Flight trajectory determination method, apparatus and controller

Abstract

The application discloses a flight trajectory determination method and device and a controller, and relates to the technical field of vehicles. The method comprises the following steps: performing data fusion on multi-source data corresponding to a previous moment of a current moment of an aircraft to obtain a globally optimal estimated pose state of the aircraft at the previous moment; determining a predicted pose state at the current moment based on the globally optimal estimated pose state and a control input parameter at the current moment, and determining predicted multi-source data at the current moment based on the predicted pose state; correcting the predicted pose state based on actual multi-source data and the predicted multi-source data at the current moment to obtain a current pose state at the current moment; and generating a target flight trajectory based on the current pose state and an expected pose state. In this way, the accuracy of flight trajectory prediction can be improved.

Description

Technical Field

[0001] This application relates to the field of aircraft technology, and in particular to a method, apparatus and controller for determining flight trajectory. Background Technology

[0002] Split-type flying cars consist of a flight module and a driving module. During the docking process between the flight module and the driving module, there is a possibility that the sensor signals (or sensor data) collected in real time by the sensors in the flight module may be interfered with or temporarily lost. This can lead to a lower accuracy in the attitude estimation of the aircraft, resulting in a lower accuracy in the determined flight trajectory of the flight module, and thus the possibility of misalignment between the flight module and the driving module.

[0003] Therefore, improving the accuracy of flight trajectory prediction by the flight module has become an urgent problem to be solved. Summary of the Invention

[0004] This application provides a method, apparatus, and controller for determining flight trajectories, which can improve the accuracy of flight trajectory prediction.

[0005] In a first aspect, embodiments of this application provide a method for determining a flight trajectory, the method comprising:

[0006] Data fusion is performed on the multi-source data corresponding to the previous time step of the current time step to obtain the global optimal estimated pose state of the aircraft at the previous time step.

[0007] Based on the globally optimal estimated pose state and the control input parameters at the current moment, the predicted pose state at the current moment is determined, and based on the predicted pose state, the predicted multi-source data at the current moment is determined.

[0008] Based on the actual multi-source data and predicted multi-source data at the current moment, the predicted pose state is corrected to obtain the current pose state at the current moment.

[0009] Generate the target flight trajectory based on the current pose state and the desired pose state.

[0010] In this embodiment, since the predicted pose state of the aircraft at the current moment is determined based on the globally optimal estimated pose state of the previous moment, and the globally optimal estimated pose state of the previous moment is determined based on the multi-source data corresponding to the aircraft at the previous moment, compared with the method of determining the real-time pose state of the aircraft based on real-time acquired sensor signals in related technologies, this method can avoid the problem of low accuracy in estimating the aircraft's pose state due to interference or temporary loss of sensor signals as much as possible. Then, based on the predicted pose state, the predicted multi-source data for the current moment is determined, and the predicted pose state is corrected based on the actual multi-source data and the predicted multi-source data for the current moment. This further improves the accuracy of the pose state at the current moment (i.e., the current pose state). Therefore, based on the highly accurate current pose state and the desired pose state, a highly accurate flight trajectory can be generated. It is evident that this method can improve the accuracy of flight trajectory prediction.

[0011] In one optional implementation of the first aspect, the predicted pose state is corrected based on the actual multi-source data and the predicted multi-source data at the current time to obtain the current pose state at the current time, including: determining a correction factor for correcting the predicted pose state; the correction factor is determined at least based on the error covariance matrix corresponding to the predicted pose state and the measurement noise covariance matrix at the current time; determining the difference between the actual multi-source data and the predicted multi-source data at the current time; and correcting the predicted pose state based on the correction factor and the difference to obtain the current pose state at the current time.

[0012] By employing this implementation method, the correction factor required to correct the predicted pose state can be determined more accurately when the sensor signal is interfered with or briefly lost, based on the covariance matrix of the predicted pose state correspondence error and the measurement noise covariance matrix at the current moment. Thus, the predicted pose state can be corrected based on the difference between the actual multi-source data and the predicted multi-source data at the current moment and the correction factor, and a high-precision current pose state can be estimated.

[0013] In an optional implementation of the first aspect, determining a correction factor for correcting the predicted pose state includes: determining an error covariance matrix corresponding to the predicted pose state based on the process noise covariance matrix at the current time, a first Jacobian matrix, and the previous error covariance matrix at the previous time; the error covariance matrix is ​​used to characterize the uncertainty of the predicted pose state; the first Jacobian matrix is ​​the Jacobian matrix of the prediction function of the error covariance matrix; and determining a correction factor for correcting the predicted pose state based on the error covariance matrix, a second Jacobian matrix, and the measurement noise covariance matrix at the current time; the second Jacobian matrix is ​​the Jacobian matrix of the prediction function of multi-source data.

[0014] Using this implementation method, the correction factor used to correct the predicted pose state can be determined simply and quickly.

[0015] In one optional implementation of the first aspect, data fusion is performed on the multi-source data corresponding to the previous time step of the aircraft at the current time step to obtain the globally optimal estimated pose state of the aircraft at the previous time step, including: determining the multi-source data corresponding to the previous time step of the aircraft at the current time step; determining the confidence levels of multiple acquisition devices associated with the multi-source data at the previous time step; determining the weights of each data in the multi-source data based on the confidence levels of each acquisition device at the previous time step, with the goal of minimizing the variance after data fusion; and performing data fusion on the multi-source data based on the weights of each data to obtain the globally optimal estimated pose state of the aircraft at the previous time step.

[0016] By adopting this implementation method, the weights of the multi-source data corresponding to the previous time are determined based on the confidence levels of multiple acquisition devices associated with the multi-source data corresponding to the aircraft at the previous time. In this way, the influence of unreliable high-precision data (a certain acquisition device itself has high precision, but if its confidence level is low due to the current environment (such as a small number of GPS satellites), unreliable high-precision data will be generated) on the fusion result (the global optimal estimated pose state at the previous time) can be avoided as much as possible, thereby improving the accuracy of the determined global optimal estimated pose state at the previous time.

[0017] In an optional implementation of the first aspect, the multi-source data includes at least two of the following: position data, millimeter-wave radar data, lidar data, and attitude data; determining the confidence levels of multiple acquisition devices associated with the multi-source data at the previous moment includes at least two of the following: determining the number of satellites, horizontal accuracy factor, and base station signal strength corresponding to the position sensor associated with the position data at the previous moment, and determining the confidence level of the position sensor at the previous moment based on the number of satellites, horizontal accuracy factor, and base station signal strength; the horizontal accuracy factor is used to represent the degree of influence of the spatial geometric distribution of satellites on the two-dimensional horizontal positioning accuracy; determining the confidence level of the millimeter-wave radar sensor associated with the millimeter-wave radar data at the previous moment. The signal-to-noise ratio (SNR) and target scattering point consistency index are used to determine the confidence level of the millimeter-wave radar at the previous time step. The target scattering point consistency index is used to measure the quality of the synthetic image corresponding to the millimeter-wave radar. The point cloud density, rainfall intensity, and relative motion ambiguity of the lidar sensor associated with the lidar data at the previous time step are determined, and the confidence level of the lidar sensor at the previous time step is determined based on these parameters. The vibration amplitude and temperature drift compensation residual of the attitude sensor associated with the attitude data at the previous time step are determined, and the confidence level of the attitude sensor at the previous time step is determined based on these residuals.

[0018] Using this implementation method, the confidence level of each acquisition device at the previous moment can be quickly determined based on the information related to each acquisition device when collecting multi-source data from the previous moment.

[0019] In one optional implementation of the first aspect, generating a target flight trajectory based on the current pose state and the desired pose state includes: constructing the parametric equation of a Bézier curve based on the current pose state and the desired pose state, and constructing an objective function for solving the target flight trajectory based on the parametric equation of the Bézier curve; determining the minimum value of the objective function as the objective, solving the objective function based on safe flight channel constraints, dynamic constraints, and obstacle constraints, and determining the target flight trajectory based on the solution result.

[0020] By adopting this implementation method, the trajectory generation problem can be transformed into a quadratic programming problem, which can quickly determine the target flight trajectory.

[0021] In one optional implementation of the first aspect, the safe flight path constraint is determined by: predicting the obstacle's flight trajectory when an obstacle is detected within a preset distance range of the aircraft, and predicting the covariance matrix corresponding to the obstacle based on the obstacle's flight trajectory; the covariance matrix is ​​used to characterize the uncertainty of the obstacle during flight; the obstacle's flight trajectory is subjected to spatiotemporal dilation processing based on the covariance matrix to obtain the processed obstacle flight trajectory; the region corresponding to the obstacle at each moment in the processed obstacle flight trajectory is projected from four-dimensional spatiotemporal to three-dimensional space, and the complement of the projected three-dimensional space is used as the safe flight path; and the safe flight path constraint is generated based on the safe flight path.

[0022] This implementation method provides a high-dimensional safe flight channel (or spatiotemporal corridor) for the generation of target flight trajectories by using predicted future trajectories of dynamic obstacles. This ensures that any trajectory planned within the safe flight channel is absolutely safe from both spatial and temporal perspectives, thereby improving the safety of subsequently generated target flight trajectories.

