation model, which models and compensates for non-orthogonal joint errors. This matrix is ​​directly applied to the kinematic model, thereby improving the accuracy of the explosion-proof robot's kinematic model and, in turn, the robot's motion control precision.

[0132] In one embodiment, reference Figure 6 , step S50 may specifically include:

[0133] Step S401: introducing the joint error compensation matrix into the original kinematic model of the explosion-proof robot to form a revised kinematic model.

[0134] The original kinematic model is a mathematical model that describes the relationship between the position, velocity, and acceleration of each joint of an explosion-proof robot before considering joint error compensation. It is based on the robot's mechanical structure, joint connections, and kinematic principles, and is a mathematical expression of the robot's ideal motion state.

[0135] In the embodiments of the present application, although the original kinematic model can describe the basic motion relationships between the robot joints, it may have deviations in actual applications because it does not take into account factors such as non-orthogonality errors between the joints. The joint error compensation matrix is ​​a quantitative representation of the non-orthogonality deviations between the joints. By introducing this matrix into the original kinematic model, the parameters in the model can be adjusted to make the model more similar to the actual motion situation.

[0136] For example, the equations in the original kinematic model might describe the relationship between a joint's position and the positions, velocities, and accelerations of other joints. When the joint error compensation matrix is ​​introduced, the elements in the matrix modify the coefficients in these equations. For example, if the coefficient for the relationship between the position of joint 1 and the velocity of joint 2 in the original model is , after the joint error compensation matrix is ​​introduced, this coefficient might be modified to based on the corresponding elements in the matrix to reflect the impact of the actual non-orthogonality between the joints on this relationship.

[0137] Introducing the joint error compensation matrix is ​​an effective way to improve the original kinematic model. It can improve the accuracy of the kinematic model, enabling it to better predict and describe the joint states of the explosion-proof robot during actual motion, and provide a more reliable model foundation for subsequent precise control and error correction.

[0138] In one embodiment: Assume that the original kinematic model can be expressed as a set of matrix equations Y=AX, where Y is a vector representing the joint position, velocity or acceleration, A is a coefficient matrix, and X is a vector related to the joint variables. The joint error compensation matrix is ​​introduced into this model and can be corrected by matrix multiplication. For example, A'=A×E is corrected to (the multiplication here is specifically defined according to the properties of the matrix and the requirements of the model, and may be ordinary matrix multiplication or other special matrix operations), and then the corrected kinematic model becomes Y=A'X. In this way, the joint error compensation matrix is ​​introduced into the original kinematic model to obtain the corrected kinematic model.

[0139] Step S402: numerically solve the modified kinematic model to obtain the actual spatial position, velocity and acceleration values ​​of the joint.

[0140] Numerical solution is the process of solving mathematical models through numerical calculations. Since kinematic models are typically composed of a set of equations, numerical solution involves using specific numerical algorithms to calculate the specific values ​​of the variables in the equations (such as the spatial position, velocity, and acceleration of joints) under given initial and boundary conditions.

[0141] In the embodiments of this application, although the modified kinematic model theoretically more accurately describes the relationship between joints, obtaining the specific joint state values ​​requires numerical solution. For example, in some complex explosion-proof robot motion scenarios, the equations in the model may be nonlinear and cannot be solved by simple analytical methods.

[0142] The numerical solution process takes into account various parameters and constraints in the model. For example, the robot's initial joint positions, velocity limits, and acceleration limits are known as initial and boundary conditions. Starting from these known conditions, the numerical solution algorithm gradually calculates the actual spatial position, velocity, and acceleration values ​​of the joints at different times. This provides a deeper understanding of the joint's actual motion state, providing accurate data support for subsequent model optimization and robot control.

[0143] Obtaining the actual spatial position, velocity, and acceleration of the joints is crucial for evaluating the accuracy of the modified kinematic model. If the numerically calculated values ​​are close to the actual measured values ​​(obtained through sensors and other equipment), the modified kinematic model effectively reflects the robot's actual motion. Otherwise, further adjustments to the model or optimization of the numerical solution algorithm may be necessary.

