Intelligent unmanned aerial vehicle food material distribution method and system

Abstract

The embodiment of the invention discloses an intelligent unmanned aerial vehicle food material delivery method and system, and the method comprises the following steps: obtaining to-be-delivered order information and parameter information of an unmanned aerial vehicle, determining a target delivery strategy according to the to-be-delivered order information and the parameter information of the unmanned aerial vehicle, and carrying out the food material delivery according to the target delivery strategy. Real-time environment information images of the unmanned aerial vehicle are obtained in real time in the distribution process; and comparing the real-time environment information image with an initial environment information image to obtain incremental image information, and adjusting a target distribution strategy or controlling an unmanned aerial vehicle to perform intelligent obstacle avoidance according to the incremental image information. According to the method, global distribution path planning and local intelligent obstacle avoidance are combined, manual frequent intervention and adjustment are not needed, the autonomy and efficiency of unmanned aerial vehicle distribution are improved, the method can adapt to complex scenes, the environment adaptability is higher, the safety of the unmanned aerial vehicle can be guaranteed, and food distribution can be efficiently completed.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of drone delivery, and particularly to a smart drone food delivery method and system. Background Art

[0002] With the fast pace of modern consumers' lives, the requirements for the timeliness and convenience of food delivery are getting higher and higher. Consumers hope to obtain fresh food quickly, and the demand for efficient delivery in the fresh food e-commerce market is also becoming increasingly urgent. For example, in Changsha, Zhihang Feigou Technology uses drones to deliver pre-cut vegetables, and the pre-cut vegetables can be delivered to consumers within 15 - 20 minutes. Traditional ground delivery methods are difficult to ensure delivery timeliness and the freshness of food when facing problems such as traffic congestion. Drone delivery provides a new delivery solution for consumers and fresh food e-commerce.

[0003] However, drone delivery also faces some challenges. For example, with the increasingly complex flight environment, there are many unpredictable threats or obstacles during the execution of tasks by drones. It is very necessary to design a smart drone delivery method to ensure the safety of drones and enable them to complete the food delivery task. Summary of the Invention

[0004] To overcome the deficiencies of the prior art, the purpose of the embodiments of the present invention is to provide a smart drone food delivery method and system, an electronic device, and a storage medium. By combining global delivery path planning and local intelligent obstacle avoidance, it is not necessary for humans to frequently intervene and adjust, which improves the autonomy and efficiency of drone delivery, can adapt to complex scenarios, has stronger environmental adaptability, can ensure the safety of drones, and can efficiently complete the food delivery.

[0005] To solve the above problems, the first aspect of the embodiments of the present invention discloses a smart drone food delivery method, which includes the following steps:

[0006] Obtain the information of the order to be delivered and the parameter information of the drone. The information of the order to be delivered includes the weight of the food, the type of the food, and the delivery destination;

[0007] Determine the target delivery strategy according to the information of the order to be delivered and the parameter information of the drone. The target delivery strategy includes the drones corresponding to each delivery destination and the corresponding drone delivery paths;

[0008] Conduct food delivery according to the target delivery strategy, and obtain the real-time environmental information image of the drone during the delivery process;

[0009] Compare the real-time environmental information image with the initial environmental information image to obtain incremental image information, and adjust the target delivery strategy or control the drone to perform intelligent obstacle avoidance according to the incremental image information.

[0010] Preferably, determining the target delivery strategy according to the to-be-delivered order information and the drone parameter information includes:

[0011] Determine the initial delivery strategy according to the to-be-delivered order information and the drone parameter information, input the initial delivery strategy data into the trained safety scoring model, and obtain the safety score;

[0012] When the safety score does not reach the preset threshold, adjust the delivery strategy until the safety score of the current delivery strategy reaches the preset threshold, and perform food delivery according to the current delivery strategy.

[0013] Preferably, determining the initial delivery strategy according to the to-be-delivered order information and the drone parameter information includes the following steps:

[0014] Establish a model with the constraints of the drone's load-bearing, the drone's flight speed, and the minimum freshness of the food, aiming at minimizing the total delivery cost, and solve for the initial delivery strategy through the adaptive ant colony algorithm.

[0015] Preferably, obtain the minimum freshness of the food according to the weight and type of the food and the delivery destination, and obtain the constraints of the drone's load-bearing and the drone's flight speed according to the parameter information of the drone.

[0016] Preferably, adjusting the target delivery strategy or controlling the drone to perform intelligent obstacle avoidance according to the incremental image information includes:

[0017] Identify the obstacles in the environmental information image, determine whether the obstacles meet the preset conditions. When the identified obstacles meet the preset conditions, adjust the target delivery strategy; when the identified obstacles do not meet the preset conditions, control and adjust the flight attitude and flight speed of the drone for intelligent obstacle avoidance.

