165 results about "Forward kinematics" patented technology

The invention provides a training method and device for a VLA model, and relates to the technical field of intelligent robots with bodies, and the method comprises the steps that the VLA model outputs a joint angle sequence of a mechanical arm of a robot with a body based on visual input and language input; the differentiable forward kinematics module outputs a real-time pose of an end effector of the mechanical arm in a task space; based on joint space loss formed by the joint angle sequence and the joint angle in the teaching data, task space loss formed by the real-time pose and the tail end pose in the teaching data, and task space constraint loss, a multi-objective loss function is constructed, and total loss is calculated; calculating the gradient of the total loss to the joint angle sequence and the real-time pose through back propagation, and optimizing preset parameters of the VLA model; the above steps are repeatedly executed until the total loss meets the expectation, and a trained target VLA model is obtained; the technical problems that when a VLA model is trained in a joint space, the hardware coupling performance is high, and the generalization ability is insufficient are solved.

The invention discloses a spine tactile sensing system and method based on a multi-modal sensor, and the method comprises the steps: obtaining an original tactile signal, carrying out the preprocessing of the original tactile signal, carrying out the feature extraction based on the preprocessed original tactile signal, and obtaining a spatial distribution feature and a time dynamic feature; on the basis of tail end coordinates collected by forward kinematics of the tail end of the mechanical arm, space mapping of the tactile signals and the spine anatomical structure is constructed, and anatomical position features are obtained on the basis of the preprocessed original tactile signals and the space mapping; constructing task-oriented feature vectors based on the spatial distribution features, the anatomical position features and the time dynamic features, wherein the task-oriented feature vectors comprise a sliding state recognition feature vector and a spine ordinal number positioning feature vector; and constructing a multi-stage adaptive evolution hidden Markov model, and realizing spine ordinal number positioning in combination with the task-oriented feature vector. According to the method, through multi-dimensional feature fusion and dynamic model adjustment, anatomical prior knowledge and real-time tactile signals are combined, and spine segments are accurately recognized.

The invention provides a method and a device for predicting an operation track of an action-free labeling robot based on body flow representation, and the method comprises the steps: taking an initial point position in the body flow of the robot, a text feature in a language task instruction and a visual feature in an initial image as multi-modal condition features; the multi-modal condition features are input into the robot trajectory prediction model to obtain body flow information, and the robot trajectory prediction model can comprehensively consider motion of the robot, object related information contained in a language instruction and visual features in an initial image, so that more accurate body flow information is generated; the robot can accurately interact with a specific object; according to the method, the motion constraint information of each joint of the robot is obtained, the body flow information is decomposed into the motion component of each joint through forward kinematics mapping based on the motion constraint information of each joint, and kinematics constraints of different joints of the robot are fully considered, so that the accurate control of the robot motion is realized.

The invention discloses a security constraint type inverse kinematics solving method and device and a storage medium. The method comprises the following steps: firstly, forming a joint angle-tail end pose pairing data set by students through forward movement of a robot; training an improved RBF (Radial Basis Function) network by using the data set to establish initial mapping from a tail end pose to a joint angle to obtain an initial approximation model; taking the prediction output of the model as an initial population of a PSO algorithm, and updating the speed and position of particles in the initial population through iteration; designing a comprehensive cost function, introducing a joint angle constraint and an obstacle collision constraint, and evaluating the advantages and disadvantages of each particle; and when the PSO algorithm reaches the maximum number of iterations or the comprehensive cost function value is lower than a set threshold value, outputting a global optimal particle position as a final inverse kinematics solution. On the premise of ensuring the solving precision, the joint limit constraint and the obstacle avoidance constraint are explicitly processed, so that an accurate, safe and feasible inverse solution is output, and the reliability and the safety of robot operation are remarkably improved.

