WO2019119926A1 - 货物保护方法、装置、系统和计算机可读存储介质 - Google Patents

货物保护方法、装置、系统和计算机可读存储介质 Download PDF

Info

Publication number
WO2019119926A1
WO2019119926A1 PCT/CN2018/108955 CN2018108955W WO2019119926A1 WO 2019119926 A1 WO2019119926 A1 WO 2019119926A1 CN 2018108955 W CN2018108955 W CN 2018108955W WO 2019119926 A1 WO2019119926 A1 WO 2019119926A1
Authority
WO
WIPO (PCT)
Prior art keywords
drone
cargo
obstacle
target point
airbag
Prior art date
Application number
PCT/CN2018/108955
Other languages
English (en)
French (fr)
Inventor
谷飞
Original Assignee
北京京东尚科信息技术有限公司
北京京东世纪贸易有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京京东尚科信息技术有限公司, 北京京东世纪贸易有限公司 filed Critical 北京京东尚科信息技术有限公司
Priority to US16/954,650 priority Critical patent/US11694294B2/en
Priority to JP2020530580A priority patent/JP7179065B2/ja
Publication of WO2019119926A1 publication Critical patent/WO2019119926A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B64—AIRCRAFT; AVIATION; COSMONAUTICS
    • B64D—EQUIPMENT FOR FITTING IN OR TO AIRCRAFT; FLIGHT SUITS; PARACHUTES; ARRANGEMENT OR MOUNTING OF POWER PLANTS OR PROPULSION TRANSMISSIONS IN AIRCRAFT
    • B64D45/00—Aircraft indicators or protectors not otherwise provided for
    • B64D45/04—Landing aids; Safety measures to prevent collision with earth's surface
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/881—Radar or analogous systems specially adapted for specific applications for robotics
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/93—Radar or analogous systems specially adapted for specific applications for anti-collision purposes
    • G01S13/933—Radar or analogous systems specially adapted for specific applications for anti-collision purposes of aircraft or spacecraft
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/08—Control of attitude, i.e. control of roll, pitch, or yaw
    • G05D1/0808—Control of attitude, i.e. control of roll, pitch, or yaw specially adapted for aircraft
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/24—Aligning, centring, orientation detection or correction of the image
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B64—AIRCRAFT; AVIATION; COSMONAUTICS
    • B64D—EQUIPMENT FOR FITTING IN OR TO AIRCRAFT; FLIGHT SUITS; PARACHUTES; ARRANGEMENT OR MOUNTING OF POWER PLANTS OR PROPULSION TRANSMISSIONS IN AIRCRAFT
    • B64D2201/00—Airbags mounted in aircraft for any use
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B64—AIRCRAFT; AVIATION; COSMONAUTICS
    • B64U—UNMANNED AERIAL VEHICLES [UAV]; EQUIPMENT THEREFOR
    • B64U2101/00—UAVs specially adapted for particular uses or applications
    • B64U2101/60—UAVs specially adapted for particular uses or applications for transporting passengers; for transporting goods other than weapons
    • B64U2101/64—UAVs specially adapted for particular uses or applications for transporting passengers; for transporting goods other than weapons for parcel delivery or retrieval
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/93—Radar or analogous systems specially adapted for specific applications for anti-collision purposes
    • G01S13/933—Radar or analogous systems specially adapted for specific applications for anti-collision purposes of aircraft or spacecraft
    • G01S13/935—Radar or analogous systems specially adapted for specific applications for anti-collision purposes of aircraft or spacecraft for terrain-avoidance
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/418—Theoretical aspects

