Robot arm grabbing control method and system based on multi-sensor data analysis

Abstract

The invention relates to the technical field of robots, in particular to a robot arm grabbing control method and system based on multi-sensor data analysis, and the method comprises the steps: obtaining the spatial position and morphological feature data of a grabbing area and surrounding objects, and the local contact state data of a target object; the internal stress distribution and the local deformation state of the flexible object in the grabbing process are predicted in real time; generating a clamping control instruction based on a prediction result, and dynamically planning a motion path and a grabbing posture in a grabbing process by combining adjacent object disturbance modeling and disturbance risk quantification to form an executable grabbing track and grabbing posture control instruction; and meanwhile, a microcosmic contact state model is constructed, the grabbing process is dynamically adjusted in combination with clamping control and a grabbing track, and real-time updating of local grabbing force and action is achieved. Through multi-sensing data fusion and dynamic modeling, the local stress change of the flexible object can be effectively sensed, and damage and sliding are inhibited.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a robotic arm grasping control method and system based on multi-sensor data analysis. Background Technology

[0002] When robotic arms grasp flexible objects, they typically rely on overall force feedback or visual information to obtain the object's state. By setting a fixed gripping force or planning the grasping path based on external image information, the object can be transported and grasped. Existing solutions mainly focus on overall force and pose control. Some solutions use simple tactile sensors to obtain local contact information, but these are usually only used to determine whether the object is being grasped. They lack real-time prediction and feedforward control of the internal stress distribution and local deformation state of the object, making it difficult to achieve fine-grained control of the grasping process.

[0003] These existing methods suffer from localized stress concentrations or deformations that are not detected and adjusted in real time when grasping flexible objects, leading to object breakage or slippage and causing unstable clamping. Summary of the Invention

[0004] To overcome the above deficiencies, this invention provides a robotic arm grasping control method and system based on multi-sensor data analysis, aiming to improve the problem of unstable gripping caused by local stress concentration or deformation not being sensed and adjusted in real time when grasping flexible objects, which leads to object damage or slippage.

[0005] In a first aspect, the present invention provides the following technical solution: a robotic arm grasping control method based on multi-sensor data analysis, comprising the following steps:

[0006] S1. Acquire spatial location data and morphological feature data of the grasping area and surrounding objects, as well as local contact state data of the target object.

[0007] S2. Based on local contact state data, use the nonlinear dynamics model of flexible objects to predict the internal stress distribution data and local deformation state data of the target object during the grasping process.

[0008] S3. Generate clamping control commands based on internal stress distribution data and local deformation state data;

[0009] S4. Based on spatial location data and morphological feature data, model and analyze the disturbance state of nearby objects during the grasping process, and generate corresponding disturbance risk quantification data.

[0010] S5. Based on the disturbance risk quantification data, plan the motion path and grasping posture during the grasping process, and generate grasping trajectory and grasping posture control commands.

[0011] S6. Construct a micro-contact state model based on local contact state data, and combine it with clamping control commands, grasping trajectory and grasping posture control commands to dynamically adjust the grasping process and generate update control commands for local grasping force and grasping action.

[0012] By adopting the above technical solution, it is possible to predict the internal stress distribution and local deformation state of the target object in real time based on local contact state data and nonlinear dynamic model of flexible object during the grasping process. The clamping control command is generated according to the prediction results for feedforward adjustment, thereby dynamically adjusting the clamping force of the gripper and the grasping action, realizing fine control of the grasping process, and solving the problem of object breakage or slippage and unstable clamping caused by the failure to detect local stress concentration or deformation in real time in the existing technology.

[0013] Preferably, acquiring local contact state data includes:

[0014] Acquire information about the contact area between the target object and the grasping part, and determine the spatial distribution of the contact area during the grasping process;

[0015] The changes in the contact area during the grasping process are tracked, and the spatial position changes of the contact area during the grasping process are recorded to form time-series status data;

[0016] Based on the spatial distribution and temporal state data of the contact area, local contact state data is generated to characterize the local contact behavior of the target object.

[0017] Preferably, the nonlinear dynamic model of the flexible object includes:

[0018] Based on local contact state data, the contact boundary conditions of the target object during the grasping process are determined;

[0019] By incorporating contact boundary conditions into the nonlinear dynamics model of flexible objects, the force state of the target object during the grasping process is constrained and modeled.

[0020] In the nonlinear dynamics model, the dynamic response of the target object during the grasping process is calculated to obtain data on the change of the object's internal state over time.

[0021] Preferably, the internal stress distribution data and local deformation state data include:

[0022] Based on the calculation results of the nonlinear dynamic model of flexible objects, the stress state inside the target object is spatially discretized to form multiple stress calculation regions.

[0023] The stress changes in each stress calculation area are calculated to obtain the internal stress distribution data of the target object during the grasping process;

[0024] Based on the spatial relationship of the stress calculation area, the deformation of each area is calculated to obtain the local deformation state data of the target object during the grasping process.

[0025] Preferably, the generation of clamping control commands includes:

[0026] Based on the internal stress distribution data, the stress constraint conditions of the target object during the grasping process are determined;

[0027] Based on local deformation state data, the deformation constraint conditions of the target object in different contact areas are determined.

