Air-ground conversion type unmanned aerial vehicle distribution method for building power failure emergency

Abstract

The invention relates to the technical field of unmanned aerial vehicles, in particular to an air-ground conversion type unmanned aerial vehicle distribution method for building power failure emergency. The method comprises the following steps: an unmanned aerial vehicle flies to an external airspace of a target floor in a first working mode according to a task instruction and hovers at an external airspace position corresponding to an adsorption starting point; the unmanned aerial vehicle gets close to the adsorption starting point, and when a preset contact condition is met, the working mode of the unmanned aerial vehicle is converted into a second working mode from a first working mode; the unmanned aerial vehicle is controlled to move towards the target window along the target outer vertical surface, and the article delivery operation is executed when the delivery condition is met; and after delivery is completed, the unmanned aerial vehicle is controlled to remove adsorption and is switched to the first working mode. According to the method, direct delivery from the delivery station to the user window is realized, the delivery efficiency and reliability in an emergency scene are improved on the premise of ensuring the task safety, and the dependence on manual operation and potential risks are reduced.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and more specifically, to an air-to-ground UAV delivery method for emergency power outages in buildings. Background Technology

[0002] With the surge in urban food delivery demand, the delivery efficiency of high-rise buildings heavily relies on elevators. In the event of natural disasters, power outages due to line failures or equipment maintenance, elevator malfunctions will disrupt traditional delivery methods: delivery personnel manually climbing stairs is time-consuming and laborious, making it difficult to meet the timeliness requirements of meals and posing safety risks; users may cancel orders or complain due to excessively long waiting times.

[0003] When faced with extreme situations such as elevators in buildings shutting down due to sudden malfunctions like power outages, existing drone delivery solutions mostly focus on "end-to-end" aerial flight. However, in urban environments with towering buildings and complex obstacles, it is difficult to achieve accurate and safe delivery from the street to a specific floor, especially in the face of complex environments such as glass curtain walls, balconies, and air conditioning units on building facades.

[0004] In summary, this paper provides an air-to-ground drone delivery method for emergency power outages in buildings. Summary of the Invention

[0005] The purpose of this invention is to provide an air-to-ground convertible drone delivery method for emergency power outages in buildings, in order to solve the problem mentioned in the background art that drone delivery cannot cope with the complex environment of building facades such as glass curtain walls, balconies, and air conditioner outdoor units.

[0006] To achieve the above objectives, the present invention aims to provide an air-to-ground switching drone delivery method for emergency power outages in buildings, comprising the following steps:

[0007] S1. The background dispatch system receives emergency delivery tasks triggered by power outages in the target building, plans the four-dimensional task path and target window information of the drone from the starting point to the outside of the target building, and generates corresponding task instructions.

[0008] S2. Control the drone to fly to the external airspace of the target floor in the first working mode according to the mission instructions, identify the target window and calculate the optimal adsorption starting point on the target facade, and control the drone to hover at the external airspace position corresponding to the adsorption starting point.

[0009] S3. Control the drone to move from the hovering position to the adsorption starting point. When the preset contact conditions are met, make the drone adsorb onto the wall and change the drone's working mode from the first working mode to the second working mode.

[0010] S4. Control the drone in the second working mode to move along the target facade towards the target window, judge the delivery conditions through environmental perception, and execute the item delivery operation when the delivery conditions are met.

[0011] S5. After delivery, control the drone to move to the preset release point. At the preset release point, control the drone to release the attachment and switch to the first working mode, and then detach from the building facade to return.

[0012] As a further improvement to this technical solution, the four-dimensional task path in S1 includes the following:

[0013] (1) The aerial flight path segment from the drone's takeoff point to the target building;

[0014] (2) Coordinates of the hovering positioning point outside the target building;

[0015] (3) The approach vector from the hovering positioning point to the preset adsorption starting point on the target facade;

[0016] (4) The expected crawling path segment on the building facade from the adsorption starting point to the target window;

[0017] (5) The crawling path segment from the target window to the departure point after delivery is completed;

[0018] (6) The flight path segment from the departure point back to the starting point.

[0019] As a further improvement to this technical solution, it also includes a step of real-time monitoring of the drone's remaining battery power. During mission planning and execution, when the battery power is lower than the first threshold, the shortest crawling path is selected; when the battery power is lower than the second threshold and step S3 has not yet started, the current mission is canceled and the drone immediately returns to charge; if step S3 has already started, the current action is stopped and the drone is controlled to detach and return to home; when the battery power is lower than the third threshold in the second working mode, step S5 is entered.

[0020] As a further improvement to this technical solution, in step S2, the adsorption initiation point satisfies the following condition:

[0021] The adsorption starting point and the target window meet the preset safety and delivery position relationship conditions.

[0022] The area of ​​the exterior facade where the adsorption initiation point is located satisfies the adsorption conditions.

[0023] The exterior facade area where the adsorption initiation point is located meets the condition of being free of obstacles;

[0024] The estimated movement path from the adsorption starting point to the target window meets the preset path efficiency optimization conditions.

