Radioactive substance intelligent sorting method and system

Abstract

The invention provides an intelligent sorting method and system for radioactive substances. The method comprises the following steps: synchronously detecting surface element composition and radiation intensity of a radioactive substance through an X-ray fluorescence spectrometer, and constructing a spatial difference matrix of element abundance distribution and a radiation gradient topological graph; the three-dimensional coordinates, the pressure threshold value and the shape adaptation parameters of the target grabbing position are determined on the basis, the curvature radius of the clamping jaw deformation contour is dynamically adjusted, and the error of the dynamic geometric envelope and the contour template is converged to be within the threshold value; finally, when the clamping jaw reaches a target position, the area of the heat shrinkage material is activated to be heated according to the radiation intensity abrupt change boundary coordinates, a sealed isolation layer synchronously expanded with the dynamic geometric envelope is generated, and safe grabbing and sealed isolation of radioactive substances are achieved. According to the technical scheme, accurate grabbing of radioactive substances, error self-adaptive compensation and synchronous isolation of radiation leakage can be achieved, and the sorting safety and efficiency are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent sorting technology, and in particular to a method and system for intelligent sorting of radioactive materials. Background Art

[0002] Sorting radioactive materials requires precise grasping and safe isolation in high-radiation environments. This requires avoiding direct human contact and adapting to the uneven element distribution, complex geometry, and dynamic changes in radiation intensity on the radioactive material surface. Accumulation of grasping positioning errors and lack of synchronization between the sealing barrier and the dynamic deformation envelope in high-radiation environments must be addressed to ensure the safety, stability, and efficiency of the sorting process. Currently, a mechanical gripper control method based on radiation intensity threshold determination is used. This method drives gripper movement and adjusts gripper closing parameters by linearly linking a preset radiation intensity gradient threshold with gripper pressure feedback. This method relies on spatial registration of a fixed pressure threshold with the radiation intensity distribution, combined with a rigid gripper's deformation compensation mechanism to achieve grasping. However, this approach lacks dynamic adaptability to local element abundance variations on the radioactive material surface. This leads to insufficient coupling between the pressure threshold and element distribution during grasping, making positioning easily misaligned due to sudden changes in radiation intensity. Furthermore, the rigid gripper's deformation compensation mechanism struggles to keep pace with dynamic geometric changes in real time, resulting in synchronization errors between the sealing barrier and the gripper's deformation envelope, increasing the risk of radiation leakage. Summary of the Invention

[0003] The embodiments of the present application provide a method and system for intelligently sorting radioactive materials, which are used to solve the problem of insufficient coupling in the prior art and the problem of positioning deviation caused by sudden changes in radiation intensity.

[0004] In a first aspect, an embodiment of the present application provides a method for intelligent sorting of radioactive materials, comprising: synchronously detecting the surface elemental composition and radiation intensity of radioactive materials by an X-ray fluorescence spectrometer to generate a spatial difference matrix containing element abundance distribution and a gradient distribution topology map of real-time radiation values; determining a three-dimensional coordinate set, a pressure threshold set, and a shape adaptation parameter vector of a target grasping position corresponding to the radioactive material based on the spatial difference matrix and the gradient distribution topology map; converting the shape adaptation parameter vector into a displacement control instruction set of a mechanical clamp array, wherein the displacement control instruction set includes a set of three-dimensional coordinates and a set of gradient distribution topology maps; The telescopic rod displacement parameters generated by the image space registration drive the telescopic rod of the corresponding clamping claw unit to move axially according to the telescopic rod displacement parameters; during the axial movement of the telescopic rod, the contact pressure data stream of the air pressure sensor is compared with the pressure threshold set in real time, and the curvature radius of the clamping claw deformation profile is adjusted according to the comparison difference, so that the error range between the dynamic geometric envelope constituted by the clamping claw deformation profile and the geometric profile template converges to within the threshold; when the clamping claw unit reaches the target grasping position, the regional heating of the heat shrinkable material is activated according to the radiation intensity mutation boundary coordinates of the gradient distribution topology map, so as to generate a sealing isolation layer that expands synchronously with the dynamic geometric envelope. Optionally, determining the three-dimensional coordinate set, pressure threshold set and shape adaptation parameter vector of the target grasping position corresponding to the radioactive material based on the spatial difference matrix and the gradient distribution topology map includes: performing convolution matching on the spatial difference matrix and the radioactive material feature library to output the three-dimensional coordinate set of the target grasping position; calculating the pressure threshold set based on the distribution of extreme points of radiation intensity of the gradient distribution topology map; and generating a shape adaptation parameter vector based on a geometric contour template pre-stored in the radioactive material feature library.

[0005] Optionally, the convolution matching of the spatial difference matrix and the radioactive material feature library to output a three-dimensional coordinate set of the target grasping position includes: segmenting the spatial difference matrix into regions according to the reference value of the element abundance distribution in the radioactive material feature library, and extracting the feature vector of each segmented region; comparing the feature vector of each segmented region with the standard feature vector of the material category corresponding to the radioactive material in the radioactive material feature library layer by layer to generate a feature similarity distribution map; identifying the peak area that meets the preset matching conditions in the feature similarity distribution map, and extracting the geometric center coordinates of the peak area as candidate coordinates; and performing spatial position compensation on the candidate coordinates according to the radiation intensity attenuation gradient of the gradient distribution topology map at the candidate coordinates to generate a three-dimensional coordinate set of the target grasping position. Optionally, the pressure threshold set is calculated based on the distribution of extreme points of radiation intensity in the gradient distribution topology map, including: marking extreme points whose radiation intensity exceeds the neighborhood average in the gradient distribution topology map, and constructing a spatial distribution set of extreme points; establishing a radiation intensity attenuation model with each extreme point as the center, and calculating the rate of change of radiation intensity within a preset radius around the extreme point; assigning a dynamic pressure weight to each extreme point based on the mapping relationship between the rate of change and the preset safe contact pressure; and performing weighted fusion of the radiation intensity in the area where the extreme point is located based on the dynamic pressure weight to generate a pressure threshold set. Optionally, generating a shape adaptation parameter vector based on a geometric contour template pre-stored in the radioactive material feature library includes: extracting a set of surface curvature feature points of the radioactive material from the geometric contour template pre-stored in the radioactive material feature library, and constructing a contour piecewise function; calculating a curvature adaptation parameter adapted to the contact surface of the gripper unit in the mechanical fixture array based on the spatial relative position relationship between the target grasping position in the three-dimensional coordinate set and the contour piecewise function; and normalizing the curvature adaptation parameter with the radiation intensity gradient value at the corresponding position in the gradient distribution topology map to generate a shape adaptation parameter vector.

[0006] Optionally, the conversion of the shape adaptation parameter vector into a displacement control instruction set of a mechanical fixture array includes: performing degree of freedom decoupling on the shape adaptation parameter vector to separate geometric constraint parameters and radiation gradient constraint parameters corresponding to the axial displacement of the gripper unit in the mechanical fixture array; orthogonally projecting the geometric constraint parameters and the spatial posture information in the three-dimensional coordinate set to generate an initial axial displacement component; extracting the radiation intensity gradient direction vector corresponding to the three-dimensional coordinate set from the gradient distribution topology map to establish a gradient direction and gripper axial displacement correction correlation factor; dynamically compensating the initial axial displacement component according to the displacement correction correlation factor to generate an adaptive displacement parameter set including a displacement direction deviation correction amount; and converting the adaptive position The displacement parameter set and the radiation gradient constraint parameters are coupled and decomposed according to the kinematic model of the gripper unit to generate the axial displacement parameters of each telescopic rod in the displacement control instruction set; a radiation intensity attenuation field model centered on the target grasping position is established in the three-dimensional coordinate set to calculate the radiation intensity attenuation rate of each axial path; a dynamic scaling coefficient of the axial displacement is generated according to the difference between the attenuation rate and the preset safety threshold; the axial displacement parameter and the dynamic scaling coefficient are multiplied point by point to obtain a telescopic rod displacement parameter set adapted to the radiation environment; based on the radiation intensity gradient correlation of adjacent coordinate points in the gradient distribution topology map, a displacement coordination constraint equation group between the gripper units is constructed, and the telescopic rod displacement parameters finally executed are obtained by solving them.

[0007] Optionally, the shape adaptation parameter vector is converted into a displacement control instruction set of a mechanical clamp array, wherein the displacement control instruction set includes telescopic rod displacement parameters generated by spatial registration of the three-dimensional coordinate set and the gradient distribution topology map, driving the telescopic rod of the corresponding clamp unit to move axially according to the telescopic rod displacement parameters, and also includes: dynamically reconstructing the movement path control parameters of the mechanical clamp array through the association mapping of the spatial difference matrix and the thickness growth rate of the sealing isolation layer; the dynamic reconstruction of the movement path control parameters of the mechanical clamp array through the association mapping of the spatial difference matrix and the thickness growth rate of the sealing isolation layer includes: extracting the spatial weight value of the element abundance distribution in the spatial difference matrix, establishing the element abundance distribution of each coordinate point and the sealing isolation layer. A dynamic response relationship model of the thickness growth rate of the separation layer; obtaining the thickness growth rate of the sealing isolation layer in each area of ​​the dynamic geometric envelope surface in real time, and calculating the path adjustment priority coefficient of the corresponding area according to the dynamic response relationship model; generating a movement path correction vector of each clamping unit of the mechanical clamp array based on the gradient change trend of the thickness growth rate in three-dimensional space; performing a spatial convolution operation on the path correction vector and the element abundance weight value in the spatial difference matrix, and outputting a real-time offset parameter set of the movement path; establishing a radiation intensity diffusion model centered on the current clamping position in the gradient distribution topology map, and dynamically reconstructing the movement path control parameters of the mechanical clamp array according to the superposition result of the real-time offset parameter set and the diffusion model.

[0008] Optionally, during the axial movement of the telescopic rod, the contact pressure data stream of the air pressure sensor is compared with the pressure threshold set in real time, and the curvature radius of the gripper deformation profile is adjusted according to the comparison difference, so that the error range between the dynamic geometric envelope formed by the gripper deformation profile and the geometric profile template converges to within the threshold, including: establishing a real-time comparison channel between the contact pressure data stream and the pressure threshold set, generating an error compensation function containing dynamic geometric envelope deformation parameters by constructing a nonlinear mapping relationship between the curvature radius of the gripper deformation profile and the contact pressure difference; establishing a deformation control model of the gripper unit based on the error compensation function, and iteratively calculating the deformation of each node of the dynamic geometric envelope surface. The curvature radius adjustment amount of the point is calculated to generate a deformation control parameter matrix containing the three-dimensional spatial curvature gradient; the deformation control parameter matrix is ​​converted into a real-time control signal stream of the gripper drive unit, and the real-time deformation data of the geometric envelope is collected through the distributed strain sensor array embedded in the gripper deformation layer, and an error field distribution model of the deformation data and the geometric contour template is constructed; a multi-objective optimization function based on the radiation intensity gradient weight is established in the error field distribution model, and by dynamically adjusting the curvature radius compensation coefficient of each node of the gripper unit, the local curvature deviation value of the dynamic geometric envelope is exponentially decayed along the radiation intensity gradient direction until the spatial integral value of the global error field reaches a preset convergence threshold.

[0009] Optionally, when the gripper unit reaches the target grasping position, the regional heating of the heat shrinkable material is activated according to the radiation intensity mutation boundary coordinates of the gradient distribution topology map, and a sealing isolation layer that expands synchronously with the dynamic geometric envelope is generated. The movement path control parameters of the mechanical clamp array are dynamically reconstructed through the correlation mapping of the spatial difference matrix and the thickness growth rate of the sealing isolation layer, including: determining the heating area of ​​the heat shrinkable material based on the radiation intensity mutation boundary coordinates, and generating a heating control parameter matrix; calculating the thickness growth rate distribution function of the sealing isolation layer according to the heating control parameter matrix, and generating a set of thickness compensation coefficients synchronized with the change of the dynamic geometric envelope curvature; monitoring the thickness distribution of the sealing isolation layer in real time through an infrared thermal imaging array, and establishing a dynamic coupling equation of the thickness growth rate distribution function and the movement path control parameters of the mechanical clamp array; calculating the compensation vector of the movement path control parameter based on the dynamic coupling equation, and performing local correction based on the three-dimensional coordinates of the element abundance abnormal area in the spatial difference matrix to generate a displacement update parameter; fusing the displacement update parameter with the original movement path control parameter to drive the gripper unit to perform the synchronous expansion operation of the sealing isolation layer along the corrected path.

[0010] In a second aspect, an embodiment of the present application provides an intelligent sorting system for radioactive materials, comprising: a generation module, which synchronously detects the surface element composition and radiation intensity of radioactive materials through an X-ray fluorescence spectrometer to generate a spatial difference matrix containing element abundance distribution and a gradient distribution topology map of real-time radiation values; a determination module, which determines the three-dimensional coordinate set, pressure threshold set and shape adaptation parameter vector of the target grasping position corresponding to the radioactive material according to the spatial difference matrix and the gradient distribution topology map; a driving module, which converts the shape adaptation parameter vector into a displacement control instruction set of a mechanical clamp array, wherein the displacement control instruction set includes a set of three-dimensional coordinates and a set of gradient distribution parameters; The telescopic rod displacement parameters generated by the spatial registration of the cloth topology map drive the telescopic rod of the corresponding clamping claw unit to move axially according to the telescopic rod displacement parameters; the adjustment module compares the contact pressure data stream of the air pressure sensor with the pressure threshold set in real time during the axial movement of the telescopic rod, and adjusts the curvature radius of the clamping claw deformation profile according to the comparison difference, so that the error range between the dynamic geometric envelope constituted by the clamping claw deformation profile and the geometric profile template converges to within the threshold; the activation module activates the regional heating of the heat shrinkable material according to the radiation intensity mutation boundary coordinates of the gradient distribution topology map when the clamping claw unit reaches the target grasping position, so as to generate a sealing isolation layer that expands synchronously with the dynamic geometric envelope.

