Radar data screening method, height detection method, device thereof, electronic equipment and unmanned aerial vehicle

Abstract

Embodiments of the present application relate to a radar data screening method, a height detection method, a device thereof, an electronic device and a UAV. The radar data screening method comprises: determining theoretical state information of a detected target based on motion information of a carrier; obtaining observed state information of the detected target by a radar; the radar is carried on the carrier; and weighting and superimposing the theoretical state information and the observed state information to obtain accurate state information of the detected target. By weighting and superimposing two different data sources, invalid data related to false objects generated by small objects and external noise can be effectively excluded from the original radar data.

Description

[Technical Field]

[0001] This invention relates to the field of radar data processing technology, and in particular to a radar data filtering method, an altitude detection method and apparatus thereof, electronic equipment and unmanned aerial vehicles. [Background Technology]

[0002] With the continuous development of technology, various types of vehicles and equipment, such as drones, have begun to appear widely in people's daily lives and can be applied to many different fields such as aerial photography, agricultural plant protection, express delivery, and disaster relief.

[0003] Therefore, ensuring the safe flight of drones has become an increasingly important issue of concern. For drones and similar unmanned aerial vehicles, obtaining accurate information about the external environment is a prerequisite for safe flight. In particular, real-time altitude information plays a crucial role in obstacle avoidance and takeoff and landing control.

[0004] Traditional height detection methods can include the following: 1) GPS-based height measurement: that is, measuring the height of a person based on the position of GPS satellites; 2) Barometer-based height measurement: that is, measuring atmospheric pressure and converting it into current height information; 3) Ultrasonic radar-based height measurement: that is, sending ultrasonic waves to the ground and calculating the echo time to obtain the height information between the person and the ground.

[0005] However, all of the aforementioned traditional methods have many drawbacks. For example, GPS altimetry requires calibration using the drone's takeoff point as the initial altitude, and it can only output the altitude difference between the drone's current position and the takeoff point. Therefore, it cannot provide accurate altitude information in areas with uneven ground. Furthermore, if low-cost GPS is used for altimetry, its low refresh rate will increase the risk of drone accidents. Barometer altimetry is easily affected by climate changes; for example, changes in airflow can introduce measurement errors. Ultrasonic radar altimetry, due to the characteristics of ultrasound, can only be used at low altitudes, with a measurement range typically less than 10 meters, which is insufficient to meet the actual needs of drone operations. [Summary of the Invention]

[0006] The radar data filtering method, altitude detection method, apparatus, electronic equipment, and UAV provided in this application can overcome at least some of the defects of existing altitude detection methods.

[0007] In a first aspect, embodiments of this application provide a radar data filtering method. The method includes: determining theoretical state information of a target based on the motion information of a vehicle; acquiring radar observation state information of the target; the radar being mounted on a vehicle; and weighted superposition of the theoretical state information and the observation state information to filter and obtain precise state information of the target.

[0008] Optionally, the weighted superposition of the theoretical state information and the observed state information to filter and obtain the precise state information of the target includes: performing several Kalman filtering processes iteratively based on the theoretical state information and the observed state information; and outputting the estimated state information of the target obtained after each Kalman filtering process as the precise state information.

[0009] Optionally, the Kalman filtering process includes: based on the state information estimate obtained after the previous Kalman filtering process, calculating the prior estimate of the theoretical state information through a preset first transformation relationship; based on the state vector estimation error covariance matrix obtained after the previous Kalman filtering process, calculating the current state vector estimation error covariance matrix; calculating the residual between the prior estimate and the observed state information according to a preset second transformation relationship; calculating the Kalman gain coefficient based on the current state vector estimation error covariance matrix, the prior estimate, and the observed state information; updating the current state vector estimation error covariance matrix based on the Kalman gain coefficient; and weighting the residual and the prior estimate using the Kalman gain coefficient as a weighting coefficient to obtain the state information estimate.

[0010] Optionally, the theoretical state information includes: the position information, velocity information, and acceleration information of the target in the three-dimensional coordinate system; the observation state information includes: the distance information between the target and the millimeter-wave radar, the azimuth angle information of the target relative to the millimeter-wave radar, the elevation angle information of the target relative to the millimeter-wave radar, and the radial velocity information of the target relative to the millimeter-wave radar; the first conversion relationship is established based on the acceleration model of the vehicle; the second conversion relationship is: the correspondence between the theoretical state information and the observation state information after linearization processing.

[0011] Secondly, embodiments of this application provide a height detection method. The method includes: acquiring precise state information of a target using the radar data filtering method described above; determining the radar tilt angle based on the vehicle's attitude information; determining whether a target exists within a search range; the search range being formed with the tilt angle as a reference, allowing for upward and downward fluctuations with an allowable error; if so, selecting the target closest to the tilt angle within the search range as a reference target; calculating the vehicle's current height information based on the precise state information of the reference target; if not, estimating the vehicle's current height information based on the vehicle's motion information and previously calculated height information.

[0012] Optionally, the radar is located at the bottom of the vehicle, close to the ground; determining the tilt angle of the radar based on the attitude information of the vehicle specifically includes: obtaining the pitch angle of the vehicle as the azimuth angle of the radar, and obtaining the roll angle of the vehicle as the pitch angle of the radar.

[0013] Optionally, calculating the current altitude information of the vehicle based on the precise state information of the reference detection target specifically includes: obtaining the distance between the reference detection target and the radar, and the azimuth angle of the radar; and calculating the current altitude information of the vehicle based on the distance between the reference detection target and the radar and the azimuth angle using trigonometric functions.

[0014] Thirdly, embodiments of this application provide a radar data filtering device. The device includes: a theoretical data acquisition module for determining the theoretical state information of a target based on the motion information of a vehicle; a radar data acquisition module for acquiring the radar's observation state information of the target; the radar is mounted on a vehicle; and a weighted superposition module for weighted superposition of the theoretical state information and the observation state information to filter and obtain the precise state information of the target.

[0015] Fourthly, embodiments of this application provide an electronic device. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the radar data filtering method and altitude detection method as described above.

[0016] Fifthly, this application provides an unmanned aerial vehicle (UAV). The UAV includes: a fuselage; a radar disposed at the bottom of the fuselage near the ground; an arm connected to the fuselage; a power unit disposed on the arm for providing flight power to the UAV; and a flight controller disposed on the fuselage and communicatively connected to the radar; wherein the flight controller is configured to execute the radar data filtering method and altitude detection method described above.

[0017] One advantage of the radar data filtering method provided in this application is that by weighting and superimposing two different data sources, invalid data related to small objects and false objects generated by external noise can be effectively excluded from the original radar data.

