A VTOL aircraft visual-inertial navigation method and system for degraded motion

Abstract

This invention discloses a visual inertial navigation method and system for VTOL aircraft with degraded motion. The method includes: Step S1, simultaneously acquiring and preprocessing data from a downward-looking camera, IMU, and barometer; Step S2, online determination of whether the motion of the VTOL aircraft causes visual inertial odometry observability degradation, identifying the degradation mode, and quantifying and evaluating scale observability and translation observability; Step S3, based on a multi-state constrained Kalman filter framework, adaptively executing corresponding information fusion strategies to perform state estimation according to the degradation mode and observability, and outputting navigation state information. The method and system of this invention solve the problem of visual inertial navigation failure under motion modes with observable degradation, and can improve the navigation robustness of VTOL aircraft in critical phases such as takeoff, landing, and hovering under denied environments.

Description

Technical Field

[0001] This invention relates to a navigation method and system, specifically to a visual inertial navigation method and system for VTOL aircraft oriented towards degenerate motion; it belongs to the field of autonomous navigation technology for aircraft. Background Technology

[0002] Vertical takeoff and landing (VTOL) aircraft typically rely on visual inertial odometry (VIO), consisting of cameras and inertial measurement units, for autonomous navigation in low-altitude denied environments. However, they often face two types of observability degradation during critical phases such as takeoff, landing, and hovering: First, vertical translation degradation, where the aircraft moves vertically along the direction of gravity for an extended period. In this mode, the scale factor of the monocular visual inertial odometry is strongly coupled with the direction of gravity, making the scale factor and altitude estimation unobservable, thus leading to state divergence. Second, hovering rotation degradation, where the aircraft performs pure rotational motion at a fixed position. In this mode, the translational state is completely unobservable, feature point triangulation fails, and position estimation diverges rapidly.

[0003] In particular, in lightweight VTOL platforms employing a downward-looking monocular camera layout, the camera's field of view is limited to the ground. During vertical motion, the parallax at feature points is extremely small, and during pure rotation, translational excitation is completely absent, making the aforementioned degradation problems particularly prominent. Existing solutions mostly assume the camera is forward-looking or has a wide field of view, failing to optimize for the specific combination of a downward-looking layout and the aircraft's motion characteristics. Some studies have attempted to introduce external signals such as Global Navigation Satellite Systems (GNSS) and Ultra-Wideband (UWB) for assistance; however, these methods are unsuitable for signal-denied environments such as indoors, underground, and urban canyons, and pose security risks from spoofing attacks.

[0004] Therefore, there is an urgent need for a visual-inertial navigation method that does not rely on external signals and can autonomously determine and suppress the effects of the aforementioned degenerative motion, so as to ensure the navigation robustness and safety of VTOL aircraft throughout the entire mission phase. Summary of the Invention

[0005] The purpose of this invention is to provide a visual inertial navigation method and system for VTOL aircraft oriented towards degenerate motion. Addressing the problem of visual inertial navigation failure during takeoff, landing, and hovering phases of VTOL aircraft, this invention fundamentally suppresses the divergence problem of visual inertial odometry under degenerate motion by online determination of two types of degenerate motion: vertical translation and hovering rotation, and triggering corresponding multi-sensor adaptive fusion and state protection strategies. This achieves highly robust autonomous navigation without external signal dependence.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] This invention first discloses a visual inertial navigation method for VTOL aircraft oriented towards degenerate motion, comprising the following steps:

[0008] S1. Sensor data acquisition and preprocessing: Simultaneously acquire image sequence data collected by the downward-looking monocular camera, angular velocity and linear acceleration sequence data measured by the inertial measurement unit (IMU), and atmospheric pressure value sequence data measured by the barometer;

[0009] S2. Degradation Mode Determination and Observability Assessment: Determine whether the motion of the VTOL aircraft causes degradation of the visual inertial odometry. The degradation mode includes at least vertical translation, hovering rotation and compound motion modes. Quantify and assess the observability of the scale and translation, and output the current motion mode label and observability assessment results.

[0010] S3. Adaptive State Filtering and Navigation Result Output: Based on the multi-state constrained Kalman filter framework, the system adaptively executes the corresponding information fusion strategy to estimate the state according to the degradation mode and observability, and outputs navigation information.

