An indoor environment prediction and stabilization control method

Abstract

This application discloses an indoor environment pre-aiming stabilization control method, relating to the field of UAV flight control. The method includes: acquiring point cloud data of the indoor environment; updating the occupancy grid map based on the point cloud data using a log-probability update mechanism; determining the planned path of the UAV; determining the environmental complexity indices of the UAV's current position and pre-aiming position based on the updated occupancy grid map and a wall disturbance model; constructing a time-varying state predictor for the UAV based on a reference model; determining the control signal and adaptive law based on the time-varying state predictor; adjusting the control signal using the adaptive law to obtain the final control signal; and controlling the UAV's flight based on the final control signal. This application can improve the flight attitude stability and trajectory tracking accuracy of indoor operational UAVs.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) flight control, and in particular to an indoor environment pre-aiming stabilization control method. Background Technology

[0002] With continuous breakthroughs in navigation, positioning, and automatic control technologies, multi-rotor drones for indoor operations have been widely used in many fields such as industrial inspection and rubble search and rescue due to their low cost, simple structure, and diverse payloads. However, because indoor environments are sufficiently close to constrained environments, aerodynamic interactions are significant, posing challenges to the anti-interference capabilities of drone control systems.

[0003] In existing technologies, various effective solutions have been proposed for stabilization control of unmanned aerial vehicles (UAVs). For example, some methods consider system uncertainties and external disturbances, attributing the effects of unmodeled dynamic disturbances and positional disturbances to the estimation and compensation of the system's "lumped disturbance." Other methods introduce the airframe dynamics model through a turbulent wind field model, designing a hybrid stabilization method combining a cascaded PID position controller and an adaptive backstepping attitude controller, achieving strong stability and robustness even under strong wind disturbances in confined spaces. Still others improve the differential tracker by combining the Sigmoid function and the sliding mode terminal attractor, designing a novel third-order adaptive variable gain finite-time dilation observer to compensate for unknown disturbances and model uncertainties experienced by quadrotor UAVs, accelerating system convergence and improving trajectory tracking performance. Finally, a robust H∞ controller based on acceleration feedback enhancement has been designed, solving the problem of the difficulty in implementing conventional high-gain acceleration feedback in practical systems, ensuring stable navigation of rotary-wing UAVs in windy environments.

[0004] Looking at various stabilization control methods for rotary-wing UAVs, existing technical solutions mainly fall into two categories: one is to reduce the system's sensitivity to disturbances by improving the robustness of the controller itself; the other is to accurately estimate disturbances and achieve real-time compensation. However, existing technical solutions are only used at the planning level, and for complex constrained environments, airflow disturbances and the proximity of surrounding obstacles are monotonically correlated, failing to incorporate information about the aircraft's surrounding environment into the control process. Currently, the robustness and rapid maneuverability of UAV passive flight controllers need to be balanced, resulting in a sacrifice of nominal performance while ensuring disturbance suppression performance. Furthermore, the compensation mechanism is based on current state feedback, resulting in a time delay in compensation, and it cannot proactively adjust robustness in advance for future external disturbances.

[0005] Therefore, based on the above problems, there is an urgent need to provide an indoor environment pre-aiming stabilization control method that can comprehensively improve the performance of the flight controller and enhance the flight attitude stability and trajectory tracking accuracy of indoor operation UAVs. Summary of the Invention

[0006] The purpose of this application is to provide an indoor environment pre-aiming stabilization control method that can comprehensively improve the performance of the flight controller and enhance the flight attitude stability and trajectory tracking accuracy of indoor operation UAVs.

[0007] To achieve the above objectives, this application provides the following solution: Firstly, this application provides an indoor environment pre-targeting stabilization control method, comprising: Acquire point cloud data of the indoor environment; Based on point cloud data and a log-probability update mechanism, the occupied grid map is updated, and the planned path of the UAV is determined; the occupied grid map is determined based on historical point cloud data. Based on the updated occupancy grid map and the wall disturbance model, the environmental complexity indices of the UAV's current location and the target location are determined respectively; the target location is the location that the UAV will reach in the future during its flight along the planned path. Based on the environmental complexity indices of the current location and the target location, a time-varying state predictor for the UAV is constructed using a reference model. The control signal and adaptive law are determined based on the time-varying state predictor; the control signal is adjusted using the adaptive law to obtain the final control signal; and the UAV is controlled to fly based on the final control signal.