[0018] Another advantage of the height detection method provided in this application is that by searching and filtering target data that are close to the tilt angle of the radar and combining them with historical data for estimation, the limitation that the normal direction of the radar needs to be perpendicular to the ground when detecting the height above the ground is overcome, which can effectively improve the practicality. [Attached Image Description]

[0019] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0020] Figure 1 This is a schematic diagram of the application environment of an embodiment of this application;

[0021] Figure 2a The schematic diagram provided in this application illustrates the detection of different types of targets by millimeter-wave radar, showing both ground targets and non-ground targets.

[0022] Figure 2b This is a schematic diagram illustrating the change in altitude information detection data of a UAV provided in an embodiment of this application, showing the movement of the UAV from time k-1 to time k;

[0023] Figure 2c This is a schematic diagram illustrating the weighted superposition of two independent data sources provided in an embodiment of this application, showing the weighted superposition relationship between theoretical state information and observed state information;

[0024] Figure 3 A flowchart illustrating the radar data filtering method provided in this application embodiment;

[0025] Figure 4 A flowchart illustrating the height information correction method provided in this application embodiment;

[0026] Figure 5 This is a flowchart of a Kalman filtering method provided in an embodiment of this application;

[0027] Figure 6 A schematic diagram of the drone and radar provided in the embodiments of this application shows the drone tilting.

[0028] Figure 7a A schematic diagram illustrating the height information correction method combining pose information provided in an embodiment of this application;

[0029] Figure 7b A schematic diagram of a height information correction method combining pose information provided in another embodiment of this application;

[0030] Figure 8 Functional block diagram of the radar data filtering device provided in the embodiments of this application;

[0031] Figure 9a A functional block diagram of the height information correction device provided in the embodiments of this application;

[0032] Figure 9b A functional block diagram of a height information correction device provided in another embodiment of this application;

[0033] Figure 10 A schematic diagram of an electronic device provided in an embodiment of this application.

Detailed Implementation Methods

[0034] To facilitate understanding of the present invention, a more detailed description is provided below with reference to the accompanying drawings and specific embodiments. It should be noted that when an element is described as being "fixed to" another element, it can be directly on the other element, or one or more intermediate elements may exist between them. When an element is described as being "connected" to another element, it can be directly connected to the other element, or one or more intermediate elements may exist between them. The terms "upper," "lower," "inner," "outer," "bottom," etc., used in this specification indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0035] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0036] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0037] Millimeter-wave radar refers to a detection radar that operates in the millimeter-wave band. It boasts an extremely high refresh rate, capable of reporting the current ground altitude in real time at millisecond-level refresh frequencies. Furthermore, based on electromagnetic waves, it is unaffected by harsh environments such as heavy fog or sandstorms. Millimeter-wave radar determines information such as the distance, relative angle, and relative velocity between itself and surrounding obstacles based on the analysis of received intermediate-frequency signals.

[0038] Typically, when a drone detects a sudden change in altitude, such as a car, tree, or power line below its flight path, or when it detects a false obstacle due to unknown external noise, the drone cannot identify or eliminate this interference. The controller can only respond to the altitude change information, adjusting the drone's flight attitude accordingly or triggering obstacle avoidance warnings. Furthermore, millimeter-wave radar requires its antenna normal to be perpendicular to the ground to obtain accurate altitude information. This also imposes many limitations on drones equipped with millimeter-wave radar for ground altitude detection.

[0039] The applicant discovered that, on the one hand, by using Kalman filtering to weight and superimpose radar data obtained from millimeter-wave radar with other relatively independent data systems (e.g., theoretical data derived from the flight motion of UAVs), more accurate data information can be selected. This effectively eliminates interference from foreign objects (e.g., small non-ground targets) and filters out invalid data, preventing UAVs and similar vehicles from taking unnecessary evasive maneuvers or triggering unnecessary obstacle avoidance warnings. On the other hand, based on three-dimensional spatial geometry, the corresponding altitude calculation method can calculate accurate altitude information even when the millimeter-wave radar is tilted. Moreover, when millimeter-wave radar cannot provide reliable data, altitude information can be estimated using the vehicle's (e.g., UAV) motion information and previously calculated historical altitude data.

[0040] Figure 1 This is a schematic diagram illustrating the application environment provided in an embodiment of this application. The application environment uses a drone as an example. Figure 1As shown, the drone 10 includes: fuselage 11, arms 12, power unit 13, and flight controller 14.

[0041] The fuselage 11 is the main structure of the UAV 10. It has a suitable size and shape to meet practical needs, providing sufficient space to accommodate one or more functional modules and components. For example, various sensor devices, such as millimeter-wave radar, can be mounted on the fuselage 11. Specifically, the millimeter-wave radar 15 can be fixed to the underside of the fuselage 11. The underside of the fuselage refers to the side of the fuselage that is close to and faces the ground during normal use.

[0042] Arm 12 is the part that extends outward from the fuselage, serving as the mounting or fixing structure for the drone's power unit, such as propellers. The arm can be integrally formed with the fuselage or connected to the fuselage in a detachable manner. Typically, on a quadcopter drone, there can be four arms, extending symmetrically along the diagonal, forming four propeller mounting positions.

[0043] The power unit 13 is a structural device used to provide flight propulsion for the UAV. It can specifically employ any suitable type of power and structural design. For example, it could be a propeller driven by an electric motor, mounted and fixed at a position at the end of the arm.

[0044] The flight controller 14 is the core of the UAV control system built into the fuselage. It can be any type of electronic device with suitable logic and computational capabilities, including but not limited to processor chips based on large-scale integrated circuits, integrated system-on-a-chip (SoC), and processors and storage media connected via a bus. Depending on the functions to be implemented (e.g., executing the joint calibration method provided in the embodiments of this application), the flight controller 14 can include several different functional modules. These functional modules can be software modules, hardware modules, or a combination of software and hardware, and are modular devices used to implement one or more functions.

[0045] It should be noted that the embodiments of this application are provided for simplicity and exemplification, illustrating the application scenario of the radar data filtering method and the altitude information correction method based on the filtering method in UAVs. However, those skilled in the art will understand that, based on similar principles, the data filtering and altitude detection methods provided in the embodiments of this application can also be applied to altitude measurement scenarios of other vehicles. The inventive concepts disclosed in the embodiments of this application are not limited to... Figure 1 The application shown on the drone can also be used on other similar vehicles.

[0046] To fully illustrate the radar data filtering and altitude information correction method provided in the embodiments of this application, Figure 1 The specific application process in the application scenario shown below, combined with Figures 2a to 2c This paper provides a detailed description of the principle of filtering radar data and eliminating invalid data based on Kalman filtering.