[0011] Preferably, in step S1, hardware triggering ensures that the camera exposure time, IMU and barometer sampling timestamps received by the airborne computer are aligned; first, zero-bias calibration compensation is performed on the IMU data, and pre-integration is performed on the IMU data within the interval of adjacent image frames; then, the FAST feature point detector is used to perform distortion correction and feature extraction on the current image, and KLT sparse optical flow is used to track feature points; finally, the current air pressure value is converted into the current altitude value using a standard atmospheric model.

[0012] More preferably, step S2 as described above includes the following sub-steps:

[0013] S21: Analyze the principal components of motion excitation based on inertial information to preliminarily determine the tendency of degenerative motion;

[0014] S22: Degradation determination and quantification based on visual geometry: Based on the optical flow field characteristics and triangulation quality of visual feature points, identify the degradation motion pattern and evaluate the degree of degradation of system translation or scale observability.

[0015] S23: Output motion mode and degradation labels: Motion mode labels include "Normal / Vertical translation / Hovering rotation"; Observability degradation labels include "No degradation / Scale degradation / Translation degradation".

[0016] More preferably, in the aforementioned step S21, the determination principle is as follows: if a continuous and dominant vertical linear acceleration excitation is detected, while the horizontal linear acceleration and angular velocity excitation are weak, it is initially determined to be a vertical translation tendency; if a continuous angular velocity excitation is detected, while the horizontal linear acceleration excitation is weak, it is initially determined to be a rotational motion tendency.

[0017] More preferably, in the aforementioned step S22, the average divergence of the optical flow field at the feature points is calculated by analyzing visual information. and mean curl Degenerative movement patterns were identified based on different degenerative movement tendencies:

[0018] (1) Regarding the vertical translation tendency, if the divergence is significantly greater than the curl and the feature points exhibit consistent radial flow, then it is determined that the vertical translation degradation mode has been entered; further, the inverse depth uncertainty of the triangulated feature points is statistically analyzed, and if the average uncertainty exceeds the threshold, then the scale observability is determined to be degraded.

[0019] (2) Regarding the tendency of rotational motion, if the curl is significantly greater than the divergence and the feature points exhibit a rotational flow pattern around the center of the image, then it is confirmed that the hovering rotational degradation mode has been entered; further, the success rate of triangulation of new feature points is statistically analyzed. If it is lower than the preset translational degradation threshold, then the translational observability is determined to be degraded.

[0020] More preferably, in step S3 above, when the mode is vertical translation and the degradation label is scale degradation, the first strategy is executed, including:

[0021] Barometer-dominated altitude updates: reduce the noise variance of barometer altitude observations and adjust for scale observability;

[0022] Virtual planar motion damping: Introducing a virtual observation assuming zero horizontal velocity as a soft constraint to suppress horizontal velocity divergence;

[0023] Visual scale update weakening: Reduce the update weight of visual features on scale factors.

[0024] More preferably, in step S3 above, when it is a hovering rotation and the degradation flag includes translational degradation, a second strategy is executed, including:

[0025] Pause the triangulation initialization of new feature points;

[0026] For the inverse depth of the triangulated feature points, the parameters corresponding to them in the state covariance matrix are amplified by the inflation factor.

[0027] Translational state covariance freezing: Locks the predicted covariance of position and linear velocity to prevent the system from drifting without effective observations.

[0028] More preferably, in step S3 above, when a composite degradation mode of simultaneous vertical translation and hovering rotation is determined, a third strategy is executed, including:

[0029] Freeze the translation state covariance and pause feature point triangulation;

[0030] Barometer observation noise is reduced to a minimum;

[0031] Pose conservative updates are performed using only visual information.

[0032] More preferably, in the aforementioned step S3, when in the mode transition period, the strategies for different degradation modes are weighted and mixed; when the degradation motion mode label is normal, the standard multi-sensor fusion update strategy is executed.

[0033] This invention also discloses a visual inertial navigation system for VTOL aircraft oriented towards degenerate motion, comprising:

[0034] The sensor module includes a global shutter downward-looking monocular camera, a six-axis IMU, and a digital barometer, used to simultaneously acquire image sequence data collected by the downward-looking monocular camera, angular velocity and linear acceleration sequence data measured by the inertial measurement unit (IMU), and atmospheric pressure values ​​measured by the barometer.

[0035] The data synchronization and preprocessing module receives the data collected from the sensor module and processes the camera data, IMU data, and barometer data.

[0036] The Degenerative Motion Pattern Determination and Evaluation Module is used to determine the degenerative motion pattern and evaluate the system's observability.