[0018] Preferably, identifying the obstacles in the environmental information image and determining whether the obstacles meet the preset conditions includes:

[0019] Input the incremental image information into the trained obstacle recognition model. Through the obstacle recognition model, determine the category and size of the obstacle recognition model. When the category of the obstacle recognition model is the preset category and the size is greater than the preset size, it is determined that the obstacle meets the preset conditions.

[0020] Preferably, when the identified obstacles do not meet the preset conditions, controlling and adjusting the flight attitude and flight speed of the drone for intelligent obstacle avoidance includes: identifying the orientation and speed of the obstacle, and adjusting the flight attitude and flight speed according to the orientation and speed of the obstacle.

[0021] In a second aspect of the embodiments of the present invention, a smart drone food delivery system is disclosed. The system includes:

[0022] A parameter acquisition unit, configured to acquire the information of the order to be delivered and the parameter information of the drone. The information of the order to be delivered includes the weight of the food ingredients, the types of food ingredients, and the delivery destination.

[0023] A delivery strategy unit, configured to determine a target delivery strategy according to the information of the order to be delivered and the drone parameter information. The target delivery strategy includes the drones corresponding to each delivery destination and the corresponding drone delivery paths.

[0024] An environmental information image unit, configured to perform food ingredient delivery according to the target delivery strategy and acquire the real-time environmental information image of the drone in real time during the delivery process.

[0025] A control and adjustment unit, configured to compare the real-time environmental information image with the original environmental information image to obtain incremental image information, and adjust the target delivery strategy or control the drone to perform intelligent obstacle avoidance according to the incremental image information.

[0026] In a third aspect of the embodiments of the present invention, an electronic device is disclosed, which includes: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the smart drone food delivery method disclosed in the first aspect of the embodiments of the present invention.

[0027] In a fourth aspect of the embodiments of the present invention, a computer-readable storage medium is disclosed, which stores a computer program. The computer program causes a computer to execute the smart drone food delivery method disclosed in the first aspect of the embodiments of the present invention.

[0028] Compared with the prior art, the beneficial effects of the embodiments of the present invention are as follows:

[0029] The method of the present invention determines a target delivery strategy based on the to-be-delivered order information and the drone parameter information. The target delivery strategy includes the drones corresponding to each delivery destination and the corresponding drone delivery paths. Then, the real-time environmental information image during the delivery process is compared with the initial environmental information image to obtain incremental image information. When the incremental image information meets the preset conditions, the delivery path is globally optimized; when the incremental image information does not meet the preset conditions, local intelligent obstacle avoidance is performed. The method of the present invention considers the global optimum, combines real-time adjustment, and through global path planning, can evaluate and plan in advance the possible risks, and try to avoid dangerous areas as much as possible. The local intelligent obstacle avoidance, on the other hand, serves as a real-time protection mechanism that can quickly respond in case of emergencies and avoid collisions between the intelligent agent and obstacles, thereby reducing the risk of system failures or damages and improving the overall operational stability. The combination of path planning and obstacle avoidance enables the intelligent agent to autonomously respond to various environmental situations without frequent manual intervention and adjustment, enhancing the autonomy and efficiency of drone delivery, being able to adapt to complex scenarios, and having stronger environmental adaptability.

[0030] Moreover, the delivery method of the present invention adjusts the target delivery strategy or controls the drone to perform intelligent obstacle avoidance according to the incremental image information, and only performs local intelligent obstacle avoidance when the identified obstacles do not meet the preset conditions, which is beneficial to reducing the collision risks caused by possible misjudgments or missed judgments of the intelligent obstacle avoidance algorithm in complex environments and insufficient detection ranges or detection angles of drone sensors. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a schematic flowchart of a method for intelligent drone food delivery provided by an embodiment of the present invention;

[0032] Figure 2 is a schematic structural diagram of a system for intelligent drone food delivery provided by an embodiment of the present invention;

[0033] Figure 3 is a schematic structural diagram of an electronic device disclosed by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] This specific embodiment is only an explanation of the embodiments of the present invention, and it does not limit the embodiments of the present invention. After reading this specification, those skilled in the art can make modifications without creative contributions to this embodiment as needed, but as long as they are within the scope of the claims of the embodiments of the present invention, they are protected by the patent law.

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the scope of protection of the embodiments of the present invention.

[0036] The term "including" and any variations thereof in the specification and claims of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0037] In the embodiments of the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present invention should not be construed as being more preferred or having more advantages than other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0038] Embodiment 1

[0039] Please refer to Figure 1 As shown, a method for delivering food ingredients by a smart drone, as Figure 1 shown, includes the following steps:

[0040] Step S1: Obtain the information of the order to be delivered and the parameter information of the drone. The information of the order to be delivered includes the weight of the food ingredients, the types of food ingredients, and the delivery destination.