[0025] As a further improvement to this technical solution, the specific steps in S3 for attaching the drone to the wall are as follows:

[0026] S301. Control the drone to move towards the adsorption starting point with a preset attitude and speed, and prepare to start the adsorption function.

[0027] S302. When the distance to the building facade is detected to reach the preset contact condition, the adsorption function is activated.

[0028] S303. Monitor the adsorption force. When the adsorption force reaches and remains at the preset safety conditions, it is determined that the adsorption is stable.

[0029] As a further improvement to this technical solution, in step S303, if the adsorption force fails to reach the preset safety condition, it is determined that the adsorption has failed, and an emergency recovery process is executed. The emergency recovery process includes:

[0030] Control the drone to detach from the building facade and remove it from its attachment.

[0031] Control the drone to return to a safe hovering state;

[0032] Reassess and select a new adsorption starting point, and try the conversion again; if the number of consecutive failures exceeds the set number, an alarm will be sent to the background scheduling system and manual instructions will be awaited.

[0033] As a further improvement to this technical solution, in step S4, the delivery conditions are determined through environmental perception, and the item delivery operation is performed when the conditions are met. Specifically, this includes:

[0034] S401. Detect the open state of the target window through environmental perception;

[0035] S402. If the target window is detected to be fully open, then perform the delivery operation.

[0036] S403. If the target window is not fully opened, a prompt message to open it is sent to the user terminal, and the delivery operation is performed after receiving the user's confirmation instruction.

[0037] As a further improvement to this technical solution, in S403, if no user confirmation instruction is received within a preset waiting time, the drone is controlled to place the item at a preset emergency delivery location outside the window and send a notification to the user.

[0038] As a further improvement to this technical solution, in step S5, when controlling the drone to detach from the building facade, the detachment action is dynamically adjusted according to the ambient wind speed.

[0039] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0040] 1. This air-to-ground switching drone delivery method for building power outage emergencies proposes an "air-to-ground switching" mode, enabling drones to both fly quickly to the destination and then adhere to and crawl to accurately deliver to windows. This completely bypasses obstacles such as elevator malfunctions and blocked building passageways during power outages, achieving direct delivery from the delivery station to the user's window, greatly improving delivery efficiency and reliability in emergency scenarios. Furthermore, through environmental perception, the drone can accurately identify target windows, determine their open or closed status, and calculate the optimal adhesion point, ensuring automation and high safety throughout the entire process from flight, adhesion, crawling to delivery and detachment, reducing reliance on manual operation and potential risks.

[0041] 2. In this air-to-ground switching drone delivery method for building power outage emergencies, intelligent power management dynamically adjusts path planning and behavior decisions based on the remaining power, maximizing the effective operating time and mission completion rate of the drone while ensuring mission safety. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the overall method of the present invention;

[0043] Figure 2 This is a system block diagram of the interaction between the UAV and the target window in this invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0045] Example:

[0046] Please see Figure 1 As shown, this embodiment provides an air-to-ground switching drone delivery method for emergency power outages in buildings, including the following steps:

[0047] S1. The background dispatch system receives emergency delivery tasks triggered by power outages in the target building, plans the four-dimensional task path and target window information of the drone from the starting point to the outside of the target building, and generates corresponding task instructions.

[0048] In S1, the four-dimensional task path includes the following:

[0049] (1) The aerial flight path segment from the drone's takeoff point to the target building;

[0050] (2) Coordinates of the hovering positioning point outside the target building;

[0051] (3) The approach vector from the hovering positioning point to the preset adsorption starting point on the target facade;

[0052] (4) The expected crawling path segment on the building facade from the adsorption starting point to the target window;

[0053] (5) The crawling path segment from the target window to the departure point after delivery is completed;

[0054] (6) The flight path segment from the departure point back to the starting point.

[0055] The detailed steps for S1 are as follows:

[0056] First, the back-end dispatch system monitors or receives power outage status information of the target building from external systems in real time. When it is confirmed that the elevator of the target building has stopped operating due to a power outage, the system automatically marks delivery orders sent to the higher floors of the building that meet the preset conditions as emergency delivery tasks.

[0057] The preset conditions include: the delivered items are takeout food, emergency medicine, or important documents; the target floor is located on the 5th floor or above; and the user has selected the "emergency drone delivery" option in the order or the system automatically determines that it is necessary.

[0058] Then, the background dispatch system obtains the target information of the emergency delivery task, including the three-dimensional geographic coordinates of the target building, the target floor number, the preset target window identification information, and the user-authorized delivery permission;

[0059] Secondly, the background scheduling system calls the preset digital building model database, or obtains the three-dimensional model of the target building's facade through the drone's preceding flight scan data. The model includes at least facade material distribution information, window position and size information, and potential obstacle information.

[0060] Finally, based on the target information and the 3D model of the facade, the background scheduling system plans a complete four-dimensional task path;

[0061] The four-dimensional mission path, target window identification information, and necessary safety parameters are packaged into a mission command, which is then sent to the control system of the designated UAV via a wireless communication network.

