Unmanned aerial vehicle path planning method and system based on intelligent window, and storage medium

Abstract

The invention provides an unmanned aerial vehicle path planning method based on an intelligent window, and belongs to the technical field of intelligent logistics, and the method comprises the steps: building an obstacle map based on a target intelligent window, carrying out global initial path planning, carrying out local obstacle avoidance adjustment, carrying out real-time path correction, and carrying out landing control. The intelligent window is used as a distributed environment sensing node, a large-scale base station or a sensor array does not need to be additionally deployed, the network coverage range can be expanded according to needs, the deployment cost is reduced, meanwhile, installation and use of the intelligent window are promoted, and development of a smart city and a smart home is assisted; on the basis of global path planning and local obstacle avoidance adjustment, real-time path correction is carried out, and path planning failure caused by blind areas is effectively avoided; non-linear weight distribution is carried out through the fuzzy rule base, real-time path correction in a complex scene is realized, frequent re-planning is not needed for emergent obstacles, the calculation efficiency is guaranteed, the stability of the distribution track of the unmanned aerial vehicle is guaranteed, and the distribution safety of the unmanned aerial vehicle is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent logistics, and particularly relates to an unmanned aerial vehicle path planning method and system based on an intelligent window and a storage medium. BACKGROUND

[0002] With the rapid development of e-commerce and the logistics industry, people have higher requirements for the efficiency and service quality of logistics distribution. Traditional logistics distribution methods gradually reveal many limitations in the face of growing distribution needs. For example, in the case of urban traffic congestion, the driving speed of ground distribution vehicles is greatly reduced, the distribution time is prolonged, and the goods cannot be delivered to the customers in time. At the same time, with the development of intelligent logistics and unmanned aerial vehicle technology, under the driving of multiple factors such as increasing distribution demand, soaring labor costs, and complex service scenarios, unmanned aerial vehicles have gradually become an indispensable infrastructure in the modern logistics industry, helping the logistics industry to achieve leap-forward development. The continuous maturity of unmanned aerial vehicle technology brings new solutions to the logistics distribution field. Unmanned aerial vehicles have the advantages of strong mobility, no ground traffic restrictions, and the ability to quickly reach the destination, which can effectively make up for the shortcomings of traditional logistics distribution methods. However, to achieve efficient and reliable unmanned aerial vehicle logistics distribution, many challenges are faced.

[0003] Currently, a new unmanned aerial vehicle distribution method based on intelligent logistics cabinets is emerging (such as postal unmanned aerial vehicle delivery and express unmanned aerial vehicle express delivery), and logistics cabinet and unmanned aerial vehicle collaborative distribution as a "non-contact" type of distribution can ensure distribution efficiency, but for high-rise or disabled residents, they still need to go to the logistics cabinet to pick up the goods. The demand for delivery methods by unmanned aerial vehicles directly to the outside of the recipient's living environment has thus emerged, and the flight path of the unmanned aerial vehicle in the city needs to be planned.

[0004] Currently, the mainstream unmanned aerial vehicle path planning method mainly relies on on-board sensors (such as laser radars, visual cameras), satellite positioning systems (such as GPS), and pre-constructed environment maps. Such methods perform well in open environments, but in urban spaces or dynamic dense obstacle scenarios, the following shortcomings exist:

[0005] 1. Environmental perception limitations: The detection range and accuracy of on-board sensors are limited by hardware performance, especially in areas with severe obstructions (such as high-rise buildings, dense vegetation), which can easily create blind spots, leading to path planning failure.

[0006] 2. Insufficient dynamic adaptability: Traditional algorithms are mostly based on static environment assumptions, and when faced with sudden obstacles (such as moving vehicles, temporary construction facilities), frequent re-planning is required, resulting in low computational efficiency and poor trajectory stability.

[0007] 3. Strong dependence on communication and positioning: GPS signals are easily disturbed in cities, and positioning schemes that rely on external base stations have high deployment costs and limited coverage.

[0008] At the same time, in recent years, intelligent window technology has been applied in the fields of building energy saving, light and heat regulation, etc. as a new type of environmental perception and interaction carrier. The intelligent window can realize real-time perception of environmental light intensity, temperature, obstacle position and other information by integrating transparent photovoltaic, micro sensor array and dynamic light transmission adjustment module, and can be linked with external devices through wireless communication network. However, the existing technology has not explored the cooperative mechanism of intelligent window and unmanned aerial vehicle system, and the potential of the intelligent window as a distributed environmental perception node has not been tapped.