[0037] An adaptive state filtering module, the strategy execution logic of which is connected to the output of the degenerate motion mode determination and evaluation module;

[0038] The navigation information output module is used to output the estimated navigation state, including position, attitude, linear velocity, and scale factor.

[0039] Preferably, the aforementioned degenerative movement pattern determination and evaluation module includes:

[0040] Inertial information analysis unit, used to make preliminary judgments on motion tendency;

[0041] The visual information analysis unit is used to determine degenerative movement patterns and assess system observability.

[0042] The advantages of this invention are:

[0043] The visual inertial navigation method and system for VTOL aircraft oriented towards degenerate motion of the present invention does not rely on external navigation signals. It can autonomously determine vertical translation, hovering rotation and their combined degenerate motion modes. By suppressing the navigation divergence caused by degenerate motion through observability assessment and adaptive filtering, it solves the defect of visual inertial navigation being prone to failure in low observability scenarios. It significantly improves the navigation robustness of VTOL aircraft in critical phases such as take-off, landing and hovering in denied environments, and ensures flight safety throughout the mission. Attached Figure Description

[0044] Figure 1 This is the overall flowchart of the method of the present invention;

[0045] Figure 2 Flowchart for determining and evaluating degenerative movement patterns;

[0046] Figure 3 Flowchart of the adaptive filtering update strategy;

[0047] Figure 4 A structural diagram of a visual inertial navigation system for a VTOL aircraft designed for degenerative motion;

[0048] Figure 5 This is a comparison chart showing the cumulative error between the method of this invention and existing technologies. Detailed Implementation

[0049] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0050] Example 1

[0051] See Figure 1 and Figure 2 This embodiment discloses a visual inertial navigation method for VTOL aircraft with degenerate motion, including the following steps:

[0052] S1. Data Synchronization and Preprocessing: Synchronously collect and preprocess data from the downward-facing camera, IMU, and barometer. Ensure that the camera exposure time, IMU, and barometer sampling timestamps received by the onboard computer are aligned through hardware triggering.

[0053] First, zero-bias calibration compensation is performed on the IMU data, and pre-integration is performed on the IMU data within the interval of adjacent image frames.

[0054] Next, the FAST feature point detector is used to perform distortion correction and feature extraction on the current image, and KLT sparse optical flow is used to track the feature points.

[0055] Finally, the current air pressure value is converted to the current altitude value using a standard atmospheric model:

[0056] ,in, The sea level reference pressure value is obtained through ground static calibration before takeoff. This is the current air pressure value.

[0057] S2. Degradation Mode Determination and Evaluation: Online determination of whether the motion of the VTOL aircraft causes visual inertial odometry observability degradation. The degradation mode includes at least vertical translation, hovering rotation and compound motion modes, and quantitative evaluation of scale observability and translation observability.

[0058] See Figure 2 The flowchart shown includes the following sub-steps for step S2:

[0059] S21. Using the inertial analysis unit, the tendency of degenerative motion is initially determined based on the inertial mechanism.

[0060] Among them, the inertial analysis unit is used to calculate the past Variance of linear acceleration and variance of angular velocity within the time window.

[0061] Set the window inside the first IMU sample values: linear acceleration angular velocity The variances of linear acceleration and angular velocity are calculated as follows:

[0062] ,

[0063] .

[0064] If the variance of linear acceleration Mainly from the variance of the vertical component And the variance of the horizontal components Below the threshold and angular velocity variance All below the threshold Then, the "vertical translation tendency" is determined, and the determination condition is as follows:

[0065] ,

[0066] in, The proportional threshold can be set to... ; The threshold value for the variance of the horizontal component of linear acceleration can be set to... ; The angular velocity variance threshold can be set to... .

[0067] Similar logic can also be used to determine "rotational motion tendency", with the following conditional expression:

[0068]

[0069] in, The threshold value for the variance of angular velocity in rotation determination can be set to... , The linear acceleration threshold can be set to... .

[0070] S22. Using the visual analysis unit, determine the degradation motion based on features and assess the degree of degradation of system translation or scale observability.

[0071] For a successfully tracked feature point, let its coordinates on the normalized image plane be... The optical flow vector is .