[0047] 1) such as Figure 2a As shown, during the flight of the UAV 10 equipped with millimeter-wave radar, it detects the environment below the UAV at a certain refresh rate and feeds back point cloud data containing one or more detected targets. In some embodiments, point cloud data that has not been processed by the Kalman filter algorithm can also be referred to as "raw detection data", while point cloud data processed by the Kalman filter algorithm can be referred to as "corrected detection data".

[0048] The drone's flight controller uses point cloud data provided by millimeter-wave radar to monitor the distance between the drone and various detected targets, detecting situations where the drone's altitude is too low. If the altitude is too low, the flight controller considers this a risk of ground collision. To ensure flight safety, it typically issues emergency measures such as obstacle avoidance warnings, climbs, or changes in course to prevent accidents.

[0049] Those skilled in the art will understand that the point cloud data provided by millimeter-wave radar may contain some targets referred to as "non-ground targets," such as power towers, power lines, and vehicles. These non-ground targets do not actually interfere with the flight of the UAV, but the flight controller often mistakenly judges that the altitude is too low due to the close proximity of the UAV to these non-ground targets, and triggers corresponding emergency measures.

[0050] The applicant found that these non-ground targets are typically smaller in size compared to the ground, and the difference can be defined as whether the drone can pass through quickly or how long the drone stays on the target. For example, Figure 2a As shown, when the detected target A is large and the drone stays on it for more than 3 seconds or other time thresholds, its large size provides a landing surface for the drone, and therefore it can be considered "ground". However, when the detected target B is small and the time spent on it is short (or the transit time is short), such as less than 3 seconds or other time thresholds, it cannot serve as a landing surface for the drone and cannot be considered "ground," but rather a non-ground target. Therefore, it is possible to combine the drone's motion information and the millimeter-wave radar's observation information to exclude invalid data representing non-ground targets from the raw detection data, achieving non-ground target filtering and avoiding false triggering of automatic obstacle avoidance.

[0051] 2.1) As Figure 2bAs shown, in the system consisting of UAV 10 and detection target 20, the state of detection target 20 at any time can be described by the state vector shown in the following formula (1-1).

[0052]

[0053] Where x(k), y(k), and z(k) are the three-dimensional coordinates of the target in the Cartesian coordinate system, the first derivatives of x(k), y(k), and z(k) are the velocities of the target on the three coordinate axes, and the second derivatives of x(k), y(k), and z(k) are the accelerations of the target on the three coordinate axes.

[0054] 2.2) Based on the acceleration model, a state transition equation is established to describe the state change pattern of the target. In other words, given the target's state vector at the previous moment, the current state vector can be calculated using the state transition equation. This state transition equation is shown in the following formula (1-2):

[0055] S(k)=FS(k-1)+W(k) (1-2)

[0056] Where S(k) is the state vector of the target at the current moment, S(k-1) is the state vector of the target at the previous moment, F is the state transition matrix, and W(k) is the control variable of the UAV.

[0057] This control variable represents the acceleration, deceleration, or other similar effects on the drone's motion caused by control commands issued by the flight controller to the drive unit (such as motors). In some alternative implementations, if the millimeter-wave radar is mounted on another vehicle that does not change its motion state due to the aforementioned control commands, this part of the control variable can be omitted.

[0058] Specifically, the state transition matrix is ​​shown in the following formula (1-3):

[0059]

[0060] 3) In the system consisting of UAV 10 and target 20, in addition to the state information of target 20 that can be obtained by deducing based on the motion state of UAV, the millimeter-wave radar of UAV 10 can also provide detection data related to target 20.

[0061] In this embodiment, the detection data used to describe the state of the detection target 20 can be represented by the detection vector of the following formula (2-1).

[0062]

[0063] Where u(k) is the current detection vector of the millimeter-wave radar, and r(k) is the distance between the millimeter-wave radar and the target. Let r(k) be the azimuth angle of the target relative to the millimeter-wave radar, θ(k) be the elevation angle of the target relative to the millimeter-wave radar, and the first derivative of r(k) be the radial velocity of the target relative to the millimeter-wave radar.

[0064] 4.1) As Figure 2c As shown, the detection vector and the state vector originate from different sources and are two relatively independent data sets. Therefore, by using Kalman filtering, the observation data D1 from the millimeter-wave radar and the theoretical state data D2 of the target obtained from the UAV's motion state can be weighted and superimposed to help select the accurate data of the target, thereby achieving the effects of filtering out non-ground foreign objects and eliminating external noise interference.

[0065] 4.2) Based on the different ways in which the state vector and detection vector describe the target, the following transformation relationship can be determined between the two vectors, as shown in formula (2-2):

[0066] u(k)=H(S(k))+v(k) (2-2)

[0067] Where H(.) is the measurement matrix, and v(k) is the measurement noise covariance matrix. This measurement matrix is ​​calculated according to the following formula (2-3):

[0068]

[0069] Where x, y, and z represent the position information of the target in the three-dimensional coordinate system; the first derivatives of x, y, and z represent the velocity information of the target in the three-dimensional coordinate system.

[0070] 4.3) As those skilled in the art will understand, a typical Kalman filter is based on a linear model. However, as shown in equation (2-2) above, the transformation relationship between the state vector and the detection vector is nonlinear.

[0071] Therefore, in order to make it applicable to Kalman filtering, the transformation relationship between the two can be linearized, resulting in the linearized relationship shown in equation (2-4):

[0072] u(k)=H(S apr (k))+J H (S apr (k))[S(k)-S apr [(k-1)]+v(k) (2-4)

[0073] Where u(k) is the current detection vector of the millimeter-wave radar, H is the measurement matrix, and S... apr (k) is the prior estimate obtained by calculating the state transition equation based on the state vector obtained after Kalman filtering at time k-1. H It is obtained by Taylor series expansion, as shown in the following formula (2-5):

[0074]

[0075] Those skilled in the art will understand that the Kalman filter algorithm can be implemented based on the state vector, detection vector, state transition equation, and linearized relationship between the state vector and detection vector provided in the embodiments of this application. The specific derivation and implementation process is well known to those skilled in the art and will not be elaborated here.

[0076] It should be noted that the specific examples of state vectors, detection vectors, state transition equations, and linearized relationships between state vectors and detection vectors provided in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of this application. Depending on practical needs or the characteristics of specific application scenarios, those skilled in the art can readily conceive of adjusting, replacing, or changing one or more steps or parameters to obtain other state vectors, detection vectors, state transition equations, and linearized relationships between state vectors and detection vectors through reasonable derivation.