[0009] In addition, the existing unmanned aerial vehicle path planning system mostly adopts centralized control architecture, which is difficult to efficiently integrate scattered environmental data sources. Especially in the over-the-horizon flight or complex building group scene, the unmanned aerial vehicle lacks real-time response capability to local environmental dynamic changes, resulting in path deviation or collision risk.

[0010] Therefore, there is an urgent need for an unmanned aerial vehicle path planning method that integrates new environmental perception technology, dynamic data driving and adapts to complex scenes to improve its autonomy, safety and task execution efficiency. SUMMARY

[0011] To solve the above problems, the present application provides an unmanned aerial vehicle path planning method and system based on intelligent window and storage medium to solve the problems in the prior art.

[0012] In order to achieve the above invention purpose, the present application provides an unmanned aerial vehicle path planning method based on intelligent window, the method is as follows:

[0013] S1, constructing an obstacle map based on a target intelligent window;

[0014] S1-1, generating a positioning coordinate (x w ,y w ,z w ) as the target landing position of the unmanned aerial vehicle through the real-time position of the built-in positioning module installed in the target intelligent window and multi-sensor fusion;

[0015] S1-2, constructing a three-dimensional dynamic obstacle map by combining the BIM model of the positioning target intelligent window with the real-time environmental data obtained by the onboard radar and visual sensor of the unmanned aerial vehicle;

[0016] S2, global initial path planning;

[0017] generating a global initial path based on A * algorithm, and the cost function is as follows:

[0018] F(n)=α1·D(n)+β1·E(n)+γ1·O(n)

[0019] Where D(n) is the Euclidean distance from node n to the positioning coordinates, E(n) is the cumulative energy consumption from the starting point to node n, O(n) is the threat coefficient of obstacles around node n, and α1, β1, and γ1 are dynamic weight coefficients.

[0020] S3, Local obstacle avoidance adjustment;

[0021] Based on the global initial path and the dynamic window method, candidate trajectories are sampled in the velocity space (v,ω). Local obstacle avoidance adjustments are made by maximizing the objective function to obtain a local obstacle avoidance path, which is then used to replace the corresponding part of the global initial path to obtain the planned path; as follows:

[0022] G(v,ω)=α2·G nav (v,ω)+β2·C obs (v,ω)+γ2·S smooth (v,ω)

[0023] Where v is the linear velocity of the drone, ω is the angular velocity of the drone, and G nav (v,ω) represents the navigation gain term of the UAV, C obs (v,ω) represents the obstacle avoidance cost of the drone, S smooth (v,ω) represents the motion smoothness term of the UAV, and α2, β2, and γ2 are obstacle avoidance weight coefficients.

[0024] S4, Real-time Path Correction;

[0025] By using the real-time communication link between the target smart window and the drone, path deviations are corrected and dynamic weighting coefficients are adjusted.

[0026] S5, Landing Control;

[0027] Once the drone approaches the target smart window, the relative pose of the drone and the smart window is calculated using a fusion positioning method combining visual recognition and ultrasonic ranging, enabling a precise landing.

[0028] This invention constructs an obstacle map based on a target smart window, enabling three-dimensional environmental visualization of the drone delivery path and obstacles near the path, facilitating real-time monitoring of delivery progress by users. Meanwhile, the drone path planning method provided by this invention mainly includes global initial path planning before drone takeoff and delivery, as well as local obstacle avoidance adjustment and real-time path correction during flight, minimizing the occurrence of blind spots in the flight path that could cause path planning failure.

[0029] Furthermore, the positioning coordinates in step S1-1 are obtained by fusing the real-time position coordinates of the built-in positioning module with GPS coordinates and multi-sensor data, as follows:

[0030]

[0031] wherein P W is the real-time position coordinate of the built-in positioning module, P GPS is the GPS coordinate of the target positioning window, is the distance of the i-th reference object measured by the laser radar sensor, f(P w ) is a coordinate conversion function, and λ is a fusion weight.