[0023] In an optional implementation of the first aspect, the method further includes: acquiring multi-source raw data from the previous moment collected during the flight of the aircraft; performing time alignment processing on the multi-source raw data based on a time synchronization protocol to obtain processed multi-source raw data; and converting all processed multi-source raw data to the body coordinate system of the aircraft's center of mass to obtain multi-source data.

[0024] This implementation method performs spatiotemporal alignment processing on the multi-source raw data from the previous time step to obtain multi-source data. This eliminates the heterogeneity between the multi-source raw data, thereby providing high-quality data for subsequent processing.

[0025] Secondly, embodiments of this application provide a flight trajectory determination device, the device comprising:

[0026] The data processing module is used to fuse the multi-source data corresponding to the previous time of the current time of the aircraft to obtain the global optimal estimated pose state of the aircraft at the previous time.

[0027] The determination module is used to determine the predicted pose state at the current moment based on the globally optimal estimated pose state and the control input parameters at the current moment, and to determine the predicted multi-source data at the current moment based on the predicted pose state.

[0028] The correction module is used to correct the predicted pose state based on the actual multi-source data and predicted multi-source data at the current moment, so as to obtain the current pose state at the current moment.

[0029] The trajectory generation module is used to generate the target flight trajectory based on the current pose state and the desired pose state.

[0030] Thirdly, embodiments of this application provide a controller, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method provided in the first aspect above.

[0031] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in the first aspect above.

[0032] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method provided in the first aspect above.

[0033] Regarding the beneficial effects of any of the technical solutions in the second to fifth aspects mentioned above, refer to the beneficial effects of the corresponding technical solutions in the first aspect; repeated examples will not be listed here. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is an optional flowchart illustrating a flight trajectory determination method provided in an embodiment of this application;

[0036] Figure 2 This is a schematic diagram of an optional process for determining real-time pose state provided in an embodiment of this application;

[0037] Figure 3 This is a schematic diagram illustrating an optional process for determining a target flight trajectory, provided in an embodiment of this application.

[0038] Figure 4 This is a schematic diagram of another optional process for a flight trajectory determination method provided in an embodiment of this application;

[0039] Figure 5 This is a schematic diagram of an optional structure of a flight trajectory determination device provided in an embodiment of this application;

[0040] Figure 6 This is a schematic diagram of an optional structure of a controller provided in an embodiment of this application. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0042] The flight trajectory determination method provided in the embodiments of this application is described below.

[0043] Please see Figure 1 , Figure 1 This is a schematic flowchart of an optional flight trajectory determination method provided in an embodiment of this application. The method can be executed by a controller. Figure 1 As shown, the method for determining the flight trajectory may include, but is not limited to, the following steps:

[0044] S101. Perform data fusion on the multi-source data corresponding to the previous time of the current time of the aircraft to obtain the global optimal estimated pose state of the aircraft at the previous time.

[0045] Optionally, multi-source data may include at least two of the following: position data, millimeter-wave radar data, lidar data, and attitude data.

[0046] In this application, the aircraft is equipped with a Global Positioning System (GPS), a millimeter-wave radar array, a lidar, and an attitude acquisition device. Optionally, the GPS may be a dual-frequency Real-Time Kinematic Global Positioning System (RTK-GPS); the attitude acquisition device may be an Inertial Measurement Unit (IMU) or an airspeed indicator. The GPS is used to acquire position data; the millimeter-wave radar array is used to acquire millimeter-wave radar data; the lidar is used to acquire lidar data; and the attitude acquisition device is used to acquire attitude data.

[0047] Among them, the global optimal estimated pose state of the aircraft at the previous moment refers to the optimal pose estimate relative to the fixed world coordinate system determined based on all observation information at the previous sampling moment.

[0048] S102. Based on the globally optimal estimated pose state and the control input parameters at the current moment, determine the predicted pose state at the current moment, and based on the predicted pose state, determine the predicted multi-source data at the current moment.

[0049] In some embodiments, the control input parameters at the current moment may include, but are not limited to, the motor speed of the motors in the aircraft and the control surface deflection angle of the aircraft.

[0050] In some embodiments, the controller determines the predicted pose state at the current moment based on the globally optimal estimated pose state and the control input parameters at the current moment. This can be achieved by using the nonlinear dynamics model of the aircraft to predict the predicted pose state at the current moment based on the globally optimal estimated pose state and the control input parameters at the current moment.

[0051] Optionally, when the controller predicts the current pose state based on the global optimal estimated pose state and the control input parameters at the current moment using the nonlinear dynamic model of the aircraft, it can use the following formula (1).

[0052] (1)

[0053] In formula (1), X k This represents the predicted pose state at the current moment; X k-1 This represents the globally optimal estimated pose state from the previous time step; u k This represents the control input parameters at the current moment; w k denoted by , where the process noise follows a Gaussian distribution; f() represents the nonlinear dynamic model of the aircraft.

[0054] In some embodiments, the pose states mentioned in this application, such as the global optimal estimated pose state at the previous moment and the predicted pose state at the current moment, can all be represented as a 13-dimensional vector as shown in the following formula (2).

[0055] (2)

[0056] In formula (2), X represents the pose state; (p_x, p_y, p_z) represents the position of the aircraft, which is the coordinate of the aircraft in the world coordinate system; (v_x, v_y, v_z) represents the velocity of the aircraft, which is the velocity of the aircraft in the body coordinate system; (q_w, q_x, q_y, q_z) represents the attitude quaternion of the aircraft, which represents the rotational attitude of the aircraft; (ω_x, ω_y, ω_z) represents the body angular velocity of the aircraft, which represents the rotational speed of the aircraft around the xyz axis of the body.

[0057] In some embodiments, the controller determines the predicted multi-source data at the current moment based on the predicted pose state, which may include: inputting the predicted pose state into a pre-determined multi-source data prediction model to obtain the predicted multi-source data at the current moment.

[0058] S103. Based on the actual multi-source data and predicted multi-source data at the current moment, the predicted pose state is corrected to obtain the current pose state at the current moment.

[0059] In some embodiments, the actual multi-source data at the current moment can be the actual data acquired by the controller through a data acquisition device. This data acquisition device may include, but is not limited to, a GPS system, millimeter-wave radar array, lidar, pose acquisition device, etc., deployed within the aircraft. Optionally, the actual multi-source data at the current moment can be stored locally on the controller or in a database accessible to the controller (such as a data fusion center), etc., without limitation.

[0060] In some embodiments, the controller corrects the predicted pose state based on the actual multi-source data and the predicted multi-source data at the current moment to obtain the current pose state at the current moment, which may include: determining a correction factor for correcting the predicted pose state; and determining the current pose state at the current moment based on the correction factor, the actual multi-source data and the predicted multi-source data at the current moment.

[0061] S104. Generate the target flight trajectory based on the current pose state and the desired pose state.

[0062] The desired attitude state refers to the target spatial state that the aircraft should reach at a certain point in the future.

[0063] In some embodiments, the aircraft in this application may be a flight module of a split-type flying car. In this embodiment, the desired pose state may be the pose state of the flight module required when the flight module and the driving module of the split-type flying car dock.

[0064] In some embodiments, the controller generates a target flight trajectory based on the current pose state and the desired pose state. This may include inputting the current pose state and the desired pose state into a trajectory generator, so that the trajectory generator plans the target flight trajectory of the aircraft based on a Bézier curve. A Bézier curve is a smooth parametric curve that defines a smooth path using a few points called control points.

[0065] In this embodiment, since the predicted pose state of the aircraft at the current moment is determined based on the globally optimal estimated pose state of the previous moment, and the globally optimal estimated pose state of the previous moment is determined based on the multi-source data corresponding to the aircraft at the previous moment, compared with the method of determining the real-time pose state of the aircraft based on real-time acquired sensor signals in related technologies, this method can avoid the problem of low accuracy in estimating the aircraft's pose state due to interference or temporary loss of sensor signals as much as possible. Then, based on the predicted pose state, the predicted multi-source data for the current moment is determined, and the predicted pose state is corrected based on the actual multi-source data and the predicted multi-source data for the current moment. This further improves the accuracy of the pose state at the current moment (i.e., the current pose state). Therefore, based on the highly accurate current pose state and the desired pose state, a highly accurate flight trajectory can be generated. It is evident that this method can improve the accuracy of flight trajectory prediction.

[0066] In one alternative implementation, Figure 1 Step S103 in the flight trajectory determination method shown, which is the method by which the controller corrects the predicted pose state based on the actual multi-source data and the predicted multi-source data at the current moment to obtain the current pose state, can be: determining a correction factor for correcting the predicted pose state; the correction factor is determined at least based on the error covariance matrix corresponding to the predicted pose state and the measurement noise covariance matrix at the current moment; determining the difference between the actual multi-source data and the predicted multi-source data at the current moment; and correcting the predicted pose state based on the correction factor and the difference to obtain the current pose state at the current moment.

