Inertial measurement device calibration method and device, electronic equipment, robot and storage medium

Abstract

The invention provides an inertial measurement device calibration method and device, electronic equipment, a robot and a storage medium. The calibration method comprises the following steps: calculating a residual value based on a state quantity predicted value of the inertial measurement device at a previous moment, an actual measurement value at a current moment and a nonlinear observation equation; wherein the actual measurement value is a multi-dimensional vector; the nonlinear observation equation is linearized to obtain an approximate observation matrix with n rows and one column, and n is equal to the dimensionality of the state quantity; calculating a Kalman gain based on an error covariance predicted value at the previous moment, a measurement noise covariance and the approximate observation matrix; calculating a posterior error covariance predicted value at the current moment according to the Kalman gain and the approximate observation matrix; and calculating a correction term according to the Kalman gain and the residual value, and calculating a posterior state quantity predicted value of the inertial measurement device at the current moment by using the correction term. According to the invention, the operation complexity can be reduced.

Description

Background Art

[0002] Inertial Measurement Unit (IMU) is widely used in robots, drones, mobile devices and navigation systems. Most IMUs have systematic errors such as scale factor and bias, and these errors will drift over time and changes in the usage environment, affecting the measurement accuracy. Therefore, it is necessary to calibrate the IMU to correct its scale factor and bias and improve its accuracy.

[0003] Conventional inertial measurement device calibration methods mostly use nonlinear least squares optimization algorithms based on gradient descent. However, this method requires caching a large amount of raw data and relies on multiple iterative calculations, resulting in high computational complexity and is not suitable for resource-constrained embedded systems.

[0004] Some related technologies have begun to use the Extended Kalman Filter (EKF) calibration method, which can directly use real-time observation data to update parameters without caching a large amount of raw data, and estimate and correct the scale factor and deviation of the sensor in real time.

[0005] In this method, it is necessary to calculate the Kalman gain K = P pred ·H T ·(H·P pred ·H T +R) -1 ; Among them, P pred is the error covariance prediction value of the previous moment, H T is the transposed matrix of the measurement matrix, H is the measurement matrix, and R is the measurement noise covariance. When the measurement value is a multidimensional vector, the calculated measurement matrix H is a multidimensional matrix, and then (H·P pred ·H T +R) is a complex multidimensional matrix, (H·P pred ·H T +R) -1 It is an inverse operation for complex multidimensional matrices.

[0006] The above matrix inversion operation is highly computationally complex and requires a large number of floating-point operations, so the computational overhead is very high. On the one hand, it increases the performance requirements for the processor, making it difficult to reduce the cost of equipment; on the other hand, it makes it difficult to run the algorithm in scenarios with insufficient computing power, limiting the scope of application.

[0007] To solve the above problem, in the Chinese patent application with publication number CN110307842A, two orthogonal vectors are constructed using the output information of the accelerometer and the geomagnetic sensor, and then the two orthogonal vectors are used to simplify the observation matrix. The simplified observation matrix is ​​as follows:

[0008]

[0009] In this patent, the simplified observation matrix F is further used k+1 (q k+1 / k ), the matrix operation in the Kalman gain is simplified to a scalar operation without matrix inversion. However, this simplification depends on the covariance matrix of the specific initial attitude error. k / k =a 2 I 4×4 This can avoid matrix inversion, which has great limitations on usage scenarios.

[0010] In the US patent application with publication number US20190266499A1, sparse computing technology is used to reduce the amount of calculation by constructing a new, smaller correlation matrix. For the covariance matrix inversion part of the above Kalman filter gain calculation, a new observation matrix HH = [H 1 , 0...; 0, H 2 ,0...,0,H n ] (n is limited to the relevant state), and the P matrix is ​​reduced to PP accordingly. In this way, the complex matrix inversion calculation is transformed into an operation on a smaller matrix, thereby reducing the computational complexity; however, this method still cannot avoid the matrix inversion operation.

[0011] Similarly, in related technologies such as the paper "On Computational Complexity Reduction Methods for Kalman Filter Extensions" (author Matti Raitoharju et al., IEEE Aerospace and Electronic Systems Magazine, August 2015), the inverse operation problem of converting a high-dimensional matrix into a matrix of smaller dimensionality is also addressed.

[0012] The applicant intends to provide a new inertial measurement device calibration method that is different from the above method to reduce the computational complexity of the extended Kalman filter calibration method. Summary of the invention

[0013] The present disclosure provides an inertial measurement device calibration method and apparatus, electronic device, robot, and storage medium, aiming to solve the problem that the performance requirements of the extended Kalman filter calibration method in the related art for the processor make it difficult to reduce the cost of the equipment, and the algorithm is difficult to run in scenarios with insufficient computing power, thereby limiting the scope of application.

[0014] Additional aspects and advantages of the disclosure will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the disclosure.

[0015] According to a first aspect of the present disclosure, there is provided a method for calibrating an inertial measurement device, comprising:

[0016] Calculate the residual value based on the predicted value of the state quantity of the inertial measurement device at the previous moment, the actual measurement value at the current moment, and the nonlinear observation equation; wherein the actual measurement value is a multidimensional vector;

[0017] Linearizing the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column, where n is equal to the dimension of the state quantity;

[0018] Calculating the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix;

[0019] According to the Kalman gain and the approximate observation matrix, the posterior error covariance prediction value at the current moment is calculated; and, according to the Kalman gain and the residual value, a correction term is calculated, and the posterior state quantity prediction value of the inertial measurement device at the current moment is calculated using the correction term.