[0011] In an embodiment of the present application, an X-ray fluorescence spectrometer is used to synchronously detect the surface elemental composition and radiation intensity of a radioactive material to generate a spatial difference matrix containing element abundance distribution and a gradient distribution topology map of real-time radiation values; based on the spatial difference matrix and the gradient distribution topology map, a three-dimensional coordinate set, a pressure threshold set, and a shape adaptation parameter vector of a target grasping position corresponding to the radioactive material are determined; the shape adaptation parameter vector is converted into a displacement control instruction set of a mechanical clamp array, wherein the displacement control instruction set includes a telescopic rod generated by spatial registration of the three-dimensional coordinate set and the gradient distribution topology map. The displacement parameter is used to drive the telescopic rod of the corresponding gripper unit to move axially according to the displacement parameter of the telescopic rod; during the axial movement of the telescopic rod, the contact pressure data stream of the air pressure sensor is compared with the pressure threshold set in real time, and the curvature radius of the gripper deformation profile is adjusted according to the comparison difference, so that the error range between the dynamic geometric envelope constituted by the gripper deformation profile and the geometric profile template converges to within the threshold; when the gripper unit reaches the target grasping position, the regional heating of the heat shrinkable material is activated according to the radiation intensity mutation boundary coordinates of the gradient distribution topology map, so as to generate a sealing isolation layer that expands synchronously with the dynamic geometric envelope.

[0012] The technical solution of the present application has the following beneficial effects: The present application uses the intelligent sorting method for radioactive materials to simultaneously detect the surface element composition and radiation intensity through an X-ray fluorescence spectrometer to achieve multimodal data fusion and provide high-precision spatial distribution characteristics for precise positioning. Based on the coupled analysis of element abundance and radiation gradient, the three-dimensional coordinates, pressure threshold and shape adaptation parameters of the target grasping position are determined to improve the robustness of grasping in complex shape and dynamic radiation intensity scenarios. By converting the shape parameters into displacement control instructions of the mechanical clamp, the clamp is driven to move axially according to the radiation gradient alignment path, reducing the risk of cumulative positioning deviation. During the movement process, the contact pressure data stream and the threshold set are compared in real time to converge the errors of the dynamic geometric envelope and the contour template, thereby suppressing radiation leakage caused by deformation errors. Finally, based on the boundary coordinates of the radiation intensity mutation, the heating of the heat shrinkable material area is activated to generate a sealing isolation layer that expands synchronously with the deformation of the clamp, realizing real-time dynamic isolation of radiation leakage, and improving the safety, stability and efficiency of the sorting process as a whole. Furthermore, the spatial difference matrix is ​​convolved with a radioactive material feature library to output the three-dimensional coordinates of the target grasping position. A set of pressure thresholds is calculated based on the extreme points of radiation intensity in the gradient distribution topology, and a shape adaptation parameter vector is generated by combining it with a pre-stored geometric contour template. Through feature library matching and radiation intensity extreme value analysis, precise positioning of the grasping position, dynamic adaptation of the pressure threshold, and intelligent generation of shape parameters are achieved, ensuring that the gripper unit achieves both spatial positioning accuracy, robust pressure control, and geometric contour adaptability when grasping complex radioactive materials.

[0013] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0015] Figure 1 A flow chart of a method for intelligent sorting of radioactive materials provided by the present application is shown; Figure 2 A schematic diagram of the structure of an intelligent radioactive material sorting system provided by the present application is shown; Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0016] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0017] Some processes described in the specification and claims of this application, as well as in the accompanying figures, include multiple operations that appear in a specific order. However, it should be understood that these operations may be executed in a different order than the order in which they appear herein, or in parallel. Operation numbers such as 101 and 102 are merely used to distinguish between different operations and do not represent any specific order of execution. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that terms such as "first" and "second" are used herein to distinguish between different messages, devices, modules, etc., and do not imply a sequential order, nor do they limit "first" and "second" to different types. This application aims to construct a spatial difference matrix and gradient distribution topology by integrating multimodal data of elemental abundance and radiation intensity of radioactive materials. This system dynamically adjusts the gripper deformation profile and the expansion of the sealing isolation layer using real-time contact pressure feedback, thereby forming a closed-loop control mechanism. This system achieves precise gripping and positioning of radioactive materials, dynamic self-compensation for geometric envelope errors, and simultaneous isolation of radiation leakage, comprehensively improving the safety, stability, and efficiency of the sorting process. 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. Figure 1 A flow chart of a method for intelligently sorting radioactive materials is provided in the embodiment of the present application. Figure 1 As shown, the method includes: 101. Synchronously detecting the surface element composition and radiation intensity of the radioactive material by an X-ray fluorescence spectrometer to generate a spatial difference matrix including the element abundance distribution and a gradient distribution topology map of the real-time radiation value; in this step, the X-ray fluorescence spectrometer is an instrument for analyzing the composition of materials, and its core principle is to irradiate the sample surface with high-energy X-rays to excite the elements in the sample and cause it to release unique "fluorescent X-rays".

[0018] Radioactive material refers to a substance that can spontaneously release radiation (such as alpha particles, beta particles, or gamma rays). The atomic nuclei of such substances are unstable and release energy through the decay process. Therefore, they need to be intelligently sorted and protected through the solution of this application. Surface elemental composition refers to the types of elements present in the surface layer of radioactive material and their relative proportions. Radiation intensity represents the amount of radiation detected per unit time, reflecting the activity level of the radioactive material. The higher the intensity, the greater the concentration of radioactive material in the area or the faster the decay rate. Element abundance distribution describes the changes in the content of elements at different locations on the surface of radioactive material. "Abundance" refers to the relative content of an element (e.g., iron accounts for 70% and copper accounts for 30% in a certain area), while "distribution" refers to the spatial difference in this content.

[0019] In an embodiment of the present application, the surface of the radioactive material is first scanned synchronously by an X-ray fluorescence spectrometer to excite and capture the surface element composition information, and then the element abundance data of each detection point is quantified by analyzing the fluorescence signal intensity. These data are spatially integrated to form an element abundance distribution, showing the aggregation or absence pattern of elements on the surface. In order to further locate the abnormal area, a spatial difference matrix is ​​generated by comparing the element abundance of each point with the overall average value, and the coordinate positions where the content is significantly higher or lower are clearly marked. At the same time, the radiation detector measures the radiation intensity of each point in real time, and combines the interpolation algorithm to convert the discrete intensity value into a continuous radiation field, and finally draws a gradient distribution topology map, using thermal color blocks and diffusion arrows to show the dynamic change trend of the radiation intensity. The entire process, from chemical composition analysis to radiation behavior mapping, realizes the visual and accurate diagnosis of the surface characteristics and radiation risks of radioactive materials through the progressive analysis of element composition, abundance data, distribution pattern and difference matrix, combined with the dynamic correlation between radiation intensity and gradient topology.

[0020] In a real-world case study, in a radioactive material sorting scenario, an X-ray fluorescence spectrometer was used to synchronously scan the target surface, using high-energy X-rays to excite surface elements and capture characteristic fluorescence signals. The elemental composition and relative abundance were analyzed, generating a spatial difference matrix reflecting the distribution patterns of the elements and locating abnormal areas of element enrichment or deficiency. Simultaneously, the radiation detector collected radiation intensity data at each point in real time, combined with an interpolation algorithm to construct a continuous radiation field model. A gradient distribution topology map was drawn using thermal color blocks and diffusion arrows, visually presenting the dynamic changes in radiation intensity and diffusion trends. By analyzing the correlation between element abundance differences and radiation gradient topology, high-risk areas (such as areas with element enrichment and sudden radiation increases) and shielding effect areas (such as areas with radiation attenuation caused by metal element enrichment) were accurately identified, providing a spatial decision-making basis for automated sorting and protection strategies, and enabling coordinated visual diagnosis of chemical composition and radiation behavior.

[0021] 102. Based on the spatial difference matrix and the gradient distribution topology map, determine the three-dimensional coordinate set, pressure threshold set and shape adaptation parameter vector of the target grasping position corresponding to the radioactive material; in this step, the shape adaptation parameter vector includes a multidimensional parameter set that describes the degree of matching between the gripper and the surface geometric contour of the radioactive material, including the radius of curvature, the clamping angle, etc. The three-dimensional coordinate set is a three-dimensional coordinate set of the target grasping position generated by mapping based on the spatial relationship between the spatial difference matrix and the gradient distribution topology map, and contains the precise spatial points to which the gripper needs to move. The pressure threshold set is a set of pressure thresholds of the gripper in different areas that is dynamically calculated based on the distribution of extreme points of radiation intensity of the gradient distribution topology map, and is used to constrain the range of clamping force. The target grasping position refers to the operating point selected from the surface of the radioactive material that meets the requirements of safety, stability and efficiency.

[0022] In this embodiment, the spatial difference matrix is ​​first convolved with a pre-stored library of radioactive material signatures to locate regions of stable elemental abundance (e.g., the ends of fuel rods). A set of target gripping locations is generated through three-dimensional coordinate interpolation. For automobiles, a set of pressure thresholds (e.g., disabling gripping in central, high-radiation areas) is calculated based on the distribution of extreme points in the radiation gradient field intensity map, combined with a material mechanics model. Finally, cylinder parameters from a geometric contour template library are used to generate a shape adaptation parameter vector (including curvature radius, gripping angle, etc.) through least squares fitting. Ultimately, integrated control instructions for the three-dimensional coordinates, pressure threshold, and shape parameters are output.

[0023] Based on the element abundance spatial difference matrix and the radiation gradient distribution topology, the system uses a feature matching algorithm to locate low-radiation areas with stable surface element distribution (such as the low-abundance area at the end of the fuel rod). Combined with three-dimensional spatial mapping, it generates a three-dimensional coordinate set for the target grasping position. Based on the radiation field intensity extreme value distribution model, it sets a clamping pressure restricted area in high-radiation areas and dynamically calculates a set of pressure thresholds in medium and low radiation gradient areas to balance clamping stability and the material's compressive limit. At the same time, the system uses a geometric contour matching engine to extract the curvature characteristics of the waste surface and uses a surface fitting algorithm to generate a shape adaptation parameter vector (such as the gripper curvature radius and opening and closing angle) to precisely match the geometric contour of the cylindrical waste canister surface. Ultimately, it outputs control instructions integrating three-dimensional coordinates, pressure thresholds, and shape parameters, driving the robotic arm to clamp the target waste canister at the appropriate angle and pressure, avoiding high-radiation and element-enriched fragile areas, and achieving safe, stable, and secondary contamination-free automated sorting operations. 103. Convert the shape adaptation parameter vector into a displacement control instruction set of a mechanical clamp array; in this step, the displacement control instruction set includes telescopic rod displacement parameters generated by spatially aligning the three-dimensional coordinate set with the gradient distribution topology map, and drives the telescopic rod of the corresponding gripper unit to move axially according to the telescopic rod displacement parameters. The telescopic rod displacement parameters are the axial movement distance and direction parameters of the telescopic rod generated by spatially aligning the three-dimensional coordinate set with the gradient distribution topology map (such as coordinate system conversion).

[0024] A mechanical gripper array is a deformable gripping device composed of multiple independently controlled gripper units. Each gripper unit includes a telescopic rod, a drive mechanism, and a pressure sensor. In this embodiment, a shape adaptation parameter vector is first converted into a stepper motor control signal. This signal is then combined with the spatial registration results of the three-dimensional coordinate set and the radiation gradient field intensity map (coordinate system alignment using an ICP algorithm) to generate telescopic rod displacement parameters. Next, a PID controller is used to regulate the axial movement speed of the telescopic rod. In high-radiation areas (e.g., X = 40-60 cm), a displacement path attenuation coefficient is introduced to reduce movement speed. Finally, a phase synchronizer coordinates the timing of multiple gripper movements, ultimately driving the gripper units to move along a safe path to the target position. The system inputs the shape adaptation parameter vector into a multi-axis linkage control module. Based on the gripper unit kinematic model and geometric matching constraints, the curvature radius and gripping angle parameters are converted into linear displacement and rotational angle instruction sets for the servo motors. A coordinate transformation algorithm is used to drive each independent gripper in the gripper array to synchronously adjust its spatial position along a preset trajectory, ensuring that its opening and closing configuration closely matches the surface contour of the target object, ensuring uniform pressure distribution on the gripping surface and eliminating localized stress concentrations. 104. During the axial movement of the telescopic rod, the contact pressure data stream from the air pressure sensor is compared in real time with the pressure threshold set. The curvature radius of the gripper deformation profile is adjusted based on the comparison difference, so that the error range between the dynamic geometric envelope formed by the gripper deformation profile and the geometric profile template converges within the threshold. In this step, the contact pressure data stream from the air pressure sensor refers to the time series data of the contact pressure between the gripper and the object surface continuously collected and transmitted by the air pressure sensor, which is used to reflect real-time pressure changes. The pressure threshold set refers to a set of pre-set safe operating pressure ranges, including key parameters such as minimum clamping force (to prevent slippage) and maximum compressive strength (to prevent damage to the object surface). The geometric profile template refers to a pre-set ideal gripper shape model, such as a geometric form that fully matches the target object surface (such as a plane or sphere). The dynamic geometric envelope refers to the geometric boundary formed by the real-time changes in the gripper deformation profile, which is used to dynamically conform to the surface shape of the target object (such as a concave, convex, or curved surface). The deformation profile of the gripper refers to the trajectory of shape change caused by force or displacement when the gripper contacts an object, such as local bending or depression caused by uneven pressure.

[0025] In an embodiment of the present application, when a mechanical gripper array is activated to grasp a nuclear waste canister with surface corrosion, the telescopic rods of each gripper unit move axially to contact the canister surface, and a pressure sensor captures a real-time contact pressure data stream. When the gripper contacts the canister, the system compares the real-time pressure data with preset minimum gripping force and maximum pressure resistance thresholds. It detects abnormally elevated pressure in one area due to surface corrosion, while insufficient pressure in another area due to loose material. Based on this pressure differential, the control unit dynamically adjusts the curvature radius of the corresponding gripper's deformation profile—reducing the curvature in high-pressure areas to sharpen the gripper tip and disperse pressure, while increasing the curvature in low-pressure areas to smooth the contact surface and enhance grip. As the gripper shape adapts in real time, its dynamic geometric envelope gradually conforms to the canister's surface irregularities. The shape error relative to the preset ideal geometric template significantly decreases from its initial significant deviation to stabilize within an acceptable range. Ultimately, the mechanical gripper securely grasps the canister without overpressurizing it and damaging it. Radiation monitoring confirmed no leaks, validating the effectiveness of the adaptive deformation control and error convergence mechanisms. During the axial movement of the robotic arm, a pressure sensor collects real-time data on the pressure distribution at the contact surface between the gripper and the radioactive material. The control center dynamically compares this data with a set of preset pressure thresholds. When the local pressure deviates from the safety threshold, a deformation compensation mechanism is triggered: the expansion and contraction of the memory alloy actuator is adjusted inversely by the curvature radius deviation, dynamically correcting the envelope of the gripper's flexible surface. This ensures that the matching error between the gripping geometry and the target surface is continuously converged to millimeter-level accuracy, preventing damage to the radioactive material due to stress overload.