[0067] The error covariance matrix is ​​used to characterize the degree of uncertainty in the predicted pose state.

[0068] The difference between the actual multi-source data and the predicted multi-source data at the current moment can be called innovation. Innovation is the deviation between theory and reality, and it contains all the new information used to optimize the next estimate.

[0069] In some embodiments, the controller determines a correction factor for correcting the predicted pose state, which may include: determining an error covariance matrix corresponding to the predicted pose state based on the process noise covariance matrix at the current time, a first Jacobian matrix, and the previous error covariance matrix at the previous time; the error covariance matrix is ​​used to characterize the uncertainty of the predicted pose state; the first Jacobian matrix is ​​the Jacobian matrix of the prediction function of the error covariance matrix; and the correction factor for correcting the predicted pose state is determined based on the error covariance matrix, a second Jacobian matrix, and the measurement noise covariance matrix at the current time; the second Jacobian matrix is ​​the Jacobian matrix of the prediction function of multi-source data. This allows for a simple and quick determination of the correction factor for correcting the predicted pose state.

[0070] The process noise covariance matrix can be used to describe the statistical characteristics of the total error between the actual motion state (actual pose state) and the predicted pose state of the aircraft.

[0071] The Jacobian matrix is ​​an important concept in calculus, used to describe the best linear approximation of a vector-valued function at a point in its domain. It is a matrix of first-order partial derivatives arranged in a specific way.

[0072] The measurement noise covariance matrix is ​​used to describe the statistical characteristics of the error between sensor measurements (i.e., actual multi-source data) and predicted multi-source data.

[0073] Optionally, when the controller determines the error covariance matrix corresponding to the predicted pose state based on the process noise covariance matrix at the current time, the first Jacobian matrix, and the previous error covariance matrix at the previous time, the following formula (3) can be used.

[0074] (3)

[0075] In formula (3), P k F represents the error covariance matrix corresponding to the predicted pose state at the current moment; k This represents the first Jacobian matrix (the Jacobian matrix (linearized) of the aforementioned nonlinear dynamics model f() of the aircraft); P k-1 Q represents the previous time step's error covariance matrix; k This represents the process noise covariance matrix at the current moment.

[0076] Optionally, the controller determines a correction factor for correcting the predicted pose state based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current time. This may include: determining the Kalman gain based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current time; and using the Kalman gain as the correction factor for correcting the predicted pose state. Optionally, the controller may determine the Kalman gain using the following formula (4).

[0077] (4)

[0078] In formula (4), K k This represents the Kalman gain; P k H represents the error covariance matrix corresponding to the predicted pose state at the current moment; k This represents the second Jacobian matrix, which is the Jacobian matrix of the multi-source data prediction function; R k This represents the measurement noise covariance matrix at the current moment.

[0079] In some embodiments, when the controller corrects the predicted pose state based on the correction factor and the difference, the following formula (5) can be used to obtain the current pose state at the current time.

[0080] (5)

[0081] In formula (5), This represents the current pose state at the current moment; X k This represents the predicted pose state at the current moment, which can be determined by the aforementioned formula (1); K k Z represents the Kalman gain (i.e., the correction factor), which can be determined by the aforementioned formula (4); k This represents the actual multi-source data at the current moment; h(X) k ) represents the predicted multi-source data at the current moment, where h() represents the multivariate data prediction function; This represents the difference between the actual multi-source data and the predicted multi-source data at the current moment, which is the aforementioned information.

[0082] In some embodiments, the controller may also update the error covariance matrix. Optionally, the controller may update the error covariance matrix using the following formula (6).

[0083] (6)

[0084] In formula (6), K represents the updated error covariance matrix; I represents the identity matrix; K represents the identity matrix. k This represents the Kalman gain; Hk This represents the second Jacobian matrix, which is the Jacobian matrix of the multi-source data prediction function; P k It represents the error covariance matrix corresponding to the predicted pose state at the current moment.

[0085] In some embodiments, the controller can also adaptively adjust the parameters of the Kalman filter based on the innovation. In this way, by monitoring the statistical characteristics of the innovation sequence in real time, autonomously identifying model mismatch or sensor anomalies, and dynamically adjusting the parameters of the Kalman filter, the robustness of determining the pose state at each moment in complex environments can be improved.

[0086] Optionally, the process by which the controller adaptively adjusts the parameters of the Kalman filter based on the innovation may include, but is not limited to, the following steps:

[0087] Step 1: At each time k, determine the corresponding new information v. k And the theoretical covariance matrix S k .

[0088] The controller determines the new information v corresponding to each time k. k When the time is right, the following formula (7) can be used.

[0089] (7)

[0090] In formula (7), Z k This represents the actual multi-source data at time k; h(X) k|k-1 ) represents the predicted multi-source data at time k, which is obtained by the controller predicting the predicted pose state at time k based on the multi-source data at time k-1, and then inputting the predicted pose state at time k into the multi-source data prediction function h().

[0091] When the controller determines the theoretical covariance matrix corresponding to each time k, the following formula (8) can be used.

[0092] (8)

[0093] In formula (8), H k P represents the Jacobian matrix of the multi-source data prediction function; k|k-1 R represents the error covariance matrix corresponding to the predicted pose state at the current time, determined based on the error covariance matrix at time k-1; k-1 This represents the measurement noise covariance matrix at time k-1.

[0094] Step 2: Estimate the actual covariance matrix C of the innovation using a sliding window of length N. vk .

[0095] When the controller estimates the actual covariance matrix of the innovation through a sliding window of length N, the following formula (9) can be used.

[0096] (9)

[0097] In formula (9), v j This represents the information at time j.

[0098] Step 3: Determine the adaptive factor.

[0099] In some embodiments, the adaptive factor may include μ k = trace(C vk ) / trace(S k This is used to quantify the degree of deviation between actual information and theoretical expectations.

[0100] Step 4: Adjust the noise covariance matrix Q k and R k .

[0101] In some embodiments, the controller may adjust the noise covariance by: increasing the process noise covariance matrix when the adaptive factor is greater than a first preset threshold; and decreasing the process noise covariance matrix when the adaptive factor is less than a second preset threshold.

[0102] The first preset threshold can include 1+ε; the second preset threshold can include 1-ε. Here, ε is the tolerance band for error fluctuations, representing the range within which the filter considers the model to be well-matched with the measurement. Only when the adaptive factor is greater than 1+ε can a significant model mismatch be determined.

[0103] If the adaptive factor is greater than the first preset threshold, it indicates that the model prediction error is underestimated or the measurement noise is abnormally increased. In this case, it can be determined according to Q. k =Q k-1 *(1+γ(μ k-1 The process noise covariance matrix is ​​increased in a manner that allows for proportional and smooth fine-tuning based on the current model's fit. If the adaptive factor is less than the second preset threshold, it indicates that the model prediction is too conservative. In this case, it can be adjusted according to Q... k =Q k-1 / (1+γ(1-μ k This reduces the process noise covariance matrix using a method that...

[0104] In some embodiments, the controller performs noise covariance adjustment, which may further include: based on the dynamic confidence ω of each sensor. i (t) Adjust the corresponding measurement noise covariance matrix.

[0105] Step 5: Adjust the noise covariance matrix Q k and R k Perform constraint processing to ensure that the adjusted Q k and R k Maintain a positive equilibrium and do not exceed the preset physical range.

[0106] Step 6: Based on the adjusted Q k and R k Perform a Kalman update to obtain the optimal X. k and P k .

[0107] By employing this implementation method, the correction factor required to correct the predicted pose state can be determined more accurately when the sensor signal is interfered with or briefly lost, based on the covariance matrix of the predicted pose state correspondence error and the measurement noise covariance matrix at the current moment. Thus, the predicted pose state can be corrected based on the difference between the actual multi-source data and the predicted multi-source data at the current moment and the correction factor, and a high-precision current pose state can be estimated.

[0108] Please see Figure 2 , Figure 2 This is a schematic diagram illustrating an optional process for determining real-time pose state provided in an embodiment of this application. For example... Figure 2 As shown, the controller can first predict the pose state X at time k based on the nonlinear dynamic model of the aircraft. k The error covariance matrix P corresponding to the predicted pose state at time k k Then wait for the sensor data (the actual multi-source data at time k) to arrive. Determine if the sensor data (the actual multi-source data at time k) is valid; if valid, determine the Kalman gain K. k And based on Kalman gain K k And the actual multi-source data Z at time k k For the predicted pose state X k The optimal pose state at time k is obtained through correction (corresponding to the aforementioned current pose state), and the error covariance matrix is ​​updated. The optimal pose state at time t and the updated error covariance matrix are output, and adaptive noise estimation is performed to adjust the noise covariance matrix Q. k and R k Based on the adjusted Q k and R k The prediction step is then re-executed (i.e., determining the next predicted pose state and the corresponding next error covariance matrix at the next moment). If this is ineffective, the update is skipped, and the nonlinear dynamics model based on the aircraft is re-executed to predict the current predicted pose state X. kThe error covariance matrix P corresponding to the predicted pose state k The steps.

