Radial speed regulation obstacle avoidance midcourse guidance method for multi-rotor unmanned aerial vehicle

Abstract

The application relates to the technical field of unmanned aerial vehicles, in particular to a radial speed regulation obstacle avoidance middle section guiding method of a multi-rotor unmanned aerial vehicle, which comprises the following steps: generating an original approaching acceleration based on target reference position information and unmanned aerial vehicle self-motion state information; taking the original approaching acceleration as a benchmark center to construct a candidate maneuver instruction set in a three-dimensional acceleration space; generating a plurality of fusion candidate acceleration instructions; obtaining a feasible candidate instruction set for each fusion candidate acceleration instruction; selecting an instruction with the minimum cost as an actual execution instruction output of middle section guiding through a comprehensive cost function; when the unmanned aerial vehicle enters a preset guiding mode switching area, dynamically fusing the actual execution instruction of middle section guiding with a terminal section guiding instruction to generate a final guiding instruction; and the method combines active radial velocity adjustment, acceleration domain obstacle avoidance and middle-terminal section guiding state inheritance, and realizes efficient, safe and smooth whole-process approaching guiding of the multi-rotor unmanned aerial vehicle in a complex urban environment.

Description

Technical Field

[0001] The embodiments of this application relate to the field of unmanned aerial vehicle (UAV) technology, and in particular to a radial speed regulation obstacle avoidance mid-course guidance method for a multi-rotor UAV. Background Technology

[0002] With the widespread application of drone technology in civilian and public safety fields, multi-rotor drones face the dual challenges of complex obstacle environments and multi-stage task coordination when performing target approach and obstacle avoidance missions in urban low-altitude environments. Existing technologies are mainly divided into two categories: Firstly, general path planning and local obstacle avoidance methods (such as the A* algorithm and the dynamic window method DWA) have the core objective of guiding the UAV safely to the preset waypoint. They lack consideration for the tracking conditions in the subsequent terminal guidance phase and cannot actively shape the initial motion state that is conducive to terminal guidance takeover; for example, a low line-of-sight angular rate and a stable approach speed.

[0003] Secondly, while proportional guidance methods based on line-of-sight angular rate suppression can effectively drive UAVs to converge toward the target, they typically treat radial approach velocity as an uncontrollable flight state, lacking active adjustment capabilities. In urban obstacle avoidance scenarios, when the UAV's approach velocity significantly decreases due to emergency avoidance or aerodynamic drag, existing methods cannot effectively compensate, resulting in low approach efficiency, a prolonged mid-to-terminal guidance handover time window, and even system oscillations at the handover moment.

[0004] In addition, existing solutions generally suffer from technical defects such as abrupt switching between guidance and obstacle avoidance modes, disconnect between control commands and underlying dynamic constraints, and discontinuous control states when switching guidance information sources in the mid-to-late stages. These deficiencies make it difficult to meet the requirements of efficient, safe, and smooth end-to-end approach missions in complex urban environments. Summary of the Invention

[0005] In view of this, embodiments of this application propose a radial speed regulation obstacle avoidance mid-course guidance method for multi-rotor UAVs, aiming to overcome the technical problems existing in the prior art when multi-rotor UAVs perform long-distance approach missions in complex obstacle environments, such as uncontrollable radial approach speed, disconnect between obstacle avoidance commands and dynamic constraints, and discontinuous transient control during mid- and terminal guidance switching. By combining active radial speed adjustment, acceleration domain obstacle avoidance, and mid- and terminal guidance state inheritance, efficient, safe, and smooth full-process approach guidance of multi-rotor UAVs in complex urban environments is achieved.

[0006] To achieve the above objectives, embodiments of this application propose a radial speed-regulating obstacle avoidance mid-course guidance method for a multi-rotor unmanned aerial vehicle (UAV), the method comprising the following steps: Acquire target reference position information, obstacle spatial distribution information, and UAV's own motion status information; Based on the target reference position information and the UAV's own motion state information, an original approach acceleration containing normal and radial components is generated; the normal component is used to suppress the line-of-sight angular rate, and the radial component is used to adjust the radial approach velocity to a preset desired value. Using the original approach acceleration as the reference center, a set of candidate maneuver commands is constructed in a three-dimensional acceleration space; each command in the set of candidate maneuver commands satisfies the physical acceleration constraints of the UAV. Based on the spatial distribution information of obstacles, an adaptive weighting factor is determined, and the adaptive weighting factor is applied to the original approach acceleration and the commands in the candidate maneuver command set to perform weighted fusion and generate multiple fused candidate acceleration commands. For each fusion candidate acceleration command, predict the continuous time trajectory within the prediction time window, and based on the spatial relationship between the continuous time trajectory and the obstacle, eliminate commands with collision risk to obtain a set of feasible candidate commands. The instructions in the feasible candidate instruction set are evaluated by a comprehensive cost function, and the instruction with the lowest cost is selected as the mid-stage guide for actual execution. When the drone enters the preset guidance mode switching zone, based on the current relative distance, the actual execution instructions for the middle guidance stage and the instructions for the final guidance stage are dynamically fused using a continuously time-varying weight function to generate the final guidance instructions.

[0007] The radial speed regulation obstacle avoidance mid-course guidance method for a multi-rotor UAV proposed in the embodiments of this application has the following technical advantages compared with the prior art: First, this application significantly improves the efficiency and certainty of UAVs in long-range approaching operations in complex obstacle environments.

[0008] Compared to existing proportional guidance methods (such as true proportional guidance), which mainly suppress line-of-sight angular rate but leave radial approach speed uncontrollable, this application generates a raw approach acceleration containing normal and radial components based on target reference position information and the UAV's own motion state information. The normal component is used to suppress line-of-sight angular rate, and the radial component is used to adjust the radial approach speed to a preset desired value. This allows the UAV to actively and quickly recover and maintain the preset desired approach speed after obstacle avoidance deceleration by introducing independent radial speed closed-loop control on top of normal guidance, thereby ensuring the high efficiency of long-range approach and creating a favorable initial motion state for subsequent terminal guidance.

[0009] Second, this application achieves a deep integration between guidance commands and the physical dynamics of the UAV, eliminating the risks of command execution delays and overload / exceedance.