[0026] 105. When the gripper unit reaches the target grasping position, regional heating of the heat-shrinkable material is activated based on the coordinates of the radiation intensity mutation boundary in the gradient distribution topology, generating a sealing isolation layer that expands synchronously with the dynamic geometric envelope. In this step, the radiation intensity mutation is based on the coordinates of the radiation intensity mutation boundary in the gradient distribution topology (e.g., intensity change rate > 150 μSv / h / mm), generating a trigger signal and heating power parameters for regional heating of the heat-shrinkable material. The dynamic geometric envelope is a control module that uses finite element simulation to calculate the real-time error between the dynamic geometric envelope and the geometric contour template and iteratively adjusts the gripper deformation parameters until the error converges to a threshold (e.g., <0.1 mm). Heat-shrinkable material refers to a polymer material (such as polyolefin or fluororubber) that shrinks in volume upon heating. Its characteristic is that it shrinks rapidly at a specific temperature and tightly wraps around the target surface. The sealing isolation layer is a dense protective layer formed by the heat-shrinkable material upon thermal contraction. It covers the target object surface or the gripping contact area, physically isolating the target object from radioactive particles or radiation leakage while preventing the intrusion of external contaminants.

[0027] In an embodiment of the present application, after the mechanical gripper unit moves to the target gripping position, the system locates the radiation intensity mutation boundary based on the gradient distribution topology map and activates the heat shrink material heating module in the corresponding area. After being heated, the heat shrink material shrinks rapidly, tightly wraps the target surface, and generates a sealing isolation layer that expands synchronously with the dynamic geometric envelope of the gripper. During this process, the expansion speed of the sealing layer is matched in real time with the adjustment of the gripper deformation profile to ensure that the isolation layer always covers the gripping contact surface, forming a "dynamic seal". After the gripper unit arrives at the target gripping position, the system analyzes the coordinates of the radiation intensity mutation boundary in the gradient distribution topology map and activates the regionalized temperature control unit of the thermal response material embedded in the clamping mechanism. Directed thermal energy is applied to the heat shrink film at the boundary of the radiation diffusion front through a micro-resistance heating grid, triggering the heat shrinkage effect of the material, and generating a continuous and closed isolation layer around the dynamic envelope surface of the gripper, synchronously blocking the escape path of radioactive particles and the contact interface with the external environment, thereby achieving fully enclosed protection for the high-risk material transfer process. In summary, steps 101 to 105 enable the robot to accurately grasp and avoid dangerous areas through dynamic identification of materials and risk areas. Tactile feedback adjusts the gripper's movements in real time to prevent secondary damage. Finally, an adaptive sealing layer isolates risks, enabling safe sorting of hazardous waste. This process is similar to how a "smart hand" grasps fragile items: first observe, then gently handle, and finally wrap for protection, all without human intervention.

[0028] To address the challenges of gripping positioning deviation, gripping overload leakage, and geometric adaptation in radioactive material sorting, a safe gripping area is screened and a three-dimensional coordinate set is generated by fusing an elemental abundance spatial difference matrix with a radiation gradient topology map. This is then combined with an algorithmic fitting algorithm using a pre-stored geometric template library to generate a shape adaptation parameter vector, achieving multi-dimensional adaptive matching of gripper position, pressure control, and target surface characteristics. In some embodiments, the spatial difference matrix and the gradient distribution topology map used in step 102 determine the three-dimensional coordinate set, pressure threshold set, and shape adaptation parameter vector for the target gripping location corresponding to the radioactive material. This includes: 201, performing convolution matching on the spatial difference matrix with a radioactive material feature library to output the three-dimensional coordinate set of the target gripping location. In step 201, the spatial difference matrix refers to a multidimensional matrix data structure used to quantify elemental abundance differences in radioactive materials at different spatial locations. The radioactive material feature library refers to a pre-stored database of radioactive material characteristics, containing standard geometric templates (e.g., cylinders and spheres), elemental abundance distribution models, and radiation characteristic parameters. Convolution matching refers to the use of convolutional neural networks (CNN) or three-dimensional convolution algorithms to perform sliding window calculations on the spatial difference matrix and the templates in the feature library to output matching response values.

[0029] In this embodiment, a convolution matching algorithm is first used to convert geometric templates (e.g., cylinders and spheres) in a radioactive material feature library into three-dimensional convolution kernels. The kernels are then slid across the spatial difference matrix region by region to calculate the degree of match. An initial response value matrix is ​​generated through the convolution operation, characterizing the geometric similarity between different regions and the template. Non-maximum suppression techniques are then used to filter out local high-response regions and eliminate low-match noise points. Finally, an interpolation algorithm is used to optimize coordinate accuracy, converting discrete response points into a set of high-precision three-dimensional coordinates. For example, the ends of a uranium fuel rod are selected as gripping targets due to their stable elemental abundance and high match with the cylindrical template. A set of pressure thresholds is calculated based on the distribution of extreme values ​​of radiation intensity in the gradient distribution topology. In step 202, the gradient distribution topology is a spatial gradient distribution model of radiation intensity, describing the direction and rate of change of radiation intensity at different locations using a vector field. The distribution of extreme values ​​of radiation intensity refers to the coordinates of regions within the gradient distribution topology where the radiation intensity is significantly higher than that of the surrounding area. These typically correspond to locations at high risk of radioactive material leakage or accumulation. The set of pressure thresholds is a set of upper limits on the clamping force dynamically calculated based on the radiation gradient values ​​of different regions. For example, the pressure threshold in the high radiation gradient area is 0N (clamping is prohibited), and the threshold in the low gradient area is relaxed. In the embodiment of the present application, the radiation intensity mutation boundary is first extracted from the gradient distribution topology map through the edge detection algorithm and marked as the extreme point; then the mechanical properties of the material and the radiation gradient value are combined to dynamically calculate the allowable clamping pressure of each area. For example, the radiation gradient value in the high radiation area triggers the pressure attenuation model, which automatically lowers the upper limit of the clamping force to avoid radiation leakage when the clamping claw contacts; finally, a set of pressure thresholds covering the entire area is generated to constrain the clamping claw action in real time. In the high radiation area of ​​the middle section of the uranium fuel rod, the system automatically sets the pressure threshold to 0N, completely prohibiting the clamping action. A shape adaptation parameter vector is generated based on the geometric contour template pre-stored in the radioactive material feature library.

[0030] In step 203, the radioactive material feature library refers to a pre-stored database of radioactive material features, containing standard geometric templates (e.g., cylinders and spheres), elemental abundance distribution models, and radiation characteristic parameters. The geometric profile matching matrix refers to the quantified data on the matching error between the target object's surface and the template, and the transformation rules for mapping deformation parameters from geometric parameters to gripper deformation parameters, generating gripper control parameters that adapt to the target's geometric features. The shape adaptation parameter vector refers to a set of control parameters for dynamically adjusting gripper deformation, including curvature radius, clamping angle, and telescopic rod displacement. In this embodiment, a pre-stored geometric template (e.g., cylinder diameter, sphere radius) is first retrieved from the radioactive material feature library and its matching degree calculated with the target object's surface contour. The template with the smallest error is selected as a reference. Subsequently, the template geometric parameters are converted into gripper deformation control parameters using parameter mapping rules. For example, the cylinder radius is mapped to the gripper curvature radius, and the deformation is compensated for by combining surface irregularities. Finally, a shape adaptation parameter vector containing key parameters such as curvature radius and clamping angle is output. For example, after the cylindrical template of the uranium fuel rod is matched, the driven gripper adaptively fits the surface with a curvature radius of 4 cm and a clamping angle of 15°.

[0031] The following is a specific example: A robot scanned the surface of a cylinder using an X-ray fluorescence spectrometer and detected stable and uniform elemental abundances at the ends, while the concentrations in the middle section fluctuated significantly due to oxide layer flaking. The system constructed the elemental distribution data into a spatial difference matrix and performed a convolution match with a "cylinder template" in the feature library. The convolution operation identified high-quality matches between the two ends and the template, generating a set of three-dimensional coordinates. These two ends were marked as safe gripping locations, while the low-matching region in the middle section was excluded. Combined with gradient distribution topology analysis, the system detected a significant increase in radiation intensity and abrupt boundaries in the middle section, declaring it a high-risk restricted area. Based on a radiation gradient attenuation model, the gripping pressure threshold in the middle section was automatically set to zero, prohibiting gripper contact. The radiation gradient at the ends was flat, and the maximum allowable gripping pressure was calculated to ensure the gripper applied force within a safe range. The system then used the cylindrical geometry template from the feature library to optimize the fit with the target object's surface contour. The system identified slight surface irregularities caused by the oxide layer and dynamically adjusted the gripper's curvature radius and opening and closing angle through deformation parameter mapping, generating shape parameters tailored to the current surface conditions. After the gripper is adjusted according to the parameters, the flexible material adaptively fits the surface of the cylinder, just like a glove fits the palm, avoiding squeezing of oxide layer fragments while ensuring firm grasping.

[0032] In summary, the convolution matching in steps 201-203 achieves high-precision positioning of the grasping position. The radiation gradient is combined to dynamically suppress the gripping force in high-risk areas, and adaptive deformation parameters are generated based on geometric templates. This solves the grasping deviation and leakage risk caused by the insufficient coupling between radiation and element distribution in traditional methods. In the nuclear waste sorting scenario, the grasping success rate is increased to over 98%, and the radiation leakage is controlled within the safety threshold, significantly improving operational safety and efficiency in high-risk environments. To address the problem of grasping positioning deviation and the risk of accidental contact in high-radiation areas caused by the uneven distribution of elements on the surface of radioactive materials, a grasping position positioning method based on convolution matching was developed by convolving the spatial difference matrix (quantifying the difference in surface element abundance) with the pre-stored geometric template in the radioactive material feature library to match areas with stable element distribution and high geometric adaptability. In some embodiments, the spatial difference matrix in step 201 is convoluted and matched with a radioactive material signature library to output a three-dimensional coordinate set of the target capture location, including: 301, segmenting the spatial difference matrix into regions based on reference values ​​for element abundance distribution in the radioactive material signature library, and extracting eigenvectors for each segmented region. In step 301, the spatial difference matrix represents the difference distribution matrix of element abundance at each location on the surface of the radioactive material compared to the overall average, with each value in the matrix corresponding to the element abundance difference at a coordinate point (e.g., whether the uranium content at a certain point is higher or lower than the regional average). The radioactive material signature library is a pre-stored database containing reference data such as typical elemental composition (e.g., uranium and plutonium abundance ratios), surface morphology (e.g., corrosion morphology), and radiation characteristics (e.g., gamma-ray intensity range) for different radioactive materials. The element abundance distribution describes the spatial distribution of the content of each element on the surface of the radioactive material. The region segmentation operator is an algorithm module that divides the spatial difference matrix into several subregions based on a threshold range of element abundance reference values ​​in the signature library. The eigenvector field is a set of quantized vectors that include features such as element distribution and geometric shape within each segmented region.

[0033] In this application example, a spatial difference matrix is ​​first intelligently segmented using a clustering algorithm (such as K-means). Regions with similar elemental abundance differences are merged into the same subregion. For example, regions with minimal fluctuations in uranium element abundance are classified as stable regions. Statistical features (such as abundance mean and variance) and geometric features (such as area and shape index) are then extracted from each subregion to generate a multidimensional feature vector, which is then used to construct a feature vector field. For example, a uranium fuel rod surface is segmented into stable regions at both ends and a corroded region in the middle, and feature vectors are extracted for each region. Step 302 compares the feature vectors of each segmented region with standard feature vectors corresponding to the material category of the radioactive material in the radioactive material feature library layer by layer to generate a feature similarity distribution map. In step 302, the feature vectors of the segmented regions refer to a representative data set for each surface region segmented in step 301, including features such as elemental abundance differences, shape parameters (such as area and boundary curvature), and radiation characteristics (such as radiation gradient direction).

[0034] The radioactive material signature library is a pre-existing database that stores standard data for the classification of different radioactive materials, including characteristic vectors for standard elemental composition, surface morphology, and radiation intensity for each category (e.g., uranium waste and plutonium waste). Layer-by-layer comparison is a module that matches segmented regions against standard features layer by layer based on elemental abundance, geometric shape, and radiation characteristic priority. A feature similarity distribution map quantifies the spatial distribution of the degree of match between segmented regions and standard features, with high values ​​indicating a high degree of match. In this application example, the cosine similarity algorithm is first used to calculate feature vector matching based on layer-by-layer priority. Regions with high elemental abundance matching are first screened, excluding subregions with excessive abundance fluctuations. Geometric shape parameters (e.g., curvature and symmetry) are then compared within the remaining regions to eliminate areas with excessive shape deviations. Finally, the gradient distribution topology map is used to verify radiation characteristic compliance. For example, the ends of a uranium fuel rod receive the highest matching scores due to their stable elemental abundance and symmetrical shape, while the corroded area in the middle is excluded due to its irregular shape. Finally, the results from each layer are integrated to generate a feature similarity distribution map. 303. Identify peak regions that meet preset matching conditions in the feature similarity distribution map, and extract the geometric center coordinates of the peak regions as candidate coordinates. In step 303, the feature similarity distribution map is a visual chart used to display the degree of match between each segmented region on the surface of the radioactive material and the standard features in the feature library. The preset matching conditions refer to pre-set judgment rules used to screen out regions that meet the conditions. The peak region refers to a module that extracts high-matching regions from the similarity distribution map through threshold segmentation and morphological processing. The geometric center coordinates refer to a set of candidate coordinates extracted from the peak region, which serves as the initial point for subsequent optimization.