[0032] The positioning coordinate in the application is the accurate coordinate of the end point of the UAV path, which is obtained by fusing the real-time position coordinate of the built-in positioning module, the GPS coordinate and the multi-sensor. The application breaks through the traditional fixed weight fusion method, greatly improves the accuracy of the end point positioning, effectively alleviates the navigation failure problem caused by GPS failure, and improves the applicability of the application to some special scenes in subsequent development.

[0033] Further, in step S2:

[0034]

[0035] wherein v t is the linear velocity when reaching the node n, ω t is the angular velocity when reaching the node n, and l is the moment of inertia of the UAV.

[0036]

[0037] wherein η is an attenuation factor, d k is the distance from the node n to the k-th obstacle, and r k is the influence radius of the k-th obstacle.

[0038] Further, the specific method of step S3 is as follows:

[0039] S3-1, sampling candidate trajectories in the velocity space (v, ω) based on the global initial path and the dynamic window method;

[0040] S3-2, constructing a maximization target function including a navigation gain term, an obstacle avoidance cost term and a motion smoothness term, and calculating the evaluation scores of each candidate trajectory respectively;

[0041] S3-3, selecting the candidate trajectory with the highest evaluation score as the local obstacle avoidance path.

[0042] The application optimizes the local path by the dynamic window method, significantly enhances the prediction accuracy and obstacle avoidance response speed of the UAV when facing sudden obstacles, maximally reduces the collision risk, and improves the safety of UAV distribution.

[0043] Further, the specific method of step S4 is as follows:

[0044] S4-1, discretize the planning path into a sequence of theoretical path points, and each theoretical path point includes latitude, longitude, height, and an allowable deviation threshold;

[0045] S4-2, the target intelligent window receives the state data stream of the unmanned aerial vehicle, and performs spatio-temporal alignment and confidence weighted fusion on the state data stream based on the Kalman filtering algorithm;

[0046] S4-3, based on the sequence of theoretical path points, dynamically adjust the lateral heading weight coefficient of the built-in path tracking controller of the target intelligent window based on the fuzzy inference rule base longitudinal speed weight coefficient k v height maintenance weight coefficient k h to meet:

[0047] and

[0048] At the same time, generate a composite control instruction containing a heading correction angle speed compensation amount Δv, height adjustment amount Δh;

[0049] S4-4, the real-time communication link encapsulates the above composite control instruction into a data packet and sends it to the flight control processor of the unmanned aerial vehicle for execution.

[0050] The present application discretizes the planning path into a sequence of theoretical path points containing parameters, and performs segmented calculation, which can effectively reduce the real-time calculation load, the flight path is smoother, the flight energy utilization rate is improved, the single-trip cost of unmanned aerial vehicle distribution is reduced, and the promotion and use of unmanned aerial vehicle distribution are facilitated.

[0051] Further, the relative pose in step S5 is calculated as follows:

[0052]

[0053] Δz=z ultra -z w ;

[0054] Where (x cam , y cam ) is the center coordinate of visual recognition, and z ultra is the ultrasonic ranging distance value.

[0055] Further, the navigation gain term in step S3-2 is the weighted sum of the lateral deviation Δd and the heading deviation Δθ of the trajectory endpoint and the global path, which is calculated as follows:

[0056] G nav (v,ω)=w1·e-|Δd| +w2·cos(Δθ)

[0057] w1 + w2 = 1;

[0058] The obstacle avoidance cost term is the minimum distance d between each point on the calculated trajectory and the nearest obstacle. min The reciprocal integral is calculated as follows:

[0059]

[0060] The motion smoothness term is used to calculate the rate of change of linear velocity. With the rate of change of angular velocity The quadratic form is calculated as follows:

[0061]

[0062] Furthermore, the obstacle avoidance weight coefficients α2, β2, and γ2 satisfy the following:

[0063] α2∈[0.5,0.8], β2∈[0.3,0.6], γ2∈[0.1,0.3].

[0064] This invention also provides a UAV path planning system based on a smart window for implementing the above method. The system includes:

[0065] The intelligent window has a built-in positioning module, path tracking controller, and path planning engine module. The positioning module is used to obtain the real-time position coordinates of the target intelligent window. The path tracking controller is used to provide composite control commands when the UAV performs real-time path correction. The path planning engine module is used to calculate and output the results when performing global initial path, local obstacle avoidance adjustment, real-time path correction, and landing control.