[0062] Essential safety parameters include flight safety envelope parameters, building facade operation constraint parameters, environmental adaptation and decision threshold parameters, and delivery interaction and fault tolerance parameters;

[0063] The flight safety envelope parameters include the maximum permissible flight speed, minimum obstacle avoidance distance, and coordinates of no-fly zones / protected areas for each segment of the path. The purpose is to define safe corridors for the autonomous flight of drones, avoiding potential risks such as narrow airspace between buildings, balconies, and billboards.

[0064] The building facade operation constraint parameters include the coordinates of the adsorption safety area and material labels, which can define the recommended adsorption plane area and its material, avoiding unreliable areas such as decorative strips and drainage pipes; it also includes the maximum allowable adsorption force and contact speed, which are preset according to different materials and drone loads to prevent the suction cup from damaging the curtain wall due to excessive pressure or causing instability due to contact impact;

[0065] Environmental adaptation and decision-making threshold parameters include wind speed operating threshold and power safety threshold;

[0066] Delivery interaction and fault tolerance parameters include window opening waiting timeout duration, alternative delivery strategy triggering conditions, and force / speed curve index of the delivery execution mechanism.

[0067] S2. Control the drone to fly to the external airspace of the target floor in the first working mode according to the mission instructions, identify the target window and calculate the optimal adsorption starting point on the target facade, and control the drone to hover at the external airspace position corresponding to the adsorption starting point.

[0068] During mission planning and execution, the remaining battery power of the drone is monitored in real time. When the battery power is lower than the first threshold, the shortest crawling path is selected.

[0069] If the battery level is below the second threshold and the S3 step has not yet started, the current mission will be canceled and the drone will immediately return to its home base to recharge. If the S3 step has already started, the current action will be stopped and the drone will be controlled to detach and return to its home base.

[0070] When the battery level is lower than the third threshold in the second operating mode, proceed to step S5.

[0071] Specifically, the first threshold is typically set between 30% and 40% of the total battery capacity. This setting is primarily based on the estimated battery power required to perform the most complex disengagement and direct return maneuvers from the furthest point of the current task (usually the target window), with a certain safety margin. This threshold indicates that the system's battery level has entered a range requiring conservative planning; when the battery level is below this threshold but above the second threshold, the system determines that it still has sufficient power to complete a full delivery and return cycle, but energy consumption needs to be optimized. Therefore, in step S4 (crawling delivery), the path planning algorithm will prioritize the path with the shortest crawling distance and fewest turns to minimize energy consumption in the higher-power-consuming "second working mode."

[0072] The second threshold is set between 20% and 25% of the total battery capacity. This capacity is assessed as the minimum safe charge required to ensure the drone safely detaches from any attached state from the building facade and flies back to the nearest alternative landing point or makes an emergency landing. It is higher than the absolute minimum requirement for simply maintaining level flight and returning to base.

[0073] If triggered before the start of step S3 (mode transition), it indicates that the drone is still in a state of completely free flight. At this time, canceling the mission and returning directly has the lowest risk and the highest success rate.

[0074] If triggered after the start of step S3 (i.e., the drone is attempting or has already completed attachment), it indicates that the system has entered a high-risk phase. At this point, regardless of whether attachment is successful, the system will immediately cease any crawling or delivery attempts and forcibly initiate an emergency detachment procedure (similar to the attachment failure recovery procedure) to ensure the drone quickly and safely detaches from the building facade, and then uses the remaining battery power to return to base. This prevents the drone from becoming stuck on the wall due to depleted battery power.

[0075] The third threshold is set between 15% and 20% of the total battery power. This power level is assessed as the absolute minimum required to complete the detachment maneuver and perform an emergency return to the nearest safe point while the drone is attached to the wall in "Second Operating Mode". When the battery level drops to this point during step S4 (crawling delivery phase), it indicates that the battery is extremely low, and no unnecessary movement or waiting is permitted. The system will immediately and forcibly interrupt the delivery process (even if the window is open) and unconditionally enter step S5 (detachment and return) to ensure that the drone does not run out of power after completing the delivery maneuver and is unable to detach from the wall.

[0076] In S2, the adsorption initiation point satisfies the following condition:

[0077] The adsorption starting point and the target window meet the preset safety and delivery position relationship conditions; that is, it is located within a preset safety distance to the side or below the target window to avoid directly blocking the window.

[0078] The exterior facade area where the adsorption initiation point is located meets the adsorption conditions; that is, the exterior facade material is flat and suitable for adsorption.

[0079] The exterior facade area where the adsorption initiation point is located meets the condition of being free of obstacles; that is, there are no obvious obstacles such as air conditioning units, railings, pipelines, etc.

[0080] The estimated movement path from the adsorption starting point to the target window meets the preset path efficiency optimization conditions; that is, the estimated crawling path from this point to the target window is relatively straight and the distance is the shortest.

