56results about How to "Eliminate jitter" patented technology

The invention provides an image optical anti-shake method for an endoscope camera system. The method comprises the following steps: acquiring motion data of an endoscope camera in real time; based on the collected motion data, using an image processing algorithm to analyze and identify the mode and degree of image shake, and calculating the compensation amount required by optical and electronic anti-shake; a high-precision optical anti-shake assembly is integrated, and according to the calculated compensation amount, an optical path or the position of a lens is rapidly adjusted through the optical anti-shake assembly so as to compensate for image offset caused by movement of the camera; an electronic anti-shake assembly is integrated, reading data of an image sensor or parameters of an image processing chip are adjusted through an electronic control system, and residual shake is further eliminated; sensor data and image quality are monitored in real time, optical anti-shake parameters are dynamically adjusted, and continuous image stability and clearness are ensured. Real-time, stable and clear output of images is achieved through the optical technology, and a better auxiliary diagnosis means can be provided for doctors.

The utility model relates to a rotary tooth skipping prevention device and a conveyor, and belongs to the technical field of conveying devices. The rotation tooth skipping prevention device comprises a U-shaped guide frame and a supporting plate, the supporting plate is arranged on the closed side of the U-shaped guide frame, and the supporting plate is used for being connected with a machine case. The conveyor comprises a machine box, a motor, a chain wheel and a chain and further comprises the rotary tooth skipping prevention device, the chain wheel is rotatably installed in the machine box, the chain is located in the machine box and surrounds the chain wheel, the motor is in transmission connection with the chain wheel, and the motor is in transmission connection with the chain wheel. A supporting plate of the rotation tooth skipping prevention device is fixed to the inner side wall of the machine box, and a U-shaped guide frame of the rotation tooth skipping prevention device is arranged on the upper side and the lower side of the chain in a half-surrounding mode. When the tooth skipping occurs, the U-shaped guide frame of the device effectively blocks the rotary part in a semi-surrounding manner, so that the motion trail of the rotary part is limited, the tooth skipping phenomenon is prevented, and the stable operation of the rotary part is ensured.

The invention provides a motor control method and device, a medium and a product, and relates to the technical field of audio noise reduction. The motor control method comprises the steps that in response to a driving instruction, current stage information is acquired, and the stage information comprises any one of a starting stage, a conventional stage and an end stage; if it is determined that the stage information meets preset conditions, parameters of a PID algorithm are adjusted according to the stage information so as to drive the direct-current motor to act, the parameters include at least one of a proportionality coefficient, an integral coefficient and a differential coefficient, and the preset conditions include that the stage information is a starting stage or an end stage of movement of the target object. According to the embodiment of the invention, the variable-speed motion of the direct-current motor can be realized, the control effect is good, the parameters of the algorithm are adjusted at the starting stage and the tail stage of the movement of the target object, the overshoot and jitter phenomena can be effectively eliminated, oscillation is not liable to occur, the operation noise is large, and the control effect is good.

The invention relates to the technical field of television image communication, in particular to an anti-explosion conference television system and a conference television implementation method. The method comprises the following steps: acquiring an original conference image stream through multi-source image acquisition, performing displacement compensation and multi-resolution fusion, generating a standardized image stream, simulating conference communication transmission, checking image feature quality, deducing equipment load, analyzing the bearing capacity of the explosion-proof conference television, and obtaining a bearing capacity change curve; and according to the curve, screening out a limit load point, carrying out load reduction segmentation, generating a low-load image stream, carrying out path redistribution on the low-load image stream, forming a path distribution matrix, and realizing real-time transmission so as to ensure effective operation of the explosion-proof conference television. According to the invention, a safer and more efficient explosion-proof conference television realization method is realized.

The invention provides a method, device and system for eliminating jitter at the tail end of a mechanical arm and a storage medium, and relates to the technical field of robot control, and the method for eliminating jitter at the tail end of the mechanical arm comprises the steps that historical motion data of the tail end of the mechanical arm to be subjected to jitter elimination is obtained and converted, and a sampling spectrogram is obtained; judging the sampling spectrogram according to a preset threshold value to obtain band-pass data; filtering the real-time motion data of the tail end of the mechanical arm to be subjected to shake elimination according to the band-pass data so as to eliminate shake of the tail end of the mechanical arm to be subjected to shake elimination in the motion process; through the jitter eliminating process, the filtered data can more accurately reflect the real motion state of the tail end of the mechanical arm, and therefore the operation precision of the mechanical arm is improved. Particularly, under the condition of fine operation or high-load working, the accuracy is crucial.