[0008] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides an indoor environment pre-aiming stabilization control method. By acquiring point cloud data of the indoor environment and updating the occupancy grid map based on a log-probability update mechanism, the method determines the planned path of the UAV. This application considers current and future environmental changes, improving the adaptability of the occupancy grid map to dynamic environments and its consistency in long-term operation. Based on the updated occupancy grid map and a wall disturbance model, the environmental complexity indices of the UAV's current and pre-aiming positions are determined, achieving accurate perception and proactive adaptation to complex indoor wall environments, thus improving the UAV's flight safety and path tracking robustness in complex spaces. Based on the environmental complexity indices of the current and pre-aiming positions, a time-varying state predictor for the UAV is constructed using a reference model to further determine the final control signal. This achieves advance compensation and disturbance suppression for the UAV's dynamic response, improving the trajectory tracking accuracy and anti-interference capability of the UAV in indoor environments. In summary, this application comprehensively improves the performance of the flight controller and enhances the flight attitude stability and trajectory tracking accuracy of indoor operational UAVs. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating an indoor environment pre-targeting stabilization control method in one embodiment of this application; Figure 2 This is a schematic diagram of an indoor environment pre-targeting stabilization control method in one embodiment of this application; Figure 3 This is a schematic diagram of wall effect intensity fitting in one embodiment of this application; Figure 4 This is a schematic diagram illustrating the determination of the comprehensive complexity index in one embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0013] In one exemplary embodiment, such as Figure 1 and Figure 2 As shown, an indoor environment pre-aiming stabilization control method is provided, including the following S1 to S5. Wherein: S1: Acquire point cloud data of the indoor environment.

[0014] Point cloud data is acquired through a depth camera or LiDAR. In an exemplary embodiment, LiDAR is used to acquire the distance and angle information of objects in the indoor environment and convert it into a three-dimensional point cloud. Then, an Inertial Measurement Unit (IMU) is used to convert the three-dimensional point cloud from the sensor coordinate system to world coordinates to achieve multi-frame alignment. Further preprocessing such as denoising and downsampling is performed on the coordinate-transformed three-dimensional point cloud to obtain a preprocessed three-dimensional point cloud.

[0015] S2: Based on point cloud data and a log-probability update mechanism, update the occupied grid map and determine the planned path of the UAV.

[0016] The occupancy grid map is determined based on historical point cloud data. Specifically, the space is discretized into voxel grids at a certain resolution, and the preprocessed 3D point cloud is mapped onto the corresponding voxel grids. Then, through ray projection, voxel grids along the laser path are marked as free, and the endpoint voxel grid is marked as occupied. Finally, a Bayesian log-probability update method is used to fuse multi-frame observations, continuously updating the occupancy probability of each voxel grid, and gradually constructing a 3D occupancy grid map. The occupancy grid map is used to store the occupancy probability of voxel grids.

[0017] When new observations are received During this process, the occupancy raster map continuously integrates new occupancy information data obtained from the preprocessed 3D point cloud, updating the original occupancy probability of the occupancy raster map. The logarithmic representation of the update probability for each frame is as follows: ; in, For the raster update occupancy probability, This represents the historical occupancy probability of a grid. For new observations, Indicates that a grid cell is occupied. Indicates an empty grid cell.

[0018] When lidar collects point cloud data, it encounters degraded environments such as smoke and reflectivity, resulting in noise. Inertial measurement units (IMUs) also exhibit drift and noise. Therefore, lidar and IMUs are the primary sources of noise when constructing occupancy grid maps; this noise is collectively referred to as sensor noise. The log-probability update mechanism enables smooth increases and decreases in voxel occupancy probabilities, thereby enhancing the robustness and stability of the occupancy grid map against sensor noise and dynamic uncertainties. The updated occupancy probability information is stored in the voxel information structures (VIS) of each voxel grid, providing a foundation for subsequent spatial aggregation of environmental disturbances.

[0019] S3: Based on the updated occupancy grid map and the wall disturbance model, determine the environmental complexity index of the UAV's current location and the target location.