[0081] S2 specifically includes the following steps:

[0082] First, after receiving the mission instructions, the drone carrying the delivery items enters the first working mode, unfolds all its rotors, and autonomously flies towards the target building along the planned aerial flight path.

[0083] Secondly, when the drone flies to the airspace near the target building and reaches the hovering positioning point, it acquires fused LiDAR point cloud data and visual images through environmental perception, performs real-time 3D reconstruction and matching of the building facade, compares the real-time perception data with the 3D model issued by the background, accurately locates the horizontal strip area where the target floor is located, and initially identifies the preset target window identification information.

[0084] Preliminary methods for identifying target window identification information include: identifying visual identifiers of specific patterns, shapes, or colors that users have pre-pasted on the window; or identifying unique grid patterns, window dimensions, and relative positional combinations that are characteristic of the window itself.

[0085] Finally, while maintaining stable hovering, the drone dynamically fine-tunes the approach vector based on real-time perception data to calculate the optimal adsorption starting point. The specific calculation steps are as follows:

[0086] Step 1: Determine the candidate region

[0087] Using the geometric center of the target window as a reference, a rectangular candidate area is delineated on the building facade. The extent of this area is defined by preset safety distance parameters.

[0088] Horizontal direction: extending to the left and right sides of the window rice( ).

[0089] Vertical direction: mainly extends downwards from the window. rice( ), extending upwards rice( (to avoid obstructing the window).

[0090] Safety distance parameters , , Based on the drone's fuselage size, rotor deployment radius, and safety margin settings, this ensures that the drone does not collide with windows or building structures during adsorption and crawling. This area ensures that the candidate point is located to the side or below the window, meeting the basic requirement of "avoiding direct obstruction of the window".

[0091] Step 2: Mesh Discretization and Attribute Labeling

[0092] The candidate regions are discretized into a set of grid points at a preset resolution (0.1m × 0.1m). For each grid point By fusing real-time sensing data (LiDAR point cloud, visual images, and infrared data), it is endowed with the following attributes:

[0093] Surface material type Such as glass, smooth paint, metal, rough stone, etc.

[0094] Local flatness : By calculating points The standard deviation of the point cloud normal vector within a small surrounding area is used for quantification; the smaller the value, the smoother the surface.

[0095] obstacle distance :point The Euclidean distance to the nearest obstacle (such as an air conditioner outdoor unit or a railing).

[0096] Estimated crawl path length to the target window Based on the wall grid diagram, search algorithms such as A* are used to calculate the value from... The shortest path length to the target window delivery preparation position (usually the center of the window slightly inside), the path must avoid known obstacles.

[0097] Step 3: Construction of Multi-Constraint Cost Function

[0098] For each candidate point Construct a multi-objective cost function A smaller function value indicates a better point. The cost function consists of the following weighted sub-terms:

[0099]

[0100] In the formula, This is the normalization factor for the path length; The weighting coefficient for the path length; For flatness The normalization function maps the smoothness to a score in the interval [0,1], with higher scores for smoother surfaces. This is the weighting coefficient for local flatness; For the material fit function, based on the point material Return a cost. In this embodiment, surfaces suitable for vacuum adsorption, such as glass and smooth ceramic tiles, have a low cost (0.1). Surfaces suitable for electromagnetic adsorption, such as metal frames, have a low cost if the drone is equipped with an electromagnetic chuck; otherwise, the cost is high. Rough, porous, or unknown materials have a high cost (0.9). The weighting coefficient for the surface material type; It should be a very small positive number to prevent division by zero errors; This is the weighting coefficient for the distance to obstacles;

[0101] Step 4: Optimal Solution Selection and Verification

[0102] (1) Traverse all candidate grid points and calculate their costs. .

[0103] (2) Select points whose cost is lower than the preset threshold to form a feasible solution set.

[0104] (3) Select the point with the minimum cost from the set of feasible solutions. As the optimal adsorption starting point.

[0105] (4) Perform final verification: Check from the current hover position to The "proximity vector" is checked for collision-free nature and is reachable by the UAV's attitude. If the verification fails, a second-best option is selected from the set of feasible solutions.

[0106] S3. Control the drone to move from the hovering position to the adsorption starting point. When the preset contact conditions are met, make the drone adsorb onto the wall and change the drone's working mode from the first working mode to the second working mode.

[0107] In this embodiment, a fusion strategy of "distance threshold as the primary factor and contact sensing verification" is adopted to set the preset contact conditions, and the specific settings are as follows:

[0108] The system uses onboard laser rangefinders, ultrasonic sensors, or stereo matching algorithms based on binocular vision to measure the vertical distance in real time between the drone's adsorption contact surface and the wall at the target's "adsorption starting point." When this distance is less than or equal to a first preset distance threshold... ( The range is 20mm to 100mm, and the value is set according to the size and speed of the drone. In this embodiment, the preferred value is 50mm. When the contact condition is initially determined to be met, it is determined that the contact condition is met.