[0077] It should be noted that the above embodiments use millimeter-wave radar as an example to describe the processing of radar data. Those skilled in the art will understand that, based on the same principle, it can also be applied to other different types of radar, and is not limited to millimeter-wave radar. Therefore, to avoid unnecessary limitations, the term "radar" will be used throughout this text.

[0078] Figure 3 This is a flowchart illustrating a radar data filtering method provided in an embodiment of this application. The radar data filtering method uses... Figures 2a-2c Based on the Kalman filter principle, this method introduces motion-related data sources to help obtain more accurate target information. For example... Figure 3 As shown, the radar data filtering method includes the following steps:

[0079] S310. Based on the motion information of the vehicle, determine the theoretical state information of the target to be detected.

[0080] Specifically, the vehicle can be any suitable type of mobile device, such as... Figure 1The drone shown is not specifically limited in its implementation. "Motion information" refers to data related to the motion of the vehicle, or data items used to describe or define the vehicle's motion, such as acceleration. "State information" refers to data directly describing the state of the target at a specific moment. For example, taking the state vector in the above specific example, it can include position, velocity, and acceleration. In this embodiment, the term "theoretical state information" is used to represent the state information calculated from the vehicle's motion.

[0081] S320: Acquire radar observation status information of the target being detected.

[0082] For example, such as Figure 1 As shown, the radar is mounted on a vehicle. To distinguish between two relatively independent sources of data, this embodiment uses the term "observation status information" to refer to the data information obtained through radar detection. For example, taking the observation vector in the specific example above, it may include relative distance, relative azimuth angle, relative elevation angle, and relative radial velocity.

[0083] S330, weighted superposition of theoretical state information and observational state information to filter and obtain accurate state information of the target.

[0084] "Weighted overlay" refers to assigning appropriate weight coefficients to two sets of data, multiplying them by the weight coefficients, and then adding them together. (See above.) Figure 2c As shown, theoretical state information and observed state information are two relatively independent data sources. By calculating the Kalman gain coefficient and then using it as a weighted superposition coefficient, the two data can be combined to select more accurate state information, thereby filtering radar data and eliminating the influence of non-ground targets or external noise.

[0085] Figure 4 This is a flowchart illustrating a height information correction method provided in an embodiment of this application. The height information correction method is implemented using a Kalman filter algorithm. Figure 4 As shown, the height information correction method includes the following steps:

[0086] S410: Acquire raw radar detection data.

[0087] As described above, the raw detection data includes observation status information of the detected target. In this embodiment, the term "observation status information" is used to represent the data information obtained by the radar sensor. It can have specific data depending on the actual situation. For example, taking the observation vector in the above specific example as an example, it can include relative distance, relative azimuth angle, relative pitch angle, and relative radial velocity.

[0088] S420. Based on the vehicle's motion information, calculate the theoretical state information of the target being detected.

[0089] The radar is mounted on a vehicle. This vehicle can be any suitable type of mobile device, such as... Figure 1 The drone shown is not specifically limited in its implementation.

[0090] "Motion information" refers to data related to the vehicle's motion, or data items used to describe or define the vehicle's motion, such as acceleration. "Theoretical state information" refers to data calculated based on the vehicle's motion, directly describing the state of the target at a specific moment. For example, taking the state vector in the above specific instance as an example, it can include position, velocity, and acceleration.

[0091] S430. Combining theoretical state information, the Kalman filter algorithm is used to eliminate invalid data in the original detection data to obtain corrected detection data.

[0092] Theoretical state information and raw detection data are two relatively independent data sources. As described in the specific example above, the two can be combined by calculating the Kalman gain coefficient, thereby eliminating invalid data in the raw detection data that is not related to ground targets.

[0093] In this embodiment, for ease of explanation, the term "invalid data" is used to refer to the observation status information of non-ground targets in the radar detection data. As described in the specific examples above, the characteristics of this non-ground target are its small size or the fact that the UAV can quickly pass by it. Therefore, the criterion can be the time required for the vehicle to pass over the detected target. In other words, if the time required for the vehicle to pass over the detected target is less than a preset threshold, the detected target can be determined to be a non-ground target.

[0094] S440. Determine the vehicle's height information by correcting the detection data.

[0095] In this process, the detection data was corrected to exclude the influence and interference of non-ground targets. The height information obtained from this correction can more accurately reflect the vehicle's true ground clearance, thus eliminating the influence of non-ground targets or external noise.

[0096] Those skilled in the art will understand that Figure 4 The "correcting detection data" step in the method steps shown can be considered as Figure 3 The method steps shown combine precise state information of one or more targets. In other words, after filtering out invalid data, the remaining precise state information of the targets is the corrected detection data.

[0097] One advantage of the altitude information correction method provided in this application is that by using the Kalman filter algorithm, the motion information of the vehicle is incorporated into the radar detection data, thereby effectively excluding the detection data of small objects that the vehicle can quickly pass over from the original radar detection data. This allows the detected vehicle altitude information to more accurately reflect the actual flight situation, thereby avoiding unnecessary obstacle avoidance warnings or flight attitude adjustments.

[0098] In some embodiments, during the process of correcting altitude information (e.g., step S430) and filtering to obtain accurate state information of the target (e.g., step S330), several sub-Kalman filtering processes can be performed iteratively, and the estimated state information of the target obtained after Kalman filtering is output as filtered data and applied to subsequent calculation of altitude information.

[0099] Specifically, after the Kalman filtering process described above, the excluded invalid data may also include observation status information of false detection targets caused by external noise. This achieves the effect of filtering out external noise generated by vehicles such as UAVs during flight, thus improving the accuracy of altitude information.

[0100] Figure 5 This is a flowchart illustrating a Kalman filtering method provided in an embodiment of this application. It exemplarily demonstrates the process of one iteration. Figure 5 As shown, the Kalman filtering process includes:

[0101] S510. Based on the state information estimate obtained after the previous Kalman filtering, calculate the prior estimate of the theoretical state information through the preset first transformation relationship.

[0102] The "first transformation relation" is a function representing the change in the theoretical state information of the target at different times. It can be established based on the vehicle's acceleration model, for example, it can be the state transition equation as described above.

[0103] In other embodiments, when calculating the prior estimate, the prior estimate can also be corrected directly using the vehicle's acceleration information.

[0104] S520. Based on the state vector estimation error covariance matrix obtained after the previous Kalman filtering, calculate the current state vector estimation error covariance matrix.

[0105] The covariance matrix is ​​a matrix that represents the relationship between different data. In this embodiment, the "state vector estimation error covariance matrix" is a symmetric matrix used to represent the correlation between multiple different data information of theoretical state information. Each element in this covariance matrix represents the relationship between different state variables.