The invention provides a digital twinning five-axis machine tool contour error prediction method and system and electronic equipment, and belongs to the technical field of numerical control machine tool precision control and digital twinning. According to the method, servo axis data are collected in real time at a numerical control system end, electromechanical parameters are identified on line, comprehensive disturbance is estimated, and an autoregressive exogenous (ARX) model is updated accordingly; an interpolation instruction is generated in a digital twin body end mirror image numerical control system process, an updated model is utilized to predict a shaft position with deviation, a tool path is obtained through kinematics positive solution, and finally contour errors are calculated through geometric alignment. According to the digital twinning five-axis machine tool contour error prediction method and system and the electronic equipment, dynamic and high-precision prediction of the five-axis machine tool machining track contour error is achieved through the online identification and digital twinning technology, the limitation of a traditional static compensation method is overcome, and the precision of the digital twinning five-axis machine tool contour error prediction is improved. And a key basis is provided for prospective process optimization and precision control, and the machining precision and efficiency are effectively improved.

The invention provides a flexible glove based on sparse tensile sensing and a three-dimensional hand fine-grained reconstruction method, and relates to the technical field of man-machine interaction. The method comprises the following steps: constructing a biological constraint sensing sensor array layout; according to the sensor array layout, a wearable glove integrated with six-channel sparse stretching sensing is designed; constructing a three-dimensional multi-modal dynamic hand data set according to the wearable glove; designing a data flow processing pipeline; inputting the data set into a data stream processing pipeline to obtain a preprocessed data set; constructing a multi-stage graph attention network; inputting the preprocessed data set into a multi-stage graph attention network to obtain space mapping data from a 6-channel signal to a 15-joint signal; the mapping data are input into a multi-scale Transform model, and motion feature data of the 15 joints on the time sequence are obtained; according to the motion feature data, a forward kinematics model is adopted for decoding, and the rotation quaternion of the 20 skeletons is output. According to the invention, the stability of the wearable interactive glove can be improved.

The invention relates to a double-robot automatic calibration method based on self-adaptive point location planning in a restricted space, and belongs to the technical field of double-robot calibration, and the method comprises the steps: building a three-dimensional simulation model of a double-robot operation space on an Rviz platform based on an ROS architecture; a collision detection module is used for randomly generating a calibration point composed of collision-free double-robot joint angle vectors; when the robot is driven to move, data of the motion capture camera and positive kinematics Cartesian coordinates of the double robots are synchronously collected, a calibration result is obtained through optimization of a least square method by adopting a hand-eye calibration algorithm, and the effectiveness of the calibration result is judged through the average value of data calibration precision until calibration of the double robots is completed. Simulation modeling, collision detection and a hand-eye calibration technology are fused, a complex calibration problem is converted into a virtual environment parameter optimization process through a self-adaptive sampling strategy and a data synchronous acquisition mechanism, calibration robustness and precision are remarkably improved, and guarantee is provided for high-precision cooperative control of double robots.

The invention discloses a multi-degree-of-freedom pose adjustment method for a large-length-diameter-ratio target, and belongs to the technical field of multi-degree-of-freedom pose adjustment of automatic spraying equipment.The multi-degree-of-freedom pose adjustment method comprises the steps that firstly, the tail end pose deviation value of the target is obtained; and establishing a five-connecting-rod model and a forward kinematics model of the pose adjusting device. A five-connecting-rod model is constructed along the tail end of the large-length-diameter-ratio target, the pitching unit, the rotation unit, the transverse moving unit and the supporting foot unit in sequence; constructing a forward kinematics transformation matrix based on a product of coordinate transformation of the five connecting rods; and based on the forward kinematics transformation matrix, inversely solving the adjustment amount of each motion axis of the pose adjustment device at the target tail end pose deviation amount. And finally, according to an inverse solution result, adjusting each motion axis of the pose adjusting device so as to realize multi-degree-of-freedom pose adjustment of the tail end of the target with the large length-diameter ratio. According to the method, high-precision adjustment of the tail end posture is achieved, the purpose of precise adjustment of the automatic spraying length-diameter ratio target posture is achieved, and good practicability is achieved.