[0035] In this application example, an adaptive threshold segmentation algorithm (such as the Otsu algorithm) is first used to extract high-matching regions (e.g., regions with a similarity ≥ 0.8) from the feature similarity distribution map. Holes within these regions are then filled using a morphological closing operation to ensure regional connectivity. The geometric center coordinates (e.g., centroid coordinates) of each connected region are then calculated, outputting a chain of candidate coordinates. For example, the stable regions at both ends of a uranium fuel rod are identified as peak regions due to their high matching, and their geometric center coordinates are extracted as candidate grasping points. Step 304: Spatial position compensation is performed on the candidate coordinates based on the radiation intensity attenuation gradient of the gradient distribution topology map at the candidate coordinates, generating a three-dimensional coordinate set for the target grasping location. In step 304, the gradient distribution topology map is a graph showing how radiation intensity on the surface of a radioactive material varies with spatial position, typically using color gradients or contour lines to depict the diffusion trend from high-radiation areas to low-radiation areas. The radiation intensity attenuation gradient refers to a model that performs spatial offset compensation on candidate coordinates based on the radiation attenuation direction of the gradient distribution topology map. The three-dimensional coordinate set is an optimization module that iteratively adjusts the offset of candidate coordinates to ensure that the final coordinates meet radiation safety requirements. In this application example, the direction of the radiation intensity attenuation gradient at the candidate coordinates is first calculated (e.g., from a high-radiation area to a low-radiation area), and a compensation vector is generated (e.g., offsetting a certain distance in the opposite direction of the gradient). An interpolation algorithm is then used to optimize the spatial continuity of the compensated coordinates and verify whether the radiation intensity is below a safety threshold. For example, the candidate coordinates for the middle section of a uranium fuel rod were compensated to the low-radiation zones at both ends due to excessive radiation gradients, ensuring that the capture position avoids high-risk areas.

[0036] The following is a specific example: In a nuclear waste disposal scenario, an intelligent robot needs to grasp a radioactive cylinder with a corrosion layer on its surface. The system first scans the cylinder's surface and segments the spatial difference matrix into stable, corroded, and transitional regions based on elemental abundance reference values. The system then extracts the elemental distribution and geometric shape feature vectors for each region. A hierarchical comparison is then performed to identify stable regions that closely match the standard cylinder template. A feature similarity distribution map is generated, and the stable regions at both ends are identified as high-matching peak regions. The geometric centers of the peak regions are extracted as candidate coordinates. A sudden change in the radiation gradient near the candidate point is detected, and spatial position compensation is performed along the low-radiation direction. Finally, a radiation-safe 3D grasping coordinate is generated, avoiding the corrosion layer, enabling the gripper to precisely grasp the object.

[0037] In summary, steps 301-304 select high-matching areas through intelligent area segmentation and hierarchical comparison, and combine radiation gradient compensation to eliminate positioning deviations, thereby achieving the dual guarantee of accurate generation of radioactive material grabbing coordinates and radiation safety. This method solves the problem of inaccurate grabbing caused by local element fluctuations or radiation interference in traditional solutions, and significantly improves the grabbing success rate and operational safety in nuclear waste sorting scenarios. In order to solve the problem of inaccurate clamping force setting due to uneven distribution of radiation intensity and leakage risk caused by pressure overload in high-radiation areas during the grabbing of radioactive materials, by analyzing the distribution of extreme points of radiation intensity in the gradient distribution topology map, identifying high-radiation mutation areas and their gradient attenuation trends, and realizing regionalized precise constraints on gripper pressure control, ensuring that the grabbing process meets both mechanical stability requirements and avoids radiation leakage risks.

[0038] In some embodiments, in step 202, calculating a pressure threshold set based on the distribution of extreme points of radiation intensity in the gradient distribution topology map includes: 401, marking extreme points in the gradient distribution topology map whose radiation intensity exceeds the neighborhood average, thereby constructing a spatial distribution set of extreme points. In step 401, the radiation intensity includes a set of points in the gradient distribution topology map where the radiation intensity is significantly higher than that of the surrounding areas, representing potential areas of high radiation leakage risk. The spatial distribution set of extreme points refers to a radiation intensity distribution model within a preset neighborhood centered on the extreme point, used to quantify local radiation fluctuation characteristics. In this example, the gradient distribution topology map is first scanned using an edge detection algorithm (such as the Sobel operator) to identify extreme points whose radiation intensity exceeds the neighborhood average. Specifically, the difference between the average radiation intensity of each point and its surrounding neighborhood (e.g., a 3×3 pixel area) is calculated, and points whose difference exceeds a preset threshold are marked as extreme points. Subsequently, a density clustering algorithm (such as DBSCAN) is used to merge spatially adjacent extreme points into clusters of high radiation risk areas to construct the spatial distribution set of extreme points. For example, if radiation intensity is significantly higher in a damaged area on the surface of a nuclear waste container than in the surrounding area, clustering results in a cluster of extreme points. 402. A radiation intensity attenuation model is established with each extreme point as the center, and the rate of change of radiation intensity within a preset radius around the extreme point is calculated. In step 402, the radiation intensity attenuation model is a mathematical model that describes the attenuation trend of radiation intensity at an extreme point with distance, used to quantify the direction and rate of radiation diffusion. The preset radius is a fixed distance range around the extreme point (e.g., a circular area with a radius of 50 cm centered on the extreme point), which defines the calculation boundary of the attenuation model. The rate of change of radiation intensity refers to the rate at which radiation intensity changes with distance or direction within a preset radius, such as the decrease in radiation intensity per unit distance (e.g., a decrease of 100 μSv / h per meter). In this example, a radiation attenuation model is first constructed with each extreme point as the center using a radial basis function interpolation algorithm to simulate the attenuation trend of radiation intensity with distance. Within a preset radius (e.g., a 10 cm range), the rate of change of radiation intensity within that area, i.e., the intensity attenuation per unit distance, is calculated. The gradient descent method is further used to determine the direction with the largest attenuation rate and generate a rate of change field. For example, the radiation intensity at the damaged part of the nuclear waste container decays exponentially to the surrounding areas, and the direction with a high attenuation rate is marked in the rate of change field as a safe clamping path. 403. According to the mapping relationship between the rate of change and the preset safe contact pressure, a dynamic pressure weight is assigned to each extreme point; in step 403, assigning a dynamic pressure weight refers to a module that dynamically adjusts the clamping force constraint weight based on the radiation intensity change rate, and assigns a low weight to the high change rate area to limit the clamping force. The preset safe contact pressure refers to a mechanical operation pressure threshold value pre-set according to different radiation intensity ranges. The mapping relationship refers to the association rule between the predefined radiation change rate and the clamping force upper limit to ensure that the clamping action in high-risk areas is limited.

[0039] In this application example, the safe contact pressure mapping table is first called to map the rate of change value to a corresponding dynamic pressure weight. For example, areas where the rate of change exceeds a threshold (where the radiation decay rate is fast) are assigned a low weight (e.g., 0.2), limiting the gripping force to avoid contact with high-risk areas. Areas with a lower rate of change (where the radiation decay is gradual) are assigned a high weight (e.g., 0.8), allowing for higher gripping force for a secure grip. A linear interpolation algorithm is used to smooth the weight distribution and avoid sudden changes. For example, the edge of a nuclear waste container has a weight of 0.8 due to its gradual radiation decay, allowing the gripper to apply normal pressure.

[0040] 404. Based on the dynamic pressure weights, the radiation intensities in the regions where the extreme points are located are weighted and fused to generate a set of pressure threshold values. In step 404, the dynamic pressure weights refer to the calculation module that performs spatial weighted fusion with the radiation intensity distribution to generate a continuous pressure threshold field. The pressure threshold set refers to the final set of maximum operating pressures allowed for each region. Radiation intensity weighted fusion refers to the process of combining the radiation intensity data of the regions where different extreme points are located with their dynamic pressure weights. In this example, a spatial weighted averaging algorithm is first used to fuse the dynamic pressure weights with the radiation intensity values ​​of the corresponding regions. Specifically, the radiation intensity values ​​in the region where each extreme point is located are weighted to generate an initial pressure threshold distribution. Subsequently, a Gaussian filtering algorithm is used to eliminate local noise interference and improve the continuity of the threshold distribution. For example, the pressure threshold of the damaged area of ​​a nuclear waste container is set to near zero due to its low weight and high radiation intensity. The threshold is relaxed to a safe upper limit for the edge area due to its high weight and low radiation intensity.

[0041] The following is a specific example: In a nuclear waste treatment plant, an intelligent robot must sort radioactive containers with damaged surfaces. These containers, due to leakage from the interior, have an abnormally high radiation intensity in a local area. The system scans the container surface using a gradient distribution topology map and detects that the radiation intensity at the damaged area is significantly higher than in the surrounding area. An edge detection algorithm identifies multiple extreme radiation points at the damaged edge, clusters them, and marks them as high-radiation exclusion zones. A spatial distribution set of these extreme points is then constructed. A radiation intensity decay model is then established, centered around the extreme point at the damaged area. Analysis reveals that the radiation intensity at the damaged area decays exponentially, with the decay rate increasing with proximity to the damaged area. The rate of change field marks the area near the damaged area as having a high decay rate, while the decay gradually decreases away from the damaged area. Based on the mapping between the radiation decay rate and pre-set safety rules, the high decay rate area near the damaged area is assigned a low dynamic pressure weight, limiting the maximum gripping force of the gripper during contact. The container edge, due to its gentle decay, is assigned a higher weight, allowing the gripper to apply normal pressure for a secure grip. The dynamic pressure weights are spatially fused with the radiation intensity distribution to generate a set of pressure thresholds. The damaged area has an extremely low weight and high radiation intensity, so the pressure threshold is set close to zero, prohibiting the gripper from contacting. The undamaged edge area of ​​the container has a high weight and low radiation intensity, so the pressure threshold is set to a safe upper limit, and the gripper can apply force normally to grasp.

[0042] In summary, steps 401 to 404 dynamically quantify the risk area through extreme point identification and radiation attenuation modeling, and combine weight distribution and fusion to generate a set of pressure thresholds. The gripper completely avoids the high-radiation damage area and only applies adaptive clamping force in the safe area. In the radioactive container sorting scenario, it not only eliminates the risk of leakage caused by clamping overload, but also ensures the stability of grasping, and achieves safe and efficient operation in high-risk environments. In order to solve the problems of insufficient gripper adaptability, unstable grasping and radiation leakage risk caused by the complex geometric shape or surface deformation of radioactive materials, the surface contour of the target object is matched with the geometric contour template in the radioactive material feature library. The optimal matching template is screened and its geometric feature parameters are extracted. The shape adaptation parameter vector is generated by combining the real-time surface deformation compensation algorithm, and the gripper deformation mechanism is driven to dynamically adjust the curvature radius and clamping angle to achieve high-precision fitting of the gripper geometric envelope with the target surface, ensuring the stable grasping and radiation safety of complex-shaped radioactive materials.

[0043] In some embodiments, generating a shape adaptation parameter vector based on a geometric contour template pre-stored in the radioactive material feature library in step 203 includes the following steps: 501. Extracting a set of radioactive material surface curvature feature points from the geometric contour template pre-stored in the radioactive material feature library and constructing a contour piecewise function. In step 501, the radioactive material feature library refers to a pre-stored database containing standard geometric contour data (e.g., three-dimensional models of cylinders, spheres, or complex surfaces) and surface characteristic parameters (e.g., curvature distribution, material hardness) for different radioactive materials. The geometric contour template refers to a standard surface shape model of radioactive materials stored in the feature library. The curvature feature point set refers to a collection of surface curvature extreme points (e.g., peaks and valleys) extracted from the pre-stored geometric contour template, used to describe the geometric characteristics of the object's contour. The contour piecewise function refers to a mathematical model that converts a discrete set of curvature feature points into a continuous piecewise function, characterizing the curvature variation patterns in different intervals.

[0044] In this application example, a geometric contour template of a target object (e.g., a three-dimensional model of a cylindrical nuclear waste tank) is retrieved from a radioactive material feature library. A curvature analysis algorithm is used to extract a set of surface curvature feature points (e.g., vertices of arcs on the tank's end face and endpoints of straight segments on the barrel). Based on the distribution pattern of the feature point set (e.g., the demarcation points between arc and straight segments), the contour is divided into multiple geometrically continuous subsegments (e.g., arc segments on the end face and straight segments on the barrel). An independent contour function is constructed for each segment (e.g., arc segments are described using arc equations and straight segments are described using linear equations). Ultimately, all the segmented functions are combined into a complete contour model, which is used to guide the generation of deformation parameters or automated detection of mechanical fixtures. Step 502: Based on the spatial relative positional relationship between the target grasping position in the three-dimensional coordinate set and the contour segmented function, curvature adaptation parameters are calculated for the contact surface of the gripper units in the mechanical fixture array. In step 502, the spatial relative position field refers to a quantitative model describing the spatial relationship between the target grasping position and the contour segmented function, which is used to locate the matching region between the gripper contact surface and the contour curve. A mechanical gripper array is a gripping device composed of multiple independently controlled gripper units, each of which can adapt to the target surface by adjusting its extension or deformation. Contact surface adaptation is the process of matching the gripper's contact surface shape (e.g., curvature and angle) with the target surface geometry to ensure uniform contact pressure distribution. Curvature adaptation parameters are a sequence of gripper curvature control parameters generated based on spatial matching relationships to ensure that the gripper's deformation profile matches the target surface geometry.

[0045] In this embodiment, the gripper's curvature parameters are calculated by matching the target grasping position with a contour segment function (e.g., arc segments or straight line segments). When the grasping point is located in an arc region, the gripper's curvature matches the geometric parameters to achieve a close fit; when located in a straight line region, the gripper's curvature is adjusted to a planar contact mode. Multiple gripper units coordinately adjust based on the contour function to ensure a seamless curvature transition (e.g., the arc-to-straight transition zone). A smooth envelope is dynamically generated to distribute pressure, and the adaptation effect is verified through real-time pressure feedback. For example, in the case of grasping a cylindrical container, the top gripper adapts to the arc curvature, while the bottom gripper adjusts to a planar mode. This coordinated force evenly distributes pressure, avoiding stress concentration and achieving safe grasping and radiation protection for complex geometries. Step 503: The curvature adaptation parameters are normalized with the radiation intensity gradient values ​​at corresponding locations in the gradient distribution topology map to generate a shape adaptation parameter vector. In step 503, the gradient distribution topology map is a graph showing how the radiation intensity of a radioactive material surface changes with spatial position, using color gradients or arrows to indicate the direction and rate of radiation intensity diffusion. A normalized fuser refers to an algorithm module that fuses the curvature adaptation parameters and the radiation gradient values ​​in a weighted ratio to balance geometric adaptation and radiation safety. The shape adaptation vector field refers to a set of shape control parameters covering the entire area generated after fusion, which drives the dynamic deformation of the gripper. In an embodiment of the present application, a linear normalization algorithm is used to scale the curvature adaptation parameters and the radiation gradient values ​​at the corresponding positions to the same dimension; then, the parameters are weightedly fused based on the radiation safety weight (e.g., high radiation areas have low weights) to generate a shape adaptation parameter vector. For example, the curvature parameters of high radiation areas are compressed to avoid overfitting, while the original curvature parameters are retained in low radiation areas.