[0020] The current location point and the preview location point are collectively referred to as the selected point. The current location point is the current position coordinate of the UAV itself, and the preview location point is the future position coordinate of the UAV during its flight along the planned path, that is, a point selected on the planned path by setting a preview time interval. S3 specifically includes: S31: Based on the updated occupancy grid map and the wall perturbation model, determine the spatial perturbation mapping operator.

[0021] When a ducted platform interacts aerodynamically with its environment at close range, a severe coupling effect exists between the two, characterized by high dimensionality, complexity, and strong nonlinearity. During takeoff and hovering near the ground, the outflow from the duct is deflected to both sides due to ground obstruction, altering the outflow state and generating complex vortices between the ducts, causing ground-based aerodynamic disturbances on the platform. When the ducted platform flies near the ceiling, the deflection of the inflow from the duct reduces the pressure in the inflow region, and the backflow from the UAV fuselage causes changes in the thrust and torque of the duct propeller, generating ceiling-based aerodynamic disturbances. When approaching a vertical wall, the inflow state on one side of the duct is affected, and the static pressure distribution at the lips of the duct is asymmetrical. At this time, the wall aerodynamic effects generate a tilting moment towards the wall, severely affecting the system stability.

[0022] The intensity of the wall effect is approximately exponentially related to the distance from the confined environment, using dimensionless parameters. This indicates the proximity of the UAV platform to the wall. The maximum wall effect disturbance occurs when the duct is in complete contact with the wall. The lateral attraction effect of the wall, i.e., the calculation formula for the wall disturbance model, is as follows: ; in, For the positive real number field, The lateral disturbance coefficient is... Indicates the proximity of the drone platform to the wall. , This represents the actual distance from the drone to the wall. Where is the propeller radius. The first fitting coefficient, The second fitting coefficient, and It is determined by the propeller airfoil, the number of propeller blades, and the duct structure.

[0023] For the boundary voxel set, let Indicates a selected point within the indoor environment The normalized distance to the boundary of the surrounding voxels. ,in, For the first The normalized distance from the individual pixel grid to the selected point. Correspondingly, define the occupancy probability weight vector. ,in, For the first The raster occupancy probability of a single-genome mesh. , For the first Historical occupancy probability of individual pixel grids This represents the total number of voxel grids.

[0024] Based on the wall perturbation model (2), a spatial perturbation mapping operator is defined. The calculation formula is as follows: ; in, The normalized distance from the voxel mesh to the selected point is Lateral disturbance coefficient at time, , For raster map resolution, For Hadamard exponentiation. For indicator functions, when When the value exceeds 5, the disturbance approaches 0.

[0025] Multiple measurement points were set up to fit and verify the wall disturbance model, such as... Figure 3 As shown, the measurement points in the near-wall region are relatively scattered. As the distance increases, the scattered points gradually converge towards the wall perturbation model curve. The overall trend is highly consistent with the wall perturbation model curve, which means that the wall perturbation model can well describe the law of the change of the lateral perturbation force coefficient with the normalized distance.

[0026] S32: Determine the environmental complexity index of the current position and the target position of the UAV based on the spatial disturbance mapping operator.

[0027] Based on the occupancy grid map, the environmental complexity index of a single point is obtained by aggregating the perturbation contributions of all obstacle voxels in the neighborhood of a selected point. The calculation formula is as follows: ; in, It is a spatial aggregation operator.

[0028] To integrate the complexity of the current and future environments, spatial aggregation of the current location and the projected location is required. The projected location is extracted from the planning path generated by the planning layer. The expression is as follows: ; in, To plan the route in Coordinate components on the axis, To plan the route in Coordinate components on the axis, To plan the route in Coordinate components on the axes.

[0029] Therefore, the aiming point position can be expressed as ,in, , Indicates the current moment. For the aiming time interval, then the first... The formula for calculating the selected point complexity of a frame, i.e., the environmental complexity index of a selected point, is as follows: ; ; in, This is an indicator of the environmental complexity at the current location. The environmental complexity index for the target location. For spatial aggregation operators, For spatial perturbation mapping operators, To determine the normalized distance from the current location of the UAV to the boundaries of surrounding voxels within the sensing area, This is the occupancy probability weight vector. This is the normalized distance from the UAV's pre-aiming position point to the boundary of the surrounding voxels within the sensing area.