[0066] The drone is equipped with a multi-sensor module to acquire the drone's real-time position coordinates, flight speed and flight attitude data, as well as visual recognition during landing;

[0067] The communication link module is used to realize the data, command transmission and status synchronization between the smart window and the drone.

[0068] Furthermore, the smart window is also equipped with a retractable landing platform; the unfolded area of ​​the landing platform is matched to the size of the drone, as follows:

[0069]

[0070] Among them, L uav Δ represents the diagonal length of the drone, and Δ represents the safety margin.

[0071] The application further provides a computer storage medium storing program instructions for executing the above method.

[0072] Compared with the prior art, the application has the following beneficial effects:

[0073] 1. The intelligent window is used as a distributed environment sensing node, without the need of deploying large-scale base stations or sensor arrays, the network coverage can be expanded on demand, the deployment cost is reduced, the installation and use of the intelligent window are promoted, and the development of smart cities, smart homes and smart logistics is facilitated;

[0074] 2. On the basis of global path planning and local obstacle avoidance adjustment, real-time path correction is performed, so that path planning failure caused by the emergence of a blind area is effectively avoided;

[0075] 3. Nonlinear weight distribution is performed through a fuzzy rule base, real-time path correction in a complex scene is realized, frequent re-planning is not needed in the face of sudden obstacles, the calculation efficiency is guaranteed, the stability of the unmanned aerial vehicle delivery trajectory is ensured, and the delivery safety of the unmanned aerial vehicle is improved. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 The application is based on an unmanned aerial vehicle path planning method based on an intelligent window.

[0077] Figure 2 The application is based on an unmanned aerial vehicle path planning method based on an intelligent window. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical scheme and advantages of the application clearer, the application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0079] Example 1

[0080] As shown in the drawings, the present embodiment is an unmanned aerial vehicle path planning method based on an intelligent window, and the method is as follows: Figure 1 S1, constructing an obstacle map based on a target intelligent window;

[0081] S1-1, generating a positioning coordinate (x w ,y w ,z w ) as an unmanned aerial vehicle target landing position through real-time position of a built-in positioning module installed in the target intelligent window and multi-sensor fusion;

[0082]

[0083] ​The positioning coordinates are obtained by fusing the real-time position coordinates of the built-in positioning module, the GPS coordinates and multiple sensors, as follows:

[0084]

[0085] wherein P W is the real-time position coordinates of the built-in positioning module, P GPS is the GPS coordinates of the target positioning window, is the distance of the i-th reference object measured by the laser radar sensor, f(P w ) is a coordinate conversion function, and λ is a fusion weight;

[0086] S1-2, real-time environment data are acquired by the radar and the visual sensor on the unmanned aerial vehicle, and a three-dimensional dynamic obstacle map is constructed in combination with the BIM model in which the target intelligent window is positioned; the update frequency of the three-dimensional dynamic obstacle map can be positively correlated with the flight speed of the unmanned aerial vehicle, and the resolution can be adaptively adjusted according to the obstacle density;

[0087] S2, global initial path planning;

[0088] The global initial path is generated based on an A * algorithm, and the cost function is as follows:

[0089] F(n)=α1·D(n)+β1·E(n)+γ1·O(n)

[0090] wherein D(n) is the Euclidean distance from the node n to the positioning coordinates, E(n) is the cumulative energy consumption from the starting point to the node n, O(n) is the threat coefficient of the obstacles around the node n, and α1, β1 and γ1 are dynamic weight coefficients;

[0091] In step S2:

[0092]

[0093] wherein v t is the linear speed when reaching the node n, ω t is the angular speed when reaching the node n, and l is the moment of inertia of the unmanned aerial vehicle;

[0094]

[0095] wherein η is a decay factor, d k is the distance from the node n to the k-th obstacle, and r k is the influence radius of the k-th obstacle;

[0096] When the unmanned aerial vehicle encounters strong wind interference during flight, after detection, the path tracking controller built-in the intelligent window increases β1 to make real-time adjustment, so as to ensure that the actual path of the unmanned aerial vehicle delivery and the planned path error is less than 0.5 m;

[0097] S3, local obstacle avoidance adjustment;

[0098] Based on the global initial path and the dynamic window method, candidate trajectories are sampled in the speed space (v, ω), the local obstacle avoidance adjustment is obtained by maximizing the objective function, the corresponding part in the global initial path is replaced, and the planned path is obtained; as follows:

[0099] G(v, ω)=α2·G nav (v, ω)+β2·C obs (v, ω)+γ2·S smooth (v, ω)

[0100] Wherein, v is the linear velocity of the unmanned aerial vehicle, ω is the angular velocity of the unmanned aerial vehicle, G nav (v, ω) is the navigation gain term of the unmanned aerial vehicle, C obs (v, ω) is the obstacle avoidance cost term of the unmanned aerial vehicle, S smooth (v, ω) is the motion smoothness term of the unmanned aerial vehicle, and α2, β2, γ2 are obstacle avoidance weight coefficients.

[0101] The specific method is as follows:

[0102] S3-1, based on the global initial path and the dynamic window method, candidate trajectories are sampled in the speed space (v, ω);

[0103] S3-2, construct a maximization objective function containing navigation gain term, obstacle avoidance cost term and motion smoothness term, and calculate the evaluation score of each candidate trajectory;

[0104] The navigation gain term is the weighted sum of the lateral deviation Δd and the heading deviation Δθ of the trajectory endpoint and the global path, which is calculated as follows:

[0105] C nav (v, ω)=w1·e -|Δd| +w2·cos(Δθ)

[0106] w1+w2=1;

[0107] The obstacle avoidance cost term is the reciprocal integral of the minimum distance d min of each point of the trajectory and the nearest obstacle, which is calculated as follows:

[0108]

[0109] The motion smoothness term is the calculation of the linear velocity change rate with angular velocity rate of change a quadratic form of the angular velocity rate of change

[0110]

[0111] And the obstacle avoidance weight coefficients a2, b2, g2 satisfy the following:

[0112] a2∈[0.5, 0.8], b2∈[0.3, 0.6], b2∈[0.1, 0.3];

[0113] S3-3, select the candidate trajectory with the highest evaluation score as the local obstacle avoidance path;

[0114] S4, real-time path correction;

[0115] Through the real-time communication link between the target intelligent window and the unmanned aerial vehicle, the path deviation is corrected, and the dynamic weight coefficient is adjusted;

[0116] The specific method is as follows:

[0117] S4-1, discretize the planned path into a theoretical path point sequence, and each theoretical path point includes latitude, longitude, height, and an allowable deviation threshold;

[0118] S4-2, the target intelligent window receives the state data stream of the unmanned aerial vehicle, and performs spatio-temporal alignment and confidence weighted fusion on the state data stream based on the Kalman filtering algorithm;

[0119] S4-3, based on the theoretical path point sequence as a reference, dynamically adjust the lateral heading weight coefficient k the longitudinal speed weight coefficient k v , the height maintenance weight coefficient k h to meet:

[0120] And

[0121] At the same time, a composite control instruction containing the heading correction angle the speed compensation amount Av, and the height adjustment amount Ah is generated;

[0122] S4-4, the real-time communication link encapsulates the above composite control instruction into a data packet and sends it to the flight control processor of the unmanned aerial vehicle for execution;

[0123] S5, landing control;

[0124] When the unmanned aerial vehicle approaches the target intelligent window, based on the fusion positioning method of visual recognition and ultrasonic ranging, the relative pose of the unmanned aerial vehicle and the intelligent window is calculated, and accurate landing is completed;

[0125] The relative pose is calculated as follows:

[0126]

[0127] Δz = z ultra -z w ;

[0128] Where (x cam , y cam ) is the center coordinate of visual recognition, and z ultra is the ultrasonic ranging distance value.

[0129] Embodiment 2

[0130] This embodiment is based on the specific path planning scheme provided in Embodiment 1, and adds a multi-unmanned aerial vehicle cooperative operation control method. The path priority is distributed by negotiation, and the priority is determined by the remaining flight power of each unmanned aerial vehicle and the urgency of the distribution task. A collision-free path is planned through a space-time corridor model to ensure that the minimum safety distance between each unmanned aerial vehicle is always greater than 2m, avoiding conflicts between unmanned aerial vehicles.

