A method and system for optimizing paths for low-altitude logistics terminal collaborative distribution

Abstract

The application relates to the technical field of logistics distribution, in particular to a path optimization method and system for low-altitude logistics terminal collaborative distribution, which comprises the following steps: determining a traffic time redundancy coefficient of each candidate stopping point based on a historical road section speed sequence; determining a hovering power consumption increase coefficient of each candidate stopping point based on surrounding obstacle distribution data; determining a total energy consumption estimation value of a UAV at each candidate stopping point based on the traffic time redundancy coefficient, the hovering power consumption increase coefficient, real-time operation parameters of the UAV, real-time operation parameters of a robot and the position of each candidate stopping point; and determining a target stopping point from the multiple candidate stopping points based on the total energy consumption estimation value of each candidate stopping point. The method can accurately estimate the energy consumption of the UAV and improve the time reliability and energy utilization efficiency of the low-altitude logistics distribution.

Description

Technical Field

[0001] This application relates to the field of logistics and distribution technology, specifically to a path optimization method and system for low-altitude logistics last-mile collaborative distribution. Background Technology

[0002] Low-altitude logistics, as a crucial component of smart cities and on-demand delivery systems, is rapidly developing. Among these, heterogeneous collaborative delivery systems combining drones and robots, leveraging the efficiency advantages of drones' ability to traverse terrain and fly in straight lines, and the convenience of robots' precise building delivery, are considered one of the effective paradigms for solving the "last mile" and even the "last hundred meters" delivery challenges. In this collaborative model, the key to achieving smooth overall system operation, energy efficiency, and mission reliability lies in how to plan an efficient rendezvous and docking point for both drones and robots.

[0003] In existing technologies, a docking point with the shortest connection time is usually calculated based on the robot's average speed and position, as well as the drone's average flight speed and position.

[0004] However, in the above methods, the robot's travel speed is affected by complex road conditions, and it may not be able to reach the docking point as expected. In this case, the drone needs to hover and wait for a long time, which may result in energy waste or even energy depletion, leading to low time reliability and energy utilization efficiency of low-altitude logistics delivery. Summary of the Invention

[0005] To address the technical problem of potential energy waste or even depletion during drone-robot handover processes, where drones need to hover for extended periods, this application aims to provide a path optimization method and system for low-altitude logistics last-mile collaborative delivery. The specific technical solution adopted is as follows: This application provides a path optimization method for low-altitude logistics last-mile collaborative delivery, including: acquiring the location information of multiple candidate stopping points, the historical road segment speed sequence and surrounding obstacle distribution data of each candidate stopping point, the real-time operating parameters of the UAV, and the real-time operating parameters of the robot. The historical road segment speed sequence includes the historical average travel speed of multiple road segments on the path between the candidate stopping point and the robot; determining the travel time redundancy coefficient of each candidate stopping point based on the historical road segment speed sequence of each candidate stopping point, the travel time redundancy coefficient being used to characterize the degree of delay risk of the robot reaching the candidate stopping point; and so on. Based on the obstacle distribution data around each candidate docking point, a hovering power consumption increase coefficient is determined for each candidate docking point. The hovering power consumption increase coefficient is used to characterize the risk of increased power consumption when the UAV hovers at the candidate docking point. Based on the travel time redundancy coefficient, the hovering power consumption increase coefficient, the real-time operating parameters of the UAV, the real-time operating parameters of the robot, and the location of each candidate docking point, the estimated total energy consumption of the UAV at each candidate docking point is determined. Based on the estimated total energy consumption of each candidate docking point, the target docking point is determined from multiple candidate docking points.

[0006] Optionally, the above-mentioned determination of the travel time redundancy coefficient of each candidate stop based on the historical road segment speed sequence of each candidate stop includes: performing frequency analysis on the historical road segment speed sequence of the first candidate stop to obtain the spatial frequency domain energy distribution of the first candidate stop, which includes multiple frequency components and the energy value of each frequency component, wherein the first candidate stop is any one of the multiple candidate stops; determining the traffic flow speed fluctuation ratio of the first candidate stop based on the energy ratio of the high-frequency components in the spatial frequency domain energy distribution, wherein the traffic flow speed fluctuation ratio is used to characterize the degree of traffic speed fluctuation of multiple road segments on the path between the candidate stop and the robot; and determining the travel time redundancy coefficient of the first candidate stop based on the preset time adjustment weight and the traffic flow speed fluctuation ratio.

[0007] Optionally, the above-mentioned determination of the hovering power consumption increase coefficient for each candidate docking point based on the surrounding obstacle distribution data of each candidate docking point includes: determining the obstacle enclosure degree of the first candidate docking point based on the surrounding obstacle distribution data of the first candidate docking point, wherein the obstacle enclosure degree is used to characterize the degree of enclosure of the surrounding environment of the candidate docking point; and determining the hovering power consumption increase coefficient of the first candidate docking point based on the obstacle enclosure degree and a preset power adjustment weight.

[0008] Optionally, determining the obstacle enclosure degree of the first candidate parking point based on the surrounding obstacle distribution data includes: emitting multiple evenly distributed virtual detection rays centered on the first candidate parking point; determining the obstacle distance at which each virtual detection ray first encounters an obstacle within a preset extension distance based on the surrounding obstacle distribution data of the first candidate parking point; and determining the obstacle enclosure degree of the first candidate parking point based on the obstacle distances of the multiple virtual detection rays, the number of the multiple evenly distributed virtual detection rays, and the preset extension distance.

