Flight path planning method for optimizing energy consumption of unmanned aerial vehicle

Abstract

The invention discloses a flight path planning method for unmanned aerial vehicle energy consumption optimization, and relates to the technical field of unmanned aerial vehicle energy consumption optimization, and the method comprises the following steps: 1, a multi-sensor fusion system is arranged on an unmanned aerial vehicle, and the multi-sensor fusion system comprises a visual sensor, an IMU inertial measurement unit and a GPS standby module; according to the flight path planning method for optimizing the energy consumption of the unmanned aerial vehicle, through a flight path optimization algorithm based on machine learning in combination with current position information, an original energy consumption optimization flight path and surrounding environment information, a connection path with the shortest connection path from a current offset position to the original flight path and the highest energy efficiency is calculated in real time; an improved path is obtained through optimization according to the current remaining electric quantity of the unmanned aerial vehicle and the target point distance, and it is ensured that the unmanned aerial vehicle can smoothly transition back to the original energy consumption optimized track and continue to fly in an energy-saving mode; the continuity and reliability of task execution of the unmanned aerial vehicle in a complex environment are improved, and the risk of task failure or great energy consumption increase caused by signal interruption is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) energy consumption optimization, and in particular to a trajectory planning method for unmanned aerial vehicle (UAV) energy consumption optimization. Background Art

[0002] With the widespread application of drone technology in various fields, such as warehouse material inventory, geographic surveying and mapping, environmental monitoring, etc., the requirements for drone trajectory planning are also increasing. In a multi-warehouse environment, especially in the warehouse material inventory scenario, drones face complex geographical environments and strict airspace control. On the one hand, the warehouse is full of shelves and the distribution of stored goods is dense and crowded, which brings many challenges to the flight of drones. On the other hand, the warehouse may block the signal transmission between the drone and the ground control station, resulting in signal interruption. Once the signal is interrupted, the drone cannot accurately obtain its own location information and receive instructions from the ground, and it is easy to deviate from the preset energy-saving optimized trajectory. Moreover, when flying in the internal area of ​​the warehouse, the drone also needs to frequently adjust its flight attitude to avoid the goods stored in the warehouse, which not only increases the complexity of the flight, but also leads to increased energy consumption.

[0003] That is to say, in the warehouse inventory scenario, the drone needs to fly according to the energy-optimized trajectory planning method in an environment with complex shelf structures and multiple no-fly zones. However, the signal interruption caused by warehouse obstruction may cause the drone to deviate from the preset energy-optimized trajectory. How to ensure that the drone can quickly return to the optimal trajectory and continue energy-saving flight after the signal is restored. Summary of the invention

[0004] In order to solve the above technical problems, the present invention is implemented by the following technical solutions: A trajectory planning method for optimizing energy consumption of unmanned aerial vehicles, comprising the following steps:

[0005] Step 1: Deploy a multi-sensor fusion system on the drone, which includes a visual sensor, an IMU inertial measurement unit, and a GPS backup module;

[0006] Step 2: The drone flies in the warehouse according to the preset original energy consumption optimization track. When the signal is interrupted due to the warehouse occlusion, the visual sensor obtains the image data of the surrounding warehouse and environment, and the IMU inertial measurement unit continuously records the motion state parameters of the drone;

[0007] Step 3: Use visual SLAM technology to build a three-dimensional local map around the drone based on image data, and combine motion state parameters to measure the relative displacement and attitude change of the drone during signal interruption to obtain the current position information;

[0008] Step 4: When the signal is restored, the current position information is compared with the original energy consumption optimized track, and the shortest connection path from the current offset position back to the original track is solved through the path planning algorithm to obtain the connection path;

[0009] Step 5: Optimize the strategy instructions according to the improved path to obtain the optimized strategy instructions, and perform maneuver adjustments on the UAV based on the optimized strategy instructions so that the UAV can smoothly transition back to the original energy consumption optimized trajectory and continue flying.

[0010] Preferably, in the multi-sensor fusion system, the visual sensor is an omnidirectional camera or a camera device with wide-angle characteristics, which is used to provide visual coverage of the surrounding environment during signal interruption.

[0011] Preferably, the motion state parameters include acceleration, angular velocity and heading information, and are continuously accumulated during signal interruption, so as to assist in estimating the relative position change of the UAV when the visual SLAM technology cannot provide the current position information.

[0012] Preferably, the three-dimensional local map is a sparse point cloud map or a key feature point map constructed by visual SLAM technology during signal interruption, so as to quickly correct the drone position and calculate the path access after the signal is restored.

[0013] Preferably, the process of obtaining the current location information is as follows:

[0014] Establish a data synchronization mechanism between the visual sensor and the IMU inertial measurement unit, and pre-process the image data collected by the visual sensor, including image filtering to remove noise interference and image enhancement to improve feature recognition. At the same time, filter the motion state parameters to remove measurement noise and outliers to improve data quality.

[0015] Use feature extraction algorithms to extract representative feature points, including corner points and edge points, from preprocessed image data; track and match feature points in continuous frame images through optical flow method and feature descriptor matching technology, determine the corresponding relationship between feature points in different image frames, and obtain the motion trajectory of feature points;

[0016] According to the motion trajectory of the feature points and the camera imaging model, the position change of the feature points in the three-dimensional space is calculated by triangulation, and the displacement of the UAV in the horizontal, vertical and depth directions is derived, and the displacement estimation result of the UAV is preliminarily estimated;

[0017] According to the acceleration and angular velocity data recorded by the IMU inertial measurement unit, the inertial navigation algorithm is used to calculate the attitude change of the drone; first, the acceleration data is integrated to obtain the velocity change data, and then the velocity change data is integrated to obtain the displacement change, and the zero-speed update ZUPT is used to correct the error; and the angular displacement of the drone is calculated through the gyroscope data, and the change data of the heading, pitch and roll attitude angles are determined to obtain the attitude estimation result;

[0018] The extended Kalman filter (EKF) or unscented Kalman filter (UKF) is used to fuse the displacement estimation result based on the visual SLAM technology and the attitude estimation result based on the IMU inertial measurement unit to obtain fusion information. During the fusion process, the weights are dynamically adjusted through the covariance matrix according to the uncertainty of the two data sources to improve the accuracy of the fusion result.

[0019] The fused information is used to optimize and adjust the previously estimated relative displacement and attitude changes, and continuously iterate and update them, so that the measurement results are closer to the actual UAV motion state, reducing the deviation caused by single sensor errors or environmental interference, and finally obtaining accurate relative displacement and attitude change data to determine the current position information of the UAV during the signal interruption.

[0020] Preferably, step four includes a trajectory optimization algorithm based on machine learning, which compares the current position information with the original energy consumption optimized trajectory, and calculates in real time the flight path with the shortest and highest energy efficiency from the current offset position back to the original trajectory to obtain the connection path.