[0072] Calculate the average divergence of all optical flow vectors and mean curl :

[0073] ,

[0074] ;

[0075] The partial derivatives are approximated by the difference between adjacent feature points. Indicates the number of feature points. and These represent the x and y coordinates of the feature points in the image coordinate system, respectively. and These represent the velocity vectors of the feature point along the x-axis and y-axis, respectively.

[0076] For each feature point, calculate the dot product of its position vector and optical flow vector: ,in, The center of the image.

[0077] The probability that the statistical dot product is positive is the radial consistency probability of the flow at the feature points. If the threshold is exceeded, the feature points are considered to exhibit a consistent radial flow pattern.

[0078] If the aforementioned step S21 determines that there is a "vertical translation tendency", and the conditions are met... , Radial consistency probability This confirms entry into the "vertical translation degradation mode." A proportionality coefficient greater than 1 can be set to 2.0. The preset absolute divergence threshold can be set to 0.05. It can be set to 0.7.

[0079] Furthermore, the inverse depth of all successfully triangulated feature points is calculated. The variance is Scale observability is defined as... .in, The minimum effective depth can be set to 0.3m. When If the value is less than a set threshold (e.g., 0.3), it is determined that "scale observability has degraded".

[0080] If the aforementioned step S21 determines that there is a "tendency towards rotational motion", and the conditions are met... , Radial consistency probability This confirms entry into "hover rotation mode". Among other things, A proportionality coefficient greater than 1 can be set to 2.0. The preset absolute curl threshold can be set to 0.03. It can be set to 0.3.

[0081] Furthermore, the success probability of triangulation of newly added feature points in the statistical image is defined as the translation observability. That is, the number of feature points that are successfully triangulated. Number of all feature points The proportion, if If the value is less than the set threshold (e.g., 0.7), it is determined that "translation observability has degraded".

[0082] S23. After determining the motion degradation mode and degree, publish the mode and degradation labels at a frequency of 10Hz. This includes: the current mode label (0: normal, 1: vertical translation, 2: hovering rotation), degradation label (0: none, 1: scale degradation, 2: translation degradation), and scale observability. Translation observability .

[0083] S3. Based on the multi-state constrained Kalman filter framework, the corresponding information fusion strategy is adaptively executed to perform state estimation according to the degradation mode and observability.

[0084] S31, Reference Figure 3 First, extended state prediction is performed, maintaining a sliding window containing 20 historical camera states. The state vector includes the IMU's pose, velocity, attitude, gyroscope and accelerometer bias, all historical camera poses, and the inverse depth of the feature points.

[0085] S32. Determine the vertical translation update strategy (first strategy).

[0086] When a "vertical translation" command is received, the noise variance of the barometer altitude observation is adjusted to: in the next filtering cycle.

[0087] ,

[0088] in, The default variance can be set to [value]. , This is the preset minimum weight coefficient, which can be set to 0.1.

[0089] When the scale degradation label is true, the virtual velocity observations are constructed as follows: the observed values ​​are... Observation matrix The velocity along the horizontal axis of the aircraft is 1, and all others are 0. The observation noise covariance is... ,in Can be set to .

[0090] Multiply the update weights of visual features to the scale factor by the decay factor. ,in Can be set to .

[0091] S33. Determine the hover rotation update strategy (second strategy).

[0092] When the command is identified as "hover rotation", the newly obtained sequence of feature points will no longer enter the triangulation candidate queue.

[0093] For the inverse depth of all triangulated feature points, calculate the corresponding elements (i.e., diagonal elements) in the covariance matrix based on the inflation factor. enlarge, Can be set to .

[0094] When the translation degradation label is true, the translation (position, linear velocity) derivatives are ignored when calculating the Jacobian matrix, only the attitude derivatives are retained, and the prediction covariance matrices of position and linear velocity are temporarily frozen.

[0095] S34. Determine the composite degradation cascade protection strategy (third strategy).

[0096] When both hovering rotation and vertical translation are simultaneously determined, a composite degradation cascade protection strategy is implemented: the predicted covariance matrix of position and linear velocity is completely frozen; the noise variance of barometer altitude observations is set to a minimum value. The pose is updated using only visual information.

[0097] S35, Transitional Update Mode: When the mode label changes in M ​​consecutive judgments, the judgment mode is in a transition period, and the transition coefficient is set to... , The transition period is linearly increasing from 0 to 1. .

[0098] During the transition from vertical translation to normal mode, the barometer observation noise variance gradually recovers to .

[0099] During the transition from hover rotation to normal mode, the translation covariance varies. The proportions are gradually thawing.