[0041] In this step, when the information of the order to be delivered includes the weight of the food ingredients, the types of food ingredients, and the delivery destination, the information of the weight of the food ingredients, the types of food ingredients, and the delivery destination can be directly read. The parameter information of the drone includes, but is not limited to, parameters such as the drone model, the upper limit of the drone's load capacity, and the flight speed of the drone.

[0042] In specific implementation, the delivery destination can be set by a drone airport or a dedicated receiving point for receiving the delivered food ingredients, and the global positioning system (GPS), Beidou satellite navigation system, etc. are used to accurately locate the coordinates of the landing point.

[0043] The information of the order to be delivered is not limited to the order information that can directly read the weight of the food ingredients and the types of food ingredients, but can also be food information. For example, the information of the order to be delivered is food information such as dish information, biscuits, and calorie information.

[0044] When the information of the order to be delivered is food information or information such as calories, the weight and type of ingredients can be obtained through a machine learning system.

[0045] Specifically, customer historical order data, ingredient supplier information, food cooking manual data, calories and nutrition contained in ingredients, etc. can be used as training data for the neural network model, and ingredient type, ingredient quantity, ingredient supplier information, etc. can be used as output data, so as to select ingredients that can better meet customer needs.

[0046] Step S2: Determine a target delivery strategy according to the information of the order to be delivered and the drone parameter information, where the target delivery strategy includes the drones corresponding to each delivery destination and the corresponding drone delivery paths;

[0047] As an embodiment, determining the target delivery strategy according to the information of the order to be delivered and the drone parameter information may include the following steps:

[0048] According to the delivery destination coordinates of the information of the order to be delivered and the position coordinates of the delivery starting point; based on the delivery area map data, generate several delivery paths of the drone;

[0049] Among them, the delivery starting point can be the warehouse of the delivery company, or the production factory or the farmland address, etc.

[0050] Determine the weight and type of ingredients in the information of the order to be delivered, and screen the delivery paths that meet the order delivery time and the load capacity and endurance of the drone. The types and weights of ingredients for different orders are different, and the ingredient weight needs to be matched according to the load capacity of the drone to ensure that the drone can safely carry the goods for delivery. The delivery time requirements for different types of ingredients are different: some orders may have urgent delivery requirements, and drones with faster flight speeds and endurance need to be arranged for delivery first. At the same time, the endurance time and flight range of the drone determine the distance and scope it can fly. According to the distance and task volume of the delivery destination, select a drone with sufficient endurance to ensure that the delivery task can be completed and the drone can return.

[0051] As another specific embodiment, step S2 includes:

[0052] S21: Establish a model with the constraints of drone load, drone flight speed, and minimum freshness of ingredients, and take the minimum total delivery cost as the goal, and solve it through an adaptive ant colony algorithm to obtain an initial delivery strategy.

[0053] Specifically, obtain the minimum freshness of ingredients according to the ingredient weight, ingredient type, delivery destination, and historical order data, and obtain the drone load and drone flight speed constraints according to the parameter information of the drone.

[0054] As a specific embodiment of the present invention, the minimum freshness of the food ingredients can be obtained according to the weight of the food ingredients, the types of food ingredients, the delivery destination, and the historical order data by the following method:

[0055] Specifically, a minimum freshness model of food ingredients is established, and the preprocessed historical order data is divided into a training set and a test set according to a certain ratio (such as 70% training set and 30% test set). The selected model is trained using the training set data. By inputting information such as the weight of the food ingredients, the types of food ingredients, and the delivery destination of a new order into the trained model, the minimum freshness of the food ingredients required for this order can be predicted.

[0056] Among them, the historical order data can be preprocessed by the following method:

[0057] Data acquisition: Extract historical order data containing information such as the weight of the food ingredients, the types of food ingredients, the delivery destination, the delivery time, the freshness of the food ingredients at the time of order delivery, and customer evaluations from the order management system.

[0058] In specific implementation, the freshness of the food ingredients at the time of order delivery is obtained by analyzing the historical order data. For example, by analyzing the freshness status of the food ingredients when they reach the customers under different delivery times and environmental conditions, the minimum freshness standard that can ensure customer satisfaction is summarized.

[0059] For example, a certain delivery company has been delivering vegetables for a long time. By sorting and analyzing the customer feedback and food ingredient detection data of a large number of past delivery orders, it is found that when the chlorophyll content of the vegetables is not less than 80% of the initial value and there is no obvious wilting and discoloration in appearance, the customer satisfaction with the freshness of the vegetables is relatively high. Therefore, this can be used as a reference for the minimum freshness of vegetable delivery.

[0060] For example, the analyzed minimum freshness standards are as follows:

[0061] Leafy vegetables: Such as spinach, lettuce, etc., the chlorophyll content should generally not be less than 75%-80% of the initial value, and the water loss rate should be controlled within 10%-15% to ensure that there is no obvious wilting and yellowing of the leaves.

[0062] Berry fruits: Such as strawberries, blueberries, etc., there should be no obvious soft rot and discoloration in appearance, the good fruit rate should reach more than 90%-95%, and the decrease in the soluble solid content should not exceed 15%-20%.