[0020] In an exemplary embodiment of the present disclosure, linearizing the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column includes:

[0021] The nonlinear observation equation is partially derived at each of the state quantity prediction values, and each obtained partial derivative value is used as an element of the approximate observation matrix.

[0022] In an exemplary embodiment of the present disclosure, the nonlinear observation equation is an ellipsoid model:

[0023] h(state)=A·(x i -x 0 ) 2 + B·(y i -y 0 ) 2 +C·(z i -z 0 ) 2

[0024] Among them, [A, B, C, x 0 ,y 0 , z 0 ] is the predicted value of the state quantity at the previous moment; A, B, and C are the squares of the scale factors of the x, y, and z axes respectively; x 0 ,y 0 、z 0are the x, y, and z axis deviations respectively; (x i ,y i , z i ) is the actual measured value, and h(state) is a scalar.

[0025] In an exemplary embodiment of the present disclosure, linearizing the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column includes:

[0026] The partial derivative of the nonlinear observation equation is obtained at each predicted value of the state quantity:

[0027]

[0028] The obtained partial derivatives are used as the approximate observation matrix H i Elements:

[0029]

[0030] In an exemplary embodiment of the present disclosure, a result of multiplying the approximate observation matrix by the error covariance prediction value at a previous moment by the transposed matrix of the approximate observation matrix is ​​a scalar.

[0031] In an exemplary embodiment of the present disclosure, the calculation of the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix includes:

[0032] K=P pred ·H i T ·(H i ·P pred ·H i T +r) -1

[0033] Where K is the Kalman gain, P pred is the error covariance prediction value of the previous moment, H i T is the transposed matrix of the approximate measurement matrix, H i is the approximate observation matrix and r is the measurement noise covariance.

[0034] In an exemplary embodiment of the present disclosure, the inertial measurement device includes one or more of an accelerometer, a gyroscope, and a magnetometer.

[0035] According to a second aspect of the present disclosure, there is provided an inertial measurement device calibration apparatus, comprising:

[0036] A residual calculation module, used to calculate the residual value based on the predicted value of the state quantity of the inertial measurement device at the previous moment, the actual measurement value at the current moment and the nonlinear observation equation; wherein the actual measurement value is a multidimensional vector;

[0037] An observation matrix calculation module, used for linearizing the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column;

[0038] A Kalman gain calculation module, used to calculate the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix;

[0039] A data updating module is used to calculate the posterior error covariance prediction value at the current moment according to the Kalman gain and the approximate observation matrix; and to calculate a correction term according to the Kalman gain and the residual value, and use the correction term to calculate the posterior state quantity prediction value of the inertial measurement device at the current moment.

[0040] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0041] Processor; and

[0042] A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement any one of the above-described methods for calibrating an inertial measurement device.

[0043] In an exemplary embodiment of the present disclosure, the processor is one or more of a microprocessor, a microcontroller unit, a single-chip microcomputer and an embedded processor.

[0044] According to a fourth aspect of the present disclosure, there is provided a robot, comprising:

[0045] Processor; and

[0046] A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement any one of the above-described methods for calibrating an inertial measurement device.

[0047] In an exemplary embodiment of the present disclosure, the robot includes any one of a legged robot, a quadrupedal robot, a bipedal robot, a wheeled robot, a wheel-legged robot, a quadrupedal robot, a humanoid robot, a cleaning robot, a transport robot, a mobile robot and a robotic arm.

[0048] In an exemplary embodiment of the present disclosure, the processor is one or more of a microprocessor, a microcontroller unit, a single-chip microcomputer and an embedded processor.

[0049] According to a fifth aspect of the present disclosure, a computer-readable storage medium is provided, on which computer program code instructions are stored. When the computer program code instructions are called by a processor of a robot, the robot executes an inertial measurement device calibration method as described in any one of the above.

[0050] It can be seen from the above technical solution that the present disclosure has at least one of the following advantages and positive effects:

[0051] On the one hand, in the related art, when the measured value is a multidimensional vector, the process of calculating the Kalman gain involves a complex multidimensional matrix inversion operation; and in the inertial measurement device calibration method disclosed in the present invention, by linearizing the nonlinear observation equation, an approximate observation matrix of n rows and one column is obtained, so that the complex multidimensional matrix inversion operation can be simplified to a scalar operation. This simplification can effectively reduce the computational complexity, reduce the floating-point operation requirements, reduce the computational overhead, improve the computational speed, and reduce the requirements for the performance of the hardware device, thereby reducing the device cost; for example, it can be adapted to MCU (Micro Controller Unit, microprocessor), single-chip microcomputer, etc., thereby greatly reducing the size, cost and power consumption of the hardware. On the other hand, in low-computing power devices, matrix inversion operations are prone to cause numerical precision loss, which leads to error accumulation and affects the accuracy of calibration. The present disclosure significantly reduces the error accumulation in a low-computing power environment by avoiding matrix inversion operations, thereby ensuring that the calibration accuracy will not be significantly affected. Therefore, based on the inertial measurement device calibration method disclosed in the present invention, the extended Kalman filter method can be applied to low-computing power devices, expanding the scope of application of the algorithm. In addition, the inertial measurement device calibration method disclosed in the present invention has low dependence on initial parameters, and even when the initial conditions are not ideal, the calibration accuracy can be continuously optimized by updating the Kalman gain in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the embodiments or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0053] Figure 1 A system architecture block diagram of an application environment of an inertial measurement device calibration method and apparatus in an embodiment of the present disclosure;

[0054] Figure 2 A schematic diagram of a flow chart of an inertial measurement device calibration method in an embodiment of the present disclosure;

