Gimbal stabilization method and device, electronic equipment and computer readable storage medium

Abstract

The application discloses a gimbal stabilization method and device, electronic equipment and computer readable storage medium, and relates to the technical field of gimbal control. The method comprises the following steps: acquiring the current distance sequence of the gimbal to the ground, accurately acquiring the current vertical acceleration and the predicted vertical displacement, and providing high-confidence preview information for subsequent control; calculating the target weight coefficient according to the current vertical acceleration and the predicted vertical displacement, constructing a variable weight cost function, and enabling the gimbal control quantity to be self-adaptive to different vibration intensities; calculating the three-axis feedforward compensation quantity based on the predicted vertical displacement, combining the preset three-axis tracking error to calculate the three-axis control quantity, and realizing the physical level active suppression of disturbance; and finally fusing the gimbal control quantity and the three-axis control quantity to realize gimbal stabilization, realize the adaptive control strategy of different jolting working conditions and the active cancellation of disturbance force, significantly improve the low-frequency jitter suppression capability, reduce the control bandwidth requirement and energy consumption, and improve the stability, responsiveness and energy efficiency balance.

Description

Technical Field

[0001] This invention relates to the field of gimbal control technology, and in particular to a gimbal stabilization method, device, electronic device, and computer-readable storage medium. Background Technology

[0002] With the popularization of mobile photography and short video creation, handheld gimbals, as key stabilization devices, directly impact image quality. Traditional handheld gimbals typically employ feedback control systems based on inertial measurement units (IMUs), which compensate for changes in device attitude by driving motors. However, this system has inherent latency, often resulting in high-frequency vibrations and uneven motion in fast-moving or complex, bumpy environments. While existing technologies employ vibration suppression methods such as improved PID algorithms and adaptive notch filters, they are essentially still passive response control systems, unable to predict future motion states and vibration changes. Consequently, they perform poorly in environments with sudden movements or complex vibration spectra, making handheld gimbal technology particularly weak in vertical stabilization. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to overcome the shortcomings of the prior art and provide a gimbal stabilization method, apparatus, electronic device and computer-readable storage medium.

[0004] This invention provides the following technical solution: In a first aspect, the present invention proposes a gimbal stabilization method, the method comprising: Obtain the current distance sequence from the gimbal to the ground, and obtain the current vertical acceleration and predicted vertical displacement based on the current distance sequence; The target weight coefficient is calculated based on the current vertical acceleration and the predicted vertical displacement. The variable weight cost function is then optimally solved based on the target weight coefficient to obtain the gimbal control quantity. The three-axis feedforward compensation is calculated based on the predicted vertical displacement, and the three-axis control quantity is calculated based on the three-axis feedforward compensation and the preset three-axis tracking error. Gimbal stabilization is performed based on gimbal control parameters and three-axis control parameters.

[0005] In one embodiment, the target weighting coefficient includes displacement weight, velocity weight, and control quantity weight. The target weighting coefficient is calculated based on the current vertical acceleration and the predicted vertical displacement, including: The aiming velocity is calculated based on the predicted vertical displacement. The first proportional factor is obtained based on the first preset adjustment coefficient and the square of the aiming velocity. The displacement weight is calculated based on the first preset reference weight and the first proportional factor. The second proportional factor is obtained based on the second preset adjustment coefficient and the absolute value of the current vertical acceleration. The velocity weight is calculated based on the second preset benchmark weight and the second proportional factor. The aiming acceleration is calculated based on the predicted vertical displacement. The third proportional factor is calculated based on the third preset adjustment coefficient and the square of the aiming acceleration. The control weight is calculated based on the third preset reference weight and the reciprocal of the third proportional factor.

[0006] In one embodiment, the target weight coefficient further includes an acceleration weight, and the method further includes: Calculate the square of the displacement difference between the gimbal body displacement and the gimbal reference displacement, and calculate the displacement tracking error term based on the displacement weight and the square of the displacement difference; The main body speed term is calculated based on the speed weight and the square of the gimbal main body speed; The main acceleration term is calculated based on the acceleration weight and the square of the gimbal's main body acceleration; Calculate the control energy consumption item based on the control quantity weight and the gimbal control quantity; The variable-weight cost function is obtained by accumulating and integrating the displacement tracking error term, the main velocity term, the main acceleration term, and the energy consumption term.

[0007] In one embodiment, the three-axis feedforward compensation includes vertical feedforward compensation and horizontal feedforward compensation; the calculation of the three-axis feedforward compensation based on the predicted vertical displacement includes: Differentiating the predicted vertical displacement yields the predicted vertical velocity and the predicted vertical acceleration. Obtain the current horizontal parameters, calculate the predicted horizontal displacement based on the current horizontal parameters, and differentiate the predicted horizontal displacement to obtain the predicted horizontal velocity and predicted horizontal acceleration. Based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted vertical velocity, predicted vertical acceleration and their corresponding prediction time, the vertical inertial force overcoming term, the vertical damping force overcoming term and the vertical spring restoring force overcoming term are calculated; The vertical feedforward compensation is obtained by summing the vertical inertial force overcoming term, the vertical damping force overcoming term, and the vertical spring restoring force overcoming term; Based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted horizontal velocity, predicted horizontal acceleration and their corresponding prediction time, the horizontal inertial force overcoming term, the horizontal damping force overcoming term and the horizontal spring restoring force overcoming term are calculated. The horizontal feedforward compensation is obtained by summing the horizontal inertial force overcoming term, the horizontal damping force overcoming term, and the horizontal spring restoring force overcoming term.