[0021] Preferably, step four also includes optimizing the connection path according to the current remaining power of the UAV, the distance to the target point, and the surrounding environmental constraints during the connection path calculation process to obtain an improved path, wherein the surrounding environmental constraints include warehouse distribution, no-fly zone location, and weather conditions; to ensure that the UAV can still maintain the energy consumption optimization characteristics of subsequent navigation after the connection is completed.

[0022] Preferably, the path planning algorithm adopts a multi-objective optimization strategy when calculating the connection path, including the shortest distance, the minimum turning radius, avoiding the no-fly zone and the comprehensive energy consumption index, and solves the flight trajectory that meets the predetermined optimization criteria through weight distribution, wherein the predetermined optimization criteria include the shortest path length, the lowest energy consumption, the flight safety, the smooth attitude adjustment, and the remaining power utilization efficiency criteria.

[0023] Preferably, the process of obtaining the connection path is as follows:

[0024] Import the current position information of the UAV determined during the signal interruption, obtain the original energy consumption optimization track data, integrate the surrounding environment information of the UAV, and build a comprehensive data model including position and environmental constraints; extract key features of the current position and the original energy consumption optimization track, including key navigation points, geometric shape features of the original energy consumption optimization track, and the azimuth and distance features of the current position relative to the original energy consumption optimization track;

[0025] Construct a graph structure representation of the original energy consumption optimization trajectory, with navigation points as nodes, connection relationships as edges and edge weights set, and represent the current position as a special node related to the graph structure; select a deep reinforcement learning algorithm as the type of machine learning algorithm, construct a model structure, and determine the input layer, output layer and hidden layer parameters, where the input layer includes the current position characteristics, the original energy consumption optimization trajectory characteristics, and environmental characteristics; the output layer includes the action sequence or path node sequence of the connection path; collect simulation or historical flight data, train the machine learning model according to the set reward function, and optimize the model parameters to learn the optimal connection strategy; where the reward function uses short path length and low energy consumption as the basis for high rewards;

[0026] The current position information, original energy-optimized track information and environmental information are input into the trained machine learning model to generate a candidate path sequence or action sequence that connects from the current offset position back to the original energy-optimized track. The weighted sum method is used to evaluate and screen the candidate paths based on the target weights of path length, energy consumption and flight safety, and the final connection path is determined by combining factors such as shortest distance, highest energy efficiency, avoiding no-fly zones and adapting to meteorological conditions.

[0027] Preferably, the process of obtaining the improved path is as follows:

[0028] Remaining power evaluation: Based on the UAV flight performance model, the energy consumption of each path segment of the connection path is estimated to determine the current remaining power of the UAV; the path segments that are difficult to support with the current remaining power are screened out and removed from the connection path to obtain a subset of paths with feasible remaining power;

[0029] Target point distance analysis: Calculate the distance from the current position to the original track target point, as well as the distance from each path end point to the target point in the subset of feasible paths with remaining power, and adjust the path based on the relationship between the two distances to avoid affecting the overall energy consumption and flight efficiency due to deviation from the target point due to path selection;

[0030] If the path is within the remaining power range, but will cause the drone to deviate too far from the target point, resulting in a significant increase in the subsequent flight distance, which may affect the overall energy consumption and flight efficiency, the path will be corrected: by inserting a new navigation point in the middle of the path or changing the direction of the path, so that it is closer to the target point of the original track while ensuring that there is sufficient remaining power;

[0031] Processing of surrounding environment constraints: Warehouse distribution processing: With the help of image data or pre-stored warehouse distribution data, the spatial relationship between the connecting path and the warehouse is analyzed. For the path segments that are close to the warehouse and affect flight safety or have high energy consumption due to detours, the flight altitude is adjusted, the direction is changed, and the complex warehouse area is avoided to obtain the path with optimized warehouse distribution. In other words, for the path segments that are close to the warehouse and affect flight safety or have high energy consumption due to detours, the path with optimized warehouse distribution is obtained by adjusting the flight altitude and changing the direction of the path to avoid the complex warehouse area;

[0032] No-fly zone location processing: Based on the no-fly zone location information, check whether the connecting path crosses the no-fly zone; if there is a crossing, use the path planning algorithm to re-plan the path segment outside the no-fly zone boundary, and replace the original path segment that crosses the no-fly zone with a path segment that bypasses the no-fly zone to ensure that the path is legal and the detour energy consumption increment is minimized;

[0033] Meteorological condition processing: Obtain meteorological condition data, including wind direction and wind speed information. For the path segment of headwind flight, calculate the increase in energy consumption of headwind flight according to the aerodynamic performance model of the UAV (such as the drag coefficient change curve under different wind direction angles); if the energy consumption increases too much, adjust the path direction to obtain a circuitous path segment, including: find a more favorable wind direction area for flight, such as adopting a circuitous strategy, flying a distance in the tangential direction of the wind and then turning to the target point direction to reduce the overall energy consumption; at the same time, consider the impact of wind speed on the flight stability of the UAV. For the path segment where the wind speed is too high and the attitude of the UAV is difficult to control, appropriately reduce the flight speed or adjust the flight altitude to ensure flight safety, and accordingly optimize the path length and flight time to maintain energy consumption optimization;

[0034] Based on the path after remaining power evaluation, target point distance analysis and surrounding environment constraint processing; the weighted sum method is used again to optimize the path to determine the improved path that meets the current remaining power of the UAV, target point distance and surrounding environment constraints. This path is the final result obtained in the connection path calculation process and is used for subsequent flight control and navigation operations; when the improved path is determined, the control unit sends adjustment instructions to the flight control system of the UAV, including altitude fine-tuning, heading change, and speed control, so that the UAV can smoothly complete the route correction without drastically increasing energy consumption.

[0035] The present invention provides a trajectory planning method for optimizing energy consumption of unmanned aerial vehicles, which has the following beneficial effects:

[0036] 1. The energy consumption optimization trajectory planning method of the UAV is to deploy a multi-sensor fusion system on the UAV. During the signal interruption period, the visual SLAM technology combined with IMU data can accurately measure the relative displacement and attitude change of the UAV, thereby determining the current position information; this allows the UAV to accurately know its own position in a complex warehouse environment even if the signal is lost, laying the foundation for subsequent trajectory recovery;

[0037] 2. The energy-optimized trajectory planning method for the UAV. When the signal is restored, the trajectory optimization algorithm based on machine learning combines the current position information, the original energy-optimized trajectory, and the surrounding environment information to calculate in real time the shortest and most energy-efficient connection path from the current offset position back to the original trajectory. In the connection path calculation process, the improved path can be further optimized according to the current remaining power of the UAV and the distance to the target point, ensuring that the UAV can smoothly transition back to the original energy-optimized trajectory and continue energy-saving flight. It improves the continuity and reliability of the UAV's mission execution in complex environments, and reduces the risk of mission failure or a significant increase in energy consumption due to signal interruption. It enables the UAV to fly according to the energy-optimized trajectory planning method in the warehouse cargo inventory scenario, where there are many shelf structures and multiple no-fly zones, to solve the problem that the signal interruption caused by warehouse occlusion may cause the UAV to deviate from the preset energy-optimized trajectory, ensuring that the UAV can quickly return to the optimal trajectory and continue energy-saving flight after the signal is restored. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The present invention is a flowchart of a trajectory planning method for optimizing energy consumption of a UAV. DETAILED DESCRIPTION