[0055] Figure 3This is an experimental data diagram of an inertial measurement device calibration method in an embodiment of the present disclosure;

[0056] Figure 4 This is an experimental data diagram of an inertial measurement device calibration method in an embodiment of the present disclosure;

[0057] Figure 5 is a structural block diagram of a robot according to an embodiment of the present disclosure;

[0058] Figure 6 is a schematic diagram of a robot according to an embodiment of the present disclosure;

[0059] Figure 7 is a schematic diagram of a robot according to another embodiment of the present disclosure;

[0060] Figure 8 is a schematic diagram of a robot according to yet another embodiment of the present disclosure;

[0061] Fig. 9 It is a structural block diagram of an inertial measurement device calibration device in one embodiment of the present disclosure;

[0062] Fig.10 The structure block diagram of the electronic device according to some embodiments of the present disclosure is schematically shown;

[0063] Fig.11 is a schematic diagram of a computer-readable storage medium according to some embodiments of the present disclosure.

[0064] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0065] In the description of the present invention, terms such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "clockwise", "counterclockwise", "axial", "radial", etc. indicating directions or positional relationships are based on the drawings and are only for the convenience of description and simplified description. They do not mean that the device or element must have a specific direction, be constructed and operated in a specific direction, and cannot be regarded as a limitation of the present invention.

[0066] The terms "first" and "second" are used for description only and do not indicate relative importance or imply the number of technical features. Therefore, the "first" and "second" features may explicitly or implicitly include at least one of the features. The meaning of "plurality" is at least two, unless otherwise explicitly defined.

[0067] Unless otherwise specified, terms such as "installed", "connected", "joined", "fixed", etc. shall be understood in a broad sense, which can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection, an electrical connection, or capable of communicating with each other; it can be directly connected, or indirectly connected through an intermediate medium, and can be the communication inside two components or the interaction relationship between two components.

[0068] In the present invention, the first feature being "on" or "under" the second feature can be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, the first feature being "above", "over" and "on top of" the second feature can be that the first feature is directly above or obliquely above the second feature, or merely indicates that the first feature has a higher horizontal height than the second feature. The first feature being "under", "beneath" and "underneath" the second feature can be that the first feature is directly below or obliquely below the second feature, or merely indicates that the first feature has a lower horizontal height than the second feature.

[0069] First, the following definitions are given for some terms that may be involved in the present disclosure:

[0070] Inertial measurement device: includes a sensor (such as an accelerometer, gyroscope, or magnetometer, etc.) or a sensing device combining multiple sensors (such as multiple of an accelerometer, gyroscope, and magnetometer, etc.), and is used to detect the motion state and direction of the device.

[0071] Scale factor: the proportionality factor of the measured value of each axis of the inertial measurement device, and is a calibration parameter used to convert the sensor output into an actual physical quantity.

[0072] Bias: the non-zero value of the measurement output of the inertial measurement device in the stationary state, representing the constant part of the measurement error.

[0073] Extended Kalman filter: a method based on the state estimation of a nonlinear system, which updates the relationship between the measured value and the state prediction value through recursive calculation to improve the estimation accuracy.

[0074] Nonlinear observation equation: a mathematical equation used to describe the nonlinear relationship between the sensor output and its state.

[0075] State quantity: a set of variables describing the system state, used to completely express the current characteristics or dynamic changes of the system, such as speed, position, attitude, error, etc.

[0076] Actual measured value: the original measurement data recorded by the sensor at the current moment.

[0077] Observation matrix: a matrix that maps the state variables to the measurement value space, and is used to calculate the relationship between the state and the measurement value.

[0078] Error covariance: describes the distribution characteristics of state estimation error and indicates the uncertainty of the estimated value.

[0079] Measurement noise covariance: A statistical parameter that describes the characteristics of sensor measurement noise and is used to adjust the weight distribution of the Kalman filter.

[0080] Kalman gain: The coefficient used to balance the weight of predicted values ​​and measured values ​​in the Kalman filter algorithm, which determines the degree of state update.

[0081] Ellipsoid model: A mathematical model used in the calibration of inertial measurement devices to describe the distribution characteristics of sensor data in three-dimensional space, usually used to correct scale factors and deviations.

[0082] Figure 1 The schematic diagram of the system architecture of an exemplary application environment in which the inertial measurement device calibration method and the inertial measurement device calibration apparatus according to the embodiment of the present disclosure can be applied is shown. Figure 1 As shown, the system architecture 100 may include an operation processing unit 101 and an inertial measurement device 102. The number of the operation processing unit 101 and the inertial measurement device 102 may be one or more, and this is not particularly limited in the present exemplary embodiment. For example, in one embodiment, the number of the operation processing unit 101 and the inertial measurement device 102 is the same and both are multiple, and each inertial measurement device 102 is connected to one operation processing unit 101. In another embodiment, there is one operation processing unit 101, and the number of the inertial measurement devices 102 is multiple, and the multiple inertial measurement devices 102 are all connected to the operation processing unit 101.

[0083] After detecting the motion state and direction of the device, the inertial measurement device 102 sends the actual measurement value to the processing unit 101. The processing unit 101 calibrates the inertial measurement device according to the actual measurement value and the inertial measurement device calibration method or inertial measurement device calibration device provided by the present disclosure to correct errors such as scale factor and deviation, thereby improving accuracy.

[0084] refer to Figure 2 As shown, the inertial measurement device calibration method in this exemplary embodiment includes the following steps:

[0085] S210, calculating a residual value based on a predicted value of a state quantity of the inertial measurement device at a previous moment, an actual measurement value at a current moment, and a nonlinear observation equation; wherein the actual measurement value is a multidimensional vector.