Definitions

  • the present disclosure relates to the field of drone technology, and in particular, to a cargo protection method, apparatus, system, and computer readable storage medium.
  • One technical problem to be solved by the present disclosure is how to improve the safety of the drone cargo and reduce the probability of cargo damage.
  • a cargo protection method includes: determining whether a drone is in a falling state according to an acceleration of a drone currently in a vertical direction and a vertical distance of the drone to the ground, When the drone is in a falling state, at least one air bag in the cargo compartment of the drone is opened to protect the cargo.
  • determining whether the drone is in a falling state according to the current acceleration of the drone in the vertical direction and the vertical distance of the drone from the ground includes: detecting the current acceleration of the drone in the vertical direction. In the case where the acceleration reaches the acceleration threshold, the vertical distance of the drone to the ground is detected; when the vertical distance is greater than the distance threshold, the drone is determined to be in a falling state.
  • opening at least one airbag in the cargo compartment of the drone includes: information obtained by scanning the ground according to the radar on the drone when the drone is in a falling state Determining the type of obstacle under the drone and the orientation of the obstacle relative to the drone; determining the timing of opening at least one airbag in the corresponding orientation according to the type of the obstacle and the orientation of the obstacle relative to the drone; At least one airbag in the corresponding orientation is opened when the timing is satisfied.
  • the method further includes: extending the predetermined time interval; opening the air bag currently located under the cargo.
  • the method further includes: adjusting the posture of the drone according to the orientation of the obstacle relative to the drone, In order to avoid obstacles; open the airbag currently located under the cargo.
  • determining the type of obstacle that the drone falls to the ground based on the information obtained by the radar on the drone scanning the ground includes: the target point in the image of the region of interest obtained by scanning the ground on the radar on the drone Constructing echoes; time-domain features and frequency-domain features of echoes extracting target points based on different azimuth angles as time-domain feature sequences and frequency-domain feature sequences of target points; time-domain feature sequences and frequency-domain feature sequences of target points
  • the hidden Markov model is input separately, and the time domain feature sequence of the output target point and the frequency domain feature sequence are obtained under the hidden Markov model respectively.
  • the time domain feature sequence and the frequency domain feature sequence according to the target point are hidden in the hidden Markov.
  • the probability under the Kraft model determines the type of obstacle corresponding to the target point.
  • reconstructing the echo of the target point in the image of the region of interest comprises: performing a two-dimensional Fourier transform on the image of the region of interest to obtain a wavenumber domain image of the region of interest;
  • the azimuth relationship maps the wavenumber domain image to the frequency and azimuth fields to obtain the frequency azimuth domain image of the region of interest;
  • the inverse frequency Fourier transform of the frequency azimuth domain image is obtained in the azimuthal direction to obtain the reconstructed image.
  • the echo of the target point comprises: performing a two-dimensional Fourier transform on the image of the region of interest to obtain a wavenumber domain image of the region of interest;
  • the azimuth relationship maps the wavenumber domain image to the frequency and azimuth fields to obtain the frequency azimuth domain image of the region of interest;
  • the inverse frequency Fourier transform of the frequency azimuth domain image is obtained in the azimuthal direction to obtain the reconstructed image.
  • the echo of the target point comprises: performing a two-dimensional Fourier transform on the image
  • determining the type of the obstacle corresponding to the target point according to the probability of the time domain feature sequence of the target point and the frequency domain feature sequence in the hidden Markov model comprises: time domain feature sequence and frequency domain of the target point
  • the probability of the feature sequence under the hidden Markov model is taken as the probability feature of the target point;
  • the obstacle type of the target point is determined according to the distribution of the probability features of the target point and the distribution of the probability features of the training samples of the obstacles.
  • a cargo protection device comprising: a state determination module configured to determine no according to an acceleration of a drone currently in a vertical direction and a vertical distance of the drone from the ground to the ground Whether the man-machine is in a falling state; the protection measure triggering module is configured to open at least one air bag in the cargo compartment of the drone to protect the cargo when the drone is in a falling state.
  • the state determination module is configured to detect the current acceleration of the drone in the vertical direction, and detect the vertical distance of the drone to the ground when the acceleration reaches the acceleration threshold, where the vertical distance is greater than the distance In the case of a threshold, it is determined that the drone is in a falling state.
  • the protective measure triggering module is configured to determine the type of obstacle under the drone and the obstacle relative to the information obtained from the radar scanning ground on the drone if the drone is in a falling state.
  • the orientation of the drone determines the timing of opening at least one airbag in the corresponding orientation according to the type of the obstacle and the orientation of the obstacle relative to the drone, and opens at least one airbag in the corresponding orientation when the timing is satisfied.
  • the protective measure triggering module is further configured to extend the predetermined time interval after opening the at least one air bag to open the air bag currently located below the cargo.
  • the protection triggering module is further configured to determine the type of obstacle below the drone and the orientation of the obstacle relative to the drone, depending on the orientation of the obstacle relative to the drone The posture is adjusted to avoid obstacles and the airbag currently located below the cargo is opened.
  • the protection measure triggering module is configured to: reconstruct an echo of a target point in the image of the region of interest obtained from the radar scan ground on the drone; extract the time domain of the echo of the target point based on different azimuth angles
  • the feature and frequency domain features are respectively used as the time domain feature sequence and the frequency domain feature sequence of the target point; the time domain feature sequence and the frequency domain feature sequence of the target point are respectively input into the hidden Markov model to obtain the output target point time respectively.
  • the probability of the domain feature sequence and the frequency domain feature sequence in the hidden Markov model; the type of the obstacle corresponding to the target point is determined according to the probability of the time domain feature sequence of the target point and the frequency domain feature sequence under the hidden Markov model.
  • the protection measure triggering module is configured to perform a two-dimensional Fourier transform on the image of the region of interest to obtain a wavenumber domain image of the region of interest; and to determine the wave number according to the relationship between frequency and azimuth in the wavenumber domain
  • the domain image is mapped into the frequency and azimuth domains to obtain the frequency azimuth domain image of the region of interest; the frequency azimuth domain image is inverse Fourier transformed along the azimuth direction to obtain the echo of the reconstructed target point.
  • the protection measure triggering module is configured to use the probability of the time domain feature sequence of the target point and the frequency domain feature sequence under the hidden Markov model as the probability feature of the target point, according to the distribution of the probability features of the target point.
  • the distribution of probability features of the training samples with each obstacle determines the type of obstacle at the target point.
  • a cargo protection system comprising: the cargo protection device of any of the preceding embodiments; and a plurality of airbags disposed around the cargo in the cargo compartment of the drone.
  • the airbag is a blasting airbag that is divided into at least one airbag set according to the orientation, and the airbags in the airbag set are connected in parallel to the same ignition device.
  • a cargo protection device comprising: a memory; and a processor coupled to the memory, the processor being configured to perform any of the foregoing implementations based on instructions stored in the memory device Example of cargo protection methods.
  • a computer readable storage medium having stored thereon a computer program, wherein the program is executed by a processor to implement the cargo protection method of any of the foregoing embodiments.
  • the airbag when the drone falls, at least one airbag is opened, and the airbag is opened more timely than the parachute, which can directly avoid the impact of the cargo and reduce the probability of damage of the drone cargo.
  • FIG. 1 shows a flow diagram of a cargo protection method of some embodiments of the present disclosure.
  • Figure 2 shows a top view of the distribution of the cargo compartment airbag of the drone of some embodiments of the present disclosure.
  • FIG. 3 shows a schematic structural view of an air bag of some embodiments of the present disclosure.
  • FIG. 4A shows a schematic flow diagram of a cargo protection method of further embodiments of the present disclosure.
  • FIG. 4B shows a schematic flow diagram of a cargo protection method of still further embodiments of the present disclosure.
  • FIG. 5 shows a schematic structural view of a cargo protection device of some embodiments of the present disclosure.
  • FIG. 6 shows a schematic structural view of a cargo protection system of some embodiments of the present disclosure.
  • FIG. 7 shows a schematic structural view of a cargo protection device according to further embodiments of the present disclosure.
  • FIG. 8 is a block diagram showing the structure of a cargo protection device according to still another embodiment of the present disclosure.
  • FIG. 1 is a flow chart of some embodiments of a cargo protection method of the present disclosure. As shown in FIG. 1, the method of this embodiment includes: steps S102 to S104.
  • step S102 it is determined whether the drone is in a falling state based on the current acceleration of the drone in the vertical direction and the vertical distance of the drone from the ground.
  • the vertical direction acceleration of the drone is first detected, and when the acceleration in the vertical direction reaches the acceleration threshold, the vertical distance of the drone to the ground is detected, and when the vertical distance reaches the distance threshold, the determination is made.
  • the drone is in a falling state.
  • the acceleration of the drone in the vertical direction can be detected by an acceleration sensor.
  • the vertical distance of the drone to the ground can be detected by an ultrasonic ranging system, a ranging radar, and the like.
  • the vertical acceleration of the drone and the vertical distance to the ground can be periodically detected. Once the fall condition is found, the subsequent protection measures are immediately implemented.
  • step S104 if the drone is in a falling state, at least one airbag in the cargo compartment of the drone is opened to secure the cargo.
  • a plurality of air bags arranged around the cargo may be provided in the cargo compartment of the drone.
  • the size, number and position of the airbags can be adjusted according to the size of the actual cargo and the space of the cargo compartment.
  • multiple airbags can be placed around the outside of the cargo hold.
  • FIG 2 there is a top view of the distribution of the cargo compartment airbags of the drone in some embodiments.
  • a plurality of air bags are arranged around the cargo and around the cargo hold.
  • multiple airbags can be placed on the drone, distributed in various parts of the drone.
  • the airbag provided on the drone can be opened simultaneously with the airbag of the corresponding orientation in the cargo compartment, and protects the drone when the drone falls.
  • the air bag can be used with a parachute mounted on the drone.
  • the parachute can reduce the speed of the drone's falling, and the airbag in the cargo compartment, around the cargo compartment or on the drone can better protect the cargo or the drone.
  • the drone can detect the current flight attitude, for example, using a gyroscope to detect the tilt angle of the drone with respect to the horizontal direction, determining the location of the ground contact when falling, and opening the airbag of the corresponding portion according to the flight attitude.
  • the drone falls parallel to the ground
  • the bottom of the drone first touches the ground
  • the airbag at the bottom of the cargo can be mainly opened
  • the airbag at the bottom of the drone can be opened
  • the airbag at the upper part of the drone or the cargo compartment can be opened.
  • open with respect to the bottom airbag
  • the head of the drone falls down
  • the head of the drone and the airbag in the cargo tank of the corresponding orientation can be preferentially opened.
  • the airbag that is preferentially opened can protect the first time when the drone falls. Further, when the impact force of some parts at the time of falling reaches the threshold value, the airbag of the corresponding part can be triggered to be opened, and the protection is continued.
  • the flight attitude can be adjusted to select the location of the ground contact when the fall is made, and the airbag of the corresponding portion is opened.
  • FIG. 3 is a schematic structural view of a single airbag.
  • the airbag is, for example, a blasting airbag, and the airbag 300 is provided with a detonator 302, which is located in the gas generator 304.
  • the detonator can be ignited by the electronic ignition device 306 through the lead 308, and further the detonator ignites the surrounding reactants to generate gas to fill the balloon.
  • the reactants are, for example, sodium azide, which can be triggered by rapid or thermal decomposition to generate a gas.
  • they can be divided into at least one airbag group according to the orientation, and the airbags in the airbag group are connected in parallel to the same ignition device, and can be simultaneously turned on.
  • the combination of the blasting airbag and the parallel arrangement can further speed up the opening of the airbag and improve the safety of the cargo.
  • a plurality of airbags arranged around the cargo are arranged in the unmanned cargo warehouse.
  • the airbag is opened more timely than the parachute, and can directly Avoid the impact of the cargo, reducing the probability of damage to the drone cargo.
  • the drone can also automatically scan the ground, determine whether there is an obstacle on the ground according to the scanned image, and determine the airbag opening mode or obstacle avoidance according to the obstacle on the ground. Further embodiments of the cargo protection method of the present disclosure are described below in conjunction with Figures 4A and 4B.
  • FIG. 4A is a flow chart of still another embodiment of the cargo protection method of the present disclosure. As shown in FIG. 4A, the method of this embodiment includes steps S402 to S416.
  • step S402 it is determined whether the drone is in a falling state based on the current acceleration of the drone in the vertical direction and the vertical distance of the drone to the ground. If it is in the falling state, step S404 is performed; otherwise, step S402 is re-executed after the preset period is interrupted.
  • step S404 an image of the region of interest obtained by the radar scanning ground on the drone is acquired.
  • a radar detecting device can be set on the drone to detect the ground and obtain a radar detecting image. From the radar detection image, an ROI (Region of Interest) image can be extracted according to the local gray feature.
  • the ROI image contains at least one target point, and the ROI image is a slant range-azimuth image.
  • step S406 the type of the obstacle on which the drone falls to the ground and the orientation of the obstacle relative to the drone are determined based on the image of the region of interest.
  • the step S406 can include the following sub-steps S4061-S4064.
  • step S4061 an echo is reconstructed for the target point in the region of interest image.
  • the image of the region of interest is subjected to two-dimensional Fourier transform to obtain a wavenumber domain image of the region of interest; and the wavenumber domain image is mapped to frequency and azimuth according to the relationship between frequency and azimuth in the wavenumber domain.
  • the frequency azimuth domain image of the region of interest is obtained; the frequency azimuth domain image is inverse Fourier transformed along the azimuth direction to obtain the echo of the reconstructed target point.
  • the ROI image is represented as f(x, r)
  • it is subjected to two-dimensional Fourier transform to obtain a wavenumber domain image.
  • x and r represent the slant range and the azimuth position, respectively
  • k x and k r represent the slant range wave number and the azimuth wave number, respectively.
  • f represents the frequency
  • c represents the speed of light.
  • Map to The domain, the frequency azimuth domain forms the frequency response of the target point at each angle of incidence.
  • Azimuth By performing an inverse Fourier transform, an echo corresponding to the target point at the azimuth can be obtained.
  • step S4062 the time domain feature and the frequency domain feature of the echo of the target point are extracted based on different azimuth angles, respectively, as the time domain feature sequence and the frequency domain feature sequence of the target point.
  • Time-frequency atomic analysis can be used to extract time-domain features and frequency-domain features of echoes of target points at different azimuths, for example, extracting time-domain bimodal spacing and frequency-domain pits.
  • Time-frequency atomic analysis is a prior art and will not be described here.