[0100] S36. Normal Update Mode: When the mode label is "Normal", the standard multi-state constrained Kalman filter update process is executed, and the sensor noise uses the default parameters.

[0101] S37. Output navigation status: Publish the odometry message obtained in step S3 at a frequency of 100Hz, including position, attitude, speed and scale factor.

[0102] Example 2

[0103] This embodiment discloses a navigation system for implementing the above-described visual-inertial navigation method, with reference to... Figure 4 The system includes:

[0104] The sensor module includes a global shutter downward-looking monocular camera, a six-axis IMU, and a digital barometer. It is used to simultaneously acquire image sequence data from the downward-looking monocular camera, angular velocity and linear acceleration sequence data measured by the inertial measurement unit (IMU), and atmospheric pressure values ​​measured by the barometer. All sensors achieve synchronous sampling through hardware synchronization. Specifically, the camera module can use a Sony IMX296, the IMU module can use a TDK ICM-42688-P, and the barometer module can use a TE MS5611.

[0105] The data synchronization and preprocessing module receives the data collected from the sensor module and processes the camera data, IMU data, and barometer data.

[0106] The Degenerate Motion Pattern Determination and Evaluation Module is used to determine the degenerate motion pattern and evaluate the system's observability. This module is the core of the system and specifically includes: an inertial information analysis unit for preliminary judgment of motion tendency; and a visual information analysis unit for determining the degenerate motion pattern and evaluating the system's observability.

[0107] The adaptive state filtering module has its strategy execution logic connected to the output of the degenerate motion mode determination and evaluation module.

[0108] The navigation information output module is used to output estimated navigation state information.

[0109] See Figure 5Traditional VIO systems experience a gradual accumulation of errors during the vertical climb phase, and the eastward position error accelerates after the hovering and rotating phase, with a simultaneous and continuous drift in the altitude direction, exhibiting a clear divergence trend overall. In contrast, the method of this invention, through real-time determination of degenerate motion and correction of state constraints, maintains good tracking of the GNSS true value across all three axes, with significantly smaller position deviations than traditional methods.

[0110] In summary, the visual inertial navigation method and system for VTOL aircraft oriented towards degenerate motion of the present invention does not rely on external signals, can autonomously determine two types of degenerate motion: vertical translation and hovering rotation, and suppress the influence of the above-mentioned degenerate motion. It solves the problem of visual inertial navigation failure in observable degenerate motion modes, and can improve the navigation robustness of VTOL aircraft in critical phases such as take-off, landing, and hovering under denied environments, so as to ensure the safety of VTOL aircraft in all mission phases.

[0111] In the description of this invention, references to terms such as “embodiment,” “specific example,” or “practical application” indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment is included in at least one embodiment or example of the invention; the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0112] The above embodiments are only used to illustrate the technical solutions of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any way. All technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion, characterized in that, Includes the following steps: S1. Sensor data acquisition and preprocessing: Simultaneously acquire image sequence data collected by the downward-looking monocular camera, angular velocity and linear acceleration sequence data measured by the inertial measurement unit (IMU), and atmospheric pressure value sequence data measured by the barometer. S2. Degradation Mode Determination and Observability Assessment: Determine whether the motion of the VTOL aircraft causes degradation of the visual inertial odometry. The degradation mode includes at least vertical translation, hovering rotation and compound motion modes. Quantify and assess the observability of the scale and translation, and output the current motion mode label and observability assessment results. Step S2 includes the following sub-steps: S21: Analyze the principal components of motion excitation based on inertial information to preliminarily determine the tendency of degenerative motion; S22: Degradation determination and quantification based on visual geometry: Based on the optical flow field characteristics and triangulation quality of visual feature points, identify the degradation motion pattern and evaluate the degree of degradation of system translation or scale observability. S23: Output motion mode and degradation label: Motion mode label includes "Normal / Vertical Translation / Hover Rotation"; Observability degradation labels include "no degradation / scale degradation / translation degradation"; In step S21, the determination principle is as follows: Analyze the IMU information. If a continuous and dominant vertical linear acceleration excitation is detected, while the horizontal linear acceleration and angular velocity excitation are weak, it is initially determined to be a vertical translation tendency. If a continuous angular velocity excitation is detected, while the horizontal linear acceleration excitation is weak, it is initially determined to be a rotational motion tendency. S3. Adaptive State Filtering and Navigation Result Output: Based on the multi-state constrained Kalman filter framework, the system adaptively executes the corresponding information fusion strategy to estimate the state according to the degradation mode and observability, and outputs navigation information.