[0109] Contact sensors such as microswitches and thin-film pressure sensors are installed on the contact surface of the drone. When any contact sensor is triggered and generates a valid signal, it is determined that physical contact has occurred and the contact condition is met.

[0110] In S3, the specific steps to attach the drone to the wall are as follows:

[0111] S301. Control the drone to move towards the adsorption starting point at a preset attitude and speed, and prepare to start the adsorption function.

[0112] S302. When the distance to the building facade is detected to reach the preset contact condition, the adsorption function is activated.

[0113] S303. Monitor the adsorption force. When the adsorption force reaches and remains at the preset safety condition, it is determined that the adsorption is stable.

[0114] Once reliable adsorption is confirmed, commands are sent to the drone to first reduce the rotor speed to idle or zero, and then control the actuator of the foldable rotor to fold all the rotors backward or towards the fuselage, so as to significantly reduce the drone's projected area and wind resistance on the building facade.

[0115] After the rotor folding is completed, the global status flag of the drone is switched from "flight mode" to "crawling mode", and a confirmation signal of "mode conversion completed, entered the second working mode" is sent to the upper control system and the background scheduling system.

[0116] In S302, the adsorption function is activated as follows:

[0117] The wall material at the adsorption initiation point has been analyzed during the approach process using environmental perception.

[0118] If the material is glass, ceramic tile, or a smooth painted surface, select to activate the vacuum suction cup array, which uses a micro vacuum pump to quickly draw air and generate negative pressure for adsorption.

[0119] If the material is a metal window frame, metal decorative strip, or steel structure, select to activate the electromagnetic chuck, which will generate magnetic attraction when powered on.

[0120] If the material is a mixture, the vacuum suction cup array should be activated first, and the electromagnetic suction cup should be activated as needed to enhance local fixation.

[0121] In S303, the preset safety conditions include the adsorption force threshold condition and the stability duration condition;

[0122] For vacuum suction cup arrays: negative pressure values ​​are monitored by pressure sensors integrated into the vacuum tubing. When the absolute value of the monitored negative pressure is greater than or equal to a preset safe pressure threshold... At that time, the force value condition is satisfied. Calculations are based on the drone's weight, payload, and safety factor (typically 1.5 to 2.5 times). For example, if the total weight is 5kg, the effective adsorption area of ​​a single suction cup is calculated as follows: It can be set in the range of -60kPa to -80kPa.

[0123] For electromagnetic chucks: the magnetic flux is estimated by monitoring the current value in its working circuit or indirectly using a Hall sensor. When the current value reaches or exceeds a preset safe current threshold... When sufficient electromagnetic attraction is generated, the force value condition is met. Also based on load and safety factor calibration.

[0124] Meeting the force threshold is not enough; the adsorption force must also be continuously stable. Therefore, the state of meeting the force threshold must be maintained continuously for a preset stable period of time. (In this embodiment) The range is 0.5 seconds to 1.5 seconds. Within a given time period, fluctuations in adsorption force (pressure or current) must be less than the allowable fluctuation range (±5%).

[0125] Only when both "adsorption force is greater than or equal to the corresponding safety threshold" and "this state remains stable at greater than or equal to" are conditions met, can the adsorption force be effectively maintained. Only when the system finally determines that the "preset safety conditions" are met and confirms that the adsorption is stable.

[0126] Furthermore, in S303, if the adsorption force fails to reach the preset safety condition, it is determined that the adsorption has failed, and an emergency recovery procedure is executed. The emergency recovery procedure includes:

[0127] Control the drone to detach from the building facade and remove it from its attachment.

[0128] Control the drone to return to a safe hovering position;

[0129] Reassess and select a new adsorption starting point, and try the conversion again; if the number of consecutive failures exceeds the set number, an alarm will be sent to the background scheduling system and manual instructions will be awaited.

[0130] The number of attempts is typically set to 2 to 3. This is based on the following: in most cases, a single failure is due to chance factors (such as a sudden gust of wind or sensor noise), and a second attempt may succeed; if two consecutive failures occur, it indicates a systemic adsorption problem in the area; the third attempt serves as final confirmation. Continuing to attempt autonomously beyond this number results in a low success rate and accumulated risk, so the system is switched to reporting to the backend, where the operator makes a decision based on real-time visuals (e.g., instructing the user to try an alternate window or canceling the current delivery). This strategy strikes a balance between automation efficiency and ultimate safety, and through an emergency recovery process with intelligent retry logic, it significantly improves the system's success rate and adaptability under non-ideal wall conditions.

[0131] S4. Control the drone in the second working mode to move along the target facade towards the target window, judge the delivery conditions through environmental perception, and execute the item delivery operation when the delivery conditions are met.

[0132] In S4, delivery conditions are determined through environmental perception, and the item delivery operation is executed when the conditions are met. Specifically, this includes:

[0133] S401. Detect the open status of the target window through environmental perception;

[0134] S402. If the target window is detected to be fully open, then perform the delivery operation.