[0063] Among them, the freshness of the ingredients feedback by customers can be obtained through real-time monitoring experiments. The sensor technology is used to monitor the freshness change of the ingredients during the distribution process. For example, intelligent tags or sensors are integrated into the packaging to collect data such as the temperature, humidity, and gas composition of the ingredients in real time, and the data is transmitted to the monitoring terminal through wireless communication technology. By analyzing these real-time data, the dynamic change of the freshness of the ingredients can be understood, and then the minimum freshness level that the ingredients need to maintain during the whole distribution process can be determined.

[0064] Data cleaning: Remove duplicate, incorrect or incomplete data records to ensure the accuracy and consistency of the data. Obtain historical data of previous similar ingredient distributions.

[0065] Among them, the minimum freshness model of the ingredients can adopt a linear regression model, a decision tree model or a random forest model, etc. As a specific embodiment, the decision tree model is selected for the minimum freshness model of the ingredients. By recursively dividing the feature space, a decision tree structure is constructed. Each internal node represents a test on a feature, each branch represents the test output, and each leaf node represents a minimum freshness value of the ingredients.

[0066] Among them, the method for selecting features is specifically described as follows:

[0067] Ingredient weight: Directly used as a continuous feature. Ingredients with different weights may have different freshness change characteristics under the same conditions. For example, heavier ingredients may be different from lighter ingredients in terms of internal heat dissipation, etc.

[0068] Ingredient type: The preservation requirements and freshness decline rates of different ingredients vary greatly. For example, the preservation conditions of meat and vegetables are completely different. After encoding the ingredient type, it is used as a feature to input into the model. For example, one-hot encoding is used to convert it into numerical data.

[0069] Delivery destination: Factors such as climate and geographical environment in different regions will affect the environmental conditions during the distribution process, and thus affect the freshness of the ingredients. Similarly, the delivery destination information is encoded and used as a feature.

[0070] Delivery distance: Calculate the distance from the delivery destination data to the starting point of the distribution. It is an important factor affecting the delivery time and the freshness of the ingredients.

[0071] Delivery time: Calculated based on the order time and delivery time in historical orders, which reflects the duration of the ingredients during transportation.

[0072] In specific implementation, the model parameters are adjusted to minimize the error between the predicted value and the actual value. For the decision tree model, methods such as cross-validation can be used to select the optimal tree structure and parameters.

[0073] Specifically, the load constraint of the drone limits the upper weight limit of each food transportation, the flight speed constraint determines the delivery time, affects the timeliness cost, and the minimum freshness constraint of the food requires that the freshness of the food be guaranteed not to be lower than the set threshold during the delivery process, which is closely related to the delivery time.

[0074] Set the objective function

[0075] Among them, c1: the cost per unit flight distance of the drone; c2: the cost of food freshness loss per unit time; c3: the penalty cost per unit overloaded weight if the drone is overloaded; C: the maximum load capacity of the drone; m i : the weight of the i-th type of food; f i : the initial freshness of the i-th type of food. λ i : the decay rate of the freshness of the i-th type of food over time; F min : the minimum freshness that the food needs to guarantee; n: the number of food types in the delivery task; d: the delivery distance; x ij : indicates whether the i-th type of food is included in the j-th delivery task; x ij = 1 means included, x ij = 0 means not included; t: the total time to complete the delivery task; i and j are natural numbers.

[0076] Among them, the drone load constraint is satisfied:

[0077] For each delivery task j, the total weight of the loaded food cannot exceed the maximum load capacity of the drone;

[0078] The drone flight speed constraint is satisfied:

[0079] The drone delivery time is determined by the delivery distance and the flight speed. v is the flight speed of the drone. At the same time, t ≤ T needs to satisfy that the delivery time is within the allowed maximum delivery time, and T is the allowed maximum delivery time.

[0080] The minimum food freshness constraint is satisfied:

[0081] f i -λ i t ≥ F min , for each type of food i, its freshness at the end of the delivery should be greater than or equal to the minimum freshness requirement.

[0082] Solve the above equations through the adaptive ant colony algorithm to minimize the objective function Z, the total delivery cost, so as to obtain the delivery route when the total delivery cost is the smallest and the drones corresponding to each delivery destination.

[0083] Specifically, initialize the parameters, where the pheromone heuristic factor α = 1, the pheromone evaporation coefficient ρ = 0.1,

[0084] the evaporation coefficient β of the heuristic function = 5; the maximum number of iterations N = 100; initialize the pheromone τ ij = 0.4, set the number of ants m. The distance matrix is determined according to the distance between the order delivery point and the delivery destination point; construct a solution. Each ant starts from the delivery starting point and selects the next delivery node according to the state transition probability formula. The state transition probability satisfies the following formula:

[0085]

[0086] where, represents the "distance" from the current node to node i, which affects the increase in cost brought by delivering this ingredient. allowed k is the set of delivery nodes not visited by ant k; η is represents the expected heuristic value from node i to node s, reflecting the quality of the path; is the set of optional nodes allowed for ant k k For all nodes s in, calculate the sum of the weighted products of the pheromone and the expected heuristic value from node i to node s.