[0046] The following is a specific example: In a smart home scenario, a robot needs to grasp a ceramic water cup with a partially overheated surface. The cup body is a standard cylindrical shape, but the handle, filled with hot water, has an abnormally high temperature, similar to a high radiation risk zone for radioactive materials. The system uses the "Cylindrical Cup with Handle" template from the feature library to extract curvature feature points from the uniform curvature section of the cup body and the curved section of the handle (for example, the curvature is constant in the middle of the cup body, while the curvature changes suddenly at the handle bend). A piecewise function is then constructed to describe the overall contour. The target grasp position is located in the cooler region in the middle of the cup body. Spatial mapping is used to match the uniform curvature section of the cup body, and the gripper calculates the constant curvature required to conform to the cylindrical surface. The system also detects an abnormal temperature in the handle area, which it simulates as a "thermal radiation gradient" and marks as a risk zone. The curvature parameters of the cup body are normalized and combined with the "thermal gradient" value of the handle. The hot handle area is assigned a low weight, and the gripper automatically relaxes its curvature to avoid contact with the hot part. The cooler area of ​​the cup body is assigned a high weight, and the gripper maintains a stable grip with a constant curvature. Ultimately, the robot gently grips the cup body with an adaptive shape, avoiding the hot zone of the handle, just as a human hand naturally avoids the hot steam at the cup's rim. In summary, steps 501 to 503 achieve precise modeling of the geometric contour through curvature feature extraction and piecewise function construction. This, combined with spatial mapping to optimize the gripper's curvature parameters, and the integration of radiation gradient data to dynamically adjust the adaptation weights, solves the gripper adaptation challenge in scenarios with complex geometry and coupled radiation distribution.

[0047] To address the issues of motion deviation, low coordination efficiency, and radiation leakage risk caused by mismatches between gripper deformation parameters and mechanical control instructions, a spatial registration algorithm is used to perform multimodal data alignment of the shape adaptation parameter vector (e.g., curvature radius, gripping angle) with the three-dimensional coordinate set and gradient distribution topology of the mechanical fixture array. This allows for a deformation-displacement coupling model to be constructed, achieving dynamic coordinated matching of the gripper's geometric envelope with the target surface characteristics. In some embodiments, the conversion of the shape adaptation parameter vector into a displacement control instruction set for the mechanical fixture array in step 103 includes the following steps: 601: performing degree-of-freedom decoupling on the shape adaptation parameter vector to separate geometric constraint parameters and radiation gradient constraint parameters corresponding to the axial displacement of the gripper units in the mechanical fixture array. In step 601, degree-of-freedom decoupling refers to decomposing the composite parameter into independent control variables, ensuring that adjustments of different degrees of freedom (e.g., axial displacement, rotation angle) do not interfere with each other. Geometric constraint parameters refer to parameters that control the gripper displacement to conform to the target shape (e.g., displacement amount, curvature matching accuracy). Radiation gradient constraint parameters refer to parameters that control the gripper displacement path to avoid radiation risks (such as moving along the direction of low radiation gradient and residence time limit).

[0048] In the embodiment of the present application, by decoupling the degree of freedom of the shape adaptation parameter vector, the composite parameters of the fusion geometric shape adaptation (such as the radius of curvature) and the radiation gradient constraint (such as the avoidance direction) are separated into independently controllable geometric constraint parameters and radiation gradient constraint parameters. Specifically, the geometric constraint parameters (such as the axial displacement of the gripper and the curvature matching accuracy) are extracted from the parameter vector to drive the gripper to fit the target surface morphology, while the radiation gradient constraint parameters (such as the displacement path direction and the moving speed) are separated to control the gripper to quickly evacuate the high-risk area along the radiation attenuation direction; the two types of parameters generate displacement instructions respectively through independent control modules. For example, the geometric parameters guide the gripper's extension and contraction amount to match the target curvature, while the radiation parameters dynamically plan the gripper's moving path to minimize radiation exposure. The two are decoupled and parallel execution and real-time feedback calibration are achieved through a collaborative optimization algorithm. 602. Orthogonally project the geometric constraint parameters onto the spatial pose information in the three-dimensional coordinate set to generate an initial axial displacement component. In step 602, the spatial pose information in the three-dimensional coordinate set refers to the position (coordinates) and posture (orientation angle, rotation angle) data of the mechanical fixture in three-dimensional space, such as the position coordinates of the gripper tip and the orientation of its axis. The orthogonal projection field is a mathematical model that maps the geometric constraint parameters into a three-dimensional coordinate space, generating a displacement component corresponding to the gripper's pose. The axial displacement component describes the dynamic relationship between the gripper's geometric deformation parameters and its spatial pose (e.g., position and orientation).

[0049] In this embodiment of the present application, an orthogonal projection algorithm is used to map geometric constraint parameters (such as the radius of curvature) to the spatial position of the gripper (such as the axial position and circumferential angle) in a three-dimensional coordinate set. For example, the radius of curvature of a cylindrical gripper is converted to the axial displacement of the telescopic rod, and the clamping angle is mapped to the circumferential rotation angle. A rigid body kinematic model is introduced into the projection process to ensure the physical consistency of the geometric parameters and the displacement components, ultimately generating an initial axial displacement component. Step 603: Extract the radiation intensity gradient direction vector corresponding to the three-dimensional coordinate set from the gradient distribution topology map, and establish a displacement correction correlation factor between the gradient direction and the gripper axial direction. In step 603, the gradient direction vector field refers to the set of vectors indicating the direction of radiation intensity variation in the gradient distribution topology map, which is used to correct the gripper movement path. The displacement correction correlation factor is a weight parameter that quantifies the effect of the gradient direction on the gripper axial displacement, such as a reverse correction coefficient. In this embodiment of the present application, vector field analysis techniques are used to extract the radiation intensity gradient direction vector (e.g., from a high radiation area to a low radiation area) from the gradient distribution topology map. The displacement correction correlation factor is established by calculating the angle between the gradient direction and the gripper axial direction. For example, if the gradient direction is opposite to the gripper movement direction (i.e., the gripper moves toward the high-radiation area), the correlation factor triggers reverse displacement compensation, driving the gripper to bypass the high-risk area.

[0050] 604. Dynamically compensate the initial axial displacement component based on the displacement correction correlation factor to generate an adaptive displacement parameter set containing a displacement directional deviation correction. In step 604, the dynamic compensator is a module that adjusts the initial displacement component in real time based on the displacement correction correlation factor to achieve path avoidance. The initial axial displacement component is a baseline displacement instruction generated through orthogonal projection, representing the original extension and contraction of the gripper along the axis. Dynamic compensation refers to adjusting the direction or magnitude of the initial displacement based on real-time data (such as changes in radiation gradient) to offset external interference or risks. The adaptive displacement parameter set is a set of displacement parameters that meet radiation safety constraints after correction, including the directional deviation correction. In this embodiment of the present application, the displacement correction correlation factor is dynamically integrated with the initial axial displacement component through a PID control algorithm. For example, when the gripper moves toward a high-radiation area, the correlation factor generates a reverse compensation to reduce the displacement speed or adjust the movement direction; in low-radiation areas, the original displacement parameters are maintained. Through real-time feedback adjustment, an adaptive displacement parameter set is generated to ensure that the gripper path avoids areas with sudden changes in radiation.

[0051] 605. The adaptive displacement parameter set and the radiation gradient constraint parameters are coupled and decomposed according to the kinematic model of the gripper unit to generate the axial displacement parameters for each telescopic rod in the displacement control instruction set. In step 605, the adaptive displacement parameter set refers to a set of correction parameters that include the overall displacement direction, distance, and speed of the gripper, used to guide the gripper to adhere to the target surface while avoiding radiation risks. The kinematic model of the gripper unit is a mathematical model that describes the relationship between the gripper's mechanical structure and motion. It can decompose the overall displacement target into independent motion parameters (such as extension amount and speed) for each telescopic rod. Coupled decomposition involves fusing the adaptive displacement parameters and the radiation constraint parameters according to the geometric relationship and physical constraints of the kinematic model to generate coordinated motion instructions for each telescopic rod. The displacement control instruction set refers to the resulting instruction set, which includes the axial displacement, direction, and timing parameters for each telescopic rod. In this embodiment of the present application, an inverse kinematic solution algorithm is used to decompose the adaptive displacement parameter set (such as axial displacement and direction correction) and the radiation gradient constraint parameters (such as gradient threshold and avoidance weight) according to the gripper's degrees of freedom. For example, the radius of curvature of a cylindrical gripper is converted into the synchronized extension and retraction of multiple telescopic rods, while the directional correction is decomposed into the compensation displacement of a specific telescopic rod. Ultimately, the axial displacement parameters of each telescopic rod are generated to ensure multi-axis coordinated motion.

[0052] 606. A radiation intensity attenuation field model centered at the target grasping position is established within the three-dimensional coordinate set, and the radiation intensity attenuation rate along each axial path is calculated. In step 606, the core point at the target grasping position where the mechanical gripper contacts or operates is typically the optimal grasping point after geometric adaptation of the radioactive material surface. The radiation attenuation field model is a spatial distribution model of the radiation intensity attenuation rate centered at the target grasping position. The axial path attenuation rate field is quantitative data describing the radiation attenuation trend along the axial movement path of the gripper.

[0053] In an embodiment of the present application, a radiation attenuation field model is constructed by a Kriging interpolation algorithm to simulate the radiation intensity attenuation trend when the gripper moves along the axial path. For example, when approaching a high-radiation area, the attenuation rate accelerates, triggering a displacement speed adjustment; when away from a high-risk area, the attenuation rate is slow, allowing normal movement. By calculating the path attenuation rate, a basis for dynamic speed control is provided. 607. Based on the difference between the attenuation rate and a preset safety threshold, a dynamic scaling coefficient for the axial displacement is generated; in step 607, the safety threshold difference field refers to the dynamic deviation data used to quantify the attenuation rate and the preset safety threshold. The dynamic scaling coefficient chain refers to a sequence of displacement speed scaling parameters generated based on the deviation between the attenuation rate and the safety threshold.

[0054] In an embodiment of the present application, a fuzzy logic control algorithm is used to map the difference between the attenuation rate and the safety threshold into a dynamic scaling factor. For example, when the attenuation rate exceeds the safety threshold (the gripper approaches a high radiation area), a scaling factor less than 1 is generated to reduce the displacement speed; when the attenuation rate is lower than the threshold, the scaling factor is set to 1 to maintain the original speed. Through real-time scaling adjustment, movement efficiency and radiation safety are balanced. 608. The axial displacement parameter is multiplied point by point by the dynamic scaling factor to obtain a set of telescopic rod displacement parameters adapted to the radiation environment; in step 608, the dynamic scaling factor refers to the scaling factor calculated based on the radiation intensity attenuation field model in step 606, which is used to dynamically adjust the position according to changes in radiation intensity. The point-by-point product operation of the displacement refers to a one-to-one multiplication operation of the axial displacement parameter of each telescopic rod with the corresponding dynamic scaling factor to generate a corrected displacement adapted to the radiation environment. The radiation-adapted displacement parameter set represents the final set of displacement parameters after integrating radiation safety constraints. In the embodiment of the present application, the system assigns a dynamic scaling coefficient to each area where the telescopic rod is located based on the spatial distribution characteristics of the radiation intensity around the target grasping position. The coefficient in the high-radiation area is less than 1 to reduce the displacement and increase the speed, shortening the exposure time; the coefficient in the low-radiation area is greater than 1 to amplify the displacement and reduce the speed to ensure grasping stability. The original axial displacement instruction is dynamically scaled according to the radiation risk through point-by-point multiplication. For example, the displacement of the telescopic rod in the high-radiation area is reduced to 70% of the original value and accelerated to retract, while the displacement in the low-radiation area is amplified to 130% and decelerated and pressurized. The final generated adaptation parameter set simultaneously optimizes the displacement, speed and path direction, so that the gripper moves along the safe path with the fastest radiation attenuation while conforming to the target surface, taking into account both operational efficiency and radiation protection requirements.

[0055] 609. Based on the correlation of radiation intensity gradients between adjacent coordinate points in the gradient distribution topology, a set of displacement coordination constraint equations is constructed between the gripper units, and the resulting telescopic rod displacement parameters are solved. In step 609, the gradient distribution topology is a graph showing the variation of radiation intensity on the surface of a radioactive material with spatial position, with the radiation diffusion direction and rate in different regions indicated by color or arrows. Radiation intensity gradient correlation refers to the linkage between radiation intensity changes between adjacent coordinate points. For example, an increase in radiation intensity at a point will cause a synchronous fluctuation in radiation values ​​in surrounding areas. The set of displacement coordination constraint equations is a set of mathematical equations that describe constraints such as synchronization of multiple gripper unit movements and path collision avoidance. The resulting telescopic rod displacement parameters are the actual displacement instructions for each telescopic rod generated by solving the set of constraint equations, including displacement, speed, and direction. In this embodiment of the present application, the displacement coordination constraint equations are constructed, including conditions such as multi-gripper synchronization, path collision avoidance, and force balance, and the optimal displacement parameters are solved using the Newton iteration method. For example, when multiple grippers grasp a cylinder, they must extend and retract synchronously to avoid unilateral overload, and path planning must avoid conflicts between adjacent gripper movements. By solving the equation, the final executable telescopic rod displacement parameters are generated.