[0009] Optionally, determining the obstacle enclosure degree of the first candidate docking point based on the obstacle distances of multiple virtual detection rays, the number of the multiple uniformly distributed virtual detection rays, and the preset extension distance includes: determining the average obstacle distance based on the sum of the obstacle distances of all virtual detection rays and the number of the multiple uniformly distributed virtual detection rays; and determining the obstacle enclosure degree of the first candidate docking point based on the ratio between the average obstacle distance and the preset extension distance.

[0010] Optionally, the robot's real-time operating parameters include the robot's current position and average speed, and the UAV's real-time operating parameters include standard cruise speed, standard cruise power, nominal hovering power, and current position. The above-mentioned determination of the total energy consumption estimate of the UAV at each candidate docking point, based on the travel time redundancy coefficient for each candidate docking point, the hovering power consumption increase coefficient for each candidate docking point, the UAV's real-time operating parameters, the robot's real-time operating parameters, and the position of each candidate docking point, includes: determining a safe navigation time based on the robot's current position and average speed, the position of the first candidate docking point, and the travel time redundancy coefficient; the safe navigation time being the maximum time for the robot to reach the candidate docking point; determining the flight time for the UAV to reach the first candidate docking point based on the UAV's standard cruise speed and current position, and the position of the first candidate docking point; and determining the total energy consumption estimate for the first candidate docking point based on the safe navigation time, the flight time, the standard cruise power, the nominal hovering power, and the hovering power consumption increase coefficient.

[0011] Optionally, determining the estimated total energy consumption of the first candidate docking point based on the safe navigation time, flight time, standard cruise power, nominal hovering power, and hovering power consumption increase factor includes: determining the potential hovering waiting time of the UAV at the first candidate docking point based on the safe navigation time and flight time, wherein the potential hovering waiting time characterizes the duration for which the UAV hovers at the candidate docking point while waiting for the robot to arrive; determining the flight energy consumption based on the flight time and standard cruise power; determining the hovering energy consumption based on the potential hovering waiting time, nominal hovering power, and hovering power consumption increase factor; and determining the sum of the flight energy consumption and the hovering energy consumption as the estimated total energy consumption of the first candidate docking point.

[0012] Optionally, the above-mentioned determination of the target stop from the plurality of candidate stops based on the total energy consumption estimate of each candidate stop includes: determining the candidate stop with the lowest total energy consumption estimate as the target stop.

[0013] Optionally, the real-time operating parameters of the drone also include the current remaining battery power. The method further includes: determining the product of the current remaining battery power and a preset safety factor as a battery safety threshold; if the estimated total energy consumption at the target docking point is less than or equal to the battery safety threshold, sending a handover command to the drone and the robot respectively, the handover command being used to instruct the handover to take place at the target docking point; if the estimated total energy consumption at the target docking point is greater than the battery safety threshold, sending a handover stop command to the drone and the robot respectively, the handover stop command being used to instruct the current handover to be stopped.

[0014] This application also provides a path optimization system for low-altitude logistics last-mile collaborative delivery, including: a data acquisition module, a risk prediction module, and a stop point determination module; the data acquisition module is used to acquire the location information of multiple candidate stop points, the historical road segment speed sequence and surrounding obstacle distribution data of each candidate stop point, the real-time operating parameters of the UAV, and the real-time operating parameters of the robot, the historical road segment speed sequence including the historical average travel speed of multiple road segments on the path between the candidate stop point and the robot; the risk prediction module is used to determine the travel time redundancy coefficient of each candidate stop point based on the historical road segment speed sequence of each candidate stop point, the travel time redundancy coefficient is used to characterize the delay risk of the robot reaching the candidate stop point. The risk assessment module is used to determine the hovering power consumption increase coefficient for each candidate docking point based on the surrounding obstacle distribution data. The hovering power consumption increase coefficient is used to characterize the risk level of power consumption increase when the UAV hovers at the candidate docking point. The risk assessment module is also used to determine the total energy consumption estimate of the UAV at each candidate docking point based on the travel time redundancy coefficient, the hovering power consumption increase coefficient, the real-time operating parameters of the UAV, the real-time operating parameters of the robot, and the location of each candidate docking point. The docking point determination module is used to determine the target docking point from multiple candidate docking points based on the total energy consumption estimate of each candidate docking point.

[0015] This application has the following beneficial effects: By acquiring historical speed sequences and surrounding obstacle distribution data for candidate docking points, the delay risk of robots on the ground and the hovering energy consumption risk of drones in the air are quantified separately. Based on the real-time operating parameters of both drones and robots, these two risks are mapped to a unified total energy consumption estimate, ultimately determining the target docking point from multiple candidate locations. This more accurate prediction of drone energy consumption improves the time reliability and energy efficiency of low-altitude logistics delivery. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating a route optimization method for low-altitude logistics last-mile collaborative delivery, provided as an embodiment of this application; Figure 2This is a structural diagram of a low-altitude logistics last-mile collaborative delivery route optimization system provided in one embodiment of this application. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a low-altitude logistics last-mile collaborative delivery method and system proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0020] The following description, in conjunction with the accompanying drawings, details a specific scheme for a path optimization method and system for low-altitude logistics last-mile collaborative delivery provided in this application.