[0010] Compared with traditional obstacle avoidance methods, this application constructs a set of candidate maneuver commands in a three-dimensional acceleration space based on the original approach acceleration. Each command in the candidate maneuver command set satisfies the physical acceleration constraints of the UAV. This allows the sampling space to be directly transferred from the velocity domain to the acceleration domain, thereby ensuring that the generated candidate commands satisfy the maximum acceleration and anisotropy constraints of the multi-rotor UAV. The downstream flight control can execute the commands directly without secondary conversion, fundamentally eliminating command conversion delay and ensuring the instantaneous accessibility and execution accuracy of the maneuver commands.

[0011] Third, this application achieves a smooth transition between guidance and obstacle avoidance modes, avoiding track oscillations and command jumps.

[0012] In existing technologies, guidance and obstacle avoidance modes often employ hard switching logic. Near the obstacle warning boundary, control commands may suddenly change, causing trajectory oscillations and affecting flight stability. This application determines an adaptive weighting factor based on obstacle spatial distribution information and applies this factor to the weighted fusion between the original approach acceleration and the commands in the candidate maneuver command set. By introducing an adaptive weight based on the obstacle distance field, which continuously changes with the distance between the UAV and the obstacle, guidance and obstacle avoidance commands are smoothly fused within the warning area. This eliminates command jumps and trajectory abrupt changes caused by mode switching, ensuring flight stability and safety.

[0013] Fourth, this application is able to construct a comprehensive decision-making mechanism that takes into account proximity trends, smoothness, and security, thus ensuring the achievement of mission objectives.

[0014] Since existing single obstacle avoidance or path planning methods often only focus on reaching the target point or avoiding obstacles, they cannot achieve a balance between multiple objectives (approach, smoothness, and safety). This application evaluates the instructions in the feasible candidate instruction set through a comprehensive cost function and selects the instruction with the lowest cost as the mid-stage guidance actual execution instruction output. By adopting a cost function that includes multiple indicators such as guidance tracking cost, motion smoothness cost, and safety margin cost, the system can comprehensively weigh the tracking of the original approach path, the smoothness of maneuvering, and the safe distance from obstacles when selecting the optimal obstacle avoidance instruction. This ensures that the UAV maintains the approach trend to the target to the maximum extent while safely avoiding obstacles, and avoids deviating too far from the target.

[0015] Fifth, this application completely eliminates system disturbances caused by transients during mid-to-late stage guidance switching, ensuring the smoothness and robustness of the entire process control.

[0016] The existing mid-course guidance (relying on external low-frequency data) and terminal guidance (relying on airborne high-frequency data) have fundamental differences in information sources and control objectives. Hard switching not only causes a step jump in the three-dimensional acceleration command, but also causes discontinuity in the state of the terminal guidance controller (especially the integral term), resulting in speed response lag and system oscillation. In contrast, when the UAV enters the preset guidance mode switching zone, this application dynamically integrates the actual mid-course guidance command and the terminal guidance command based on the current relative distance using a continuously time-varying weight function to generate the final guidance command. By setting a handover corridor and using a continuously time-varying weight function such as cosine, a smooth transition of commands is achieved.

[0017] Based on this, this application can overcome the technical problems of uncontrollable radial approach speed, disconnect between obstacle avoidance commands and dynamic constraints, and discontinuous transient control during mid- and terminal guidance switching when multi-rotor UAVs perform long-distance approach missions in complex obstacle environments. By combining active radial speed adjustment, acceleration domain obstacle avoidance, and mid- and terminal guidance state inheritance, this application achieves efficient, safe, and smooth full-process approach guidance for multi-rotor UAVs in complex urban environments.

[0018] Optionally, based on the target reference position information and the UAV's own motion state information, a raw approach acceleration including a normal component and a radial component is generated, including: calculating the unit vector and line-of-sight angular velocity of the UAV pointing towards the target; generating the normal component; wherein the magnitude of the normal component is proportional to the product of the line-of-sight angular velocity and the approach velocity, and its direction is perpendicular to the line-of-sight direction; generating the radial component; wherein the magnitude of the radial component is the result of proportional-integral control based on the deviation between the expected value and the actual approach velocity, and its direction is along the line-of-sight direction.

[0019] Optionally, an adaptive weighting factor is determined based on the spatial distribution information of obstacles, including: obtaining the distance between the current position of the UAV and the nearest obstacle; when the distance is greater than or equal to a preset warning radius, the weighting factor is zero; when the distance is less than the warning radius, the weighting factor is determined based on the warning radius and the distance; wherein, the weighting factor is negatively correlated with the distance.

[0020] Optionally, the adaptive weighting factor is calculated using the following formula: ; in, The preset warning radius, This represents the distance between the drone's current position and the nearest obstacle. Sensitivity coefficient; The formula for calculating the fused candidate acceleration commands is as follows: ; in, Indicates the instructions in the candidate maneuver instruction set; This represents the initial approach acceleration.

[0021] Optionally, for each fusion candidate acceleration command, predict the continuous time trajectory within the prediction time window, including: for each fusion candidate acceleration command In the prediction time window Continuous time trajectory prediction is performed within the time frame. The calculation formula for the continuous time trajectory is as follows: ; in, This represents the three-dimensional position vector of the UAV at the current moment; This represents the three-dimensional velocity vector of the drone at the current moment; For predicting time variables.

[0022] Optionally, based on the spatial relationship between the continuous time trajectory and obstacles, instructions with collision risks are eliminated to obtain a set of feasible candidate instructions, including: Through security detection function The command determines whether the distance between all points on the continuous time trajectory and the obstacle envelope is greater than a preset safety radius, thus identifying a collision risk. Specifically, if at any point within the prediction time window... Security detection function The calculation formula is as follows: ; in, Indicates the predicted location of the drone; indicates Obstacle boundary points; Indicates the safety radius.

[0023] Optionally, the instructions in the feasible candidate instruction set are evaluated, and the instruction with the lowest cost is selected as the output instruction for mid-stage guidance execution, including: Let the set of feasible candidate instructions be . ; The comprehensive cost function The expression is as follows: ; in, To guide the tracking cost, For the cost of smooth motion, The cost of a safety margin; , , These are the corresponding weighting coefficients; By global traversal The optimal solution with the minimum cost is calculated. And use it as the final intermediate guide for the actual execution instructions.

[0024] Optionally, when the UAV enters the preset guidance mode switching zone, based on the current relative distance, the actual execution commands for the mid-stage guidance and the final guidance commands are dynamically fused using a continuously time-varying weighting function to generate the final guidance command; wherein, the final guidance command... The calculation formulas include: ; in, This represents a continuously time-varying weighting function. This indicates that the middle section guides the actual execution of instructions. This indicates the final boot instruction.