[0030] S4: Based on the environmental complexity index of the current location and the target location, construct a time-varying state predictor for the UAV based on the reference model.

[0031] S4 specifically includes: S41: Smooth the environmental complexity indices of the current location and the target location respectively; and perform weighted fusion to obtain a comprehensive complexity index.

[0032] Use a queue structure for storage Select the point complexity index data in the frame and apply the moving average filtering method to each queue.

[0033] Since the environmental complexity index sequence at the current location point reflects both current and historical disturbance characteristics, the Exponential Moving Average (EMA) method is used to smooth the environmental complexity index at the current location point, achieving gradual decay of historical data. This effectively improves the robustness and real-time response capability of UAV flight control. The calculation formula is as follows: ; in, For the first The environmental complexity index of the current position point after frame smoothing. As the first smoothing weight, For the first The environmental complexity index of the current position point after frame smoothing. For the first The environmental complexity metric for the current position of the frame.

[0034] The environmental complexity index sequence of the pre-aimed location points characterizes the environmental disturbance characteristics of future path segments. To maintain the validity and representativeness of each frame of data in the environmental complexity index sequence of the pre-aimed location points, a simple moving average (SMA) method is used to smooth the environmental complexity index of the pre-aimed location points to avoid premature decay of prediction information. The calculation formula is as follows: ; in, For the first The environmental complexity index of the preview position point after frame smoothing. For the first The environmental complexity index of the preview position point after frame smoothing. This represents the number of data frames stored in the filter queue. For the first Environmental complexity metrics for frame preview location points.

[0035] like Figure 4 As shown, by integrating the complexity indices of the current position and the target position, a comprehensive complexity index is obtained, calculated using the following formula: ; in, For the first Frame complexity metrics The environmental complexity index for the current location point after smoothing. The environmental complexity index for the pre-aimed location points after smoothing. The adaptive fusion weights are used to adjust the contribution of the aiming information to ensure the continuity and smoothness of weight changes. , This serves as the lower bound for the adaptive fusion weights. This is the upper bound of the adaptive fusion weights. The predicted changes reflect the degree of environmental degradation along the future route segment. , This is an indicator of the environmental complexity of the previewed position points after smoothing in the first frame. For the first Environmental complexity index of the pre-aimed position point after frame smoothing S42: Determine performance variables based on the comprehensive complexity index.

[0036] Based on the comprehensive complexity index, determine the performance variables of the reference model. Also known as performance parameters, triangular forms are used to ensure the smooth continuity of performance variables in the reference model. Function as mapping Its Lipschitz constant is With the boundary set, the analytical expression for the performance variable is: ; in, For performance variables, To provide a comprehensive complexity index, The function expression form, To synthesize the mapping from complexity to performance parameters, this mapping adopts a C∞ form of trigonometric function mapping, which ensures the continuous and bounded nature of multiple derivatives, thereby guaranteeing the stable change of controller performance parameters.

[0037] Will Abbreviated as ,Will Abbreviated as The following formula is derived: ; in, This is the upper bound for performance variables. This is the lower bound for performance variables. This is the upper bound of the overall complexity index.

[0038] S43: Based on performance variables and a reference model, construct a time-varying state predictor for the UAV.

[0039] Indoor operational drones typically operate at low speeds, flying forward and hovering, without requiring large-angle maneuvers. Their operational range lies around the equilibrium point of the flight control system. In this state, the flight control system exhibits weak nonlinearity, satisfying the small-angle assumption. The coupling between channels is treated as an uncertainty, and the controller channels are decoupled. The platform's flight mode inner loop is a non-Hulwezian stable system. The closed-loop stability of the system needs to be achieved through a static parameter PID controller via the inner loop controller. Taking a single roll channel as an example, an adaptive controller is used to perform stability enhancement compensation by estimating system uncertainties in real time. A stable drone system with uncertainties and disturbances can be represented as follows: ; in, This is the output signal of the actual UAV system. For control signals, This represents the transfer function of the actual UAV system, but its structure is unknown. For the uncertainties and disturbance signals of the actual UAV system, , For the perturbation function related to time and state variables, This is a Laplace transform.

[0040] The time constant determines the dynamic performance of the system; a time-varying reference model is considered. Its state-space implementation can be written as: ; in, For observable states, For the actual observation variables of the UAV system, For the system matrix, , For the control matrix, , To control the input, For uncertainties and disturbances.