[0144] In one embodiment, the Runge-Kutta method can be used for numerical solution. The modified kinematic model is converted into a system of first-order ordinary differential equations. For example, if the modified kinematic model describes the relationship between the joint position x(t), velocity v(t), and acceleration a(t), it can be expressed as a system of equations in the form of dx / dt=v(t), dv / dt=a(t), etc. Then, based on the known initial conditions, such as x(0), v(0), etc., the iterative formula of the Runge-Kutta method is used for calculation. In each iteration, the joint position, velocity, and acceleration values ​​for the next time step are calculated based on the current state value. Through multiple iterations, the actual spatial position, velocity, and acceleration values ​​of the joint at different times are obtained.

[0145] Step S403: Generate a modified kinematic model according to the actual spatial position, velocity and acceleration values ​​of the joint.

[0146] The present invention can further optimize or reconstruct the modified kinematic model based on the actual joint state values ​​obtained by the numerical solution. These actual values ​​reflect the actual state of the joints in actual motion, and incorporating them into the kinematic model can make the model more accurately describe the relationship between the joints.

[0147] For example, if the actual velocities and accelerations of certain joints are found to deviate significantly from the values ​​predicted by the modified kinematic model during the numerical solution, the relevant parameters or equation structure in the model can be adjusted based on these actual values. This may involve re-evaluating the coupling relationships between the joints in the model or modifying the equations that describe the changes in velocity and acceleration.

[0148] The goal of generating a revised kinematic model is to obtain a model that more accurately reflects the actual motion state of the explosion-proof robot. This model will be used in subsequent control command generation and other operations to ensure that the robot operates according to the expected motion trajectory and state, improving the accuracy and reliability of the robot's motion.

[0149] In one embodiment: first, the actual spatial position, velocity and acceleration values ​​of the joint obtained by numerical solution are statistically analyzed. For example, the statistical quantities such as the mean value and variance of these values ​​are calculated. Then, the parameters in the revised kinematic model are adjusted according to these statistical quantities. If it is found that the actual velocity variance of a certain joint is large, it means that the motion stability of the joint is poor, and it may be necessary to add a constraint term for velocity fluctuation in the kinematic model or adjust the coefficient related to the joint. Through this statistical analysis and parameter adjustment based on actual values, a revised kinematic model that is more in line with the actual situation is reconstructed.

[0150] In one embodiment, step S403 may specifically include:

[0151] Calculate the errors between the spatial position, velocity and acceleration values ​​of the joint and their corresponding theoretical values; if all errors are less than a preset threshold, determine the modified kinematic model as the final model; if at least one error is greater than or equal to the preset threshold, iteratively optimize the joint error compensation matrix until the error meets the requirements.

[0152] In the kinematic model of an explosion-proof robot, theoretical values ​​refer to the spatial position, velocity, and acceleration of the joints calculated based on an ideal kinematic model (without considering actual errors). These values ​​are determined based on the robot's design parameters, ideal joint relationships, and kinematic principles, and represent the state the joints should achieve, assuming no errors or interference.

[0153] The purpose of this step is to quantify the degree of deviation between the actual motion state and the ideal state. For spatial position, by comparing the joint position coordinates obtained by actual measurement or numerical solution with the position coordinates calculated theoretically, the deviation of the joint position in three-dimensional space can be determined. For example, in a simple planar joint motion, if the joint should theoretically be at the coordinate (x 0 , y0), and the actual numerical solution position is (x 1 , y1), then the position error can be obtained by calculating the distance formula between the two points.

[0154] Error calculation is equally important for velocity and acceleration values. Velocity error reflects the difference between the actual joint motion velocity and the theoretical velocity. For example, the theoretical velocity may be a constant value preset based on task requirements and the robot's motion plan, while the actual velocity may vary due to various factors such as joint friction and external interference. The same is true for acceleration error, which reflects the inconsistency between the actual joint acceleration and the theoretical acceleration. This error calculation helps to comprehensively evaluate the accuracy of the kinematic model, as position, velocity, and acceleration are key factors in describing the joint motion state.

[0155] Calculating these errors is the basis for determining whether the kinematic model meets the requirements and for further optimization. Only by accurately knowing the deviation between the actual and theoretical values ​​can we determine whether the model needs to be improved and how to improve it.