The invention discloses an abnormal behavior identification method based on a visual large model and a cognitive Agent, and relates to the technical field of computer vision, and the method comprises the steps: collecting an RGB image and an infrared thermal imaging IR image in a monitoring scene, and extracting a human body key point sequence; performing time sequence difference processing on the key point sequence, and constructing a skeleton velocity vector containing a motion trend; inputting the skeleton velocity vector into a Mamba model, adjusting model parameters by using a selective state space mechanism, and carrying out long-time-sequence modeling on the action sequence; predicting the speed variation of the next frame by adopting a residual prediction architecture and reconstructing a human body posture; and calculating an abnormal score according to the prediction error, triggering a cognitive Agent when the score exceeds a preset threshold value, calling a visual large model (VLM) to perform semantic analysis on the current scene environment, and generating a natural language warning containing an abnormal type and an environmental cause. According to the method, the forgetting problem of a traditional model in a long sequence and the jitter problem of a prediction action are solved.

This application discloses a low-power temperature-compensated RC oscillator, comprising a low-voltage linear buffer, an RC core circuit, a comparator, a Schmitt trigger, a non-overlapping clock generator, and a level shifter. The low-voltage linear buffer uses intrinsic transistors to reduce operating voltage and power consumption; the RC core circuit employs temperature compensation technology, including a series low-temperature drift resistor and a temperature-dependent current source, causing the equivalent resistance to increase with temperature; the comparator compares a triangular wave signal with a reference voltage and outputs a reference clock; the Schmitt trigger filters out clock glitches; the non-overlapping clock generator generates a two-phase clock with dead time to control charging and discharging; and the level shifter converts the low-voltage domain clock to the high-voltage domain. This oscillator eliminates the influence of comparator offset voltage through two-phase switching control and employs low-voltage and temperature compensation technology to achieve a high-precision, low-temperature drift reference frequency in a small area with low power consumption, making it suitable for compact sensing and measurement systems.

This application provides a mobile communication message transmission and reception system and method. The system includes a BBU unit, a HUB unit, and an RRU unit. The HUB unit includes a message receiving module, a first FIFO memory, a timer module, a message reading module, a message decompression module, a message merging and compression module, a second FIFO memory, and a message sending module. The timer module sends a reminder message to the message reading module when a set time is reached. The message reading module reads the message data sent by each RRU unit stored in the first FIFO memory. Upon receiving the reminder message from the timer module, it sends the currently read message data from each RRU unit to the message decompression module. In this embodiment, because a timer is used, messages can be sent to the BBU unit at a constant interval in different RRU unit relocation scenarios, eliminating message jitter and saving the capacity of the second FIFO memory.

This invention belongs to the interdisciplinary field of power management and radiation measurement, aiming to solve the problem of inaccurate counting caused by shared power supply in radiation measurement. It provides a seamless power switching device and method for radiation measurement equipment. The switching device includes a controlled power-off unit, a transient competitive power supply network, and a static path maintenance unit. The controlled power-off unit is configured to connect the external power supply to the transient competitive power supply network in non-measurement states, and to disconnect the electrical connection between the external power supply and the transient competitive power supply network upon receiving a radiation measurement trigger signal. The transient competitive power supply network is configured to automatically transfer the power supply path of the static path maintenance unit from the external power supply to the internal energy storage power supply when the output voltage of the controlled power-off unit drops below the voltage of the internal energy storage power supply. The static path maintenance unit is configured to maintain connection with the radiation measurement equipment. This invention can solve the problem of inaccurate counting in radiation measurement equipment under shared power supply scenarios.

The application relates to a video frame processing method and device, computer equipment, a storage medium and a computer program product. The method can be applied to cloud conference, cloud storage, artificial intelligence and intelligent transportation scenes, and the method comprises the following steps: determining the posture of a key part of a target object in each video frame carrying noise; determining the deflection degree value of the key part in each video frame according to the posture of the key part; generating a control signal based on the deflection degree value of the key part in each video frame; performing back gap filtering on the control signal according to a back gap filtering coefficient through a back gap filter to obtain a deflection update value of the key part in each video frame; and sequentially performing image processing on each video frame based on the deflection update value in each video frame. The method can effectively eliminate the jitter caused by noise when the video frame is subjected to image processing, and is beneficial to improving the stability of the video frame after image processing.