[0106] S530. Calculate the residual between the prior estimate and the observed state information according to the preset second transformation relationship.

[0107] Here, the "second transformation relationship" refers to the correspondence between observed state information and theoretical state information. As described in the specific example above, when applied to Kalman filtering, this second transformation relationship is a linearized transformation relationship.

[0108] S540. Calculate the Kalman gain coefficient based on the current state vector estimation error covariance matrix, prior estimates, and observed state information.

[0109] The "Kalman gain coefficient" represents the relative reliability of theoretical or observed state information. In other words, it reflects the degree of certainty regarding the accuracy of the theoretical or observed state information. Understandably, in the weighted superposition process, the components with higher reliability or greater accuracy should have higher weights.

[0110] S550. Using the Kalman gain coefficient as the weighting coefficient, the residual and the prior estimate are weighted and superimposed to obtain the estimated state information of the target.

[0111] Once the Kalman gain coefficient is determined, more accurate state data can be obtained through weighted superposition. In this embodiment, the term "state information estimate" is used to represent the point cloud data obtained after Kalman filtering. Because it combines the vehicle's motion information and the millimeter-wave radar's detection information, the observation state data corresponding to small non-ground targets that the vehicle can quickly pass through are considered invalid data and excluded from the correction data due to the smaller Kalman gain coefficient assigned to them.

[0112] S560. Update the current state vector estimation error covariance matrix according to the Kalman gain coefficient.

[0113] In addition to the state information estimate after processing, the state vector estimation error covariance matrix also needs to be updated so that it can be applied to the next Kalman filtering process, thus achieving multiple iterations.

[0114] It should be noted that, Figure 5The flowchart shown is for illustrative purposes only and is not intended to limit the execution order of the steps. The order of independent steps can be arbitrarily adjusted and is not limited to this. Figure 5 As shown. For example, the update step of the state vector estimation error covariance is independent of the calculation of the state information estimate; it can be performed before or after the calculation of the state information estimate.

[0115] To fully illustrate the specific execution process of Kalman filtering, the following uses the state transition equations, state vectors, observation vectors, and the linearized transformation relationship between them described in equations (1-1) to (2-5) of the above specific example as examples to illustrate the process. Figure 5 The method and steps shown are described in detail.

[0116] 1) When performing step S510, the prior estimate can be calculated according to the following formula (3-1):

[0117] S apr (k)=FS(k-1) (3-1)

[0118] Among them, S apr (k) is the prior estimate, S(k-1) is the state information estimate obtained after the previous Kalman filtering process, and F is the state transition matrix.

[0119] 2) When performing step S520, the covariance matrix can be calculated according to the following formula (3-2):

[0120] P apr (k)=FP(k-1)F T +Q(k-1) (3-2)

[0121] Among them, P aqr (k) is the current state vector estimation error covariance matrix; P(k-1) is the updated state vector estimation error covariance matrix after the previous Kalman filtering process; Q(k-1) is the process noise covariance matrix.

[0122] "Process noise" refers to untraceable external forces that affect the system due to unforeseen circumstances, such as a drone being affected by sudden wind. These uncertainties can be modeled uniformly, thus adding corresponding terms to equation (3-2).

[0123] 3) When performing step S530, the residual between the prior estimate and the observed state information can be calculated according to the following formula (3-3):

[0124] y(k)=u(k)-H(S apr(k)) (33-33)

[0125] Where y(k) is the residual, u(k) is the observation state information, and S aqr (k) is the prior estimate; H(.) is the measurement matrix.

[0126] 4) When performing step S540, the Kalman gain coefficient can be calculated according to the following formula (3-4):

[0127]

[0128] Among them, P aqr (k) is the current covariance matrix, S aqr R(k) is the prior estimate, and R(k) is the measurement noise covariance matrix.

[0129] "Measurement noise" refers to the fluctuation of readings from sensors such as radar within a specific range due to Gaussian noise. In other words, radar data acquired by radar will appear as Gaussian spots with a range.

[0130] 5) When performing step S550, the estimated value of the state information can be calculated according to the following formula (3-5):

[0131] S(k)=Sapr(k)+K(k)y(k) (3-5)

[0132] Where S(k) is the estimated state information value; K(k) is the Kalman gain coefficient; S aqr y(k) is the prior estimate; y(k) is the residual.

[0133] 6) When performing step S560, the posterior error covariance matrix can be updated according to the following formula (3-6):

[0134] P(k)=P apr (k)-K(k)J H (S apr (k))P apr (k) (3-6)

[0135] Where P(k) is the covariance matrix of the updated state vector estimation error, P aqr (k) is the current state vector estimation error covariance matrix, K(k) is the Kalman gain coefficient, and S aqr (k) is the prior estimate.

[0136] One advantage of the altitude detection method provided in this application embodiment is that by using Kalman filtering, the theoretical state information and the observed state information of the target are combined to filter out accurate point cloud data that meets the needs of use, thereby achieving the effect of filtering out small-volume interference objects and false foreign objects caused by external noise that appear in the flight phase of vehicles such as UAVs.

[0137] In addition to using Kalman filtering to eliminate interference from small foreign objects, another embodiment of this application provides a method for obtaining specific height information by combining vehicle attitude. In this application, to distinguish it from the above-described method for eliminating interference using Kalman filtering, it is hereinafter referred to as the "height detection method combined with vehicle attitude".

[0138] To fully illustrate the height detection method for vehicle attitude provided in the embodiments of this application, Figure 1 The specific application process in the application scenario shown below, combined with Figure 6 The principle of calculating ground clearance by combining the attitude of the UAV is described in detail.

[0139] First, during flight, the roll and pitch angles of the drone inevitably change, which means that the radar's normal direction cannot always remain perpendicular to the ground.

[0140] The applicant discovered that when the radar is installed on a drone, the normal direction of the millimeter-wave radar and the drone's Z-axis (i.e., the drone's body coordinate axis) can be obtained based on the drone's attitude information. Figure 6 The angle between the vertical downward direction shown is used to further determine or correct the drone's altitude information.

[0141] by Figure 6 Taking the radar installation method shown as an example, the radar 15 can be fixedly installed on the underside 14 of the UAV fuselage. The underside 14 refers to the part of the UAV fuselage facing the ground when it is in flight.

[0142] Wherein, point C is the antenna phase center of radar 15, CB is the normal direction of the radar, CA is the altitude of the UAV, and CB is perpendicular to AB. The pitch angle of the UAV corresponds to the azimuth angle of the radar (hereinafter referred to as azimuth angle for simplicity), while the roll angle of the UAV corresponds to the pitch angle of the radar (hereinafter referred to as pitch angle for simplicity). The two angles are consistent in terms of positive and negative values.

