294 results about "Dynamic motion" patented technology

In various examples, systems and methods are disclosed relating to generation of four-dimensional (4D) content models, such as 4D content models to render realistic sequences of frames of 3D data. The systems can initialize a 3D component of the 4D content model, such as a 3D Gaussian splatting representation, based at least on a prompt for the 4D content. The system can configure motion and / or dynamics for the sequence of frames by evaluating frames rendered from the 4D content model using one or more latent diffusion models (LDMs), including a video LDM. The system can perform operations such as autoregressive generation of frames to create long sequences of content, motion amplification to facilitate realistic, dynamic motion generation, and regularization to facilitate generation of complex dynamics.

The invention provides a dynamic scene reconstruction method and system based on spatial decomposition and Gaussian splashing, and belongs to the technical field of computer vision and graphics. The method comprises the following steps: decomposing a dynamic scene into a static standard space and a dynamic playground; reconstructing the standard space by using three-dimensional Gaussian splashing to obtain Gaussian primitives of the standard space; decomposing the characteristics of the motion field into standard space characteristics and a plurality of motion subspace characteristics based on tensor decomposition; fusing the standard space feature and the multiple motion subspace features through a motion information decoder, and decoding motion information of Gaussian primitives in the standard space; and applying the motion information to Gaussian primitives of a standard space, generating any moment representation of a dynamic scene, and rendering a target view angle image through a snowball throwing method to realize reconstruction of the dynamic scene. Through the method, the reconstruction precision and the rendering quality are effectively improved, and the real-time rendering capability is kept.

The invention discloses an intelligent analysis system and method for orthopedic nursing, and relates to the technical field of medical information, and the method comprises the steps: collecting and preprocessing medical image data, combining the daily motion state data of a patient, constructing a multi-dimensional data set containing skeleton morphological features and motion features, extracting data features, and carrying out the analysis of the data. Adaptive fusion is carried out by adopting an attention mechanism, and a digital feature model containing static morphological features and dynamic motion features is established; based on the digital feature model, joint stress features and coupling features are obtained, and a comprehensive evaluation index system is constructed to achieve quantitative evaluation of the bone health condition; and according to the evaluation index system, constructing multi-level risk early warning to carry out real-time monitoring and early warning on abnormal conditions, and meanwhile, based on the rehabilitation progress evaluation indexes, adaptively generating personalized rehabilitation schemes and exercise suggestions. According to the rehabilitation parameter adjusting method combined with double-objective optimization, the safety of the rehabilitation process is guaranteed, and the pertinence of the rehabilitation effect is improved.

The invention discloses a multi-view child motion coordination ability evaluation system and method based on deep learning, belongs to the technical field of motion evaluation, and solves the problems that child motion evaluation in the prior art mainly depends on manual observation and simple physical testing, and multiple angles and key details of child actions are difficult to synchronously track. The method comprises the following steps: training to obtain a skeleton point detection model and an athletic ability evaluation model, acquiring real-time video streams of personnel entering a field based on an acquisition camera, identifying skeleton key points in a preprocessing data set by the skeleton point detection model, performing multi-person detection on a personnel matching result based on an athletic area division method, and evaluating the athletic ability of the personnel. The exercise ability evaluation model carries out quantitative analysis on multi-person detection results; according to the invention, visual identification and motion state detection technologies are combined, a front view angle and side view angle dual-camera layout is adopted, and a deep learning algorithm is matched, so that automatic children dynamic motion evaluation is realized. The action process can be completely captured, and the detection accuracy and efficiency are improved.

The invention discloses a two-arm robot self-adaptive skill learning method based on a dynamic motion primitive. The method comprises the steps of obtaining teaching data, establishing a dynamic motion primitive model, learning shape features of tracks from the teaching data through the dynamic motion primitive model and two arms, generating corresponding track shape functions, and obtaining an initial track control weight; obtaining a new task target point, and generating an initial track similar to the teaching track by combining the learned initial track control weight and the track shape function with the new task target point; optimizing the initial trajectory to obtain an optimized trajectory shape, and obtaining an optimized trajectory control weight; and the optimized trajectory control weight is substituted into the dynamic motion primitive model, adaptive adjustment is performed by using an adaptive time scaling factor based on a trajectory change rate, and an optimized trajectory of the two arms is generated. It is ensured that the robot efficiently and accurately completes tasks.