[0025] Optionally, a continuously time-varying weighting function It is a cosine function, specifically: ; in, Indicates the width of the handover corridor; Indicates the distance to the entrance of the handover corridor. The relative position vector of the drone pointing at the target. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies of this application will be briefly introduced below. Obviously, the following drawings 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. The drawings described herein are only used to explain this application and are not intended to limit this application.

[0027] Figure 1 This is a flowchart of a mid-course guidance method for radial speed regulation obstacle avoidance of a multi-rotor unmanned aerial vehicle provided in one embodiment of this application; Figure 2 This is a schematic diagram of a radial speed regulation true proportional mid-section guide approach law provided in one embodiment of this application; Figure 3 This is a schematic diagram of an acceleration space dynamic window provided in one embodiment of this application; Figure 4 This is a schematic diagram of flexible fusion of obstacle distance field provided in one embodiment of this application; Figure 5 This is a schematic diagram of continuous-time collision prediction provided in one embodiment of this application; Figure 6This is a schematic diagram of a mid-to-late stage guidance handover corridor provided in one embodiment of this application; Figure 7 This is a schematic diagram of controller state inheritance provided in one embodiment of this application; Figure 8 This is a schematic diagram of the overall process of a mid-stage guidance method provided in one embodiment of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. Those skilled in the art will understand that many technical details have been presented in the embodiments of this application to facilitate better understanding. However, the technical solutions claimed in this application can be implemented even without these technical details and various variations and modifications based on the following embodiments. The division of the following embodiments is for ease of description and should not constitute any limitation on the specific implementation of this application. The following embodiments can be combined with and referenced by each other without contradiction.

[0029] With the widespread application of drone technology in civilian and public safety fields, multi-rotor drones face the dual challenges of complex obstacle environments and multi-stage mission coordination when performing target approach and obstacle avoidance tasks in urban low-altitude environments. Existing technologies are mainly divided into two categories: Firstly, general path planning and local obstacle avoidance methods (such as the A* algorithm and the dynamic window method DWA) have the core objective of guiding the UAV safely to the preset waypoint. They lack consideration for the tracking conditions in the subsequent terminal guidance phase and cannot actively shape the initial motion state that is conducive to terminal guidance takeover; for example, a low line-of-sight angular rate and a stable approach speed.

[0030] Secondly, while proportional guidance methods based on line-of-sight angular rate suppression can effectively drive UAVs to converge toward the target, they typically treat radial approach velocity as an uncontrollable flight state, lacking active adjustment capabilities. In urban obstacle avoidance scenarios, when the UAV's approach velocity significantly decreases due to emergency avoidance or aerodynamic drag, existing methods cannot effectively compensate, resulting in low approach efficiency, a prolonged mid-to-terminal guidance handover time window, and even system oscillations at the handover moment.

[0031] In addition, existing solutions generally suffer from technical defects such as abrupt switching between guidance and obstacle avoidance modes, disconnect between control commands and underlying dynamic constraints, and discontinuous control states when switching guidance information sources in the mid-to-late stages. These deficiencies make it difficult to meet the requirements of efficient, safe, and smooth end-to-end approach missions in complex urban environments.

[0032] In view of this, embodiments of this application propose a radial speed regulation obstacle avoidance mid-course guidance method for multi-rotor UAVs, aiming to overcome the technical problems existing in the prior art when multi-rotor UAVs perform long-distance approach missions in complex obstacle environments, such as uncontrollable radial approach speed, disconnect between obstacle avoidance commands and dynamic constraints, and discontinuous transient control during mid- and terminal guidance switching. By combining active radial speed adjustment, acceleration domain obstacle avoidance, and mid- and terminal guidance state inheritance, efficient, safe, and smooth full-process approach guidance of multi-rotor UAVs in complex urban environments is achieved.

[0033] One embodiment of this application proposes a mid-course guidance method for radial speed regulation obstacle avoidance of a multi-rotor UAV, applied to an electronic device, wherein the electronic device can be a terminal or a server. This embodiment and the following embodiments will use a server as an example for description. The implementation details of the mid-course guidance method for radial speed regulation obstacle avoidance of a multi-rotor UAV proposed in this embodiment will be described in detail below. The following implementation details are provided for ease of understanding and are not necessary for implementing this solution.

[0034] The specific process of the radial speed regulation obstacle avoidance mid-course guidance method for a multi-rotor UAV proposed in this embodiment can be described as follows: Figure 1 As shown, it includes: Step 101: Obtain target reference position information, obstacle spatial distribution information, and UAV's own motion status information.

[0035] In this embodiment, the target reference location information can be coarse target location information provided by an external collaborative platform (such as a ground station or other drones) via a data link. For example, the target location... Predicted target location or optional target velocity .

[0036] Of course, the coarse location information of the target can also be provided by external detection equipment, ground stations, mission planning results, multi-source fusion estimation or other existing methods, and this application embodiment does not impose specific limitations on this.

[0037] The spatial distribution information of obstacles can come from pre-installed high-precision 3D city maps, building envelope models, 3D occupancy grids, or obstacle distance fields that are detected and constructed in real time by airborne sensors (such as lidar and vision cameras).

[0038] The drone's own motion status information can include the drone's current position. ,speed Current acceleration or previous guidance instructions, etc.

[0039] The output of this method is the actual acceleration or overload command output during the mid-course guidance phase, as well as the merged mid-to-terminal guidance command within the handover corridor. This command can be converted by the downstream flight control system into attitude, thrust, angular velocity, or motor control quantities, as shown in Table 1.

[0040]

[0041] It should be noted that the contents of Table 1 can be found in the detailed descriptions of steps 101 to 107 in this implementation.

[0042] Step 102: Based on the target reference position information and the UAV's own motion state information, generate the original approach acceleration, which includes normal and radial components.

[0043] The normal component is used to suppress the line-of-sight angular rate, while the radial component is used to adjust the radial approach speed to a preset desired value.

[0044] like Figure 2 As shown, the core of this step lies in using a true proportional guidance law with active radial velocity adjustment capability to generate the initial approach acceleration during the mid-course guidance phase. The figure illustrates the process by which the normal true proportional guidance component and the radial velocity adjustment component jointly generate the initial approach acceleration for mid-course guidance.

