An unmanned aerial vehicle airborne lightweight perception error adaptive calibration method

Abstract

The application discloses a man-machine airborne lightweight sensing error adaptive calibration method, comprising the following steps: establishing an observation equation of a nacelle installation angle error; establishing an observation model based on the observation equation of the nacelle installation angle error; estimating the nacelle installation angle error based on the observation model; decoupling the nacelle installation angle error in the yaw and pitch channels to reduce the coupling error of the estimated nacelle installation angle error, and realizing the calibration of the installation angle error. The man-machine airborne lightweight sensing error adaptive calibration method disclosed by the application solves the problem of constant deviation in angle measurement of a low-cost unmanned aerial vehicle airborne nacelle under the influence of lightweight sensing error.

Description

Technical Field

[0001] This invention relates to an adaptive calibration method for lightweight sensing errors on unmanned aerial vehicles (UAVs), belonging to the field of flight control technology. Background Technology

[0002] Low-cost UAVs equipped with airborne electro-optical pods are the most common observation method. Therefore, target localization often employs passive positioning schemes using multiple UAVs measuring angles of arrival (AOA) or active positioning schemes using single UAVs measuring angles and distances. Thus, for low-cost UAV target localization, angle measurement accuracy significantly impacts target positioning accuracy. However, the installation angle error is unavoidable when using airborne electro-optical pods, resulting in significant deviations from the constant angle measurement value.

[0003] Traditional calibration methods require specialized calibration instruments and rely on repeated manual adjustments, which are demanding in terms of environmental conditions and complex in operation. In addition, traditional calibration methods are mostly designed for expensive electro-optical pods mounted on large aircraft, requiring the collection of flight data over a long period of time for calibration, and are not suitable for the rapid batch calibration of installation angle errors of low-cost UAV pods.

[0004] Existing research on UAV target localization often uses large airborne optoelectronic pods or targets with relatively small distances, thus ignoring the influence of installation angle errors and mainly employing methods such as optimizing UAV trajectories and multi-point measurement optimal estimation to eliminate the influence of random errors.

[0005] While existing methods address the impact of pod installation angle errors on target positioning, these methods primarily employ long-range flight data acquisition for cooperative targets or utilize high-precision instruments for calibration. These methods are mainly suitable for large, sophisticated airborne electro-optical pods and are insufficient for the mass calibration needs of low-cost UAVs. For example, Wang Chenxin et al. briefly described a given benchmark method for calibrating the deterministic error of angle measurement using high-precision instruments to address the AOA positioning problem of high-speed aircraft swarms; Liu Yunhua et al. analyzed the factors affecting the angle measurement accuracy of electro-optical pods and controlled the installation angle error through high-precision goniometers and high-precision mounting surface processing technology; Tan Renlong et al. proposed an aerial calibration method for tracking cooperative targets to address the installation angle error calibration problem of electro-optical pods. The method reduced the average positioning error from 1700m to 450m after calibration at a relative distance of 50km. However, this method requires the UAV to collect multiple sets of flight data at an altitude of 3000m, which places a huge burden on the batch calibration of the installation angle error of low-cost UAVs. Zhou Yong et al. addressed the camera extrinsic parameter calibration problem by using the square of the deviation between the estimated target position and the actual position as the cost function and calibrating the constant errors of the pod installation angle and attitude angle using an optimization method. However, this method also requires the collection of multiple sets of flight data and timestamp alignment, making it difficult to apply to the batch calibration of low-cost UAVs.

[0006] It is evident that existing methods do not take into account the installation angle error of low-cost airborne optoelectronic pods, or the calibration methods are complex to implement, making it difficult to achieve simple and rapid calibration.

[0007] Therefore, it is necessary to conduct research on the adaptive calibration of airborne lightweight sensing errors for low-cost UAVs in order to solve the above problems. Summary of the Invention

[0008] To overcome the above problems, in-depth research was conducted, and an adaptive calibration method for lightweight human-machine interface perception error was proposed, including the following steps:

[0009] S1. Establish the observation equation for the pod installation angle error;

[0010] S2. Establish an observation model based on the observation equations for the pod installation angle error;

[0011] S3. Estimate the pod installation angle error based on the observation model.

[0012] In a preferred embodiment, the method further includes S4, decoupling the pod installation angle error of the yaw and pitch channels to reduce the coupling error of the estimated pod installation angle error and achieve calibration of the installation angle error.