The invention relates to the technical field of optical engineering, and discloses a handheld laser directional energy system which comprises a Shack-Hartmann wavefront detection system, a visual observation system, an atmospheric turbulence correction system, a laser emission system, a laser ranging system and an anti-shake system. By integrating wavefront detection and adaptive correction, coaxial design, target background difference enhancement through light splitting, MEMS reflector dynamic adjustment, jitter compensation and other technologies, the problems that the imaging quality is interfered by atmospheric turbulence, correction is inaccurate due to system space separation, the contrast ratio of a dark and weak target is poor, aiming is difficult, and jitter miss is caused are solved; the effects of high-resolution imaging, energy concentration, easy target discovery, automatic locking and stable aiming are achieved.

The invention relates to a damping-adjustable hydraulic lock suitable for a heightening oil cylinder of a coal mining machine, which comprises a valve block, and a first hydraulic control one-way valve, a second hydraulic control one-way valve and a one-way adjustable throttle valve which are arranged on the valve block. An inlet of the first hydraulic control one-way valve and a hydraulic control port of the second hydraulic control one-way valve are both communicated with a first inlet of the valve block, an outlet of the first hydraulic control one-way valve is communicated with an inlet of the one-way adjustable throttling valve, and an outlet of the one-way adjustable throttling valve is communicated with a first outlet of the valve block. An inlet of the second hydraulic control one-way valve and a hydraulic control port of the first hydraulic control one-way valve are communicated with a second inlet of the valve block, and an outlet of the second hydraulic control one-way valve is communicated with a second outlet of the valve block. When the rocker arm is heightened, frequent impact and vibration between the valve element and the valve seat are avoided, so that the rocker arm is long in service life, and the rocker arm can descend stably without shaking.

The present invention belongs to the field of image acquisition technology, and solves the technical problem in the prior art that the image acquisition system uses a motor to drive a telephoto camera to aim at a target, resulting in optical path jitter and unclear imaging, and provides a device, method and electronic device for acquiring indoor documents. The device includes: a first camera mechanism, used to acquire a target image; a brushless motor, used to control the shooting angle of the first camera mechanism; a braking mechanism, which controls the brushless motor to stop rotating when the first camera mechanism is at a specified shooting angle, and when the brushless motor is in a braking state, the braking mechanism and the brushless motor both stop power output, and the optical path of the first camera mechanism to capture the target image is in a stable state. The present invention can accurately aim at indoor target document images for acquisition, and can avoid the problem of unclear imaging caused by motor jitter during acquisition, thereby achieving a high-precision document shooting effect.

The embodiment of the invention provides an electronic image stabilization method, which is used for processing visible light video data of an airborne photoelectric hanging bin, and comprises the following steps of: 1, acquiring video sequence basic information; step 2, feature point detection and motion estimation; step 3, performing abnormal processing on the motion model; step 4, motion trail smoothing processing; 5, image stabilization transformation and post-processing are carried out; and step 6, outputting a result. Aiming at visible light video data of an airborne photoelectric pod, various forms of jitter motion can be accurately captured and compensated through multi-scale feature point tracking and intelligent motion estimation. Compared with a traditional method, the method shows higher precision and stability when a complex motion mode is processed. Especially in scenes with slight motion blur, illumination variation or partial shielding, the method can still keep reliable image stabilization performance. The smooth processing of the motion trail enables the finally output video to eliminate uncomfortable jittering and maintain the intentional motion of the camera, thereby conforming to the visual perception characteristics of human beings.

The utility model provides a high-speed cable assembly and a server. The high-speed cable assembly comprises a cable piece and a connecting plug, the connecting plug comprises an AEC circuit board, the AEC circuit board comprises a board main body and a Retimer chip, a welding position is arranged on the surface of the board main body, one end of the cable piece is welded to the welding position, the Retimer chip is arranged on the board main body and electrically connected with the welding position, and the connecting plug is electrically connected with the Retimer chip. And a CDR circuit is arranged in the Regimer chip. According to the utility model, the Retimer chip is additionally arranged in the board main body, the CDR circuit is arranged in the Retimer chip, the CDR circuit can convert a digital signal and an analog signal, when an external analog signal enters the CDR circuit, the analog signal is restored to the digital signal and then changed into the analog signal again to be sent out, that is, a signal arrangement process is added to the Retimer chip, the jitter of the signal is eliminated, and the reliability of the signal processing is improved. And thus, the signal transmission distance is relatively long.