The invention provides a method for predicting the vibration reduction effect of exercise compression pants, and relates to the technical field of functional costume design, and the method comprises the steps: collecting a static costume pressure value of a lower limb target muscle group and soft tissue amplitude data in exercise, constructing a pressure-amplitude mathematical model through data processing, and verifying the model; constructing a human body-clothing finite element model, inputting fabric mechanical parameters into the model, and verifying the effectiveness of the model after iterative optimization; inputting fabric mechanical parameters into the verified finite element model, inputting a pressure prediction value output by the model into the pressure-amplitude mathematical model, and predicting soft tissue amplitude in a dynamic motion scene; according to the predicted soft tissue amplitude in the dynamic motion scene, the vibration reduction effect of the motion compression trousers in the target motion scene is evaluated. The method provided by the invention solves the problem that the vibration reduction effect evaluation of the traditional sports compression pants depends on a large number of repeated physical experiments and is difficult to realize accurate quantitative prediction.

The invention relates to the technical field of mechanical arm control, and discloses a mechanical arm path planning method and system.The method comprises the steps that in a simulation environment, a reward function is generated based on the current pose and the target pose of the tail end of a mechanical arm, a deep reinforcement learning network is trained based on the reward function, and a global planning path is obtained; constructing a dynamic motion primitive system by taking the global planning path as teaching; obstacle information is obtained, and an obstacle potential field model is constructed under the condition that an obstacle affects the working space of the tail end of the mechanical arm; based on the obstacle potential field model, a rejection item in the dynamic motion primitive system is constructed, and an obstacle avoidance path is generated based on the rejection item and is used for guiding the tail end of the mechanical arm to move away from the obstacle until the tail end of the mechanical arm reaches the target pose, so that the technical problem of simultaneously realizing real-time obstacle avoidance and global optimum in mechanical arm path planning is solved.

The invention provides a mechanical arm dynamic obstacle avoidance trajectory planning method based on a dynamic motion element and a steering force field, and belongs to the technical field of robots. Secondly, learning the teaching track preprocessed in the step 1 by using an improved dynamic motion primitive (DMP) algorithm, so that the DMP can generate a motion mode similar to the teaching track; through a depth camera installed around a mechanical arm or at a fixed position, obstacle information, shape and motion state are obtained in real time, obstacle avoidance repulsive force is calculated, steering force is quantified, and obstacle avoidance of free motion obstacles is achieved; a Lyapunov equation is constructed, based on the Lyapunov stability theory, the stability of the improved DMP algorithm is proved, and it is ensured that the motion trail of the mechanical arm in the dynamic environment is converged to a target point; and finally, carrying out an obstacle avoidance experiment to verify the dynamic obstacle avoidance capability of the mechanical arm. According to the invention, the trajectory generalization ability of the mechanical arm in a narrow working space is obviously improved.

The invention relates to the technical field of motion capture and data processing, in particular to a method for realizing real-time dynamic capture of a three-dimensional space of a human spine by applying an IMU (Inertial Measurement Unit). According to the method, a personalized three-dimensional dynamic spine model is constructed based on CT image data of a user, and a mathematical model of the spine is established by fusing a physiological anatomical structure of the spine and a robot mechanism motion theory so as to accurately describe motion characteristics of the spine. An IMU (inertial measurement unit) sensor is adhered to a key part of the back of a user, an individualized spine model is driven by acquiring and processing IMU data in real time, real-time reconstruction of dynamic movement of the spine is realized, and space angles among vertebral sections of the spine are accurately calculated. The dynamic change of the spine in the movement process can be accurately captured, the movement posture of the spine can be deeply analyzed, and the movement effect of the spine can be accurately and visually tracked and quantified in real time.