[0008] In one embodiment, the current horizontal parameters include the current horizontal acceleration, the current horizontal velocity, and the current horizontal position. Calculating the predicted horizontal displacement based on the current horizontal parameters includes: Based on Newtonian kinematics, the predicted horizontal displacement is calculated using the current horizontal acceleration, current horizontal velocity, current horizontal position, and prediction time.

[0009] In one embodiment, the three-axis control quantity is calculated based on the three-axis feedforward compensation amount and the preset three-axis tracking error, including: The rate of change of the preset triaxial tracking error is obtained by differentiating the preset triaxial tracking error. The three-axis control quantity is obtained by weighting and summing the preset three-axis tracking error, rate of change, and three-axis feedforward compensation amount based on the preset proportional gain, preset differential gain, and preset aiming feedforward gain.

[0010] In one embodiment, obtaining the current vertical acceleration and predicted vertical displacement based on the current distance sequence includes: Denoise the current distance sequence to obtain the denoised distance sequence; Acquire accelerometer data, and calculate the current vertical acceleration based on the accelerometer data and the denoised distance sequence; The vertical velocity estimate is obtained by performing differential processing on the denoised distance sequence. The vertical displacement is predicted based on the denoised distance sequence, the current vertical acceleration, and the estimated vertical velocity.

[0011] Secondly, the present invention provides a gimbal stabilization device, the device comprising: The acquisition module is used to acquire the current distance sequence from the gimbal to the ground, and to obtain the current vertical acceleration and predicted vertical displacement based on the current distance sequence; The first calculation module is used to calculate the target weight coefficient based on the current vertical acceleration and the predicted vertical displacement, and to perform optimal solution on the variable weight cost function based on the target weight coefficient to obtain the gimbal control quantity; The second calculation module is used to calculate the three-axis feedforward compensation amount based on the predicted vertical displacement, and to calculate the three-axis control amount based on the three-axis feedforward compensation amount and the preset three-axis tracking error. The stabilization module is used to stabilize the gimbal based on the gimbal control input and the three-axis control input.

[0012] Thirdly, the present invention proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the gimbal stabilization method as described in the first aspect.

[0013] Fourthly, the present invention proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the gimbal stabilization method as described in the first aspect.

[0014] The gimbal stabilization method, device, electronic equipment, and computer-readable storage medium disclosed in this invention acquire the current distance sequence from the gimbal to the ground, accurately obtain the current vertical acceleration and predicted vertical displacement, and provide high-confidence pre-aiming information for subsequent control; calculate the target weight coefficient based on the current vertical acceleration and predicted vertical displacement, and construct a variable weight cost function so that the gimbal control quantity can adapt to different vibration intensities; calculate the three-axis feedforward compensation quantity based on the predicted vertical displacement, and calculate the three-axis control quantity in combination with the preset three-axis tracking error to achieve physical-level active suppression of disturbances; finally, integrate the gimbal control quantity and the three-axis control quantity to stabilize the gimbal, so that the control strategy can adapt to different turbulence conditions and actively cancel disturbance forces, significantly improve the mid-to-low frequency jitter suppression capability, reduce control bandwidth requirements and energy consumption, and improve the balance between stability, responsiveness and energy efficiency. Attached Figure Description

[0015] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope of protection of the present invention. In the various drawings, similar components are numbered similarly.

[0016] Figure 1 A flowchart of the gimbal stabilization method proposed in this embodiment is shown; Figure 2 A schematic diagram of the handheld three-axis gimbal system proposed in this embodiment is shown; Figure 3 Another flowchart of the gimbal stabilization method proposed in this embodiment is shown; Figure 4 This diagram illustrates another step in the process of the gimbal stabilization method proposed in this embodiment. Figure 5 Another schematic diagram of the gimbal stabilization method proposed in this embodiment is shown; Figure 6 A schematic diagram of the gimbal stabilization device proposed in this embodiment is shown.

[0017] Explanation of reference numerals in the attached diagram: 200 - Handheld three-axis gimbal system; 201 - Sensor module; 2011 - Pre-aiming sensor; 2012 - Inertial measurement unit; 2013 - Position sensor; 202 - Control module; 2021 - Main controller; 2022 - Communication interface; 2023 - Storage unit; 203 - Execution module; 2031 - FOC driver; 600 - Gimbal stabilization device; 601 - Acquisition module; 602 - First calculation module; 603 - Second calculation module; 604 - Stabilization module. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0020] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0021] Furthermore, the terms "first," "second," and "third" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0022] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0023] Example 1 This disclosure provides a gimbal stabilization method for acquiring pre-aiming information and combining it with a variable weight cost function to actively suppress vertical vibrations and improve the stability of the gimbal in turbulent scenarios.

[0024] Please see Figure 1 The gimbal stabilization method includes steps S101 to S104; please refer to [link / reference]. Figure 2The data processing in steps S101 to S104 is handled by the main controller 2021 in the control module 202 of the handheld three-axis gimbal system 200. The processed data and signals are transmitted and stored through the communication interface 2022 and the storage unit 2023. The execution module 203 then executes the commands to achieve predictive stabilization of the gimbal. The following provides a detailed explanation of each step.

[0025] Step S101: Obtain the current distance sequence from the gimbal to the ground, and obtain the current vertical acceleration and predicted vertical displacement based on the current distance sequence.