The utility model provides a constant-tension pay-off drawing mechanism for tungsten filaments, which comprises a base, a pay-off servo motor, a pay-off spool, a wire passing wheel set, a traction wheel set and a drawing die, and traction motors are arranged at the end parts of a primary traction winding wheel and a secondary traction winding wheel; the pay-off servo motor and the traction motor are both connected with the HMI setting unit and the motor driver, a PID current loop is arranged in the motor driver, and the HMI setting unit and the motor driver are both connected with the controller. The servo motor can feed back the change of the torque, at the moment, the servo control system adjusts a current loop of the paying-off servo motor in real time, stable control over the target tension is achieved, the amplitude of vibration is greatly weakened, the improvement effect is quite visual, the quality abnormity problems such as wire breakage caused by shaking are solved, and the purposes of improving the product quality and the production efficiency are achieved.

The present invention proposes an electronic mechanical brake system and discloses a control method for the electronic mechanical brake system. The method installs a force sensor on at least one wheel of a vehicle, sets a wheel with a force sensor, obtains clamping force information through the force sensor, performs contact identification in combination with the motor current, updates the contact point stroke of the piston, and calculates the piston position according to the motor angle information. For wheels without force sensors, contact calculation is performed by obtaining vehicle parameters and motor current to obtain piston position information. The clamping force information of the wheel with a force sensor and the motor parameters of the wheel with a forceless sensor are used to perform stiffness fitting to determine the estimated clamping force of other wheels. Finally, a clamping or releasing operation is performed based on the obtained information. The present invention realizes precise control of the clamping force and effective monitoring of the wear of the friction lining, thereby improving the performance and reliability of the electronic brake system.

The invention discloses an automatic wire plugging device of a flat wire motor, and relates to the field of wire plugging of the flat wire motor, and the automatic wire plugging device comprises a four-axis robot, the bottom of the four-axis robot is provided with a double-claw wire clamping mechanism, the side edge of the four-axis robot is provided with a conveying line, the top of the conveying line is provided with a wire cup, and the bottom of the wire cup is provided with a shaping structure. A jacking rotating mechanism is arranged at the bottom of the shaping structure, a shaping block is installed at the top of the shaping structure, a rotating disc is rotationally connected into an inner cavity of the shaping structure, and bosses are symmetrically installed on the two sides of the bottom of the rotating disc. According to the automatic wire plugging equipment of the flat wire motor, a complex mold is not needed, the tool design is greatly simplified, the mold manufacturing cost is greatly reduced, the flexibility is high, automatic remodeling can be achieved through program switching, a four-axis robot is used for transferring, the structural design of the whole equipment is simplified, the debugging time can be effectively shortened, and the debugging efficiency is improved. And the wire plugging efficiency and quality are greatly improved.

The utility model discloses a kind of hardware anti-shaking electronic switch circuit, including signal input filter voltage stabilizing circuit, signal anti-shaking inverting circuit and signal buffer circuit;Wherein, the input end of signal input filter voltage stabilizing circuit is connected after the input signal needing anti-shaking outside, the output end of signal input filter voltage stabilizing circuit is connected with the signal anti-shaking inverting circuit, signal buffer circuit;The output end of signal anti-shaking inverting circuit is connected with the input end of signal buffer circuit, and the output end of signal buffer circuit is connected to the external circuit of receiving anti-shaking signal.The utility model can effectively eliminate mechanical contact shaking, ensure the stability and accuracy of signal, applicable to most circuits needing signal anti-shaking.Meanwhile, circuit structure is easy to realize, and cost is low.

The invention provides a multi-sensing measurement method and system for displacement signals of a micro-resistance spot welding large cantilever electrode, and the method comprises the steps: S1, collecting sensor data in an air compression process, and calibrating the rigidity of an electrode pressurization mechanism; s2, sample welding is conducted, and dynamic signals in the welding process are collected; s3, according to the rigidity of the electrode pressurization mechanism, the pressure signal, the displacement signal and the acceleration signal are coupled to obtain a far-end displacement signal and a near-end vibration signal; and S4, performing vibration filtering on the far-end displacement signal by using the near-end vibration signal to obtain a smooth electrode displacement signal, and completing the measurement process. According to the method, the problem of sensor installation interference occurring when electrode displacement signals are measured in a near-end welding area is solved, the micro spot welding process can be monitored on line under the condition that the welding process is not interfered, the problem of signal jitter caused by insufficient rigidity of a large cantilever electrode pressurization mechanism is solved, and the precision of far-end displacement signal measurement is improved.