[0143] Therefore, it can be determined that the roll angle and pitch angle of the UAV are equivalent to the angle between the radar normal direction and the Z-axis of the body coordinate system.

[0144] It should be noted that the specific examples of radar and UAV installation methods provided in this application are for illustrative purposes only and are not intended to limit the scope of this application. Based on practical needs or the characteristics of specific usage scenarios, those skilled in the art can readily conceive of adjusting, replacing, or changing one or more steps or parameters, and through reasonable deduction, obtain the correspondence between the radar's normal direction and the UAV's body coordinate axis Z-axis, and the UAV's attitude information in other installation methods.

[0145] Figure 7a This is a flowchart illustrating a method for detecting the altitude of a vehicle's attitude, as provided in an embodiment of this application. Figure 7a As shown, the height detection method for combined vehicle attitude includes the following steps:

[0146] S701. Determine the radar tilt angle based on the vehicle's attitude information.

[0147] Here, "attitude information" refers to the tilt angle of the vehicle in three-dimensional space. It can be described in any suitable way, such as quaternions and Euler angles. "Tilt angle" refers to the deviation between the radar's normal direction and the direction perpendicular to the ground. In some embodiments, such as... Figure 6 As shown, the tilt angle can be a combination of the UAV's pitch angle and roll angle (equivalent to the radar's azimuth and pitch angle).

[0148] S703. Determine whether there is a target within the search range in the corrected detection data. If yes, proceed to step S705; otherwise, proceed to step S709.

[0149] The search range refers to a numerical range formed by floating within an allowable error, with the tilt angle as the reference. This allowable error is an empirical value and can be set by those skilled in the art according to the actual needs. Preferably, the allowable error can be set to 5°.

[0150] S705. Within the search range, select the detection target that is closest to the tilt angle as the baseline detection target.

[0151] There may be one or more targets within the search range. If there is only one target, it can be directly identified as the baseline target without selection. If there are multiple targets, the closest one is selected as the baseline target.

[0152] S707. Calculate the current altitude information of the vehicle based on the observation status information of the benchmark detection target.

[0153] As described above, "observation status information" refers to a series of data signals corresponding to ground targets obtained by radar detection. The altitude above the ground can be calculated using one or more of these data signals, serving as the vehicle's current altitude information.

[0154] S709. Estimate the current height of the vehicle based on the vehicle's motion information and the previously calculated height information.

[0155] "Motion information" refers to the vehicle's movement in three-dimensional space. Based on the calculation of motion information, the relative change in altitude before and after movement can be obtained. Therefore, even without detecting a target, the current altitude of the vehicle can be approximated using pre-determined historical altitude information and the relative change determined by motion information.

[0156] In other embodiments, it is also possible to employ Figure 7b The method shown achieves the same effect by filtering and identifying the detection target closest to the tilt angle. For example... Figure 7b As shown, the method includes:

[0157] S702. Determine the radar tilt angle based on the vehicle's attitude information.

[0158] Here, "attitude information" refers to the tilt angle of the vehicle in three-dimensional space. It can be described in any suitable way, such as quaternions and Euler angles. "Tilt angle" refers to the deviation between the radar's normal direction and the direction perpendicular to the ground. In some embodiments, such as... Figure 5 As shown, the tilt angle can be a combination of the UAV's pitch angle and roll angle (equivalent to the radar's azimuth and pitch angle).

[0159] S704. Identify the target closest to the tilt angle.

[0160] In this method, radar point cloud data can be used to filter and determine the target closest to the tilt angle. In some embodiments, when there are multiple different tilt angle data, the "closest" can be determined based on the difference between the target and all data items.

[0161] S706. Determine whether the nearest detection target is within the preset search range. If yes, proceed to step S708; otherwise, proceed to step S710.

[0162] The search range refers to a numerical range formed by floating within an allowable error based on the tilt angle. This allowable error is an empirical value and can be set by those skilled in the art according to the actual needs. Preferably, the allowable error can be set to 5°, that is, the search range is (α, β) ± 5°.

[0163] S708. Calculate the current altitude of the vehicle based on the observation status information of the target.

[0164] The "observation status information," as described above, refers to a series of data signals obtained by radar detection. The altitude above the ground can be calculated using one or more of these data signals, serving as the vehicle's current altitude information.

[0165] S710. Estimate the current height of the vehicle based on the vehicle's motion information and the previously calculated height information.

[0166] "Motion information" refers to the vehicle's movement in three-dimensional space. Based on the calculation of motion information, the relative change in altitude before and after movement can be obtained. Therefore, even without detecting a target, the current altitude of the vehicle can be approximated using pre-determined historical altitude information and the relative change determined by motion information.

[0167] In some embodiments, when the vehicle's pose information is represented using quaternions, the following steps may be further included: First, obtaining the quaternion of the vehicle. Then, converting the quaternion into Euler angles to obtain the vehicle's pitch and roll angles.

[0168] Specifically, the conversion between quaternions and Euler angles can be represented by the following formula (4):

[0169]

[0170] The quaternion is represented as [q1, q2, q3, q4]; the Euler angle is represented as [α, β, γ].

[0171] In some embodiments, step S707 or step S708 specifically includes: first, obtaining the distance between the reference detection target and the radar, and the azimuth angle of the radar; then, calculating the current altitude information of the vehicle based on the distance between the reference detection target and the radar and the azimuth angle using trigonometric functions.

[0172] Specifically, such as Figure 6 As shown, the ground clearance of the vehicle can be calculated using the following formula (5):

[0173] H=R cosβ (5)

[0174] Where H represents the current altitude of the vehicle, R represents the distance between the reference target and the radar, and β represents the azimuth angle of the radar.

[0175] In some embodiments, step S690 or step S610 specifically includes: estimating the current height information of the vehicle using the following formula (6):

[0176] H(k)=H(k-1)+vt+0.5at 2 (6)

[0177] Wherein, H(k) is the current altitude information of the vehicle, H(k-1) is the previously calculated altitude information; v is the velocity of the vehicle in the altitude direction; a is the acceleration of the vehicle in the altitude direction; t is the time elapsed between H(k) and H(k-1), such as the refresh time of millimeter-wave radar.

[0178] One advantage of the altitude information detection method provided in this application is that when a drone or similar vehicle tilts during flight and the radar antenna normal is not perpendicular to the ground, it can be corrected to obtain accurate altitude information.