[0028] Based on stress and deformation constraints, the clamping parameters corresponding to the contact area with the target object are calculated, and clamping control commands are generated according to the clamping parameters.

[0029] Preferably, the modeling analysis includes:

[0030] Based on the spatial location data and morphological feature data of the grasped area and surrounding objects, determine the spatial relationship between neighboring objects;

[0031] Based on spatial correlation, a disturbance propagation model is established between the grasping action and nearby objects;

[0032] In the disturbance propagation relationship model, the state changes of neighboring objects caused by the grasping process are calculated to obtain the disturbance state description results of neighboring objects during the grasping process.

[0033] Preferably, the disturbance risk quantification data includes:

[0034] Based on the perturbation state description results of neighboring objects, perturbation feature parameters that characterize the degree of state change of neighboring objects are extracted.

[0035] The perturbation characteristic parameters are normalized to form a set of comparable perturbation indices;

[0036] Based on the set of disturbance indicators, the corresponding quantitative data of disturbance risk is calculated.

[0037] Preferably, the planning of the motion path and grasping posture during the grasping process includes:

[0038] Based on the disturbance risk quantification data and the disturbance state description results of nearby objects, a set of feasible paths is established within the capture area;

[0039] From the set of feasible paths, the grasping trajectory is filtered according to the spatial constraints of nearby objects and the grasping posture constraints of the target object;

[0040] Based on the selected grasping trajectory, the corresponding grasping posture parameters are determined, and a set of grasping trajectory and grasping posture control instructions is generated;

[0041] The set of grasping trajectory and grasping posture control instructions is calculated based on spatial constraints to generate executable grasping trajectory and grasping posture control instructions.

[0042] Preferably, the construction of the microscopic contact state model includes:

[0043] The spatial location data and the change process data of the contact area of ​​the target object are discretized to form a set of micro-contact units;

[0044] A corresponding mechanical property model is established for each micro-contact unit, including the characterization of local stress, deformation and contact boundary relationship;

[0045] The micro-contact unit and its mechanical property model are combined to form a micro-contact state model, which describes the local contact state distribution of the target object during the grasping process.

[0046] During the grasping process, the micro-contact state model is dynamically updated, and the state of the micro-contact unit is adjusted according to the gripping control command, grasping trajectory and grasping posture control command to generate the corresponding local grasping force and grasping action control command.

[0047] Secondly, the present invention provides the following technical solution: a robotic arm grasping control system based on multi-sensor data analysis, the system comprising:

[0048] The data acquisition module is used to acquire spatial location data and morphological feature data of the grasping area and surrounding objects, as well as local contact state data of the target object.

[0049] The dynamic modeling module is used to predict the internal stress distribution and local deformation state data of the target object during the grasping process based on local contact state data and using a nonlinear dynamic model of flexible objects.

[0050] The clamping control generation module is used to generate clamping control commands based on internal stress distribution data and local deformation state data.

[0051] The disturbance modeling module is used to model and analyze the disturbance state of nearby objects during the grasping process based on spatial location data and morphological feature data, and generate corresponding disturbance risk quantification data.

[0052] The grasping planning module is used to plan the motion path and grasping posture during the grasping process based on the disturbance risk quantification data, and generate grasping trajectory and grasping posture control commands.

[0053] The micro-contact state module is used to construct a micro-contact state model based on local contact state data, and to dynamically adjust the gripping process by combining gripping control commands, gripping trajectory and gripping posture control commands, and to generate control commands for local gripping force and gripping action.

[0054] The present invention has the following beneficial effects:

[0055] 1. In this invention, by acquiring local contact state data of the target object and predicting the internal stress distribution and local deformation state of the object based on the nonlinear dynamic model of the flexible object, and then generating clamping control commands according to the prediction results, the feedforward control of the gripper during the grasping process is realized, which solves the problems of damage, slippage and unstable clamping caused by local stress concentration or deformation of the flexible object at the moment of grasping.

[0056] 2. In this invention, by establishing spatial relationships between neighboring objects based on the spatial location and morphological features of the grasping area and surrounding objects, and constructing a disturbance propagation relationship model, the grasping trajectory and grasping posture are dynamically optimized using the disturbance state description results and disturbance risk quantification data. This achieves path avoidance and posture adjustment in complex stacking environments during the grasping process, and solves the problem of grasping failure or stacking damage caused by disturbances of neighboring objects.

[0057] 3. In this invention, by discretizing the contact area of ​​the target object into micro-contact units and establishing a mechanical property model, a micro-contact state model is formed. By combining the clamping control command, the grasping trajectory and the grasping posture control command, the state of the micro-contact units is dynamically adjusted, thereby achieving precise control of local grasping force and motion. This solves the problem that traditional overall force feedback cannot suppress micro-slippage and cope with the instability of grasping caused by the heterogeneity of the surface of flexible objects. Attached Figure Description

[0058] Figure 1 This is a flowchart of a robotic arm grasping control method based on multi-sensor data analysis proposed in this invention;

[0059] Figure 2 This is an architecture diagram of a robotic arm grasping control system based on multi-sensor data analysis proposed in this invention. Detailed Implementation

[0060] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Example 1:

[0062] In a first embodiment of the present invention, the present invention provides a robotic arm grasping control method based on multi-sensor data analysis, such as... Figure 1 As shown, it includes the following steps:

[0063] S1. Acquire spatial location data and morphological feature data of the grasping area and surrounding objects, as well as local contact state data of the target object.