[0039] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0040] like Figure 1 As shown, the present invention provides a technical solution: a trajectory planning method for optimizing energy consumption of unmanned aerial vehicles, comprising the following steps:

[0041] Step 1: Deploy a multi-sensor fusion system on the drone, which includes a visual sensor, an IMU inertial measurement unit, and a GPS backup module;

[0042] Step 2: The drone flies in the warehouse according to the preset original energy consumption optimization track. When the signal is interrupted due to the warehouse occlusion, the visual sensor obtains the image data of the surrounding warehouse and environment, and the IMU inertial measurement unit continuously records the motion state parameters of the drone;

[0043] Step 3: Use visual SLAM technology to build a three-dimensional local map around the drone based on image data, and combine motion state parameters to measure the relative displacement and attitude change of the drone during signal interruption to obtain the current position information;

[0044] Step 4: When the signal is restored, the current position information is compared with the original energy consumption optimized track, and the shortest connection path from the current offset position back to the original track is solved through the path planning algorithm to obtain the connection path;

[0045] Step 5: Optimize the strategy instructions according to the improved path to obtain the optimized strategy instructions, and perform maneuver adjustments on the UAV based on the optimized strategy instructions so that the UAV can smoothly transition back to the original energy consumption optimized trajectory and continue flying.

[0046] In the multi-sensor fusion system, the visual sensor is an omnidirectional camera or a camera device with wide-angle characteristics, which is used to provide visual coverage of the surrounding environment during signal interruption.

[0047] The motion state parameters include acceleration, angular velocity and heading information, and are continuously accumulated during signal interruption. They are used to assist in estimating the relative position change of the drone when the visual SLAM technology cannot provide the current position information.

[0048] The three-dimensional local map is a sparse point cloud map or key feature point map constructed by visual SLAM technology during signal interruption, so as to quickly correct the drone position and calculate the path access after the signal is restored.

[0049] The process of obtaining the current location information is as follows;

[0050] Establish a data synchronization mechanism between the visual sensor and the IMU inertial measurement unit, and pre-process the image data collected by the visual sensor, including image filtering to remove noise interference and image enhancement to improve feature recognition. At the same time, filter the motion state parameters to remove measurement noise and outliers to improve data quality.

[0051] Use feature extraction algorithms to extract representative feature points, including corner points and edge points, from preprocessed image data; track and match feature points in continuous frame images through optical flow method and feature descriptor matching technology, determine the corresponding relationship between feature points in different image frames, and obtain the motion trajectory of feature points;

[0052] According to the motion trajectory of the feature points and the camera imaging model, the position change of the feature points in the three-dimensional space is calculated by triangulation, and the displacement of the UAV in the horizontal, vertical and depth directions is derived, and the displacement estimation result of the UAV is preliminarily estimated;

[0053] According to the acceleration and angular velocity data recorded by the IMU inertial measurement unit, the inertial navigation algorithm is used to calculate the attitude change of the drone; first, the acceleration data is integrated to obtain the velocity change data, and then the velocity change data is integrated to obtain the displacement change, and the zero-speed update ZUPT is used to correct the error; and the angular displacement of the drone is calculated through the gyroscope data, and the change data of the heading, pitch and roll attitude angles are determined to obtain the attitude estimation result;

[0054] The extended Kalman filter (EKF) or unscented Kalman filter (UKF) is used to fuse the displacement estimation result based on the visual SLAM technology and the attitude estimation result based on the IMU inertial measurement unit to obtain fusion information. During the fusion process, the weights are dynamically adjusted through the covariance matrix according to the uncertainty of the two data sources to improve the accuracy of the fusion result.

[0055] The fused information is used to optimize and adjust the previously estimated relative displacement and attitude changes, and continuously iterate and update them, so that the measurement results are closer to the actual UAV motion state, reducing the deviation caused by single sensor errors or environmental interference, and finally obtaining accurate relative displacement and attitude change data to determine the current position information of the UAV during the signal interruption.

[0056] Step 4 includes a trajectory optimization algorithm based on machine learning, which compares the current position information with the original energy-optimized trajectory, and calculates in real time the shortest and most energy-efficient flight path from the current offset position back to the original trajectory to obtain the connection path.

[0057] Step 4 also includes optimizing the connection path according to the current remaining power of the UAV, the distance to the target point, and the surrounding environmental constraints during the connection path calculation process to obtain an improved path, where the surrounding environmental constraints include warehouse distribution, no-fly zone location, and weather conditions; to ensure that after the connection is completed, the UAV can still maintain the energy consumption optimization characteristics of subsequent navigation.

[0058] The path planning algorithm adopts a multi-objective optimization strategy when calculating the connecting path, including the shortest distance, minimum turning radius, avoiding no-fly zones and comprehensive energy consumption indicators. It solves the flight trajectory that meets the predetermined optimization criteria through weight allocation. The predetermined optimization criteria include shortest path length, lowest energy consumption, flight safety, smooth attitude adjustment, and efficient use of remaining power.

[0059] The process of obtaining the connection path is as follows: first, data preparation and model construction are carried out, and data is imported and fused. First, the current position information of the UAV determined during the signal interruption is imported into the system, and the pre-planned original energy consumption optimization track data and the surrounding environment information of the UAV obtained by sensors are obtained, including warehouse distribution, no-fly zone location, and meteorological conditions; then, these data are fused to build a comprehensive data model containing position and environmental constraints; reflecting the current status of the UAV and the impact of the surrounding environment on the track planning;

[0060] Optimize the track according to the current position and original energy consumption, extract key features, including image features, track features and environmental features; use the omnidirectional camera or wide-angle camera device in the visual sensor to obtain image data, and use the feature extraction algorithm SIFT and SURF to extract representative feature points in the image to obtain image features, including corner points, edges or texture-rich areas in the image;

[0061] Based on the original energy consumption optimized track, the key navigation points and geometric shape features of the original energy consumption optimized track are extracted to obtain the track features; wherein, by analyzing the track of energy consumption optimization of the UAV, the track is first segmented according to the flight phase to obtain the take-off segment, the cruise segment, and the descent segment; then, through the track planning algorithm, the key navigation points in the optimized track are identified, including the turning point, the flight altitude change point, the speed change point, the acceleration mutation point, the starting point, the end point, and the turning point;