[0026] In this embodiment, please refer to Figure 2 The distance sequence from the gimbal to the ground is obtained through the pre-aiming sensor 2011 in the sensor module 201 of the handheld three-axis gimbal system 200, such as a lidar. In the formula, for t Distance values ​​measured at any given time For the actual distance, For measuring noise.

[0027] Furthermore, the current vertical acceleration and predicted vertical displacement are calculated using the current distance sequence. The predicted vertical displacement is used as pre-aiming information to provide a reliable input basis for subsequent adaptive weighting and physical feedforward.

[0028] Please see Figure 3 In one specific embodiment, step S101 includes steps S1011 to S1014, and each step is described in detail below.

[0029] Step S1011: Denoise the current distance sequence to obtain the denoised distance sequence.

[0030] In this embodiment, the current distance sequence is denoised using an adaptive Kalman filter. The denoised distance sequence includes: In the formula, for t The estimated value of the true distance after time-to-time noise reduction. Here is the state transition matrix. For the observation matrix, For adaptive Kalman gain.

[0031] Unlike the standard Kalman filter, Adaptive adjustment based on the gimbal's motion status: In the formula, To estimate the covariance a priori, The time-varying noise covariance is dynamically adjusted according to the current intensity of the motion, i.e., it increases when the motion is intense. This increases the system's confidence in new measurements.

[0032] Understandably, by introducing a time-varying noise covariance matrix and adaptively adjusting the Kalman gain based on the intensity of motion, the robustness of the filter is significantly improved—enhancing the confidence of new observations and suppressing oversmoothing when the gimbal is jittering at high speed (such as running or off-roading); and reducing the risk of noise amplification when holding the device steadily.

[0033] It should be noted that the state transition matrix F This model is derived from the system's motion model. In this embodiment, a uniform acceleration motion model is used, but considering the randomness of the handheld gimbal's motion, jerk is introduced as a state variable to form a uniform acceleration model (i.e., a constant jerk model).

[0034] The state vector is: In the formula, For vertical displacement, Vertical velocity, Vertical acceleration, This is the vertical jerk.

[0035] According to Newtonian kinematics, at the sampling time (Determined by the system sampling frequency, for example, 500Hz corresponds to) Within this range, the state transition relationship is: .

[0036] Therefore, the state transition matrix F is: .

[0037] Observation matrix H A relationship between the state vector and the observation vector was established. The observation vector includes displacement, velocity, and acceleration measured by the inertial measurement unit 2012 (three-axis gyroscope and three-axis accelerometer), and displacement measured by the lidar. H for: .

[0038] Assume a linear relationship exists between the observed values ​​and the state values, and that the observation noise is zero-mean Gaussian white noise. Therefore, the observation matrix... H Designed as follows: In the formula, the observation matrix H First line: IMU displacement observations corresponding to state displacement ; Observation matrix H Second line: IMU velocity observation corresponding to state velocity ; Observation matrix H Third line: IMU acceleration observations corresponding to state acceleration ; Observation matrix HFourth line: State displacement corresponding to lidar displacement observation In this context, it is assumed that both the IMU and lidar displacement observations are observations of the same state displacement, and therefore both are directly related to the state displacement.

[0039] Prior estimation of covariance : , In the formula, It is the posterior estimate of the covariance from the previous time step. It is the identity matrix. Let be the Kalman gain at time t-1.

[0040] Time-varying noise covariance The covariance matrix represents the observation noise, with its diagonal elements being the noise variance of each observation. It varies over time and adjusts according to the sensor's operating state and external environment. Time-varying noise covariance. expression: In the formula, The variance of IMU displacement observation noise can be estimated by the variance of the displacement observation data of the IMU in a stationary state; The variance of IMU velocity observation noise is also estimated using the variance of velocity observation data under stationary conditions. The variance of IMU acceleration observation noise is estimated by using the variance of acceleration observation data under stationary conditions; The noise variance of lidar ranging is estimated by measuring the variance of a fixed distance under stationary conditions.

[0041] Step S1012: Obtain accelerometer data and calculate the current vertical acceleration based on the accelerometer data and the denoised distance sequence.

[0042] In this embodiment, the current vertical acceleration is calculated by fusing accelerometer data acquired from a triaxial accelerometer with a denoised distance sequence. .

[0043] Understandably, through tight coupling fusion under state-space constraints, the current vertical acceleration estimate can possess both high-frequency responsiveness (from the accelerometer) and long-term stability (from the distance observation), providing a reliable transient excitation intensity characterization for subsequent dynamic weight adjustment.

[0044] Step S1013: Perform differential processing on the denoised distance sequence to obtain the vertical velocity estimate.

[0045] In this embodiment, the vertical velocity estimate is... The calculation formula is: , The sampling interval is denoted as .

[0046] Understandably, the first-order forward difference can obtain an initial velocity value with a signal-to-noise ratio significantly better than that of the IMU velocity integral. This predicted value is used as an intermediate variable to build a more accurate prediction model and avoid the divergence of predicted displacement due to velocity error propagation.

[0047] Step S1014: Calculate and predict the vertical displacement based on the denoised distance sequence, the current vertical acceleration, and the estimated vertical velocity.

[0048] In this embodiment, vertical displacement is predicted. The calculation formula is: In the formula, The aiming time is adaptively adjusted based on the dominant vibration frequency.