[0179] Another advantage of the altitude information detection method provided in this application embodiment is that when the radar does not detect ground target information, the current altitude information can be estimated by combining motion information to calculate the relative change in altitude and historical altitude information, ensuring that the radar-based altitude detection method can be used even in non-vertical states.

[0180] Based on the corrected height information obtained by the height information correction method provided in this application, this application also provides an automatic obstacle avoidance method. This automatic obstacle avoidance method includes the following steps:

[0181] First, a more accurate vehicle height information can be calculated using one or more of the height information correction methods described above. Then, it is determined whether this height information is less than a preset height threshold. Finally, if it is less than the preset height threshold, an obstacle avoidance warning is triggered. If it is not less than the preset height threshold, the original trajectory of the drone or other vehicle can be maintained.

[0182] It should be noted that the preset altitude threshold is an empirical value and can be adjusted or set by technicians according to actual needs. In some embodiments, it can also be adjusted to other suitable judgment criteria (e.g., altitude change rate threshold), and is not limited to the altitude threshold. In other embodiments, triggering an obstacle avoidance warning can be replaced by other suitable obstacle avoidance operations, such as automatically increasing flight altitude or changing course, in response to changes in altitude information.

[0183] One advantage of the automatic obstacle avoidance method provided in this application is that the point cloud data provided by the above-mentioned height information correction method excludes invalid data corresponding to small non-ground targets. Therefore, the height information calculated therefrom will not be affected by these non-ground targets, thus avoiding the accidental triggering of obstacle avoidance warnings and / or changes in flight trajectory by vehicles such as drones during automatic obstacle avoidance.

[0184] Figure 8 A radar data filtering device is provided in an embodiment of this application. The radar data filtering device 800 includes: a theoretical data acquisition module 810, a radar data acquisition module 820, and a weighted superposition module 830.

[0185] The theoretical data acquisition module 810 is used to determine the theoretical state information of the target based on the motion information of the vehicle. The radar data acquisition module 820 is used to acquire the radar's observation state information of the target; the radar is mounted on the vehicle. The weighted superposition module 830 is used to weight and superimpose the theoretical state information and the observation state information to filter and obtain the precise state information of the target.

[0186] In some embodiments, the weighted superposition module 830 is specifically used to: perform several iterations of Kalman filtering based on observed state information and theoretical state information; and output the estimated state information of the target obtained after each Kalman filtering process as the accurate state information of the target.

[0187] Figure 9a This is a functional block diagram of the height information correction device provided in an embodiment of this application. Figure 9a As shown, the altitude information correction device 900 may include: a radar data acquisition module 910, a theoretical data acquisition module 920, a data correction module 930, and an altitude information calculation module 940.

[0188] The radar data acquisition module 910 acquires the radar's raw detection data, including the observation status information of the target. The theoretical data acquisition module 920 calculates the theoretical status information of the target based on the vehicle's motion information. The radar is mounted on the vehicle. The data correction module 930, based on the theoretical status information, uses a Kalman filter algorithm to eliminate invalid data from the raw detection data to obtain corrected detection data. The altitude information calculation module 940 determines the vehicle's altitude information using the corrected detection data.

[0189] Specifically, invalid data refers to the observation status information of non-ground targets. Non-ground targets are: detection targets whose elapsed time is less than a preset threshold; the elapsed time is the time required for the vehicle to pass the detection target.

[0190] In some embodiments, the data correction module 930 is specifically used to perform several iterations of Kalman filtering based on the observed state information and the theoretical state information; and output the estimated state information of the target obtained after each Kalman filtering process as the corrected detection data.

[0191] Specifically, the theoretical state information includes: the position, velocity, and acceleration information of the target in the three-dimensional coordinate system; the observation state information includes: the distance between the target and the radar, the azimuth angle of the target relative to the millimeter-wave radar, the elevation angle of the target relative to the radar, and the radial velocity of the target relative to the radar.

[0192] In other embodiments, please refer to Figure 9b In addition to the aforementioned functional modules, the height information correction device 900 may also include: an inclination angle acquisition module 950, a target search module 960, a height calculation module 970, and a height estimation module 980.

[0193] The tilt angle acquisition module 950 is used to determine the tilt angle of the radar based on the attitude information of the vehicle; the target search module 960 is used to determine whether there is a target within the search range in the corrected detection data; the search range is formed with the tilt angle as a reference, with an allowable fluctuation range. The altitude calculation module 970 is used to select the target closest to the tilt angle as the reference target when there is a target within the search range; and calculate the current altitude information of the vehicle based on the observation data of the reference target; the altitude estimation module 980 is used to estimate the current altitude information of the vehicle based on the motion information of the vehicle and the previously calculated altitude information when there is no target within the search range.

[0194] Specifically, the radar is located at the bottom of the vehicle close to the ground, and the tilt angle of the radar includes: the elevation angle of the vehicle, which is the azimuth angle of the radar; and the roll angle of the vehicle, which is the elevation angle of the millimeter-wave radar.

[0195] In some embodiments, the altitude calculation module 970 is specifically used to: obtain the distance between the reference detection target and the radar, and the azimuth angle of the radar; and calculate the current altitude information of the vehicle based on the distance between the reference detection target and the radar and the azimuth angle using trigonometric functions.

[0196] It should be noted that, in the embodiments of this application, functionally named modules are used as examples to describe in detail the method steps to be implemented by the device provided in the embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. Those skilled in the art will realize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0197] Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application. The computer software can be stored in a computer-readable storage medium, and when executed, the program can include the processes of the embodiments of the methods described above. The storage medium can be a magnetic disk, optical disk, read-only memory, or random access memory, etc.

[0198] Figure 10 The diagram illustrates the structure of an electronic device according to an embodiment of this application. This application does not limit the specific implementation of the electronic device. For example, it could be made of... Figure 1 The flight controller carried by the drone shown.

[0199] like Figure 10 As shown, the electronic device may include: a processor 1002, a communications interface 1004, a memory 1006, and a communications bus 1008.

[0200] The processor 1002, communication interface 1004, and memory 1006 communicate with each other via communication bus 808. Communication interface 1004 is used to communicate with other network elements such as clients or other servers. The processor 1002 executes program 1010, specifically performing the relevant steps in the embodiments of the altitude information correction method and radar data filtering method described above.

[0201] Specifically, program 1010 may include program code, which includes computer operation instructions. Specifically, it can be used to cause processor 1002 to execute the height information correction method in any of the above method embodiments.

[0202] In the embodiments of this application, depending on the type of hardware used, the processor 1002 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0203] The memory 1006 is used to store the program 1010. The memory 1006 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage, flash memory device or other non-volatile solid-state storage device.