[0062] Extract the geometric features of the track: the curvature of the track, the turning angle, and the heading change rate; use the spline interpolation and Bezier curve in the curve fitting technology to smooth the optimized track, remove unnecessary oscillations and sharp changes, and obtain the geometric shape characteristics of the original energy-optimized track; it helps to analyze the smoothness and stability of the track, and can also effectively reduce the energy waste caused by excessive track curvature;

[0063] The surrounding environment information of the drone obtained by sensors: warehouse distribution, no-fly zone location, weather conditions, and the use of data processing algorithms to extract features related to trajectory planning to obtain environmental characteristics;

[0064] The process of obtaining environmental characteristics is as follows: using the barometer, temperature sensor, wind speed sensor and GPS carried by the drone to collect environmental parameters that may affect energy consumption during flight in real time. Environmental parameters include wind speed and direction, temperature, air pressure, humidity, and terrain undulations; and using Kalman filtering, mean filtering, and low-pass filtering to denoise the collected environmental data to improve data quality and ensure the accuracy of feature extraction; at the same time, the denoised environmental data is normalized or standardized to make data from different sources comparable, which is convenient for subsequent analysis and obtains standard environmental data;

[0065] According to the real-time wind speed and direction data of the UAV during flight, a fluid mechanics model or a physics-based wind field modeling algorithm is used to simulate the wind field distribution in the flight area; the CFD computational fluid dynamics simulation model is used to predict the impact of wind on the flight path, and combined with real-time climate data, the impact of climate factors on energy consumption is analyzed through regression analysis or support vector machine and random forest to obtain climate conditions. Subsequently, the flight path is analyzed using the digital elevation model DEM to extract the terrain undulation characteristics, and the terrain analysis algorithm is used to identify the terrain obstacles that may affect the track and the trend of flight altitude changes; thereby extracting geographical environmental characteristics; among which, climate data includes temperature, humidity and air pressure; terrain undulation characteristics include mountains and valleys, and terrain obstacles include mountains and buildings;

[0066] The PCA method is used to reduce the dimension of the standard environmental data and extract the most representative environmental features. K-means is used to cluster the flight path data under different environmental conditions to identify the typical flight mode under specific meteorological or geographical conditions. ARIMA is used to predict the environmental change trend for the wind speed and temperature changes with time series characteristics in the standard environmental data, so as to provide a reference for the dynamic changes of the environment for track planning in advance.

[0067] Combined with environmental characteristics, random forest regression is used to establish the mapping relationship between environmental characteristics and energy consumption. By training the regression model, the energy consumption changes under different environmental conditions can be predicted, thus providing a basis for track optimization. Variance analysis is used to analyze the influence of different environmental characteristics: wind speed, temperature, and air pressure on energy consumption, to find and obtain the key environmental factors that affect track energy consumption.

[0068] Combine data from multiple sensors, including GPS, IMU, anemometer, and barometer, and use Kalman filtering to comprehensively process environmental information to obtain accurate environmental characteristics; and combine external environmental information, including topographic maps and meteorological data obtained by remote sensing satellites or drones, and use image processing and pattern recognition technology to extract environmental characteristics of the flight area, including wind field distribution and weather conditions, to provide external environmental references for track planning; and analyze the relationship between wind speed, wind direction, temperature, and air pressure in environmental characteristics and track changes to obtain the interaction between environmental conditions and tracks, from which track patterns under different environmental conditions are extracted; the track pattern includes that in a strong wind environment, the track needs to be adjusted to reduce the impact of wind on flight; by modeling the interaction between environmental characteristics and track planning, the energy consumption of the flight path is optimized;

[0069] Environmental features include the warehouse’s boundaries, height, internal obstacle distribution, and the location and range of no-fly zones;

[0070] Among them, for the extraction of key navigation points, by analyzing the track data, the key navigation points in the track are identified: the starting point, end point, turning point or marking point of a specific location; the coordinates, altitude and speed information of these points are extracted to provide basic data for subsequent path planning;

[0071] For the extraction of geometric shape features, the line segment fitting and curve fitting in the geometric algorithm are used to fit and analyze the track to extract the geometric shape features of the track; the geometric shape features include the length of the straight segment of the track, the curvature of the curve segment, and the turning point angle; the geometric shape features can reflect the smoothness, complexity and energy consumption characteristics of the track.

[0072] and the azimuth and distance characteristics of the current position relative to the original energy consumption optimized track;

[0073] Constructing the graph structure and determining the algorithm type: Constructing the graph structure representation of the original energy consumption optimization track, with the navigation points in the track as nodes, the connection relationship between the navigation points as edges, and setting the edge weights according to the actual situation; at the same time, the current position of the drone is represented as a special node related to the graph structure; then, the deep reinforcement learning algorithm is selected as the machine learning algorithm type; then the model structure is constructed, and the input layer, output layer and hidden layer parameters are determined; the input layer includes the current position characteristics, the original energy consumption optimization track characteristics, and the environmental characteristics; the output layer includes the action sequence or path node sequence of the connecting path;

[0074] The deep reinforcement learning algorithm is composed of the agent, environment, state, action, reward function, deep neural network DNN and training algorithm Q-Learning. It is combined with the energy consumption optimization trajectory planning requirements of drones to form a complete technical architecture.

[0075] The agent refers to the flight control system responsible for selecting actions based on the current environmental state; the agent's task is to select an appropriate flight path or strategy based on the environmental information collected during the flight: climate, wind speed, terrain, in order to maximize energy efficiency and flight performance;

[0076] The environment is the object that the agent interacts with, and it contains all external factors and conditions. The flight environment of the drone includes but is not limited to: meteorological data: climate factors such as wind speed, air pressure, temperature, and humidity; flight status data: the speed, altitude, heading, attitude, acceleration and other dynamic states of the drone; terrain information: terrain undulations, obstacles, flight areas, etc.; mission requirements: waypoints, flight targets, predetermined paths, etc.; the environment provides the agent with information about the current state and returns feedback (rewards or penalties) based on the agent's actions;

[0077] The state is a description of the environment at a certain moment, including: the flight state of the drone: such as the current position (coordinates), speed, flight altitude, heading, acceleration, etc.; environmental characteristics: such as wind speed, wind direction, temperature, air pressure, etc.; task-related information: such as the location of the target waypoint, task progress, etc.; the state is represented by a vector or tensor, which serves as the input of the neural network for the agent to make decisions;

[0078] Actions are choices that an agent makes in a given state; in drone trajectory planning, the action space may include:

[0079] Flight path adjustment: change flight direction, speed, and heading;

[0080] Flight altitude change: adjust flight altitude to avoid obstacles or optimize energy consumption;

[0081] Track correction: fine-tune the flight path based on real-time environmental changes (such as wind speed changes); the action space can be discrete (such as selecting a specific track) or continuous (such as adjusting flight angle and speed);