[0109] In this application, an adaptive extended Kalman filter algorithm is used, which can not only perform simple fusion of sensor data, but also perform dynamic model prediction and sensor measurement update, and estimate and adjust the system noise statistical characteristics in real time. Thus, even when the sensor signal is disturbed or briefly lost, it can still perform high-precision pose state estimation.

[0110] In one alternative implementation, Figure 1 Step S101 in the flight trajectory determination method shown, which is the method by which the controller fuses the multi-source data corresponding to the previous time of the aircraft at the current time to obtain the global optimal estimated pose state of the aircraft at the previous time, can be as follows: determine the multi-source data corresponding to the previous time of the aircraft at the current time; determine the confidence levels of the multiple acquisition devices associated with the multi-source data at the previous time; determine the weights of each data in the multi-source data based on the confidence levels of each acquisition device at the previous time, with the goal of minimizing the variance after data fusion; and fuse the multi-source data based on the weights of each data to obtain the global optimal estimated pose state of the aircraft at the previous time.

[0111] In some embodiments, multi-source data of the aircraft during flight may be stored in a data fusion center; in this case, the controller may determine the multi-source data corresponding to the previous time of the aircraft at the current time by obtaining the multi-source data corresponding to the previous time of the aircraft at the current time from the data fusion center.

[0112] In some embodiments, the multi-source data may include at least two of the following: location data, millimeter-wave radar data, lidar data, and attitude data; the controller determines the confidence levels of multiple acquisition devices associated with the multi-source data at the previous time step, including at least two of the following: determining the number of satellites, horizontal accuracy factor, and base station signal strength corresponding to the location sensor associated with the location data at the previous time step, and determining the confidence level of the location sensor at the previous time step based on the number of satellites, horizontal accuracy factor, and base station signal strength; the horizontal accuracy factor is used to represent the degree of influence of the spatial geometric distribution of satellites on the two-dimensional horizontal positioning accuracy; determining the signal-to-noise ratio corresponding to the millimeter-wave radar sensor associated with the millimeter-wave radar data at the previous time step. The confidence level of the millimeter-wave radar at the previous time step is determined based on the consistency index of the target scattering point and the signal-to-noise ratio (SNR). The target scattering point consistency index measures the quality of the synthetic image corresponding to the millimeter-wave radar. The point cloud density, rainfall intensity, and relative motion ambiguity of the lidar sensor associated with the lidar data at the previous time step are determined, and the confidence level of the lidar sensor at the previous time step is determined based on these parameters. Similarly, the vibration amplitude and temperature drift compensation residual of the attitude sensor associated with the attitude data at the previous time step are determined, and the confidence level of the attitude sensor at the previous time step is determined based on these residuals. This allows for the rapid determination of the confidence level of each acquisition device at the previous time step based on the information related to each acquisition device when collecting multi-source data from the previous time step.

[0113] The location sensor may include dual-frequency RTK-GPS.

[0114] In this embodiment, the controller determines the confidence level of the position sensor at the previous moment based on the number of satellites, the horizontal accuracy factor, and the base station signal strength. This can include: inputting the number of satellites into a satellite number confidence factor calculation function (denoted as function f1) to obtain a first confidence factor corresponding to the position sensor; inputting the horizontal accuracy factor into a horizontal accuracy factor confidence factor calculation function (denoted as function f2) to obtain a second confidence factor corresponding to the position sensor; inputting the base station signal strength into a base station signal strength confidence factor calculation function (denoted as function f3) to obtain a third confidence factor corresponding to the position sensor; and determining the confidence level of the position sensor at the previous moment based on the first confidence factor, the second confidence factor, and the third confidence factor corresponding to the position sensor.

[0115] Among them, functions f1, f2, and f3 are all empirical functions.

[0116] The Horizontal Dilution of Precision (HDOP) describes the degree to which the spatial geometric distribution of satellites affects the accuracy of two-dimensional horizontal positioning (longitude and latitude). The smaller the HDOP value, the wider the satellite geometric distribution, the smaller the potential positioning error, and the higher the horizontal positioning accuracy.

[0117] Optionally, function f1 may include a piecewise function that increases with the number of satellites N. For example, when N is less than a first preset satellite number threshold, the value of function f1 may be 0; when N is greater than or equal to the first preset satellite number threshold and less than or equal to a second preset satellite number threshold, the value of function f1 may linearly increase from 0.2 to 0.6; when N is greater than the second preset satellite number threshold and less than or equal to a third preset satellite number threshold, the value of function f1 may linearly increase from 0.6 to 1; and when N is greater than the third preset satellite number threshold, the value of function f1 may be 1. The first, second, and third preset satellite number thresholds may be determined based on expert experience, multiple experiments, or manually defined, etc., and are not limited here. Optionally, the value of the first preset satellite number threshold may be any one of 4, 5, and 6; the value of the second preset satellite number threshold may be any one of 7, 8, and 9; and the value of the third preset satellite number threshold may be any one of 11, 12, and 13. For example, the first preset satellite number threshold can be 5, the second preset satellite number threshold can be 7, and the third preset satellite number threshold can be 12.

[0118] Optionally, the function f2 may include an exponential function that decreases as the horizontal precision factor increases, and its expression may be f2(HDOP)=e -(HDOP-0.8) When HDOP is greater than a preset threshold (e.g., 2.5), it indicates poor satellite distribution (e.g., all satellites are clustered on one side of the sky). In this case, f2(HDOP) will output a very low coefficient (approaching 0). At this time, the controller can automatically reduce the confidence and weight of the data corresponding to the position sensor (e.g., GPS data) to minimize the impact of the inherently inaccurate data on the final data fusion result.

[0119] For example, the reference standard for HDOP values ​​can be shown in Table 1 below.

[0120] Table 1

[0121]

[0122] Optionally, function f3 may include a mapping function based on an sigmoid growth function, the center of which corresponds to a signal strength of -88dBm; when the base station signal strength is higher than -88dBm, f3 (base station signal strength) may approach 1; when the base station signal strength is lower than a preset strength threshold (e.g., -100dBm), f3 (base station signal strength) may approach 0, indicating that the differential link is unavailable.

[0123] Optionally, when the controller determines the confidence level of the position sensor at the previous moment based on the first confidence factor, the second confidence factor, and the third confidence factor corresponding to the position sensor, the following formula (10) can be used.

[0124] (10)

[0125] In formula (10), This represents the confidence level of the position sensor at time t.

[0126] In this embodiment, the controller determines the confidence level of the millimeter-wave radar at the previous moment based on the signal-to-noise ratio (SNR) and the target scattering point consistency index. This can include: inputting the SNR into the SNR confidence factor calculation function (denoted as function g1) to obtain the first confidence factor corresponding to the millimeter-wave radar; inputting the target scattering point consistency index into the target scattering point consistency index confidence factor calculation function (denoted as function g2) to obtain the second confidence factor corresponding to the millimeter-wave radar; and determining the confidence level of the millimeter-wave radar at the previous moment based on the first and second confidence factors corresponding to the millimeter-wave radar.

[0127] Optionally, function g1 may include an sigmoid function, the center of which corresponds to a signal-to-noise ratio (SNR) of 12 dB. When the SNR is higher than a first preset SNR threshold (e.g., 20 dB), the value of function g1 can approach 1, indicating that the detection result is highly reliable; when the SNR is lower than a second SNR threshold (e.g., 5 dB), the value of function g1 can approach 0, indicating that the echo signal is overwhelmed by noise and the confidence level of the millimeter-wave radar is extremely low.

[0128] Optionally, function g2 may include an exponential decay function based on the standard deviation C of observations of the same target from multiple receiving channels of the millimeter-wave radar. In some embodiments, the expression for function g2 may be g2(C) = e -(C² / (2*L²))The values ​​of g2 are as follows: C represents the standard deviation of observations of the same target from multiple receiving channels of the millimeter-wave radar; L represents the preset tolerance threshold based on the type of observation data (e.g., range tolerance L_range = 0.5m, angle tolerance L_angle = 1.5°). The smaller the standard deviation C, the closer g2 is to 1, indicating a higher degree of target authenticity. When the standard deviation C exceeds L, the value of g2 decreases rapidly, effectively filtering out false targets caused by multipath reflection or interference.

[0129] Optionally, when the controller determines the confidence level of the millimeter-wave radar at the previous moment based on the first confidence factor and the second confidence factor corresponding to the millimeter-wave radar, the following formula (11) can be used.

[0130] (11)

[0131] In formula (11), This represents the confidence level of the millimeter-wave radar at time t.

[0132] In this embodiment, the controller determines the confidence level of the lidar sensor at the previous moment based on point cloud density, rainfall intensity, and relative motion ambiguity. This can be achieved by: inputting the point cloud density into a point cloud density confidence factor calculation function (denoted as function h1) to obtain the first confidence level of the lidar sensor; inputting the rainfall intensity into a rainfall intensity confidence factor calculation function (denoted as function h2) to obtain the second confidence level of the lidar sensor; inputting the relative motion ambiguity into a relative motion ambiguity calculation function (denoted as function h3) to obtain the third confidence level of the lidar sensor; and determining the confidence level of the lidar sensor at the previous moment based on the first, second, and third confidence levels of the lidar sensor.