[0041] The performance variables are introduced into the state space of the time-varying reference model, and the reference model is rewritten as follows: ; in, This is the output signal of the actual UAV system. As a time-varying reference model, For control signals, For the uncertain input and the deviation after rewriting, , This represents the transfer function of the actual UAV system, but its structure is unknown. This refers to the uncertainties and disturbance signals in the actual system of the unmanned aerial vehicle (UAV).

[0042] Next, based on the state-space representation of the reference model, a time-varying state predictor for the UAV is constructed, calculated using the following formula: ; in, For performance variables, For control signals, For adaptive laws, The derivative of the UAV state estimate, For the state estimation of the UAV, Observations for UAV state estimation.

[0043] S5: Determine the control signal and adaptive law based on the time-varying state predictor; adjust the control signal using the adaptive law to obtain the final control signal; and control the UAV flight based on the final control signal.

[0044] Adaptive laws employ projection operators The calculation, which uses a smooth transformation of the vector field to limit the uncertainty estimation boundary, thereby preventing parameter drift in the adaptive scheme, is as follows: ; in, The derivative of the uncertainty estimate, Uncertainty estimation For projection operators, For adaptive gain constrained by a lower bound, increasing the adaptive gain can improve the convergence speed of system uncertainty estimation and make the actual system output arbitrarily close to the reference model. However, for systems with high adaptive gain, the degree to which they are affected by changes in initial conditions, reference input, and parameter uncertainties increases proportionally, which reduces the system's robustness margin. This is the error signal between the actual system state output and the state predictor output. , This is the state estimate of the unmanned aerial vehicle (UAV) system. This represents the actual state of the unmanned aerial vehicle (UAV) system. The system matrix is ​​a solution to the Lyapunov equation. This is the control matrix.

[0045] Based on a time-varying state predictor and an adaptive law, the control signal, i.e., the control law, is determined. The calculation formula is as follows: ; in, for The purpose of this first-order low-pass filter is to decouple the controller's uncertainty estimation capability from the system's robustness. The Laplace transform of the input signal, , for, For uncertainty estimation, These are variables in the complex frequency domain.

[0046] The system must meet the following conditions: ; in, The transfer function is used to ensure stability. for The mediator transfer function, It is a constant, based on the Lipschitz continuity assumption. definition, for Norm, and Let be any two variable trajectories, represented by . The derivative is bounded, where the uncertainty perturbation satisfies the above equation. It can be any function.

[0047] , and The norm is defined as follows: .

[0048] It is easy to derive and It is strictly regular and stable.

[0049] In summary, the adaptive controller architecture based on environment prediction is represented as follows: .

[0050] This application improves the efficiency of utilizing environmental perception information of UAVs, enabling the estimation of airflow intensity from obstacles within the forward-looking range during indoor UAV flight operations. This reveals trends for the controller to proactively adjust its performance parameters, enhancing the UAV's adaptability to complex environments and its disturbance suppression capabilities. Secondly, this application can proactively adjust the UAV's performance parameters in advance using forward-looking environmental data, resolving the difficult trade-off between robustness and speed in UAV flight control. Thirdly, this application uses the grid occupancy probability constructed from point cloud data as weights to perform spatial aggregation within a limited space between the platform's current position and the pre-aiming position, combined with empirical formulas for near-wall effects. It also considers multi-frame complexity for temporal smoothing and variable-weight fusion of pre-aiming points to form environmental complexity information. Furthermore, this application designs a mapping mechanism between environmental complexity and reference system performance parameters, enabling proactive adjustment of the controller's robustness performance in advance based on forward-looking environmental information to address uncertainties arising from interactions with the current and future environments. Real-time estimation and effective compensation for uncertainties within the system's effective bandwidth are achieved through adaptive gain real-time updates and output low-pass filtering shaping.

[0051] 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 specification.