[0131] Embodiment 3

[0132] This embodiment provides an unmanned aerial vehicle path planning system based on an intelligent window, which is used to implement the method in Embodiment 1. As shown in Figure 2 , the system includes:

[0133] An intelligent window, which is built-in with a positioning module, a path tracking controller, and a path planning engine module. The positioning module is used to obtain the real-time position coordinates of the target intelligent window. The path tracking controller is used to provide a composite control instruction when performing real-time path correction on the unmanned aerial vehicle. The path planning engine module is used to calculate and output results when performing global initial path, local obstacle avoidance adjustment, real-time path correction, and landing control.

[0134] An unmanned aerial vehicle, which is equipped with a multi-sensor module, is used to obtain real-time position coordinates, flight speed and flight attitude data of the unmanned aerial vehicle, and visual recognition when landing.

[0135] A communication link module is used to realize data, instruction transmission and state synchronization between the intelligent window and the unmanned aerial vehicle.

[0136] The intelligent window is also provided with a retractable parking platform. The unfolded area of the parking platform matches the size of the unmanned aerial vehicle, as follows:

[0137]

[0138] Where L uav is the diagonal length of the unmanned aerial vehicle, and Δ is the safety margin.

[0139] Embodiment 4

[0140] This embodiment is based on the intelligent window-based UAV path planning system provided in Embodiment 3, and adds a trajectory simulation module for forward simulation of candidate trajectories in the local obstacle avoidance process, and timely verification of the candidate trajectory selected as the local obstacle avoidance path, to improve the obstacle avoidance reliability.

[0141] Embodiment 5

[0142] This embodiment provides a computer storage medium storing program instructions for executing the method in Embodiment 1.

[0143] It should be understood that although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or sub-steps or stages of other steps.

[0144] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware. The above-mentioned program can be stored in a non-volatile computer readable storage medium, and when the program is executed, it can include the processes of the above-mentioned embodiments of each method. Any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0145] Any combination of the technical features in the above-described embodiments can be made, and for the sake of brevity, not all possible combinations are described, however, as long as there is no conflict, any combination of the technical features should be considered within the scope of the present disclosure.

[0146] The above-described embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the patent scope of the present application. It should be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for unmanned aerial vehicle path planning based on smart windows, characterized in that: The method is as follows: S1, constructing an obstacle map based on the target intelligent window; S1-1, through the real-time position of the built-in positioning module installed in the target intelligent window and multi-sensor fusion, generate positioning coordinates (x w ,y w ,z w ) as the target landing position of the unmanned aerial vehicle; S1-2, acquiring environmental data in real time through the onboard radar and visual sensor of the unmanned aerial vehicle, and combining the BIM model in which the target intelligent window is located to construct a three-dimensional dynamic obstacle map; S2, global initial path planning; Based on A * The algorithm generates a global initial path, and the cost function is as follows: F(n) = α1·D(n) + β1·E(n) + γ1·O(n) Wherein, D(n) is the Euclidean distance from node n to the positioning coordinates, E(n) is the cumulative energy consumption from the starting point to node n, O(n) is the threat coefficient of the obstacles around node n, and α1, β1, γ1 are dynamic weight coefficients; S3, local obstacle avoidance adjustment; Based on the global initial path and the dynamic window method, candidate trajectories are sampled in the speed space (v, w), local obstacle avoidance adjustment is performed by maximizing the target function, a local obstacle avoidance path is obtained, and the corresponding part of the global initial path is replaced to obtain the planned path; as follows: G(v, ω) = α2·G nav (v, ω) + β2·C obs (v, ω) + γ2·S smooth (v, ω) where v is the linear velocity of the UAV, ω is the angular velocity of the UAV, G nav (v, ω) is a navigation gain term of the UAV, C obs (v, ω) is an obstacle avoidance cost term of the UAV, S smooth (v, ω) is a motion smoothness term of the UAV, and α2, β2, and γ2 are obstacle avoidance weight coefficients. S4, real-time path correction; Through the real-time communication link between the target intelligent window and the unmanned aerial vehicle, path deviation is corrected, and dynamic weight coefficients are adjusted; S5, landing control; When the unmanned aerial vehicle approaches the target intelligent window, the relative pose of the unmanned aerial vehicle and the intelligent window is calculated based on a visual recognition and ultrasonic ranging fusion positioning method, and accurate landing is completed.