[0021] Please see Figure 1 The diagram illustrates a method flowchart for a low-altitude logistics last-mile collaborative delivery route optimization method provided in one embodiment of this application.

[0022] like Figure 1 As shown, the path optimization method for low-altitude logistics last-mile collaborative delivery includes S101-S105.

[0023] S101. Obtain the location information of multiple candidate stopping points, the historical road segment speed sequence and surrounding obstacle distribution data of each candidate stopping point, the real-time operating parameters of the UAV and the real-time operating parameters of the robot.

[0024] The historical road segment speed sequence includes the historical average travel speed of multiple road segments on the path between the candidate stop and the robot.

[0025] The embodiment of this application describes a method for obtaining the location information of multiple candidate stops as follows: obtaining a high-precision vector map of the city, parsing the topology of the road network from the map, filtering out nodes that meet the physical stopping conditions as candidate stops, and assigning a unique location identifier to each candidate stop, which is the location information of the candidate stop.

[0026] Among them, nodes that meet the physical parking conditions include, but are not limited to, building entrances and exits, public parking lots, dedicated loading and unloading areas, and open spaces near non-motorized vehicle lane intersections.

[0027] The method for obtaining the historical road segment speed sequence for each candidate stop is as follows (taking the first candidate stop as an example): Taking the first candidate stop as the endpoint and the robot's current position as the starting point, a path for the robot to reach the first candidate stop is generated based on a preset path planning algorithm (such as Dijkstra's algorithm). Then, the path is parsed and decomposed into multiple consecutive road segments according to the road network topology. For each road segment, historical records with the same date type and time period as the current time are queried in the historical traffic database. The same date type includes weekdays or holidays, and the same time period includes morning peak hours or evening peak hours, etc. The historical average traffic speed of each road segment is extracted from the historical records. Finally, the obtained historical average traffic speeds are arranged in order from the starting road segment to the ending road segment to obtain the historical road segment speed sequence.

[0028] Optionally, if the data for this road segment is missing in the historical traffic database, the nominal speed limit value of this road segment in the high-precision vector map attributes is read as the historical average traffic speed.

[0029] It should be noted that the first candidate stop can be any one of multiple candidate stops.

[0030] The method for obtaining the surrounding obstacle distribution data for each candidate parking point is as follows (taking the first candidate parking point as an example): A circular scanning area with a preset radius is defined, centered on the location of the first candidate parking point. The height information and outline coordinates of all buildings and structures within this circular scanning area are extracted from a high-precision vector map. Then, the extracted information is mapped into a two-dimensional raster matrix centered on the first candidate parking point. This two-dimensional raster matrix serves as the surrounding obstacle distribution data for the first candidate parking point. In this two-dimensional raster matrix, grids occupied by obstacles are marked with a first value (e.g., 1), and grids without obstacles are marked with a second value (e.g., 0).

[0031] For example, the size of a single grid cell in this two-dimensional grid matrix can be 0.5 meters.

[0032] It should be understood that the surrounding obstacle distribution data records the space occupancy within the area centered on the candidate parking point, which is used to calculate the degree of enclosure of the environment.

[0033] Optionally, the value of the preset extension distance is determined based on the effective sensing range of the UAV positioning sensor and the distance at which the rotor flow field is susceptible to interference. In this embodiment, an empirical value of 50 meters is used.

[0034] It should be understood that the robot in the embodiments of this application refers to a ground mobile robot.

[0035] Optionally, the robot's real-time operating parameters include the robot's current position and average speed.

[0036] Optionally, the real-time operating parameters of the drone include standard cruise speed, standard cruise power, nominal hovering power, current location, and current remaining battery power.

[0037] It should be understood that standard cruise speed is the preset speed value when the drone is flying under normal conditions, standard cruise power is the average power of the drone when flying at standard cruise speed, and nominal hovering power is the average power of the drone when it is hovering in the air.

[0038] It should be noted that the locations obtained in the embodiments of this application are all latitude and longitude coordinates.

[0039] S102. Determine the travel time redundancy coefficient of each candidate stop based on the historical road segment speed sequence of each candidate stop.

[0040] Among them, the travel time redundancy coefficient is used to characterize the degree of delay risk of the robot arriving at the candidate docking point.

[0041] It should be understood that a path that frequently alternates between congested nodes (very low speed) and unobstructed nodes (high speed) has a highly uneven speed distribution. This unevenness statistically corresponds to a very high risk of long-tail delays, meaning that the robot's actual arrival time often far exceeds the predicted value based on the average speed.

[0042] In one alternative implementation, frequency analysis can be performed on the historical road segment speed sequence of the first candidate stop to obtain the spatial frequency domain energy distribution of the first candidate stop; based on the energy proportion of high-frequency components in the spatial frequency domain energy distribution, the traffic flow speed fluctuation proportion of the first candidate stop can be determined; based on the preset time adjustment weight and the traffic flow speed fluctuation proportion, the travel time redundancy coefficient of the first candidate stop can be determined.

[0043] The spatial frequency domain energy distribution includes multiple frequency components and the energy value of each frequency component. The traffic flow speed fluctuation ratio is used to characterize the degree of traffic speed fluctuation on multiple road segments along the path between the candidate stop point and the robot.

[0044] In this embodiment of the application, the historical road segment speed sequence is regarded as a discrete signal distributed along the spatial dimension, and frequency analysis is performed on the discrete signal.