[0135] S403. If the target window is not fully opened, a prompt message to open it is sent to the user terminal, and the delivery operation is performed after receiving the user's confirmation instruction.

[0136] If no confirmation instruction is received from the user within the preset waiting time, the drone will be controlled to place the item at the preset emergency delivery location outside the window and send a notification to the user.

[0137] S4 is the stage of climbing along the facade, precise window positioning, and safe delivery. The specific process is as follows:

[0138] Facade crawling and moving sub-step: Drive the drone to move on the exterior facade of the building according to the preset or real-time planned crawling path; the movement method is to control the adsorption / release sequence of different suction cup groups, and coordinate with the fine-tuning of the thrusters or tracks to achieve lateral, longitudinal or diagonal stepping movement.

[0139] Window precise positioning steps: During the movement, the target window identification information is locked by a high-resolution visual camera, and the relative coordinate error between the UAV's own position and attitude and the center of the target window is controlled within the centimeter level by combining the laser radar ranging.

[0140] Intelligent window status judgment sub-step: When the drone moves to the preset delivery preparation position directly in front of the target window, it detects the temperature field change in the window area through infrared sensors, and combines visual analysis of the window gaps, shadows and light transmittance to comprehensively judge whether the target window is in a "fully open", "partially open" or "fully closed" state.

[0141] Human-machine collaborative interaction sub-step: If it is determined that the window is "completely closed" or "partially open" but the opening is insufficient for safe delivery, the drone will initiate a strong reminder interaction request to the user's mobile terminal corresponding to the order through its communication module. The request content includes at least: "The drone has arrived outside your window. Please fully open the window for safe delivery."

[0142] Upon receiving the user's confirmation command, the delivery execution mechanism in the drone is triggered: the door of the mission storage compartment in the delivery execution mechanism opens, and the built-in electric telescopic push rod or synchronous belt drive mechanism pushes the delivery item out of the storage compartment at a smooth speed and force, through the window, and places it in a preset safe area indoors or at least 0.3 meters away from the edge of the window; the mission storage compartment is a modular compartment with heat preservation, shock absorption or specific shape adaptation functions; in the S45 delivery execution sub-step, the delivery execution mechanism dynamically adjusts the push speed and force curve according to the weight and center of gravity of the item in the compartment to ensure the smooth transfer of the item.

[0143] Delivery completion confirmation sub-step: After delivery is executed, the delivery action is recorded as successful if the item has left the storage compartment by checking the changes in the weight sensor inside the compartment or by visual confirmation, and the item has entered the room by checking the miniature camera facing the room.

[0144] In the fourth stage, S44 sub-step, if no user confirmation instruction is received within the preset waiting time, the drone will be controlled to place the item at the preset emergency delivery location outside the window and send a notification to the user.

[0145] S5. After delivery, control the drone to move to the preset release point. At the preset release point, control the drone to release the attachment and switch to the first working mode, and then detach from the building facade to return.

[0146] When controlling a drone to detach from a building facade, the detachment action is dynamically adjusted according to the ambient wind speed:

[0147] If the real-time wind speed is below the safety threshold, the drone will immediately detach; if the wind speed is high, the drone will be controlled to provide an initial acceleration in the direction away from the building at the moment of detachment, so as to quickly enter the stable airflow zone.

[0148] In S5, the specific steps for controlling the drone to return to home are as follows:

[0149] Delivery and detachment preparation sub-step: After confirming that the delivery is completed, control the drone to move along the facade from the target window back to the preset detachment point. This point is usually selected on the side, below or at the corner of the window to avoid affecting the window during detachment.

[0150] Adsorption release and flight system preparation sub-steps: After the release point is stabilized, first control the UAV to gradually release the adsorption force, and at the same time control the actuator of the foldable rotor to unfold the rotor from the folded state to the flight preparation position.

[0151] Safe detachment and mode reset sub-steps: After confirming that all rotors are deployed and locked in place, issue a detachment command: In the final stage of releasing the suction force, make the rotors quickly reach a speed that provides slight lift, assisting the drone to detach smoothly and cleanly from the building facade;

[0152] Return flight sub-steps: After completely disengaging, switch the drone's global status flag back to "flight mode" and control the drone to fly back to the starting point, transit station, or to the next mission point according to the planned return flight path;

[0153] Mission closure and data upload sub-steps: During the return journey or after landing, the UAV will package and upload all the data of this mission, including path execution status, mode conversion logs, sensor data, delivery result images and abnormal event records, to the background scheduling system for mission settlement, system optimization and data analysis.

[0154] An example of this solution is: delivering lunch to an office building that has experienced a power outage due to a malfunction.

[0155] The interaction process between the drone and the target window used in this embodiment is as follows: Figure 2As shown;

[0156] The drone is a quadcopter with a frame made of carbon fiber, making it lightweight and sturdy. At its core is a high-performance flight controller that integrates an air-to-ground switching control module.

[0157] The building surface moving module is located at the bottom and around the sides of the unit. It includes two arrays of eight miniature vacuum chucks, each connected to an independent vacuum generator; and two sets of electromagnetic chucks located on both sides of the unit, suitable for metal frames.