[0086] S220. Linearize the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column, where n is equal to the dimension of the state quantity.

[0087] S230, calculating the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance, and the approximate observation matrix.

[0088] S240, calculating the posterior error covariance prediction value at the current moment according to the Kalman gain and the approximate observation matrix; and calculating the correction term according to the Kalman gain and the residual value, and using the correction term to calculate the posterior state quantity prediction value of the inertial measurement device at the current moment.

[0089] On the one hand, in the related art, when the measurement value is a multidimensional vector, the process of calculating the Kalman gain involves a complex multidimensional matrix inversion operation; and in the inertial measurement device calibration method disclosed in the present invention, by linearizing the nonlinear observation equation, an approximate observation matrix of n rows and one column is obtained, so that the complex multidimensional matrix inversion operation can be simplified to a scalar operation. This simplification can effectively reduce the computational complexity, reduce the floating-point operation requirements, reduce the computational overhead, improve the computing speed, and reduce the requirements for the performance of the hardware device, thereby reducing the equipment cost. On the other hand, in low-computing power devices, matrix inversion operations are prone to cause numerical precision loss, which in turn leads to error accumulation and affects the accuracy of the calibration. The present disclosure significantly reduces the error accumulation in a low-computing power environment by avoiding the matrix inversion operation, thereby ensuring that the calibration accuracy will not be significantly affected. Therefore, based on the inertial measurement device calibration method disclosed in the present invention, the extended Kalman filter method can be applied to low-computing power devices, expanding the scope of application of the algorithm. In addition, the inertial measurement device calibration method disclosed in the present invention has low dependence on initial parameters, and even when the initial conditions are not ideal, the calibration accuracy can be continuously optimized by updating the Kalman gain in real time.

[0090] The inertial measurement device calibration method in this example embodiment will be described in detail below.

[0091] In step S210, a residual value is calculated based on the predicted value of the state quantity of the inertial measurement device at the previous moment, the actual measurement value at the current moment, and the nonlinear observation equation; wherein the actual measurement value is a multidimensional vector.

[0092] In this exemplary embodiment, "previous moment" refers to the most recent time point at which the state quantity prediction and update have been completed. At this time point, we have the updated state quantity prediction value and its uncertainty (i.e., the error covariance matrix described below). During the first iteration calibration, the initialized state quantity prediction value and error covariance prediction value can be used as the state quantity prediction value at the previous moment and the error covariance prediction value at the previous moment. "Current moment" refers to the time point at which the state quantity prediction is to be performed.

[0093] In the calibration method based on the extended Kalman filter, each iteration combines the state quantity prediction value at the previous moment, the error covariance prediction value at the previous moment, the actual measurement value at the current moment, and the nonlinear observation equation to perform a prediction-update cycle. In this way, the accuracy of the state quantity prediction value estimation can be continuously improved, making the next calibration result closer to the actual state quantity.

[0094] The above-mentioned state quantity is usually expressed in vector form. In this exemplary embodiment, the state quantity may include the scale factor parameter and the deviation parameter of the inertial measurement device. Taking the three-axis inertial measurement device as an example, the state quantity prediction value can be expressed using [A, B, C, x 0 ,y 0 , z 0 ]; A, B, C are the squares of the scale factors of the x, y, and z axes respectively; x 0 ,y 0 、z 0 They are the x-axis, y-axis, and z-axis deviations, respectively. In other embodiments of the present disclosure, the state quantity may also include other parameters, such as gyroscope misalignment, accelerometer zero bias, etc., so as to adapt to different types of inertial measurement devices. In principle, the dimension of the state quantity is the sum of the number of parameters related to the calibration. When other parameters need to be expanded, the corresponding dimensions can be added.

[0095] The above actual measurement values ​​are the observed outputs of the inertial measurement device at the current moment. Taking the three-axis inertial measurement device as an example, the actual measurement values ​​of the three dimensions x i ,y i , z i Represents the original measurement value in the x, y, and z axis directions at the current moment. The actual measurement value may be the acceleration component (m / s 2 ), the geomagnetic intensity component (μT) output by the magnetometer, or the angular velocity component (rad / s) output by the gyroscope, etc., depending on the type of inertial measurement device to be calibrated. In addition, in some embodiments, x i ,y i , z i They can also all be combined measurement values; for example, when the three-axis inertial measurement device also uses other types of sensors, such as optical flow sensors, odometers or GPS (Global Positioning System), the outputs of various forms can also be combined to obtain measurement values ​​to ensure consistency with the input form requirements of the nonlinear observation equation.

[0096] The above nonlinear observation equation is used to map the state quantity to the measurement space, that is, to give a predicted value of the state quantity. In this exemplary embodiment, taking the ellipsoid model as an example, when the deviation and scale factor of the inertial measurement device satisfy the ellipsoid distribution characteristics, its nonlinear observation equation can be defined as:

[0097] h(state)=A·(x i -x 0 ) 2 + B·(y i -y 0 ) 2 +C·(z i -z 0 ) 2

[0098] Among them, [A, B, C, x 0 ,y 0 , z 0 ] is the predicted value of the state quantity at the previous moment; A, B, and C are the squares of the scale factors of the x, y, and z axes respectively; x 0 ,y 0 、z 0 are the x, y, and z axis deviations respectively; (x i ,y i , z i ) are actual measured values.