[0052] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for predictive stabilization control of indoor environment, characterized in that, The method includes: Acquire point cloud data of the indoor environment; Based on point cloud data and a log-probability update mechanism, the occupied grid map is updated, and the planned path of the UAV is determined; the occupied grid map is determined based on historical point cloud data. Based on the updated occupancy grid map and the wall disturbance model, the environmental complexity indices of the UAV's current location and the target location are determined respectively; the target location is the location that the UAV will reach in the future during its flight along the planned path. Based on the environmental complexity indices of the current location and the target location, a time-varying state predictor for the UAV is constructed using a reference model. The control signal and adaptive law are determined based on the time-varying state predictor; the control signal is adjusted using the adaptive law to obtain the final control signal; and the UAV is controlled to fly based on the final control signal.

2. The indoor environment pre-aiming stabilization control method according to claim 1, characterized in that, The step of updating the occupied raster map based on point cloud data and a log-probability update mechanism specifically includes: Using formula Update the occupied grid map; in, For the raster update occupancy probability, For the historical occupancy probability of the grid, For new observations, Indicates that a grid cell is occupied. Indicates an empty grid cell.

3. The indoor environment pre-targeting stabilization control method according to claim 1, characterized in that, The step involves determining the environmental complexity indices of the UAV's current and target locations based on the updated occupancy grid map and the wall perturbation model. Specifically, this includes: Based on the updated occupancy grid map, the spatial perturbation mapping operator is determined according to the wall perturbation model; Based on the spatial perturbation mapping operator, the environmental complexity index of the UAV's current position and the pre-aiming position is determined.

4. The indoor environment pre-aiming stabilization control method according to claim 3, characterized in that, The step of determining the environmental complexity index of the UAV's current position and pre-aiming position based on the spatial perturbation mapping operator specifically includes: Using formula Determine the environmental complexity index of the current location of the drone. ; Using formula Determine the environmental complexity index of the drone's target location. ; in, For spatial aggregation operators, For spatial perturbation mapping operators, To determine the normalized distance from the current location of the UAV to the boundaries of surrounding voxels within the sensing area, This is the occupancy probability weight vector. This is the normalized distance from the UAV's pre-aiming position point to the boundary of the surrounding voxels within the sensing area.

5. The indoor environment pre-targeting stabilization control method according to claim 1, characterized in that, The step of constructing a time-varying state predictor for the UAV based on the environmental complexity index of the current location and the pre-aimed location, using a reference model, specifically includes: The environmental complexity indices of the current location and the target location are smoothed separately; then weighted and fused to obtain a comprehensive complexity index. Based on the comprehensive complexity index, determine the performance variables; Based on the performance variables and a reference model, a time-varying state predictor for the UAV is constructed.

6. The indoor environment pre-aiming stabilization control method according to claim 5, characterized in that, The smoothing process for the environmental complexity indices of the current location and the target location includes: The environmental complexity index at the current location is smoothed using the exponential moving average method. The environmental complexity index of the pre-aimed location point is smoothed using a simple moving average method.

7. The indoor environment pre-aiming stabilization control method according to claim 5, characterized in that, The above steps involve weighted fusion to obtain a comprehensive complexity index, which specifically includes: Using formula Obtain the comprehensive complexity index ; in, The environmental complexity index for the current location point after smoothing. The environmental complexity index for the pre-aimed location points after smoothing. For adaptive fusion weights.

8. The indoor environment pre-aiming stabilization control method according to claim 5, characterized in that, The determination of performance variables based on the comprehensive complexity index specifically includes: Using formula Determine performance variables ; in, To comprehensively measure complexity, This is a mapping from overall complexity to performance parameters.

9. The indoor environment pre-aiming stabilization control method according to claim 5, characterized in that, The step of constructing a time-varying state predictor for the UAV based on the performance variables and a reference model specifically includes: Using formula Construct a time-varying state predictor for drones; in, For performance variables, For control signals, For adaptive laws, The derivative of the UAV state estimate, For the state estimation of the UAV, Observations for UAV state estimation.

10. The indoor environment pre-aiming stabilization control method according to claim 1, characterized in that, The determination of the control signal and adaptive law based on the time-varying state predictor specifically includes: Using formula Determine the control signals of the drone ; Using formula Determine the adaptive law for the UAV; in, for A first-order low-pass filter, The Laplace transform of the input signal, For uncertainty estimation, The derivative of the uncertainty estimate, For projection operators, For adaptive gain constrained by a lower bound, This is the error signal between the actual system state output and the state predictor output. The system matrix is ​​a solution to the Lyapunov equation. For the control matrix, These are variables in the complex frequency domain.

Images