[0133] Among them, rainfall intensity is the main attenuation factor, which can be estimated by rain gauge data from automatic wipers or by camera image recognition.

[0134] Optionally, the function h1 may include a saturation function that approaches 1 as the point cloud density ρ increases, and its expression may be... ,in This represents a preset reference density based on the performance of the lidar and typical docking distance. When the point cloud density ρ is lower than the first preset density threshold (e.g., 10 points / m²), h1(ρ) can approach 0, indicating that the data is unavailable; when ρ reaches the second preset density threshold (e.g., 150 points / m²), h1(ρ) exceeds 0.95, indicating that the density is sufficient.

[0135] Optionally, the function h1 can include a function that decreases linearly with increasing rainfall intensity R, and its expression can be h2(R) = max(0, 1 - 0.05 * R). When the rainfall intensity R is less than a first preset rainfall intensity threshold (e.g., 10 mm / h (light rain)), h2(R) can be higher than 0.5; when R reaches a second preset rainfall intensity threshold (e.g., 20 mm / h (moderate rain)), h2(R) can drop to 0, indicating that the lidar performance is severely degraded and its data is unusable in this scenario.

[0136] Optionally, function h3 can be used to evaluate the degree of point cloud distortion caused by relative high-speed motion, and its expression can be... In this equation, M = V * θ represents the ambiguity metric, V represents the relative lateral velocity, θ represents the radar angular resolution, and M0 represents the preset tolerance threshold (e.g., 0.1). The relative lateral velocity V can be directly and accurately measured by millimeter-wave radar. The value of function h3 decreases rapidly as the relative lateral velocity increases, thus effectively quantifying the negative impact of motion ambiguity on the reliability of point cloud matching.

[0137] Optionally, when the controller determines the confidence level of the lidar sensor at the previous moment based on the first confidence level, the second confidence level, and the third confidence level corresponding to the lidar sensor, the following formula (12) can be used.

[0138] (12)

[0139] In formula (12), This represents the confidence level of the lidar sensor at time t.

[0140] In this embodiment, the controller determines the confidence level of the attitude sensor at the previous moment based on the vibration amplitude and the temperature drift compensation residual. This can include: inputting the vibration amplitude into the vibration amplitude confidence factor calculation function (denoted as function k1) to obtain the first confidence level corresponding to the attitude sensor; inputting the temperature drift compensation residual into the temperature drift compensation residual confidence factor calculation function (function k2) to obtain the second confidence level corresponding to the attitude sensor; and determining the confidence level of the attitude sensor at the previous moment based on the first and second confidence levels corresponding to the attitude sensor.

[0141] Among them, the confidence level of the attitude sensor is lower when the vibration is too large (such as due to strong wind) or the temperature changes drastically.

[0142] Optional, function It can be used to evaluate the impact of high-frequency vibration on the data quality of attitude sensors (e.g., time-based IMUs), and its expression can be: Here, A represents the vibration amplitude, which can be obtained by calculating the standard deviation of the high-frequency components (>50Hz) of the accelerometer. When A is less than a first preset threshold (e.g., 0.5m / s²), A value approaching 1 indicates that the vibration is within the allowable range; when A is greater than the second preset threshold (e.g., 1.5 m / s²), A value close to 0 indicates that the attitude sensor is under severe vibration, and the reliability of its output angular velocity and acceleration data is severely reduced.

[0143] Optional, function It can be used to evaluate the effectiveness of attitude sensor temperature drift compensation residuals, and its expression can be: Here, ΔB represents the temperature drift compensation residual, which can be calculated by monitoring the standard deviation of the gyroscope output during system quiet periods (such as before takeoff or after docking). When ΔB is less than a first preset residual threshold (e.g., 0.1° / s), A value approaching 1 indicates good temperature compensation; when ΔB is greater than the second preset residual threshold (e.g., 0.5° / s), If the value approaches 0, it indicates that the temperature drift is too large, and the output angular velocity data has a significant and unpredictable error with extremely low confidence.

[0144] In some embodiments, the expression for the variance after data fusion can be shown in the following formula (13).

[0145] (13)

[0146] In formula (13), W represents the variance after data fusion. i This represents the fusion weight of the i-th acquisition device, which can be determined by the following formula (14); This represents the measurement variance (or measurement noise variance) of the i-th acquisition device.

[0147] (14)

[0148] In formula (14), This represents the dynamic confidence level of the i-th data acquisition device. .

[0149] In some embodiments, the controller can determine the globally optimal estimated pose state by the following formula (15).

[0150] (15)

[0151] In formula (15), X fW represents the globally optimal estimated pose state. i Xi represents the fusion weight of the i-th acquisition device, which can be determined by the aforementioned formula (14); Xi represents the data acquired by the i-th acquisition device.

[0152] By adopting this implementation method, the weights of the multi-source data corresponding to the previous time are determined based on the confidence levels of multiple acquisition devices associated with the multi-source data corresponding to the aircraft at the previous time. In this way, the influence of unreliable high-precision data (a certain acquisition device itself has high precision, but if its confidence level is low due to the current environment (such as a small number of GPS satellites), unreliable high-precision data will be generated) on the fusion result (the global optimal estimated pose state at the previous time) can be avoided as much as possible, thereby improving the accuracy of the determined global optimal estimated pose state at the previous time.

[0153] In one alternative implementation, Figure 1 Step S104 in the flight trajectory determination method shown, that is, the way the controller generates the target flight trajectory based on the current pose state and the desired pose state, can be as follows: based on the current pose state and the desired pose state, construct the parametric equation of the Bézier curve, and based on the parametric equation of the Bézier curve, construct the objective function for solving the target flight trajectory; determine the minimum value of the objective function as the objective, solve the objective function based on the safe flight channel constraints, dynamic constraints and obstacle constraints, and determine the target flight trajectory based on the solution result.

[0154] In some embodiments, the parameter equation of the Bézier curve constructed by the controller based on the current pose state and the desired pose state can be shown in the following formula (16).

[0155] (16)

[0156] In formula (16), B(t) represents the constructed Bézier curve; P i This represents the coordinates of the i-th position of the aircraft during flight, where P0 represents the starting constraint, which is the current coordinate in the current pose state of the aircraft, P0 = X. current.position ;P n The endpoint constraint is represented by the coordinates in the desired pose state, P. n =X tager.position The tangent direction of the curve at the starting point can be determined based on P1-P0 and can be constrained to the direction of the current velocity. The tangent direction of the curve at the ending point can be determined based on P... n -P n-1(Assuming the aircraft is a flight module within a flying car, and the desired pose state is the one required for docking with the driving module within the flying car, the tangent direction of the endpoint can be constrained to vertically downwards (for easier vertical docking), i.e., P) n-1 =P n +[0,0,k] T Using formula (16), the trajectory generation problem can be transformed into a constrained quadratic programming (QP) problem, where the optimization variables are the intermediate control points P1, P2, ..., P of the Bézier curve. n-1 .

[0157] In some embodiments, the objective function for solving the target flight trajectory can be shown in the following formula (17).

[0158] (17)

[0159] In formula (17), J represents the objective function used to solve the target flight trajectory; B(t) represents the constructed Bézier curve.

[0160] As can be seen from formula (17), determining the minimum value of the objective function is equivalent to determining the energy of the entire trajectory, which is the square integral of the acceleration. This ensures smoothness and optimal energy.

[0161] Among them, safe flight path constraints are used to ensure that all control points P i All are located within a safe flight corridor (or spacetime corridor), which is a time-varying, unobstructed region. In other words, the safe flight corridor constraint is used to determine P. i ∈Corridor(t i ) for all i.

[0162] Among them, dynamic constraints are used to ensure that the aircraft's flight speed is less than the speed limit and the acceleration is less than the acceleration limit.

[0163] Among them, obstacle constraints are used to ensure that all control points on the Bézier curve maintain a minimum distance d from the dynamic obstacle at any time t. min . d min It is a dynamic, risk-based buffer distance, which can also be called a safety distance.

[0164] In some embodiments, dynamic constraints and obstacle constraints may be determined by the controller by: obtaining the maximum permissible speed and maximum permissible acceleration corresponding to the aircraft; taking the speed of the aircraft at any time during flight as less than the maximum permissible speed and the acceleration at any time as less than the maximum permissible acceleration as dynamic constraints; obtaining a predetermined safe distance and taking the distance between the position of the aircraft and the predicted position of the obstacle at the same time as greater than the safe distance as obstacle constraints.

[0165] Optionally, the dynamic constraints can be expressed as: ,and Among them, V max This indicates the maximum permissible speed; A max This indicates the maximum permissible acceleration.

[0166] Optionally, the safe distance can be determined by the controller based on the physical radius of the obstacle, the perception error buffer, and the dynamic response buffer δ, wherein the perception error buffer is obtained based on the state estimation covariance, and the dynamic response buffer is obtained based on the relative velocity. When the controller determines the safe distance based on the physical envelope radius, the perception error buffer, and the dynamic response buffer, the following formula (18) can be used.