[0082] The reward function defines the reward for an agent taking an action in a certain state. In energy-optimized trajectory planning, the reward function is usually closely related to flight energy consumption. It includes: when the trajectory selected by the drone can effectively reduce flight energy consumption, a positive reward is given, that is, a reward for low energy consumption; when the trajectory leads to increased energy consumption, a negative reward is given, that is, a penalty for high energy consumption; when the task is successfully completed (such as reaching the destination or completing a specific flight mission), an additional reward is given; it also includes environmental adaptation rewards: for example, using wind power to fly or adjusting the flight path according to climate change to optimize energy efficiency;

[0083] Among them, the reward function is the driving force of the agent's learning strategy, ensuring that the agent always optimizes in the direction of reducing energy consumption and improving task execution efficiency during flight;

[0084] In deep reinforcement learning, deep neural networks (DNNs) are used to approximate Q-value functions, policy functions or value functions, and process complex environmental state spaces; deep neural network models include: deep Q networks approximate Q-value functions through neural networks. In discrete action spaces, the DQN algorithm optimizes the Q-value function so that the agent can choose the optimal action; policy networks are neural networks used to output action probability distributions; in policy gradient methods, policy networks can directly output the probability of selecting each action in a given state to guide decision-making; Actor-Critic architecture: combines the policy network Actor and the value network Critic; Actor is responsible for generating action strategies, while Critic evaluates the value of the current strategy and provides feedback; by optimizing these two networks at the same time, the Actor-Critic method can learn the optimal strategy more effectively; based on the model contained in the deep neural network, its parameters are optimized through training, so that the agent can make effective decisions in complex state spaces and optimize flight paths and energy consumption;

[0085] The training process of deep reinforcement learning includes the interaction between the agent and the environment, collecting experience and updating the parameters of the neural network through back propagation; the training algorithms include Q-Learning, Policy Gradient, A3C and DDPG; Q-Learning finds the optimal strategy by updating the Q value function, and DQN is a deep learning extension of Q-learning; Policy Gradient directly optimizes the policy function so that the agent can choose the optimal action from the current state; A3C accelerates the training process through multiple parallel agents and optimizes the strategy and value function; DDPG is suitable for deep reinforcement learning algorithms in continuous action space and can be optimized in high-dimensional state space and continuous action space;

[0086] The components of deep reinforcement learning algorithms in drone energy-optimized trajectory planning cover everything from agent interaction, environment modeling, reward design, to deep neural network training and optimization. Through the continuous interaction between agents and the environment, the guidance of reward mechanisms, and the effective application of deep neural networks, deep reinforcement learning can adaptively optimize flight paths in complex dynamic environments, reduce energy consumption, and improve flight efficiency.

[0087] The flight data is collected through actual UAV flight or simulated flight. The flight data includes the position, speed, acceleration, attitude, and energy consumption of the UAV. At the same time, the environmental information during the flight is recorded: wind direction, wind speed, and temperature. Then, according to the mechanical structure and power system characteristics of the UAV, a flight performance model of the UAV is established to simulate the energy consumption of the UAV under different flight conditions. The collected flight data is then input into the machine learning model for training. The model's goal is set to output the optimal connection path based on the current position information, the original energy consumption optimization track, and the surrounding environmental information.

[0088] During the training process, different weight allocations are set to optimize different objectives: path length, energy consumption, and flight safety balance. At the same time, the flight performance model is used to estimate the energy consumption of candidate paths, and energy consumption is used as an optimization goal for model training.

[0089] When the candidate path generated by the model is shorter and has lower energy consumption, a higher reward value is given, otherwise a lower reward value is given; by continuously adjusting the model parameters, the model can learn the optimal connection strategy, that is, it can accurately generate connection paths that meet the requirements under different circumstances;

[0090] Generate candidate path sequences or action sequences: Input the current position information, the original energy-optimized track information and the environmental information into the trained machine learning model. The model will generate candidate path sequences or action sequences from the current offset position back to the original energy-optimized track based on the learned strategy. These candidate paths are generated based on the model's comprehensive consideration of various factors and have different lengths, energy consumption and flight safety.

[0091] Comprehensively evaluate each candidate path by combining the shortest distance, highest energy efficiency, avoiding no-fly zones, and adapting to meteorological conditions. For the shortest distance, directly calculate the actual length of the candidate path. For energy efficiency, estimate the energy consumption based on the flight performance model of the drone and the characteristics of the path. For avoiding no-fly zones, check whether the path crosses the no-fly zone. For adapting to meteorological conditions, consider the impact of wind direction and wind speed on flight energy consumption and stability, including: the increase in energy consumption of flying against the wind and the impact of excessive wind speed on flight attitude.

[0092] The energy consumption period is estimated according to the flight performance model and path characteristics of the drone. The flight performance model based on the drone includes thrust, lift, drag, and aerodynamic characteristics. The dynamic model of the drone is established through the physical characteristics and flight mode of the drone, and the dynamic model reflects the energy consumption changes of the drone under different flight conditions. Specifically, the flight performance model includes the following key factors:

[0093] Relationship between thrust and speed: The thrust requirement of a drone is closely related to flight speed, air density, and aerodynamic drag; the magnitude of thrust determines the energy consumption of the drone, which in turn affects energy efficiency;

[0094] Lift and gravity balance: The lift and gravity balance of a drone affects flight stability and energy efficiency; the lift and drag relationship of a drone at different flight altitudes, as well as changes in lift coefficient during flight, will affect energy consumption;

[0095] Aerodynamic drag and speed: The relationship between the drag of a drone and its flight speed can be modeled using classical aerodynamic theory. Generally, drag increases with increasing flight speed.

[0096] Flight path and flight phase analysis: The flight path is divided into take-off phase, cruising phase, and landing phase, and the energy consumption of each phase is calculated separately; the flight altitude change, route curvature, and flight speed change of the flight path will also affect the energy consumption of the drone;

[0097] During the take-off phase, the drone needs to overcome gravity and accelerate to the flight speed. Therefore, the thrust requirement is large and the energy consumption is also high. The energy consumption during the take-off phase is estimated by calculating the relationship between thrust and take-off speed. During the cruising phase, the drone usually maintains a relatively constant speed and flight altitude, and the energy consumption is relatively stable. Based on the aerodynamic efficiency and flight speed of the drone, the performance model of the drone is used to calculate the energy consumption during the cruising phase. During the landing phase, the energy consumption during the landing phase is estimated by analyzing the changes in thrust and speed of the drone during the landing process.

[0098] Impact of environmental factors: Meteorological conditions and terrain have a significant impact on energy consumption. Airflow, wind speed, temperature, and air pressure will affect the aerodynamic efficiency of the drone, and thus affect energy consumption. Therefore, when estimating energy consumption, the wind speed, wind direction, and air pressure of real-time meteorological data are added, and the height and slope of terrain data are added for correction. By combining the digital elevation model DEM for terrain compensation, the optimal solution for the flight path is calculated to reduce the additional energy consumption caused by terrain changes. Based on the flight performance model and environmental factors, dynamic programming DP and particle swarm optimization PSO are used to optimize the flight path and select the flight path with the lowest energy consumption. The algorithm can dynamically adjust the flight path, flight altitude, and speed according to the real-time flight status and environmental changes to reduce energy loss. The energy consumption model is verified and optimized using the flight simulation tool MATLAB. The energy consumption under different flight conditions is simulated and the energy efficiency performance under different paths and flight conditions is evaluated, thereby verifying the accuracy and practicality of the energy consumption estimation method.