2. The unmanned aerial vehicle path planning method based on an intelligent window according to claim 1, characterized in that: The positioning coordinates in step S1-1 are obtained by fusing the real-time position coordinates of the built-in positioning module, GPS coordinates, and multiple sensors, as follows: where P W is the real-time position coordinate of the built-in positioning module, P GPS is the GPS coordinate of the target positioning window, is the distance of the i-th reference object measured by the laser radar sensor, f(P w ) is the coordinate conversion function, and λ is the fusion weight. 3.The method of claim 1, wherein: In step S2: wherein v t is the linear velocity at the node n, ω t is the angular velocity at the node n, and l is the moment of inertia of the UAV. where η is the attenuation factor, d k is the distance of node n to the kth obstacle, r k is the influence radius of the kth obstacle. 4.The intelligent window based UAV path planning method of claim 1, wherein: The specific method of step S3 is as follows: S3-1, sampling candidate trajectories in the speed space (v, ω) based on the global initial path and the dynamic window method; S3-2, constructing a maximization target function including a navigation gain term, an obstacle avoidance cost term, and a motion smoothness term, and calculating the evaluation scores of each candidate trajectory respectively; S3-3, selecting the candidate trajectory with the highest evaluation score as the local obstacle avoidance path.

5. The method of claim 1, wherein: The specific method of step S4 is as follows: S4-1, discretizing the planned path into a theoretical path point sequence, and each theoretical path point including longitude, latitude, height, and an allowable deviation threshold; S4-2, the target intelligent window receives the state data stream of the unmanned aerial vehicle, and performs spatio-temporal alignment and confidence weighted fusion on the state data stream based on a Kalman filtering algorithm; S4-3, based on the fuzzy inference rule base, dynamically adjust the lateral heading weight coefficient of the target intelligent window built-in path tracking controller with the theoretical path point sequence as the reference longitudinal speed weight coefficient k v height maintenance weight coefficient k h to satisfy: and The heading correction angle is generated simultaneously with the generation of the composite control command a composite control command of the speed compensation amount Δv and the height adjustment amount Δh S4-4, the real-time communication link encapsulates the above composite control instructions into a data packet and sends it to the flight control processor of the unmanned aerial vehicle for execution.

6. The method of claim 1, wherein: The relative pose in step S5 is calculated as follows: Δz = z ultra - z w ; Wherein, (x cam ,y cam ) is the center coordinate of visual identification, z ultra is the ultrasonic ranging distance value.

7. The method of claim 4, wherein: In step S3-2 The navigation gain term is the weighted sum of the lateral deviation Δd and the heading deviation Δθ of the trajectory endpoint and the global path, which is calculated as follows: G nav (v, ω) = w1 · e -|Δd| + w2 · cos(Δθ) w1+w2=1; The obstacle avoidance cost term is the inverse integral of the minimum distance d of each point of the trajectory to the nearest obstacle min is computed as follows: The motion smoothness term is a quadratic form of the linear velocity change rate and the angular velocity change rate is calculated as follows: And the obstacle avoidance weight coefficients α2, β2, and γ2 satisfy the following: α2∈[0.5,0.8],β2∈[0.3,0.6],γ2∈[0.1,0.3]。 8. An intelligent window based UAV path planning system for implementing the method of any one of claims 1 to 7, characterized by: The system comprises: The intelligent window is internally provided with a positioning module, a path tracking controller and a path planning engine module; the positioning module is used to obtain real-time position coordinates of the target intelligent window; the path tracking controller is used to provide a composite control instruction when performing real-time path correction on the unmanned aerial vehicle; and the path planning engine module is used to calculate and output results when performing global initial path, local obstacle avoidance adjustment, real-time path correction and landing control; The unmanned aerial vehicle is provided with a multi-sensor module, which is used to obtain real-time position coordinates, flight speed and flight attitude data of the unmanned aerial vehicle, and visual identification when landing; A communication link module is used to realize data and instruction transmission and state synchronization between the intelligent window and the unmanned aerial vehicle.

9. The UAV path planning system based on smart windows according to claim 8, wherein: The intelligent window is further provided with a retractable landing platform; the unfolded area of the landing platform matches the size of the unmanned aerial vehicle, as follows: where L uav is the diagonal length of the UAV, and Δ is the safety margin.

10. A computer storage medium, characterized in that: The computer storage medium stores program instructions for executing the method of any one of claims 1 to 7.

Images