The invention relates to a vehicle trajectory tracking control method considering dynamic motion requirements. Comprising the following steps: 1, establishing a nonlinear vehicle dynamics model by comprehensively considering longitudinal motion, lateral motion and yawing motion of a vehicle; 2, searching saddle node bifurcation points with qualitative change of the stability state of the vehicle through a particle swarm optimization algorithm and a Lyapunov criterion, and constructing a dynamic stability domain boundary of the vehicle under the change of the longitudinal vehicle speed and the road adhesion coefficient; and 3, designing a vehicle multi-target trajectory tracking controller based on a nonlinear model predictive control theory, quantitatively evaluating the stability state of the vehicle by adopting a stability index, and meanwhile, designing a controller built-in parameter adjustment strategy to match the dynamic motion requirements of the distributed driving electric vehicle in different stability states. According to the method, the driving safety and the movement comfort of the distributed driving electric automobile are effectively ensured, and a set of solution is provided for realizing the optimal cooperative control of the chassis subsystem of the automatic driving automobile.

Architectures and techniques are provided for improved vibration compensation for a boom lift-robot-mounted (BLMR) system, which can result in improved BLMR operating precision and safety at a reduced cost. Boom lifts typically contain extendable large-scale, variable-stiffness structures, subject to complex nonlinear static deformation and dynamic motion / gust-induced vibrations. Said vibrations can be compensated for at an end effector of a robot via a Jacobian-based vibration compensation via IMU (JVCI) procedure. The output of the JVCI procedure can rely on IMU feedback and can be combined with other vibration-mitigation techniques, including various feedforward techniques.

A wearable system for pedestrians / users is adapted to detect threats posed by, for example, approaching vehicles, bikers, and issue an alert. The wearable system may include an inertial measurement unit to measure the gait of the pedestrian and actively adjust alert generation thresholds to account for the noisy and large movements of the gait using present and past IMU data. The thresholds can be raised during active phases of the gait, and lowered in the quieter (i.e., less active) phases of the gait, keeping the range, and therefore time to react, relatively steady with respect to the pedestrian / user's motion. The alert generation threshold may be combined with the processed receiver data to determine threats and alert the pedestrian / user using any combination of haptic, audio, vibrational, or visual outputs.

The invention discloses a flow field numerical simulation method for autonomous swimming of underwater bionic fish, and particularly relates to the field of bionic fish swimming flow field numerical simulation, which comprises the step of describing the shape of a fish body by establishing a contour line equation of a horizontal section. The global ground coordinate system and the fish body coordinate system are associated through the gravity center position and the pitch angle of the fish body, and a time sequence fish body motion data set is established. And generating a time sequence flow field data set by combining unsteady flow field data acquired by the distributed underwater sensor array. And based on the flow field control equation and the dynamic motion characteristics of the fish body, constructing a static interactive coupling model, extracting key dynamic characteristic quantities, and establishing a flow field prediction model. And inputting interactive coupling data to the static model to calculate a global coupling solution for error analysis, and generating a dynamic interactive coupling model. And according to the flow field prediction result and the dynamic interactive coupling model, correcting the motion characteristic data of the fish body, optimizing the motion trail of the bionic fish, and realizing closed-loop optimization of the flow field and the fish body motion.

The invention discloses a five-dimensional motion vector real-time construction and calibration method based on multi-sensor fusion, and relates to the technical field of multi-sensor information fusion and dynamic state estimation, and the method comprises the steps: collecting and preprocessing motion carrier data, obtaining a preprocessing data set and a feature data set, inputting the feature data set into a long short-term memory network model, and obtaining a multi-sensor fusion model; and outputting a sensor error offset prediction vector to the extended Kalman filtering model, outputting a preliminary five-dimensional motion vector, and performing consistency verification and correction on the preliminary five-dimensional motion vector through a kinematics constraint equation to obtain a corrected five-dimensional motion vector so as to drive a virtual model corresponding to a motion carrier to perform synchronous position and attitude updating. According to the method, the sensor error is predicted through the long-short-term memory network model, and the dynamic motion model constraint module is additionally arranged to perform physical constraint correction, so that the problems of inaccurate error compensation and lack of physical authenticity of the calculation result under the dynamic working condition are solved, and the construction precision and reliability of the five-dimensional motion vector are improved.