[0045] In one possible embodiment, based on the target reference position information and the UAV's own motion state information, a raw approach acceleration including normal and radial components is generated, including: calculating the unit vector of the line of sight of the UAV pointing towards the target and the line of sight angular velocity; generating the normal component; generating the radial component.

[0046] The magnitude of the normal component is proportional to the product of the line-of-sight angular rate and the approach speed, and its direction is perpendicular to the line-of-sight direction; the magnitude of the radial component is the result of proportional-integral control based on the deviation between the desired value and the actual approach speed, and its direction is along the line-of-sight direction.

[0047] For example, the target could be a building, a crane waiting to approach, or other moving objects such as aircraft.

[0048] First, based on the obtained target location and drone location Calculate the relative position vector of the UAV pointing at the target. and its unit line-of-sight vector .

[0049] ; ; in, To prevent positive numbers with excessively small denominators, This is the relative position vector of the UAV pointing towards the target.

[0050] Then calculate the approach speed and line-of-sight angular rate .

[0051] For example, approach speed The calculation formula is as follows: ; Specifically, the approach speed is defined. Let be the relative approach speed between the UAV and the target along the line of sight, and let be the line-of-sight angular velocity. . The unit vector representing the line of sight, by Normalized, we obtain a value representing the unit direction from the UAV to the target, which is used for subsequent directional decomposition of the normal guidance component and the radial velocity component.

[0052] To enable rapid approach to the target during the mid-course guidance phase, and to recover the desired approach speed after speed decay due to obstacle avoidance braking or aerodynamic drag, this application employs a true proportional mid-course guidance approach law with radial adjustment capability. This approach law includes a normal true proportional guidance component and a radial speed adjustment component. The normal component is used to suppress line-of-sight angular velocity, enabling the UAV to quickly form an approach heading towards the target; the radial component is used to adjust the approach speed, allowing the UAV to maintain the preset desired approach speed during the mid-course guidance phase.

[0053] In other words, this application does not rely on dynamic speed reference, but uses a fixed or preset mid-range guidance reference speed to enable the UAV to actively recover its approach speed after the approach speed decreases due to obstacle avoidance braking, obstacle maneuvering, or aerodynamic drag.

[0054] The initial approach acceleration is composed of the superposition of the normal and radial components:

[0055] The normal component is used to suppress the line-of-sight angular rate, enabling the UAV to quickly form an approach heading towards the target. Its calculation formula is: ; in, For mid-range guidance and navigation ratio, This represents the low-pass filtering result of the line-of-sight angular velocity or angular rate estimation. The unit vector in the direction of the line of sight normal is specifically... .

[0056] Understandably, this normal component ensures that the drone can quickly converge to the heading toward the target.

[0057] The radial component is used to actively adjust and maintain the preset desired approach speed. It employs a proportional-integral (PI) control structure, and the calculation formula is as follows: ; in, To guide the expected approach speed in the middle section, and These are the proportional gain and integral gain of the mid-section guided radial speed controller, respectively.

[0058] Understandably, this radial component causes the UAV's approach speed to be affected by obstacle avoidance braking, obstacle maneuvering, or aerodynamic drag. During descent, it can actively and quickly recover to the desired speed, thereby ensuring high efficiency in long-distance approach.

[0059] As an example, the mid-section guidance expects to approach speed. It can be set to a fixed cruising speed. That is, a preset constant value. (The value can be...) In a head-on approach scenario where the target's flight speed is 80% to 150% of its maximum speed. It can be less than the target's flight speed, in a tail-chase approach scenario. The speed must be greater than the target's flight speed. For example, if the target is a civilian quadcopter drone, its typical speed is 15 m / s. The range should be 12~23 m / s. This fixed reference speed is used to ensure that the UAV maintains a stable approach to the target during the mid-course guidance phase, and actively restores the approach speed through the radial speed controller after the approach speed decreases due to obstacle avoidance braking, obstacle maneuvering, or aerodynamic drag.

[0060] ; For the final desired speed, The reference speed is used during normal mid-course guidance under normal circumstances. Values When there are obstacle avoidance and mission requirements, based on Make dynamic adjustments.

[0061] As another example It can also be dynamically adjusted based on obstacle risk, handover distance, or flight mission status. The definition is as follows: ; in, To ensure a safe approach speed limit, a preset constant parameter (typically 5-10 m / s) is used. When the drone enters a high-obstacle-risk area, the speed is dynamically adjusted. restrained The following ensures that the drone has sufficient braking distance and maneuver response time in confined spaces. This can be determined by obstacle risk, distance field weights, or task state. For example, when obstacle risk is low, Increasing the cruising approach speed can be advantageous; when entering an obstacle warning zone or needing to maneuver around obstacles. The threshold can be appropriately lowered to enhance the safety margin; when entering the handover area, The relative motion should be reduced to a gentler state during the final guidance phase; when in the rapid approach mission phase, A relatively high speed should be maintained when performing high-maneuverability target tracking missions. The speed should be reduced to reserve more normal overload for tracking.

[0062] It should be noted that the above-mentioned dynamic reference speed form is an optional implementation method and does not limit this application to adopt dynamic speed reference.

[0063] Optionally, the mid-course guidance desired approach speed can be a fixed cruise approach speed, or it can be adaptively adjusted according to obstacle risk, remaining distance, handover conditions, or target maneuver intensity.

[0064] Understandably, compared to approach methods that only output normal proportional guidance commands, this radial speed-regulating true proportional approach law can actively restore the approach speed when the approach speed decreases due to obstacle avoidance braking, obstacle maneuvering, or aerodynamic drag, while maintaining the characteristic of true proportional guidance to quickly suppress line-of-sight rotation.

[0065] Optionally, the radial speed controller in the true proportional approximation law of radial speed regulation may also be a PI, PID, sliding mode, active disturbance rejection, model predictive control or a speed controller with a disturbance observer. The embodiments of this application do not impose specific limitations on this.

[0066] Step 103: Using the original approach acceleration as the reference center, construct a set of candidate maneuver commands in the three-dimensional acceleration space.

[0067] Each instruction in the candidate maneuver instruction set satisfies the physical acceleration constraints of the UAV.

[0068] For example, the candidate maneuver command set is obtained by discrete sampling in three-dimensional space with the original approach acceleration as the center and a preset step size, and the magnitude of each candidate command does not exceed the maximum acceleration value of the UAV in the corresponding axis.