[0045] Optionally, a fast Fourier transform algorithm can be used to perform frequency domain conversion on the historical road segment speed sequence, mapping the spatial speed sequence to an energy distribution in the frequency domain, thereby obtaining the spatial frequency domain energy distribution of the first candidate stop.

[0046] It should be understood that in this spatial frequency domain energy, low-frequency components represent the gradual trend of speed changes along the path (such as the entire road being congested or unobstructed), while high-frequency components represent drastic changes in speed (such as frequent start-stop switching). The energy value of a frequency component is the square of the amplitude obtained after the fast Fourier transform of that frequency component, which represents the energy contribution of that frequency component in the original road segment speed sequence.

[0047] It should be noted that, due to the symmetry of the Fast Fourier Transform, only the first half of the frequency components can be effectively analyzed. The sum of the energies of the high-frequency components and the sum of the total energies are both determined based on this first half of the frequency components.

[0048] Optionally, multiple frequency components can be sorted from low to high frequency starting from index 0 to obtain the index value of each frequency component. An index threshold is set, and frequency components with index values ​​greater than the index threshold are identified as high-frequency components.

[0049] Optionally, the index threshold can be set to 1 / 4 of the total number of frequency components. For example, assuming there are 10 frequency components and the index range to be analyzed is 0-5, the index threshold can be set to 2.5, that is, the high-frequency components are 3, 4, and 5.

[0050] It is understandable that the energy proportion of high-frequency components is the ratio of the sum of the energy values ​​of high-frequency components to the sum of the energy of all components (the first half of the frequency components).

[0051] In this embodiment of the application, the energy proportion of high-frequency components is determined as the traffic flow speed fluctuation proportion of the first candidate stop.

[0052] Optionally, the percentage of traffic flow speed fluctuations satisfies the following formula:

[0053] in, Indicates candidate stops The percentage of traffic flow speed fluctuations Indicates the index threshold. Indicates the robot and candidate docking points The total number of road segments on the path between them also represents the total number of frequency components. Represents frequency components amplitude, Represents frequency components The energy represents the numerical stability constant, taking a very small positive number (e.g., 10). -6 (), used to prevent the denominator from being zero and to ensure calculation stability.

[0054] In this formula, This represents the sum of the energies of the high-frequency components. It represents the sum of total energy.

[0055] It should be noted that in this formula, The value can be an integer, when When it is an odd number, Round down to the nearest integer.

[0056] It should be understood that the greater the proportion of traffic flow speed fluctuation, the more fragmented and disordered the speed distribution on the path corresponding to the candidate stop is. That is, the path contains a large number of traffic nodes that start and stop frequently, the higher the degree of road disorder, and the greater the probability that the robot will encounter unexpected long-tail delays.

[0057] In one alternative implementation, a time adjustment weight can be set to map the dimensionless traffic flow speed fluctuation ratio to a specific time increment ratio.

[0058] Optionally, the value of the time adjustment weight can be obtained through offline data calibration. The calibration method is as follows: collect historical operating data of the system in the target area, extract the deviation value between the robot's estimated navigation time and actual arrival time from a large number of samples; construct a loss function with the goal of minimizing the root mean square error of this deviation value, and use the gradient descent method to train and optimize the time adjustment weight, thereby obtaining the weight value that best matches the local traffic characteristics. For example, it can be set to 0.5.

[0059] Alternatively, the following formula must be satisfied:

[0060] in, To indicate candidate stops The redundancy coefficient of travel time, Indicates time-adjusted weights. Indicates candidate stops The percentage of traffic flow speed fluctuations.

[0061] In this formula, because and All are non-negative numbers, therefore The value is always greater than or equal to 1. The greater the proportion of traffic flow speed fluctuation, the greater the redundancy coefficient of travel time.

[0062] The aforementioned method for determining the travel time redundancy coefficient obtains the spatial frequency domain energy distribution by performing frequency analysis on historical road segment speed sequences, and determines the proportion of traffic flow speed fluctuations based on the energy proportion of high-frequency components. Then, it generates the travel time redundancy coefficient by combining preset time adjustment weights. This method can effectively identify road condition characteristics with similar average speeds but huge differences in stability, and quantify road segments with long-tail delay risks as the degree of delay risk, providing an accurate data foundation for subsequent robot prediction of arrival time.

[0063] S103. Based on the surrounding obstacle distribution data of each candidate docking point, determine the hovering power consumption increase coefficient of each candidate docking point.

[0064] Among them, the hovering power consumption increase coefficient is used to characterize the risk of increased power consumption when the drone hovers at the candidate docking point.

[0065] It should be understood that building obstruction can cause multipath effects on Global Navigation Satellite System (GNSS) signals, forcing UAVs to consume additional power for high-frequency attitude correction to maintain positioning. Near-wall effects and complex airflow in narrow spaces require UAVs to maintain higher power redundancy. Therefore, the more obstacles around a candidate docking point, the higher the power consumed by the UAV when hovering there.

[0066] In one implementation of this application, taking the first candidate docking point as an example, the obstacle enclosure degree of the first candidate docking point can be determined based on the surrounding obstacle distribution data of the first candidate docking point, and then the hovering power consumption increase coefficient of the first candidate docking point can be determined based on the obstacle enclosure degree and the preset power adjustment weight.