The present invention provides a base station front-haul interface parsing processor and a message parsing method, belonging to the field of communication technology, wherein the processor comprises a register group module composed of a plurality of registers, an operation module, a block moving module, a data recovery module and a cache, wherein the operation module processes the message data under the drive of a parsing instruction sequence when the message type of the message data is a U-plane message and generates a block moving control word and a data recovery control word, the block moving module moves the message data from the register group module based on the block moving control word and writes the message data into the cache, the data recovery module loads the message data from the cache based on the data recovery control word for recovery, and sends the recovered message data until the sending of a message data is completed. The present invention configures the control words for realizing block moving and data recovery in the processor through the operation module, constructs a new architecture for collaborative parsing of message data by ASIP and accelerator, and realizes a high-speed and high-throughput processor under the parsing of message data of the front-haul interface.

The invention provides a T trigger, a multimode frequency divider and a phase-locked loop, and relates to the technical field of electronic circuits, the output end of a first D trigger is connected with the grid electrode and the drain electrode of a first triode, the reverse output end of the first D trigger is connected with the grid electrode and the drain electrode of a second triode, and the grid electrode of a sixth triode is connected with the signal input end of the T trigger; the drain electrode and the grid electrode of the third triode are connected with the source electrode of the first triode and the inverted input control end of the T trigger, the source electrode is connected with the source electrode of a fourth triode and the drain electrode of a fifth triode, the grid electrode of the fourth triode is connected with the input control end, and the grid electrode of the fifth triode is connected to the signal inverted input end of the T trigger; the drain electrode of the tail current tube is connected with the source electrode of the sixth triode and the source electrode of the fifth triode, the grid electrode is connected with bias voltage, and the source electrode is grounded; the output ends and the reverse output ends of the two-stage first D flip-flops are correspondingly connected with the input ends and the reverse input ends of the two-stage first D flip-flops, so that multi-bit mode control can be realized, and a continuous and programmable frequency division value is realized.

A defect detection method and system for photovoltaic cells, an initial point pair set corresponding to each line laser profiler is obtained, a first correction slope corresponding to the line laser profiler is calculated, and then the initial point pair set corresponding to the line laser profiler is profile corrected to obtain a real-time point pair set corresponding to the line laser profiler; a first type of defect detection result is obtained according to the real-time point pair set corresponding to each line laser profiler; and / or a second type of defect detection result is obtained according to the real-time point pair set corresponding to each line laser profiler and the standard profile data of the three-dimensional vision sensor corresponding thereto. The error caused by mechanism jitter, installation non-parallelism and the like is suppressed or removed through profile correction and one-dimensional signal filtering and the like, so as to improve the repeatability and stability of measurement, so that the measurement data is close to the true value, and then the accuracy of the defect detection result is ensured, and on this basis, integrated measurement of various defects is realized.

The present application discloses a detection frame de-jittering method based on an optimization algorithm, and the detection frame de-jittering method comprises the following steps: inputting a video, detecting: performing object detection on each frame of the input video, tracking: using the SORT algorithm to track the detection frame, extracting all detection frames of a certain ID, where k represents the frame number, and using the Levenburg‑Marquardt algorithm to solve the minimization problem. The application is beneficial in that the second-order difference operator is used as a constraint to ensure the smoothness of the detection frames of adjacent frames. In each iteration of the optimization, the constructed matrix is ​​sparse, and the sparse algorithm is used to solve the linear equation with extremely high efficiency, which fully meets the real-time requirements. Compared with other methods, this method has good robustness. No matter how drastically the trajectory of the object changes, it can eliminate jitter well while ensuring that it will not deviate too far from the original frame.

The invention provides an anti-jitter method and device based on a vehicle-mounted pavement disease recognition system. The technical problem that jitter cannot be eliminated in a high-quality mode in the prior art is solved. The method comprises the following steps: acquiring preprocessed road picture data; converting coordinates of the preprocessed road picture data in the local coordinate system into coordinates in a camera coordinate system; extracting each feature point in the camera coordinate system; and tracking each feature point by using an optical flow estimation method to obtain a motion track of the feature point, and carrying out smoothing processing to obtain road image data after jitter elimination. The anti-jitter method can be widely applied to the technical field of anti-jitter of the vehicle-mounted platform.