[0064] Furthermore, acquiring local contact state data includes:

[0065] Acquire information about the contact area between the target object and the grasping part, and determine the spatial distribution of the contact area during the grasping process;

[0066] The changes in the contact area during the grasping process are tracked, and the spatial position changes of the contact area during the grasping process are recorded to form time-series status data;

[0067] Based on the spatial distribution and temporal state data of the contact area, local contact state data is generated to characterize the local contact behavior of the target object.

[0068] Specifically, the grasping control system first acquires spatial position data and morphological feature data of the grasping area and surrounding objects to form a three-dimensional spatial model of the grasping area and its surrounding environment. The spatial position data may include the position coordinates (x, y, z) of the object in the three-dimensional coordinate system and its offset relative to the grasping reference system. The morphological feature data may include information such as the object's surface contour, geometric dimensions, and curvature distribution. At the same time, the system acquires the contact area information between the target object and the grasping part, and identifies the spatial distribution state of the contact area during the grasping process through scanning, imaging, or other sensing means. The spatial distribution state can be represented as the center position coordinates, area size, and shape description parameters of the contact area, such as using a grid to represent the discretized set of points of the contact surface shape.

[0069] In the specific implementation process, the changes in the contact area during the grasping process are tracked. The system continuously collects spatial location data of the contact area at different time points to form temporal state data; the temporal state data can be represented as:

[0070] T = {t1, t2, ..., t} n}, C={c1,c2,…,c m},c i =(x i ,y i ,z i );

[0071] Where c iThe position of the contact unit in three-dimensional space is represented by n, where n is the total number of sampling time points and m is the total number of contact units. By recording the spatial position changes of the contact area, the dynamic evolution process of the contact area can be reflected. During the tracking process, optical sensors, force sensors, or flexible tactile sensors can be used to acquire the displacement and deformation information of the contact unit over continuous time. The collected data is preprocessed, including noise reduction, coordinate correction, and time synchronization, to ensure the accuracy and consistency of the time-series state data.

[0072] Based on the spatial distribution and temporal state data of the contact area, local contact state data is generated to characterize the local contact behavior of the target object. This local contact state data includes contact pressure distribution, local friction coefficient, and contact deformation information. The local pressure and friction force of the contact element can be calculated using the following formula:

[0073]

[0074] Where P i (t) represents the local pressure of contact element i at time t, F i (t) represents the normal force of contact element i at time t, A i (t) represents the corresponding contact area, F fi (t) represents the frictional force, μ i (t) represents the friction coefficient of contact unit i at time t; the system integrates the time-series data of each contact unit to form a microscopic contact behavior model:

[0075] C local (t)={P i (t),F fi (t),δ i (t)};

[0076] Where δ i (t) represents the deformation of contact unit i at time t, which fully describes the spatial and temporal distribution characteristics of the local contact state of the target object during the grasping process.

[0077] Local contact state data can be used for subsequent dynamic modeling and gripping control strategy generation. By real-time monitoring and updating of pressure, friction, and deformation in the contact area, the system can establish contact constraints between the target object and the gripping part, providing initial boundary conditions for the nonlinear dynamic model of the flexible object. The system can iteratively calculate the model based on the real-time collected local contact state data, realizing dynamic prediction and adjustment of the contact state during the gripping process. In application, this local contact state data not only provides a basis for gripping force calculation, but can also be used to assess the disturbance risk of nearby objects and provide input information for gripping path planning.

[0078] S2. Based on local contact state data, use the nonlinear dynamics model of flexible objects to predict the internal stress distribution data and local deformation state data of the target object during the grasping process.

[0079] Furthermore, the nonlinear dynamics model of flexible objects includes:

[0080] Based on local contact state data, the contact boundary conditions of the target object during the grasping process are determined;

[0081] By incorporating contact boundary conditions into the nonlinear dynamics model of flexible objects, the force state of the target object during the grasping process is constrained and modeled.

[0082] In the nonlinear dynamics model, the dynamic response of the target object during the grasping process is calculated to obtain data on the change of the object's internal state over time.

[0083] Furthermore, the internal stress distribution data and local deformation state data include:

[0084] Based on the calculation results of the nonlinear dynamic model of flexible objects, the stress state inside the target object is spatially discretized to form multiple stress calculation regions.

[0085] The stress changes in each stress calculation area are calculated to obtain the internal stress distribution data of the target object during the grasping process;

[0086] Based on the spatial relationship of the stress calculation area, the deformation of each area is calculated to obtain the local deformation state data of the target object during the grasping process.