The invention provides a virtual fitting method and device, equipment and a storage medium. The method comprises the following steps: acquiring a video containing a figure and a clothing image; determining a character potential feature map according to the video potential features obtained from the video without the clothing part and the depth posture in the video; determining a clothing feature map and clothing semantic latent features according to the clothing image; performing first self-attention calculation according to the character latent feature map and the clothing feature map to obtain a first feature map; performing first cross attention calculation according to the clothing semantic latent features and the first feature map to obtain a second feature map; performing space-time attention processing according to the second feature map to obtain a third feature map; and generating a virtual fitting image according to the third feature map and the clothing semantic latent features. According to the method provided by the invention, learning is carried out for clothing properties through space-time attention processing, and an efficient global space-time attention mechanism is realized, so that a more real and coherent try-on image is generated, and the visual fidelity and dynamic motion consistency of clothing are improved.

The invention belongs to the technical field of aerospace, and relates to a trailing edge flap driving mechanism for adjusting a rotor cone and dynamic balance and controlling vibration, a driver transmits high-frequency dynamic motion to a trailing edge flap through a push-pull rod, and the trailing edge flap is driven to dynamically deflect at high frequency to change aerodynamic load and aerodynamic-elastic coupling response of a blade; when the trailing edge flap dynamically deflects at high frequency, the push-pull rod drives the trailing edge flap to change the initial deflection angle through stretching and retracting of the piston rod so as to change the aerodynamic load and elastic deformation of the paddle; the trailing edge flap driving mechanism drives the trailing edge flap to dynamically deflect at high frequency, and passing frequency vibration of a rotor wing can be restrained; meanwhile, the push-pull rod can change the initial deflection angle of the trailing edge flap, so that the rotor cone is actively adjusted, and the vibration load of the rotor is reduced.

The invention discloses a moving object grabbing method based on deep learning, and the method comprises the steps: collecting a dynamic scene image through an industrial camera, and detecting the 2D image features of a target object in a dynamic motion process through an improved YOLOv11 algorithm; a depth sensor is used for collecting 3D point cloud data of a target object to be grabbed in advance for improved YOLOv11 model training, preprocessing and feature extraction are carried out on the 3D point cloud data, a 2D detection result detected by an improved YOLOv11 algorithm is fused with the 3D point cloud data, then registration of the 3D point cloud data and a 2D image is carried out, and the three-dimensional attitude of the dynamic target object is estimated. On the basis of the fused and registered data and the motion track of the dynamic target object, the optimal grabbing point and posture are predicted, and an improved S-shaped curve track planning algorithm is adopted to generate a grabbing path of the mechanical arm; the mechanical arm completes the grabbing action according to the planned path, and meanwhile, the grabbing force and posture are adjusted in real time through force sense feedback; and the grabbed target object is placed at a designated position, and the next grabbing task is prepared.

The invention relates to the technical field of garment tailoring, in particular to an automatic garment pattern making method and system, and the method comprises the following steps: constructing a garment preliminary pattern based on fabric characteristics; the method comprises the following steps: dynamically simulating a preliminary garment model by collecting accurate human body data; carrying out accurate cutting based on a dynamic simulation result; according to the invention, the dynamic response of human body motion data and cloth is considered through the clothing preliminary pattern dynamic simulation module, the clothing pattern is optimized, the clothing is ensured to keep fit and comfort in dynamic motion, and the pattern is dynamically adjusted through a deep learning algorithm to adapt to the motion requirements and human body form changes of different users; the tailoring path is optimized by adopting deep reinforcement learning, and the intelligent agent can continuously adjust and optimize the tailoring path in actual production through interaction with the environment; the deep learning algorithm and the reinforcement learning algorithm can be adaptively adjusted according to feedback, and the optimization precision is gradually improved.

The invention discloses a behavior analysis method and device, electronic equipment and a computer program product, and relates to a big data technology. The method comprises the following steps: processing received real-time video stream data to obtain continuous frame images; when it is detected that the continuous frames of images contain the target object, respectively extracting limb features of the target object in each frame of image based on a key point detection model; determining track features of the target object based on the limb features in each frame of image and the corresponding timestamp; and performing behavior analysis on the target object according to the limb features and the trajectory features. According to the method, the static limb state and the dynamic movement track of the target object are comprehensively considered, and the behaviors of the target object are accurately analyzed from multiple dimensions. Compared with the existing scheme of monitoring depending on manpower, the method has the advantages that the investment of labor cost is reduced, and the situations of false alarm and missing alarm are effectively reduced.