[0099] Through the energy consumption estimation method of the present invention, the energy consumption of the UAV can be accurately predicted in the flight mission planning stage, and the flight path can be optimized based on the flight performance model and environmental factors to maximize the flight energy efficiency; this method not only improves the execution efficiency of the flight mission, but also extends the flight time of the UAV and reduces operating costs.

[0100] To avoid no-fly zones, check whether the path crosses the no-fly zones. The process is as follows: obtain no-fly zone data through the API interface via information released by air traffic control agencies and the Civil Aviation Administration. The no-fly zone data contains the geographical information of the no-fly zones; the no-fly zone data includes polygonal areas, circular areas and composite areas, among which polygonal areas are used to represent no-fly zones with complex boundaries, such as urban airspace and no-fly zones in specific areas; circular areas are often used to represent no-fly zones around airports, represented by the center and radius; composite areas are no-fly zones composed of multiple geometric shapes, which are processed in combination with different area shapes; use the ray method to determine whether the path crosses the polygonal area of ​​the no-fly zone. When the no-fly zone is circular, determine whether the path enters the no-fly zone by calculating the intersection of the path and the circular area; for no-fly zones composed of multiple geometric shapes, use the intersection and union in Boolean operations to determine whether the path crosses the no-fly zone; thereby comprehensively checking whether the path crosses the no-fly zone to obtain the inspection result

[0101] Based on the pre-set target weights of path length, energy consumption, and flight safety, the candidate paths are comprehensively evaluated using the weighted summation method;

[0102] If the path length weight is 0.4, the energy consumption weight is 0.3, and the flight safety weight is 0.3, for the candidate path X, calculate its weighted score (path length score × 0.4 + energy consumption score × 0.3 + flight safety score × 0.3); by comparing the weighted scores of each candidate path, select the path with the highest score as the final connection path; this path achieves a balanced optimization among various goals while meeting the requirements of shortest distance, highest energy efficiency, avoiding no-fly zones, and adapting to meteorological conditions, ensuring that the UAV can smoothly transition back to the original energy consumption optimized trajectory and continue energy-saving flight.

[0103] Among them, the implementation process of building a comprehensive data model is as follows:

[0104] Obtain current location information: During signal interruption, the current location information of the drone is obtained through visual SLAM technology combined with IMU inertial measurement unit. Among them, visual SLAM technology uses visual sensors: omnidirectional cameras to collect image data of surrounding warehouses and environments, and build a three-dimensional local map around the drone;

[0105] At the same time, the IMU inertial measurement unit continuously records the motion state parameters of the drone, including acceleration, angular velocity and heading information; by establishing a data synchronization mechanism between the visual sensor and the IMU inertial measurement unit, the collected data is subjected to image filtering, image enhancement, and motion state parameter filtering, and then feature extraction algorithms, optical flow methods, feature descriptor matching technology, triangulation methods, and inertial navigation algorithms are used to calculate the motion trajectory of feature points, the displacement of the drone in three-dimensional space, and attitude changes, and ultimately obtain accurate current position information;

[0106] Obtaining original energy consumption optimized track data: Obtaining original energy consumption optimized track data from pre-stored track planning data, which contains the flight path information planned by the UAV according to the energy consumption optimization principle under normal flight conditions, including: coordinates of each navigation point in the track, flight direction, and speed setting;

[0107] Collecting surrounding environment information: Collecting warehouse distribution information through pre-stored warehouse layout data, obtaining no-fly zone location information from geographic information systems or airspace management regulations, and obtaining meteorological condition information, including wind direction and wind speed data, through meteorological sensors or connection with meteorological data service platforms;

[0108] Data fusion and model building:

[0109] Data fusion processing: The acquired current position information, original energy consumption optimization track data and surrounding environment information are fused and processed; during the fusion process, the appropriate fusion method is determined according to the characteristics and mutual relationship of the data; including: coordinate conversion and spatial alignment of position information and track data to ensure that they are processed in the same reference system; for environmental information, it is associated with position and track information to analyze the impact of environmental factors on the flight path;

[0110] Construct a comprehensive data model: Based on the fused data, construct a comprehensive data model that includes position and environmental constraints; use a graph structure to use the navigation points in the original energy consumption optimization track as nodes, and the connection relationship between nodes as edges, and set the edge weights according to the characteristic distance and energy consumption of the track; at the same time, represent the current position as a special node related to the graph structure, and integrate the surrounding environmental information: warehouses, no-fly zones, weather conditions; as node or edge attributes into the model; the comprehensive data model constructed in this way can fully reflect the current status of the drone, the target track, and the constraints of the surrounding environment, providing a complete data foundation for the subsequent path planning algorithm, so that the algorithm can solve the optimal connection path while considering multiple factors.

[0111] The process of obtaining the improvement path is as follows:

[0112] According to the UAV flight performance model, the energy consumption of each path segment of the connection path is estimated to determine the current remaining power of the UAV; the path segments that are difficult to support with the current remaining power are screened out and removed from the connection path to obtain a subset of paths with feasible remaining power;

[0113] Calculate the distance from the current position to the original track target point, as well as the distance from each path end point to the target point in the subset of feasible paths with remaining power, and adjust the path according to the two distance relationships; use geometric algorithms to calculate the straight-line distance from the current position to the original track target point and the straight-line distance from each path end point to the target point in the subset of feasible paths with remaining power, and obtain distance information. This distance information can reflect the relative position relationship between the drone and the target point and the advantages and disadvantages of different paths; based on the calculated distance information, select a path that meets the remaining power requirements and is as close to the target point as possible as the final path. If multiple paths meet the conditions, further consider the path length, energy consumption, and flight safety for comprehensive evaluation and selection; after the final path is selected, if the path is within the remaining power range but causes the drone to deviate too far from the target point, it is adjusted by inserting a new navigation point in the middle of the path or changing the direction of the path. The adjusted path should be closer to the target point of the original track while ensuring sufficient remaining power; avoid deviating too far from the target point due to path selection, which affects the overall energy consumption and flight efficiency; let the distance from the current position to the target point of the original track be D1, and the distance from each path end point to the target point in the subset of feasible paths with remaining power be D2; use the current position information of the drone (obtained through visual SLAM technology and IMU data fusion) and the position information of the target point of the original track to calculate D1; in the subset of paths with remaining power, calculate the distance between the position information of each path end point and the position information of the target point of the original track to obtain the set D2; if D1 is less than all D 2, that is, the distance from the current position directly returning to the original track target point is shorter than returning through any subset of feasible paths with remaining power, then the UAV should directly return to the original track along the shortest path as a direct return strategy; if D1 is greater than some D2, that is, there is a situation where the return to the original track target point through a subset of feasible paths with remaining power is shorter, then these paths are analyzed, and the path with the smallest D2 and meeting the energy consumption optimization requirements is selected as the adjusted path as a path optimization strategy; based on the above analysis, the adjusted path is determined, and adjustment instructions are sent to the flight control system of the UAV through the control unit, including altitude fine-tuning, heading change, and speed control, to complete path planning and implementation;