[0156] In one embodiment: For the calculation of spatial position error, the theoretical position coordinates and actual position coordinates of the joint are first converted to the same coordinate system (if necessary). Then, the Euclidean distance is calculated as the position error based on the coordinate difference. For the velocity error, the theoretical velocity value and the actual velocity value are subtracted to obtain the velocity difference. In order to avoid the influence of positive and negative values, the absolute value of the difference can be taken as the velocity error. For the acceleration error, a similar method is used to subtract the theoretical acceleration value and the actual acceleration value and take the absolute value to obtain the acceleration error. Finally, these position, velocity and acceleration errors are respectively stored in a data structure (such as an array or matrix) for subsequent processing and analysis.

[0157] The preset threshold is a pre-set value used to determine whether the error between the actual and theoretical values ​​of the joint's spatial position, velocity, and acceleration is within an acceptable range. This threshold is determined based on factors such as the specific application requirements of the explosion-proof robot, the required accuracy, and previous experience.

[0158] In one embodiment, if all errors are less than a preset threshold, the revised kinematic model is determined to be the final model. This means that when the deviations between the actual values ​​of the joint's spatial position, velocity, and acceleration and the theoretical values ​​are sufficiently small, within a preset acceptable range, the current revised kinematic model is able to accurately describe the explosion-proof robot's motion state. For example, in scenarios where the accuracy of the explosion-proof robot's motion is not particularly demanding, such as performing simple hazardous material handling tasks in relatively open spaces, as long as the joint position error is within ±5 mm, the velocity error is within ±0.1 m / s², and the acceleration error is within ±0.05 m / s² (these values ​​are for example only), the current model can be considered the final model that meets the requirements.

[0159] If at least one error is greater than or equal to the preset threshold, the joint error compensation matrix is ​​iteratively optimized until the error meets the requirements. When one or some errors exceed the acceptable range, it means that the current joint error compensation matrix may not be accurate enough and needs further optimization. For example, if the position error of a joint is too large, it may be because the previous modeling of the joint non-orthogonality error is not accurate enough, and the relevant elements in the joint error compensation matrix need to be readjusted. Through iterative optimization, the joint error compensation matrix is ​​continuously adjusted, the corrected kinematic model is recalculated, and then the error between the actual value and the theoretical value is calculated again until all errors are less than the preset threshold. This iterative process can gradually improve the accuracy of the kinematic model, making it more consistent with the actual movement of the explosion-proof robot.

[0160] The embodiment of the present application can obtain a final kinematic model that can accurately describe the movement of the explosion-proof robot through comparison with preset thresholds and iterative optimization, thereby providing a reliable model basis for subsequent control instruction generation and accurate movement of the robot.

[0161] In one embodiment: a loop structure can be set to implement an iterative optimization process. In each iteration, it is first determined whether all errors are less than a preset threshold. If so, the iteration is ended and the current corrected kinematic model is determined to be the final model. If not, the joint error compensation matrix is ​​adjusted according to the size and direction of the error. For example, if the position error is too large and is a positive value, it may be necessary to increase the element value related to the joint position in the joint error compensation matrix; if the speed error is too large and is a negative value, it may be necessary to reduce the element value related to the speed in the joint error compensation matrix. After adjustment, the corrected kinematic model is recalculated according to the previous steps, and then the new error value is calculated to enter the next iteration until all errors meet the requirements.

[0162] In one embodiment, reference Figure 7 , step 20 may specifically include:

[0163] Step S201: extract key features from the spatial position information to obtain position features of the joints.

[0164] Key feature extraction is an operation that filters out information that is crucial for describing joint characteristics from raw spatial position information. Spatial position information may contain a large number of data points, but not all data are equally important for accurately describing joint position characteristics. Key feature extraction aims to identify information that represents the essential characteristics of joint position.

[0165] Spatial position information is typically a series of coordinate points representing joints in three-dimensional space. This data can be complex and redundant. Simply using all of this data to describe joint position features increases computational complexity and may include unnecessary noise. Key feature extraction can simplify this data and highlight key characteristics of joint positions.

[0166] For example, for an explosion-proof robot with multiple joints, the joint positions may fluctuate within a certain range. Within this fluctuation range, there may be key coordinate points or coordinate intervals that can effectively define the joint position characteristics. For example, the extreme position of a joint in a specific direction and the average position of a joint in a stable state are key characteristics. These characteristics can effectively reduce the amount of data without losing key information about the joint positions, thereby improving the efficiency of subsequent processing.

[0167] Obtaining joint position features is an important step in constructing a joint feature set. Joint position features are the basis for subsequent analysis of joint motion and error correction. Accurate position features provide the necessary reference for extracting other features, such as posture and velocity.