2. The visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion according to claim 1, characterized in that, In step S1, hardware triggering ensures that the camera exposure time, IMU, and barometer sampling timestamps received by the airborne computer are aligned. First, zero-bias calibration compensation is performed on the IMU data, and pre-integration is performed on the IMU data within the interval of adjacent image frames. Next, the FAST feature point detector is used to perform distortion correction and feature extraction on the current image, and KLT sparse optical flow is used to track feature points. Finally, the current air pressure value is converted into the current altitude value using a standard atmospheric model.

3. The visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion according to claim 1, characterized in that, In step S22, the average divergence and average curl of the optical flow field at the feature points are calculated by analyzing visual information, and the degradation motion mode is identified for different degradation motion tendencies: (1) Regarding the vertical translation tendency, if the divergence is significantly greater than the curl and the feature points exhibit consistent radial flow, then it is determined that the vertical translation degradation mode has been entered; further, the inverse depth uncertainty of the triangulated feature points is statistically analyzed, and if the average uncertainty exceeds the threshold, then the scale observability is determined to be degraded. (2) Regarding the tendency of rotational motion, if the curl is significantly greater than the divergence and the feature points exhibit a rotational flow pattern around the center of the image, then it is confirmed that the hovering rotational degradation mode has been entered; further, the success rate of triangulation of new feature points is statistically analyzed. If it is lower than the preset translational degradation threshold, then the translational observability is determined to be degraded.

4. The visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion according to claim 3, characterized in that, In step S3, when the mode is vertical translation and the degradation label is scale degradation, the first strategy is executed, including: Barometer-dominated altitude updates: reduce the noise variance of barometer altitude observations and adjust for scale observability; Virtual planar motion damping: Introducing a virtual observation assuming zero horizontal velocity as a soft constraint to suppress horizontal velocity divergence; Visual scale update weakening: Reduce the update weight of visual features on scale factors.

5. The visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion according to claim 3, characterized in that, In step S3, when it is a hover rotation and the degradation flag includes translation degradation, the second strategy is executed, including: Pause the triangulation initialization of new feature points; For the inverse depth of the triangulated feature points, the parameters corresponding to them in the state covariance matrix are amplified by the inflation factor. Translational state covariance freezing: Locks the predicted covariance of position and linear velocity to prevent the system from drifting without effective observations.

6. The visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion according to claim 3, characterized in that, In step S3, when a composite degradation mode involving both vertical translation and hovering rotation is determined, a third strategy is executed, including: Freeze the translation state covariance and pause feature point triangulation; Barometer observation noise is reduced to a minimum; Pose conservative updates are performed using only visual information.

7. The visual-inertial navigation method for VTOL aircraft oriented towards degenerate motion according to claim 3, characterized in that, In step S3, when in the mode transition period, the strategies for different degradation modes are weighted and mixed; when the degradation motion mode label is normal, the standard multi-sensor fusion update strategy is executed.

8. A visual inertial navigation system for VTOL aircraft oriented towards degenerate motion, characterized in that, To implement the method according to any one of claims 1-7, comprising: The sensor module includes a global shutter downward-looking monocular camera, a six-axis IMU, and a digital barometer, used to simultaneously acquire image sequence data collected by the downward-looking monocular camera, angular velocity and linear acceleration sequence data measured by the inertial measurement unit (IMU), and atmospheric pressure values ​​measured by the barometer. The data synchronization and preprocessing module receives the data collected from the sensor module and processes the camera data, IMU data, and barometer data. The Degenerative Motion Pattern Determination and Evaluation Module is used to determine the degenerative motion pattern and evaluate the system's observability. An adaptive state filtering module, the strategy execution logic of which is connected to the output of the degenerate motion mode determination and evaluation module; The navigation information output module is used to output the estimated navigation state, including position, attitude, linear velocity, and scale factor.

9. A visual inertial navigation system for VTOL aircraft oriented towards degenerate motion according to claim 8, characterized in that, The degenerative movement pattern determination and evaluation module includes: Inertial information analysis unit, used to make preliminary judgments on motion tendency; The visual information analysis unit is used to determine degenerative movement patterns and assess system observability.

Images