[0158] Environmental perception module: The nose is equipped with a pair of binocular cameras and a single-line lidar. The belly is equipped with an infrared thermal imager and a miniature camera facing the delivery direction.

[0159] Mission storage compartment: It is a modular insulated compartment with grooves and elastic straps for accommodating meal boxes. The compartment door is an electric sliding cover.

[0160] Delivery mechanism: An electric push rod controlled by a stepper motor is installed at the bottom of the compartment.

[0161] Back-end dispatch system: running on a cloud server, possessing a city digital building model database, and can interface with food delivery platform order systems and power grid fault alarm systems.

[0162] The specific implementation process of using the above modules to execute the drone delivery method is as follows:

[0163] Step S1: Task Triggering and Planning

[0164] The backend dispatch system detected a power outage in a building due to a line fault, causing the elevators to stop working. Almost simultaneously, the system received a food delivery order for room 2208 on the 22nd floor of the building. Based on the rules (floor > 5 and food), the system automatically classified it as an emergency delivery task.

[0165] The system retrieves the building's digital model, which shows that the windows of apartment 2208 on the north facade of the 22nd floor are glass curtain walls with metal frames. The system plans the following path: The drone takes off from "Delivery Station A" next to the building and flies in a straight line to the hovering point on the north side of the building (coordinates...). , (Corresponding to a height of 22 floors), the preset adsorption starting point is the glass wall 0.8 meters below the right side of the window in room 2208, the crawling path is vertically upward 0.8 meters to the window, and the detachment point is 1 meter to the left of the window. The mission command is issued to drone number 03 at delivery station A.

[0166] Step S2: Flight Approach

[0167] Drone No. 03, loaded with a takeout container, received instructions. The air-to-ground control module put it into its first operating mode, its rotors deployed, and it took off. The drone integrated GPS signals and visual odometry data, flying along a planned path. Approximately 90 seconds later, it reached its hovering point and hovered stably.

[0168] At this point, the airborne LiDAR began scanning the north facade, generating point cloud data, which was then matched with the model sent from the backend to quickly locate the 22nd floor area. Simultaneously, the dual cameras identified the blue circular marker (already registered in the digital model) pre-attached by the user on the window of room 2208, completing the initial positioning.

[0169] Step S3: Mode switching and adsorption

[0170] The air-to-ground conversion control module controls the drone to approach the preset adsorption starting point (glass wall) vertically and smoothly at a speed of approximately 0.2 meters per second. When the drone is about 5 centimeters away from the wall, the visual proximity sensor is triggered.

[0171] Adsorption Trigger: Once the environmental sensing module confirms that the adsorption point is made of clean glass, the air-to-ground conversion control module instructs the vacuum suction cup array to start. The miniature vacuum pump starts working instantly, the suction cups adhere tightly to the glass, and negative pressure is rapidly established.

[0172] Adsorption confirmation: If the pressure sensor inside the suction cup returns a pressure value of -80 kPa and remains stable for more than 0.5 seconds, it is determined to be reliable adsorption.

[0173] Flight system shutdown and folding: The controller immediately sends a command, the four rotor motors stop, and then the servos move to rotate and fold the four rotor arms backward, close to the fuselage. The drone's status indicator switches to "crawling mode" and notifies the backend: "Ready to attach".

[0174] Step S4: Crawling and Delivery

[0175] Crawling: The air-to-ground conversion control module controls the vacuum suction cup array to move in a stepping manner. First, the suction cup group on the left side of the fuselage maintains adhesion, while the suction cup group on the right side briefly releases the vacuum and moves 10 centimeters to the right using a miniature electric push rod, then re-adheres; then the left side group moves in the same way, enabling the drone to move horizontally to the right as a whole until it reaches directly below the window.

[0176] Precise positioning and status assessment: After the drone arrived below the window, the infrared thermal imager detected the window area. The image showed a significant temperature difference between the inside of the window (approximately 22°C) and the outside (approximately 10°C), and there was a noticeable "cold seam" at the bottom of the window, indicating that the window was "fully open".

[0177] Human-machine collaboration: Since the window is already open, skip the window opening prompt step.

[0178] Delivery executed: The belly hatch slides open. An electric push rod smoothly pushes the meal box out at a preset medium speed, based on its weight (measured by sensors), passing through the window and placing it on a desk approximately 0.5 meters from the window sill inside the cabin. A miniature camera confirms the meal box has entered the cabin and is stably positioned.

[0179] Delivery complete: The lever retracts and the hatch closes. The system records delivery successful.

[0180] Step S5: Disengagement and Return

[0181] Prepare to disengage: Control the drone to crawl 1 meter to the left to the disengagement point (a flat glass wall).

[0182] Mode Reset and Detachment: First, deploy and lock the rotors. Then, while gradually reducing the negative pressure of the vacuum suction cup, start the rotor motors at low speed (providing approximately 1.5 times the weight of the drone). When the suction force approaches zero, the rotor lift gently detaches the drone from the wall.