[0168] In one embodiment: a cluster analysis method can be used to extract key features. First, all coordinate points in the spatial position information are regarded as a data set. Then, a suitable clustering algorithm (such as the K-Means clustering algorithm) is selected to divide the data set into several clusters based on the distance relationship between the coordinate points. After the clustering is completed, the central coordinate point of each cluster or the boundary coordinate point of the cluster can be used as a key feature point. For example, for the spatial position information of a joint, if it is divided into 3 clusters using the K-Means clustering algorithm, then the central coordinate points of these 3 clusters can be used as the key feature points of the joint position. These points can represent the position characteristics of the joint in different states.

[0169] Step S202: performing frequency domain transformation on the motion trajectory information to obtain the posture features of the joint.

[0170] Frequency domain transformation is a mathematical transformation method that converts a signal (in this case, motion trajectory information) from the time domain to the frequency domain. In the frequency domain, the frequency characteristics of the signal can be better analyzed. This transformation can reveal information such as periodicity and frequency components hidden in the motion trajectory, which is closely related to the posture characteristics of the joint.

[0171] Motion trajectory information is the path left by a joint's movement in space over time. In the time domain, it is a series of chronologically ordered position points. However, extracting posture features directly from this time-domain motion trajectory information can be difficult. Instead, the motion trajectory information can be converted to a frequency-domain representation through frequency-domain transformations, such as Fourier transforms.

[0172] In the frequency domain, different frequency components correspond to different motion patterns. For example, lower-frequency components may correspond to slow oscillation of a joint or overall translational motion, while higher-frequency components may be associated with subtle vibrations or rapid adjustments to the joint's posture. By analyzing these frequency components and their amplitudes in the frequency domain, we can derive the joint's posture characteristics. For example, a larger amplitude at a certain frequency may indicate a specific posture tendency in the motion pattern corresponding to that frequency.

[0173] Determining joint posture features from frequency domain transformation is an effective feature extraction method. It provides another dimension of information for comprehensively describing joint characteristics. Combined with position features, it can more accurately analyze the joint's motion state, providing more basis for subsequent velocity and acceleration feature extraction and error correction.

[0174] In one embodiment: the motion trajectory information is transformed into the frequency domain using fast Fourier transform (FFT). First, the motion trajectory information is represented as a discrete time series, where each time point corresponds to the position coordinates of the joint in space. Then, FFT calculation is performed on this time series to obtain a frequency domain representation. In the frequency domain, the frequency components and their corresponding amplitudes are analyzed. In order to extract posture features, some frequency thresholds can be set. For example, the frequency components below a certain low-frequency threshold and above a certain high-frequency threshold are regarded as noise, and the frequency components in the intermediate frequency range and their amplitude relationships are analyzed in detail. Based on these analysis results, a posture feature vector of the joint is constructed, where the elements of the vector may include the frequency values ​​of the main frequency components, the amplitude values, and the proportional relationship between the frequencies.

[0175] Step S203: Extracting the velocity and acceleration features of the joints based on the key position features and posture features of the joints.

[0176] The position and posture features of the joints contain information about the position and posture of the joints in space. This information provides the basis for extracting velocity and acceleration features. Velocity is the rate of change of joint position over time, and acceleration is the rate of change of velocity over time.

[0177] The coordinates of key position features and their changes can be used to calculate velocity features. For example, if the key position coordinates of a joint are known at different times, the velocity can be calculated by calculating the change in position between adjacent time points and dividing it by the time interval. Acceleration features can be obtained by further calculating the change in velocity and dividing it by the time interval. Posture features also influence the extraction of velocity and acceleration features. For example, the posture of a joint (such as rotation angle and tilt direction) can affect its velocity and the direction and magnitude of its acceleration.

[0178] Extracting velocity and acceleration features further quantifies the joint's motion state. Together with position and posture features, these features form a complete joint feature set, more comprehensively reflecting the joint's motion characteristics and providing richer data for subsequent error correction and kinematic model refinement.

[0179] In one embodiment, to extract velocity features, the key position coordinates of the joint at different times can be determined based on the key position features. Furthermore, to consider the impact of posture features on velocity and acceleration, a transformation matrix can be established based on the posture features. The calculated velocity and acceleration vectors can be multiplied by the transformation matrix to obtain velocity and acceleration features that account for the influence of posture.