[0069] Understandably, unlike traditional velocity space dynamic windows, this application uses a mid-range guiding original approach acceleration generated around a radial speed-regulating true proportional approach law within the acceleration space. Construct a candidate maneuver set. This allows candidate commands to directly satisfy the maximum acceleration or maximum overload constraints of the multi-rotor UAV, making it easier for downstream flight control to execute.

[0070] like Figure 3 As shown, the innovation of this step lies in transferring the traditional velocity-space dynamic window method to the acceleration space. The figure illustrates the construction of a candidate acceleration set centered on the mid-section guided initial approach acceleration, which is constrained by the maximum acceleration or overload boundary.

[0071] Specifically, the original approach acceleration generated by the above embodiments Using this as the sampling center, discrete sampling is performed in the surrounding three-dimensional acceleration space at certain step sizes to generate a set of candidate maneuver commands. : ; in, This represents the acceleration search space, which is the set of all candidate acceleration vectors. This set is defined by... The sampling center is used to generate discrete samples in the three-dimensional acceleration space at a certain step size; For search space The k-th candidate acceleration vector (three-dimensional vector) in the vector. Each All of them will be sent to the subsequent collision prediction module and cost function for evaluation and screening. The three-dimensional acceleration vector, which represents the original approach acceleration guided by the mid-section and output by the radial speed regulation true proportional approach law, is the sampling center (i.e., the reference command) of the acceleration search space. The acceleration sampling step sizes in the x, y, and z axes determine the resolution of the search space. Smaller step sizes result in denser candidate instructions and higher computational cost; larger step sizes result in lower accuracy. Typical values ​​range from 0.2 to 1.0 m / s². These are the unit direction vectors of the x, y, and z axes of an inertial coordinate system (such as the NED North-East-Earth coordinate system), used to project scalar sampling increments into three-dimensional space. This is a set of constraints for the maximum acceleration (or maximum overload) of a multi-rotor UAV in each axial direction. Because the dynamic characteristics of a multi-rotor differ in the horizontal and vertical directions, these constraints are typically anisotropic, for example: Horizontal direction (x, y axes): =10~15m / s²; Vertical direction (z-axis): =5~8m / s².

[0072] By sampling directly in the acceleration space, all candidate commands are guaranteed. All of them meet the physical overload limits of UAVs, eliminating the lag and unreachability problems that may be caused by the conversion of speed commands to acceleration commands in traditional methods.

[0073] To further ensure command smoothness, an acceleration rate of change constraint can be added: ; in, Output the command from the previous moment. The upper limit of the allowable rate of change of acceleration, To control the cycle.

[0074] Optionally, the acceleration sampling method can be a regular grid, polar coordinate sampling, random sampling, low-difference sequence sampling, local rolling sampling based on the optimal solution of the previous time step, or optimization solution.

[0075] Step 104: Based on the spatial distribution information of obstacles, determine the adaptive weighting factor, and apply the adaptive weighting factor to the original approach acceleration and the commands in the candidate maneuver command set to perform weighted fusion and generate multiple fused candidate acceleration commands.

[0076] In one possible embodiment, determining an adaptive weighting factor based on obstacle spatial distribution information includes: obtaining the distance between the current position of the UAV and the nearest obstacle; when the distance is greater than or equal to a preset warning radius, the weighting factor is zero; when the distance is less than the warning radius, the weighting factor is determined based on the warning radius and the distance.

[0077] Among them, the weighting factor is negatively correlated with the distance.

[0078] like Figure 4 As shown, this step aims to achieve a flexible fusion of guidance mode and obstacle avoidance mode, avoiding sudden changes in instructions caused by hard switching. Figure 4 The demonstration shows the construction of a candidate acceleration set centered on the initial approach acceleration guided by the mid-section, constrained by the maximum acceleration or overload boundary. First, based on obstacle spatial information, the distance from the UAV's current position or predicted trajectory to the nearest obstacle surface is calculated. .

[0079] For example, obstacles can be represented as spheres, cylinders, cuboids, building envelopes, polyhedra, three-dimensional occupancy grids, or distance fields. An adaptive weighting factor based on the distance field is pre-introduced before performing collision detection on the sampled acceleration.

[0080] The formula for calculating the adaptive weighting factor is as follows: ; in, The preset warning radius, This represents the distance between the drone's current position and the nearest obstacle. Sensitivity coefficient; This weighting factor is used to assign weight to each sampling instruction within the search space. With current guidance instructions Smoothing weighting is performed to obtain the fused acceleration command used for subsequent prediction and verification. : ; in, Indicates the instructions in the candidate maneuver instruction set; This represents the initial approach acceleration.

[0081] Understandably, this fusion approach ensures a seamless transition of control commands from pure guidance commands to pure obstacle avoidance commands near obstacle warning boundaries, greatly improving flight smoothness and safety.

[0082] Optionally, the adaptive weighting factor based on the distance field can also be replaced by an exponential function, a cosine function, a hyperbolic tangent function, a logistic function, a piecewise polynomial function, or a discrete weight based on the risk level. The embodiments of this application do not impose specific limitations on this.

[0083] Step 105: For each fusion candidate acceleration command, predict the continuous time trajectory within the prediction time window, and based on the spatial relationship between the continuous time trajectory and the obstacle, eliminate commands with collision risk to obtain a set of feasible candidate commands.

[0084] like Figure 5 As shown in the figure, this diagram illustrates how candidate fused accelerations generate predicted trajectories within the prediction time window and perform collision detection with the obstacle safety envelope. This step involves processing each candidate fused acceleration command... Conduct a collision risk assessment. Within the prediction time window. Inside, assume the drone accelerates at a constant speed. Motion, and continuous time prediction of its future trajectory.

[0085] In one possible embodiment, for each fusion candidate acceleration command, predicting a continuous time trajectory within a prediction time window includes: For each fusion candidate acceleration command In the prediction time window Continuous time trajectory prediction is performed within the time frame. The calculation formula for the continuous time trajectory is as follows: ; in, This represents the three-dimensional position vector of the UAV at the current moment; This represents the three-dimensional velocity vector of the drone at the current moment; For predicting time variables.

[0086] Then, define the security detection function. If, within the entire prediction time window, any point on the predicted trajectory of the drone... With any obstacle surface point The distances are all greater than the preset safety radius. If the instruction is deemed safe, then the instruction is considered safe.