The invention relates to the technical field of bionic robots, and particularly provides a robot motion trail generation method and device, and the method comprises the steps: obtaining a corresponding behavior sequence based on the motion behavior of a robot, determining an optimal variable corresponding to each time node based on a target dynamic model for each time node in the behavior sequence, and determining a target trajectory corresponding to the motion behavior based on the optimal variable. In the embodiment of the invention, the target dynamics model for optimizing the high-dynamic motion track is constructed based on the robot mass center dynamics and the whole-body kinematics, the track optimization algorithm is simplified, the whole-body motion information of the robot in the high-dynamic motion process is effectively reflected, and the corresponding behavior sequence is coded for the high-dynamic motion, so that the high-dynamic motion track optimization is realized. Motion information in various high-dynamic motion processes is accurately reflected, and the trajectory planning precision is improved.

The invention belongs to the field of underwater target navigation and positioning, and discloses an underwater multi-station speed measurement technology based on acoustic beacons, which comprises the following specific steps: 1) establishing a coordinate system, acquiring measurement information, and estimating the initial position of a navigation body; 2) sound velocity correction and position parameter estimation are carried out point by point, and speed parameter estimation is carried out after iterative calculation of position parameter convergence; and 3) estimating a constraint polynomial coefficient, and optimizing the position and speed parameters of the navigation body. The technology has the characteristics that the observation process does not depend on a navigation body, the measurement precision is high, the result can be accurately obtained in real time and the like, reliable estimation of position and speed parameters based on polynomial constraint under a complex navigation working condition is realized, and the problems of functional expression of observation motion under a high-dynamic motion condition and provision of continuous and smooth motion parameters are solved; the method has good adaptability to the complex motion process of an underwater target, has a wide technical application prospect, and can provide a data processing and analysis technology for underwater motion performance analysis and evaluation of a navigation body.

The invention belongs to the technical field of robot application, and particularly relates to an acupoint recognition and mechanical arm control method for assisting expectoration, which uses a lightweight Inverted Residual convolution module containing depth separable convolution and residual connection to carry out high-precision recognition on 13 acupoints on the back, and on the basis of RGB three-channel image input, the acupoint recognition and mechanical arm control method for assisting expectoration is realized. A spine central axis channel is added, and a user-defined loss function including MSE loss, spine symmetry constraint and acupoint position constraint is designed to optimize the deep learning model, so that the accuracy and symmetry of acupoint recognition are further improved; after identification is completed, the system combines an improved near-end strategy optimization reinforcement learning algorithm and a visual servo technology to design a self-defined reward function which comprehensively considers distance reward, angle reward, cross punishment and target completion reward. And a voice coil motor of an end effector of the mechanical arm is guided to approach the target acupuncture point and align with the normal of the back of the human body, so that accurate control and dynamic motion planning of the voice coil motor are realized, and auxiliary sputum excretion actions are effectively executed.

The invention provides a mechanical arm movement track generation method and device, a storage medium and a program product. The method comprises the steps that observation data of the current environment and state data of the mechanical arm at the current moment are obtained; respectively carrying out feature coding on the observation data and the state data to obtain an observation feature vector and a state feature vector; the observation feature vector, the state feature vector and the state data are input into a trained mechanical arm motion trail generation model, and the motion trail of the mechanical arm is obtained; wherein the mechanical arm motion trail generation model comprises a denoising network and a probabilistic dynamic motion primitive ProDMP decoder, the denoising network is used for generating ProDMP weight parameters based on the observation feature vector and the state feature vector, and the ProDMP decoder is used for generating a motion trail according to the ProDMP weight parameters and the state data. The continuity, smoothness and boundary controllability of the motion trail of the mechanical arm are improved.

The present invention relates to articulating spacecraft chassis and methods of making and using same. The present invention relates to spacecraft chassis and methods of making and using same. Such spacecraft chassis have a dynamic movement capability that allows the spacecraft to alter its structure while still maintaining industry volumetric launch standards. This capability increases opens up a wide range of achievable volumetric states and increases the ability to meet mission requirements by introducing a new tunable parameter. In addition, the judicious selection of certain dynamic movement parameters can result increased payload capabilities and improved maneuverability.