[0180] Step S204: combining the position features, posture features, velocity features and acceleration features into a joint feature set.

[0181] The joint feature set is a comprehensive collection of features that can fully describe the characteristics of a joint. Position features describe the static position of a joint in space, posture features reflect the posture state of the joint, and velocity and acceleration features reflect the dynamic motion characteristics of the joint.

[0182] By combining these features, a complete description of the joint can be formed. This joint feature set can be directly used as input in subsequent processing, such as in skew compensation models or during kinematic model correction. This joint feature set provides a comprehensive data foundation for analyzing joint relationships and performing error correction.

[0183] The ultimate goal of the entire feature extraction process is to combine these features into a joint feature set. This feature set can more effectively represent the state of the joints. Compared to using any one feature alone, it can provide more accurate and comprehensive information, helping to improve the accuracy and effectiveness of the kinematic error correction method for explosion-proof robots.

[0184] In one embodiment, reference Figure 8 , step S50 may specifically include:

[0185] Step S501: Calculating the target position, target velocity, and target acceleration of the explosion-proof robot joints according to the modified kinematic model.

[0186] The target position is the desired spatial position of the explosion-proof robot joint, calculated based on the modified kinematic model. This position is the ideal coordinates for the joint's ultimate motion, taking into account the robot's mission requirements, current state, and error correction.

[0187] The target speed is the desired speed of the robot's joints as they move toward their target positions. This speed is calculated based on a modified kinematic model, taking into account the robot's dynamic characteristics, the required motion efficiency, and other factors. It is used to control the speed of joint movement.

[0188] The target acceleration is the acceleration that a joint is expected to generate during motion to achieve its target velocity and position. It reflects the speed of change in joint velocity and is determined based on the modified kinematic model, taking into account factors such as the robot's structural limitations and motion smoothness.

[0189] The modified kinematic model in this application's embodiment already accounts for joint error correction, enabling a more accurate description of the relationships between joints. For target position calculation, the model calculates the target position coordinates for each joint by analyzing the current joint positions, the robot's structural parameters, and the kinematic relationships, based on the robot's mission objectives, such as moving the end effector to a specific work area or precisely docking with a target object.

[0190] After determining the target position, the target velocity must be calculated to ensure the joint reaches the target position within an appropriate timeframe while ensuring smooth and efficient motion. For example, if the target position is far from the current position, a higher target velocity can be set without violating the robot's motion performance limits. If the distance is closer or precise control is required, a lower target velocity may be appropriate. The target acceleration further refines joint motion control. It is determined based on the target velocity's changing requirements and the robot's dynamic characteristics. For example, to avoid shock during joint motion, the acceleration must be within a reasonable range, neither too high nor too low.

[0191] Calculating the target position, target velocity, and target acceleration is crucial for precisely controlling the motion of the explosion-proof robot's joints. These values ​​form the basis for generating subsequent control commands. Accurate calculations ensure the robot operates according to the intended trajectory and motion state, improving the accuracy and reliability of the robot's mission execution.

[0192] In one embodiment, matrix operations are used to calculate the target position, target velocity, and target acceleration. First, the modified kinematic model is expressed in the form of a matrix equation. Assuming that the joint variable matrix is ​​J, the position, velocity, and acceleration of the end effector are respectively related to the joint variables by a matrix relationship P=M1J, V=M2J, and A=M3J, where M 1、 M 2、 M3 is the coefficient matrix determined by the kinematic model. According to the task requirements, the target position Pt, target velocity Vt and target acceleration At of the end effector are determined, and then the equation Jt=M1 is solved. -1 Pt、Jt=M2 -1 Vt, Jt=M3 -1 At (here we assume that the matrix M 1、 M 2、 M3 is reversible), and the target position Jt, target velocity Jt, and target acceleration Jt of the joint are obtained (the target position, velocity, and acceleration of the joint are represented accordingly in the joint variable matrix).

[0193] Step S502: converting the target position, target velocity and target acceleration into control instructions, and sending the control instructions to the explosion-proof robot to drive the joints to perform corresponding motion operations.

[0194] A control instruction is a command signal that can be recognized and executed by the drive system of an explosion-proof robot. It contains information such as the target position, target velocity, and target acceleration of the joint, and is used to control the movement of the robot joint so that the robot operates in the expected manner.