[0013] In a preferred embodiment, in S1, the observation equation for the pod installation angle error is expressed as:

[0014]

[0015] Among them, P t P represents the location of the target in the geographic coordinate system. u Let r be the position of the UAV in the geographic coordinate system, and r be the relative distance between the UAV and the target. Let be the transformation matrix from the base coordinate system to the optical axis coordinate system. This is the transformation matrix from the body coordinate system to the base coordinate system. This is the transformation matrix from the geographic coordinate system to the body coordinate system, with the superscript T indicating transpose.

[0016] In a preferred embodiment, step S2, establishing the observation model, includes the following sub-steps:

[0017] The projection of the line vector connecting the UAV and the target in the body coordinate system is used as the first projection.

[0018] The projection of the vector connecting the UAV and the target in the base coordinate system is used as the second projection.

[0019] An observation model for the pod installation angle error is constructed based on the first projection, the second projection, and the coordinate system transformation matrix.

[0020] In a preferred embodiment, the first projection V b Represented as

[0021]

[0022] The second projection V s Represented as

[0023]

[0024] In a preferred embodiment, the first projection V b Second projection V s Normalize them separately, and they are expressed as:

[0025]

[0026] Where, x b ,y b ,z b Let x be the first projected coordinate after normalization. s ,y s ,z s These are the normalized second projected coordinates.

[0027] In a preferred embodiment, the established observation model is represented as follows:

[0028]

[0029] Where, x = [δ x ,δ y ,δ z ] T ,

[0030]

[0031] x represents the state variable to be estimated in the observation model. This is the observation matrix with noise values.

[0032] In a preferred embodiment, observations are performed multiple times using an observation model, as follows:

[0033]

[0034] Where k is the total number of observations, Let be the observation matrix for the i-th observation, y be the number of observations, and V be the observation noise;

[0035] The least squares method is used to solve the above equation to obtain the estimated pod installation angle error:

[0036]

[0037] in, This represents the estimated pod installation angle error.

[0038] In a preferred embodiment, in S4, the pitch installation angle error δ is... y During calibration, the pitch installation angle error equation is obtained. Solving this equation yields the roll installation angle error and pitch installation angle error, thus achieving calibration. The pitch installation angle error equation can be expressed as:

[0039]

[0040] x b =r x cosψ+r y sinψ,y b =-r x sinψ+r y cosψ

[0041] In a preferred embodiment, in S4, the yaw installation angle error δ z During calibration, the yaw installation angle error equation is obtained. Solving this equation yields the roll installation angle error and pitch installation angle error, thus achieving calibration. The yaw installation angle error equation can be expressed as:

[0042]

[0043] x b =r x cosθ-r z sinθ,z b =r x sinθ+r z cosθ

[0044] The beneficial effects of this invention include:

[0045] (1) Solve the problem of constant deviation in angle measurement under the influence of lightweight perception error (i.e., pod installation angle error) of low-cost UAV airborne pods;

[0046] (2) The calibration of the pod installation angle error is highly accurate and stable. Attached Figure Description

[0047] Figure 1 The diagram illustrates a flowchart of an adaptive calibration method for airborne lightweight sensing errors of an unmanned aerial vehicle (UAV) according to a preferred embodiment of the present invention. Detailed Implementation

[0048] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Through these descriptions, the features and advantages of the present invention will become clearer and more apparent.

[0049] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments. Although various aspects of embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated otherwise.

[0050] An adaptive calibration method for lightweight sensing error onboard UAVs, provided by the present invention, includes the following steps:

[0051] S1. Establish the observation equation for the pod installation angle error;

[0052] S2. Establish an observation model based on the observation equations for the pod installation angle error;

[0053] S3. Estimate the pod installation angle error based on the observation model.

[0054] Preferably, the step further includes S4, decoupling the pod installation angle error of the yaw and pitch channels to reduce the coupling error of the estimated pod installation angle error and achieve calibration of the installation angle error.

[0055] In S1, the observation equation for the pod installation angle error is expressed as:

[0056]

[0057] Among them, P t P represents the location of the target in the geographic coordinate system. u Let r be the position of the UAV in the geographic coordinate system, and r be the relative distance between the UAV and the target. Let be the transformation matrix from the base coordinate system to the optical axis coordinate system. This is the transformation matrix from the body coordinate system to the base coordinate system. This is the transformation matrix from the geographic coordinate system to the body coordinate system, with the superscript T indicating transpose.