[0099] The above residual is the deviation R between the actual measured value and the theoretical measured value given by the nonlinear observation equation. k , which is used to indicate the degree of deviation between the predicted value of the state quantity at the current moment and the actual observed value. The predicted value of the state quantity at the previous moment is substituted into the nonlinear observation equation to calculate the theoretical measurement value h(state); h(state) is a scalar; the value of h(state) will also be different depending on the inertial measurement device being calibrated; for example, if the accelerometer is calibrated, h(ssate) can be the first-order norm of the local gravity acceleration, such as the modulus of the gravity acceleration vector; if the magnetometer is calibrated, h(ssase) can be the first-order norm of the local geomagnetic field vector, such as the modulus of the geomagnetic field vector; for other sensors (such as gyroscopes), h(state) can be defined as other theoretical output values ​​related to sensor characteristics.

[0100] It should be noted that in other exemplary embodiments of the present disclosure, the nonlinear observation equation may also be a polynomial or other types of nonlinear models, as long as the state quantity can be mapped to the measurement domain. In addition, if there is a non-uniform noise distribution in the inertial measurement device, additional nonlinear terms or additional correction terms may be considered in the nonlinear observation equation, which also fall within the scope of protection of the present disclosure.

[0101] In step S220, the nonlinear observation equation is linearized to obtain an approximate observation matrix with n rows and one column, where n is equal to the dimension of the state quantity.

[0102] The principle of the linearization process in this exemplary embodiment is to use Taylor series expansion to transform the nonlinear observation equation into a local linear approximate model, and to approximate the originally complex function form in the form of a polynomial. Specifically, the nonlinear observation equation can be partially derived at each of the above-mentioned state quantity prediction values, and each obtained partial derivative value is used as an element of the approximate observation matrix.

[0103] Taking the nonlinear observation equation as the above ellipsoid model as an example, the partial derivative operation of the nonlinear observation equation at the predicted value of each state quantity can be as follows:

[0104]

[0105] Since the state quantity is a 6-dimensional vector, 6 partial derivatives can be obtained by taking the partial derivative of each state quantity prediction value; then, the 6 partial derivatives can be combined into an observation matrix H with n rows and one column. i :

[0106]

[0107] In actual operation, to ensure accuracy, [A, B, C, x 0 ,y 0 , z 0 ] and (x i ,y i , z i ) can be stored in floating point or fixed point high precision format. In addition, you can also calculate the partial derivative before (x i ,y i , z i ) and (x 0 ,y 0 , z 0 ) to calculate the difference and store the intermediate result, such as Δx = x i -x 0 , Δy=y i -y 0 , Δz=z i -z 0 , in order to reduce repeated operations and reduce numerical errors.

[0108] It should be noted that the linearization described in this disclosure is not limited to the use of standard analytical partial derivatives. In the absence of analytical expressions, numerical differentiation methods such as differential approximation can also be used to find (θ j is the predicted value of the state quantity), that is, a small perturbation Δθ is made to each predicted value of the state quantity j , the partial derivative approximation is calculated according to the following formula:

[0109]

[0110] where e j is a unit vector that is 1 in the j-th coordinate direction and 0 in the remaining directions. This method also belongs to the scope of protection of the present disclosure.

[0111] In addition, in some embodiments, if the dimension n of the state quantity changes (for example, more parameters are introduced, such as additional error terms of the gyroscope, etc.), in this step, the partial derivatives of the newly added parameters of the nonlinear observation equation can be obtained according to the same principle. Correspondingly, the dimension of the above approximate observation matrix H i will change accordingly; however, the basic calculation framework does not need to be changed.

[0112] In step S230, the Kalman gain is calculated based on the predicted value of the error covariance at the previous moment, the measurement noise covariance, and the approximate observation matrix.

[0113] In the extended Kalman filter algorithm, the Kalman gain is a weighting factor used to balance the predicted value of the state quantity and the actual measurement information. By applying the Kalman gain to the residual, the predicted value of the state quantity can be corrected, so as to obtain a posteriori estimate value closer to the true state. In this exemplary embodiment, the Kalman gain is calculated based on the predicted value of the error covariance at the previous moment, the measurement noise covariance, and the above approximate observation matrix:

[0114] K = P pred ·H i T ·(H i ·P pred ·H i T + r) -1

[0115] where K is the Kalman gain, P pred is the predicted value of the error covariance at the previous moment, H i T is the transpose matrix of the approximate observation matrix, H i is the approximate observation matrix, and r is the measurement noise covariance.

[0116] In the related art, when the measurement value is a multi-dimensional vector, the calculated observation matrix H is a multi-dimensional matrix, and then (H i ·P pred ·H i T + r) is a complex multi-dimensional matrix, (H i ·P pred ·H i T + r) -1It is the inverse operation of a complex multi-dimensional matrix. In the inertial measurement device calibration method of the present disclosure, by linearizing the non-linear observation equation, an approximate observation matrix H with n rows and one column is obtained i , so that H i ·P pred ·H i T The calculation result of is a scalar, and then (H i ·P pred ·H i T +r) -1 is simplified to a scalar operation. Next, a more detailed description of the calculation of the Kalman gain is given:

[0117] The above predicted value P of the error covariance pred is used to reflect the estimation of the uncertainty of the state quantity during the prediction process; the size of the error covariance is n×n, where n is the dimension of the state quantity. The above measurement noise covariance is a measure of the uncertainty of the actual measurement value; since in this exemplary embodiment H i ·P pred ·H i T The calculation result of is a scalar, and the measurement noise covariance r in this exemplary embodiment is also set as a scalar.