  • the time domain feature is represented, for example, as t
  • the frequency domain feature is represented, for example, as f
  • f (f 1 , f 2 , ... f N ).
  • step S4063 the time domain feature sequence and the frequency domain feature sequence of the target point are respectively input into the hidden Markov model, and the time domain feature sequence and the frequency domain feature sequence of the output target point are respectively obtained under the hidden Markov model. Probability.
  • the time domain feature and the frequency domain feature based on the azimuth extraction are sequences that change with time.
  • the HMM Hidden Markov Model
  • the time domain feature and the frequency domain feature can be utilized to identify the category of the target point.
  • the parameters of the HMM mainly include an initial state distribution ⁇ , a state transition probability matrix B, and a probability matrix U that produces a visible state in an implicit state.
  • the state transition probability matrix B ⁇ b mn
  • m, n 1, 2, ..., N ⁇ , which conforms to the following formula:
  • the HMM training process is simplified to a probability matrix U that produces a visible state in the hidden state.
  • U can be determined according to the probability of occurrence of the observed value.
  • ⁇ ) can be obtained for the time domain feature sequence t of the target point, that is, under the HMM model ⁇ .
  • ⁇ ) can be obtained for the time domain feature sequence t of the target point, that is, under the HMM model ⁇ .
  • ⁇ ) can be obtained for the time domain feature sequence t of the target point, that is, under the HMM model ⁇ .
  • the probability of generating the time domain feature sequence t, for the frequency domain feature sequence f of the target point can obtain P(f
  • the HMM model is an existing model, and the determination process of the HMM model can refer to the existing method.
  • step S4064 the type of the obstacle corresponding to the target point is determined according to the probability of the time domain feature sequence of the target point and the frequency domain feature sequence under the hidden Markov model.
  • the probability of the time domain feature sequence of the target point and the frequency domain feature sequence under the hidden Markov model is taken as the probability feature of the target point; the distribution of the probability features according to the target point and the training samples of the obstacles The distribution of probability features determines the type of obstacle at the target point.
  • the time domain feature sequence and the frequency domain feature sequence of the target point are used as training samples to train the HMM model respectively, and two HMM models can be obtained. Further, the probability P(t i
  • the training sample contains a plurality of training samples of various obstacles as target points, and the type of the corresponding obstacle can be marked for each training sample.
  • the corresponding v i of each training sample ie, each target point i
  • V the vector corresponding to the training sample set
  • M is the number of training samples and is a positive integer.
  • the distribution of the probability features of the target point can be represented by the distance feature of the target point to the distance of the training sample set.
  • the distance from the probability set of each training sample (ie, each target point i) to the training sample set is calculated according to the following formula.
  • the distribution of the distance from the training samples of each type of obstacle to the training sample set is calculated, and other types of obstacles are used as clutter, and the distribution of the distance of the clutter to the training sample set is statistically analyzed to distinguish the type of obstacle and the miscellaneous wave. For example, for an obstacle such as a tree, the distance between the probability features of the training samples labeled as trees to the training sample set is counted, and the remaining training samples are all used as clutter, and the probability features of the statistical clutter are added to the training sample set. distance. When the training sample is a tree Clutter Then you can distinguish between trees and other obstacles. Of course, according to the actual situation, the probability characteristics of different obstacles may present different distributions, and the distribution characteristics of a probability feature can be determined for each obstacle through training and statistics.
  • each target point in the detected image is taken as a sample, and the probabilistic feature is obtained by the above method.
  • the obtained probabilistic features are compared with the distribution characteristics of each of the previously estimated obstacle probabilities, and the obstacle type of the target point can be determined.
  • step S408 the timing of opening at least one airbag in the corresponding orientation is determined according to the type of the obstacle and the orientation of the obstacle relative to the drone.
  • the orientation of the obstacle is obtained by radar detection of the image.
  • the timing of opening at least one airbag of the corresponding orientation may be separately set for different obstacles.
  • the corresponding airbag priority opening time is searched according to the type of the obstacle, and the at least one airbag corresponding to the obstacle orientation relative to the unmanned aircraft orientation is preferentially opened with respect to the other airbags according to the priority opening time.
  • the type of obstacle can be divided according to actual needs. For example, trees, buildings, telephone poles, and other obstacles that are higher than the ground, the drones will first collide with these obstacles before they fall to the ground, and the airbags of the corresponding orientation of these obstacles can be compared with other
  • the azimuth airbag opens in advance for a long time.
  • the airbags in other orientations may be opened later, or depending on the subsequent posture of the obstacle, it is determined which airbags are opened.
  • the priority on time is determined, for example, based on the statistical height of the obstacle.
  • the timing of opening at least one airbag in the respective orientation is determined based on the measured vertical distance of the drone to the ground and the type of obstacle.
  • the distance from the drone to the obstacle is determined based on the radar detection, and the timing of opening at least one airbag in the corresponding orientation is determined according to the distance from the drone to the obstacle and the type of the obstacle.
  • determining the distance of the drone to the obstacle and the orientation of the obstacle relative to the drone based on the radar detection determining from the distance of the drone to the obstacle and the orientation of the obstacle relative to the drone The timing of opening at least one airbag in the corresponding orientation.
  • the time of the drone to the obstacle can be determined according to the distance and speed of the drone relative to the obstacle, and then the at least one airbag in the corresponding orientation can be opened for a certain time. This embodiment can be substituted for steps S404 to S408.
  • step S410 at least one airbag in the corresponding orientation is opened with the timing satisfied.
  • the airbag is opened after 0.5 s.
  • the drone opens the airbag on the east side of the cargo after 0.5 s to prevent the drone from first hitting the tree.
  • the drone detects that there is a stone in the west direction with respect to the drone in the horizontal direction, and determines that the airbag is opened for a delay of 1 s when the obstacle is a rock, the drone is delayed by 1 s to open.
  • the airbag is located on the west side of the cargo.
  • the airbags are all disposable products, the replacement is complicated. Therefore, the number of opening of the airbag can be saved according to the actual situation. For example, when the obstacle is a grass or a sand pile, the number of airbag opening can be reduced. And the airbag may be automatically deflated after opening. To ensure the protection effect, a part of the airbag can be opened preferentially, and then the other airbags can be opened.
  • step S412 may also be included after step S410.
  • the predetermined time interval is extended to open the airbag currently located under the cargo. That is, after the UAV has opened a part of the airbag, the corresponding predetermined time interval is determined according to the previously determined obstacle type, and after the corresponding predetermined time interval, the airbag currently located under the cargo is opened.
  • the drone can determine the current attitude based on the gyroscope and open the airbag located below the cargo.
  • the predetermined time interval may be separately set according to the type of the obstacle. For example, when the obstacle is a tree, the predetermined time interval is 1 s, and when the obstacle is a car, the predetermined time interval is 0.5 s.
  • steps S414-S416 may also be included after step S406.
  • the drone adjusts the attitude of the drone according to the orientation of the obstacle so as to avoid the obstacle.
  • the airbag currently located below the cargo is opened.
  • the drone can also be equipped with a pushing device, such as a jet device, which can also push the drone horizontally when it is dropped, so that obstacles can be avoided.
  • a pushing device such as a jet device
  • Obstacles such as mounds can not be avoided. Different obstacles have different volumes and the distances to be avoided are different.
  • the corresponding avoidance scheme is searched according to the type of the obstacle, and the obstacle is avoided according to the avoidance scheme, and the avoidance scheme includes the distance that the drone moves horizontally from the current position.
  • Steps S404 to S410 are not only suitable for the case where the drone is dropped, but also configured to be a case where the drone is normally flying or the drone cannot be turned and the like is out of control.
  • the drone When the drone is flying normally, it may be judged according to the methods of steps S404 to S410 whether or not an obstacle is hit, and the airbag is opened according to the type of the obstacle.
  • the airbag opening timing, the avoidance scheme, and the like can all be designed and adjusted according to the actual flight environment of the drone.
  • the drone can identify the ground obstacle and take corresponding protective measures to further improve the safety of the cargo.
  • the solution in the present disclosure is not only applicable to the protection of the cargo when the drone is falling, but also to the protection of the drone, and further, it can also be applied to the scene in which the obstacle is recognized during the flight of the drone to avoid obstacles.
  • the present disclosure also provides a cargo protection device, which is described below in conjunction with FIG.
  • Figure 5 is a block diagram of some embodiments of the cargo protection device of the present disclosure.
  • the apparatus 50 of this embodiment includes a state determination module 502 and a protection measure triggering module 504.
  • the state determination module 502 is configured to determine whether the drone is in a fall state based on the current acceleration of the drone in the vertical direction and the vertical distance of the drone from the ground.
  • the state determination module 502 is configured to detect the current acceleration of the drone in the vertical direction, and detect the vertical distance of the drone to the ground when the acceleration reaches the acceleration threshold, where the vertical distance is greater than In the case of the distance threshold, it is determined that the drone is in a falling state.
  • the protective measure triggering module 504 is configured to open at least one air bag in the cargo compartment of the drone to protect the cargo when the drone is in a falling state.
  • the airbag is an airbag that is placed around the cargo in the cargo compartment of the drone.
  • the protective measure triggering module 504 is configured to determine the type of obstacle under the drone and the relative obstacles based on information obtained by radar scanning the ground on the drone if the drone is in a falling state. In the orientation of the drone, depending on the type of the obstacle and the orientation of the obstacle relative to the drone, the timing of opening at least one airbag in the corresponding orientation is determined, and at least one airbag in the corresponding orientation is opened when the timing is satisfied.
  • the protective measure triggering module 504 is further configured to extend the predetermined time interval after opening the at least one air bag to open the air bag currently located below the cargo.
  • the protective measure trigger module 504 is further configured to determine an orientation of the obstacle relative to the drone after determining an obstacle type under the drone and an orientation of the obstacle relative to the drone The drone's attitude is adjusted to avoid obstacles and open the airbag currently located below the cargo.
  • protection measure triggering module 504 is configured to perform the following steps:
  • the time domain features and the frequency domain features of the echoes extracting the target points based on different azimuth angles are respectively used as the time domain feature sequence and the frequency domain feature sequence of the target point;
  • the time domain feature sequence and the frequency domain feature sequence of the target point are respectively input into the hidden Markov model, and the time domain feature sequence of the output target point and the frequency domain feature sequence are obtained under the hidden Markov model respectively.
  • the type of the obstacle corresponding to the target point is determined according to the probability of the time domain feature sequence of the target point and the frequency domain feature sequence under the hidden Markov model.
  • the protection measure triggering module 504 is configured to perform a two-dimensional Fourier transform on the image of the region of interest to obtain a wavenumber domain image of the region of interest, and map the wavenumber domain image to the relationship between the frequency and the azimuth angle in the wave number domain.
  • the frequency azimuth image of the region of interest is obtained, and the frequency azimuth image is inverse Fourier transformed along the azimuth direction to obtain the echo of the reconstructed target point.
  • the protection measure triggering module 504 is configured to use the probability of the time domain feature sequence of the target point and the frequency domain feature sequence in the hidden Markov model as the probability feature of the target point, according to the distribution of the probability features of the target point and the obstacles.
  • the distribution of the probability characteristics of the training samples of the object determines the type of obstacle at the target point.
  • the present disclosure also provides a cargo protection system, which is described below in connection with FIG.
  • Figure 6 is a block diagram of some embodiments of the cargo protection system of the present disclosure. As shown in Fig. 6, the system 6 of this embodiment comprises: the cargo protection device 50 of any of the preceding embodiments; and a plurality of airbags 61 disposed around the cargo in the cargo compartment of the drone.
  • the air bag 61 is a blasting air bag, and the air bag 61 is divided into at least one air bag group according to the orientation, and the air bags in the air bag group are connected in parallel to the same ignition device.
  • the cargo protection devices in embodiments of the present disclosure may each be implemented by various computing devices or computer systems, as described below in connection with Figures 7 and 8.
  • Figure 7 is a block diagram of some embodiments of the cargo protection device of the present disclosure.
  • the apparatus 70 of this embodiment includes a memory 710 and a processor 720 coupled to the memory 710, the processor 720 being configured to perform any of the implementations of the present disclosure based on instructions stored in the memory 710. The method of cargo protection in the example.
  • the memory 710 may include, for example, a system memory, a fixed non-volatile storage medium, or the like.
  • the system memory stores, for example, an operating system, an application, a boot loader, a database, and other programs.
  • FIG. 8 is a block diagram of another embodiment of the cargo protection device of the present disclosure.
  • the apparatus 80 of this embodiment includes a memory 810 and a processor 820, which is similar to the memory 710 and the processor 720, respectively.
  • the cargo protection device 80 may also include an input and output interface 830, a network interface 840, a storage interface 850, and the like. These interfaces 830, 840, 850 and the memory 810 and the processor 820 can be connected, for example, via a bus 860.
  • the input/output interface 830 provides a connection interface for input and output devices such as a display, a mouse, a keyboard, and a touch screen.
  • the network interface 840 provides a connection interface for various networked devices, such as a database server or a cloud storage server.
  • the storage interface 850 provides a connection interface for an external storage device such as an SD card or a USB flash drive.
  • a computer readable storage medium having stored thereon a computer program, wherein the program is executed by a processor to implement the cargo protection method of any of the foregoing embodiments.
  • embodiments of the present disclosure can be provided as a method, system, or computer program product. Accordingly, the present disclosure may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware aspects. Moreover, the present disclosure may take the form of a computer program product embodied on one or more computer-usable non-transitory storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code. .
  • the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
  • the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps that are configured to implement the functions specified in one or more blocks of the flowchart or in a block or blocks of the flowchart.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Remote Sensing (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Electromagnetism (AREA)
  • Software Systems (AREA)
  • Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Operations Research (AREA)
  • Automation & Control Theory (AREA)
  • Probability & Statistics with Applications (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Algebra (AREA)
  • Multimedia (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Quality & Reliability (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Robotics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Radar Systems Or Details Thereof (AREA)
  • Traffic Control Systems (AREA)