The invention relates to an excavator trajectory planning method based on an improved multi-objective particle swarm algorithm, and belongs to the technical field of engineering machinery trajectory planning. The method comprises the following steps: establishing a forward kinematics model through a D-H homogeneous coordinate transformation method; based on the forward kinematics model, establishing an inverse kinematics model through a geometric method; setting six path points in the track according to the mining track; based on an inverse kinematics model, converting the space coordinates of the six path points into space angles of joints; interpolation is carried out on the joint path points through a five-time B spline curve, and a track is generated; determining an optimization objective function according to actual application requirements, and setting constraint conditions according to a specific excavator type; and finally, the optimization problem is solved through an IADEMOPSO algorithm, an optimal Pareto solution is selected by using an average optimal evaluation formula, and an optimal track is generated. The track planning of the excavator can be realized.

A medical devices en-vivo positional determination system using a plurality of optical time of flight (TDF) transceivers along with gyroscopic angular and mechanical distance sensors for the determination of absolute linear and angular positional information of the tool bit tip for the purposes of more accurate hand-held drilling or cutting, on the workpiece using forward kinematic equations calculated on a microcomputer to provide a real-time display of the 3-linear position and 2-angular orientation of the tool bit (such as a drill, scalpel, or wire driver).

The invention discloses a mechanical arm autonomous obstacle avoidance planning method based on an improved artificial potential field method, and relates to the field of mechanical arm path planning and obstacle avoidance, in particular to the mechanical arm autonomous obstacle avoidance planning method based on the improved artificial potential field method. The objective of the invention is to solve the problem of low efficiency in an obstacle avoidance planning process of an existing method. The method comprises the following steps: constructing a forward kinematics transformation matrix from a base coordinate system to an end effector of the mechanical arm; a collision-free feasible path of the tail end of the mechanical arm from a starting point to a target point is solved by considering obstacles of a static part; key path points are extracted; the key path points containing the mechanical arm joint angle information are sequentially stored in a path point queue from the starting point to the end point to serve as temporary target points for improving the artificial potential field method; and key path points are selected from the path point queue to serve as current temporary target points, and an improved artificial potential field method and a collision early warning mechanism are used for guiding the mechanical arm to advance towards the temporary path points till the mechanical arm moves to the target points.

The invention discloses an intelligent robot grabbing control method and system based on AI vision, original vision and kinematics flow are fused through space-time alignment, and a physical-level dynamic shielding mask is accurately extracted through forward kinematics and reverse perspective projection. Then, the limitation of traditional pure semantic matching is broken through, a narrow-band gradient reference field of a shielding boundary is explicitly extracted to serve as a spatial topology bias injection attention mechanism, feature illusion completion guided by physical continuity prior is executed in a hidden space, and high-fidelity perspective features are generated; on the basis, the network synchronously regresdes six-degree-of-freedom grabbing pose and execution uncertainty, the uncertainty serves as an independent variable to dynamically attenuate and adjust the impedance rigidity of the mechanical arm, finally, self-adaptive compliant interaction in a blind area is achieved in a bottom layer physical servo link, and space positioning precision and operation safety are both considered.

The application discloses a mobile redundant manipulator real-time grabbing and trajectory tracking method and system based on optimization theory. In the method, the mobile redundant manipulator platform carries out real-time distance measurement on a target object based on a camera, and when the distance is less than a set threshold, a grabbing operation is performed. The application constructs a mobile redundant manipulator platform dynamics model based on robot forward kinematics, establishes a mobile redundant manipulator platform optimization model based on kinetic energy loss, designs a corresponding objective function, and sets constraint conditions of the corresponding optimization model in combination with physical limit conditions of the mobile redundant manipulator platform. By sampling time, inputting a current state of the mobile redundant manipulator platform, solving joint angular velocity and mobile platform speed at a next time, real-time trajectory tracking of the mobile redundant manipulator platform is realized. The application greatly increases the working range of the manipulator, the model is simple and easy to solve, accurate trajectory tracking can be realized, and the application prospect is wide.