[0204] It has a program storage area and a data storage area, which are used to store program 1010 and its corresponding data information, respectively. For example, non-volatile software programs, non-volatile computer executable programs and modules are stored in the program storage area, or the calculation results, radar data and flight data are stored in the data storage area.

[0205] This application also provides a computer-readable storage medium. This computer-readable storage medium can be a non-volatile computer-readable storage medium. This computer-readable storage medium stores a computer program.

[0206] When the computer program is executed by a processor, it implements one or more steps of the height information correction method disclosed in the embodiments of this application. The complete computer program product is embodied on one or more computer-readable storage media (including but not limited to, disk storage, CD-ROM, optical storage, etc.) containing the computer program disclosed in the embodiments of this application.

[0207] In summary, this application proposes a radar data filtering method based on Kalman filtering. This method uses a weighted superposition approach, combined with the UAV's flight attitude information, to eliminate some invalid radar data. The altitude information correction method established based on this filtering method can treat small, easily passable objects that cause significant changes in radar altitude measurement as invalid data, thus avoiding abrupt changes in altitude information. Furthermore, this invalid data exclusion method can also simultaneously eliminate external noise interference to the radar, eliminating abrupt changes in altitude information caused by false targets generated by noise.

[0208] Another aspect of this application proposes a height detection method that corrects radar detection points by incorporating UAV flight attitude information. This method can correct for accurate height information when the UAV tilts during flight and the radar antenna normal is not perpendicular to the ground. Furthermore, when the radar does not detect ground targets, it estimates the current height by combining aircraft attitude and historical height information.

[0209] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; under the concept of the present invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the present invention as described above, which are not provided in detail for the sake of brevity; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A radar data filtering method, characterized in that, include: Based on the vehicle's motion information, the theoretical state information of the target is determined; Acquire radar observation status information of the target; The radar is mounted on a vehicle; The theoretical state information and the observed state information are weighted and superimposed to filter and obtain the precise state information of the target. The weighted superposition of the theoretical state information and the observed state information to filter and obtain the precise state information of the target includes: Based on the theoretical state information and the observed state information, Kalman filtering is performed iteratively several times. The estimated state information of the target obtained after each Kalman filter process is output as the precise state information. The Kalman filtering process includes: Based on the state information estimate obtained after the previous Kalman filtering process, the prior estimate of the theoretical state information is calculated through a preset first transformation relationship. Based on the state vector estimation error covariance matrix obtained after the previous Kalman filtering process, the current state vector estimation error covariance matrix is ​​calculated. Based on the preset second transformation relationship, the residual between the prior estimate and the observed state information is calculated; Calculate the Kalman gain coefficient based on the current state vector estimation error covariance matrix, the prior estimate, and the observed state information; Based on the Kalman gain coefficients, the current state vector estimation error covariance matrix is ​​updated, and The residual and the prior estimate are weighted and summed using the Kalman gain coefficient as the weighting coefficient to obtain the state information estimate.

2. The method according to claim 1, characterized in that, The theoretical state information includes: the position information, velocity information, and acceleration information of the target in the three-dimensional coordinate system; The observation status information includes: distance information between the target and the millimeter-wave radar, azimuth angle information of the target relative to the millimeter-wave radar, elevation angle information of the target relative to the millimeter-wave radar, and radial velocity information of the target relative to the millimeter-wave radar. The first conversion relationship is established based on the acceleration model of the vehicle; the second conversion relationship is the correspondence between the theoretical state information and the observed state information after linearization.

3. A height detection method, characterized in that, include: The radar data filtering method described in claim 1 or 2 is used to obtain accurate state information of the target being detected. The tilt angle of the radar is determined based on the vehicle's attitude information; Determine whether a target exists within the search range; the search range is formed based on the tilt angle, with allowable upward and downward fluctuations. If so, within the search range, select the detection target that is closest to the tilt angle as the reference detection target; Based on the precise status information of the benchmark detection target, calculate the current altitude information of the vehicle; If not, estimate the current height of the vehicle based on the vehicle's motion information and the previously calculated height information.

4. The method according to claim 3, characterized in that, The radar is located at the bottom of the vehicle, close to the ground. Determining the tilt angle of the radar based on the vehicle's attitude information specifically includes: The elevation angle of the vehicle is obtained as the azimuth angle of the radar, and The roll angle of the vehicle is obtained as the pitch angle of the radar.

5. The method according to claim 4, characterized in that, The step of calculating the current altitude information of the vehicle based on the precise state information of the benchmark detection target specifically includes: The distance between the reference detection target and the radar, as well as the azimuth angle of the radar, are obtained; Using trigonometric functions, the current altitude information of the vehicle is calculated based on the distance between the reference target and the radar, and the azimuth angle.

6. A radar data filtering device, characterized in that, include: The theoretical data acquisition module is used to determine the theoretical state information of the target based on the vehicle's motion information; A radar data acquisition module is used to acquire the radar's observation status information of the target; the radar is mounted on a vehicle. The weighted superposition module is used to weight and superimpose the theoretical state information and the observed state information to filter and obtain the accurate state information of the detection target; Specifically, the weighted superposition module is used to: iteratively perform several Kalman filtering processes based on the theoretical state information and the observed state information; and output the estimated state information of the target obtained after each Kalman filtering process as the precise state information. The Kalman filtering process includes: Based on the state information estimate obtained after the previous Kalman filtering process, the prior estimate of the theoretical state information is calculated through a preset first transformation relationship. Based on the state vector estimation error covariance matrix obtained after the previous Kalman filtering process, the current state vector estimation error covariance matrix is ​​calculated. Based on the preset second transformation relationship, the residual between the prior estimate and the observed state information is calculated; Calculate the Kalman gain coefficient based on the current state vector estimation error covariance matrix, the prior estimate, and the observed state information; Based on the Kalman gain coefficients, the current state vector estimation error covariance matrix is ​​updated, and The residual and the prior estimate are weighted and summed using the Kalman gain coefficient as the weighting coefficient to obtain the state information estimate.

7. An electronic device, characterized in that, The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the radar data filtering method as described in claim 1 or 2 and / or the altitude detection method as described in any one of claims 3-5.

8. A drone, characterized in that, include: body; A millimeter-wave radar is installed at the bottom of the fuselage, which is close to the ground; The arm is connected to the machine body; A power unit, located on the arm, is used to provide the drone with the power to fly; as well as A flight controller, located on the fuselage and communicatively connected to the millimeter-wave radar; wherein the flight controller is configured to execute the radar data filtering method as described in claim 1 or 2 and / or the altitude detection method as described in any one of claims 3-5.

Images