Abstract

本公开涉及一种货物保护方法、装置、系统和计算机可读存储介质，涉及无人机技术领域。本公开的方法包括：根据无人机当前在竖直方向上的加速度和无人机当前到地面的垂直距离确定无人机是否处于坠落状态，在无人机处于坠落状态的情况下，打开无人机货舱内的至少一个气囊，以便对货物进行安全保护。本公开当发生无人机坠落情况时，至少一个气囊被打开，气囊相对于降落伞打开更为及时，能够直接避免货物遭受到撞击，降低了无人机货物损毁的概率。

Description

货物保护方法、装置、系统和计算机可读存储介质
相关申请的交叉引用
本申请是以CN申请号为201711363753.9，申请日为2017年12月18日的申请为基础，并主张其优先权，该CN申请的公开内容在此作为整体引入本申请中。
技术领域
本公开涉及无人机技术领域，特别涉及一种货物保护方法、装置、系统和计算机可读存储介质。
背景技术
随着无人机的应用越来越多，无人机的安全问题也开始受到大家的关注。
在货运方面无人机技术目前已经比较成熟，安全性能也较好，但仍然存在故障的可能性。当无人机高空作业时，一旦飞行系统失灵，可能导致贵重货物损毁，给客户和商家带来不可挽回的损失。
目前对无人机坠落的保护技术越来越多，这些技术大多是通过内置的芯片智能监控飞机坠落，在危险情况下会自动开启无人机降落伞，保护无人机的同时也防止坠落所造成的伤害。
发明内容
发明人发现：目前检测出飞机失控、失速状态，并在空中自动抛伞进行保护，只对无人机本体进行保护，对无人机承载物品保护不足，对易碎品，贵重物品无法进行保护。并且降落伞开启反应迟钝，需要时长较多。失控状态下对降落地点不能自主选择，地面有障碍物时容易发生降落失败，即使有坠落保护装置也不能避免无人机和货物的损坏。
本公开所要解决的一个技术问题是：如何提高无人机货物的安全性，减少货物损坏的概率。
根据本公开的一些实施例，提供的一种货物保护方法，包括：根据无人机当前在竖直方向上的加速度和无人机当前到地面的垂直距离确定无人机是否处于坠落状态，在无人机处于坠落状态的情况下，打开无人机货舱内的至少一个气囊，以便对货物进 行安全保护。
在一些实施例中，根据无人机当前在竖直方向上的加速度和无人机当前到地面的垂直距离确定无人机是否处于坠落状态包括：检测无人机当前在竖直方向上的加速度；在加速度达到加速度阈值的情况下，检测无人机当前到地面的垂直距离；在垂直距离大于距离阈值的情况下，确定无人机处于坠落状态。
在一些实施例中，在无人机处于坠落状态的情况下，打开无人机货舱内至少一个气囊包括：在无人机处于坠落状态的情况下，根据无人机上的雷达扫描地面得到的信息，确定无人机下方的障碍物类型以及障碍物相对于无人机的方位；根据障碍物的类型以及障碍物相对于无人机的方位，确定打开相应方位上的至少一个气囊的时机；在时机满足的情况下打开相应方位上的至少一个气囊。
在一些实施例中，在打开至少一个气囊后，还包括：延长预定的时间间隔；将当前位于货物下方的气囊打开。
在一些实施例中，在确定无人机下方的障碍物类型以及障碍物相对于无人机的方位后，还包括：根据障碍物相对于无人机的方位对无人机的姿态进行调整，以便躲避障碍物；将当前位于货物下方的气囊打开。
在一些实施例中，根据无人机上的雷达扫描地面得到的信息，确定无人机坠落地面的障碍物的类型包括：对无人机上的雷达扫描地面得到的感兴趣区域图像中的目标点重构回波；基于不同方位角提取目标点的回波的时域特征和频域特征分别作为目标点的时域特征序列和频域特征序列；将目标点的时域特征序列和频域特征序列分别输入隐马尔可夫模型，分别得到输出的目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率；根据目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率确定目标点对应的障碍物的类型。
在一些实施例中，对感兴趣区域图像中的目标点重构回波包括：对感兴趣区域图像进行二维傅里叶变换，得到该感兴趣区域的波数域图像；根据波数域中频率与方位角的关系，将波数域图像映射到频率和方位角域中，得到感兴趣区域的频率方位角域图像；将频率方位角域图像沿方位角方向进行逆傅里叶变换，得到重构的目标点的回波。
在一些实施例中，根据目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率确定目标点对应的障碍物的类型包括：将目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率作为目标点的概率特征；根据目标点的概率特征的分 布与各障碍物的训练样本的概率特征的分布，确定目标点的障碍物类型。
根据本公开的另一些实施例，提供的一种货物保护装置，包括：状态确定模块，被配置为根据无人机当前在竖直方向上的加速度和无人机当前到地面的垂直距离确定无人机是否处于坠落状态；保护措施触发模块，被配置为在无人机处于坠落状态的情况下，打开无人机货舱内的至少一个气囊，以便对货物进行安全保护。
在一些实施例中，状态确定模块被配置为检测无人机当前在竖直方向上的加速度，在加速度达到加速度阈值的情况下，检测无人机当前到地面的垂直距离，在垂直距离大于距离阈值的情况下，确定无人机处于坠落状态。
在一些实施例中，保护措施触发模块被配置为在无人机处于坠落状态的情况下，根据无人机上的雷达扫描地面得到的信息，确定无人机下方的障碍物类型以及障碍物相对于无人机的方位，根据障碍物的类型以及障碍物相对于无人机的方位，确定打开相应方位上的至少一个气囊的时机，在时机满足的情况下打开相应方位上的至少一个气囊。
在一些实施例中，保护措施触发模块还被配置为在打开至少一个气囊后，延长预定的时间间隔，将当前位于货物下方的气囊打开。
在一些实施例中，保护措施触发模块还被配置为在确定无人机下方的障碍物类型以及障碍物相对于无人机的方位后，根据障碍物相对于无人机的方位对无人机的姿态进行调整，以便躲避障碍物，将当前位于货物下方的气囊打开。
在一些实施例中，保护措施触发模块被配置为：对无人机上的雷达扫描地面得到的感兴趣区域图像中的目标点重构回波；基于不同方位角提取目标点的回波的时域特征和频域特征，分别作为目标点的时域特征序列和频域特征序列；将目标点的时域特征序列和频域特征序列分别输入隐马尔可夫模型，分别得到输出的目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率；根据目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率确定目标点对应的障碍物的类型。
在一些实施例中，保护措施触发模块被配置为：对感兴趣区域图像进行二维傅里叶变换，得到该感兴趣区域的波数域图像；根据波数域中频率与方位角的关系，将波数域图像映射到频率和方位角域中，得到感兴趣区域的频率方位角域图像；将频率方位角域图像沿方位角方向进行逆傅里叶变换，得到重构的目标点的回波。
在一些实施例中，保护措施触发模块被配置为将目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率作为目标点的概率特征，根据目标点的概率特征的 分布与各障碍物的训练样本的概率特征的分布，确定目标点的障碍物类型。
根据本公开的又一些实施例，提供的一种货物保护系统，包括：前述任一个实施例的货物保护装置；以及位于无人机货舱内围绕货物布置的多个气囊。
在一些实施例中，气囊为燃爆式气囊，气囊根据方位被划分为至少一个气囊组，气囊组内的气囊并联至同一个点火装置。
根据本公开的再一些实施例，提供的一种货物保护装置，包括：存储器；以及耦接至存储器的处理器，处理器被配置为基于存储在存储器设备中的指令，执行如前述任一个实施例的货物保护方法。
根据本公开的又一些实施例，提供的一种计算机可读存储介质，其上存储有计算机程序，其中，该程序被处理器执行时实现前述任一个实施例的货物保护方法。
本公开当发生无人机坠落情况时，至少一个气囊被打开，气囊相对于降落伞打开更为及时，能够直接避免货物遭受到撞击，降低了无人机货物损毁的概率。
通过以下参照附图对本公开的示例性实施例的详细描述，本公开的其它特征及其优点将会变得清楚。
附图说明
此处所说明的附图用来提供对本公开的进一步理解，构成本申请的一部分，本公开的示意性实施例及其说明被配置为解释本公开，并不构成对本公开的不当限定。在附图中：
图1示出本公开的一些实施例的货物保护方法的流程示意图。
图2示出本公开的一些实施例的无人机货舱气囊分布的俯视图.
图3示出本公开的一些实施例的气囊的结构示意图。
图4A示出本公开的另一些实施例的货物保护方法的流程示意图。
图4B示出本公开的又一些实施例的货物保护方法的流程示意图。
图5示出本公开的一些实施例的货物保护装置的结构示意图。
图6示出本公开的一些实施例的货物保护系统的结构示意图。
图7示出本公开的另一些实施例的货物保护装置的结构示意图。
图8示出本公开的又一些实施例的货物保护装置的结构示意图。
具体实施方式
下面将结合本公开实施例中的附图，对本公开实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本公开一部分实施例，而不是全部的实施例。以下对至少一个示例性实施例的描述实际上仅仅是说明性的，决不作为对本公开及其应用或使用的任何限制。基于本公开中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本公开保护的范围。
针对目前无人机运输时对货物的保护不足，容易造成坠落时货物损毁的问题，提出本方案。下面结合图1描述本公开的货物保护方法。
图1为本公开货物保护方法一些实施例的流程图。如图1所示，该实施例的方法包括：步骤S102～S104。
在步骤S102中，根据无人机当前在竖直方向上的加速度和无人机当前到地面的垂直距离确定无人机是否处于坠落状态。
在一些实施例中，首先检测无人机竖直方向加速度，在竖直方向加速度达到加速度阈值的情况下，再检测无人机到地面的垂直距离，在垂直距离达到距离阈值的情况下，确定无人机处于坠落状态。可以通过加速度传感器检测无人机竖直方向加速度。可以通过超声波测距系统、测距雷达等检测无人机到地面的垂直距离。对于无人机竖直方向加速度和到地面的垂直距离可以进行周期性检测，一旦发现符合坠落状态的情况，立即执行后续保护措施。