[0049] Understandably, the prediction time is adaptively adjusted according to the dominant vibration frequency (e.g., long prediction for low-frequency turbulence and short prediction for high-frequency jitter), combined with a uniform acceleration (jerk) motion model, so that the predicted vertical displacement accurately reflects the actual disturbance trajectory of the gimbal platform at future moments, rather than empirical extrapolation.

[0050] Step S102: Calculate the target weight coefficient based on the current vertical acceleration and the predicted vertical displacement, and perform optimal solution on the variable weight cost function based on the target weight coefficient to obtain the gimbal control quantity.

[0051] In this embodiment, considering the highly random nature of handheld gimbal motion, the target weight coefficient is calculated using the current vertical acceleration and predicted vertical displacement, and the weights of each performance index are dynamically adjusted. Furthermore, the gimbal control quantity is optimally solved for the variable weight cost function based on the target weight coefficient, thereby achieving the optimization of tracking accuracy, response speed, and energy consumption cost based on real-time vibration characteristics. Under strong disturbances, displacement tracking is prioritized, while under weak disturbances, energy consumption and comfort are considered, achieving Pareto optimal control under all operating conditions.

[0052] Please see Figure 4 In one specific embodiment, the target weight coefficient includes displacement weight, velocity weight and control quantity weight. Step S102 includes steps S1021 to S1023. Each step is described in detail below.

[0053] Step S1021: Calculate the aiming velocity based on the predicted vertical displacement, obtain the first proportional factor based on the first preset adjustment coefficient and the square of the aiming velocity, and calculate the displacement weight based on the first preset reference weight and the first proportional factor.

[0054] In this embodiment, the formula for calculating the displacement weight is: In the formula, Pre-aiming speed (reflecting the intensity of the movement). As the first scaling factor, For displacement weights, As the first preset benchmark weight, This is the first preset adjustment coefficient. To predict vertical displacement, For time. The displacement weight is positively correlated with the square of the aiming velocity, reflecting the strong tracking requirement for large-stroke disturbances.

[0055] It should be noted that the first preset benchmark weight The adjustment under stable motion was determined through nominal operating condition tests. Find the balance point between tracking error and control energy. First preset adjustment coefficient. It is determined by analyzing the degree of influence of the velocity square term on performance. For example, .

[0056] Step S1022: Obtain the second proportional factor based on the second preset adjustment coefficient and the absolute value of the current vertical acceleration, and calculate the velocity weight based on the second preset benchmark weight and the second proportional factor.

[0057] In this embodiment, the formula for calculating the velocity weight is: In the formula, As the second scaling factor, For speed weights, As the second preset benchmark weight, This is the second preset adjustment coefficient. This represents the current vertical acceleration (reflecting the intensity of the impact). The velocity weight is positively correlated with the absolute value of the current vertical acceleration, reflecting the need for rapid suppression of sudden impacts.

[0058] It should be noted that the second preset benchmark weight Frequency domain analysis was used to determine and suppress vibrations in a specific frequency range; a second preset adjustment coefficient was used. The required damping characteristics are determined by analyzing different accelerations.

[0059] Step S1023: Calculate the aiming acceleration based on the predicted vertical displacement, calculate the third proportional factor based on the third preset adjustment coefficient and the square of the aiming acceleration, and calculate the control weight based on the third preset reference weight and the reciprocal of the third proportional factor.

[0060] In this embodiment, the formula for calculating the control quantity weight is: In the formula, It is the square of the aiming acceleration (reflecting the degree of abrupt change in motion). As the third proportional factor, To control the weight of quantities, As the third preset benchmark weight, This is the third preset adjustment coefficient. The control quantity weight is positively correlated with the inverse of the aiming acceleration, reflecting the forward-looking suppression of the actuator saturation risk under high-frequency jitter.

[0061] It should be noted that the third preset benchmark weight Based on the maximum torque of the motor and the system bandwidth, the calculation formula is as follows: Third preset adjustment coefficient It is determined by analyzing the response requirements of the control variable to sudden motion changes.

[0062] Understandably, when a violent vertical movement is detected, the system automatically increases the displacement weight. and speed weight At the same time, reduce the control quantity weight. This improves the system's ability to suppress vibration; conversely, during steady motion, it saves energy and reduces noise.

[0063] In one specific embodiment, the target weight coefficient further includes an acceleration weight. The method further includes: calculating the square of the displacement difference between the gimbal body displacement and the gimbal reference displacement; calculating a displacement tracking error term based on the displacement weight and the square of the displacement difference; calculating a body velocity term based on the velocity weight and the square of the gimbal body velocity; calculating a body acceleration term based on the acceleration weight and the square of the gimbal body acceleration; calculating a control energy consumption term based on the control quantity weight and the gimbal control quantity; and accumulating and integrating the displacement tracking error term, body velocity term, body acceleration term, and energy consumption term to obtain a variable weight cost function.

[0064] In this embodiment, the expression for the variable weight cost function is: In the formula, For displacement tracking error term, For the displacement of the gimbal body, For gimbal reference displacement, The main velocity term, For the speed of the gimbal body, As the main acceleration term, Accelerate the main body of the gimbal. To control energy consumption items, For gimbal control parameters, This represents the acceleration weight. However, in practical applications, acceleration is generally not directly controlled here. The value is 0.

[0065] Understandably, by substituting each known variable into the variable weight cost function, the optimal gimbal control quantity is obtained, thereby improving the overall system performance. Minimum.

[0066] Step S103: Calculate the three-axis feedforward compensation amount based on the predicted vertical displacement, and calculate the three-axis control amount based on the three-axis feedforward compensation amount and the preset three-axis tracking error.