[0058] In this invention, the geographic coordinate system OX n Y n Z n Let the origin be the center of mass of the UAV, OX. n The axis is parallel to the horizontal plane of the earth, pointing eastward as positive; OY n The axis is parallel to the horizontal plane of the earth, pointing north (positive); OZ n The axis is perpendicular to the ground and points upwards; according to the right-hand rule, pointing upwards is positive.

[0059] The body coordinate system OX b Y b Z b Let the origin be the center of mass of the UAV, OX. bParallel to the fuselage axis, pointing towards the nose is positive, OY b The axis is located in the OX region b In the transverse plane of the axis and perpendicular to OX b The axis is positive to the left; OZ b The axis is perpendicular to the other two axes, determined by the right-hand rule, with upward being positive;

[0060] The base coordinate system OX s Y s Z s The origin is located at the center of mass of the base of the optoelectronic pod, OX. s Axis and OY s The shaft is on the horizontal plane of the base, OX s The axis points directly in front of the photoelectric sphere; forward is positive. (OY) s The axis is perpendicular to OX s The axis, pointing to the left, is positive; OZ s The axis is perpendicular to OX s Y s The plane must satisfy the right-hand rule, and the direction pointing upwards towards the photoelectric sphere is positive;

[0061] The optical axis coordinate system OX g Y g Z g The origin is set at the camera's center of mass on the optoelectronic pod, OX. g The axis points in the direction of the optical axis of the optoelectronic pod, with forward being positive; OY g Axis and OZ g The axis is in the camera plane, OY g Axis and OX g The axis is vertical, with left being positive; OZ g Axis and OY g Axis, OZ g The axis satisfies the right-hand rule, with upward pointing being positive.

[0062] There is an installation angle error between the base coordinate system and the body coordinate system. In the optical axis coordinate system, the pod's optical axis always points towards the target.

[0063] Preferably, the transformation matrix is ​​set as follows:

[0064]

[0065] Where γ represents the UAV roll angle, θ represents the UAV pitch angle, ψ represents the UAV yaw angle, and δ x Indicates the roll-on installation angle, δ y Indicates the pitch installation angle, δ z Indicates the yaw installation angle. Indicates the corner of the outer frame of the pod. This indicates the angle of the inner frame of the pod.

[0066] In S2, the study found that the translation error in the pod installation error has a small impact on the target positioning accuracy of the UAV, while the angle error has a large impact. Therefore, the influence of the translation error in the pod installation is ignored, and only the estimation of the pod installation angle error is considered.

[0067] In S2, establishing the observation model includes the following sub-steps:

[0068] The projection of the line vector connecting the UAV and the target in the body coordinate system is used as the first projection.

[0069] The projection of the vector connecting the UAV and the target in the base coordinate system is used as the second projection.

[0070] An observation model for the pod installation angle error is constructed based on the first projection, the second projection, and the coordinate system transformation matrix.

[0071] Furthermore, the first projection, namely the projection V of the vector connecting the UAV and the target in the body coordinate system, is... b Represented as

[0072]

[0073] Preferably, for the first projection V b Normalization is performed to eliminate the influence of ranging errors, expressed as:

[0074]

[0075] Where, x b ,y b ,z b These are the first projected coordinates after normalization.

[0076] The second projection, namely the projection V of the vector connecting the UAV and the target in the base coordinate system. s Represented as:

[0077]

[0078] Preferably, for the second projection V s Normalization is performed to eliminate the influence of ranging errors, expressed as:

[0079]

[0080] Where, x s ,y s ,z s These are the normalized second projected coordinates.

[0081] The first projection, the second projection, and the coordinate system transformation matrix satisfy the following:

[0082]

[0083] Preferably, since the installation angle error of the pod is relatively small, a smaller angle reduction method can be adopted. Simplify the parameters:

[0084] cos(δ)=1, sin(δ)=δ

[0085] After simplifying and removing higher-order minor quantities, the relationship between the first projection, the second projection, and the coordinate system transformation matrix can be obtained as follows:

[0086]

[0087] Based on the above equation, the constructed observation model is expressed as:

[0088]

[0089] Where, x = [δ x ,δ y ,δ z ] T ,

[0090]

[0091] x represents the state variable to be estimated in the observation model. This is the observation matrix with noise values.

[0092] In S3, the least squares method is used to estimate the pod installation angle error based on the observation model.

[0093] Specifically, multiple observations are performed using an observation model, as shown below:

[0094]

[0095] Where k is the total number of observations, Let be the observation matrix for the i-th observation, y be the number of observations, and V be the observation noise.

[0096] The least squares method is used to solve the above equation to obtain the estimated pod installation angle error:

[0097]

[0098] in, This represents the estimated pod installation angle error.