[0087] In one possible embodiment, based on the spatial relationship between the continuous time trajectory and obstacles, instructions with collision risks are eliminated to obtain a set of feasible candidate instructions, including: Through security detection function The command determines whether the distance between all points on the continuous time trajectory and the obstacle envelope is greater than a preset safety radius, thus identifying a collision risk. Specifically, if at any point within the prediction time window... Security detection function The calculation formula is as follows: .

[0088] For example, the safety radius The typical value is 2~5m, which consists of the following parts: the envelope radius of the UAV body is about 0.3~0.8m (depending on the UAV wheelbase), the positioning and perception error margin is about 0.5~1.5m, and the safety buffer distance is about 1~3m.

[0089] Specific example: For a medium-sized multi-rotor drone with a wheelbase of 0.5m, the following can be used when flying between urban buildings: The value is 3, which means the body is 0.5m, the error is 1.0m, and the buffer is 1.5m.

[0090] Alternatively, collision prediction can be replaced by velocity barrier method, occupied grid ray detection, ellipsoidal safe domain detection, control barrier function constraint, quadratic programming constraint or model predictive control constraint.

[0091] Step 106: Evaluate the instructions in the feasible candidate instruction set through a comprehensive cost function, and select the instruction with the lowest cost as the mid-stage guidance actual execution instruction output.

[0092] Understandably, all the fusion candidate acceleration commands that passed the security test constituted a set of feasible candidate commands.

[0093] In one possible embodiment, the instructions in the feasible candidate instruction set are evaluated, and the instruction with the lowest cost is selected as the mid-stage guidance instruction for actual execution output, including: Let the set of feasible candidate instructions be . ; In the set of feasible candidate instructions In this process, a multi-index comprehensive cost function is minimized. To select the optimal instruction : The comprehensive cost function The expression is as follows: ; in, The cost of guiding and tracking is used to evaluate whether the obstacle avoidance maneuver trajectory conforms to the original guiding path to the greatest extent possible, avoiding excessive deviation from the target; The motion smoothing cost is used to evaluate the current instruction compared to the previous instruction. The rate of change between them ensures the smoothness of maneuvering; The safety margin cost is used to assess the proximity of the predicted trajectory's endpoint to obstacles; the greater the distance, the lower the cost. , , These are the corresponding weighting coefficients; For the predicted trajectory corresponding to the candidate acceleration at the end of the time window, the distance from the UAV's predicted position to the nearest obstacle envelope surface is given.

[0094] By global traversal The optimal solution with the minimum cost is calculated. It then issues the final mid-course guidance command to the flight control system.

[0095] Step 107: When the UAV enters the preset guidance mode switching zone, based on the current relative distance, the actual execution instructions of the mid-stage guidance and the instructions of the final stage guidance are dynamically fused through a continuous time-varying weight function to generate the final guidance instructions.

[0096] As the drone gradually approaches the target, the relative distance reaches the maximum reliable detection range of the onboard sensors. At this time, the system needs to switch from mid-stage guidance to terminal guidance. Because there are inherent deviations in coordinate system alignment and observation accuracy between the external low-frequency guidance data and the high-frequency data from airborne sensors, a hard switch can cause a step change in the calculated line-of-sight angular rate, leading to abrupt changes in control system commands and aircraft oscillations. To ensure the smooth intervention of the omnidirectional overload constraint strategy, this application designs a dynamic weighted fusion strategy.

[0097] like Figure 6 As shown, Figure 6The demonstration shows how mid-stage guidance commands are gradually transitioned to terminal guidance commands via continuous weighting within the handover corridor. When the UAV approaches the target and the relative distance ||r|| enters the guidance mode switching zone, a smooth handover mechanism is activated. The initial distance of the handover corridor can be equal to the maximum reliable detection range of the airborne sensors. The width of the handover process is defined as... The handover starting distance is The handover ends at a distance of [distance missing]. .

[0098] In one possible embodiment, when the drone enters a preset guidance mode switching zone ( At that time, based on the current relative distance, a continuously time-varying weighting function guides the actual execution of instructions in the middle section. With the final guidance command Dynamic fusion is performed to generate the final boot instructions; among which, the final boot instructions The calculation formulas include: ; in, This represents a continuously time-varying weighting function. This indicates that the middle section guides the actual execution of instructions. This indicates the final boot instruction.

[0099] In one possible embodiment, a continuously time-varying weighting function The function is a cosine function, which smoothly and monotonically increases from 0 to 1 as the relative distance decreases, specifically: ; in, Indicates the width of the handover corridor; Indicates the distance to the entrance of the handover corridor. The relative position vector of the drone pointing at the target.

[0100] In addition, to prevent the integral term of the radial velocity controller from accumulating from zero during the final guidance stage, which would lead to response lag, this method also introduces a controller state inheritance mechanism.

[0101] For example, the continuous time-varying weight function can also be replaced by a cosine function with a cubic polynomial, a quintic polynomial, a spline function, an exponential function, or an adaptive weight function based on the terminal guided confidence. The embodiments of this application do not impose specific limitations on this.

[0102] The radial control in the final stage employs a PI controller with an integral anti-saturation mechanism to eliminate steady-state errors caused by air resistance. This is to prevent errors at the transition moment (i.e., weighting). When the value reaches 1, the integral term of the final stage guide PI controller starts from zero and accumulates, resulting in a lag in speed response. Therefore, a state inheritance mechanism must be implemented.

[0103] like Figure 7 As shown, Figure 7 This demonstrates how the integral state of the mid-stage guide radial velocity controller is transferred to the terminal guide radial velocity controller during handover. At the instant of entering the handover corridor, the initial integral value of the terminal guide radial PI controller is forcibly inherited from the stable integral value of the mid-stage guide velocity controller at the previous moment. ; in, The handover start moment is the moment when the relative distance R between the UAV and the target first decreases to the distance to the handover corridor entrance. At this moment, the state inheritance operation is triggered.

[0104] in, For the speed tracking error of the final stage guide radial speed controller, This refers to the speed tracking error of the mid-section guided radial speed controller.

[0105]

[0106] Understandably, the significance of state inheritance is that the integral term of the final guidance controller does not accumulate from zero, but starts from the steady-state integral value already reached in the mid-stage guidance, thereby avoiding radial velocity response lag and overshoot caused by integral restart at the moment of handover.