[0195] For example, for some robot drive systems, control instructions may be a series of bytes arranged in a specific order, where each byte represents a different meaning. For example, some bytes represent the joint number, some bytes represent the target position, and some bytes represent the target velocity and acceleration. These encoding rules and data format requirements must be taken into account when converting numerical values ​​into control instructions.

[0196] The generated control commands are then sent to the explosion-proof robot's drive joints. This transmission process can be accomplished via wired communication (such as industrial Ethernet, CAN bus, etc.) or wireless communication (such as Wi-Fi, Bluetooth, etc.). Upon receiving the control commands, the drive joints drive actuators such as motors according to the instructions, causing the joints to move according to the target position, target velocity, and target acceleration, thereby achieving overall motion control of the explosion-proof robot.

[0197] Accordingly, in order to better implement the above method, the embodiment of the present application further provides a kinematic error correction system 80 for an explosion-proof robot, wherein the explosion-proof robot includes a plurality of movable joints. Figure 9 As shown, the kinematic error correction system 80 of the explosion-proof robot includes:

[0198] The acquisition module 801 is used to acquire the spatial position information and motion trajectory information of each joint of the explosion-proof robot based on a laser calibration instrument with a preset accuracy;

[0199] A feature extraction module 802 is configured to extract features from the spatial position information and motion trajectory information to obtain a joint feature set, wherein the joint feature set includes position features, posture features, and motion features of the joint, and the motion features include velocity and acceleration of the joint;

[0200] An error compensation module 803 is configured to input the joint feature set into a skew compensation model, model and compensate for non-orthogonal joint errors using the skew compensation model, and generate a joint error compensation matrix, wherein the joint error compensation matrix is ​​used to quantify the non-orthogonality deviations between the joints;

[0201] a correction module 804, configured to correct the kinematic model of the explosion-proof robot according to the joint error compensation matrix to obtain a corrected kinematic model, wherein the kinematic model describes the correlation between the position, velocity, and acceleration of each joint of the explosion-proof robot;

[0202] The instruction generation module 805 is used to generate corresponding control instructions based on the modified kinematic model, and send the control instructions to the explosion-proof robot to perform motion operations.

[0203] The implementation of each module can be specifically referred to the above method embodiment, which will not be described in detail here. The technical effects achieved by each module and device are described in the above method embodiment.

[0204] like Figure 10As shown, an embodiment of the present application further provides a computer device 90, which includes a processor 901 and a memory 902, wherein the memory 902 stores a computer program, and when the computer program is executed by the processor 901, the processor 901 performs the steps of any of the methods described above.

[0205] The above disclosure is only a preferred embodiment of the present application, and certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A kinematic error correction method for an explosion-proof robot, characterized in that: The explosion-proof robot includes a plurality of movable joints, and the method includes: A laser calibration instrument with preset accuracy collects the spatial position information and motion trajectory information of each joint of the explosion-proof robot; Performing key feature extraction on the spatial position information to extract position features of the joints; performing frequency domain transformation on the motion trajectory information to obtain posture features of the joints; extracting velocity and acceleration features of the joints based on the key position features and posture features of the joints; combining the position features, posture features, velocity features, and acceleration features into a joint feature set, wherein the joint feature set includes the position features, posture features, and motion features of the joints, and the motion features include the velocity and acceleration of the joints; Analyzing the sources of non-orthogonality errors between joints based on the joint feature set to construct an initial parameter matrix of the skew compensation model; fitting the initial parameter matrix using the least squares method to obtain preliminary parameters of the skew compensation model; regularizing the preliminary parameters to obtain optimal parameters of the skew compensation model; generating a joint error compensation matrix based on the optimal parameters; the joint error compensation matrix is ​​used to quantify the non-orthogonality deviations between the joints; Introducing the joint error compensation matrix into the original kinematic model of the explosion-proof robot to form a revised kinematic model; numerically solving the revised kinematic model to obtain the actual spatial position, velocity, and acceleration values ​​of the joint; Generating a final corrected kinematic model according to the actual spatial position, velocity, and acceleration values ​​of the joint, specifically comprising: calculating the errors between the spatial position, velocity, and acceleration values ​​of the joint and their corresponding theoretical values; if all errors are less than a preset threshold, determining the corrected kinematic model as the final model; if at least one error is greater than or equal to the preset threshold, iteratively optimizing the joint error compensation matrix until the error meets the requirements; wherein the kinematic model describes the correlation between the position, velocity, and acceleration of each joint of the explosion-proof robot; A corresponding control instruction is generated based on the modified kinematic model, and the control instruction is sent to the explosion-proof robot to perform motion operations.