[0067] In this embodiment, the three-axis inertial-damping-elastic full compensation torque covering the vertical and horizontal directions is calculated based on the predicted vertical displacement to obtain the three-axis feedforward compensation amount. The three-axis control amount is further calculated using the three-axis feedforward compensation amount and the preset three-axis tracking error, which fundamentally offsets the dynamic reaction force generated by the load disturbance and significantly reduces the tracking error bandwidth pressure of the closed-loop controller.

[0068] Please see Figure 5 In one specific embodiment, the three-axis feedforward compensation includes vertical feedforward compensation and horizontal feedforward compensation. Step S103 includes S1031 to S1033. Each step is described in detail below.

[0069] Step S1031: Differentiate the predicted vertical displacement to obtain the predicted vertical velocity and predicted vertical acceleration; obtain the current horizontal parameters, calculate the predicted horizontal displacement based on the current horizontal parameters, and differentiate the predicted horizontal displacement to obtain the predicted horizontal velocity and predicted horizontal acceleration.

[0070] In this embodiment, the predicted vertical displacement is differentiated once and twice to obtain the predicted vertical velocity and predicted vertical acceleration, respectively.

[0071] Simultaneously, based on Newtonian kinematics, the predicted horizontal displacement is obtained by extrapolating from the current horizontal parameters, ensuring that the horizontal and vertical predictions maintain dynamic consistency in time and amplitude, and avoiding phase conflicts caused by inter-axis decoupling in feedforward compensation. The predicted horizontal displacement includes the prediction of… Axial displacement and prediction Axial displacement, current horizontal parameters include current horizontal acceleration, current horizontal velocity, and current horizontal position; current horizontal acceleration includes current... Axis acceleration, current Axial acceleration, current horizontal velocity including current Axis speed, current Axis speed, current horizontal position including current Axis position, current Axis position.

[0072] Exemplary, predictive Axial displacement The calculation formula is: In the formula, and Each is the current Axis acceleration, current Axial acceleration, measured by an IMU; and Each is the current Axis speed, current Axial velocity is obtained by integrating acceleration. and Each is the current Axis position, current The axis position is obtained by double integration of acceleration.

[0073] Furthermore, regarding prediction Axial displacement and The axial displacement is differentiated first and second, respectively, to obtain the predicted values. Axial velocity and prediction Axial acceleration, and prediction Axial velocity and prediction Axial acceleration.

[0074] Step S1032: Based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted vertical velocity, predicted vertical acceleration and their corresponding prediction time, calculate the vertical inertial force overcoming term, vertical damping force overcoming term and vertical spring restoring force overcoming term; based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted horizontal velocity, predicted horizontal acceleration and their corresponding prediction time, calculate the horizontal inertial force overcoming term, horizontal damping force overcoming term and horizontal spring restoring force overcoming term.

[0075] In this embodiment, the expression for overcoming the vertical inertial force is: The expression for the term that overcomes the vertical damping force is: The expression for the vertical spring restoring force overcoming term is: In the formula, The equivalent mass is preset (obtained through static experiments or dynamic modeling). The damping ratio (determined through step response experiments, typically 0.7~1.0). The vertical natural frequency (determined by measuring the resonant frequencies of each axis through system frequency response experiments).

[0076] The terms for overcoming horizontal inertial forces include The term to overcome axial inertial force and The term for overcoming inertial forces in the axial direction. The expression for overcoming the axial inertial force is: , The expression for overcoming the axial inertial force is: .

[0077] The horizontal damping force overcoming terms include Axial damping force overcoming term and Axial damping force overcoming term. The expression for the term that overcomes the axial damping force is: , The expression for the term that overcomes the axial damping force is: .

[0078] The horizontal spring restoring force overcoming items include Axial direction spring restoring force overcoming term and The term for overcoming the spring restoring force in the axial direction. The expression for the term that overcomes the spring restoring force in the axial direction is: , The expression for the term that overcomes the spring restoring force in the axial direction is: .

[0079] In the formula, , They are respectively axial natural frequency and Natural frequency along the axis.

[0080] Step S1033: The vertical feedforward compensation is obtained by summing the vertical inertial force overcoming term, the vertical damping force overcoming term, and the vertical spring restoring force overcoming term; the horizontal feedforward compensation is obtained by summing the horizontal inertial force overcoming term, the horizontal damping force overcoming term, and the horizontal spring restoring force overcoming term.

[0081] In this embodiment, the vertical feedforward compensation amount The calculation formula is: .

[0082] Horizontal feedforward compensation includes Axial direction feedforward compensation and Axial direction feedforward compensation. Axial direction feedforward compensation The expression is: ; Axial direction feedforward compensation The expression is: .

[0083] In one specific embodiment, step S103 includes: differentiating the preset three-axis tracking error to obtain the rate of change of the preset three-axis tracking error; and weighting and summing the preset three-axis tracking error, the rate of change, and the three-axis feedforward compensation amount based on the preset proportional gain, the preset differential gain, and the pre-aiming feedforward gain to obtain the three-axis control amount.

[0084] In this embodiment, the formula for calculating the three-axis control quantity is: , , , These are three-axis control variables. , , For three-axis tracking error, , For the proportional gain matrix and the differential gain matrix, This is the aiming feedforward gain matrix.