The invention provides a digital twinning five-axis machine tool kinematics transformation method, system and device and a medium, and belongs to the field of numerical control technology and digital twinning crossing technology. The method comprises the steps that axis parameters are collected from a five-axis machine tool controller; constructing a unified five-axis parameter set comprising axis types, axis reference points and direction vectors; a hierarchical homogeneous transformation chain is constructed in a machine tool base coordinate system based on the parameter set, and forward kinematics transformation from a mechanical coordinate system to a workpiece coordinate system is achieved; and based on the unified parameter set and the transformation chain, reversely solving an axis position vector from a tool position instruction of the workpiece coordinate system and mapping the axis position vector back to a controller coordinate. By adopting the digital twin five-axis machine tool kinematics transformation method, system and equipment and the medium, the problems of poor model reusability and inconsistent virtual-real mapping caused by machine tool structure difference and numerical control system diversity are solved, cross-platform and cross-model general reversible kinematics transformation is realized, and the method has the advantages of high practicability and high practicability. And the engineering applicability and deployment efficiency of the digital twin model are improved.

The invention discloses an IMU (Inertial Measurement Unit) drift error calibration method for long-term attitude measurement of an industrial robot. The IMU drift error calibration method comprises the following steps: deducing a pose matrix between a robot base coordinate system and a navigation coordinate system; deriving a pose matrix between a robot tail end coordinate system and a sensor body coordinate system; kinematics modeling is conducted on the industrial robot, and motion parameters are obtained; deriving a homogeneous transformation matrix between two adjacent joints according to the kinematic parameters; acquiring actual joint corner information when the robot runs and actual information measured by a tail end IMU when the robot runs; wherein the actual information of the IMU comprises actual angular velocity information and actual attitude information; the collected actual joint rotation angle information is substituted into the homogeneous transformation matrix, theoretical attitude information of the tail end of the robot is obtained, and the theoretical attitude information is converted into a navigation coordinate system; converting the theoretical attitude information converted into the navigation coordinate system into an Euler angle; designing a Kalman filter; according to a Kalman filter, data fusion of theoretical attitude information and actual attitude information is completed, and calibration is achieved. According to the method, joint angle information, the industrial robot end theoretical attitude obtained through forward kinematics calculation of the kinematics model and actual attitude data obtained through IMU measurement are fused, and accumulated error calibration formed in long-term attitude measurement of the IMU is achieved.

The invention discloses a robot whole body control method and system based on real-time inertia motion capture and reinforcement learning, and belongs to the field of robot control. The system comprises a motion capture module, a data processing and forward kinematics module, a motion redirection and optimization module, a general motion strategy control module and a real-time control and driving module which are connected in sequence, and an end-to-end closed-loop control assembly line is formed through a shared memory and a high-speed communication protocol. According to the method, human body joint data are collected through inertia motion capture equipment, high-quality mapping from a human body posture to a robot model is achieved through posture calculation, non-uniform scaling and two-stage differential inverse kinematics optimization, and a stable joint instruction is generated in combination with a reinforcement learning strategy trained by a PPO algorithm. The method solves the problems of lack of feedback closed loop, joint deadlock, contradiction between fidelity and stability and the like in the prior art, has the characteristics of high robustness and high expansibility, and can be widely applied to scenes of remote operation, action reproduction and the like.

The invention discloses a robot multi-component low-cost high-precision calibration method based on a consumer-level camera, and the method comprises the steps: photographing a calibration plate through the consumer-level camera, and optimizing the internal parameters of the camera through corner detection and nonlinear optimization; a camera is installed at the tail end of the mechanical arm, about 50 sets of calibration plate images are shot, the pose is obtained by means of forward kinematics of the mechanical arm, and the pose relation between the camera and the mechanical arm is analyzed through the formula AX = XB; then the chassis and the camera on the mechanical arm are used for joint calibration, the calibration relation between the mechanical arm and the trolley is solved by moving the trolley and the mechanical arm observation calibration plate, and finally the calibration precision is verified by calculating the re-projection error. The method does not need to depend on high-precision equipment and a professional calibration field, calibration can be completed by using a common consumer camera and a simple calibration board, the method can adapt to various cameras, the requirement for computing equipment is not high, and the requirement can be met by a common workstation. The method can be carried out in a dynamic scene, the calibration time is within 1 hour, and the precision is within 5 mm.