[0087] Specifically, based on the acquired local contact state data, the contact boundary conditions of the target object during the grasping process are first determined. The contact boundary conditions include parameters such as the position of the contact surface of the target object, the contact area, the contact pressure, and the friction coefficient. These boundary conditions are obtained in real time by sensors or calculated according to the aforementioned local contact state model, and are used as input to the nonlinear dynamics model of the flexible object to constrain the force state of the target object during the grasping process.

[0088] In the specific implementation process, contact boundary conditions are introduced into the nonlinear dynamics model of the flexible object to establish the force constraint relationship of the target object during the grasping process. This model can use the finite element method or other nonlinear dynamics calculation methods to simulate the force distribution of the object's overall and local structures under the grasping action. The motion of each element in the model follows nonlinear dynamic equations:

[0089]

[0090] Where M is the mass matrix, C is the damping matrix, K(u(t)) is the stiffness matrix varying with displacement, and u(t) is the nodal displacement vector. and Here, denoted as velocity and acceleration vectors, and F(t) as external force vector, including grasping force and contact reaction force; by solving this equation, the dynamic response data of the target object during the grasping process can be obtained, including nodal displacement, velocity and acceleration information.

[0091] Based on the calculation results of the nonlinear dynamic model, the internal force state of the target object is spatially discretized to form multiple stress calculation regions. Each stress calculation region can correspond to a finite element on the surface or inside of the object. The stress change is calculated for each stress calculation region to obtain the internal stress distribution data of the target object during the grasping process, and the internal stress σ i (t) can be expressed as:

[0092]

[0093] Where, σ i (t) represents the stress in region i at time t, F i (t) represents the total force acting on this region, A i The area or volume of this region; based on the spatial relationship of the stress calculation region, the deformation ε of each region is further calculated. i (t), for example, using the strain formula:

[0094]

[0095] Where, ΔL i (t) represents the length change of region i over time t, L 0i Given the initial length, this calculation can be used to obtain the local deformation state data of the target object during the grasping process;

[0096] Internal stress distribution data and local deformation state data are used to generate subsequent clamping control strategies and optimize gripping actions. The system can dynamically update clamping parameters according to the state of each stress calculation area, realizing real-time prediction and control of mechanical behavior during the gripping of flexible objects. These data can also be used for disturbance analysis of adjacent objects and construction of micro-contact state models, providing accurate input for gripping path and attitude planning.

[0097] S3. Generate clamping control commands based on internal stress distribution data and local deformation state data;

[0098] Furthermore, generating clamping control commands includes:

[0099] Based on the internal stress distribution data, the stress constraint conditions of the target object during the grasping process are determined;

[0100] Based on local deformation state data, the deformation constraint conditions of the target object in different contact areas are determined.

[0101] Based on stress and deformation constraints, the clamping parameters corresponding to the contact area with the target object are calculated, and clamping control commands are generated according to the clamping parameters.

[0102] Specifically, it receives internal stress distribution data and local deformation state data calculated by a nonlinear dynamic model. These data reflect the stress and corresponding deformation of the target object in different regions during the grasping process, including the internal stress distribution data σ. i (t) represents the stress in region i at time t, and the local deformation state data ε. i (t) represents the deformation of region i at time t; the system performs constraint analysis on the stress and deformation of the target object during the grasping process based on these data to determine the safe grasping range and allowable deformation range of each contact area, ensuring that the clamping action will not cause damage or instability to the object.

[0103] In the specific implementation process, the system is based on the stress constraint condition σ of each contact area. i (t)≤σ max,i and deformation constraint ε i (t)≤ε max,i Calculate the clamping force F required for each contact area. c,i (t), the calculation formula can be expressed as:

[0104] F c,i (t)=k i ·(σ max,i -σ i (t))+α i ·(ε max,i -ε i (t));

[0105] Where, k i With α i These are the adjustment coefficients used for corresponding stress and deformation constraints, F, respectively. c,i (t) represents the clamping force acting on contact region i at time t; through the clamping force F on all contact regions i c,i (t) By performing comprehensive calculations, the overall clamping strategy can be obtained, which can be used to generate specific clamping control instructions;

[0106] The clamping control commands include parameters such as clamping force magnitude, clamper position, clamper posture, and action time. The system integrates the clamping force and deformation constraints of each contact area and generates a continuous and executable command sequence through a control algorithm, so that the clamper acts on the target object according to the predetermined mechanical constraints during the gripping process. This command sequence can be adjusted in real time, and combined with the dynamic change data sensed during the gripping process, the clamping force and motion trajectory are dynamically modified to achieve stable gripping and precise control of the target object.

[0107] The generated gripping control commands are not only used to execute the gripping actions of the robotic arm, but can also be linked with the microscopic contact state model and the gripping trajectory planning module to achieve global optimization of the gripping action. By collecting the stress and deformation state of each contact area in real time, the system can continuously update the gripping control commands to ensure that the target object maintains structural integrity and stability during the gripping process, and provide accurate data support for subsequent gripping path adjustment and disturbance analysis.

[0108] S4. Based on spatial location data and morphological feature data, model and analyze the disturbance state of nearby objects during the grasping process, and generate corresponding disturbance risk quantification data.