2. The kinematic error correction method of the explosion-proof robot according to claim 1, characterized in that: The laser calibrator with preset accuracy collects spatial position information and motion trajectory information of each joint of the explosion-proof robot, including: using the laser calibrator with preset accuracy to collect three-dimensional spatial coordinate data of each joint of the explosion-proof robot in static and dynamic states; detecting trajectory changes of the joints during continuous movement, and performing time series analysis on the detected trajectory change information to extract motion trajectory information, wherein the motion trajectory information includes the displacement curve of the joint and its time derivative.

3. The kinematic error correction method of the explosion-proof robot according to claim 2, characterized in that: Use a laser calibration instrument with preset accuracy to collect the three-dimensional spatial coordinate data of each joint of the explosion-proof robot in static and dynamic states, including: Use a laser calibration instrument with preset accuracy to collect the mechanical structure surface of each joint of the explosion-proof robot in static and dynamic states to obtain the spatial coordinate point cloud data of the joint; Performing filtering on the spatial coordinate point cloud data and extracting the geometric center coordinates of key nodes; The three-dimensional space coordinate data of the joint is calculated according to the geometric center coordinates of the key node.

4. The kinematic error correction method for an explosion-proof robot according to claim 3, wherein filtering processing is performed on the spatial coordinate point cloud data, comprising: Calculate the average distance between each point and its adjacent points in the spatial coordinate point cloud data; The spatial coordinate point cloud data is filtered according to the average distance.

5. The kinematic error correction method for an explosion-proof robot according to any one of claims 1 to 4, characterized in that: The generating corresponding control instructions based on the modified kinematic model and sending the control instructions to the explosion-proof robot to perform motion operations includes: According to the modified kinematic model, the target position, target velocity and target acceleration of the explosion-proof robot joint are calculated; the target position, target velocity and target acceleration are converted into control instructions, and the control instructions are sent to the explosion-proof robot drive joint to perform corresponding motion operations.

6. A kinematic error correction system for an explosion-proof robot, characterized in that: The explosion-proof robot includes a plurality of movable joints, and the system includes: The acquisition module is used to collect the spatial position information and motion trajectory information of each joint of the explosion-proof robot based on a laser calibration instrument with preset accuracy; A feature extraction module is configured to extract key features from the spatial position information to extract position features of the joints; perform frequency domain transformation on the motion trajectory information to obtain posture features of the joints; extract velocity and acceleration features of the joints based on the key position features and posture features of the joints; and combine the position features, posture features, velocity features, and acceleration features into a joint feature set, wherein the joint feature set includes the position features, posture features, and motion features of the joints, and the motion features include the velocity and acceleration of the joints. An error compensation module is configured to analyze sources of non-orthogonality errors between joints based on the joint feature set to construct an initial parameter matrix of a skew compensation model; fit the initial parameter matrix using a least squares method to obtain preliminary parameters of the skew compensation model; regularize the preliminary parameters to obtain optimal parameters of the skew compensation model; and generate a joint error compensation matrix based on the optimal parameters; the joint error compensation matrix is ​​used to quantify non-orthogonality deviations between joints; A correction module is used to introduce the joint error compensation matrix into the original kinematic model of the explosion-proof robot to form a corrected kinematic model; numerically solve the corrected kinematic model to obtain the actual spatial position, velocity and acceleration values ​​of the joint; generate a final corrected kinematic model based on the actual spatial position, velocity and acceleration values ​​of the joint, specifically including: calculating the error between the spatial position, velocity and acceleration values ​​of the joint and their corresponding theoretical values; if all errors are less than a preset threshold, determining that the corrected kinematic model is the final model; if at least one error is greater than or equal to the preset threshold, iteratively optimizing the joint error compensation matrix until the error meets the requirement; wherein, the kinematic model describes the correlation between the position, velocity and acceleration of each joint of the explosion-proof robot; The instruction generation module is used to generate corresponding control instructions based on the modified kinematic model, and send the control instructions to the explosion-proof robot to perform motion operations.

Images