The invention relates to the technical field of intelligent control, and particularly to a gantry platform kinematics modeling method considering parallelism and perpendicularity errors, which is applied to a gantry platform device. The gantry platform device comprises a crossbeam, a first guide rail and a second guide rail, one end of the crossbeam is in sliding fit with the first guide rail, and the other end of the crossbeam is in sliding fit with the second guide rail; the first guide rail is provided with a first longitudinal motor, and the second guide rail is provided with a second longitudinal motor; and the crossbeam is provided with a transverse motor configured to drive the gantry platform to move along the crossbeam. The method comprises: establishing a two-dimensional coordinate system; constructing a forward kinematics solution model; and constructing an inverse kinematics solution model.

The invention belongs to the technical field of semiconductor manufacturing safety control, and particularly relates to a collision detection method and system for a wafer transfer robot, and the method comprises the steps: receiving a joint angle sequence and height coordinates outputted by a robot trajectory planning module; cartesian coordinates of all nodes of the mechanical arm are calculated through a forward kinematics solver; preferentially checking whether the node is located in a predefined safety area or not by utilizing a point area judgment and line segment area detection algorithm; performing regionalized wafer collision detection on the wafer outside the safe region; if the wafer collision does not occur, constructing an oriented bounding box (OBB) for the risk connecting rod; detecting whether the vertex of the OBB crosses a boundary or not, and performing line segment intersection detection on each edge of the OBB and a chamber boundary line segment of a corresponding area; and finally outputting a collision judgment result, and deciding whether to execute motion or not by a robot control system. According to the invention, efficient and accurate collision detection is realized through a multi-stage detection mechanism and highly adaptive section selection, and the method has the advantages of high real-time performance, low cost, easy integration and the like.

Embodiments of the application disclose a screw-based gesture recognition method and device, electronic equipment and a storage medium. A first key point position corresponding to a hand key point in a continuous frame image is detected. A three-dimensional coordinate of each hand key point in a corresponding first coordinate system is calculated. The first coordinate system is a three-dimensional coordinate system determined by adjacent key points away from the fingertips of the hand key point. The three-dimensional coordinates are input into a preset screw-based forward kinematics model to obtain a pose transformation set of the hand key point corresponding to each frame of image. According to the pose transformation set, a Jacobian matrix corresponding to each frame of image and a predicted three-dimensional coordinate corresponding to each hand key point are calculated, and a corresponding second key point position is calculated according to the predicted three-dimensional coordinate. The pose transformation set is optimized based on the positions of the key points and the Jacobian matrix. Parameters are output based on the optimized pose transformation set. The application ensures the consistency of the processing of the shapes of the front and back hands, thereby improving the accuracy of the processing results.

The invention discloses an XR gesture tracking and recognition method based on multi-view depth fusion. The method comprises the following steps: constructing a virtual camera by adopting a perspective cutting method, and dynamically extracting a hand region after geometric correction through gesture detection and perspective cutting of each frame of image; in combination with internal and external parameters of a camera, through position coding and a multi-head attention mechanism, a geometrical relationship between visual angles is modeled, semantic relevance of multi-visual-angle 3D features is deeply excavated, globally consistent hand representation is constructed, then six-degree-of-freedom information of wrist joints and rotation angles of the joints are output based on a regression device, and forward kinematics and a linear skin algorithm are combined, so that the degree of freedom of the wrist joints is calculated. Generating a complete hand posture and skeleton key point coordinates; and based on the key point position and the rotation angle of each joint, gesture recognition is realized through template matching. According to the method, multi-view image information and hand kinematics constraints are combined, high-precision and high-robustness three-dimensional gesture attitude estimation and classification are achieved while real-time performance is guaranteed, and the method is suitable for various XR scenes.