[0085] The tracking error is generally calculated by the deviation between the actual position and the desired position. In a three-axis gimbal system, the real-time measured actual position (or attitude) of each axis is typically obtained through the Hall element and permanent magnet ring in the position sensor 2013, and the difference is calculated between this and the preset desired position (or reference trajectory). =Desired position x Actual position x, =Desired position y Actual position y, =Desired position z Actual position z.

[0086] It should be noted that, The proportional gain matrix is ​​determined through system identification or frequency domain response experiments. It is typically achieved using pole placement methods or based on LQR (linear quadratic regulator) design, ensuring that the closed-loop poles of the system are located in the desired region, balancing response speed and stability.

[0087] The (differential gain matrix) is determined in conjunction with the system's damping characteristics. The desired damping ratio (e.g., ζ=0.7) can be set by experimentally measuring the system's open-loop frequency response. 1.0ζ=0.7 (1.0), and then deduce the differential gain. It can also be obtained through simulation optimization.

[0088] (The feedforward gain matrix) is designed based on the system's inverse dynamics model. Ideally, Kpre should match the feedforward control quantity with the system's dynamic model, thus achieving complete compensation. In practice, adjustments are made through simulation or experiments to ensure that the feedforward term effectively reduces tracking error. Common methods include model matching, least squares identification, or adaptive feedforward control.

[0089] Step S104: Perform gimbal stabilization based on gimbal control values ​​and three-axis control values.

[0090] In this embodiment, control commands are generated based on gimbal control quantities and three-axis control quantities. The two control commands, which are from different sources and have complementary functions, are synergistically fused and applied to the actuators such as the three-axis brushless motors (pitch axis motor, roll axis motor, and yaw axis motor) in the FOC driver 2031 to achieve a comprehensive stabilization effect with high precision, low latency, and strong robustness.

[0091] Understandably, the gimbal control input is the globally optimal control input for the kinematic model of the gimbal itself, focusing on dynamically balancing displacement tracking accuracy, velocity responsiveness, acceleration smoothness, and control energy consumption under the current vibration conditions, ensuring that the overall attitude trajectory of the gimbal conforms to the preset reference law; while the three-axis control input is a dynamic compensation command for external load disturbances, focusing on actively counteracting the vertical inertial force, damping force, and elastic restoring force caused by ground bumps, and synchronously generating horizontal compensation components along the three-axis coupling propagation path, thereby suppressing the transmission of disturbances from the source.

[0092] The gimbal control variables form a stable reference layer, ensuring the overall convergence and comfort of the gimbal's movement; the three-axis control variables form a disturbance suppression layer, undertaking the rapid open-loop response to high-frequency / sudden disturbances. In the final execution stage, the system maps the two sets of control variables to the corresponding motor drivers along the axes. After multi-level servo adjustment through current loop, speed loop, and position loop, the brushless motor is driven to output precise torque, correcting the gimbal's attitude deviation in real time. This ensures that the mounted camera device always maintains spatial pointing stability, effectively eliminating image shaking, ghosting, and defocusing caused by hand-held shakiness, terrain undulations, or motion impacts, ultimately achieving the goal of all-scene, adaptive, and high-fidelity video stabilization.

[0093] The gimbal stabilization method proposed in this embodiment obtains the current vertical acceleration and predicted vertical displacement by acquiring the current distance sequence from the gimbal to the ground, providing high-confidence pre-aiming information for subsequent control. Based on the current vertical acceleration and predicted vertical displacement, target weight coefficients are calculated to construct a variable-weight cost function, enabling the gimbal control quantity to adapt to different vibration intensities. Three-axis feedforward compensation is calculated based on the predicted vertical displacement, and three-axis control quantity is calculated in conjunction with a preset three-axis tracking error, achieving physical-level active suppression of disturbances. Finally, the gimbal control quantity and three-axis control quantity are fused for gimbal stabilization, enabling the control strategy to adapt to different turbulent conditions and actively cancel disturbance forces, significantly improving the ability to suppress mid-to-low frequency jitter, reducing control bandwidth requirements and energy consumption, and improving the balance between stability, responsiveness, and energy efficiency.

[0094] Example 2 Furthermore, this disclosure provides a gimbal stabilization device 600, please refer to [link to relevant documentation]. Figure 6 The device includes: The acquisition module 601 is used to acquire the current distance sequence from the gimbal to the ground, and to acquire the current vertical acceleration and predicted vertical displacement based on the current distance sequence; The first calculation module 602 is used to calculate the target weight coefficient based on the current vertical acceleration and the predicted vertical displacement, and to perform optimal solution on the variable weight cost function based on the target weight coefficient to obtain the gimbal control quantity; The second calculation module 603 is used to calculate the three-axis feedforward compensation amount based on the predicted vertical displacement, and to calculate the three-axis control amount based on the three-axis feedforward compensation amount and the preset three-axis tracking error. The stabilization module 604 is used to stabilize the gimbal based on the gimbal control input and the three-axis control input.

[0095] Optionally, the target weight coefficients include displacement weight, velocity weight, and control weight. The first calculation module 602 is further configured to calculate the aiming velocity based on the predicted vertical displacement, obtain a first proportional factor based on a first preset adjustment coefficient and the square of the aiming velocity, calculate the displacement weight based on a first preset reference weight and the first proportional factor; obtain a second proportional factor based on a second preset adjustment coefficient and the absolute value of the current vertical acceleration, calculate the velocity weight based on the second preset reference weight and the second proportional factor; calculate the aiming acceleration based on the predicted vertical displacement, calculate a third proportional factor based on a third preset adjustment coefficient and the square of the aiming acceleration, and calculate the control weight based on the third preset reference weight and the reciprocal of the third proportional factor.