在步骤S104中，在无人机处于坠落状态的情况下，打开无人机货舱内的至少一个气囊，以便对货物进行安全保护。
无人机货舱内可以设置围绕货物布置的多个气囊。气囊的大小、数量和位置可以根据实际货物的大小和货舱的空间进行调整。为进一步加强对货物的保护，可以在货舱外侧环绕设置多个气囊。如图2所示，为一些实施例中无人机货舱气囊分布的俯视图。在货物周围以及货舱周围布置了多个气囊。同时，还可以在无人机上设置多个气囊，分布在无人机的各个部位。无人机上设置的气囊可以和货舱内的相应方位的气囊同时开启，在无人机坠落时对无人机起到保护作用。气囊可以和安装于无人机上的降落伞配合使用。在无人机处于坠落状态的情况下，降落伞可以降低无人机下落的速度，配合开启货舱内、货舱周围或无人机上的气囊能够对货物或无人机起到更好的保护。
在一些实施例中，无人机可以检测当前的飞行姿态，例如利用陀螺仪检测无人机相对于水平方向的倾角，根据飞行姿态确定坠落时接触地面的部位，打开相应部位的气囊。例如，当无人机平行于地面坠落时，无人机底部首先接触地面，可以主要打开 货物底部的气囊，还可以打开无人机底部的气囊，而无人机或者货舱上部的气囊可以不打开或者相对于底部气囊滞后打开。又例如，当无人机头部向下坠落时，可以优先打开无人机头部以及相应方位的货舱内的气囊。优先打开的气囊可以在无人机坠落时第一时间起到保护作用。进一步，坠落时某些部位受到的冲击力达到阈值的情况下，可以触发相应部位的气囊开启，持续进行保护。
进一步，在无人机可以自动调整飞行姿态的情况，则可以调整飞行姿态选择坠落时的接触地面的部位，并打开相应部位的气囊。
如图3所示为单体气囊的结构示意图。气囊例如为燃爆式气囊，气囊300上设置有雷管302，雷管位于气体发生器304内。可以由电子点火装置306通过引线308引燃雷管，进一步雷管引燃周围的反应物产生气体充满气囊。反应物例如为叠氮化钠，受撞击或受热可以引发快速热分解反应，产生气体。当存在多个气囊时，可以根据方位被划分为至少一个气囊组，气囊组内的气囊并联至同一个点火装置，可以同时开启。燃爆式气囊和并联的设置方式能够进一步加快气囊打开的速度，提高货物的安全性。
上述实施例的方法，在无人机货仓内设置多个围绕货物布置的多个气囊，当发生无人机坠落情况时，至少一个气囊被打开，气囊相对于降落伞打开更为及时，能够直接避免货物遭受到撞击，降低了无人机货物损毁的概率。
在无人机坠落的情况下，无人机还可以自动对地面进行扫描，根据扫描图像确定地面是否存在障碍物，并根据地面的障碍物确定气囊打开方式或避障。下面结合图4A和4B描述本公开货物保护方法的另一些实施例。
图4A为本公开货物保护方法另一些实施例的流程图。如图4A所示，该实施例的方法包括：步骤S402～S416。
在步骤S402中，根据无人机当前在竖直方向上的加速度和无人机到地面的垂直距离，判断无人机是否处于坠落状态。如果处于坠落状态，则执行步骤S404，否则，间隔预设周期后重新执行步骤S402。
在步骤S404中，获取无人机上的雷达扫描地面得到的感兴趣区域图像。
无人机上可以设置雷达探测装置，对地面进行探测，获得雷达探测图像。从雷达探测图像中可根据局部灰度特征提取ROI(Region of Interest，感兴趣区域)图像。ROI图像中包含至少一个目标点，ROI图像为斜距-方位图像。
在步骤S406中，根据感兴趣区域图像确定无人机坠落地面的障碍物的类型以及障碍物相对于无人机的方位。
在一些实施例中，如图4B所示，该步骤S406可以包括以下子步骤S4061-S4064。
在步骤S4061中，对感兴趣区域图像中的目标点重构回波。
在一些实施例中，对感兴趣区域图像进行二维傅里叶变换，得到该感兴趣区域的波数域图像；根据波数域中频率与方位角的关系，将波数域图像映射到频率和方位角域中，得到感兴趣区域的频率方位角域图像；将频率方位角域图像沿方位角方向进行逆傅里叶变换，得到重构的目标点的回波。
假设ROI图像表示为f(x，r)，对其进行二维傅里叶变换，得到波数域图像
Figure PCTCN2018108955-appb-000001
x、r分别表示斜距和方位位置，k x、k r分别表示斜距波数和方位波数。根据后向投影成像算法模型，存在以下公式：
Figure PCTCN2018108955-appb-000002
其中，
Figure PCTCN2018108955-appb-000003
f表示频率，c表示光速。根据公式(1)可以将
Figure PCTCN2018108955-appb-000004
映射到
Figure PCTCN2018108955-appb-000005
域中，即频率方位角域，形成目标点在各个入射角下的频率响应。在
Figure PCTCN2018108955-appb-000006
域中，沿方位角
Figure PCTCN2018108955-appb-000007
做逆傅里叶变换，就能得到目标点在该方位角对应的回波。
在步骤S4062中，基于不同方位角提取目标点的回波的时域特征和频域特征，分别作为目标点的时域特征序列和频域特征序列。
可以采用时频原子分析法提取目标点的回波在不同方位角的情况下的时域特征和频域特征，例如，提取时域双峰间距和频域凹点。时频原子分析法为现有技术，在此不再赘述。
时域特征例如表示为t，频域特征例如表示为f，对目标点的回波进行方位等角度N次采样，得到t＝(t 1，t 2，......t N)，f＝(f 1，f 2，......f N)。
在步骤S4063中，将目标点的时域特征序列和频域特征序列分别输入隐马尔可夫模型，分别得到输出的目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率。
由于无人机在进行雷达探测时是移动的，即雷达信号相对于目标点的方位角是随时间变化的，因此，基于方位角提取的时域特征和频域特征是随时间变化的序列。从而，可以采用HMM(隐马尔可夫模型)对目标点的时域特征序列和频域特征序列进行识别。
不同类型的目标点(例如树、石块、楼房、汽车、人等)由于形状，结构等不同造成的回波不同，进而时域特征和频域特征也是不同的。因此，可以利用时域特征和频域特征识别目标点的类别。
HMM的参数主要包括初始状态分布π，状态转移概率矩阵B和隐状态下产生可见状态的概率矩阵U。设每一个方位采样为一个隐状态，表示为X n，n＝1，2，......，N，N表示隐状态的数量，为正整数。X n的初始概率为1，即P(X n)＝1。根据时域特征序列或频域特征序列提取过程可知，对应的隐状态是固定的。因此，状态转移概率矩阵B＝{b mn|m，n＝1，2，......，N}中，符合以下公式：
Figure PCTCN2018108955-appb-000008
HMM训练过程简化为求隐状态下产生可见状态的概率矩阵U。根据时域特征序列或频域特征序列为连续随机变量的特征，U可以根据观测值出现的概率进行确定。进一步，确定隐状态下产生可见状态的概率矩阵U后，假设训练后的HMM模型为γ，则针对目标点的时域特征序列t，可以得到P(t|γ)，即在HMM模型γ下产生时域特征序列t的概率，针对目标点的频域特征序列f，可以得到P(f|γ)，即在HMM模型γ下产生频域特征序列f的概率。HMM模型为现有模型，HMM模型的确定过程可以参考现有方法。
在步骤S4064中，根据目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率确定目标点对应的障碍物的类型。
在一些实施例中，将目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率作为目标点的概率特征；根据目标点的概率特征的分布与各障碍物的训练样本的概率特征的分布，确定目标点的障碍物类型。
在训练HMM模型时，目标点的时域特征序列和频域特征序列分别作为训练样本训练HMM模型，可得到两种HMM模型。进一步，可得到每个目标点i的时域特征序列t i在其对应的HMM模型γ t下的概率P(t i|γ t)，i为正整数。还可以得到每个目标点i的频域特征序列f i在其对应的HMM模型γ t下的概率P(f i|γ f)。将目标点i对应的P(t i|γ t)和P(f i|γ f)构成新的矢量v i＝(P(t i|γ t)，P(f i|γ f))，作为目标点i的概率特征。
训练样本中包含了各类障碍物作为目标点的多个训练样本，并可以针对每个训练样本标注其对应的障碍物的类型。利用训练样本完成HMM模型的训练后，则可以得到每个训练样本(即每个目标点i)对应的v i，进而组成训练样本集对应的矢量V＝(v 1，v 2，......，v M)，M为训练样本的数量，为正整数。目标点的概率特征的分布可以用目标点的概率特征到训练样本集的距离表示。根据以下公式计算每个训练样本(即每个目标点i)的概率特征到训练样本集的距离。
Figure PCTCN2018108955-appb-000009
Figure PCTCN2018108955-appb-000010
Figure PCTCN2018108955-appb-000011
为训练样本(即每个目标点i)到训练样本集的距离，
Figure PCTCN2018108955-appb-000012
为v i的均值。统计每种类型的障碍物的训练样本到训练样本集的距离的分布，同时将其他类型的障碍物作为杂波，统计杂波到训练样本集的距离的分布，以区分该类型障碍物和杂波。例如，针对树这种障碍物，统计各个标注为树的训练样本的概率特征到训练样本集的距离，同时，将剩余的训练样本都作为杂波，统计杂波的概率特征到训练样本集的距离。当训练样本为树时
Figure PCTCN2018108955-appb-000013
而杂波的
Figure PCTCN2018108955-appb-000014
则可以区分树与其他障碍物。当然，根据实际情形，不同的障碍物的概率特征可能呈现不同的分布，通过训练与统计可以针对每一种障碍物确定一种概率特征的分布特征。
训练结束后，当无人机利用雷达进行探测，将探测图像中的每个目标点都作为样本，利用上述方法得到其概率特征。将得到的概率特征与之前统计的每一种障碍物概率特征的分布特征进行比对，进而可以确定目标点的障碍物类型。