[0118] Since H i is a matrix of size n×1, and P pred is a matrix of size n×n, then H i ·P pred is a matrix of size 1×n; H i T is a matrix of size 1×n, then H i ·P pred ·H i T is a matrix of size 1×1, that is, a scalar. For convenience of description, this scalar is denoted as S. Furthermore, the Kalman gain K is:

[0119]

[0120] Denote the intermediate quantity P pred ·H i T as w. Since P pred is a matrix of size n×n, and H i T is a matrix of size 1×n, so w is a matrix of size n×1. Multiply the matrix w by to obtain K. Since is a scalar, perform a scalar multiplication on each component of w. Finally, K is a matrix of size n×1, and the elements are denoted as K i,satisfy:

[0121]

[0122] In the above process, all operations are simple multiplication and addition, which completely avoids the original multi-dimensional matrix inversion problem, thereby significantly reducing the computational complexity and hardware requirements, and helping to stably and efficiently execute the extended Kalman filter calibration process on a resource-limited platform.

[0123] In step S240, the a posteriori error covariance prediction value at the current moment is calculated based on the Kalman gain and the approximate observation matrix; and a correction term is calculated based on the Kalman gain and the residual value, and the correction term is used to calculate the a posteriori state prediction value of the inertial measurement device at the current moment.

[0124] In the extended Kalman filter algorithm, by updating the error covariance, the uncertainty change of the state quantity prediction value at the current moment after correction using the new measurement information can be reflected. In this exemplary embodiment, the posterior error covariance prediction value P can be calculated by the following formula post :

[0125] P post =(I-KH i ) pred

[0126] Where I is the identity matrix of size n×n.

[0127] The above correction term is used to correct the state quantity prediction value, that is, to perform offset correction on the state quantity prediction value after weighting the residual, so that the result is closer to the actual state. In this exemplary embodiment, the posterior state quantity prediction value can be obtained by adding the state quantity prediction value and the correction term:

[0128] x k|k =x k|k-1 +K·R k

[0129] Among them, x k|k is the predicted value of the posterior state quantity at the current moment, x k|k-1 is the predicted value of the state quantity at the previous moment, R k is the above residual value, K·R k For correction items.

[0130] Furthermore, in a low computing power environment, in order to reduce memory access and instruction count, in this exemplary embodiment, K and H can also be i The storage structure of KH is designed as a continuous memory block, and loop unrolling technology is used to speed up the outer product calculation. iis a matrix constructed by outer product, and its element values ​​may be small or large, depending on the magnitude of the Kalman gain and the approximate observation matrix. In order to avoid the loss of precision caused by too small values ​​or the overflow caused by too large values, in this exemplary embodiment, K and H may also be i The components are scaled or normalized to keep the values ​​within a reasonable range.

[0131] Next, taking a magnetometer as an example, the complete calibration process using the inertial measurement device calibration method in this exemplary embodiment is described.

[0132] In this exemplary embodiment, the state quantity of the magnetometer includes the square parameters of the scale factors of the three axes x, y, and z and the deviation parameters of each axis, a total of 6 dimensions:

[0133] state = [A, B, C, x 0 ,y 0 , z 0 ] T

[0134] Among them: A, B, C are the squares of the scale factors of the three axes x, y, and z; 0 ,y 0 , z 0 are the x, y, and z axis deviations.

[0135] In this exemplary embodiment, the initial state quantity prediction value is set as:

[0136] x 0|0 =[1.2, 0.9, 1.1, 2.0, -1.0, 0.5] T

[0137] The initial error covariance prediction value is set as a diagonal matrix:

[0138] P 0|0 =diag(0.1, 0.1, 0.1, 0.05, 0.05, 0.05)

[0139] The measurement noise covariance scalar is set to:

[0140] r=0.02

[0141] Assume that the magnetometer is stationary at a certain location, where the nominal strength of the geomagnetic field is:

[0142] M ref =70μT

[0143] To match the quadratic structure of the nonlinear observation equation, the reference value is squared:

[0144]

[0145] Actual measured value at the current moment:

[0146] (x i ,y i , z i )=(48.5,-49.0,10.2)μT

[0147] Nonlinear observation equation (ellipsoid model):

[0148] h(state)=A(x i -x 0 ) 2 +B(y i -y 0 ) 2 +C(z i -z 0 ) 2

[0149] First, x 0|0 Substitute: A = 1.2, B = 0.9, C = 1.1, x 0 =2.0,y 0 =-1.0, z 0 = 0.5 Calculate the difference for subsequent direct calls:

[0150] Δx=x i -x 0 =46.5

[0151] Δy=y i -y o =-48.0

[0152] Δz=z i -z 0 =9.7

[0153] Next, calculate the theoretical measurement value:

[0154] h(x 0|0 )=1.2·46.5 2 +0.9 (-48.0) 2 +1.1 9.7 2 ≈4771.8

[0155] Then, the residual value can be calculated:

[0156] r k =z k -h(x 0|0 )=128.2

[0157] Then, linearization is performed to obtain an approximate observation matrix H with 6 rows and 1 column. The partial derivative calculation process is as follows:

[0158]

[0159] Combining the partial derivatives gives an approximate observation matrix H of size 6×1:

[0160] H=[2162.25, 2304, 94.09, -111.6, 86.4, -21.34] T

[0161] Next, calculate the Kalman gain K. The Kalman gain formula is:

[0162] K=P pred H(H T P pred H+r) -1

[0163] Where S = H T P pred H≈1,000,278.15, then (S+r) -1 ≈9.9972×10 -7

[0164] Then Kalman gain K=(P pred H)(S+r) -1 ≈[2.162×10 -4 , 2.304×10 -4 , 9.41×10 -6 , -5.58×10 -6 , 4.32×10 -6 , -1.067×10 -6 ] T