[0183] Return: After the drone has completely detached, it switches back to "flight mode" and flies back to "delivery station A" along the planned return path, landing accurately.

[0184] Mission closed loop: The drone uploads the entire log (including the adsorption force curve, crawling trajectory, and delivery images) to the backend, and the mission ends.

[0185] This example demonstrates the complete application of the invention in a real power outage scenario. The drone operates entirely automatically, eliminating the need for delivery personnel to climb stairs and users to descend. Even when elevators are out of service, it can still deliver meals quickly, accurately, and safely to a designated window on the 22nd floor, effectively solving the "last 100 meters" problem of emergency delivery in high-rise buildings, demonstrating significant technological advantages and social value.

[0186] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A method for air-to-ground switching drone delivery for emergency power outages in buildings, characterized in that, Includes the following steps: S1. The background dispatch system receives emergency delivery tasks triggered by power outages in the target building, plans the four-dimensional task path and target window information of the drone from the starting point to the outside of the target building, and generates corresponding task instructions. S2. Control the drone to fly to the external airspace of the target floor in the first working mode according to the mission instructions, identify the target window and calculate the optimal adsorption starting point on the target facade, and control the drone to hover at the external airspace position corresponding to the adsorption starting point. S3. Control the drone to move from the hovering position to the adsorption starting point. When the preset contact conditions are met, make the drone adsorb onto the wall and change the drone's working mode from the first working mode to the second working mode. S4. Control the drone in the second working mode to move along the target facade towards the target window, judge the delivery conditions through environmental perception, and execute the item delivery operation when the delivery conditions are met. S5. After delivery, control the drone to move to the preset release point. At the preset release point, control the drone to release the attachment and switch to the first working mode, and then detach from the building facade to return.

2. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 1, characterized in that: In S1, the four-dimensional task path includes the following: (1) The aerial flight path segment from the drone's takeoff point to the target building; (2) Coordinates of the hovering positioning point outside the target building; (3) The approach vector from the hovering positioning point to the preset adsorption starting point on the target facade; (4) The expected crawling path segment on the building facade from the adsorption starting point to the target window; (5) The crawling path segment from the target window to the departure point after delivery is completed; (6) The flight path segment from the departure point back to the starting point.

3. The air-to-ground switching drone delivery method for building power outage emergencies according to claim 1, characterized in that: It also includes steps for real-time monitoring of the drone's remaining battery power. During mission planning and execution, when the battery power is below the first threshold, the shortest crawling path is selected; when the battery power is below the second threshold and step S3 has not yet started, the mission is canceled and the drone immediately returns to charge; if step S3 has already started, the current action is stopped and the drone is controlled to detach and return to home; when the battery power is below the third threshold in the second working mode, step S5 is entered.

4. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 1, characterized in that: In step S2, the adsorption initiation point satisfies the following condition: The adsorption starting point and the target window meet the preset safety and delivery position relationship conditions. The area of ​​the exterior facade where the adsorption initiation point is located satisfies the adsorption conditions. The exterior facade area where the adsorption initiation point is located meets the condition of being free of obstacles; The estimated movement path from the adsorption starting point to the target window meets the preset path efficiency optimization conditions.

5. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 4, characterized in that: In step S3, the specific steps for attaching the drone to the wall are as follows: S301. Control the drone to move towards the adsorption starting point with a preset attitude and speed, and prepare to start the adsorption function. S302. When the distance to the building facade is detected to reach the preset contact condition, the adsorption function is activated. S303. Monitor the adsorption force. When the adsorption force reaches and remains at the preset safety conditions, it is determined that the adsorption is stable.

6. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 5, characterized in that: In step S303, if the adsorption force fails to reach the preset safety condition, it is determined that the adsorption has failed, and an emergency recovery process is executed. The emergency recovery process includes: Control the drone to detach from the building facade and remove it from its attachment. Control the drone to return to a safe hovering state; Reassess and select a new adsorption starting point, and try the conversion again; if the number of consecutive failures exceeds the set number, an alarm will be sent to the background scheduling system and manual instructions will be awaited.

7. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 1, characterized in that: In step S4, the delivery conditions are determined through environmental perception, and the item delivery operation is executed when the conditions are met. Specifically, this includes: S401. Detect the open state of the target window through environmental perception; S402. If the target window is detected to be fully open, then perform the delivery operation. S403. If the target window is not fully opened, a prompt message to open it is sent to the user terminal, and the delivery operation is performed after receiving the user's confirmation instruction.

8. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 7, characterized in that: In step S403, if no user confirmation instruction is received within a preset waiting time, the drone is controlled to place the item at a preset emergency delivery location outside the window and send a notification to the user.

9. The air-to-ground switching drone delivery method for building power outage emergency as described in claim 8, characterized in that: In S5, when controlling the drone to detach from the building facade, the detachment action is dynamically adjusted according to the ambient wind speed.

Images