[0109] Further modeling and analysis include:

[0110] Based on the spatial location data and morphological feature data of the grasped area and surrounding objects, determine the spatial relationship between neighboring objects;

[0111] Based on spatial correlation, a disturbance propagation model is established between the grasping action and nearby objects;

[0112] In the disturbance propagation relationship model, the state changes of neighboring objects caused by the grasping process are calculated to obtain the disturbance state description results of neighboring objects during the grasping process.

[0113] Furthermore, the quantitative data on disturbance risk includes:

[0114] Based on the perturbation state description results of neighboring objects, perturbation feature parameters that characterize the degree of state change of neighboring objects are extracted.

[0115] The perturbation characteristic parameters are normalized to form a set of comparable perturbation indices;

[0116] Based on the set of disturbance indicators, the corresponding quantitative data of disturbance risk is calculated.

[0117] Specifically, collect spatial location data P of the grasping area and surrounding objects. j =(x j ,y j ,z j ) and morphological feature data M jWhere j represents the index of a neighboring object, spatial location data represents the coordinates of each neighboring object in three-dimensional space, and morphological feature data describes the object's size, geometry, centroid location, and surface features; the system establishes the spatial relationship R between neighboring objects based on this data. jk , where R jk Describe the relative position, distance, and potential collision probability between object j and object k, which is used to analyze the possible impact of the grasping action on nearby objects;

[0118] In the specific implementation process, the system is further based on the spatial correlation R jk Construct a perturbation propagation relationship model D j (t), this model is used to describe the force response and possible motion offset of object j during the grasping process. The perturbation propagation relationship can be expressed as:

[0119]

[0120] Where, N j F is the set of neighboring objects in space related to object j. c (k,t) represents the force acting on neighboring object k during the grasping process, and f is the disturbance transfer function used to calculate the influence of the grasping action on the state of neighboring object j. The system uses this model to calculate the state changes of each neighboring object during the grasping process, including displacement, rotation, and force changes, and obtains the disturbance state description result S of the neighboring objects. j (t), this data is used for subsequent disturbance risk analysis;

[0121] The system is based on the disturbance state description result S j (t) Extract the perturbation feature parameter E that characterizes the degree of change in the state of neighboring objects. j Examples include displacement amplitude, angular deviation, and force variation amplitude; the disturbance characteristic parameters are normalized to form a set of disturbance indices. To facilitate comparisons between different objects and different attributes; the system calculates the corresponding disturbance risk quantification data R based on the disturbance index set. j Weighted summation or other statistical methods can be used, for example:

[0122]

[0123] Among them, w i These are the weighting coefficients for different perturbation characteristics. This represents the i-th disturbance index of neighboring object j; the disturbance risk quantification data R... j The system can further guide the path planning and posture adjustment of the grasping action, and realize the quantitative assessment and control of disturbances of nearby objects in the grasping area, thereby ensuring the stability and safety of the grasping operation.

[0124] S5. Based on the disturbance risk quantification data, plan the motion path and grasping posture during the grasping process, and generate grasping trajectory and grasping posture control commands.

[0125] Furthermore, planning the motion path and grasping posture during the grasping process includes:

[0126] Based on the disturbance risk quantification data and the disturbance state description results of nearby objects, a set of feasible paths is established within the capture area;

[0127] From the set of feasible paths, the grasping trajectory is filtered according to the spatial constraints of nearby objects and the grasping posture constraints of the target object;

[0128] Based on the selected grasping trajectory, the corresponding grasping posture parameters are determined, and a set of grasping trajectory and grasping posture control instructions is generated;

[0129] The set of grasping trajectory and grasping posture control instructions is calculated based on spatial constraints to generate executable grasping trajectory and grasping posture control instructions.

[0130] Specifically, receive disturbance risk quantification data R j and the description results of the disturbance state of neighboring objects S j (t), combining the initial grasping position of the target object and the grasping target position, establish a set of feasible paths within the grasping area. Each path P i This represents the continuous motion trajectory of the grasping tool in three-dimensional space, including time, velocity, and acceleration information related to path points, used to describe the grasper's motion state along the path; feasible paths can be generated using a three-dimensional mesh-based method, dividing the grasping area into multiple small cells, each cell being assigned a disturbance risk weight w in space. cell The weights are quantified from the neighboring object disturbance risk data R. j This mapping is used so that high-risk areas can be automatically avoided.

[0131] In the specific implementation process, the system sets feasible paths. Filtering is performed based on the spatial constraint C of neighboring objects. s and target object grasping posture constraints C a The grasping trajectory is constrained and optimized; spatial constraint C s Including boundary constraints of nearby objects, minimum safe distance, and potential collision areas, grasping posture constraints C a This includes limitations on the gripping angle range, clamping direction, and robotic arm degrees of freedom; path selection can be achieved through optimization functions, such as minimizing the total disturbance risk of the gripping path.