[0096] Optionally, the target weighting coefficient also includes acceleration weight. The first calculation module 602 is also used to calculate the square of the displacement difference between the gimbal body displacement and the gimbal reference displacement, calculate the displacement tracking error term based on the displacement weight and the square of the displacement difference; calculate the body velocity term based on the velocity weight and the square of the gimbal body velocity; calculate the body acceleration term based on the acceleration weight and the square of the gimbal body acceleration; calculate the control energy consumption term based on the control quantity weight and the gimbal control quantity; and accumulate and integrate the displacement tracking error term, body velocity term, body acceleration term, and energy consumption term to obtain the variable weight cost function.

[0097] Optionally, the three-axis feedforward compensation includes vertical feedforward compensation and horizontal feedforward compensation; the second calculation module 603 is also used to differentiate the predicted vertical displacement to obtain the predicted vertical velocity and predicted vertical acceleration; obtain the current horizontal parameters, calculate the predicted horizontal displacement based on the current horizontal parameters, differentiate the predicted horizontal displacement to obtain the predicted horizontal velocity and predicted horizontal acceleration; and calculate the vertical inertial force overcoming term and vertical resistance term based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted vertical velocity, predicted vertical acceleration and their corresponding prediction time. The vertical inertial force overcoming term and the vertical spring restoring force overcoming term are calculated. The vertical feedforward compensation is obtained by summing the vertical inertial force overcoming term, the vertical damping force overcoming term, and the vertical spring restoring force overcoming term. Based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted horizontal velocity, predicted horizontal acceleration, and their corresponding prediction time, the horizontal inertial force overcoming term, the horizontal damping force overcoming term, and the horizontal spring restoring force overcoming term are calculated. The horizontal feedforward compensation is obtained by summing the horizontal inertial force overcoming term, the horizontal damping force overcoming term, and the horizontal spring restoring force overcoming term.

[0098] Optionally, the current horizontal parameters include the current horizontal acceleration, the current horizontal velocity, and the current horizontal position. The second calculation module 603 is also used to calculate the predicted horizontal displacement based on Newtonian kinematics, according to the current horizontal acceleration, the current horizontal velocity, the current horizontal position, and the prediction time.

[0099] Optionally, the second calculation module 603 is also used to differentiate the preset three-axis tracking error to obtain the rate of change of the preset three-axis tracking error; and to perform a weighted summation of the preset three-axis tracking error, the rate of change, and the three-axis feedforward compensation based on the preset proportional gain, the preset differential gain, and the pre-aiming feedforward gain to obtain the three-axis control quantity.

[0100] Optionally, the acquisition module 601 is also used to denoise the current distance sequence to obtain a denoised distance sequence; acquire accelerometer data, calculate the current vertical acceleration based on the accelerometer data and the denoised distance sequence; perform differential processing on the denoised distance sequence to obtain a vertical velocity estimate; and calculate and predict the vertical displacement based on the denoised distance sequence, the current vertical acceleration, and the vertical velocity estimate.

[0101] The apparatus provided in this embodiment can perform the steps of the gimbal stabilization method provided in Embodiment 1. To avoid repetition, it will not be described again.

[0102] The gimbal stabilization device proposed in this embodiment obtains the current vertical acceleration and predicted vertical displacement by acquiring the current distance sequence from the gimbal to the ground, providing high-confidence pre-aiming information for subsequent control. Based on the current vertical acceleration and predicted vertical displacement, it calculates the target weight coefficient and constructs a variable-weight cost function, enabling the gimbal control quantity to adapt to different vibration intensities. It calculates the three-axis feedforward compensation quantity based on the predicted vertical displacement and combines it with the preset three-axis tracking error to calculate the three-axis control quantity, achieving physical-level active suppression of disturbances. Finally, it integrates the gimbal control quantity and the three-axis control quantity for gimbal stabilization, enabling the control strategy to adapt to different turbulent conditions and actively cancel disturbance forces, significantly improving the ability to suppress mid-to-low frequency jitter, reducing control bandwidth requirements and energy consumption, and improving the balance between stability, responsiveness, and energy efficiency.

[0103] Example 3 Furthermore, this disclosure provides an electronic device including a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the gimbal stabilization method described in Embodiment 1.

[0104] It should be noted that electronic devices such as action cameras and live streaming cameras achieve gimbal stabilization through the gimbal stabilization method provided in Example 1.

[0105] The device provided in this embodiment can perform the steps of the gimbal stabilization method provided in Embodiment 1. To avoid repetition, the steps will not be repeated.

[0106] Example 4 This disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the gimbal stabilization method described in Embodiment 1.

[0107] In this embodiment, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0108] The computer-readable storage medium provided in this embodiment can implement the gimbal stabilization method provided in Embodiment 1. To avoid repetition, it will not be described again here.

[0109] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limitations; therefore, other examples of exemplary embodiments may have different values.

[0110] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0111] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A gimbal stabilization method, characterized in that, The method includes: Obtain the current distance sequence from the gimbal to the ground, and obtain the current vertical acceleration and predicted vertical displacement based on the current distance sequence; The target weight coefficient is calculated based on the current vertical acceleration and the predicted vertical displacement. The variable weight cost function is then optimally solved based on the target weight coefficient to obtain the gimbal control quantity. The three-axis feedforward compensation amount is calculated based on the predicted vertical displacement, and the three-axis control amount is calculated based on the three-axis feedforward compensation amount and the preset three-axis tracking error. Gimbal stabilization is performed based on the gimbal control parameters and the three-axis control parameters.