[0165] Finally, update the state quantity prediction value to obtain the posterior state quantity prediction value at the current moment:

[0166] x k|k =x 0|0 +Kr k =[1.2, 0.9, 1.1, 2.0, -1.0, 0.5] T +K 128.2

[0167] ≈[1.2277, 0.9295, 1.101205, 1.999285, -0.999446, 0.499863] T

[0168] Similarly, update the posterior error covariance:

[0169] P post =(I-KH)P pred

[0170] where KH is the n×n matrix formed by the outer product. By calculating this matrix and combining it with P pred Multiplying them together, we can obtain the corrected posterior error covariance P post The specific value can be obtained by multiplying the components of K and H element by element and adding them to P pred Multiply them together to get .

[0171] The above process shows the complete process and data example of using the extended Kalman filter calibration algorithm to calibrate and update the magnetometer parameters under the given initial state, error covariance, measurement value and reference value. In the next round of calibration, repeat the above process and perform a new round of calibration. In practical applications, by continuously repeating this cycle, the calibration accuracy of the magnetometer can be iteratively improved.

[0172] refer to Figure 3 , Figure 4 As shown in the table below, in the experimental structure statistics, for the same magnetometer calibrated using the extended Kalman filter method of sensing and the calibration method in the exemplary embodiment of the present disclosure, the calibration errors of the two methods are roughly equivalent, but the calibration method in the exemplary embodiment of the present disclosure consumes less computing power.

[0173]

[0174] In another exemplary embodiment of the present disclosure, a robot is also provided. Figure 5 As shown, the robot 500 at least includes a processor 501 and a memory 502; of course, the robot 500 also includes necessary components such as an inertial measurement device 503, a motor driver 504, and a joint module 505. The memory 502 stores computer-readable instructions, and when the computer-readable instructions are executed by the processor 501, the inertial measurement device calibration method described above is implemented.

[0175] Furthermore, the present disclosure significantly reduces the complexity of calculating the Kalman gain by linearizing the nonlinear observation equation and simplifying the observation matrix into a form of n rows and one column, thereby avoiding multi-dimensional matrix inversion operations. Since there is no need to perform complex matrix inversion operations, the calibration method of the present disclosure significantly reduces the demand for processor computing power. This means that even on processors with limited operating resources and weak computing power, the calibration method of the present disclosure can still be efficiently executed. Therefore, in this exemplary embodiment, the processor 501 can be one or more of a microprocessor, a microcontroller unit, a single-chip microcomputer, and an embedded processor. Furthermore, the present disclosure can reduce the cost of the entire machine; and low-power processors consume less power, which helps to extend battery life or reduce energy consumption.

[0176] In one embodiment of the present disclosure, a robot configured with an inertial measurement device calibration function includes any one of a legged robot, a quadrupedal robot, a bipedal robot, a wheeled robot, a wheel-legged robot, a quadrupedal robot, a humanoid robot, a cleaning robot, a transport robot, a mobile robot and a robotic arm.

[0177] like Figure 6 As shown, a structural schematic diagram of a bipedal robot is exemplarily shown.

[0178] like Figure 7 In one embodiment, the biped robot and the quadruped robot may include a leg support structure, a leg-trunk connection structure, a knee joint assembly structure, a foot transmission structure, and other suitable robot structural components.

[0179] like Figure 8 , which exemplarily shows a schematic diagram of the structure of a humanoid robot. In one embodiment, the humanoid robot may include an arm structure, a connection structure between an arm and a trunk, a wrist joint assembly structure, a waist assembly structure, an elbow joint assembly structure, a thigh support structure, a hip connection structure, a transmission structure, and other suitable robot structural components.

[0180] Next, in an embodiment of the present disclosure, a calibration device for an inertial measurement device is also provided. Fig. 9 As shown, the inertial measurement device calibration apparatus 900 includes the following modules:

[0181] A residual calculation module 910 is used to calculate the residual value based on the state quantity prediction value of the inertial measurement device at the previous moment, the actual measurement value at the current moment and the nonlinear observation equation; wherein the actual measurement value is a multidimensional vector; an observation matrix calculation module 920 is used to linearize the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column; a Kalman gain calculation module 930 is used to calculate the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix; a data update module 940 is used to calculate the posterior error covariance prediction value at the current moment according to the Kalman gain and the approximate observation matrix; and, calculate the correction term according to the Kalman gain and the residual value, and use the correction term to calculate the posterior state quantity prediction value of the inertial measurement device at the current moment.

[0182] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above-mentioned cross-platform robot operation method is also provided.

[0183] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware embodiments, complete software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0184] Refer to the following Fig.10 hereinafter describes the electronic device 1000 according to such an embodiment of the present disclosure. Fig.10 The electronic device 1000 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0185] like Fig.10 As shown, the electronic device 1000 is in the form of a general computing device. The components of the electronic device 1000 may include but are not limited to: the at least one processing unit 1010, the at least one storage unit 1020, a bus 1030 connecting different system components (including the storage unit 1020 and the processing unit 1010), and a display unit 1040.

[0186] Among them, the storage unit stores program code, and the program code can be executed by the processing unit 1010, so that the processing unit 1010 executes the steps described in the above "Exemplary Method" section of this specification according to various exemplary embodiments of the present disclosure; in this exemplary embodiment, the processing unit 1010 can be one or more of a microprocessor, a microcontroller unit, a single-chip microcomputer and an embedded processor.