[0132]

[0133] in Represents path P i The k-th path point on the path, P represents the disturbance risk weight of the cell containing that point. * This is the optimal grasping trajectory after filtering. After filtering, the system calculates the corresponding grasping posture parameter Θ = θ based on the spatial position of each point on the path and its relative position to neighboring objects. x ,θ y ,θ z , where θ x ,θ y ,θ z These represent the rotation angles of the gripper along the x, y, and z axes, respectively, and are used to generate the gripping trajectory and the set of gripping attitude control commands.

[0134] In the specific implementation process, the system controls the set of grasping trajectory and grasping posture instructions. Further calculations are performed to ensure that the trajectory and attitude satisfy the spatial constraint C. s Attitude constraints C a and the kinematic constraints of the robotic arm C k The joint angle φ of the gripper at each path point can be calculated using inverse kinematics models and dynamic simulations. i and joint velocity v i This generates executable grasping trajectories and grasping posture control commands, enabling the robotic arm to move along an optimized path during grasping while maintaining the stability of the target object's posture and reducing the risk of disturbance from nearby objects. The generated grasping trajectories and grasping posture control commands can be used to directly drive the gripper to perform actions, or they can be combined with subsequent control modules to achieve dynamic adjustment and closed-loop control.

[0135] S6. Construct a micro-contact state model based on local contact state data, and combine it with clamping control commands, grasping trajectory and grasping posture control commands to dynamically adjust the grasping process and generate update control commands for local grasping force and grasping action.

[0136] Furthermore, the construction of the microscopic contact state model includes:

[0137] The spatial location data and the change process data of the contact area of ​​the target object are discretized to form a set of micro-contact units;

[0138] A corresponding mechanical property model is established for each micro-contact unit, including the characterization of local stress, deformation and contact boundary relationship;

[0139] The micro-contact unit and its mechanical property model are combined to form a micro-contact state model, which describes the local contact state distribution of the target object during the grasping process.

[0140] During the grasping process, the micro-contact state model is dynamically updated. The state of the micro-contact unit is adjusted according to the gripping control command, grasping trajectory and grasping posture control command, and corresponding local grasping force and grasping action control command are generated.

[0141] Specifically, the system receives local contact state data C(x,y,z,t) of the target object. This data reflects the spatial position of the contact area between the target object and the gripper during the grasping process, as well as its state over time. The system then spatially discretizes the contact area, dividing the continuous contact surface into several micro-contact units u. i This forms a collection of micro-contact units. Each micro-contact unit corresponds to the local spatial range of the contact area of ​​the target object. The discretization method can be uniform grid division or adaptive division based on contact density to ensure complete spatial coverage and accurate representation of contact state information.

[0142] In the specific implementation process, for each micro-contact unit u i Establish a mechanical property model M i Including local stress σ i (t), Local deformation ε i (t) and contact boundary condition B i , where σ i (t) represents the force state of the element during the grasping process, ε i (t) represents the local deformation of the element, B i Define the contact constraint relationships between elements, such as force transfer and friction constraints between adjacent elements; the mechanical model can be calculated using the finite element method or a simplified nonlinear elastic model, and the formula can be expressed as:

[0143]

[0144] Where E i V is the elastic modulus of the micro-contact unit. i Let f be the element volume, and g be the force-deformation mapping function and the deformation-force mapping function, respectively, used to describe the dynamic response relationship of the micro-contact element. All micro-contact elements and their mechanical property models are combined to form a complete micro-contact state model. Used to describe the local contact state distribution and internal mechanical properties of the target object during the grasping process;

[0145] In the specific implementation process, the system uses a microscopic contact state model during the capture and execution process. Dynamic updates are performed based on the clamping control command F. c , capture trajectory P * The gripping posture control command Θ adjusts the state of the micro-contact units; the local stress and deformation of each unit are updated in real time according to the current gripping force and the gripper's motion state, and iterative calculation methods can be used, such as:

[0146]

[0147] Where Δt is the time step. and The stress and deformation rate of the micro-contact element in the current state are calculated by the dynamic model; by dynamically updating the micro-contact state model, the system generates the corresponding local gripping force F. i Update control commands with grabbing action Used to drive the robotic arm to adjust the gripping force and motion posture, so as to achieve precise grasping of the target object, while maintaining the stability of the local contact state during the grasping process, ensuring the safety and reliability of the grasping action.

[0148] In the specific implementation process, the updating of the micro-contact state model and the generation of local grasping force can be combined with closed-loop sensing feedback to monitor the state changes of the contact unit in real time. This enables the robotic arm to dynamically adjust the grasping strategy based on local contact information, ensuring that the grasping process can maintain controllable grasping under different target object shapes, flexible characteristics and surrounding environmental disturbances.

[0149] Example 2:

[0150] In complex stacked material environments, the target flexible object has multiple points of contact and potential interference with surrounding objects. Traditional grasping methods cannot perceive local stress changes and disturbances of nearby objects in real time, easily leading to object breakage or slippage, resulting in a high grasping failure rate. To solve the above problems, this invention provides a robotic arm grasping control system based on multi-sensor data analysis, the structure of which is as follows: Figure 2 As shown. The specific implementation process of this system is as follows:

[0151] The data acquisition module is used to acquire spatial location data and morphological feature data of the grasping area and surrounding objects, as well as local contact state data of the target object.