2. The gimbal stabilization method according to claim 1, characterized in that, The target weight coefficient includes displacement weight, velocity weight, and control quantity weight. The calculation of the target weight coefficient based on the current vertical acceleration and the predicted vertical displacement includes: The aiming velocity is calculated based on the predicted vertical displacement, a first scaling factor is obtained based on the first preset adjustment coefficient and the square of the aiming velocity, and the displacement weight is calculated based on the first preset reference weight and the first scaling factor. The second proportional factor is obtained based on the second preset adjustment coefficient and the absolute value of the current vertical acceleration, and the velocity weight is calculated based on the second preset reference weight and the second proportional factor. The aiming acceleration is calculated based on the predicted vertical displacement. The third proportional factor is calculated based on the third preset adjustment coefficient and the square of the aiming acceleration. The control quantity weight is calculated based on the third preset reference weight and the reciprocal of the third proportional factor.

3. The gimbal stabilization method according to claim 2, characterized in that, The target weighting coefficient also includes an acceleration weight, and the method further includes: Calculate the square of the displacement difference between the gimbal body displacement and the gimbal reference displacement, and calculate the displacement tracking error term based on the displacement weight and the square of the displacement difference; The main body speed term is calculated based on the speed weight and the square of the gimbal main body speed; The main body acceleration term is calculated based on the acceleration weights and the square of the gimbal main body acceleration. Calculate the control energy consumption item based on the control quantity weight and the gimbal control quantity; The variable weight cost function is obtained by accumulating and integrating the displacement tracking error term, the main velocity term, the main acceleration term, and the energy consumption term.

4. The gimbal stabilization method according to claim 1, characterized in that, The three-axis feedforward compensation includes vertical feedforward compensation and horizontal feedforward compensation; the calculation of the three-axis feedforward compensation based on the predicted vertical displacement includes: Differentiating the predicted vertical displacement yields the predicted vertical velocity and the predicted vertical acceleration; Obtain the current horizontal parameters, calculate the predicted horizontal displacement based on the current horizontal parameters, and differentiate the predicted horizontal displacement to obtain the predicted horizontal velocity and the predicted horizontal acceleration. Based on the preset equivalent mass, damping ratio, vertical natural frequency, predicted vertical velocity, predicted vertical acceleration and their corresponding prediction time, the vertical inertial force overcoming term, the vertical damping force overcoming term and the vertical spring restoring force overcoming term are calculated; The vertical feedforward compensation amount is obtained by summing the vertical inertial force overcoming term, the vertical damping force overcoming term, and the vertical spring restoring force overcoming term; Based on the preset equivalent mass, the damping ratio, the vertical natural frequency, the predicted horizontal velocity, the predicted horizontal acceleration and their corresponding prediction time, the horizontal inertial force overcoming term, the horizontal damping force overcoming term and the horizontal spring restoring force overcoming term are calculated. The horizontal feedforward compensation is obtained by summing the horizontal inertial force overcoming term, the horizontal damping force overcoming term, and the horizontal spring restoring force overcoming term.

5. The gimbal stabilization method according to claim 4, characterized in that, The current horizontal parameters include the current horizontal acceleration, the current horizontal velocity, and the current horizontal position. The step of calculating the predicted horizontal displacement based on the current horizontal parameters includes: Based on Newtonian kinematics, the predicted horizontal displacement is calculated according to the current horizontal acceleration, the current horizontal velocity, the current horizontal position, and the predicted time.

6. The gimbal stabilization method according to claim 1, characterized in that, The calculation of the three-axis control quantity based on the three-axis feedforward compensation amount and the preset three-axis tracking error includes: The rate of change of the preset triaxial tracking error is obtained by differentiating the preset triaxial tracking error. The preset three-axis tracking error, the rate of change, and the three-axis feedforward compensation are weighted and summed based on the preset proportional gain, preset differential gain, and preset aiming feedforward gain, respectively, to obtain the three-axis control quantity.

7. The gimbal stabilization method according to claim 1, characterized in that, The step of obtaining the current vertical acceleration and predicted vertical displacement based on the current distance sequence includes: The current distance sequence is denoised to obtain a denoised distance sequence; Acquire accelerometer data, and calculate the current vertical acceleration based on the accelerometer data and the denoised distance sequence; The denoised distance sequence is differentially processed to obtain the vertical velocity estimate; The predicted vertical displacement is calculated based on the denoised distance sequence, the current vertical acceleration, and the estimated vertical velocity.

8. A gimbal stabilization device, characterized in that, The device includes: The acquisition module is used to acquire the current distance sequence from the gimbal to the ground, and to acquire the current vertical acceleration and predicted vertical displacement based on the current distance sequence; The first calculation module is used to calculate the target weight coefficient based on the current vertical acceleration and the predicted vertical displacement, and to perform optimal solution on the variable weight cost function based on the target weight coefficient to obtain the gimbal control quantity; The second calculation module is used to calculate the three-axis feedforward compensation amount based on the predicted vertical displacement, and to calculate the three-axis control amount based on the three-axis feedforward compensation amount and the preset three-axis tracking error. The stabilization module is used to stabilize the gimbal based on the gimbal control quantity and the three-axis control quantity.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, implements the gimbal stabilization method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the gimbal stabilization method as described in any one of claims 1 to 7.

Images