[0099] Although the above can provide an estimated pod installation angle error, the observation matrix... V in b Components and V s Each component includes the UAV's own position, body attitude angles γ, θ, ψ, and pod frame angle. The coupling error.

[0100] In S4, errors in the observation matrix are eliminated by decoupling the pod installation angle errors of the yaw and pitch channels.

[0101] For pitch installation angle error δ y During calibration, the UAV and the target are placed on the same horizontal plane at a sufficient relative distance, with the UAV's nose deviating significantly from the line of sight. At this point, the aircraft's roll angle γ, pitch angle θ, and pod pitch frame angle are measured. All are relatively small angles, only the fuselage yaw angle ψ and the pod yaw frame angle. It is a relatively large angle.

[0102] At this time, V b The component can be represented as:

[0103]

[0104] Wherein, δP u The positional error of the drone itself is due to the relative distance |P t -P u |Large enough to eliminate δP u The impact, r x ,r y These are the normalized components of the vector connecting the UAV and the target in the geographic coordinate system;

[0105] Right now

[0106] V s The component can be represented as:

[0107]

[0108] Right now

[0109] By combining the estimated pod installation angle error, the pitch installation angle error equation is obtained. Solving this equation yields the roll installation angle error and pitch installation angle error, thus enabling calibration. The pitch installation angle error equation can be expressed as:

[0110]

[0111] x b =r x cosψ+r y sinψ,y b =-r x sinψ+r y cosψ

[0112] It can be seen that the error equation is not observable for the yaw installation angle error at this time, that is, the roll and pitch installation angle errors δ can be calibrated under this condition. x ,δ y And the observation matrix The error in the system originates only from the yaw angle error ψ, which can greatly reduce the coupling error.

[0113] Similarly, for the yaw installation angle error δ z During calibration, position the UAV horizontally with its nose pointing downwards, the target stationary at a relatively high altitude, at a sufficient distance, and with the nose pointing in the line of sight. At this point, measure the UAV's roll angle γ, yaw angle ψ, and pod yaw frame angle. All are relatively small angles, only the fuselage pitch angle θ and the pod pitch frame angle. It is a relatively large angle.

[0114] At this time, V b The component can be represented as:

[0115]

[0116] Right now

[0117] V s The component can be represented as:

[0118]

[0119] Right now

[0120] By combining the estimated pod installation angle error, the yaw installation angle error equation is obtained. Solving this equation yields the roll installation angle error and pitch installation angle error, thus achieving calibration. The yaw installation angle error equation can be expressed as:

[0121]

[0122] x b =r x cosθ-r z sinθ,z b =r x sinθ+r z cosθ

[0123] It can be seen that the error equation is not observable for pitch installation angle error at this time, meaning that the roll and yaw installation angle errors δ can be calibrated under this condition. x ,δ z And the observation matrix The error in the system originates only from the pitch angle error θ of the aircraft, which can greatly reduce the coupling error.

[0124] In this invention, although the pitch installation angle error equation and the yaw installation angle error equation can both be used to solve for the roll installation angle error δ, x However, in actual target positioning, the pitch installation angle error δ y This will result in a large positioning error in the Z-axis coordinate (i.e., in the vertical direction); yaw installation angle error δ z This will result in larger positioning errors in the X and Y axis coordinates (i.e., in the horizontal direction); roll installation angle error δ x This mainly leads to pitch installation angle error δ y and yaw installation angle error δ z The coupling, while the pitch installation angle error δ y Yaw installation angle error δ z After all are calibrated and compensated, the roll installation angle error δ x The impact on target positioning is negligible. That is, in practical use, only the yaw installation angle error δ needs to be considered. z Pitch installation angle error δ y without needing to account for the roll installation angle error δ x Perform calibration.

[0125] Example

[0126] Example 1

[0127] The simulation experiment was conducted to calibrate the installation angle error, including the following steps:

[0128] S1. Establish the observation equation for the pod installation angle error;

[0129] S2. Establish an observation model based on the observation equations for the pod installation angle error;

[0130] S3. Estimate the pod installation angle error based on the observation model.

[0131] In S1, the observation equation for the pod installation angle error is expressed as:

[0132]

[0133] In S2, establishing the observation model includes the following sub-steps:

[0134] The projection of the line vector connecting the UAV and the target in the body coordinate system is used as the first projection.

[0135] The projection of the vector connecting the UAV and the target in the base coordinate system is used as the second projection.