[0152] The dynamic modeling module is used to predict the internal stress distribution and local deformation state data of the target object during the grasping process based on local contact state data and using a nonlinear dynamic model of flexible objects.

[0153] The clamping control generation module is used to generate clamping control commands based on internal stress distribution data and local deformation state data.

[0154] The disturbance modeling module is used to model and analyze the disturbance state of nearby objects during the grasping process based on spatial location data and morphological feature data, and generate corresponding disturbance risk quantification data.

[0155] The grasping planning module is used to plan the motion path and grasping posture during the grasping process based on the disturbance risk quantification data, and generate grasping trajectory and grasping posture control commands.

[0156] The micro-contact state module is used to construct a micro-contact state model based on local contact state data, and to dynamically adjust the gripping process by combining gripping control commands, gripping trajectory and gripping posture control commands, and to generate control commands for local gripping force and gripping action.

[0157] Specifically, the data acquisition module obtains spatial location and morphological feature data of the grasping area and surrounding objects, while simultaneously collecting local contact state data of the target object to provide input for subsequent dynamic modeling. Based on the local contact state data, the dynamic modeling module uses a nonlinear dynamic model of flexible objects to predict the internal stress distribution and local deformation state of the target object during the grasping process. The calculation formula can be expressed as:

[0158]

[0159] Where U(t) is the displacement vector of the target object, F contact (t) represents the contact force vector, and M, C, and K are the mass matrix, damping matrix, and stiffness matrix, respectively. and These are the velocity and acceleration vectors, respectively. Based on the dynamic prediction results, the clamping control generation module calculates the clamping parameters corresponding to the contact area and generates clamping control commands to adjust the gripper force and grasping posture in real time.

[0160] During the grasping process, the disturbance modeling module establishes a model of the propagation relationship of disturbances from neighboring objects based on spatial location data and morphological feature data. It calculates the possible state changes of neighboring objects caused by the grasping action and generates disturbance risk quantification data, providing a set of feasible paths for the grasping planning module. The grasping planning module selects grasping trajectories that meet spatial and attitude constraints from the set of feasible paths based on the disturbance risk quantification data and the description results of the disturbance states of neighboring objects, and generates a corresponding set of grasping attitude control commands. The micro-contact state module discretizes the local contact state data into a set of micro-contact units and establishes a mechanical property model for each unit, including local stress, deformation, and contact boundary relationships. By dynamically updating the micro-contact state model, it realizes real-time adjustment of clamping force and grasping action, and finally generates updated control commands for local grasping force and grasping action to ensure stable grasping of flexible objects in a stacked environment. Throughout the implementation process, the system coordinates the data flow and control signals of each module to achieve real-time prediction and dynamic control of the internal stress distribution, local deformation, and disturbances of neighboring objects of the target object, effectively solving the grasping failure problem in complex stacked environments.

[0161] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A machine arm grasping control method based on multi-sensor data analysis, characterized by, The method comprises the following steps: S1, obtaining spatial position data and morphological feature data of the grabbing area and surrounding objects, and local contact state data of the target object; S2, based on the local contact state data, using a flexible object nonlinear dynamics model to predict internal stress distribution data and local deformation state data of the target object during the grabbing process; S3, generating clamping control instructions according to the internal stress distribution data and the local deformation state data; S4, based on the spatial position data and the morphological feature data, modeling and analyzing the disturbance state of the adjacent objects during the grabbing process, and generating corresponding disturbance risk quantization data; S5, according to the disturbance risk quantization data, planning the motion path and the grabbing posture in the grabbing process, and generating grabbing trajectory and grabbing posture control instructions; S6, based on the local contact state data, constructing a micro contact state model, and combining the clamping control instructions and the grabbing trajectory and grabbing posture control instructions to dynamically adjust the grabbing process, and generating updated control instructions of local grabbing force and grabbing action. 2.The machine arm grasping control method based on multi-sensor data analysis of claim 1, wherein, The local contact state data includes: obtaining contact area information between the target object and the grabbing part, and determining the spatial distribution state of the contact area in the grabbing process; tracking the change process of the contact area in the grabbing process, recording the spatial position change of the contact area in the grabbing process to form time sequence state data; based on the spatial distribution state and the time sequence state data of the contact area, generating local contact state data for representing the local contact behavior of the target object. 3.The machine arm grasping control method based on multi-sensor data analysis of claim 1, wherein, The flexible object nonlinear dynamics model includes: based on the local contact state data, determining the contact boundary condition of the target object in the grabbing process; introducing the contact boundary condition into the flexible object nonlinear dynamics model to constrain the stress state of the target object in the grabbing process; in the nonlinear dynamics model, calculating the dynamic response of the target object in the grabbing process to obtain the data of the internal state of the object changing with time.

4. The machine arm grasping control method based on multi-sensor data analysis according to claim 1, wherein, The internal stress distribution data and the local deformation state data include: based on the calculation results of the flexible object nonlinear dynamics model, spatially discretizing the stress state of the target object to form a plurality of stress calculation regions; calculating the stress change in each stress calculation region to obtain the internal stress distribution data of the target object in the grabbing process; based on the spatial position relationship of the stress calculation regions, calculating the deformation amount of each region to obtain the local deformation state data of the target object in the grabbing process.