[0087] During the selection process, it is necessary to ensure that the total weight of the selected ingredients does not exceed the carrying weight of the drone, and meet the delivery time and ingredient freshness constraints. If no node that meets the conditions can be selected, then backtrack and adjust the previous selection, and update the pheromone. After all ants complete the construction of a delivery path, update the pheromone according to the delivery cost corresponding to the path taken by each ant. The pheromone update formula is:

[0088]

[0089] where, τ ij (t) is the pheromone concentration of the path from city i to city j at time t, τ ij (t + 1) is the concentration in the next round, represents the amount of pheromone left by the k-th ant on the path (i, j), where, Q is a constant, is the delivery cost corresponding to the path taken by the k-th ant. The lower the delivery cost, the more pheromone the ant leaves, and the greater the probability that subsequent ants will choose this path. Adaptively adjust the parameters. According to the relationship between the current number of iterations and the maximum number of iterations, dynamically adjust the pheromone evaporation factor ρ, the pheromone heuristic factor α, and the evaporation coefficient β of the heuristic function.

[0090] For example, in the initial stage of iteration, α can be appropriately increased to make the algorithm more inclined to select paths using existing pheromones and accelerate the convergence speed; in the later stage of iteration, β is increased to enhance the local search ability of the algorithm and avoid falling into local optima. For the termination condition judgment, if the maximum number of iterations N is reached, the algorithm is terminated and the optimal delivery strategy is output; otherwise, the path table passed by the ants in this round is cleared and the iteration continues.

[0091] Step S22: Input the initial delivery strategy data into the trained safety scoring model to obtain a safety score.

[0092] As a specific embodiment, the safety scoring model adopts a random forest regression model. The input is various safety-related features of the UAV delivery system, and the output is a comprehensive safety score (such as from 0 to 10 points). Through the training data, the model can learn the relationship between features and scores, so as to predict the scores of new data.

[0093] Specifically, the various safety-related features of the UAV delivery system may include features in five categories: flight safety, food safety, environmental adaptability, equipment reliability, and operation compliance. The specific input features are as follows in the table:

[0094] Table 1: Input Feature Table

[0095]

[0096]

[0097] The output label is the comprehensive safety score, ranging from 0 to 10 points. The score can be obtained by setting the weights of each input feature and calculating by expert evaluation or historical data. The output label is the safety score, ranging from 0 to 10 points.

[0098] Step 23: When the safety score fails to reach the preset threshold, adjust the delivery strategy until the safety score of the current delivery strategy reaches the preset threshold, and perform food delivery according to the current delivery strategy.

[0099] As a specific embodiment, the preset threshold can be set to 8. When the safety score is greater than or equal to 8, it indicates that the path has relatively high safety under the current conditions, and the UAV can normally execute the delivery task without adjusting the strategy. The current delivery strategy is the target delivery strategy.

[0100] In this embodiment, the present invention performs delivery by presetting a target delivery strategy, comprehensively considering the delivery cost, globally planning the delivery route of the drone, and then performing a safety score, taking into account various factors within the entire flight area, such as terrain, obstacles, no-fly zones, etc., improving the safety of delivery. While greatly reducing the delivery cost, it also improves the delivery safety, and plans a delivery strategy for the drone with high safety and low cost.

[0101] Step S3: Perform food delivery according to the target delivery strategy, and obtain the real-time environmental information image of the drone in real time during the delivery process;

[0102] In specific implementation, the environmental information image can be collected through the sensors carried by the drone itself. For example, it can be collected through a vision sensor. The drone uploads the collected real-time environmental information image to the cloud platform, and the cloud platform can store, analyze, and process these data.

[0103] Step S4: Compare the real-time environmental information image with the initial environmental information image to obtain incremental image information, and adjust the target delivery strategy or control the drone to perform intelligent obstacle avoidance according to the incremental image information.

[0104] In this embodiment, the initial environmental information image is the environmental information image before the drone delivery. Specifically, the initial environmental parameters can be set through professional map software, such as the coordinates of the delivery destination, the general terrain and landform types of the flight area, the roads, buildings, water areas, etc. in the flight area for drawing; it can also obtain the detailed image of the flight area through software such as GIS system or online map, for example, the environmental information image including altitude, slope, aspect, etc.

[0105] Among them, the incremental image information may include the variable image information during the drone delivery process. For example, the variable image information caused by sudden situations due to some environmental factors, such as newly cut and piled big trees, or buildings, wires and cables omitted in the initial environmental information image.