在步骤S408中，根据障碍物的类型以及障碍物相对于无人机的方位，确定打开相应方位上的至少一个气囊的时机。
障碍物的方位通过雷达探测图像即可获得。针对不同的障碍物可以分别设置打开相应方位的至少一个气囊的时机。在一些实施例中，根据障碍物的类型查找对应的气囊优先开启时间，将障碍物相对于无人机方位对应的至少一个气囊按照优先开启时间相对于其他气囊优先打开。障碍物的类型可以根据实际需求进行划分。例如，树木、楼房、电线杆这种高出地面较多的障碍物，无人机坠落过程首先会碰撞到这些障碍物进而才会坠落地面，可以将这些障碍物的相应方位的气囊相对于其他方位的气囊提前较长时间打开。而其他方位的气囊可以晚一些打开，或者，根据撞到障碍物后续的姿态再确定打开哪些气囊。优先开启时间例如是根据障碍物的统计高度确定的。
在一些实施例中，在无人机坠落的情况下，根据测得的无人机到地面的垂直距离和障碍物的类型，确定打开相应方位上的至少一个气囊的时机。或者，根据雷达探测确定无人机到障碍物的距离，并根据无人机到障碍物的距离和障碍物的类型，确定打开相应方位上的至少一个气囊的时机。
在一些实施例中，根据雷达探测确定无人机到障碍物的距离和障碍物相对于无人机的方位，根据无人机到障碍物的距离和障碍物相对于无人机的方位，确定打开相应方位上的至少一个气囊的时机。可以根据无人机相对于障碍物的距离和速度，确定无人机到障碍物的时间，进而提取一定时间打开相应方位上至少一个气囊。该实施例可 以替代步骤S404～S408。
在步骤S410中，在时机满足的情况下打开相应方位上的至少一个气囊。
例如，当无人机检测到相对于无人机在水平方向上位于正东方向存在障碍物为树木，并根据确定在障碍物为树木的情况下在0.5s后打开气囊。无人机则在0.5s后打开位于货物正东侧的气囊，以免无人机首先撞到树木损坏。或者，当无人机检测到相对于无人机在水平方向上位于正西方向存在石块，并确定在障碍物为石块的情况下延时1s打开气囊，无人机则延时1s打开位于货物正西侧的气囊。
由于气囊均为一次性产品，更换较为复杂。因此，可以根据实际情况节省气囊的开启数量，例如，在障碍物为草地或沙堆等情况，可以减少气囊开启数量。并且气囊打开后可能会自动放气，为保证保护效果，可以优先开启一部分气囊，之后再开启其他气囊。
在一些实施例中，在步骤S410之后还可以包括步骤S412。在步骤S412中，延长预定的时间间隔，将当前位于货物下方的气囊打开。即在无人机已开启部分气囊后，根据之前确定的障碍物类型，确定对应的预定时间间隔，并在对应的预定时间间隔之后，打开当前位于货物下方的气囊。无人机可以根据陀螺仪确定当前的姿态，打开位于货物下方的气囊。预定的时间间隔可以根据障碍物的类型分别进行设置，例如，障碍物为树木时，预定的时间间隔为1s，障碍物为汽车时，预定的时间间隔为0.5s。
在一些实施例中，在步骤S406之后还可以包括步骤S414-S416。在步骤S414中，无人机根据障碍物的方位对无人机的姿态进行调整，以便躲避障碍物。在步骤S416中，将当前位于货物下方的气囊打开。无人机上还可以安装推动装置，例如喷气装置，在坠落时还可以推动无人机水平移动，则可以对障碍物进行躲避。例如，对于人、汽车这类障碍物，为避免无人机坠落带来安全问题，需要进行躲避；对于树木，电线杆等为避免无人机坠落后无法取下来也可以进行躲避；而对于草地、土堆等障碍物可以不躲避。不同的障碍物体积不同，需要躲避的距离也不同。因此，可以根据各种障碍物设置不同的躲避方案，坠落时根据障碍物的类型查找对应躲避方案，根据躲避方案对障碍物进行躲避，躲避方案包括无人机由当前位置水平移动的距离。
步骤S404～S410不仅适被配置为无人机坠落的情况，还适被配置为无人机正常飞行或无人机无法转向等失控的故障的情况。无人机正常飞行时也可以根据步骤S404～S410的方法判断是否会撞到障碍物，根据障碍物的类型进行气囊的开启。
上述实施例的方案，气囊开启时机、躲避方案等都可以根据无人机实际的飞行环 境进行设计和调整。利用上述实施例中的方案，无人机可以识别地面障碍物进而采取相应的保护措施，进一步提高货物的安全性。本公开中的方案不仅适用于无人机坠落时对于货物的保护，也适用于对于无人机的保护，进一步，还可以应用于无人机飞行过程中识别障碍物进行避障的场景。
本公开还提供一种货物保护装置，下面结合图5进行描述。
图5为本公开货物保护装置的一些实施例的结构图。如图5所示，该实施例的装置50包括：状态确定模块502，保护措施触发模块504。
状态确定模块502，被配置为根据无人机当前在竖直方向上的加速度和无人机当前到地面的垂直距离确定无人机是否处于坠落状态。
在一些实施例中，状态确定模块502被配置为检测无人机当前在竖直方向上的加速度，在加速度达到加速度阈值的情况下，检测无人机当前到地面的垂直距离，在垂直距离大于距离阈值的情况下，确定无人机处于坠落状态。
保护措施触发模块504，被配置为在无人机处于坠落状态的情况下，打开无人机货舱内的至少一个气囊，以便对货物进行安全保护。
气囊为位于无人机货舱内，围绕货物布置的气囊。在一些实施例中，保护措施触发模块504被配置为在无人机处于坠落状态的情况下，根据无人机上的雷达扫描地面得到的信息，确定无人机下方的障碍物类型以及障碍物相对于无人机的方位，根据障碍物的类型以及障碍物相对于无人机的方位，确定打开相应方位上的至少一个气囊的时机，在时机满足的情况下打开相应方位上的至少一个气囊。
在一些实施例中，保护措施触发模块504还被配置为在打开至少一个气囊后，延长预定的时间间隔，将当前位于货物下方的气囊打开。
在一些实施例中，保护措施触发模块504还被配置为在确定无人机下方的障碍物类型以及障碍物相对于无人机的方位后，根据障碍物相对于所述无人机的方位对无人机的姿态进行调整，以便躲避障碍物，将当前位于货物下方的气囊打开。
进一步，保护措施触发模块504被配置为执行以下步骤：
对无人机上的雷达扫描地面得到的感兴趣区域图像中的目标点重构回波；
基于不同方位角提取目标点的回波的时域特征和频域特征分别作为目标点的时域特征序列和频域特征序列；
将目标点的时域特征序列和频域特征序列分别输入隐马尔可夫模型，分别得到输出的目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率；
根据目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率确定目标点对应的障碍物的类型。
进一步，保护措施触发模块504被配置为对感兴趣区域图像进行二维傅里叶变换，得到该感兴趣区域的波数域图像，根据波数域中频率与方位角的关系，将波数域图像映射到频率和方位角域中，得到感兴趣区域的频率方位角域图像，将频率方位角域图像沿方位角方向进行逆傅里叶变换，得到重构的目标点的回波。
进一步，保护措施触发模块504被配置为将目标点的时域特征序列和频域特征序列在隐马尔可夫模型下的概率作为目标点的概率特征，根据目标点的概率特征的分布与各障碍物的训练样本的概率特征的分布，确定目标点的障碍物类型。
本公开还提供一种货物保护系统，下面结合图6进行描述。
图6为本公开货物保护系统的一些实施例的结构图。如图6所示，该实施例的系统6包括：前述任一个实施例中的货物保护装置50；以及位于无人机货舱内围绕货物布置的多个气囊61。
在一些实施例中，气囊61为燃爆式气囊，气囊61根据方位被划分为至少一个气囊组，气囊组内的气囊并联至同一个点火装置。
本公开的实施例中的货物保护装置可各由各种计算设备或计算机系统来实现，下面结合图7以及图8进行描述。
图7为本公开货物保护装置的一些实施例的结构图。如图7所示，该实施例的装置70包括：存储器710以及耦接至该存储器710的处理器720，处理器720被配置为基于存储在存储器710中的指令，执行本公开中任意一些实施例中的货物保护方法。
其中，存储器710例如可以包括系统存储器、固定非易失性存储介质等。系统存储器例如存储有操作系统、应用程序、引导装载程序(Boot Loader)、数据库以及其他程序等。
图8为本公开货物保护装置的另一些实施例的结构图。如图8所示，该实施例的装置80包括：存储器810以及处理器820，存储器810以及处理器820分别与存储器710以及处理器720类似。货物保护装置80还可以包括输入输出接口830、网络接口840、存储接口850等。这些接口830，840，850以及存储器810和处理器820之间例如可以通过总线860连接。其中，输入输出接口830为显示器、鼠标、键盘、触摸屏等输入输出设备提供连接接口。网络接口840为各种联网设备提供连接接口，例如可以连接到数据库服务器或者云端存储服务器等。存储接口850为SD卡、U盘等外置 存储设备提供连接接口。
根据本公开的一些实施例，提供的一种计算机可读存储介质，其上存储有计算机程序，其中，该程序被处理器执行时实现前述任一个实施例的货物保护方法。
本领域内的技术人员应当明白，本公开的实施例可提供为方法、系统、或计算机程序产品。因此，本公开可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本公开可采用在一个或多个其中包含有计算机可用程序代码的计算机可用非瞬时性存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本公开是参照根据本公开实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解为可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生被配置为实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供被配置为实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
以上所述仅为本公开的较佳实施例，并不用以限制本公开，凡在本公开的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本公开的保护范围之内。