[0107] Optionally, state inheritance can be extended from integral term inheritance to filter state, velocity estimation state, target tracking state, overload allocation state, or flight control feedforward state inheritance.

[0108] Through the above dynamic handover strategy, the drone can leave the handover corridor ( After that, the system will smoothly transition to the end-stage guidance spatiotemporal coupling dynamic overload allocation mode, effectively avoiding transient disturbances caused by information source switching and ensuring the accuracy of the end-stage approach.

[0109] In one possible implementation, if the available markers for the final guidance phase are lost during the handover process, the target temporarily fails to be stably tracked and identified, or the obstacle risk increases again, the handover weight can be frozen, reduced, or reset. Maintain the hybrid mode or revert to the mid-range guided obstacle avoidance mode. If all candidate accelerations pose a collision risk, conservative strategies such as braking, hovering, ascending, bypassing, or waiting safely can be employed.

[0110] like Figure 8 As shown, Figure 8This is a schematic diagram of the overall process of the mid-course guidance method, showing the relationship between target coarse position information, obstacle spatial information, UAV status input, and radial speed regulation true proportional approach, acceleration space obstacle avoidance, continuous prediction, cost optimization, and mid-to-final handover. First, the system acquires external input information, including target coarse position information and UAV status, while also accessing environmental perception data, specifically including obstacle spatial distribution information (ground / air / fixed field), and flight and handover altitudes. Then, it enters the core algorithm processing stage. The system first performs high-proportion approach trajectory filtering on the target information to generate a smooth reference path. Subsequently, a dynamic window acceleration space is constructed based on the UAV's current dynamic capabilities. This space strictly limits the range of acceleration commands that the aircraft can execute under current physical conditions. Next, the algorithm performs risk assessment and prediction. Through the collision probability and loop prediction module, combined with obstacle distribution information, it quantifies flight risks and predicts potential motion trends. After generating candidate commands, the system uses a cost function optimization module to comprehensively evaluate possible maneuver commands. This module weighs multiple indicators such as tracking accuracy, energy consumption, and safety, selecting the optimal solution from numerous candidate schemes. Ultimately, the system transforms the optimized results into mid-to-final guidance handover commands, thereby generating smooth control commands that meet physical constraints and safety requirements, enabling the UAV to fly autonomously and safely in complex environments.

[0111] The radial speed regulation obstacle avoidance mid-course guidance method for a multi-rotor UAV proposed in the embodiments of this application has the following technical advantages compared with the prior art: First, this application significantly improves the efficiency and certainty of UAVs in long-range approaching operations in complex obstacle environments.

[0112] Compared to existing proportional guidance methods (such as true proportional guidance), which mainly suppress line-of-sight angular rate but leave radial approach speed uncontrollable, this application generates a raw approach acceleration containing normal and radial components based on target reference position information and the UAV's own motion state information. The normal component is used to suppress line-of-sight angular rate, and the radial component is used to adjust the radial approach speed to a preset desired value. This allows the UAV to actively and quickly recover and maintain the preset desired approach speed after obstacle avoidance deceleration by introducing independent radial speed closed-loop control on top of normal guidance, thereby ensuring the high efficiency of long-range approach and creating a favorable initial motion state for subsequent terminal guidance.

[0113] Second, this application achieves a deep integration between guidance commands and the physical dynamics of the UAV, eliminating the risks of command execution delays and overload / exceedance.

[0114] Compared with traditional obstacle avoidance methods, this application constructs a set of candidate maneuver commands in a three-dimensional acceleration space based on the original approach acceleration. Each command in the candidate maneuver command set satisfies the physical acceleration constraints of the UAV. This allows the sampling space to be directly transferred from the velocity domain to the acceleration domain, thereby ensuring that the generated candidate commands satisfy the maximum acceleration and anisotropy constraints of the multi-rotor UAV. The downstream flight control can execute the commands directly without secondary conversion, fundamentally eliminating command conversion delay and ensuring the instantaneous accessibility and execution accuracy of the maneuver commands.

[0115] Third, this application achieves a smooth transition between guidance and obstacle avoidance modes, avoiding track oscillations and command jumps.

[0116] In existing technologies, guidance and obstacle avoidance modes often employ hard switching logic. Near the obstacle warning boundary, control commands may suddenly change, causing trajectory oscillations and affecting flight stability. This application determines an adaptive weighting factor based on obstacle spatial distribution information and applies this factor to the weighted fusion between the original approach acceleration and the commands in the candidate maneuver command set. By introducing an adaptive weight based on the obstacle distance field, which continuously changes with the distance between the UAV and the obstacle, guidance and obstacle avoidance commands are smoothly fused within the warning area. This eliminates command jumps and trajectory abrupt changes caused by mode switching, ensuring flight stability and safety.

[0117] Fourth, this application is able to construct a comprehensive decision-making mechanism that takes into account proximity trends, smoothness, and security, thus ensuring the achievement of mission objectives.

[0118] Since existing single obstacle avoidance or path planning methods often only focus on reaching the target point or avoiding obstacles, they cannot achieve a balance between multiple objectives (approach, smoothness, and safety). This application evaluates the instructions in the feasible candidate instruction set through a comprehensive cost function and selects the instruction with the lowest cost as the mid-stage guidance actual execution instruction output. By adopting a cost function that includes multiple indicators such as guidance tracking cost, motion smoothness cost, and safety margin cost, the system can comprehensively weigh the tracking of the original approach path, the smoothness of maneuvering, and the safe distance from obstacles when selecting the optimal obstacle avoidance instruction. This ensures that the UAV maintains the approach trend to the target to the maximum extent while safely avoiding obstacles, and avoids deviating too far from the target.

[0119] Fifth, this application completely eliminates system disturbances caused by transients during mid-to-late stage guidance switching, ensuring the smoothness and robustness of the entire process control.

[0120] The existing mid-course guidance (relying on external low-frequency data) and terminal guidance (relying on airborne high-frequency data) have fundamental differences in information sources and control objectives. Hard switching not only causes a step jump in the three-dimensional acceleration command, but also causes discontinuity in the state of the terminal guidance controller (especially the integral term), resulting in speed response lag and system oscillation. In contrast, when the UAV enters the preset guidance mode switching zone, this application dynamically integrates the actual mid-course guidance command and the terminal guidance command based on the current relative distance using a continuously time-varying weight function to generate the final guidance command. By setting a handover corridor and using a continuously time-varying weight function such as cosine, a smooth transition of commands is achieved.