[0106] Specifically, comparing the real-time environmental information image with the initial environmental information image to obtain incremental image information may include:

[0107] Step S41: Preprocess the real-time environmental information image and the initial environmental information image respectively, such as size adjustment and normalization processing, to ensure that the real-time environmental information image and the initial environmental information image have the same size and brightness.

[0108] Step S42: Perform feature extraction and feature matching on the preprocessed real-time environmental information image and the initial environmental information image respectively.

[0109] Specifically, feature extraction can extract key points and corresponding feature descriptors in the image through an algorithm. For example, points with unique features in the image, such as corner points, edge points, etc. Feature matching can calculate the Euclidean distance between feature descriptors after extracting the features of two images.

[0110] Step S43: For the corresponding point pairs obtained by feature matching, calculate geometric transformation matrices such as affine transformation and perspective transformation to align the real-time image, so that the real-time environmental information image and the initial environmental information image coincide as much as possible in space.

[0111] Step S44: After the images are aligned, calculate the difference between the corresponding pixels of the two images, and use the difference image as the incremental image information. Among them, the pixel area with a larger difference indicates that the area has changed between the real-time image and the initial image.

[0112] Specifically, adjusting the delivery strategy or controlling the drone for intelligent obstacle avoidance according to the incremental image information includes:

[0113] Identify obstacles in the environmental information image, and determine whether the obstacles meet the preset conditions. When the identified obstacles meet the preset conditions, adjust the target delivery strategy;

[0114] When the identified obstacles do not meet the preset conditions, control and adjust the flight attitude and flight speed of the drone for intelligent obstacle avoidance.

[0115] In this embodiment, adjusting the target delivery strategy can be specifically carried out by first adjusting the initial delivery strategy and then adjusting the target delivery strategy.

[0116] For example, perform corresponding updates and adjustments to the previous adaptive ant colony algorithm, adaptively adjust the pheromone evaporation factor ρ, the pheromone heuristic factor α, and the heuristic function evaporation coefficient β to obtain different initial delivery strategies, and then perform a safety assessment on the initial delivery strategies to select the target delivery strategy that meets the safety requirements.

[0117] Optionally, identifying obstacles in the environmental information image and determining whether the obstacles meet the preset conditions includes:

[0118] Input the incremental image information into the trained obstacle recognition model. Through the obstacle recognition model, determine the category and size of the obstacle recognition model. When the category of the obstacle recognition model is the preset category and the size is greater than the preset size, it is determined that the obstacle meets the preset conditions.

[0119] In specific implementation, the preset size is adjusted according to the flight altitude. For example, the preset size is usually set to 1.5 to 2 times the size of the UAV fuselage. Because if the lateral size of an obstacle is greater than 1.5 to 2 times the size of the UAV fuselage, it may pose a greater threat to the UAV flight and the path needs to be re-planned.

[0120] However, if the UAV is flying at a low altitude, such as less than 5 meters above the ground, due to the limited space for obstacle avoidance, the tolerance for the size of obstacles will decrease. At this time, even an obstacle with a diameter of about 15 cm, such as a tree branch, may require re-planning the path to avoid collision. When flying at a high altitude, such as more than 50 meters, for some small obstacles, the UAV can avoid them by appropriately adjusting its attitude without changing the path.

[0121] Specifically, input the incremental image information into the trained obstacle recognition model. Through the obstacle recognition model, determine the category and size of the obstacle recognition model, including:

[0122] Divide the incremental image information to obtain several sub-images.

[0123] In specific implementation, each sub-image can be expanded outward. After expansion, there is an overlapping part between sub-images to ensure that the cut part can be found in another overlapping sub-image.

[0124] The obstacle recognition model uses a convolutional network. The detection layer of the obstacle recognition model uses multiple preset sizes to detect obstacles of different sizes. Determine which category the detected target belongs to, such as buildings, people, vehicles, animals, etc. The detection layer outputs the probability of each category through activation functions such as softmax, selects the category with the highest probability as the category of the target, and accurately frames the position and size of the target.

[0125] Among them, the training of the detection layer uses cross-entropy loss to measure the difference between the classification result and the true label. The training data uses the category label and bounding box information containing the target. By continuously adjusting the parameters of the detection layer, the loss function is minimized, thereby improving the performance of the detection layer.

[0126] When the recognized obstacle does not meet the preset conditions, control and adjust the flight attitude and flight speed of the UAV for intelligent obstacle avoidance, including:

[0127] Identify the orientation and speed of the obstacle, and adjust the flight attitude and flight speed according to the orientation and speed of the obstacle. Specifically, it may include:

[0128] Determine the orientation of the obstacle relative to the drone based on sensor data. By continuously collecting multiple frames of sensor data, analyze the position changes of the obstacle in different frames and calculate its speed.

[0129] Based on the orientation and speed of the obstacle, establish a corresponding decision-making model to determine the adjustments to the flight attitude and speed that the drone should take.