[0067] The obstacle enclosure degree is used to characterize the degree of enclosure of the surrounding environment of the candidate stopping point. The higher the obstacle enclosure degree, the more severe the enclosure.

[0068] In one optional implementation, the method for determining the obstacle enclosure degree of the first candidate docking point is as follows: taking the first candidate docking point as the center, multiple evenly distributed virtual detection rays are emitted in all directions; based on the obstacle distribution data around the first candidate docking point, the obstacle distance at which each virtual detection ray first encounters an obstacle within a preset extension distance is determined; based on the obstacle distances of the multiple virtual detection rays, the number of the multiple evenly distributed virtual detection rays, and the preset extension distance, the obstacle enclosure degree of the first candidate docking point is determined.

[0069] Optionally, a virtual detection ray can be emitted outward from the first candidate docking point as the origin, at 10-degree intervals. The virtual detection ray is used to simulate the detection path extending outward from the first candidate docking point and to determine the distance at which each virtual detection ray first encounters an obstacle within a preset extension distance.

[0070] Optionally, the average obstacle distance can be determined based on the sum of the obstacle distances of all virtual probe rays and the number of the multiple uniformly distributed virtual probe rays; the obstacle enclosure degree of the first candidate docking point can be determined based on the ratio between the average obstacle distance and the preset extension distance.

[0071] Optionally, the ratio between the sum of obstacle distances of all virtual probe rays and the number of uniformly distributed virtual probe rays can be determined as the average obstacle distance, the ratio between the average obstacle distance and the preset extension distance can be determined as the free space ratio, and the difference between 1 and the free space ratio can be determined as the obstacle enclosure degree.

[0072] It should be understood that the larger the ratio between the average obstacle distance and the preset extension distance, the more open the space around the first candidate stopping point and the smaller the obstacle enclosure.

[0073] In one alternative implementation, after determining the obstacle enclosure degree, a power adjustment weight can be set to convert the obstacle enclosure degree into a specific power increment ratio.

[0074] It should be understood that the power increment ratio is the ratio between the hovering increment power and the nominal hovering power, and the hovering increment power is the difference between the actual hovering power and the nominal hovering power.

[0075] Optionally, the power adjustment weight can be obtained through engineering experiments. The calibration process is as follows: under typical test scenarios with different obstacle enclosure degrees, collect the actual average power data of the UAV hovering for a long time; construct a scatter plot with obstacle enclosure degree as the abscissa and power increment ratio as the ordinate; use a quadratic curve fitting algorithm to fit the data points in the scatter plot, solve for the parameter with the smallest fitting error, and determine it as the optimal power adjustment weight so that it can accurately reflect the nonlinear influence of the environmental enclosure degree on the nominal hovering power.

[0076] As an example, the power regulation weight can be set to 0.8.

[0077] Optionally, the hovering power consumption increase factor satisfies the following formula:

[0078] in, Indicates candidate stops The hovering power consumption increase coefficient, Indicates power adjustment weight. Indicates candidate stops Obstacle enclosure degree.

[0079] In this formula, because and All are non-negative numbers, therefore Always greater than or equal to 1, the higher the obstacle enclosure, the greater the increase in hovering power consumption. Squaring the obstacle enclosure reveals the engineering characteristic that hovering power consumption increases non-linearly with environmental complexity: as the environment gradually becomes more enclosed, the rate of increase in additional energy consumption gradually accelerates, rather than a simple linear relationship. Find the square such that when When the increment is small, close to 0, the increment is small; when When the value approaches 1, the increment increases dramatically, which is consistent with the nonlinear deterioration characteristics of near-wall effect and multipath effect in an extremely closed environment.

[0080] The method described above for determining the hovering power consumption increase coefficient uses virtual probe rays to simulate the claustrophobia of the space in which the UAV is located. Based on the obstacle distances, numbers, and preset extension distances of all virtual probe rays, the obstacle enclosure degree is determined, achieving precise quantification of the microenvironment enclosure degree. Then, based on this obstacle enclosure degree and power adjustment weights, a hovering power consumption increase coefficient is generated, establishing a correlation between hovering power and the geometric enclosure characteristics of the environment. This quantifies the additional energy consumption caused by dense building environments for UAV positioning and control, providing a data foundation for subsequent determination of hovering energy consumption.

[0081] S104. Based on the travel time redundancy coefficient of each candidate docking point, the hovering power consumption increase coefficient of each candidate docking point, the real-time operating parameters of the UAV, the real-time operating parameters of the robot, and the position of each candidate docking point, determine the estimated total energy consumption of the UAV at each candidate docking point.

[0082] It should be understood that the estimated total energy consumption of a drone at a candidate docking point includes the energy consumption of the drone on its way to the candidate docking point and the energy consumption during the waiting process at the candidate docking point.

[0083] Understandably, based on the redundancy coefficient of the travel time at each candidate docking point, the time required for the robot to reach each candidate docking point in the event of a delay can be determined. Based on the hovering power consumption increase coefficient at each candidate docking point, the energy required for the drone to consume when hovering power consumption increases can be determined.

[0084] In one implementation of this application, taking the first candidate docking point as an example, the safe navigation time can be determined based on the robot's current position and average speed, the position of the first candidate docking point, and the travel time redundancy coefficient; the flight time of the drone to reach the first candidate docking point can be determined based on the standard cruise speed and current position of the drone, and the position of the first candidate docking point; and the total energy consumption estimate of the first candidate docking point can be determined based on the safe navigation time, the flight time, the standard cruise power, the nominal hovering power, and the hovering power consumption increase coefficient.