[0167] (18)

[0168] In formula (18), d min This indicates a safe distance; R physical V represents the physical radius of the obstacle; K represents the safety factor; σ represents the standard deviation of the positioning error; V rel T represents the relative velocity vector between the aircraft and the obstacle; delay This indicates the calibrated total response time.

[0169] Optionally, the obstacle constraint can be expressed as: distance(B(t),Obstacle(t))>d min Here, Obstacle(t) represents the predicted trajectory of the obstacle.

[0170] It should be noted that the constraint parameter V mentioned above... max A max , and d min It is not a fixed value, but an adaptive parameter that is dynamically calculated based on the platform's physical limits, real-time environmental conditions, and its own state confidence.

[0171] In some embodiments, the safe flight path constraint can be determined by the controller in the following manner: when an obstacle is detected within a preset distance range of the aircraft, the obstacle's flight trajectory is predicted, and based on the obstacle's flight trajectory, the corresponding covariance matrix is ​​predicted; the covariance matrix is ​​used to characterize the uncertainty of the obstacle during flight; the obstacle's flight trajectory is subjected to spatiotemporal dilation based on the covariance matrix to obtain the processed obstacle flight trajectory; the region corresponding to the obstacle at each moment in the processed obstacle flight trajectory is projected from four-dimensional spatiotemporal space to three-dimensional space, and the complement of the projected three-dimensional space is used as the safe flight path; based on the safe flight path, a safe flight path constraint is generated. In this way, by providing a high-dimensional safe flight path (or spatiotemporal corridor) for the generation of the target flight trajectory based on the predicted future trajectory of the dynamic obstacle, any trajectory planned within the safe flight path is absolutely safe from both spatial and temporal perspectives, thereby improving the safety of the subsequently generated target flight trajectory.

[0172] In this embodiment, the controller performs spatiotemporal dilation processing on the obstacle flight trajectory based on the covariance matrix to obtain the processed obstacle flight trajectory. This can include: determining the dilation distance of the obstacle based on the covariance matrix and the physical radius of the obstacle; determining the region corresponding to the obstacle at each time step based on the dilation distance; the distance between each point in the region and the position of the obstacle at the current time step is less than or equal to the dilation distance; and obtaining the processed obstacle flight trajectory based on the region corresponding to the obstacle at each time step.

[0173] The area corresponding to the obstacle at each moment refers to the set of spaces occupied by the obstacle and its safety buffer zone at each moment.

[0174] Optionally, when the controller determines the expansion distance of the obstacle based on the covariance matrix and the physical radius of the obstacle, it can use the following formula (19).

[0175] (19)

[0176] In formula (19), d inflated (t) represents the expansion distance of the obstacle at time t; R physical This represents the physical radius of the obstacle; This represents the covariance matrix; K represents the safety factor; σ(t) represents the standard deviation of the positioning error at time t. This represents a safety buffer based on the covariance matrix. The larger the covariance matrix (e.g., obstacles accelerate sudden changes), the better. The larger the value, the greater the safety buffer.

[0177] Optionally, when the controller determines the area corresponding to the obstacle at each time based on the expansion distance, it can use the expression shown in the following formula (20).

[0178] (20)

[0179] In formula (20), S safe (t) represents the safety buffer zone corresponding to the obstacle at time t, and formula (20) represents the distance X between all points in the safety buffer zone corresponding to the obstacle and the obstacle at time t. obs (t) is the expansion distance of the obstacle at time t, which is less than or equal to t. Where X obs (t)=[p x (t),p y (t),p z (t)].

[0180] In this embodiment, the controller projects the regions corresponding to the obstacles at each moment in the processed obstacle flight trajectory from four-dimensional spacetime to three-dimensional space, and uses the complement of the projected three-dimensional space as a safe flight channel. This can include: searching for connected free spaces that are not occupied by the expanded region segment by segment on the time axis from the start point to the end point of the processed obstacle flight trajectory; and using multiple segments of free space as safe flight channels.

[0181] A safe flight path can be represented as a sequence of polyhedra (such as cuboids, hexagonal prisms, etc.) that change over time. For example, a safe flight path can be represented as For t in[t k , t k+1 ]: C(t) = {(x,y,z)|A k (t) *[x;y;z;t]≤b k (t)}, that is, within the time period [t] k , t k+1 Within [the specified range], the position of the aircraft must satisfy the linear inequality constraint A. k (t) *[x;y;z;t]≤b k (t). The safe flight path is formed by connecting these polyhedra end to end.

[0182] In some embodiments, the controller, with the objective of determining the minimum value of the objective function, solves the objective function based on safe flight path constraints, dynamic constraints, and obstacle constraints. This may include: with the objective of determining the minimum value of the objective function, solving the objective function using the interior point method or the active set method based on safe flight path constraints, dynamic constraints, and obstacle constraints.

[0183] In some embodiments, the solution result is the optimal set of control point coordinates; the controller determines the target flight trajectory based on the solution result, which may include: generating a target Bézier curve based on the optimal set of control point coordinates, and using the target Bézier curve as the target flight trajectory.

[0184] Please see Figure 3 , Figure 3 This is a schematic diagram illustrating an optional process for determining a target flight trajectory, provided in an embodiment of this application. For example... Figure 3 As shown, the controller can generate Bézier curves based on the current and desired pose states of the aircraft, and provide convex hull constraints on the Bézier curves based on obstacle prediction and map-generated safe flight paths; initialize Bézier curve control points, and construct objective functions and constraints for solving the target flight trajectory based on the parametric equations of the Bézier curves; solve for the optimal control points; output the initial flight trajectory based on the optimal control points; verify the initial flight trajectory to check its corresponding dynamic feasibility and safe flight path constraints. If the verification passes, the initial flight trajectory is output as the target flight trajectory. If the verification fails, the safe flight path boundary is adjusted or the problem weights are optimized, and the optimal control points are solved again to regenerate the target flight trajectory.

[0185] By adopting this implementation method, the trajectory generation problem can be transformed into a constrained quadratic programming problem, which can quickly determine the target flight trajectory.

[0186] In one alternative implementation, Figure 1 In the flight trajectory determination method shown, the controller can also acquire multi-source raw data from the previous moment of the current moment collected during the flight of the aircraft; perform time alignment processing on the multi-source raw data based on the time synchronization protocol to obtain processed multi-source raw data; and transform the processed multi-source raw data into the body coordinate system of the aircraft's center of mass to obtain multi-source data.

[0187] In some embodiments, the time synchronization protocol may include a Precision Time Protocol (PTP).

[0188] This implementation method performs spatiotemporal alignment processing on the multi-source raw data from the previous time step to obtain multi-source data. This eliminates the heterogeneity between the multi-source raw data, thereby providing high-quality data for subsequent processing.

[0189] The following is combined with Figure 4 This paper provides an overall description of the flight trajectory determination method provided in the embodiments of this application. Please refer to [link to relevant documentation]. Figure 4 , Figure 4 This is a schematic diagram of another optional process for a flight trajectory determination method provided in an embodiment of this application. For example... Figure 4 As shown, the method for determining the flight trajectory may include, but is not limited to, the following steps:

[0190] S401. Acquire multi-source raw data from the previous moment of the current moment collected during the flight of the aircraft.

[0191] In some embodiments, the multi-source raw data may include location data, millimeter-wave radar data, lidar data, and attitude data, etc.

[0192] S402. Perform spatiotemporal alignment processing on the multi-source raw data of the previous time step to obtain the multi-source data of the previous time step.

[0193] In some embodiments, the controller performs spatiotemporal alignment processing on the multi-source raw data of the previous moment to obtain the multi-source data of the previous moment. This may include: performing time alignment processing on the multi-source raw data based on a time synchronization protocol to obtain the processed multi-source raw data; and converting all the processed multi-source raw data to the body coordinate system of the aircraft's center of mass to obtain the multi-source data.

[0194] S403. Determine the confidence levels of the multiple acquisition devices associated with the multi-source data at the previous time.

[0195] In some embodiments, the multi-source data may include location data, millimeter-wave radar data, lidar data, and attitude data; the controller determines the confidence levels of multiple acquisition devices associated with the multi-source data at the previous moment, which may include: determining the number of satellites, horizontal accuracy factor, and base station signal strength corresponding to the location sensor associated with the location data at the previous moment, and determining the confidence level of the location sensor at the previous moment based on the number of satellites, horizontal accuracy factor, and base station signal strength; the horizontal accuracy factor is used to represent the degree of influence of the spatial geometric distribution of satellites on the two-dimensional horizontal positioning accuracy; determining the signal-to-noise ratio and target scattering point corresponding to the millimeter-wave radar sensor associated with the millimeter-wave radar data at the previous moment. Consistency indices are used to determine the confidence level of the millimeter-wave radar at the previous time step, based on the signal-to-noise ratio and target scattering point consistency indices. The target scattering point consistency index is used to measure the quality of the synthetic image corresponding to the millimeter-wave radar. The point cloud density, rainfall intensity, and relative motion ambiguity of the lidar sensor associated with the lidar data at the previous time step are determined, and the confidence level of the lidar sensor at the previous time step is determined based on these indices. The vibration amplitude and temperature drift compensation residuals of the attitude sensor associated with the attitude data at the previous time step are determined, and the confidence level of the attitude sensor at the previous time step is determined based on these indices.