[0130] Specifically, the decision-making model can output the following decisions: For example, when it is detected that the obstacle is at a relatively far distance or the volume of the obstacle is small, control the drone to make a small-angle turn, generally between 15 - 30°. This fine-tuning allows the drone to bypass the obstacle without significantly deviating from the original flight path. If the obstacle is close or appears suddenly, control the drone to turn by 45 - 90°. When it is detected that there are obstacles below, such as trees, vehicles, etc., the drone can increase the lift to rise the flight altitude so as to cross the obstacle. Also, for example, when the obstacle is in front of the drone and has a high speed, the drone should decelerate in time or change the flight direction to the left or right.

[0131] In specific implementation, the decision-making model can also be implemented using existing technologies.

[0132] For example, it can be implemented through a deep reinforcement learning algorithm. Use a deep neural network to approximate the Q function and learn the optimal obstacle avoidance strategy through trial and error. Specifically, it includes multiple stages such as data preparation, data marking, network construction, model training, and model optimization.

[0133] Data preparation: Use simulation software to construct diverse virtual scenarios, simulate different flight environments and obstacle situations, control the drone to fly in the simulated environment and collect data;

[0134] Data marking: For the collected image or point cloud data, label information such as the position, category, and boundary of the obstacle in it;

[0135] Network construction: Construct a deep Q-network model. By taking the environmental state as the input, the network outputs numerical estimates of different actions, and the drone selects the optimal action based on these numerical values;

[0136] Model training: Use the marked data for model training, and use cross-entropy loss as the loss function. Calculate the loss value according to the loss function, and then calculate the gradient and update the model parameters through the backpropagation algorithm. Continuously repeat this process until the performance of the model reaches a satisfactory level;

[0137] Model optimization: In practical applications, continuously collect new data, continuously optimize and update the model. The incremental learning method can be used to fine-tune the model using new data without retraining the entire model to adapt to different environmental and task requirements.

[0138] In this embodiment, by adjusting the target delivery strategy according to the incremental image information or controlling the drone for intelligent obstacle avoidance, when the identified obstacle meets the preset conditions, the drone delivery path is replanned. When the preset conditions are not met, the flight attitude and flight speed of the drone are controlled to adjust for intelligent obstacle avoidance. Considering the global optimum and combining real-time adjustment, through global path planning, possible risks can be evaluated and planned in advance to avoid dangerous areas as much as possible. Local intelligent obstacle avoidance, as a real-time protection mechanism, can quickly respond in case of emergencies to prevent the agent from colliding with obstacles, thereby reducing the risk of system failure or damage, improving the overall operational stability. The combination of path planning and obstacle avoidance enables the agent to autonomously handle various environmental situations without frequent manual intervention and adjustment, enhancing the autonomy and efficiency of drone delivery, being able to adapt to complex scenarios, and having stronger environmental adaptability.

[0139] Embodiment Two

[0140] Please refer to Figure 2 as shown, a smart drone food delivery system, such as Figure 2 which includes:

[0141] A parameter acquisition unit 210, configured to acquire the information of the order to be delivered and the parameter information of the drone. The information of the order to be delivered includes the weight of the food ingredients, the types of food ingredients, and the delivery destination;

[0142] A delivery strategy unit 220, configured to determine a target delivery strategy according to the information of the order to be delivered and the drone parameter information. The target delivery strategy includes the drones corresponding to each delivery destination and the corresponding drone delivery paths;

[0143] An environmental information image unit 230, configured to perform food ingredient delivery according to the target delivery strategy and acquire the real-time environmental information image of the drone in real time during the delivery process;

[0144] A control and adjustment unit 240, configured to compare the real-time environmental information image with the original environmental information image to obtain incremental image information, and adjust the target delivery strategy or control the drone for intelligent obstacle avoidance according to the incremental image information.

[0145] Embodiment Three

[0146] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of an electronic device disclosed in an embodiment of the present invention. As Figure 3 shown, the electronic device may include:

[0147] A memory 310 storing executable program code;

[0148] A processor 320 coupled to a memory 310;

[0149] Wherein, the processor 320 calls the executable program code stored in the memory 310 and executes some or all of the steps in a smart drone food delivery method in Embodiment 1.

[0150] An embodiment of the present invention discloses a computer-readable storage medium that stores a computer program, wherein the computer program causes a computer to execute some or all of the steps in an automatic adjustment method of an intelligent seat in Embodiment 1.

[0151] An embodiment of the present invention also discloses a computer program product, wherein when the computer program product runs on a computer, it causes the computer to execute some or all of the steps in a smart drone food delivery method in Embodiment 1.

[0152] An embodiment of the present invention also discloses an application publishing platform, wherein the application publishing platform is used to publish a computer program product, and when the computer program product runs on a computer, it causes the computer to execute some or all of the steps in a smart drone food delivery method in Embodiment 1.