Claims (20)

  1. 一种货物保护方法，包括：
    根据无人机当前在竖直方向上的加速度和所述无人机当前到地面的垂直距离确定所述无人机是否处于坠落状态；
    在所述无人机处于坠落状态的情况下，打开所述无人机货舱内的至少一个气囊，以便对货物进行安全保护。
  2. 根据权利要求1所述的货物保护方法，其中，所述根据无人机当前在竖直方向上的加速度和所述无人机当前到地面的垂直距离确定所述无人机是否处于坠落状态包括：
    检测所述无人机当前在竖直方向上的加速度；
    在所述加速度达到加速度阈值的情况下，检测所述无人机当前到地面的垂直距离；
    在所述垂直距离大于距离阈值的情况下，确定所述无人机处于坠落状态。
  3. 根据权利要求1所述的货物保护方法，其中，在所述无人机处于坠落状态的情况下，打开所述无人机货舱内至少一个气囊包括：
    在所述无人机处于坠落状态的情况下，根据所述无人机上的雷达扫描地面得到信息，确定所述无人机下方的障碍物类型以及所述障碍物相对于所述无人机的方位；
    根据所述障碍物的类型以及所述障碍物相对于所述无人机的方位，确定打开相应方位上的至少一个气囊的时机；
    在所述时机满足的情况下打开相应方位上的至少一个气囊。
  4. 根据权利要求3所述的货物保护方法，其中，在打开所述至少一个气囊后，还包括：
    延长预定的时间间隔；
    将当前位于货物下方的气囊打开。
  5. 根据权利要求3所述的货物保护方法，其中，在确定所述无人机下方的障碍物类型以及所述障碍物相对于所述无人机的方位后，还包括：
    根据所述障碍物相对于所述无人机的方位对所述无人机的姿态进行调整，以便躲避所述障碍物；
    将当前位于货物下方的气囊打开。
  6. 根据权利要求3所述的货物保护方法，其中，所述根据所述无人机上的雷达扫 描地面得到的信息，确定所述无人机坠落地面的障碍物的类型包括：
    对所述无人机上的雷达扫描地面得到的感兴趣区域图像中的目标点重构回波；
    基于不同方位角提取所述目标点的回波的时域特征和频域特征，分别作为所述目标点的时域特征序列和频域特征序列；
    将所述目标点的时域特征序列和频域特征序列分别输入隐马尔可夫模型，分别得到输出的所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率；
    根据所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率确定所述目标点对应的障碍物的类型。
  7. 根据权利要求6所述的货物保护方法，其中，所述对所述感兴趣区域图像中的目标点重构回波包括：
    对所述感兴趣区域图像进行二维傅里叶变换，得到该感兴趣区域的波数域图像；
    根据波数域中频率与方位角的关系，将所述波数域图像映射到频率和方位角域中，得到所述感兴趣区域的频率方位角域图像；
    将频率方位角域图像沿方位角方向进行逆傅里叶变换，得到重构的所述目标点的回波。
  8. 根据权利要求6所述的货物保护方法，其中，所述根据所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率确定所述目标点对应的障碍物的类型包括：
    将所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率作为所述目标点的概率特征；
    根据所述目标点的概率特征的分布与各障碍物的训练样本的概率特征的分布，确定所述目标点的障碍物类型。
  9. 一种货物保护装置，包括：
    状态确定模块，被配置为根据无人机当前在竖直方向上的加速度和所述无人机当前到地面的垂直距离确定所述无人机是否处于坠落状态；
    保护措施触发模块，被配置为在所述无人机处于坠落状态的情况下，打开所述无人机货舱内的至少一个气囊，以便对货物进行安全保护。
  10. 根据权利要求9所述的货物保护装置，其中，
    所述状态确定模块被配置为检测所述无人机当前在竖直方向上的加速度，在所述 加速度达到加速度阈值的情况下，检测所述无人机当前到地面的垂直距离，在所述垂直距离大于距离阈值的情况下，确定所述无人机处于坠落状态。
  11. 根据权利要求9所述的货物保护装置，其中，
    所述保护措施触发模块被配置为在所述无人机处于坠落状态的情况下，根据所述无人机上的雷达扫描地面得到的信息，确定所述无人机下方的障碍物类型以及所述障碍物相对于所述无人机的方位，根据所述障碍物的类型以及所述障碍物相对于所述无人机的方位，确定打开相应方位上的至少一个气囊的时机，在所述时机满足的情况下打开相应方位上的至少一个气囊。
  12. 根据权利要求11所述的货物保护装置，其中，
    所述保护措施触发模块还被配置为在打开所述至少一个气囊后，延长预定的时间间隔，将当前位于货物下方的气囊打开。
  13. 根据权利要求11所述的货物保护装置，其中，
    所述保护措施触发模块还被配置为在确定所述无人机下方的障碍物类型以及所述障碍物相对于所述无人机的方位后，根据所述障碍物相对于所述无人机的方位对所述无人机的姿态进行调整，以便躲避所述障碍物，将当前位于货物下方的气囊打开。
  14. 根据权利要求11所述的货物保护装置，其中，
    所述保护措施触发模块被配置为：
    对所述无人机上的雷达扫描地面得到的感兴趣区域图像中的目标点重构回波；
    基于不同方位角提取所述目标点的回波的时域特征和频域特征，分别作为所述目标点的时域特征序列和频域特征序列；
    将所述目标点的时域特征序列和频域特征序列分别输入隐马尔可夫模型，分别得到输出的所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率；
    根据所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率确定所述目标点对应的障碍物的类型。
  15. 根据权利要求14所述的货物保护装置，其中，
    所述保护措施触发模块被配置为：
    对所述感兴趣区域图像进行二维傅里叶变换，得到该感兴趣区域的波数域图像；
    根据波数域中频率与方位角的关系，将所述波数域图像映射到频率和方位角域中，得到所述感兴趣区域的频率方位角域图像；
    将频率方位角域图像沿方位角方向进行逆傅里叶变换，得到重构的所述目标点的回波。
  16. 根据权利要求14所述的货物保护装置，其中，
    所述保护措施触发模块被配置为将所述目标点的时域特征序列和频域特征序列在所述隐马尔可夫模型下的概率作为所述目标点的概率特征，根据所述目标点的概率特征的分布与各障碍物的训练样本的概率特征的分布，确定所述目标点的障碍物类型。
  17. 一种货物保护系统，包括：权利要求9-16任一项所述的货物保护装置；以及
    位于无人机货舱内围绕货物布置的多个气囊。
  18. 根据权利要求17所述的货物保护系统，其中，
    所述气囊为燃爆式气囊，所述气囊根据方位被划分为至少一个气囊组，所述气囊组内的气囊并联至同一个点火装置。
  19. 一种货物保护装置，包括：
    存储器；以及
    耦接至所述存储器的处理器，所述处理器被配置为基于存储在所述存储器中的指令，执行如权利要求1-8任一项所述的货物保护方法。
  20. 一种计算机可读存储介质，其上存储有计算机程序，其中，该程序被处理器执行时实现权利要求1-8任一项所述方法的步骤。
PCT/CN2018/108955 2017-12-18 2018-09-30 货物保护方法、装置、系统和计算机可读存储介质 WO2019119926A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US16/954,650 US11694294B2 (en) 2017-12-18 2018-09-30 Cargo protection method, device and system, and non-transitory computer-readable storage medium
JP2020530580A JP7179065B2 (ja) 2017-12-18 2018-09-30 貨物保護の方法、デバイスおよびシステム、ならびに非一時的コンピュータ可読記憶媒体