[0114] It should be further explained that, in the specific implementation process, if the path is within the remaining power range, but will cause the drone to deviate too far from the target point, resulting in a significant increase in the subsequent flight distance, which may affect the overall energy consumption and flight efficiency, the path will be corrected: by inserting a new navigation point in the middle of the path or changing the direction of the path, so that it is closer to the target point direction of the original track while ensuring that the remaining power is sufficient;

[0115] Processing of surrounding environment constraints: Warehouse distribution processing: With the help of image data or pre-stored warehouse distribution data, the spatial relationship between the connecting path and the warehouse is analyzed. For the path segments that are close to the warehouse and affect flight safety or have high energy consumption due to detours, the flight altitude is adjusted, the direction is changed, and the complex warehouse area is avoided to obtain the path after warehouse distribution optimization; It should be further explained that in the specific implementation process, for the path segments that are close to the warehouse and affect flight safety (such as the flight altitude is lower than the warehouse altitude and the distance is too close) or the energy consumption is too high due to detours (such as frequent turns), the path after warehouse distribution optimization is obtained by adjusting the flight altitude and changing the direction of the path to avoid the complex warehouse area;

[0116] No-fly zone location processing: Based on the no-fly zone location information, check whether the connecting path crosses the no-fly zone; if there is a crossing, use the path planning algorithm to re-plan the path segment outside the no-fly zone boundary, and replace the original path segment that crosses the no-fly zone with a path segment that bypasses the no-fly zone to ensure that the path is legal and the detour energy consumption increment is minimized;

[0117] Meteorological condition processing: Obtain meteorological condition data, including wind direction and wind speed information. For the path segment of headwind flight, calculate the increase in energy consumption of headwind flight according to the aerodynamic performance model of the UAV (such as the drag coefficient change curve under different wind direction angles); if the energy consumption increases too much, adjust the path direction to obtain a circuitous path segment. It should be further explained that in the specific implementation process, find a more favorable wind direction area for flight, adopt a circuitous strategy, fly a distance in the tangential direction of the wind and then turn to the direction of the target point to reduce the overall energy consumption; at the same time, consider the impact of wind speed on the flight stability of the UAV. For the path segment where the wind speed is too high and the attitude of the UAV is difficult to control, appropriately reduce the flight speed or adjust the flight altitude to ensure flight safety, and optimize the path length and flight time accordingly to maintain energy consumption optimization;

[0118] Based on the path after remaining power evaluation, target point distance analysis and surrounding environment constraint processing; the weighted sum method is used again to optimize the path to determine the improved path that meets the current remaining power of the UAV, target point distance and surrounding environment constraints. This path is the final result obtained in the connection path calculation process and is used for subsequent flight control and navigation operations; when the improved path is determined, the control unit sends adjustment instructions to the flight control system of the UAV, including altitude fine-tuning, heading change, and speed control, so that the UAV can smoothly complete the route correction without drastically increasing energy consumption.

[0119] It should be further explained that, in the specific implementation process, by deploying a multi-sensor fusion system on the drone, during the signal interruption, the relative displacement and attitude change of the drone can be accurately measured by using visual SLAM technology combined with IMU data, so as to determine the current position information; this enables the drone to accurately know its own position in a complex warehouse environment even if the signal is lost, laying the foundation for subsequent track recovery;

[0120] When the signal is restored, the machine learning-based track optimization algorithm combines the current position information, the original energy-optimized track, and the surrounding environment information to calculate in real time the shortest and most energy-efficient connection path from the current offset position back to the original track. In addition, during the connection path calculation process, the improved path can be further optimized based on the current remaining power of the drone and the distance to the target point, ensuring that the drone can smoothly transition back to the original energy-optimized track and continue energy-saving flight. This improves the continuity and reliability of drone missions in complex environments and reduces the risk of mission failure or a significant increase in energy consumption due to signal interruption.

[0121] In the process of trajectory planning, a multi-objective optimization strategy is adopted to comprehensively consider factors such as the shortest distance, minimum turning radius, avoiding no-fly zones, and comprehensive energy consumption indicators, and solve the flight trajectory that meets the predetermined optimization criteria through weight distribution; this enables the UAV to choose a better path during flight, reduce unnecessary flight distance and maneuvers, and thus reduce energy consumption; for example, by reasonably planning the path to avoid headwind areas or reduce the number of turns, the increase in energy consumption caused by air resistance and attitude adjustment is effectively reduced;

[0122] Whether during signal interruption or normal flight, this method can fully consider the impact of surrounding environmental constraints on energy consumption; when faced with changing weather conditions, it can timely adjust the path direction, find a more favorable wind direction area to fly, and maintain energy consumption optimization; at the same time, the path is dynamically adjusted according to the current remaining power of the drone to ensure optimal energy consumption under different power states, thereby improving the drone's endurance and enabling it to complete more tasks with limited power;

[0123] The above technical solution enables the drone to fly according to the energy-optimized trajectory planning method in a warehouse cargo inventory scenario where there are many shelf structures and multiple no-fly zones. This solves the problem of signal interruption caused by warehouse occlusion that may cause the drone to deviate from the preset energy-optimized trajectory, and ensures that the drone can quickly return to the optimal trajectory and continue energy-saving flight after the signal is restored.