[0196] S404. With the goal of minimizing the variance after data fusion, determine the weights of each data point in the multi-source data based on the confidence levels of each acquisition device at the previous time.

[0197] In some embodiments, the variance after data fusion can be as shown in the aforementioned formula (13).

[0198] S405. Based on the weights corresponding to each data point, the multi-source data are fused to obtain the global optimal estimated pose state of the aircraft at the previous moment.

[0199] In some embodiments, when the controller performs data fusion on the multi-source data based on the weights corresponding to each data point to obtain the global optimal estimated pose state of the aircraft at the previous moment, the aforementioned formula (15) can be used.

[0200] S406. Based on the globally optimal estimated pose state and the control input parameters at the current moment, determine the predicted pose state at the current moment, and based on the predicted pose state, determine the predicted multi-source data at the current moment.

[0201] In some embodiments, when the controller determines the predicted pose state at the current moment based on the globally optimal estimated pose state and the control input parameters at the current moment, the aforementioned formula (1) may be used.

[0202] In some embodiments, the controller determines the predicted multi-source data at the current moment based on the predicted pose state, which may include: inputting the predicted pose state into a pre-determined multi-source data prediction model to obtain the predicted multi-source data at the current moment.

[0203] S407. Based on the process noise covariance matrix and the first Jacobian matrix at the current moment, and the previous error covariance matrix at the previous moment, determine the error covariance matrix corresponding to the predicted pose state.

[0204] Among them, the error covariance matrix is ​​used to characterize the degree of uncertainty in the predicted pose state; the first Jacobian matrix is ​​the Jacobian matrix of the prediction function of the error covariance matrix.

[0205] In some embodiments, when the controller determines the error covariance matrix corresponding to the predicted pose state based on the process noise covariance matrix at the current time, the first Jacobian matrix, and the previous error covariance matrix at the previous time, the aforementioned formula (3) may be used.

[0206] S408. Based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current time, determine the correction factor used to correct the predicted pose state.

[0207] The second Jacobian matrix is ​​the Jacobian matrix of the multi-source data prediction function.

[0208] In some embodiments, the controller determines a correction factor for correcting the predicted pose state based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current time. This may include: determining the Kalman gain using the aforementioned formula (4) based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current time; and using the Kalman gain as a correction factor for correcting the predicted pose state.

[0209] S409. Based on the correction factor, the actual multi-source data at the current time and the predicted multi-source data, the predicted pose state is corrected to obtain the current pose state at the current time.

[0210] In some embodiments, when the controller corrects the predicted pose state based on the correction factor, the actual multi-source data at the current time and the predicted multi-source data, the aforementioned formula (5) can be used to obtain the current pose state at the current time.

[0211] S410. Based on the current pose state and the desired pose state, construct the parametric equation of the Bézier curve, and based on the parametric equation of the Bézier curve, construct the objective function for solving the target flight trajectory.

[0212] S411. To determine the minimum value of the objective function, the objective function is solved based on the constraints of safe flight path, dynamic constraints, and obstacle constraints, and the target flight trajectory is determined based on the solution results.

[0213] In some embodiments, the specific processes of steps S410 and S411 can be found in the relevant descriptions above, and will not be repeated here.

[0214] In this embodiment, the weights of the multi-source data corresponding to the previous time are determined based on the confidence levels of multiple acquisition devices associated with the multi-source data corresponding to the aircraft at the previous time. This minimizes the impact of unreliable high-precision data on the fusion result (the globally optimal estimated pose state at the previous time), thereby improving the accuracy of the determined globally optimal estimated pose state at the previous time. The predicted pose state of the aircraft at the current time is determined based on the globally optimal estimated pose state of the previous time, thus minimizing the problem of low accuracy in the aircraft's pose state estimation due to interference or temporary loss of sensor signals. Furthermore, based on the predicted pose state, the predicted multi-source data for the current time is determined, and the predicted pose state is corrected based on the actual multi-source data and the predicted multi-source data at the current time. This further improves the accuracy of the pose state at the current time (i.e., the current pose state). Therefore, based on the highly accurate current pose state and the desired pose state, a highly accurate flight trajectory can be generated. It is evident that this method can improve the accuracy of flight trajectory prediction. Furthermore, by transforming the trajectory generation problem into a constrained quadratic programming problem, the target flight trajectory can be determined quickly.

[0215] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0216] Based on the same inventive concept, this application also provides a flight trajectory determination device for implementing the flight trajectory determination method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more flight trajectory determination device embodiments provided below can be found in the limitations of the flight trajectory determination method described above, and will not be repeated here.

[0217] Please see Figure 5 , Figure 5 This is a schematic diagram of an optional structure of a flight trajectory determination device provided in an embodiment of this application. For example... Figure 5 As shown, the flight trajectory determination device may include, but is not limited to:

[0218] The data processing module 501 is used to perform data fusion on the multi-source data corresponding to the previous time of the current time of the aircraft to obtain the global optimal estimated pose state of the aircraft at the previous time.

[0219] The determination module 502 is used to determine the predicted pose state at the current time based on the globally optimal estimated pose state and the control input parameters at the current time, and to determine the predicted multi-source data at the current time based on the predicted pose state.

[0220] The correction module 505 is used to correct the predicted pose state based on the actual multi-source data and the predicted multi-source data at the current time, so as to obtain the current pose state at the current time.

[0221] The trajectory generation module 504 is used to generate the target flight trajectory based on the current pose state and the desired pose state.

[0222] In some embodiments, when the correction module 505 corrects the predicted pose state based on the actual multi-source data and the predicted multi-source data at the current time to obtain the current pose state, the determination module 502 is further used to determine a correction factor for correcting the predicted pose state; the correction factor is determined at least based on the error covariance matrix corresponding to the predicted pose state and the measurement noise covariance matrix at the current time; the difference between the actual multi-source data and the predicted multi-source data at the current time is determined; the correction module 505 is used to correct the predicted pose state based on the correction factor and the difference to obtain the current pose state at the current time.

[0223] In some embodiments, when determining the correction factor for correcting the predicted pose state, the determining module 502 specifically performs the following steps: based on the process noise covariance matrix at the current time, the first Jacobian matrix, and the previous error covariance matrix at the previous time, it determines the error covariance matrix corresponding to the predicted pose state; the error covariance matrix is ​​used to characterize the uncertainty of the predicted pose state; the first Jacobian matrix is ​​the Jacobian matrix of the prediction function of the error covariance matrix; based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current time, it determines the correction factor for correcting the predicted pose state; the second Jacobian matrix is ​​the Jacobian matrix of the prediction function of multi-source data.

[0224] In some embodiments, when the data processing module 501 performs data fusion on the multi-source data corresponding to the previous time of the aircraft at the current time to obtain the globally optimal estimated pose state of the aircraft at the previous time, the determination module 502 is further used to determine the multi-source data corresponding to the previous time of the aircraft at the current time; determine the confidence levels of the multiple acquisition devices associated with the multi-source data at the previous time; and determine the weights of each data in the multi-source data based on the confidence levels of each acquisition device at the previous time, with the goal of minimizing the variance after data fusion; the data processing module 501 performs data fusion on the multi-source data based on the weights corresponding to each data to obtain the globally optimal estimated pose state of the aircraft at the previous time.

[0225] In some embodiments, the multi-source data includes at least two of the following: location data, millimeter-wave radar data, lidar data, and attitude data; when determining the confidence levels of multiple acquisition devices associated with the multi-source data at the previous moment, the determining module 502 specifically performs at least two of the following: determining the number of satellites, horizontal accuracy factor, and base station signal strength corresponding to the location sensor associated with the location data at the previous moment, and determining the confidence level of the location sensor at the previous moment based on the number of satellites, horizontal accuracy factor, and base station signal strength; the horizontal accuracy factor is used to represent the degree of influence of the spatial geometric distribution of satellites on the two-dimensional horizontal positioning accuracy; determining the confidence level of the millimeter-wave radar sensor associated with the millimeter-wave radar data at the previous moment... The signal-to-noise ratio (SNR) and target scattering point consistency index are determined, and based on these indices, the confidence level of the millimeter-wave radar at the previous moment is determined. The target scattering point consistency index is used to measure the quality of the synthetic image corresponding to the millimeter-wave radar. The point cloud density, rainfall intensity, and relative motion ambiguity of the lidar sensor associated with the lidar data at the previous moment are determined, and based on these indices, the confidence level of the lidar sensor at the previous moment is determined. The vibration amplitude and temperature drift compensation residual of the attitude sensor associated with the attitude data at the previous moment are determined, and based on these indices, the confidence level of the attitude sensor at the previous moment is determined.