[0056] The following is a specific example: In a smart home scenario, a robot needs to grasp a ceramic cup filled with hot water. The cup body is intact, but the handle poses a scalding risk due to high temperatures. The system decouples shape adaptation parameters, separating the cup body geometry (uniform cylindrical curvature) from the handle temperature distribution (marking the hot zone). The cup body curvature parameters are projected into the initial displacement of the gripper's telescopic rod, planning the gripper to grasp from the center of the cup. The temperature gradient direction in the handle area (increasing temperature from the cup body to the handle) is extracted to generate a reverse displacement correction factor, prompting the gripper to move away from the handle. The initial displacement path is corrected, allowing the gripper to fine-tune from the center of the cup body toward the cooler zone, avoiding the hot zone on the handle. The corrected displacement parameters are decomposed into multiple sets of telescopic rod movements, driving the gripper to wrap around the cup in an arc. A temperature decay model for the handle area is constructed to detect a sharp temperature rise near the handle. A dynamic scaling factor is generated, causing the gripper to automatically decelerate to a "tentative touch" mode as it approaches the handle. The gripper's displacement is adjusted to ensure only light contact with the cup near the handle, while the displacement in the hot zone is compressed to a safe distance. Coordinate the synchronous movement of multiple groups of grippers to ensure that the cup body is evenly stressed while completely avoiding the high-temperature area of ​​the handle. In summary, steps 601 to 609 achieve radiation safety adaptation and multi-axis precision coordination of the gripper displacement path through degree of freedom decoupling, dynamic compensation and collaborative constraint solving, solving the path intrusion, speed mismatch and action conflict problems caused by insufficient radiation-geometry coupling control in traditional solutions. In order to solve the problems of incomplete sealing, increased risk of radiation leakage and path conflict caused by the mismatch between the moving path of the mechanical fixture and the expansion rate of the sealing layer, a nonlinear correlation model of the spatial difference matrix (characterizing the difference in the distribution of elements on the surface of radioactive materials) and the growth rate of the thickness of the sealing isolation layer is established to drive the fixture array to synchronously match the expansion rate of the sealing layer during movement, thereby realizing dynamic coupling of path speed, direction and sealing protection, and ensuring complete coverage of the sealing layer in high-risk areas and safe coordination of the fixture action.

[0057] In some embodiments, the dynamic reconstruction of the movement path control parameters of the mechanical fixture array by mapping the spatial difference matrix with the thickness growth rate of the sealing isolation layer in step 105 includes the following steps: 701: extracting spatial weights of element abundance distribution from the spatial difference matrix and establishing a dynamic response relationship model between the element abundance distribution at each coordinate point and the thickness growth rate of the sealing isolation layer. In step 701, the element abundance spatial weights refer to parameters in the spatial difference matrix that quantify the stability of element distribution in different regions. A high weight indicates uniform element abundance and suitable for stable expansion of the sealing layer. The dynamic response relationship model is a mathematical model established through machine learning that describes the dynamic influence of element abundance distribution on the thickness growth rate of the sealing layer. In this embodiment, a random forest algorithm is used to analyze the element abundance data in the spatial difference matrix to identify the weight differences between stable element distribution regions (e.g., the ends of the uranium fuel rods) and fluctuating regions (e.g., oxidation corrosion zones). A nonlinear mapping relationship between element abundance weights and sealing layer thickness growth rate is established by training historical sealing layer expansion data. For example, regions with stable element abundance are predicted to be areas of rapid sealing layer expansion, while fluctuating regions are mapped to areas of slow expansion. 702. Obtain in real time the thickness growth rate of the sealing isolation layer in each area of ​​the dynamic geometric envelope surface, and calculate the path adjustment priority coefficient of the corresponding area according to the dynamic response relationship model; in step 702, the sealing isolation layer refers to a protective layer formed by heat shrinkable materials or coatings, which covers the surface of the target object and is used to isolate the leakage of radioactive substances or the invasion of external contaminants. The dynamic response relationship model refers to a mathematical model that describes the relationship between the thickness growth of the sealing layer and the path adjustment of the mechanical fixture. For example, if the thickness grows too fast, the movement speed needs to be reduced to strengthen the seal, and if the thickness is insufficient, the movement needs to be accelerated to supplement it. The path adjustment priority coefficient refers to the regional priority parameter generated according to the deviation between the actual expansion rate of the sealing layer and the model prediction value. A high coefficient indicates that the path needs to be adjusted urgently.

[0058] In this embodiment, the sealant layer thickness growth rate is monitored in real time and compared to the predicted value from the dynamic response relationship model to calculate the deviation rate. Regions with excessive deviation rates (e.g., where sealant layer expansion stalls at a damaged area) are marked as high priority, and a priority coefficient matrix is ​​generated. For example, in a damaged area, where the actual growth rate is significantly lower than the predicted value, the trigger priority coefficient is set to the highest, causing the fixture to prioritize adjusting the path in that area.

[0059] 703. Based on the gradient variation trend of the thickness growth rate in three-dimensional space, a movement path correction vector is generated for each gripper unit of the mechanical gripper array. In step 703, the thickness growth rate refers to the real-time thickening rate of the sealing isolation layer in a specific area of ​​the dynamic geometric envelope surface, typically expressed as the thickness change per unit time (e.g., mm / minute). The gradient variation trend in three-dimensional space refers to the pattern of thickness growth rate changes along different directions in the three-dimensional coordinate system (X / Y / Z axes), for example, the thickness growth rate of a certain area decreases in the northwest direction and increases in the southeast direction. The movement path correction vector refers to the gripper path adjustment parameters generated based on the gradient trend, including a directional offset and a speed correction. In this embodiment of the present application, a gradient descent algorithm is used to analyze the thickness gradient variation trend field to determine the direction with the fastest sealing layer expansion rate (e.g., from a low-radiation area to a high-radiation area). The movement path correction vector is generated to drive the gripper along the expansion direction, and the gripper movement speed is dynamically adjusted based on the rate difference. For example, when the sealing layer expansion rate is low at the damaged area, the gripper decelerates and increases local pressure to accelerate the sealing filling. 704. Perform a spatial convolution operation on the path correction vector and the element abundance weight values ​​in the spatial difference matrix to output a real-time offset parameter set for the movement path. In step 704, the element abundance weight value refers to the weight coefficient assigned to the abundance difference of different elements (such as uranium and plutonium) based on the hazard level or treatment priority of the radioactive material (for example, a weight of 0.9 for uranium-exceeding areas and a weight of 0.3 for oxygen-stable areas). Spatial convolution is a mathematical operation that uses a sliding window (convolution kernel) to perform a weighted summation on local areas of the spatial difference matrix and combines it with the path correction vector to generate a comprehensive offset. The real-time offset parameter set refers to the set of displacement adjustment parameters formed by the fusion of the path correction vector and the element abundance weight.

[0060] In this embodiment, a three-dimensional convolution kernel is used to spatially weight the path correction vector and element abundance weights. Path correction vectors in high-weight regions (such as the ends of a container with stable element abundances) are amplified to ensure the gripper precisely adheres to the seal expansion path. Noise correction is suppressed in low-weight regions (such as corroded sections). For example, at damaged locations with low element abundance weights, the correction vector is weakened to prevent over-adjustment that could cause tearing of the seal. 705. A radiation intensity diffusion model centered on the current gripper position is established within the gradient distribution topology map. The movement path control parameters of the mechanical gripper array are dynamically reconstructed based on the superposition of the real-time offset parameter set and the diffusion model. In step 705, the radiation intensity diffusion model is a dynamic model that predicts the radiation intensity diffusion trend along the gripper's movement path, used to avoid high-risk areas. The movement path control parameters are the final motion instructions for the mechanical gripper array, including the displacement, direction, speed, and coordinated timing of each gripper unit. Dynamic reconstruction refers to the process of updating path planning parameters based on real-time input data (such as radiation diffusion changes and offset adjustments). Among them, the dynamic reconstruction process includes: identifying the emergency avoidance area that requires path optimization based on the abnormal mutation characteristics of the growth rate of the thickness of the sealing isolation layer in the corresponding area of ​​the spatial difference matrix; establishing a composite constraint condition of the element abundance threshold and the radiation gradient direction in the emergency avoidance area to generate the dynamic deflection angle parameter of the movement direction of the gripper unit; based on the dynamic deflection angle parameter, the collaborative motion trajectory parameter is corrected section by section to form a radiation protection enhanced mobile path control parameter set. In an embodiment of the present application, a radiation intensity diffusion model is constructed based on the gradient distribution topology map to predict the radiation intensity change trend on the gripper movement path. The real-time offset parameter is superimposed on the diffusion model, and the path parameters are optimized using a genetic algorithm to avoid high-risk areas of radiation diffusion. For example, when the gripper approaches a high radiation diffusion area, the path automatically detours to a safe area completely covered by the sealing layer to ensure the safety of the operator and the environment.

[0061] The following is a specific example: In a household scenario, an intelligent robot repairs a kettle with a damaged inner wall. High-temperature steam leaks within the kettle pose a radioactive risk. The robot scans the inner wall of the kettle to identify intact areas (stable stainless steel) and damaged areas (fluctuating iron oxidation). A model predicts that the sealing layer (thermal insulation coating) in the intact areas will expand faster. The damaged area, detected to be slow to expand due to high-temperature oxidation, is marked as the highest priority, triggering the robot to prioritize its work. A path correction vector is generated, causing the gripper to slow down as it approaches the damaged area and pressurized spray insulation material to accelerate the filling process. By incorporating element weights (higher weight for stainless steel areas), the correction vector is weakened at the damaged area to avoid overspray and uneven coating. The high-temperature steam diffusion model indicates an increased risk at the damaged area. After path reconstruction, the gripper detours to a nearby safe zone, filling the gap layer by layer until the seal is complete. In summary, steps 701 to 705, through the integration of dynamic modeling of element abundance and sealing rate and path correction, achieve real-time coordination between the gripper path and the expansion of the sealing layer, addressing the risks of incomplete sealing and leakage caused by path mismatch in traditional solutions. In order to solve the problems of unstable grasping, surface damage and radiation leakage risk caused by the dynamic mismatch between the gripper deformation profile and the target geometric profile, a dynamic geometric envelope optimization method based on pressure-deformation closed-loop control was developed. Dual closed-loop control of geometric matching accuracy and radiation safety is achieved during the grasping process. In some embodiments, in step 105, during the axial movement of the telescopic rod, the contact pressure data stream of the air pressure sensor is compared with the pressure threshold set in real time, and the curvature radius of the gripper deformation profile is adjusted according to the comparison difference so that the error range between the dynamic geometric envelope formed by the gripper deformation profile and the geometric profile template converges to within the threshold. The method includes: 801, obtaining the contact pressure data stream of the air pressure sensor, filtering and denoising the contact pressure data stream, and extracting the real-time contact pressure value as the pressure threshold set; in step 801, the contact pressure data stream refers to the original pressure time series signal collected in real time by the distributed air pressure sensor on the gripper surface, which contains mechanical vibration noise and environmental interference. Filtering and denoising involves processing the raw pressure data stream to remove high-frequency noise (such as mechanical vibration) and abnormal fluctuations (such as transient shock), while preserving the effective pressure trend. The real-time contact pressure value, obtained after filtering, represents the actual contact force between the fixture and the target object.

[0062] In this embodiment, a wavelet transform filtering algorithm is used to perform multi-scale decomposition on the raw pressure data stream. Transient noise caused by mechanical vibration (such as jitter in the gripper extension rod) is removed through wavelet high-frequency coefficient thresholding. A sliding window mean filter is used to eliminate baseline offsets caused by temperature drift, preserving the true contact pressure trend. The filter window size is adaptively adjusted based on the gripper's movement speed, shrinking the window at high speeds to preserve detail and expanding it at low speeds to enhance smoothness. Ultimately, a stable and reliable real-time contact pressure value is output. For example, when the gripper contacts a raised surface, the pressure peak is accurately captured while noise fluctuations are suppressed. 802. The pressure threshold set is obtained, and the pressure threshold corresponding to the current gripper deformation profile is extracted as a contact pressure comparison benchmark. In step 802, the pressure threshold set refers to the dynamic or preset pressure range set generated in step 801, which includes the upper and lower safety pressure limits (e.g., 50N to 70N for gripping mode) for different operating modes (e.g., gripping, sealing, and releasing). The deformation profile of the gripper refers to the real-time shape change caused by the force after the gripper contacts the target object, such as geometric features such as the curvature, contact area, or displacement of the telescopic rod. The contact pressure comparison benchmark refers to the pressure threshold range extracted from the pressure threshold set and matched with the current gripper deformation profile, which is used to determine whether the real-time contact pressure is in a safe range. In an embodiment of the present application, the dynamic matching of the deformation profile and the threshold is achieved through a spatial hash mapping algorithm: the parameters such as the current gripper curvature radius and contact angle are hashed into a unique index value. The upper limit of the pressure of the corresponding area is extracted from the threshold table according to the hash index (such as a curvature radius of 4 cm corresponding to a threshold of 50N). Bilinear interpolation is used to smooth the transition threshold of the deformation profile boundary area to avoid sudden changes. For example, when the gripper grasps the side wall of a cylinder, the hash code is located in the "cylindrical surface-middle section" threshold area, and the pressure reference value under the radiation safety constraint of this area is extracted.

[0063] 803. Calculate the difference between the real-time contact pressure value and the contact pressure comparison benchmark to generate a contact pressure comparison difference. In step 803, the real-time contact pressure value refers to the actual pressure measurement between the contact surface of the mechanical fixture and the target object at the current moment (e.g., 52.3 N), filtered and output by the air pressure sensor. Difference calculation involves comparing the real-time contact pressure value with the upper and lower limits of the comparison benchmark (e.g., if the real-time value exceeds the upper limit, the difference is positive; if it falls below the lower limit, the difference is negative). The contact pressure comparison difference is used to quantify the deviation of the real-time pressure from the benchmark (e.g., +5 N indicates overpressure, -8 N indicates underpressure) and guide subsequent pressure adjustment strategies. In this embodiment, a weighted sliding standard deviation algorithm is used to calculate the pressure difference: a weight of 0.8 is assigned to high-radiation areas and a weight of 0.2 to emphasize the impact of the difference in high-risk areas. The deviation between the real-time pressure value and the benchmark is multiplied by a weight coefficient to generate a weighted difference signal. A sliding window standard deviation analysis is performed on the difference signal. Areas with three consecutive window standard deviations exceeding the limit are marked as abnormal. For example, when the gripper contacts the high-radiation area of ​​a nuclear waste container, the pressure difference is weighted amplified, triggering adjustments for even slight overshoots, while brief fluctuations are permitted in normal areas. 804. Based on the contact pressure comparison difference, the curvature radius of the gripper deformation profile is dynamically adjusted in combination with the current curvature radius of the gripper deformation profile, the elastic modulus of the gripper material, and historical deformation adjustment records, generating an updated gripper deformation profile. In step 804, the contact pressure comparison difference refers to the degree of deviation between the real-time contact pressure calculated in step 803 and the baseline value (e.g., +5N indicates overpressure, -3N indicates underpressure). The elastic modulus of the gripper material refers to the material's physical property of resisting elastic deformation (e.g., the elastic modulus of titanium alloy is 110 GPa), which determines the range of recoverable deformation of the gripper after being subjected to force. Historical deformation adjustment records store the gripper's deformation adjustment data from past operations (e.g., curvature radius adjustment, pressure deviation correction records), which are used to optimize dynamic adjustment strategies. The updated gripper deformation profile refers to the adjusted gripper geometry parameters (such as the new curvature radius R=55mm) to ensure that the contact pressure is evenly distributed and meets the safety threshold.