The invention discloses a method and device for controlling a robot to open a door and a nonvolatile storage medium. The method comprises the steps that pose information corresponding to a target handle and position information corresponding to a mechanical arm and a chassis in a target robot are obtained; inputting the pose information, the position information corresponding to the mechanical arm and the position information corresponding to the chassis into the target dynamic motion primitive model to obtain a predicted motion track; state information corresponding to the mechanical arm and the chassis is obtained; based on the state information corresponding to the mechanical arm and the chassis, corresponding dynamic models are established; an action sequence is generated based on the predicted motion trail, kinetic models corresponding to the mechanical arm and the chassis and a prediction control model based on the model; and controlling the mechanical arm and the chassis to perform door opening motion based on the action sequence. The technical problems that a mechanical arm and a chassis of an existing robot are poor in coordination, a handle can not be accurately grabbed or the door opening action can not be completed when the robot moves according to a fixed track, and accidents are likely to happen are solved.

The invention relates to the technical field of radar motion recognition, in particular to a method and system for recognizing a human motion state through radar, and the method comprises the following steps: receiving echo signals of millimeter wave radar through a sensor, analyzing the amplitude and phase change of the signals, calculating the average value and standard deviation of the amplitude, and comparing the received signals. And recording an amplitude extreme point and a phase abrupt change point to obtain a radar echo signal feature. According to the method, the amplitude and phase change of the radar echo signals are meticulously analyzed, the execution process optimizes the monitoring of the motion state of the human body, while key signal features are extracted, the amplitude average value and the standard deviation of the signals are calculated in a continuous time period, and the sensitivity and the accuracy of signal processing are further improved; by analyzing the gradient value of the phase change and comparing the change of the adjacent time points, the method effectively recognizes the fine change of the human body movement speed, and greatly improves the capturing capability of the dynamic movement track.

The invention provides a target object 6D pose estimation method and system oriented to electronic equipment AR assembly guidance. The method comprises the steps of electronic part coarse positioning based on improved YOLOv5s-CBAM, multi-mode (RGB-point cloud) feature fusion fine registration and LSTM dynamic motion prediction compensation. The system is developed based on an ROS framework and a PyTorch platform, and depends on an Open3D point cloud library and a TensorRT acceleration engine. According to the method, the problems of small target detection, metal reflection and shielding in an electronic assembly scene are solved through a cascaded visual positioning technology, the problem of insufficient precision caused by dependence on mark points or general SLAM in a traditional AR guide system is avoided, and the pose estimation robustness in a complex working condition is improved. Compared with an existing assembling method depending on manual calibration or two-dimensional drawings, the assembling efficiency and the first-pass yield are remarkably improved.

Apparatus and associated methods relate to a system configured to automatically identify a kinematic profile of a user and generate a recommendation library. In an illustrative example, a dynamic kinematic intelligence platform (DKIP) may activate a sensor module in response to a neuromuscular activation signal. A sequence of N kinematic evaluation, for example, may be retrieved to assess a user's health conditions. For example, a movement code generation operation may include, for i = 1 to N, transmit an i-th kinematic evaluation from the sequence of kinematic evaluation to a user device, and generate an i-th movement code of the user corresponding to the i-th kinematic evaluation performed by the user. For example, the movement code may include a first key comprising balance, core, and flexibility values, and a second key comprising posterior, vertical, lateral, anterior, and core lines values. Various embodiments may advantageously generate an individualized full-body enhancement library.

The invention discloses a robot autonomous carving operation method and system for a complex curved surface. The method comprises the steps that S1, a character pattern segmentation method is designed, and a letter segmentation information base is constructed; constructing a sub-skill primitive library; generating a plane reference trajectory by adopting a discrete dynamic motion primitive model; s2, a 3D camera is adopted to obtain point cloud of an interested complex curved surface, eye calibration outside the hand is conducted on the camera, the transformation relation from a camera coordinate system to a mechanical arm base coordinate system is obtained, and a curved surface reference track and a surface normal vector are obtained; s3, parameter identification and gravity compensation are conducted on the mechanical arm tail end carving tool based on the six-dimensional force sensor; and S4, a self-adaptive admittance controller based on the radial basis function neural network is adopted to generate an expected engraving track, so that the interaction force of the environment and the mechanical arm tracks the expected interaction force, and the engraving operation of the mechanical arm is realized. The robot can conduct carving operation on a complex curved surface, the carving precision is high, the carving effect is good, and practicability is high.