[0187] The storage unit 1020 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit 1021 (RAM) and / or a cache storage unit 1022, and may further include a read-only storage unit 1023 (ROM).

[0188] The storage unit 1020 may also include a program / utility having a set (at least one) of program modules 1024, such program modules 1024 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0189] Bus 1030 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0190] The electronic device 1000 may also communicate with one or more external devices 1070 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1000, and / or communicate with any device that enables the electronic device 1000 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1050. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1060. As shown, the network adapter 1060 communicates with other modules of the electronic device 1000 via a bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0191] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0192] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible embodiments, various aspects of the present disclosure can also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of the present specification.

[0193] refer to Fig.11 1, a program product 1100 for implementing the above-mentioned inertial measurement device calibration method according to an embodiment of the present disclosure is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto, and in this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus, or a device.

[0194] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0195] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0196] The program code contained on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, electromagnetic waves, etc., or any suitable combination of the foregoing.

[0197] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0198] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0199] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0200] Finally, the above preferred embodiments are only used to illustrate the technical solutions of the present application and are not restrictive. Although the present application has been described in detail, those skilled in the art should understand that changes in form and details can be made without departing from the scope defined by the claims of the present application. The dimensions of the drawings have nothing to do with the specific objects, and the dimensions of the objects can be changed arbitrarily.

Claims

1. A method for calibrating an inertial measurement device, characterized in that: include: Calculate the residual value based on the predicted value of the state quantity of the inertial measurement device at the previous moment, the actual measurement value at the current moment, and the nonlinear observation equation; wherein the actual measurement value is a multidimensional vector; Linearizing the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column, where n is equal to the dimension of the state quantity; Calculating the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix; According to the Kalman gain and the approximate observation matrix, the posterior error covariance prediction value at the current moment is calculated; and, according to the Kalman gain and the residual value, a correction term is calculated, and the posterior state quantity prediction value of the inertial measurement device at the current moment is calculated using the correction term.

2. The inertial measurement device calibration method according to claim 1, characterized in that: The linearizing of the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column includes: The nonlinear observation equation is partially derived at each of the state quantity prediction values, and each obtained partial derivative value is used as an element of the approximate observation matrix.

3. The inertial measurement device calibration method according to claim 1, characterized in that: The nonlinear observation equation is an ellipsoid model: h(state)=A·(x i -x0) 2 +B·(y i -y0) 2 +C·(z i -z0) 2 Among them, [A, B, C, x0, y0, z0] is the predicted value of the state quantity at the previous moment; A, B, C are the squares of the scale factors of the x, y, and z axes respectively; x0, y0, z0 are the deviations of the x, y, and z axes respectively; (x i ,y i , z i ) is the actual measured value; h(state) is a scalar.

4. The inertial measurement device calibration method according to claim 3, characterized in that: The linearizing of the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column includes: The partial derivative of the nonlinear observation equation is obtained at each predicted value of the state quantity: The obtained partial derivatives are used as the approximate observation matrix H i Elements:

5. The inertial measurement device calibration method according to any one of claims 1 to 4, characterized in that: The result of multiplying the approximate observation matrix by the error covariance prediction value at the previous moment and multiplying the transposed matrix of the approximate observation matrix is ​​a scalar.

6. The inertial measurement device calibration method according to claim 5, characterized in that: The calculating of the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix includes: K=P pred ·H i T ·(H i ·P pred ·H i T +r) -1 Where K is the Kalman gain, P pred is the error covariance prediction value of the previous moment, H i T is the transposed matrix of the approximate measurement matrix, H i is the approximate observation matrix and r is the measurement noise covariance.

7. The inertial measurement device calibration method according to claim 5, characterized in that: The inertial measurement device includes one or more of an accelerometer, a gyroscope, and a magnetometer.

8. An inertial measurement device calibration device, characterized in that: include: A residual calculation module, used to calculate the residual value based on the predicted value of the state quantity of the inertial measurement device at the previous moment, the actual measurement value at the current moment and the nonlinear observation equation; wherein the actual measurement value is a multidimensional vector; An observation matrix calculation module, used for linearizing the nonlinear observation equation to obtain an approximate observation matrix with n rows and one column; A Kalman gain calculation module, used to calculate the Kalman gain based on the error covariance prediction value at the previous moment, the measurement noise covariance and the approximate observation matrix; A data updating module is used to calculate the posterior error covariance prediction value at the current moment according to the Kalman gain and the approximate observation matrix; and to calculate a correction term according to the Kalman gain and the residual value, and use the correction term to calculate the posterior state quantity prediction value of the inertial measurement device at the current moment.

9. An electronic device, characterized in that: include: processor; as well as A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement the inertial measurement device calibration method according to any one of claims 1 to 7.

10. The electronic device according to claim 9, characterized in that: The processor is one or more of a microcontroller unit, a single-chip microcomputer and an embedded processor.

11. A robot, characterized in that: include: processor; as well as A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement the inertial measurement device calibration method according to any one of claims 1 to 7.

12. The robot according to claim 11, characterized in that: The robot includes any one of a legged robot, a quadruped robot, a bipedal robot, a wheeled robot, a wheel-legged robot, a quadrupedal robot, a humanoid robot, a cleaning robot, a transport robot, a mobile robot and a robotic arm.

13. The robot according to claim 11 or 12, characterized in that: The processor is one or more of a microprocessor, a microcontroller unit, a single chip microcomputer and an embedded processor.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program code instructions, and when the computer program code instructions are called by a processor of a robot, the robot executes the inertial measurement device calibration method according to any one of claims 1 to 7.

Images