[0124] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A trajectory planning method for optimizing energy consumption of unmanned aerial vehicles, characterized in that: The steps include: Step 1: Deploy a multi-sensor fusion system on the drone, which includes a visual sensor, an IMU inertial measurement unit, and a GPS backup module; Step 2: The drone flies in the warehouse according to the preset original energy consumption optimization track. When the signal is interrupted, the visual sensor obtains image data of the surrounding warehouse and environment, and the IMU inertial measurement unit continuously records the drone's motion state parameters; Step 3: Use visual SLAM technology to build a three-dimensional local map around the drone based on image data, and combine motion state parameters to measure the relative displacement and attitude change of the drone during signal interruption to obtain the current position information; Step 4: When the signal is restored, the current position information is compared with the original energy consumption optimized track, and the shortest connection path from the current offset position back to the original track is solved through the path planning algorithm to obtain the connection path; Step 5: Optimize the strategy instructions according to the improved path to obtain the optimized strategy instructions, and perform maneuver adjustments on the UAV based on the optimized strategy instructions, so that the UAV smoothly transitions back to the original energy consumption optimized trajectory and continues to fly; In the process of obtaining the current position information, the feature extraction algorithm is first used to extract representative feature points from the image data. The feature points in the continuous frame images are tracked and matched through the optical flow method and feature descriptor matching technology to determine the corresponding relationship between the feature points in different image frames and obtain the motion trajectory of the feature points. Then, according to the motion trajectory of the feature points and the camera imaging model, the position change of the feature points in the three-dimensional space is calculated through the triangulation method, and the displacement of the drone in the horizontal, vertical and depth directions is derived to obtain the displacement estimation result of the drone. Then, based on the acceleration and angular velocity data recorded by the IMU inertial measurement unit, the inertial navigation algorithm is used to calculate the attitude change of the drone; the angular displacement of the drone is calculated through the gyroscope data, and the change data of the heading, pitch and roll attitude angles are determined to obtain the attitude estimation result; In the process of obtaining the connection path, the current position information of the UAV determined during the signal interruption is first imported, and the original energy consumption optimization track data is obtained at the same time, and the surrounding environment information of the UAV is integrated to build a comprehensive data model containing position and environmental constraints; key features are extracted for the current position and the original energy consumption optimization track, and the key features include the key navigation points of the original energy consumption optimization track, geometric shape features, and the azimuth and distance features of the current position relative to the original energy consumption optimization track; among them, by analyzing the energy consumption optimization track of the UAV, the track is first segmented according to the flight stage to obtain the take-off segment, the cruising segment, and the descent segment; then, through the track planning algorithm, the key navigation points in the optimized track are identified, including turning points, flight altitude change points, speed change points, acceleration mutation points, starting points, end points, and turning points; The geometric features of the track are extracted, including the curvature, turning angle, and heading change rate of the track. Based on the geometric features of the track, the spline interpolation and Bezier curve in the curve fitting technology are used to smooth the optimized track, remove unnecessary oscillations and sharp changes, and obtain the geometric shape characteristics of the original energy consumption optimized track.

2. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 1, characterized in that: In the multi-sensor fusion system, the visual sensor is an omnidirectional camera or a camera device with wide-angle characteristics, which is used to provide visual coverage of the surrounding environment during signal interruption.

3. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 2, characterized in that: The motion state parameters include acceleration, angular velocity and heading information, and are continuously accumulated during signal interruption, and are used to assist in estimating the relative position change of the drone when the visual SLAM technology cannot provide current position information.

4. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 3, characterized in that: The three-dimensional local map is a sparse point cloud map or a key feature point map constructed by visual SLAM technology during signal interruption.

5. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 4, characterized in that: The extended Kalman filter EKF or unscented Kalman filter UKF is used to fuse the displacement estimation result based on visual SLAM technology and the attitude estimation result based on IMU inertial measurement unit to obtain fusion information; the fusion information is used to optimize and adjust the previously estimated relative displacement and attitude change, and it is continuously updated iteratively to finally obtain accurate relative displacement and attitude change data, so as to determine the current position information of the UAV during the signal interruption.

6. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 5, characterized in that: Step 4 includes a trajectory optimization algorithm based on machine learning, which compares the current position information with the original energy-optimized trajectory, and calculates in real time the shortest and most energy-efficient flight path from the current offset position back to the original trajectory to obtain the connection path.

7. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 6, characterized in that: Step 4 also includes optimizing the connection path during the connection path calculation process according to the current remaining power of the drone, the distance to the target point, and the surrounding environmental constraints to obtain an improved path, wherein the surrounding environmental constraints include warehouse distribution, no-fly zone location, and weather conditions.

8. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 7, characterized in that: The path planning algorithm adopts a multi-objective optimization strategy when calculating the connection path, including the shortest distance, the minimum turning radius, avoiding the no-fly zone and the comprehensive energy consumption index, and solves the flight trajectory that meets the predetermined optimization criteria through weight distribution. The predetermined optimization criteria include the shortest path length, the lowest energy consumption, the flight safety, the smooth attitude adjustment and the remaining power utilization efficiency criteria.

9. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 8, characterized in that: The original energy consumption optimization trajectory is constructed through the graph structure, with navigation points as nodes, connection relationships as edges and edge weights set, and the current position is represented as a special node related to the graph structure; the deep reinforcement learning algorithm is selected as the machine learning algorithm type, the model structure is constructed, and the input layer, output layer and hidden layer parameters are determined; simulation or historical flight data is collected, and the machine learning model is trained according to the set reward function, and the model parameters are optimized to learn the optimal connection strategy; among which, the reward function takes short path length and low energy consumption as the basis for high rewards; The current position information, original energy-optimized track information and environmental information are input into the trained machine learning model to generate a candidate path sequence or action sequence that connects from the current offset position back to the original energy-optimized track. The weighted sum method is used to evaluate and screen the candidate paths based on the target weights of path length, energy consumption and flight safety, and the final connection path is determined by combining factors such as shortest distance, highest energy efficiency, avoiding no-fly zones and adapting to meteorological conditions.

10. A method for trajectory planning for optimizing energy consumption of unmanned aerial vehicles according to claim 9, characterized in that: The process of obtaining the improvement path is as follows: Remaining power evaluation: Based on the UAV flight performance model, the energy consumption of each path segment of the connection path is estimated to determine the current remaining power of the UAV; the path segments that are difficult to support with the current remaining power are screened out and removed from the connection path to obtain a subset of paths with feasible remaining power; Target point distance analysis: Calculate the distance from the current position to the original track target point, as well as the distance from each path end point to the target point in the subset of feasible paths with remaining power, and adjust the path based on the relationship between the two distances; Processing of surrounding environment constraints: Warehouse distribution processing: With the help of image data or pre-stored warehouse distribution data, the relationship between the connecting path and the warehouse space is analyzed. For the path sections that are close to the warehouse and affect safety or have high energy consumption, the flight altitude is adjusted, the direction is changed, and the complex warehouse area is avoided to obtain the path with optimized warehouse distribution; No-fly zone location processing: Based on the no-fly zone location information, check whether the connecting path crosses the no-fly zone; if there is a crossing, use the path planning algorithm to re-plan the path segment outside the no-fly zone boundary, and replace the original path segment that crosses the no-fly zone with a path segment that bypasses the no-fly zone; Meteorological condition processing: Obtain meteorological condition data, including wind direction and wind speed information. For the path segment of headwind flight, calculate the increase in energy consumption of headwind flight based on the aerodynamic performance model of the UAV; if the energy consumption increases too much, adjust the path direction to obtain a circuitous path segment; Based on the path after remaining power evaluation, target point distance analysis and surrounding environment constraints; the weighted sum method is used to optimize the path again to determine the improved path that meets the current remaining power of the UAV, target point distance and surrounding environment constraints; when the improved path is determined, the control unit sends adjustment instructions to the flight control system of the UAV, including altitude fine-tuning, heading change, and speed control, so that the UAV can smoothly complete the route correction without drastically increasing energy consumption.

Images