[0153] In various embodiments of the present invention, it should be understood that the magnitudes of the sequence numbers of the various processes do not necessarily mean the order of execution is necessarily prior or posterior, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0154] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0155] In addition, in each embodiment of the present invention, the various functional units may be integrated in one processing unit, or each unit may exist physically separately, or two or more units may be integrated in one unit. The integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0156] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several requests for causing a computer device (which can be a personal computer, a server, or a network device, etc., specifically, the processor in the computer device) to execute some or all of the steps of the methods described in the various embodiments of the present invention.

[0157] In the embodiments provided by the present invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0158] Those of ordinary skill in the art can understand that some or all of the steps in the various methods of the embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electrically-erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc memories, magnetic disk memories, tape memories, or any other computer-readable medium that can be used to carry or store data.

[0159] The above has introduced in detail a method, device, electronic device and storage medium for intelligent drone food delivery disclosed in the embodiments of the present invention. Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A smart drone food delivery method, characterized in that: It includes the following steps: Obtaining order information to be delivered and parameter information of the drone, wherein the order information to be delivered includes food weight, food type, and delivery destination; Determine a target delivery strategy based on the order information to be delivered and the drone parameter information, wherein the target delivery strategy includes drones corresponding to each delivery destination and corresponding drone delivery paths; Delivering food according to the target delivery strategy, and acquiring real-time environmental information images of the drone in real time during the delivery process; The real-time environmental information image is compared with the initial environmental information image to obtain incremental image information, and the target delivery strategy is adjusted or the UAV is controlled to perform intelligent obstacle avoidance according to the incremental image information.

2. The intelligent drone food delivery method according to claim 1 is characterized in that: The determining of the target delivery strategy according to the to-be-delivered order information and the drone parameter information includes: Determine an initial delivery strategy based on the order information to be delivered and the drone parameter information, and input the initial delivery strategy data into the trained safety scoring model to obtain a safety score; When the safety score does not reach the preset threshold, the delivery strategy is adjusted until the safety score of the current delivery strategy reaches the preset threshold, and food is delivered according to the current delivery strategy.

3. The intelligent drone food delivery method according to claim 2 is characterized in that: Determining the initial delivery strategy based on the order information to be delivered and the drone parameter information includes the following steps: Based on the constraints of drone load capacity, drone flight speed, and minimum freshness of ingredients, a model was established with the goal of minimizing the total delivery cost, and the initial delivery strategy was obtained through the adaptive ant colony algorithm.

4. The intelligent drone food delivery method according to claim 1, characterized in that: The minimum freshness of the ingredients is obtained according to the weight of the ingredients, the type of ingredients, and the delivery destination, and the drone load and the drone flight speed constraints are obtained according to the parameter information of the drone.

5. The intelligent drone food delivery method according to claim 1, characterized in that: The step of adjusting the target delivery strategy or controlling the drone to perform intelligent obstacle avoidance according to the incremental image information includes: Identify obstacles in the environmental information image and determine whether the obstacles meet the preset conditions. When the identified obstacles meet the preset conditions, adjust the target delivery strategy; when the identified obstacles do not meet the preset conditions, control and adjust the flight attitude and flight speed of the drone for intelligent obstacle avoidance.

6. The intelligent drone food delivery method according to claim 5 is characterized in that: The identifying an obstacle in the environment information image and determining whether the obstacle meets a preset condition includes: The incremental image information is input into a trained obstacle recognition model, and the category and size of the obstacle recognition model are determined through the obstacle recognition model. When the category of the obstacle recognition model is a preset category and the size is greater than a preset size, it is determined that the obstacle meets the preset conditions.

7. The intelligent drone food delivery method according to claim 5, characterized in that: When the identified obstacle does not meet the preset conditions, the flight attitude and flight speed of the UAV are controlled and adjusted to perform intelligent obstacle avoidance, including: identifying the direction and speed of the obstacle, and adjusting the flight attitude and flight speed according to the direction and speed of the obstacle.

8. A smart drone food delivery system, characterized in that: It includes: A parameter acquisition unit, used to acquire the order information to be delivered and the parameter information of the drone, wherein the order information to be delivered includes the weight of the food, the type of food, and the delivery destination; A delivery strategy unit, configured to determine a target delivery strategy according to the to-be-delivered order information and the drone parameter information, wherein the target delivery strategy includes drones corresponding to each delivery destination and corresponding drone delivery paths; An environmental information image unit is used to deliver food according to the target delivery strategy and obtain real-time environmental information images of the drone in real time during the delivery process; The control adjustment unit is used to compare the real-time environmental information image with the original environmental information image to obtain incremental image information, and adjust the target delivery strategy or control the UAV to perform intelligent obstacle avoidance according to the incremental image information.

9. An electronic device, characterized in that: It includes: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the smart drone food delivery method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: It stores a computer program, wherein the computer program enables a computer to execute the smart drone food delivery method described in any one of claims 1-7.

Images