The invention discloses a continuum robot position tracking and joint limiting avoidance method, and belongs to the field of robot control. According to the method, a current path is obtained through forward kinematics modeling, joint limiting constraint and ideal path generation are combined, a nonlinear programming problem with the maximum path deviation and the minimum path deviation as the target is constructed, and joint increment is iteratively solved to achieve end tracking and overall path fitting. According to the method, the problem that inverse kinematics is difficult to solve is avoided, joint overrun is explicitly avoided, and the method is suitable for high-precision and high-reliability control in a narrow space and an obstacle environment.

The embodiment of the invention provides a regular hole cutting method and device, a cutting robot and a storage medium, and relates to the technical field of robots. Geometric parameters and cutting parameters corresponding to the to-be-cut regular holes set by a user are obtained, and a plurality of cutting point positions corresponding to the to-be-cut regular holes are determined according to the geometric parameters; under the condition that the current cutting mode is determined to be wrist cutting according to the cutting parameters, the robot Cartesian position of the six-axis body when the sixth joint position serves as the zero-degree position is calculated through a kinematics positive solution; according to the Cartesian position of the robot, the point position Cartesian position of each cutting point position and the tool parameters of the cutting tool, the wrist joint position of the wrist mechanism under each cutting point position is calculated; and the wrist mechanism is driven to move according to the positions of all the wrist joints, so that the cutting tool is moved to all the cutting point positions for regular hole cutting, and therefore the machining efficiency can be improved while the regular hole cutting precision and the automation degree are improved.

This invention discloses a multi-loop hierarchical control method and system for hydraulic robotic arms based on hybrid modeling, belonging to the field of hydraulic robot control technology. The method includes: utilizing historical operating state data of the hydraulic robotic arm, aiming to minimize data loss, physical loss, and regularization constraints, and constructing a physical-data hybrid dynamics model through physical information neural network for system dynamics identification; using the physical-data hybrid dynamics model to track the trajectory of the hydraulic robotic arm's end effector, and combining forward kinematics to determine the desired joint angle, desired joint angular velocity, and desired joint angular acceleration, which are then input into an active disturbance rejection servo controller; and using a nonlinear feedback control law to perform servo control on the hydraulic robotic arm. This method can improve the operating efficiency and accuracy of the hydraulic robotic arm, solving the problem that the trajectory tracking accuracy of existing hydraulic robotic arms under complex working conditions needs improvement.

The invention relates to an autonomous capturing method and system for an earth orbit non-cooperative target. The method comprises the following steps: acquiring left and right images of the earth orbit non-cooperative target; inputting the left image and the right image into a DETR model for target detection, and outputting a semantic identifier of a target component; performing nonlinear mapping and binocular parallax calculation on the semantic identifier to obtain two-dimensional world plane coordinates and depth information, and fusing the semantic identifier, the two-dimensional world plane coordinates and the depth information to generate three-dimensional space positioning of the target component; the method comprises the following steps: converting a task instruction in a natural language form into a standardized control instruction based on a framework combining semantic analysis and a rule base; and mechanical arm forward kinematics is established on the basis of a D-H parameterized model, a standardized control instruction is converted into a target tail end pose through inverse kinematics solution, and high-precision capture of a non-cooperative target is achieved. Compared with the prior art, the method provided by the invention has the advantages of precision and real-time performance of spatial non-cooperative target autonomous capture and the like.

The invention relates to the technical field of robot control, in particular to a hybrid control method for high-precision parking of a four-wheel independent steering robot, which comprises the following steps of: constructing a robot system model, establishing an inverse kinematics model, a forward kinematics model and an actuator model, sensing global pose parameters of the robot by using the forward kinematics model, and in an advancing stage, controlling the robot to stop in a high-precision manner. A non-linear predictive control method is adopted, a PID control method is adopted in the stopping stage, the two control methods are comprehensively adopted in the transition stage, a robot body motion instruction is output, target control parameters of all wheels are worked out through an inverse kinematics model, execution parameters of all the wheels are worked out and executed through an actuator model, and the wheels are driven to move. The robot can be guided to efficiently and smoothly advance along the preset path in the advancing stage, high-precision stopping is rapidly completed in the stopping stage, and the control method is seamlessly and smoothly switched in the transition stage.