[0085] The safe navigation time is the maximum time for the robot to reach the candidate docking point, that is, the maximum time required for the robot to reach the first candidate docking point in the worst case, taking into account the delays that may be caused by disorderly fluctuations in road conditions.

[0086] Specifically, based on the robot's current position and the position of the first candidate docking point, the path length between the robot and the first candidate docking point can be determined. Dividing the path length by the robot's average speed gives the normal travel time for the robot to reach the first candidate docking point. Multiplying the normal travel time by the travel time redundancy coefficient gives the safe navigation time.

[0087] Then, based on the current position of the drone and the position of the first candidate docking point, the path length between the drone and the first candidate docking point is determined. The path length is then divided by the standard cruising speed of the drone to obtain the flight time.

[0088] In one alternative implementation, the potential hovering waiting time of the UAV at the first candidate docking point can be determined based on the safe navigation time and the flight time; the flight energy consumption can be determined based on the flight time and the standard cruise power; the hovering energy consumption can be determined based on the potential hovering waiting time, the nominal hovering power, and the hovering power consumption increase factor; and the sum of the flight energy consumption and the hovering energy consumption can be determined as the estimated total energy consumption at the first candidate docking point.

[0089] The potential hovering waiting time is used to characterize the duration for which the drone hovers at the first candidate docking point while waiting for the robot to arrive.

[0090] Alternatively, the potential hovering wait time can be obtained by subtracting the flight time from the safe navigation time.

[0091] It should be noted that the safe navigation time minus the flight time may be negative. In this case, the potential hovering wait time can be set to 0.

[0092] Optionally, the product of non-flight time and standard cruise power is determined as flight energy consumption; the product of potential hovering waiting time, nominal hovering power, and hovering power consumption increase factor is determined as hovering energy consumption.

[0093] It should be understood that the hovering energy consumption is the estimated hovering energy consumption of the drone under the condition that the robot arrives late and the nominal hovering power is increased due to the surrounding environment of the first candidate docking point.

[0094] The method described above for determining the total energy consumption estimate converts the delay risk in the time dimension (reflected by the potential hovering waiting time) and the environmental risk in the power dimension (reflected by the hovering power consumption increase coefficient) into energy units in a unified manner, realizing the comparability of heterogeneous risks under the same metric scale, and laying the computational foundation for global optimal decision-making.

[0095] S105. Based on the estimated total energy consumption of each candidate docking point, determine the target docking point from multiple candidate docking points.

[0096] In one alternative implementation, the candidate docking point with the lowest estimated total energy consumption can be selected as the target docking point.

[0097] In another alternative implementation, the drone's current remaining battery power may be insufficient to support its arrival at the target docking point or to support its return. In this case, the product of the current remaining battery power and a preset safety factor can be determined as the battery safety threshold. If the estimated total energy consumption at the target docking point is less than or equal to the battery safety threshold, a handover command is sent to both the drone and the robot. If the estimated total energy consumption at the target docking point is greater than the battery safety threshold, a handover stop command is sent to both the drone and the robot.

[0098] The handover instruction is used to indicate that the handover should take place at the target docking point, and the handover stop instruction is used to indicate that the handover should be stopped.

[0099] It should be understood that the preset safety factor is between 0 and 1, used to reserve a certain amount of power margin to cope with extreme situations or model errors. For example, an empirical value of 0.8 can be taken, that is, reserving 20% ​​of the remaining power as a safety buffer to ensure that the drone still has enough power to perform return or emergency operations even in the worst case.

[0100] Understandably, if the estimated total energy consumption at the target docking point is less than or equal to the safe energy threshold, it means that even in the worst-case scenario considering all potential risks, the drone still has sufficient power to complete the flight mission and reserves a safety margin. In this case, handover instructions can be sent to both the drone and the robot, including the location of the target docking point, so that the drone and robot can proceed to that point. If the estimated total energy consumption at the target docking point exceeds the safe energy threshold, it indicates that the current overall environmental risk is too high. Even if the target docking point with the lowest energy consumption is selected, the collaborative task still faces a significant risk of running out of power. In this case, a handover stop command can be sent to both the drone and the robot to cancel the handover task and stop the handover process.

[0101] Optionally, if the estimated total energy consumption at the target docking point exceeds the safe energy threshold, a return-to-home command or a nearby landing command can be issued to the drone, and a command can be issued to the robot to independently complete the remaining delivery tasks or stand by in place, in order to ensure equipment safety.

[0102] In one alternative implementation, the methods provided in S101-S105 above can be executed periodically, for example, once every 60 seconds, to re-collect data and evaluate the target stop point, so as to adjust the optimal stop point in a timely manner based on real-time traffic conditions.

[0103] The methods provided in S101-S105 above quantify the delay risk of the robot on the ground and the hovering energy consumption risk of the drone by acquiring historical road segment speed sequences and surrounding obstacle distribution data of candidate docking points. Based on the real-time operating parameters of the drone and the robot, the two risks are mapped to a unified total energy consumption estimate, and finally the target docking point is determined from multiple candidate docking points. This more accurate prediction of drone energy consumption improves the time reliability and energy efficiency of low-altitude logistics delivery.