[0064] In this embodiment, a reinforcement learning algorithm is used to dynamically adjust the curvature radius. Inputs include the current pressure difference, the material elastic modulus (e.g., silicone hardness coefficient), and historical adjustment records (e.g., the directions of the last five curvature changes). A Q-learning model selects the optimal adjustment direction (increase / decrease curvature) and amplitude (fine adjustment / large adjustment). The adjusted pressure difference change and the error convergence rate serve as reward signals to update the policy network parameters. For example, when the gripper exceeds the pressure limit due to contact with a protrusion, the model, based on historical success cases, selects the "curvature radius + 0.5 cm" action to quickly release the pressure. At step 805, a multi-dimensional error calculation is performed between the dynamic geometric envelope formed by the updated gripper deformation profile and a geometric profile template to generate a dynamic geometric envelope error value. In step 805, the geometric profile template refers to a pre-stored standard geometric shape model of the target object (e.g., a cylinder, sphere, or custom surface) used to compare the actual contact profile. Multi-dimensional error calculation involves calculating the difference between the dynamic geometric envelope and the template in multiple dimensions, such as position, curvature, and contact area (e.g., a position deviation of ±0.5 mm, a curvature deviation of 3%). The dynamic geometric envelope error value refers to a set of values ​​that quantifies the degree to which the actual contact profile deviates from the template (such as total error value = position error + curvature error + area error). In the embodiment of the present application, the analytic hierarchy process (AHP) is used for multi-dimensional error fusion. The first-level indicator is the geometric shape error (weight 40%), the second-level is the contact mechanics error (weight 30%), and the third-level is the radiation safety error (weight 30%). In the early stage of grasping, emphasis is placed on rapid matching of geometric shapes (the geometric weight is increased to 60%), and in the later stage, emphasis is placed on pressure balance (the mechanical weight is increased to 50%). The errors in each dimension are normalized and weighted summed to output a standardized error value of 0-1. For example, in the initial stage of gripping, a 5% shape error is allowed for rapid fitting, and in the later stage, the error needs to be converged to within 1% to ensure sealing.

[0065] 806. Determine whether the dynamic geometric envelope error exceeds a preset error threshold. If so, iteratively adjust the curvature radius of the gripper deformation profile based on the error distribution characteristics and historical adjustment trends until the dynamic geometric envelope error converges within the preset error threshold. In step 806, the dynamic geometric envelope error refers to the multi-dimensional difference (e.g., position, curvature, and area errors) between the actual gripper deformation profile calculated in step 805 and the geometric profile template. The preset error threshold refers to the allowable error range set based on operational safety and accuracy requirements (e.g., position error ≤ 1 mm, curvature error ≤ 5%, area error ≥ 90%). The error distribution characteristics refer to the spatial distribution pattern of the dynamic geometric envelope error (e.g., local high error areas are concentrated in the gripper bend section, and the error direction deviates to the northwest). The historical adjustment trend refers to the stored gripper deformation adjustment records (e.g., curvature radius adjustment amount, pressure correction effect), which are used to predict the optimal direction and magnitude of the current adjustment. In an embodiment of the present application, a closed-loop feedback adjustment mechanism is triggered by determining whether the dynamic geometric envelope error value exceeds a preset error threshold: if the error exceeds the limit, the system dynamically and iteratively adjusts the curvature radius of the jaw deformation profile based on the spatial characteristics of the error distribution (such as the high error area is concentrated in the jaw bending section or the contact edge) and the historical adjustment trend (such as the relationship between the past curvature correction amount and the pressure response). For example, for curvature-dominated errors, the jaw bending curvature is gradually fine-tuned to match the target geometric template; for position offset errors, the contact surface inclination angle is optimized in combination with the elastic modulus of the jaw material, and the error attenuation trend after each adjustment is recorded until the multi-dimensional error values ​​(position, curvature, contact area) converge to the preset threshold, and finally a stable jaw deformation profile that fits the target surface with high precision is generated to ensure that sealing or grasping operations in radioactive environments meet the dual requirements of safety and geometric adaptation.

[0066] The following is a specific example: In a museum artifact restoration scenario, a robot grasped an ancient pottery jar with a fragile relief. When the gripper contacted the relief, high-frequency vibration noise was removed by wavelet filtering, extracting the true pressure peak at the relief protrusions. A hash map was used to locate the "relief-fragile zone" threshold (with an upper pressure limit of 20N) to prevent damage to the artifact caused by the gripping force. When the pressure at the protrusion reached 25N, the weighted difference triggered an anomaly flag, and the reinforcement learning model initiated emergency adjustments. Combining the silicone flexibility parameters with historical restoration data, the curvature radius was increased by 0.3cm to release pressure, and the effect of this adjustment was recorded. The geometric error (5% deviation in relief curvature) and mechanical error (8% uneven pressure distribution) were calculated, and a weighted composite error value of 0.62 was generated. After three iterations, the simulated annealing algorithm resolved local oscillations and ultimately adjusted the curvature to an error of 0.8% (below the 1% threshold), allowing the gripper to gently fit the relief without damage. In summary, steps 801-806 achieve high-precision dynamic matching between the gripper deformation profile and the complex surface through adaptive filtering, reinforcement learning adjustment and multi-dimensional error fusion.

[0067] To address radiation leakage and path conflicts caused by uneven expansion of the sealing layer during radioactive material capture, a dynamic sealing method based on coordinated control of radiation gradient and sealing rate has been developed. This method achieves complete coverage of high-risk areas with the sealing layer and dynamic avoidance coordination of the gripper path, ensuring zero radiation leakage and operational safety. In some embodiments, in step 105, when the gripper unit reaches the target gripping position, regional heating of the heat-shrinkable material is activated based on the radiation intensity mutation boundary coordinates of the gradient distribution topology, generating a sealing isolation layer that expands synchronously with the dynamic geometric envelope. Dynamically reconstructing the movement path control parameters of the mechanical gripper array by mapping the spatial difference matrix with the thickness growth rate of the sealing isolation layer comprises the following steps: 901: Determining the heating region of the heat-shrinkable material based on the radiation intensity mutation boundary coordinates and generating a heating control parameter matrix. In step 901, the radiation intensity mutation boundary coordinates refer to the coordinate set of the region where the radiation intensity on the surface of the radioactive material significantly changes, typically determined through gradient analysis or threshold segmentation. Heat-shrinkable material refers to a polymer material (such as polyolefin heat-shrink tubing) that undergoes directionally shrinking upon heating and is used to seal or wrap the surface of the radioactive material to prevent leakage and spread. The heating area refers to the local area that needs to be heated, which is delineated based on the boundary of the sudden change in radiation intensity, ensuring that the heat shrink material accurately covers the high-radiation risk area. The heating control parameter matrix refers to a quantitative parameter matrix that describes the temperature, power, and duration of each heating area, and is used to directionally control the deformation of the material. In the embodiment of the present application, the U-Net image segmentation algorithm is used to extract the radiation sudden change boundary coordinate clusters from the gradient distribution topology map, and the adjacent coordinate points are merged into continuous heating areas through density clustering (such as the DBSCAN algorithm); the heat diffusion rate of each area is calculated using a heat conduction finite element model, and the heating control parameter matrix is ​​generated in combination with the thermal response characteristics of the material (such as the thermal shrinkage coefficient). For example, the damaged edge of the nuclear waste container is divided into a high-priority heating zone due to the sudden change in the radiation gradient, and high temperature (such as 300°C) and long duration (such as 10 seconds) parameters are assigned to ensure rapid activation of the sealing material.

[0068] 902. Calculate the thickness growth rate distribution function of the sealing barrier layer based on the heating control parameter matrix, and generate a set of thickness compensation coefficients synchronized with the curvature changes of the dynamic geometric envelope. In step 902, the heating control parameter matrix refers to the matrix generated in step 901, which contains the temperature, duration, and power parameters for each heating zone. The sealing barrier layer thickness growth rate distribution function is a mathematical model that describes the thickness growth pattern of a heat-shrinkable material over time under different heating parameters. The thickness compensation coefficient set is a set of parameters that dynamically adjust the thickness growth rate based on curvature changes. In this embodiment of the present application, a nonlinear mapping relationship between heating parameters (temperature, power) and the sealing layer thickness growth rate is established using a finite element simulation model. After inputting the heating control parameter matrix, the expansion rate distribution of each zone is predicted. Simultaneously, curvature-thickness compensation coefficients are generated by combining real-time curvature data of the gripper deformation profile (e.g., laser scanning point cloud). For example, when the gripper grasps a concave area (curvature radius <5 cm), the compensation coefficient is set to 1.5 to increase the thickness of the sealing layer; for convex areas (curvature radius >10 cm), the coefficient is set to 0.8 to prevent material accumulation. 903. Use an infrared thermal imaging array to monitor the thickness distribution of the sealing isolation layer in real time, and establish a dynamic coupling equation between the thickness growth rate distribution function and the mechanical fixture array movement path control parameters. In step 903, the infrared thermal imaging array refers to a monitoring system composed of multiple groups of infrared sensors that captures differences in thermal radiation from the object surface to invert the material thickness distribution in real time. The sealing isolation layer thickness distribution refers to the set of thickness values ​​of the sealing layer formed by the heat-shrinkable material on the target surface at different locations, reflecting the uniformity of material coverage. The mechanical fixture array movement path control parameters are a set of instructions for controlling the fixture movement direction, speed, dwell time, and other actions. The dynamic coupling equation refers to a set of differential equations describing the thickness growth rate and the gripper movement path parameters (speed, direction, acceleration), used for coordinated control.

[0069] In this embodiment, a Kalman filter algorithm is used to fuse infrared thermal imaging thickness field data with finite element predictions to calibrate the thickness distribution in real time. Based on this calibrated thickness data, a dynamic coupling equation is constructed using partial differential equation modeling to determine the thickness growth rate and the gripper path parameters. For example, if the sealing layer expands too rapidly in a certain area, the equation outputs a command to reduce the gripper's movement speed to ensure uniform expansion of the sealing layer.

[0070] 904. Calculate a compensation vector for the movement path control parameters based on the dynamic coupling equations. Local corrections are performed based on the three-dimensional coordinates of the element abundance anomaly region in the spatial difference matrix to generate displacement update parameters. In step 904, the compensation vector refers to the path parameter correction calculated by the dynamic coupling equations, which is used to compensate for thickness distribution deviations. Local correction refers to adjusting the path parameters for specific coordinate points in the anomaly region, rather than a global uniform adjustment. The displacement update parameters refer to a set of corrected clamp movement path control instructions, including updated values ​​for speed, direction, and local dwell time. In this embodiment, the dynamic coupling equations are solved using a gradient descent optimization algorithm to generate a path compensation vector. Local weighted corrections are performed on the compensation vectors by locating element abundance anomalies (e.g., uranium abundance fluctuations >15%) based on the spatial difference matrix. Areas with anomalies in element abundance (e.g., uranium abundance fluctuations >15%) are assigned a high correction weight (e.g., 0.9). For example, if the damaged edge has an abnormal uranium abundance, the path offset increases by 30%, and the gripper detours to a stable abundance region to prevent tearing of the sealing layer due to material inhomogeneities.

[0071] 905. The displacement update parameter is merged with the original movement path control parameter to drive the gripper unit to perform the synchronous expansion operation of the sealing isolation layer along the corrected path. In step 905, the displacement update parameter refers to the corrected path control instruction generated in step 904, which includes the adjustment values ​​of speed, direction and local residence time. Parameter fusion refers to weighted superposition of the updated parameters and the original parameters according to the spatial position to generate the final execution path. Synchronous expansion operation refers to the coordinated movement of multiple gripper units according to the fused path to ensure that the sealing material is evenly extended to cover the target surface to avoid local tearing or accumulation. In the embodiment of the present application, a weighted superposition algorithm is used to fuse the compensation vector with the original path parameter. For example, the direction offset angle takes 70% of the compensation value to smooth the path; the fused path parameters are converted into the displacement of each gripper telescopic rod through the kinematic inverse solution model; the heating power and the gripper movement speed are adjusted synchronously. For example, when the sealing layer expansion rate increases by 20%, the gripper speed is synchronously increased by 15%, and the heating power is increased by 10%, so as to achieve dynamic coordination of deformation and sealing.

[0072] The following is a specific example: a nuclear waste disposal robot repaired a leak in a high-temperature liquid metal container. A gradient distribution topology map revealed a sudden change in radiation intensity at the container's sidewall. U-Net segmentation located the heating zone around the leak, generating a high-temperature (400°C) and short-duration (5-second) heating parameter matrix. The finite element model predicted a slow expansion rate of the sealing layer at the leak site, so the concave surface compensation coefficient was set to 2.0. After infrared thermal imaging calibration, the actual thickness growth rate increased to 1.2 times the predicted value. Dynamic coupling equations triggered the gripper to slow down to 60% at the leak site and increase the heating power to 120% to ensure uniform expansion of the sealing layer. The spatial difference matrix detected abnormal fluctuations in uranium abundance around the leak site, and the path compensation vector offset was increased by 25% to avoid the unstable abundance zone. After integrating the corrected path, the gripper moved at a low speed along the compensation path, with the heating power dynamically matched to the sealing layer expansion rate, ultimately forming a leak-free, complete sealing layer. In summary, according to steps 901-905, the uniform expansion of the sealing layer and the radiation safety coordination of the clamping path are achieved through precise positioning of the heating area, dynamic coupling of thickness and path, and abundance anomaly avoidance control.