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711363753.9A CN109934521B (zh) 2017-12-18 2017-12-18 货物保护方法、装置、系统和计算机可读存储介质
CN201711363753.9 2017-12-18

Publications (1)

Publication Number Publication Date
WO2019119926A1 true WO2019119926A1 (zh) 2019-06-27

Family

ID=66982379

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/108955 WO2019119926A1 (zh) 2017-12-18 2018-09-30 货物保护方法、装置、系统和计算机可读存储介质

Country Status (4)

Country Link
US (1) US11694294B2 (zh)
JP (1) JP7179065B2 (zh)
CN (1) CN109934521B (zh)
WO (1) WO2019119926A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110948525A (zh) * 2019-12-25 2020-04-03 西南石油大学 一种爬壁机器人安全防护联合保护装置

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109934521B (zh) * 2017-12-18 2021-07-13 北京京东尚科信息技术有限公司 货物保护方法、装置、系统和计算机可读存储介质
KR102524296B1 (ko) * 2018-12-07 2023-04-24 현대자동차주식회사 적재물 추락 발생 감지에 따른 차량 제어 방법 및 시스템
CN111038709B (zh) * 2019-10-11 2021-03-23 湖北航天化学技术研究所 一种直升机安全气囊触发方法及其装置
CN111027105B (zh) * 2019-12-20 2022-05-03 上海海鸥数码照相机有限公司 一种坠落自检的数据防泄漏方法、装置及无人机
CN111232233B (zh) * 2020-03-04 2021-09-17 北京三快在线科技有限公司 载货无人机及其检测货物掉落的方法和装置
CN111977003A (zh) * 2020-07-17 2020-11-24 苏州大晋影视传媒有限公司 一种基于无人机气囊组的智能坠落防护方法及其系统
CN112486217B (zh) * 2020-11-23 2023-07-21 北京三快在线科技有限公司 坠落保护方法、装置、存储介质及电子设备
CN113988119B (zh) * 2021-10-14 2023-04-07 电子科技大学 一种多旋翼无人机识别中的块概率分布特征提取方法
CN116088559B (zh) * 2021-11-05 2024-03-26 北京三快在线科技有限公司 一种无人机控制系统、方法及无人机

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105857590A (zh) * 2016-04-28 2016-08-17 东北大学 一种四旋翼飞行器的飞行控制方法
CN206096454U (zh) * 2016-09-29 2017-04-12 上海华测导航技术股份有限公司 一种用于无人机的防摔保护与防丢失装置
CN106927057A (zh) * 2017-03-25 2017-07-07 林娟娟 用于无人机防摔的气囊装置
CN107087429A (zh) * 2016-03-25 2017-08-22 深圳前海达闼云端智能科技有限公司 飞行器的控制方法和装置

Family Cites Families (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102005061928B4 (de) 2005-12-23 2010-04-22 Airbus Deutschland Gmbh Sicherheitssystem zum Verringern einer Aufprallenergie eines Nutzlastbehälters
SG193450A1 (en) 2011-03-14 2013-10-30 Albert Galick Method for uncovering hidden markov models
CN103034170B (zh) * 2012-11-27 2014-10-29 华中科技大学 一种基于区间的数控机床性能预测方法
EP3459859B1 (en) * 2014-02-27 2020-09-02 SZ DJI Technology Co., Ltd. Impact protection apparatus
CN103895870B (zh) * 2014-03-18 2016-08-24 华南农业大学 一种用于多旋翼无人机的紧急伞降装置及方法
US10404299B1 (en) * 2016-03-07 2019-09-03 Hrl Laboratories, Llc System for parallelized cognitive signal denoising
US10137986B1 (en) 2015-03-25 2018-11-27 Amazon Technologies, Inc. Airlift package protection airbag container
CN105046835A (zh) * 2015-08-26 2015-11-11 广州极飞电子科技有限公司 物品接收方法和装置、物品投递方法及系统
JP2017056822A (ja) 2015-09-16 2017-03-23 株式会社マルチコプターラボ 無線操縦マルチコプター
US10336453B2 (en) * 2016-01-14 2019-07-02 Elwha Llc System and method for payload management for unmanned aircraft
US20180022310A1 (en) * 2016-03-24 2018-01-25 Rhombus Systems Group, Inc. Airbag system for use with unmanned aerial vehicles
US10252795B2 (en) * 2016-04-08 2019-04-09 Ecole Polytechnique Federale De Lausanne (Epfl) Foldable aircraft with protective cage for transportation and transportability
KR20170129528A (ko) 2016-05-17 2017-11-27 엘지이노텍 주식회사 에어백 유닛 및 이를 포함하는 무인 비행체
CN106228319A (zh) * 2016-06-18 2016-12-14 张阳康 一种基于无人机的自动给递送传系统和方法
CN106716285A (zh) * 2016-06-30 2017-05-24 深圳市大疆创新科技有限公司 农业无人机作业方法、系统及农业无人机
JP7034091B2 (ja) 2016-12-20 2022-03-11 日本化薬株式会社 飛行体用のエアバッグ装置
US11027838B2 (en) * 2017-05-06 2021-06-08 Karman, Inc. In flight charging system
US10671960B2 (en) * 2017-10-13 2020-06-02 Dash Systems, Inc. System and method for performing precision guided air to ground package delivery
CN109934521B (zh) * 2017-12-18 2021-07-13 北京京东尚科信息技术有限公司 货物保护方法、装置、系统和计算机可读存储介质
CN112154102A (zh) * 2018-03-21 2020-12-29 马西莫·马萨 具有数据恢复的可展开飞行数据记录器和方法
JP2022530463A (ja) * 2019-04-25 2022-06-29 ジョビー エアロ インク 垂直離着陸航空機
US11506512B2 (en) * 2019-05-22 2022-11-22 TDK Japan Method and system using tightly coupled radar positioning to improve map performance
US20210339842A1 (en) * 2020-04-29 2021-11-04 United Therapeutics Corporation Structural features of a vehicle having utility to transport high value things
US11747480B2 (en) * 2020-08-18 2023-09-05 IntelliShot Holdings, Inc. Automated threat detection and deterrence apparatus

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107087429A (zh) * 2016-03-25 2017-08-22 深圳前海达闼云端智能科技有限公司 飞行器的控制方法和装置
CN105857590A (zh) * 2016-04-28 2016-08-17 东北大学 一种四旋翼飞行器的飞行控制方法
CN206096454U (zh) * 2016-09-29 2017-04-12 上海华测导航技术股份有限公司 一种用于无人机的防摔保护与防丢失装置
CN106927057A (zh) * 2017-03-25 2017-07-07 林娟娟 用于无人机防摔的气囊装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110948525A (zh) * 2019-12-25 2020-04-03 西南石油大学 一种爬壁机器人安全防护联合保护装置
CN110948525B (zh) * 2019-12-25 2021-11-30 西南石油大学 一种爬壁机器人安全防护联合保护装置

Also Published As

Publication number Publication date
US20200385142A1 (en) 2020-12-10
US11694294B2 (en) 2023-07-04
CN109934521A (zh) 2019-06-25
JP7179065B2 (ja) 2022-11-28
CN109934521B (zh) 2021-07-13
JP2021506644A (ja) 2021-02-22

Similar Documents

Publication Publication Date Title
WO2019119926A1 (zh) 货物保护方法、装置、系统和计算机可读存储介质
CN106054208B (zh) 多线激光雷达车辆目标识别方法和汽车的防撞装置
CN109212531A (zh) 确定目标车辆取向的方法
CN104181534B (zh) 交叉路口碰撞警示系统的概率目标选择和威胁评估方法以及应用
WO2018071106A1 (en) Systems and methods for landing a drone on a moving base
WO2020132082A1 (en) Object classification using extra-regional context
CN106371105A (zh) 单线激光雷达车辆目标识别方法、装置和汽车
CN110134145A (zh) 无人机避障控制方法、飞行控制器及无人飞行器
CN107087429B (zh) 飞行器的控制方法和装置
US20120116662A1 (en) System and Method for Tracking Objects
CN106291506A (zh) 基于单线点云数据机器学习的车辆目标识别方法及装置
CN110286389A (zh) 一种用于障碍物识别的栅格管理方法
US10379542B2 (en) Location and mapping device and method
CN106339669A (zh) 多线点云数据机器学习的人形目标识别方法和防撞装置
US11726176B2 (en) Annotation of radar-profiles of objects
CN106353769A (zh) 多线激光雷达人形目标识别方法和汽车的防撞装置
EP3842316B1 (en) Method and device for controlling operation of self-driving car
CN110543187A (zh) 基于激光雷达的定位和避障无人机装置及方法
CN113703470A (zh) 基于架空输电线电磁场探测的飞行防撞方法及装置
CN108986555B (zh) 一种飞行防碰撞感知方法、系统、存储介质及设备
US11948367B2 (en) Multi-object tracking for autonomous vehicles
JP2020034451A (ja) データ構造、記憶媒体及び記憶装置
JP2020034772A (ja) 静止物データ生成装置、制御方法、プログラム及び記憶媒体
CN109353302A (zh) 一种车辆防碰撞系统、方法、装置、设备、介质及车辆
JP2020034452A (ja) 位置推定装置、制御方法、プログラム及び記憶媒体

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18890895

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2020530580

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 18.09.2020)

122 Ep: pct application non-entry in european phase

Ref document number: 18890895

Country of ref document: EP

Kind code of ref document: A1