The embodiment of the invention provides a short video generation method, and the method comprises the steps: automatically recognizing a character dialogue in a to-be-processed long video through a large model of automatic voice recognition and deep learning, and automatically intercepting a wonderful segment in the long video according to the semantic analysis of the character dialogue; and combining the wonderful clips, and automatically adding a transition special effect at a switching position between different clips to obtain an edited short video. According to the technical scheme provided by the embodiment of the invention, the bright spot of the video can be more accurately captured when the short video is generated, and the visual effect of the short video is improved.

The invention discloses an automatic nipple opening positioning system and guiding system for duodenal nipple intubation guidance, solves the problem of how to realize high-precision and high-robustness positioning of a nipple opening under multiple constraints, and belongs to the technical field of medical image processing and computer-aided surgical navigation. According to the invention, a generative model is utilized to actively deduce anatomical features of a shielded area, and continuous tracking in an instrument shielding scene is realized; establishing an absolute mapping relation between a pixel position and a target coordinate through injection explicit space geometry prior, and obtaining a nipple opening pixel-level coordinate; visual uncertainty estimation and instrument-tissue kinematics coupling are fused, a physically constrained adaptive filtering mechanism is constructed, and smooth coordinates are obtained; the positioning system is used for constructing a guiding system, closed-loop interaction of instrument control through a handle and inlet coordinate positioning feedback is achieved in an operation environment, and dual-scene application of preoperative training and intraoperative navigation is supported.

The invention relates to the technical field of puncture in the interventional department, and provides an intelligent depth-adjustable puncture device for the interventional department, which comprises a universal flexible arm, and one end of the universal flexible arm is provided with a fixed seat connected with external equipment and an angle adjusting assembly. The angle adjusting assembly is installed at the other end of the universal flexible arm through a connecting frame, the puncture assembly is arranged on the angle adjusting assembly, through the arrangement of the structures such as the multi-section flexible arm and the transmission line, the multi-section flexible arm is composed of a plurality of hinged joint units, and the transmission line is connected with the multi-section flexible arm. The transmission lines are symmetrically arranged on the two sides of the arm body and connected to the tail end joint, the retracting and releasing amount is accurately controlled through the first power end and the second power end, the arm body is accurately bent in a three-dimensional space, it is ensured that an end effector stably reaches a preset target spot area, the problem of path deviation caused by the fact that the bending direction of a traditional flexible pipe is uncontrollable is solved, and the bending efficiency is improved. The problems of easy deviation and insufficient stability in the prior art are solved.

The utility model provides a material belt buffer which comprises a machine frame, a material conveying device, a buffer area and a distance sensor, the material conveying device, the buffer area and the distance sensor are all arranged on the machine frame, the material conveying device is used for conveying a material belt from an uncoiler to the buffer area, and the buffer area is used for buffering the material belt and used for being connected to a punching machine of a follow-up process. The distance sensor is used for detecting the relative distance between the distance sensor and the material belt in the cache area so as to provide a corresponding electric signal for an external controller, and the external controller is promoted to control the driving speed of the material conveying device. The buffering area is arranged for buffering the material belt, the distance of the material belt is recognized through the distance sensor, and therefore the stock of the material belt in the buffering area is judged, when too many materials are buffered, the driving force of the material conveying device is reduced, when the stock is too small, the driving force of the material conveying device is increased, and the material conveying device is linked with a punching machine to protect material belt raw materials; a buffering effect is achieved between the uncoiler and the punching machine, shaking from the punching machine can be eliminated, the laser coding effect is guaranteed, and the product quality is improved.

This invention discloses a method for generating smooth trajectories for robotic arms that combines visual keypoint cues with a hierarchical diffusion model, applicable to complex robotic arm operations based on visual and verbal commands. The method constructs a hierarchical robotic arm operation framework. The high-level layer utilizes a sub-target generator guided by a visual-language model to predict the six-DOF pose sequence of the end effector as operation sub-targets from visual observations and verbal commands. The low-level layer proposes a dual diffusion model trained based on a random sampling strategy to generate smooth, low-curvature joint position trajectories between sub-targets. A keypoint extraction module that requires no training is introduced to provide visual cues. This invention improves the success rate of operations, reduces trajectory jitter and redundant motion, and enhances the smoothness and execution stability of robotic arm operations through the synergy of hierarchical planning and diffusion trajectory optimization.