5. The method of claim 1, wherein, The generation of clamping control instructions includes: based on the internal stress distribution data, determining the stress constraint condition of the target object in the grabbing process; based on the local deformation state data, determining the deformation constraint condition of the target object corresponding to different contact areas; based on the stress constraint condition and the deformation constraint condition, calculating the clamping parameters corresponding to the contact area of the target object, and generating the clamping control instructions according to the clamping parameters.

6. The machine arm grasping control method based on multi-sensor data analysis according to claim 1, wherein, The modeling analysis includes: based on the spatial position data and the morphological feature data of the grabbing area and the surrounding objects, determining the spatial correlation relationship between the adjacent objects; based on the spatial correlation relationship, establishing a disturbance propagation relationship model between the grabbing action and the adjacent objects; In the disturbance propagation relationship model, the state changes of the adjacent objects caused by the grabbing process are calculated to obtain a disturbance state description result of the adjacent objects in the grabbing process.

7. The method of claim 1, wherein, The disturbance risk quantization data includes: Based on the disturbance state description result of the adjacent objects, a disturbance feature parameter representing the state change degree of the adjacent objects is extracted; The disturbance feature parameter is normalized to form a disturbance index set that can be compared; Based on the disturbance index set, corresponding disturbance risk quantization data is calculated. 8.The machine arm grasping control method based on multi-sensor data analysis of claim 1, wherein, The planning of the motion path and the grabbing posture in the grabbing process includes: Based on the disturbance risk quantization data and the disturbance state description result of the adjacent objects, a feasible path set in the grabbing area is established; In the feasible path set, the grabbing trajectory is filtered according to the spatial constraints of the adjacent objects and the grabbing posture constraints of the target object; According to the filtered grabbing trajectory, corresponding grabbing posture parameters are determined to generate a grabbing trajectory and grabbing posture control instruction set; According to the spatial constraints, the grabbing trajectory and the grabbing posture control instruction set are calculated to generate an executable grabbing trajectory and grabbing posture control instruction. 9.The machine arm grasping control method based on multi-sensor data analysis of claim 1, wherein, The construction of the micro contact state model includes: Discretizing the spatial position data and the contact area change process data of the target object contact area to form a micro contact element set; For each micro contact element, a corresponding mechanical property model is established, including the representation of local stress, deformation, and contact boundary relationship; The micro contact element and its mechanical property model are combined to form a micro contact state model to describe the local contact state distribution of the target object in the grabbing process; In the grabbing process, the micro contact state model is dynamically updated, and the state of the micro contact element is adjusted according to the clamping control instruction, the grabbing trajectory, and the grabbing posture control instruction to generate corresponding local grabbing force and grabbing action control instruction.

10. A machine arm grasping control system based on multi-sensory data analysis, characterized by, A machine arm grabbing control method based on multi-sensor data analysis for any one of claims 1-9, the system comprising: A data acquisition module for acquiring spatial position data and morphological feature data of the grabbing area and surrounding objects, and local contact state data of the target object; A dynamics modeling module for predicting internal stress distribution data and local deformation state data of the target object in the grabbing process based on the local contact state data using a flexible object nonlinear dynamics model; A clamping control generation module for generating clamping control instructions based on the internal stress distribution data and the local deformation state data; A disturbance modeling module for modeling and analyzing the disturbance state of the adjacent objects in the grabbing process based on the spatial position data and the morphological feature data, and generating corresponding disturbance risk quantization data; A grabbing planning module for planning the motion path and the grabbing posture in the grabbing process based on the disturbance risk quantization data, and generating a grabbing trajectory and a grabbing posture control instruction; A micro contact state module for constructing a micro contact state model based on the local contact state data, and dynamically adjusting the grabbing process in combination with the clamping control instruction and the grabbing trajectory and grabbing posture control instruction to generate local grabbing force and grabbing action control instructions. A machine arm grabbing control method based on multi-sensor data analysis for any one of claims 1-9, the system comprising: A data acquisition module for acquiring spatial position data and morphological feature data of the grabbing area and surrounding objects, and local contact state data of the target object; A dynamics modeling module for predicting internal stress distribution data and local deformation state data of the target object in the grabbing process based on the local contact state data using a flexible object nonlinear dynamics model; A clamping control generation module for generating clamping control instructions based on the internal stress distribution data and the local deformation state data; A disturbance modeling module for modeling and analyzing the disturbance state of the adjacent objects in the grabbing process based on the spatial position data and the morphological feature data, and generating corresponding disturbance risk quantization data; A grabbing planning module for planning the motion path and the grabbing posture in the grabbing process based on the disturbance risk quantization data, and generating a grabbing trajectory and a grabbing posture control instruction; A micro contact state module for constructing a micro contact state model based on the local contact state data, and dynamically adjusting the grabbing process in combination with the clamping control instruction and the grabbing trajectory and grabbing posture control instruction to generate local grabbing force and grabbing action control instructions.

Images