[0073] Figure 2 A schematic diagram of the structure of a radioactive material intelligent sorting device (or system) is provided for the embodiment of the present application, as shown in FIG. Figure 2 As shown, the device includes: a generation module 21, which synchronously detects the surface element composition and radiation intensity of the radioactive material through an X-ray fluorescence spectrometer to generate a spatial difference matrix containing element abundance distribution and a gradient distribution topology map of real-time radiation values; a determination module 22, which determines the three-dimensional coordinate set, pressure threshold set and shape adaptation parameter vector of the target grasping position corresponding to the radioactive material based on the spatial difference matrix and the gradient distribution topology map; a driving module 23, which converts the shape adaptation parameter vector into a displacement control instruction set of a mechanical clamp array, wherein the displacement control instruction set includes a set of three-dimensional coordinates generated by spatial registration with the gradient distribution topology map. The telescopic rod displacement parameter drives the telescopic rod of the corresponding clamping jaw unit to move axially according to the telescopic rod displacement parameter; the adjustment module 24 compares the contact pressure data stream of the air pressure sensor with the pressure threshold set in real time during the axial movement of the telescopic rod, and adjusts the curvature radius of the clamping jaw deformation profile according to the comparison difference, so that the error range of the dynamic geometric envelope constituted by the clamping jaw deformation profile and the geometric profile template converges to within the threshold; the activation module 25 activates the regional heating of the heat shrinkable material according to the radiation intensity mutation boundary coordinates of the gradient distribution topology when the clamping jaw unit reaches the target grasping position, so as to generate a sealing isolation layer that expands synchronously with the dynamic geometric envelope.

[0074] Figure 2 The radioactive material intelligent sorting device can perform Figure 1The implementation principle and technical effects of the radioactive material intelligent sorting method described in the embodiment are not described in detail here. The specific way in which each module and unit performs operations in the radioactive material intelligent sorting device in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here. In one possible design, Figure 2 The radioactive material intelligent sorting device of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32; the storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32. The processing component 32 is used for the above Figure 1 The embodiment of the intelligent radioactive material sorting method includes a processing component 32 that may include one or more processors to execute computer instructions to complete all or part of the steps in the method described above. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the method described above. The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0075] Of course, the computing device may also include other components, such as input / output interfaces, display components, communication components, etc. The input / output interface provides an interface between the processing component and the peripheral interface module, and the above-mentioned peripheral interface module may be an output device, an input device, etc. The communication component is configured to facilitate wired or wireless communication between the computing device and other devices. Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device may refer to a cloud server, and the above-mentioned processing components, storage components, etc. may be basic server resources rented or purchased from the cloud computing platform. The embodiment of the present application also provides a computer storage medium storing a computer program, which can achieve the above-mentioned when executed by a computer. Figure 1The radioactive material intelligent sorting method of the illustrated embodiment. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0076] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of these modules may be selected to achieve the objectives of the present embodiments based on practical needs. Those skilled in the art will be able to understand and implement these embodiments without inventive effort. Through the description of the above embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a required general-purpose hardware platform, or, of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or portions thereof. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for intelligent sorting of radioactive materials, characterized in that: include: The surface elemental composition and radiation intensity of radioactive materials are simultaneously detected by X-ray fluorescence spectrometer to generate a spatial difference matrix containing element abundance distribution and a gradient distribution topology map of real-time radiation values; Determining a three-dimensional coordinate set, a pressure threshold set, and a shape adaptation parameter vector of a target grasping position corresponding to the radioactive material based on the spatial difference matrix and the gradient distribution topology map; Converting the shape adaptation parameter vector into a displacement control instruction set of a mechanical clamp array, wherein the displacement control instruction set includes telescopic rod displacement parameters generated by spatial registration of the three-dimensional coordinate set and the gradient distribution topology map, and driving the telescopic rods of the corresponding gripper units to perform axial movement according to the telescopic rod displacement parameters; During the axial movement of the telescopic rod, the contact pressure data stream of the air pressure sensor is compared with the pressure threshold set in real time, and the curvature radius of the deformation profile of the gripper is adjusted according to the comparison difference, so that the error range between the dynamic geometric envelope formed by the deformation profile of the gripper and the geometric profile template converges to within the threshold; When the gripper unit reaches the target grasping position, regional heating of the heat shrinkable material is activated according to the radiation intensity mutation boundary coordinates of the gradient distribution topology map, generating a sealing isolation layer that expands synchronously with the dynamic geometric envelope.

2. The method according to claim 1, characterized in that Determining a three-dimensional coordinate set, a pressure threshold set, and a shape adaptation parameter vector of a target grasping position corresponding to the radioactive material based on the spatial difference matrix and the gradient distribution topology map includes: Perform convolution matching on the spatial difference matrix and the radioactive material feature library to output a three-dimensional coordinate set of the target grasping position; Calculating a pressure threshold set based on the distribution of extreme points of radiation intensity of the gradient distribution topology map; A shape adaptation parameter vector is generated according to the geometric contour template pre-stored in the radioactive material feature library.

3. The method according to claim 2, characterized in that The convolution matching of the spatial difference matrix and the radioactive material feature library to output a three-dimensional coordinate set of the target grasping position includes: Segmenting the spatial difference matrix into regions according to reference values ​​of element abundance distribution in a radioactive material feature library, and extracting a feature vector of each segmented region; Comparing the feature vector of each segmented region with the standard feature vector of the material category of the corresponding radioactive material in the radioactive material feature library layer by layer to generate a feature similarity distribution map; Identifying a peak area that meets a preset matching condition in the feature similarity distribution map, and extracting the geometric center coordinates of the peak area as candidate coordinates; According to the radiation intensity attenuation gradient of the gradient distribution topology at the candidate coordinates, spatial position compensation is performed on the candidate coordinates to generate a three-dimensional coordinate set of the target grasping position.

4. The method according to claim 2, characterized in that The calculating of the pressure threshold set based on the distribution of extreme points of radiation intensity of the gradient distribution topology graph includes: Marking extreme points whose radiation intensity exceeds the neighborhood average in the gradient distribution topology map, and constructing a spatial distribution set of extreme points; A radiation intensity attenuation model is established with each extreme point as the center, and the rate of change of the radiation intensity within a preset radius around the extreme point is calculated; assigning a dynamic pressure weight to each extreme point according to a mapping relationship between the change rate and a preset safe contact pressure; The radiation intensity of the area where the extreme point is located is weightedly fused based on the dynamic pressure weight to generate a pressure threshold set.

5. The method according to claim 2, characterized in that Generating a shape adaptation parameter vector according to a geometric contour template pre-stored in the radioactive material feature library includes: Extracting a set of curvature feature points of the radioactive material surface from a geometric contour template pre-stored in the radioactive material feature library, and constructing a contour segmentation function; Calculating curvature adaptation parameters adapted to the contact surface of the gripper unit in the mechanical fixture array according to the spatial relative position relationship between the target gripping position in the three-dimensional coordinate set and the contour piecewise function; The curvature adaptation parameter and the radiation intensity gradient value at the corresponding position in the gradient distribution topology map are normalized to generate a shape adaptation parameter vector.

6. The method according to claim 1, characterized in that Convert the shape adaptation parameter vector into a displacement control instruction set for the mechanical fixture array, including: Decoupling the shape adaptation parameter vector to obtain geometric constraint parameters and radiation gradient constraint parameters corresponding to the axial displacement of the gripper unit in the mechanical fixture array; Performing orthogonal projection on the geometric constraint parameters and the spatial pose information in the three-dimensional coordinate set to generate an initial axial displacement component; Extracting the radiation intensity gradient direction vector corresponding to the three-dimensional coordinate set from the gradient distribution topology map, and establishing a displacement correction correlation factor between the gradient direction and the axial direction of the gripper; Dynamically compensating the initial axial displacement component according to the displacement correction correlation factor to generate an adaptive displacement parameter set including a displacement direction deviation correction amount; The adaptive displacement parameter set and the radiation gradient constraint parameter are coupled and decomposed according to the kinematic model of the gripper unit to generate the axial displacement parameters of each telescopic rod in the displacement control instruction set; Establishing a radiation intensity attenuation field model centered at the target grasping position in the three-dimensional coordinate set, and calculating the radiation intensity attenuation rate of each axial path; generating a dynamic scaling factor of the axial displacement according to the difference between the decay rate and a preset safety threshold; Performing a point-by-point multiplication operation on the axial displacement parameter and the dynamic scaling coefficient to obtain a telescopic rod displacement parameter set adapted to the radiation environment; Based on the correlation of the radiation intensity gradients of adjacent coordinate points in the gradient distribution topology, a set of displacement coordination constraint equations between the gripper units is constructed, and the final displacement parameters of the telescopic rod are obtained by solving them.

7. The method according to claim 1, characterized in that Also includes: The moving path control parameters of the mechanical fixture array are dynamically reconstructed by mapping the spatial difference matrix with the thickness growth rate of the sealing isolation layer. The dynamic reconstruction of the movement path control parameters of the mechanical fixture array by associating the spatial difference matrix with the thickness growth rate of the sealing isolation layer includes: Extracting the spatial weight value of the element abundance distribution in the spatial difference matrix, and establishing a dynamic response relationship model between the element abundance distribution at each coordinate point and the thickness growth rate of the sealing isolation layer; Acquire in real time the thickness growth rate of the sealing isolation layer in each area of ​​the dynamic geometric envelope surface, and calculate the path adjustment priority coefficient of the corresponding area according to the dynamic response relationship model; generating a movement path correction vector of each clamping unit of the mechanical clamp array based on the gradient change trend of the thickness growth rate in the three-dimensional space; Performing a spatial convolution operation on the path correction vector and the element abundance weight value in the spatial difference matrix to output a real-time offset parameter set of the moving path; A radiation intensity diffusion model centered on the current gripper position is established in the gradient distribution topology map, and the movement path control parameters of the mechanical fixture array are dynamically reconstructed according to the superposition result of the real-time offset parameter set and the diffusion model.

8. The method according to claim 1, characterized in that During the axial movement of the telescopic rod, the contact pressure data stream of the air pressure sensor is compared with the pressure threshold set in real time, and the curvature radius of the gripper deformation profile is adjusted according to the comparison difference, so that the error range between the dynamic geometric envelope formed by the gripper deformation profile and the geometric profile template converges to within the threshold, including: A real-time comparison channel between the contact pressure data stream and the pressure threshold set is established. By constructing a nonlinear mapping relationship between the curvature radius of the gripper deformation profile and the contact pressure difference, an error compensation function containing dynamic geometric envelope deformation parameters is generated. A deformation control model of the gripper unit is established based on the error compensation function, and a deformation control parameter matrix including a three-dimensional spatial curvature gradient is generated by iteratively calculating the curvature radius adjustment amount of each node of the dynamic geometric envelope surface; The deformation control parameter matrix is ​​converted into a real-time control signal stream of the gripper drive unit, and the real-time deformation data of the geometric envelope is collected through a distributed strain sensor array embedded in the gripper deformation layer to construct an error field distribution model between the deformation data and the geometric contour template; A multi-objective optimization function based on the radiation intensity gradient weight is established in the error field distribution model. By dynamically adjusting the curvature radius compensation coefficient of each node of the gripper unit, the local curvature deviation value of the dynamic geometric envelope decays exponentially along the radiation intensity gradient direction until the spatial integral value of the global error field reaches the preset convergence threshold.

9. The method according to claim 1, characterized in that When the gripper unit reaches the target grasping position, regional heating of the heat shrinkable material is activated according to the boundary coordinates of the radiation intensity mutation of the gradient distribution topology map, generating a sealing isolation layer that expands synchronously with the dynamic geometric envelope. The movement path control parameters of the mechanical fixture array are dynamically reconstructed through the correlation mapping between the spatial difference matrix and the thickness growth rate of the sealing isolation layer, including: Determine the heating area of ​​the heat shrinkable material based on the coordinates of the boundary of the radiation intensity mutation and generate a heating control parameter matrix; Calculating a thickness growth rate distribution function of the sealing isolation layer according to the heating control parameter matrix, and generating a set of thickness compensation coefficients synchronized with the change of the dynamic geometric envelope curvature; The thickness distribution of the sealing isolation layer is monitored in real time by an infrared thermal imaging array, and a dynamic coupling equation is established between the thickness growth rate distribution function and the control parameters of the movement path of the mechanical fixture array. Calculate the compensation vector of the movement path control parameter based on the dynamic coupling equation, perform local correction based on the three-dimensional coordinates of the element abundance abnormality area in the spatial difference matrix, and generate the displacement update parameter; The displacement update parameters are integrated with the original movement path control parameters to drive the gripper unit to perform a synchronous expansion operation of the sealing isolation layer along the revised path.

10. An intelligent sorting system for radioactive materials, characterized in that: include: The generation module uses an X-ray fluorescence spectrometer to simultaneously detect the surface element composition and radiation intensity of radioactive materials, generating a spatial difference matrix containing element abundance distribution and a gradient distribution topology map of real-time radiation values; a determination module, which determines a three-dimensional coordinate set, a pressure threshold set, and a shape adaptation parameter vector of a target grasping position corresponding to the radioactive material based on the spatial difference matrix and the gradient distribution topology map; a driving module, configured to convert the shape adaptation parameter vector into a displacement control instruction set for a mechanical clamp array, wherein the displacement control instruction set includes telescopic rod displacement parameters generated by spatial registration of the three-dimensional coordinate set with the gradient distribution topology map, and drive the telescopic rods of the corresponding gripper units to perform axial movement according to the telescopic rod displacement parameters; a convergence module, which compares the contact pressure data stream of the air pressure sensor with the pressure threshold set in real time during the axial movement of the telescopic rod, and adjusts the curvature radius of the gripper deformation profile according to the comparison difference, so that the error range between the dynamic geometric envelope formed by the gripper deformation profile and the geometric profile template converges to within the threshold; The activation module activates regional heating of the heat shrinkable material according to the radiation intensity mutation boundary coordinates of the gradient distribution topology when the gripper unit reaches the target grasping position, thereby generating a sealing isolation layer that expands synchronously with the dynamic geometric envelope.

Images