[0104] like Figure 2 As shown in the figure, this application embodiment also provides a route optimization system for low-altitude logistics last-mile collaborative delivery. The route optimization system 20 for low-altitude logistics last-mile collaborative delivery includes a data acquisition module 201, a risk prediction module 202, and a stop point determination module 203.

[0105] The data acquisition module 201 is used to acquire the location information of multiple candidate stopping points, the historical road segment speed sequence and surrounding obstacle distribution data of each candidate stopping point, the real-time operating parameters of the UAV and the real-time operating parameters of the robot. The historical road segment speed sequence includes the historical average travel speed of multiple road segments on the path between the candidate stop and the robot.

[0106] The risk prediction module 202 is used to determine the travel time redundancy coefficient of each candidate stop based on the historical road segment speed sequence of each candidate stop.

[0107] Among them, the travel time redundancy coefficient is used to characterize the degree of delay risk of the robot arriving at the candidate docking point.

[0108] The risk prediction module 202 is also used to determine the hovering power consumption increase coefficient of each candidate docking point based on the surrounding obstacle distribution data of each candidate docking point.

[0109] Among them, the hovering power consumption increase coefficient is used to characterize the risk of increased power consumption when the drone hovers at the candidate docking point.

[0110] The risk prediction module 202 is also used to determine the total energy consumption prediction of the UAV at each candidate docking point based on the passage time redundancy coefficient of each candidate docking point, the hovering power consumption increase coefficient of each candidate docking point, the real-time operating parameters of the UAV, the real-time operating parameters of the robot, and the location of each candidate docking point.

[0111] The docking point determination module 203 is used to determine the target docking point from multiple candidate docking points based on the estimated total energy consumption of each candidate docking point.

[0112] It should be noted that the low-altitude logistics last-mile collaborative delivery route optimization system 20 can also execute any of the above-mentioned low-altitude logistics last-mile collaborative delivery route optimization methods.

[0113] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0114] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A route optimization method for low-altitude logistics last-mile collaborative delivery, characterized in that, include: The system acquires the location information of multiple candidate stopping points, the historical road segment speed sequence and surrounding obstacle distribution data of each candidate stopping point, the real-time operating parameters of the drone and the robot. The historical road segment speed sequence includes the historical average travel speed of multiple road segments on the path between the candidate stopping point and the robot. The passage time redundancy coefficient of each candidate stop is determined based on the historical road segment speed sequence of each candidate stop, including: performing frequency analysis on the historical road segment speed sequence of the first candidate stop to obtain the spatial frequency domain energy distribution of the first candidate stop. The spatial frequency domain energy distribution includes multiple frequency components and the energy value of each frequency component. The first candidate stop is any one of the multiple candidate stops. Based on the energy proportion of high-frequency components in the spatial frequency domain energy distribution, the traffic flow speed fluctuation proportion of the first candidate stop is determined. The traffic flow speed fluctuation proportion is used to characterize the degree of traffic speed fluctuation of multiple road segments on the path between the candidate stop and the robot. Based on preset time adjustment weights and the proportion of traffic flow speed fluctuations, the passage time redundancy coefficient of the first candidate stopping point is determined; the passage time redundancy coefficient is used to characterize the degree of delay risk of the robot arriving at the candidate stopping point; Based on the surrounding obstacle distribution data for each candidate docking point, the hovering power consumption increase coefficient for each candidate docking point is determined, including: Based on the obstacle distribution data around the first candidate parking point, the obstacle enclosure degree of the first candidate parking point is determined, including: emitting multiple evenly distributed virtual detection rays centered on the first candidate parking point; determining the obstacle distance at which each virtual detection ray first encounters an obstacle within a preset extension distance, based on the obstacle distribution data around the first candidate parking point; determining the obstacle enclosure degree of the first candidate parking point based on the obstacle distances of the multiple virtual detection rays, the number of evenly distributed virtual detection rays, and the preset extension distance; the obstacle enclosure degree is used to characterize the degree of enclosure of the surrounding environment of the candidate parking point; Based on the obstacle enclosure degree and the preset power adjustment weight, the hovering power consumption increase coefficient of the first candidate docking point is determined; the hovering power consumption increase coefficient is used to characterize the risk of increased power consumption when the UAV hovers at the candidate docking point. Based on the travel time redundancy coefficient of each candidate docking point, the hovering power consumption increase coefficient of each candidate docking point, the real-time operating parameters of the UAV, the real-time operating parameters of the robot, and the location of each candidate docking point, the estimated total energy consumption of the UAV at each candidate docking point is determined. Based on the estimated total energy consumption of each candidate stop, the target stop is determined from multiple candidate stops.

2. The route optimization method for low-altitude logistics last-mile collaborative delivery according to claim 1, characterized in that, The method of determining the obstacle enclosure degree of the first candidate docking point based on the obstacle distance of multiple virtual detection rays, the number of the multiple uniformly distributed virtual detection rays, and the preset extension distance includes: The average obstacle distance is determined based on the sum of the obstacle distances of all virtual probe rays and the number of the multiple uniformly distributed virtual probe rays; The obstacle enclosure degree of the first candidate stopping point is determined based on the ratio between the average obstacle distance and the preset extension distance.