[0136] An observation model for the pod installation angle error is constructed based on the first projection, the second projection, and the coordinate system transformation matrix.

[0137] The projection V of the line vector connecting the UAV and the target in the body coordinate system b Represented as

[0138]

[0139] For projection V b Normalization is performed to eliminate the influence of ranging errors, expressed as:

[0140]

[0141] Projection V of the line vector connecting the UAV and the target in the base coordinate system s Represented as:

[0142]

[0143] For projection V s Normalization is performed to eliminate the influence of ranging errors, expressed as:

[0144]

[0145] The first projection, the second projection, and the coordinate system transformation matrix satisfy the following:

[0146]

[0147] Small keratinization can be used Simplify the parameters:

[0148] cos(δ)=1, sin(δ)=δ

[0149] After simplifying and removing higher-order minor quantities, the relationship between the first projection, the second projection, and the coordinate system transformation matrix can be obtained as follows:

[0150]

[0151] The constructed observation model is represented as follows:

[0152]

[0153] Where, x = [δ x ,δ y ,δ z ] T ,

[0154]

[0155] In S3, based on the observation model, the least squares method is used to estimate the pod installation angle error, and the estimated pod installation angle error is obtained:

[0156]

[0157] During the simulation, the following conditions are set: In the geographic coordinate system (n-system), the target location is the origin: P t =[0,0,0] T The drone's location is 800m west, 600m north, and 100m altitude: P t = [-800, 600, 100] T True value of installation angle error: [δ x ,δ y ,δ z ] T =[1°,2°,3°] T .

[0158] Throughout the simulation process, 5000 Monte Carlo simulations were performed, with each simulation using k=500 observation data points. The random error for UAV target positioning was set as shown in Table 1.

[0159] Table 1

[0160]

[0161]

[0162] The simulation calibration results of the installation angle error are shown in Table 2.

[0163] Table 2

[0164]

[0165] Comparing Table 1 and Table 2, it can be seen that the calibrated installation angle error results approximately satisfy a normal distribution, and the estimated mean error is ≤0.1°, which can be regarded as an unbiased estimate.

[0166] Example 2

[0167] The same experiment as in Example 1 was conducted, except that it also included S4, decoupling the pod installation angle error of the yaw and pitch channels, in order to reduce the coupling error of the estimated pod installation angle error and achieve calibration of the installation angle error.

[0168] In S4, the pitch installation angle error equation is obtained, and the roll installation angle error and pitch installation angle error are solved to obtain the pitch installation angle error. The pitch installation angle error equation can be expressed as follows:

[0169]

[0170] x b =r x cosψ+r y sinψ,y b =-r x sinψ+r ycosψ

[0171] During the simulation, the following conditions were set:

[0172] 1) Calibrate the pitch installation angle error δ y At that time, in the geographic coordinate system, the target location is the origin: P t =[0,0,0] T The drone's location is 1000m to the west: P t =[-1000,0,0] T True values ​​of the aircraft's attitude angles: [γ,θ,ψ] T =[0°,0°,30°] T True value of installation angle error: [δ x ,δ y ,δ z ] T =[1°,2°,3°] T .

[0173] The simulation calibration results of the installation angle error are shown in Table 3.

[0174] Table 3

[0175]

[0176] As shown in Table 3, the calibrated installation angle errors approximately follow a normal distribution, and the dispersion is smaller than that in Table 2, with a standard deviation ≤ 0.03°. Specifically, the roll installation angle error calibration accuracy is significantly improved compared to the method in Example 1, with a standard deviation reduction of 93.6%, achieving a calibration accuracy of ≤ 0.2°; the pitch installation angle error calibration accuracy is also significantly improved compared to the method in Example 1, with a standard deviation reduction of 59.4%, achieving a calibration accuracy of ≤ 0.1°.

[0177] Example 3

[0178] The same experiment as in Example 1 was conducted, except that it also included S4, decoupling the pod installation angle error of the yaw and pitch channels, in order to reduce the coupling error of the estimated pod installation angle error and achieve calibration of the installation angle error.

[0179] The yaw installation angle error equation is obtained, and the roll installation angle error and pitch installation angle error are solved to obtain the yaw installation angle error. The yaw installation angle error equation can be expressed as:

[0180]

[0181] x b =r x cosθ-r z sinθ,z b =r x sinθ+rz cosθ

[0182] 2) Calibrate the yaw installation angle error δ z At that time, in the geographic coordinate system, the target location is: P t =[0,0,100] T The drone's location is 1000m to the west: P t =[-800,0,0] T True values ​​of the aircraft's attitude angles: [γ,θ,ψ] T = [0°, -30°, 0°] T True value of installation angle error: [δ x ,δ y ,δ z ] T =[1°,2°,3°] T .