The application discloses a Tai Chi action posture recognition and deviation correction training system and relates to the technical field of action recognition and sports training, and solves the technical problems that it is difficult to quantitatively recognize the coherence of Tai Chi actions, the guidance lags behind the actions, and the template does not match the body shape of a user. The system comprises a posture acquisition module, which acquires a user bone frame sequence and a joint angular velocity change rate; a motion trend prediction module, which adopts a rotation matrix to predict the head and a forward kinematics chain to recursively output a next frame prediction coordinate; a ghost generation module, which sets a differential matching window according to four stages of Tai Chi, generates a semi-transparent ghost by performing size normalization mapping on a three-bone segment group scale factor weighted by square roots of joint numbers, and performs size normalization mapping on a three-bone segment group scale factor weighted by square roots of joint numbers; a deviation perception module, which constructs a dynamic weight smoothness penalty term to generate a deviation correction signal; and a multi-modal feedback execution module, which maps a deviation amount into a line color and a vibration parameter, and outputs feedback according to the time sequence of vision, touch and voice.

The invention discloses a multi-object picking and placing method for an autonomous mobile mechanical arm robot, and belongs to the field of robots and artificial intelligence. Comprising the steps of target positioning based on a YOLOv11 algorithm, task planner processing based on a Prompt technology and motion planner processing based on a neural network and differentiable forward kinematics. RGB information of the top camera is used as input of a YOLOv11 algorithm, a detection frame and an article name are obtained, the coarse position of the corresponding article is obtained, a large language model deduces the execution action of the robot according to an input task and the information obtained by the top camera, and the chassis moving position and the mechanical arm joint angle corresponding to the target article are deduced through a neural network. And after the mechanical arm reaches the coarse position, the accurate position is obtained through RGB information and depth information of the camera on the arm, and the final action is completed. The robustness of the method is improved through the large language model, and the execution efficiency is improved in combination with the neural network of the differentiable forward kinematics algorithm.

The invention discloses a kinematics semi-analytical modeling method and terminal for a human wrist-simulated parallel mechanism, and the method comprises the steps: constructing a global coordinate system, a first coordinate system and a second coordinate system when the wrist parallel mechanism is subjected to kinematics positive solution, and then obtaining a target D-H expression corresponding to the tail end pose of the human wrist-simulated parallel mechanism, a target D-H expression is obtained, the target D-H expression shows the relation between the tail end pose and a first joint angle and the relation between the tail end pose and a second joint angle, then a target modular length expression is obtained, the target modular length expression shows the relation between a first modular length and the first joint angle and the relation between a second modular length and the second joint angle, and finally, the target D-H expression shows the relation between the tail end pose and the second joint angle. And obtaining numerical values of a first modulus length and a second modulus length, and substituting the numerical values of the first modulus length and the second modulus length into the target D-H expression and the target modulus length expression to obtain a tail end pose. The invention provides a solution capable of completing operation on a low-computing-power platform for a parallel mechanism simulating a human wrist.

The application provides a three-dimensional displacement monitoring method and system based on GNSS and fusion of inclination sensors, and relates to the technical field of displacement monitoring.The application determines key monitoring points of a to-be-measured structure and arranges an inclination sensor chain through modal analysis, sets a GNSS-inclination integrated monitoring unit at an end point to establish a unified reference, obtains reference absolute displacement and high-frequency attitude angle data, and then takes the reference displacement as a starting point to perform forward kinematics recursion along the sensor chain to reconstruct the theoretical three-dimensional displacement of each point, and finally outputs the fused three-dimensional displacement as a deformation monitoring result as an indirect correction of the recursion result is performed based on the principle of minimum strain energy with the reference displacement as a fixed boundary constraint.The application realizes the complementary advantages of low-frequency precision and high-frequency response through a double-layer fusion architecture of front-end enhancement and rear-end reconstruction, that is, the long-term drift of the inclination sensor is constrained by the GNSS absolute reference, and the dynamic details of the GNSS are compensated by the high-frequency response of the inclination sensor.