3. The route optimization method for low-altitude logistics last-mile collaborative delivery according to claim 1, characterized in that, The robot's real-time operating parameters include its current position and average speed. The drone's real-time operating parameters include standard cruise speed, standard cruise power, nominal hovering power, and current position. The method for determining the estimated total energy consumption of the drone at each candidate docking point, based on the travel time redundancy coefficient, the hovering power consumption increase coefficient, the drone's real-time operating parameters, the robot's real-time operating parameters, and the position of each candidate docking point, includes: Based on the robot's current position and average speed, the position of the first candidate docking point, and the travel time redundancy coefficient, a safe navigation time is determined, which is the maximum time for the robot to reach the candidate docking point. Based on the drone's standard cruising speed and current location, as well as the location of the first candidate docking point, determine the flight time for the drone to reach the first candidate docking point; Based on the safe navigation time, flight time, standard cruise power, nominal hovering power, and hovering power consumption increase factor, the total energy consumption estimate of the first candidate docking point is determined.

4. The route optimization method for low-altitude logistics last-mile collaborative delivery according to claim 3, characterized in that, The determination of the total energy consumption estimate for the first candidate docking point based on the safe navigation time, flight time, standard cruise power, nominal hovering power, and hovering power consumption increase factor includes: Based on the safe navigation time and the flight time, the potential hovering waiting time of the UAV at the first candidate docking point is determined. The potential hovering waiting time is used to characterize the duration for which the UAV hovers at the candidate docking point while waiting for the robot to arrive. The flight energy consumption is determined based on the flight time and standard cruise power. The hovering energy consumption is determined based on the potential hovering wait time, nominal hovering power, and hovering power consumption increase factor. The sum of flight energy consumption and hovering energy consumption is used as the estimated total energy consumption for the first candidate docking point.

5. The route optimization method for low-altitude logistics last-mile collaborative delivery according to claim 1, characterized in that, The process of determining the target stop from the plurality of candidate stops based on the estimated total energy consumption of each candidate stop includes: The candidate stop with the lowest estimated total energy consumption is selected as the target stop.

6. The route optimization method for low-altitude logistics last-mile collaborative delivery according to claim 1, characterized in that, The drone's real-time operating parameters also include the current remaining battery power, and the method further includes: The product of the current remaining power and the preset safety factor is determined as the power safety threshold. If the estimated total energy consumption at the target docking point is less than or equal to the power safety threshold, a handover instruction is sent to the drone and the robot respectively, the handover instruction being used to instruct the handover to take place at the target docking point; If the estimated total energy consumption at the target docking point exceeds the power safety threshold, a handover stop command is sent to both the drone and the robot, indicating that the handover should be stopped.

7. A route optimization system for low-altitude logistics last-mile collaborative delivery, characterized in that, include: Data acquisition module, risk assessment module, and docking point determination module; The data acquisition module is used to acquire the location information of multiple candidate stopping points, the historical road segment speed sequence and surrounding obstacle distribution data of each candidate stopping point, the real-time operating parameters of the UAV and the real-time operating parameters of the robot. The historical road segment speed sequence includes the historical average passage speed of multiple road segments on the path between the candidate stopping point and the robot. The risk prediction module is used to determine the travel time redundancy coefficient of each candidate stop point based on the historical road segment speed sequence of each candidate stop point. The travel time redundancy coefficient is used to characterize the degree of delay risk of the robot arriving at the candidate stop point. The risk prediction module is also used to perform frequency analysis on the historical road segment speed sequence of the first candidate stop point to obtain the spatial frequency domain energy distribution of the first candidate stop point. The spatial frequency domain energy distribution includes multiple frequency components and the energy value of each frequency component. The first candidate stop point is any one of the multiple candidate stop points. Based on the energy proportion of high-frequency components in the spatial frequency domain energy distribution, the traffic flow speed fluctuation proportion of the first candidate stop is determined. The traffic flow speed fluctuation proportion is used to characterize the degree of traffic speed fluctuation of multiple road segments on the path between the candidate stop and the robot. Based on the preset time adjustment weight and the proportion of traffic flow speed fluctuation, the passage time redundancy coefficient of the first candidate stop is determined; The risk prediction module is also used to determine the hovering power consumption increase coefficient of each candidate docking point based on the surrounding obstacle distribution data of each candidate docking point. The hovering power consumption increase coefficient is used to characterize the risk of increased power consumption when the UAV hovers at the candidate docking point. The risk prediction module is further configured to determine the obstacle enclosure degree of the first candidate parking point based on the obstacle distribution data around the first candidate parking point, including: emitting multiple evenly distributed virtual detection rays centered on the first candidate parking point; determining the obstacle distance at which each virtual detection ray first encounters an obstacle within a preset extension distance based on the obstacle distribution data around the first candidate parking point; and determining the obstacle enclosure degree of the first candidate parking point based on the obstacle distances of the multiple virtual detection rays, the number of the multiple evenly distributed virtual detection rays, and the preset extension distance; the obstacle enclosure degree is used to characterize the degree of enclosure of the surrounding environment of the candidate parking point. Based on the obstacle enclosure degree and the preset power adjustment weight, the hovering power consumption increase coefficient of the first candidate docking point is determined; The risk prediction module is also used to determine the total energy consumption prediction of the UAV at each candidate docking point based on the passage time redundancy coefficient of each candidate docking point, the hovering power consumption increase coefficient of each candidate docking point, the real-time operating parameters of the UAV, the real-time operating parameters of the robot, and the position of each candidate docking point. The docking point determination module is used to determine the target docking point from multiple candidate docking points based on the estimated total energy consumption of each candidate docking point.

Images