[0226] In some embodiments, when the trajectory generation module 504 generates a target flight trajectory based on the current pose state and the desired pose state, it specifically performs the following steps: constructing the parametric equation of the Bézier curve based on the current pose state and the desired pose state, and constructing an objective function for solving the target flight trajectory based on the parametric equation of the Bézier curve; determining the minimum value of the objective function as the objective, solving the objective function based on safe flight channel constraints, dynamic constraints, and obstacle constraints, and determining the target flight trajectory based on the solution results.

[0227] In some embodiments, the determining module 502 is further configured to predict the flight trajectory of an obstacle when an obstacle is detected within a preset distance range of the aircraft, and predict the covariance matrix corresponding to the obstacle based on the obstacle flight trajectory; the covariance matrix is ​​used to characterize the uncertainty of the obstacle during flight; perform spatiotemporal dilation processing on the obstacle flight trajectory based on the covariance matrix to obtain the processed obstacle flight trajectory; project the region corresponding to the obstacle at each moment in the processed obstacle flight trajectory from four-dimensional spatiotemporal to three-dimensional space, and use the complement of the projected three-dimensional space as a safe flight channel; and generate a safe flight channel constraint based on the safe flight channel.

[0228] In some embodiments, the device may further include an acquisition module; the acquisition module is used to acquire multi-source raw data collected during the flight of the aircraft at the previous time of the current time; the data processing module 501 is also used to perform time alignment processing on the multi-source raw data based on a time synchronization protocol to obtain processed multi-source raw data; and to convert the processed multi-source raw data to the body coordinate system of the aircraft's center of mass to obtain multi-source data.

[0229] It is understood that the specific implementation of each module in the flight trajectory determination device provided in this application embodiment and the beneficial effects that can be achieved can be referred to the description of the aforementioned flight trajectory determination method embodiment, and will not be repeated here.

[0230] Each module in the aforementioned flight trajectory determination device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the vehicle-mounted terminal device in hardware form or stored in the memory of the flight trajectory determination device in software form, so that the processor can call and execute the corresponding operations of each module.

[0231] In one exemplary embodiment, a controller is provided, the internal structure of which can be shown in the following diagram. Figure 6As shown, the controller includes a processor, memory, input / output interfaces, a communication interface, and input devices. The processor, memory, and input / output interfaces are connected via a system bus, while the communication interface, display unit, and input devices are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a flight trajectory determination method.

[0232] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the controller to which the present application is applied. A specific controller may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0233] In one exemplary embodiment, this application provides a controller, including a memory and a processor, wherein the memory stores a computer program; when the processor executes the computer program, it implements the steps in the flight trajectory determination methods described above.

[0234] In one exemplary embodiment, this application provides a computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements the steps in the flight trajectory determination methods described above.

[0235] In one exemplary embodiment, this application provides a computer program product, including a computer program. When executed by a processor, the computer program implements the steps in the flight trajectory determination methods described above.

[0236] It should be noted that the data involved in this application (including but not limited to data used for analysis, data stored, data displayed, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0237] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0238] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0239] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining a flight trajectory, characterized in that, The method includes: Data fusion is performed on the multi-source data corresponding to the previous time of the current time of the aircraft to obtain the global optimal estimated pose state of the aircraft at the previous time. Based on the globally optimal estimated pose state and the control input parameters at the current moment, the predicted pose state at the current moment is determined, and based on the predicted pose state, the predicted multi-source data at the current moment is determined. Based on the actual multi-source data and the predicted multi-source data at the current moment, the predicted pose state is corrected to obtain the current pose state at the current moment. Based on the current pose state and the desired pose state, the target flight trajectory is generated.

2. The method according to claim 1, characterized in that, The step of correcting the predicted pose state based on the actual multi-source data and the predicted multi-source data at the current moment to obtain the current pose state at the current moment includes: A correction factor is determined to correct the predicted pose state; the correction factor is determined based at least on the error covariance matrix corresponding to the predicted pose state and the measurement noise covariance matrix at the current time. Determine the difference between the actual multi-source data and the predicted multi-source data at the current moment; Based on the correction factor and the difference, the predicted pose state is corrected to obtain the current pose state at the current moment.

3. The method according to claim 2, characterized in that, The determination of the correction factor used to correct the predicted pose state includes: Based on the process noise covariance matrix and the first Jacobian matrix at the current moment, and the previous error covariance matrix at the previous moment, the error covariance matrix corresponding to the predicted pose state is determined; the error covariance matrix is ​​used to characterize the uncertainty of the predicted pose state; the first Jacobian matrix is ​​the Jacobian matrix of the error covariance matrix prediction function. Based on the error covariance matrix, the second Jacobian matrix, and the measurement noise covariance matrix at the current moment, a correction factor for correcting the predicted pose state is determined; the second Jacobian matrix is ​​the Jacobian matrix of the multi-source data prediction function.

4. The method according to claim 1, characterized in that, The process of fusing multi-source data corresponding to the previous time step of the current time step to obtain the globally optimal estimated pose state of the aircraft at the previous time step includes: Determine the multi-source data corresponding to the previous time step of the current time step; Determine the confidence levels of the multiple acquisition devices associated with the multi-source data at the previous time. With the goal of minimizing the variance after data fusion, the weights of each data point in the multi-source data are determined based on the confidence levels of each acquisition device at the previous time. Based on the weights corresponding to each of the data sources, the multi-source data are fused to obtain the global optimal estimated pose state of the aircraft at the previous moment.

5. The method according to claim 4, characterized in that, The multi-source data includes at least two of the following: position data, millimeter-wave radar data, lidar data, and attitude data; determining the confidence levels of the multiple acquisition devices associated with the multi-source data at the previous moment includes at least two of the following: The number of satellites, horizontal accuracy factor, and base station signal strength corresponding to the location sensor associated with the location data at the previous time are determined, and the confidence level of the location sensor at the previous time is determined based on the number of satellites, the horizontal accuracy factor, and the base station signal strength; the horizontal accuracy factor is used to represent the degree of influence of the spatial geometric distribution of satellites on the two-dimensional horizontal positioning accuracy. The signal-to-noise ratio (SNR) and target scattering point consistency index of the millimeter-wave radar sensor associated with the millimeter-wave radar data at the previous time step are determined, and the confidence level of the millimeter-wave radar at the previous time step is determined based on the SNR and the target scattering point consistency index; wherein, the target scattering point consistency index is used to measure the quality of the synthetic image corresponding to the millimeter-wave radar. Determine the point cloud density, rainfall intensity, and relative motion ambiguity of the lidar sensor associated with the lidar data at the previous moment, and determine the confidence level of the lidar sensor at the previous moment based on the point cloud density, the rainfall intensity, and the relative motion ambiguity. Determine the vibration amplitude and temperature drift compensation residual of the attitude sensor associated with the attitude data at the previous moment, and determine the confidence level of the attitude sensor at the previous moment based on the vibration amplitude and the temperature drift compensation residual.

6. The method according to claim 1, characterized in that, The step of generating the target flight trajectory based on the current pose state and the desired pose state includes: Based on the current pose state and the desired pose state, construct the parametric equation of the Bézier curve, and based on the parametric equation of the Bézier curve, construct the objective function for solving the target flight trajectory. The objective function is to be minimized. Based on safe flight path constraints, dynamic constraints, and obstacle constraints, the objective function is solved, and the target flight trajectory is determined based on the solution.

7. The method according to claim 6, characterized in that, The safe flight path constraints are determined in the following ways: If an obstacle is detected within a preset distance range of the aircraft, the flight trajectory of the obstacle is predicted, and based on the flight trajectory of the obstacle, the covariance matrix corresponding to the obstacle is predicted; the covariance matrix is ​​used to characterize the uncertainty of the obstacle during flight. Based on the covariance matrix, the obstacle flight trajectory is subjected to spatiotemporal dilation processing to obtain the processed obstacle flight trajectory; The regions corresponding to the obstacles at each moment in the processed obstacle flight trajectory are projected from four-dimensional spacetime to three-dimensional space, and the complement of the projected three-dimensional space is used as a safe flight channel. Based on the safe flight path, the safe flight path constraints are generated.

8. The method according to any one of claims 1 to 7, characterized in that, The method further includes: Acquire multi-source raw data from the previous moment at the current moment, collected during the flight of the aircraft; The multi-source raw data is time-aligned based on a time synchronization protocol to obtain the processed multi-source raw data. The processed multi-source raw data are all transformed into the body coordinate system of the aircraft's center of mass to obtain the multi-source data.

9. A flight trajectory determination device, characterized in that, The device includes: The data processing module is used to perform data fusion on the multi-source data corresponding to the previous time of the current time of the aircraft to obtain the global optimal estimated pose state of the aircraft at the previous time. The determination module is used to determine the predicted pose state at the current time based on the globally optimal estimated pose state and the control input parameters at the current time, and to determine the predicted multi-source data at the current time based on the predicted pose state. The correction module is used to correct the predicted pose state based on the actual multi-source data and the predicted multi-source data at the current time, so as to obtain the current pose state at the current time. The trajectory generation module is used to generate the target flight trajectory based on the current pose state and the desired pose state.

10. A controller comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images