[0121] Based on this, this application can overcome the technical problems of uncontrollable radial approach speed, disconnect between obstacle avoidance commands and dynamic constraints, and discontinuous transient control during mid- and terminal guidance switching when multi-rotor UAVs perform long-distance approach missions in complex obstacle environments. By combining active radial speed adjustment, acceleration domain obstacle avoidance, and mid- and terminal guidance state inheritance, this application achieves efficient, safe, and smooth full-process approach guidance for multi-rotor UAVs in complex urban environments.

[0122] The steps described above are for clarity only. In implementation, they can be combined into one step, or some steps can be broken down into multiple steps, as long as they involve the same logical relationship, they are all within the scope of protection of this application. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, without changing the core design of the algorithm and process, are also within the scope of protection of this application.

[0123] Those skilled in the art will understand that the above embodiments are specific implementations of this application, and in practical applications, various changes can be made in form and detail without departing from the spirit and scope of this application. For those skilled in the art, several improvements and modifications can be made without departing from the principles of this application, and these improvements and modifications are also considered to be within the scope of protection of this application.

Claims

1. A radial speed regulation obstacle avoidance mid-course guidance method for a multi-rotor unmanned aerial vehicle, characterized in that, The method includes: Acquire target reference position information, obstacle spatial distribution information, and UAV's own motion status information; Based on the target reference position information and the UAV's own motion state information, an original approach acceleration containing normal and radial components is generated; the normal component is used to suppress the line-of-sight angular rate, and the radial component is used to adjust the radial approach velocity to a preset desired value. Using the original approach acceleration as the reference center, a set of candidate maneuver commands is constructed in a three-dimensional acceleration space; each command in the set of candidate maneuver commands satisfies the physical acceleration constraints of the UAV. Based on the spatial distribution information of obstacles, an adaptive weighting factor is determined, and the adaptive weighting factor is applied to the original approach acceleration and the commands in the candidate maneuver command set to perform weighted fusion and generate multiple fused candidate acceleration commands. For each fusion candidate acceleration command, predict the continuous time trajectory within the prediction time window, and based on the spatial relationship between the continuous time trajectory and the obstacle, eliminate commands with collision risk to obtain a set of feasible candidate commands. The instructions in the feasible candidate instruction set are evaluated by a comprehensive cost function, and the instruction with the lowest cost is selected as the mid-stage guide for actual execution. When the drone enters the preset guidance mode switching zone, based on the current relative distance, the actual execution instructions for the middle guidance stage and the instructions for the final guidance stage are dynamically fused using a continuously time-varying weight function to generate the final guidance instructions.

2. The method according to claim 1, characterized in that, The process of generating the original approach acceleration, which includes normal and radial components, based on the target reference position information and the UAV's own motion state information, includes: Calculate the unit vector and line-of-sight angular velocity of the UAV pointing at the target; Generate a normal component; wherein the magnitude of the normal component is proportional to the product of the line-of-sight angular velocity and the approach velocity, and its direction is perpendicular to the line-of-sight direction; Generate a radial component; wherein the magnitude of the radial component is the result of proportional-integral control based on the deviation between the desired value and the actual approach speed, and its direction is along the line of sight.

3. The method according to claim 1, characterized in that, The candidate maneuver command set is obtained by discrete sampling in three-dimensional space with the original approach acceleration as the center and a preset step size, and the magnitude of each candidate command does not exceed the maximum acceleration value of the UAV in the corresponding axis.

4. The method according to claim 1, characterized in that, The determination of adaptive weighting factors based on obstacle spatial distribution information includes: Obtain the distance between the drone's current position and the nearest obstacle; When the distance is greater than or equal to the preset warning radius, the weight factor is zero; When the distance is less than the warning radius, a weighting factor is determined based on the warning radius and the distance; the weighting factor is negatively correlated with the distance.

5. The method according to claim 4, characterized in that, The formula for calculating the adaptive weighting factor is as follows: ； in, The preset warning radius, This represents the distance between the drone's current position and the nearest obstacle. Sensitivity coefficient; The formula for calculating the fused candidate acceleration commands is as follows: ； in, Indicates the instructions in the candidate maneuver instruction set; This represents the initial approach acceleration.

6. The method according to claim 5, characterized in that, The step of predicting a continuous-time trajectory within a prediction time window for each fusion candidate acceleration command includes: For each fusion candidate acceleration command In the prediction time window Continuous time trajectory prediction is performed within the time frame. The calculation formula for the continuous time trajectory is as follows: ； in, This represents the three-dimensional position vector of the UAV at the current moment; This represents the three-dimensional velocity vector of the drone at the current moment; For predicting time variables.

7. The method according to claim 6, characterized in that, Based on the spatial relationship between the continuous time trajectory and obstacles, instructions with collision risks are eliminated, resulting in a set of feasible candidate instructions, including: Through security detection function The command determines whether the distance between all points on the continuous time trajectory and the obstacle envelope is greater than a preset safety radius, thus identifying a collision risk. Specifically, if at any point within the prediction time window... Security detection function The calculation formula is as follows: ； in, Indicates the predicted location of the drone; indicates Obstacle boundary points; Indicates the safety radius.

8. The method according to claim 7, characterized in that, The process of evaluating instructions in the feasible candidate instruction set and selecting the instruction with the lowest cost as the mid-stage guidance actual execution instruction output includes: Let the set of feasible candidate instructions be . ; The comprehensive cost function The expression is as follows: ； in, To guide the tracking cost, For the cost of smooth motion, The cost of a safety margin; , , These are the corresponding weighting coefficients; By global traversal The optimal solution with the minimum cost is calculated. And use it as the final intermediate guide for the actual execution instructions.

9. The method according to claim 8, characterized in that, When the UAV enters the preset guidance mode switching zone, based on the current relative distance, the actual execution commands for the middle guidance stage and the final guidance stage are dynamically fused using a continuously time-varying weighting function to generate the final guidance command; wherein, the final guidance command... The calculation formulas include: ； in, This represents a continuously time-varying weighting function. This indicates that the middle section guides the actual execution of instructions. This indicates the final boot instruction.

10. The method according to claim 9, characterized in that, The continuous time-varying weighting function It is a cosine function, specifically: ； in, Indicates the width of the handover corridor; Indicates the distance to the entrance of the handover corridor. The relative position vector of the drone pointing at the target.

Images