[0183] The simulation calibration results of the installation angle error are shown in Table 4.

[0184] Table 4

[0185]

[0186] As can be seen from Table 4, the calibrated installation angle error results approximately follow a normal distribution, and the dispersion is smaller than that of the estimated results in Table 2, with a standard deviation ≤ 0.05°. Among them, the calibration accuracy of the roll installation angle error is calibrated with a standard deviation reduced by 82.4% compared to the method in Example 1; the calibration accuracy of the yaw installation angle error is calibrated with a standard deviation reduced by 86.9% compared to the method in Example 1, indicating improved calibration error stability.

[0187] The present invention has been described above with reference to preferred embodiments; however, these embodiments are merely exemplary and illustrative. Various substitutions and modifications can be made to the present invention based on these embodiments, all of which fall within the scope of protection of the present invention.

Claims

1. A method for adaptive calibration of lightweight human-machine interface sensing errors, characterized in that, Includes the following steps: S1. Establish the observation equation for the pod installation angle error; S2. Establish an observation model based on the observation equations for the pod installation angle error; S3. Estimate the pod installation angle error based on the observation model.

2. The adaptive calibration method for lightweight human-machine interface sensing error according to claim 1, characterized in that, It also includes S4, decoupling the yaw and pitch channels for pod installation angle errors, in order to reduce the coupling error of the estimated pod installation angle error and achieve calibration of the installation angle error.

3. The adaptive calibration method for human-machine interface lightweight sensing error according to claim 1, characterized in that, In S1, the observation equation for the pod installation angle error is expressed as: Among them, P t P represents the location of the target in the geographic coordinate system. u Let r be the position of the UAV in the geographic coordinate system, and r be the relative distance between the UAV and the target. Let be the transformation matrix from the base coordinate system to the optical axis coordinate system. This is the transformation matrix from the body coordinate system to the base coordinate system. This is the transformation matrix from the geographic coordinate system to the body coordinate system, with the superscript T indicating transpose.

4. The adaptive calibration method for human-machine interface lightweight sensing error according to claim 1, characterized in that, In S2, establishing the observation model includes the following sub-steps: The projection of the line vector connecting the UAV and the target in the body coordinate system is used as the first projection. The projection of the vector connecting the UAV and the target in the base coordinate system is used as the second projection. An observation model for the pod installation angle error is constructed based on the first projection, the second projection, and the coordinate system transformation matrix.

5. The adaptive calibration method for human-machine interface lightweight sensing error according to claim 4, characterized in that, The first projection V b Represented as The second projection V s Represented as 6. The adaptive calibration method for human-machine interface lightweight sensing error according to claim 5, characterized in that, For the first projection V b Second projection V s Normalize them separately, and they are expressed as: Where, x b ,y b ,z b Let x be the first projected coordinate after normalization. s ,y s ,z s These are the normalized second projected coordinates.

7. The adaptive calibration method for lightweight human-machine interface sensing error according to claim 1, characterized in that, The established observation model is represented as follows: where x = [δ x , δ y , δ z T ,​ x represents the state variable to be estimated in the observation model. This is the observation matrix with noise values.

8. The adaptive calibration method for human-machine interface lightweight sensing error according to claim 7, characterized in that, Multiple observations were performed using the observation model, as shown below: Where k is the total number of observations, Let be the observation matrix for the i-th observation, y be the number of observations, and V be the observation noise; The least squares method is used to solve the above equation to obtain the estimated pod installation angle error: in, This represents the estimated pod installation angle error.

9. The adaptive calibration method for lightweight human-machine interface sensing error according to claim 2, characterized in that, In S4, the pitch installation angle error δ y During calibration, the pitch installation angle error equation is obtained. Solving this equation yields the roll installation angle error and pitch installation angle error, thus achieving calibration. The pitch installation angle error equation can be expressed as: x b =r x cosψ+r y sinψ,y b =-r x sinψ+r y cosψ。 10. The adaptive calibration method for lightweight human-machine interface sensing error according to claim 2, characterized in that, In S4, the yaw installation angle error δ z During calibration, the yaw installation angle error equation is obtained. Solving this equation yields the roll installation angle error and pitch installation angle error, thus achieving calibration. The yaw installation angle error equation can be expressed as:

Images