U.S. patent application number 17/468864 was filed with the patent office on 2022-05-26 for obstacle avoiding method and apparatus for unmanned aerial vehicle based on multi-signal acquisition and route planning model.
The applicant listed for this patent is Guangdong Polytechnic Normal University. Invention is credited to Xuan Chu, Yiqing Fu, Qiwei Guo, Chaojun Hou, Huasheng Huang, Jiahao Li, Shaoming Luo, Aimin Miao, Yu Tang, Jiepeng Yang, Jiajun Zhuang, Xincai Zhuang.
Application Number | 20220163979 17/468864 |
Document ID | / |
Family ID | 1000005882563 |
Filed Date | 2022-05-26 |
United States Patent
Application |
20220163979 |
Kind Code |
A1 |
Tang; Yu ; et al. |
May 26, 2022 |
OBSTACLE AVOIDING METHOD AND APPARATUS FOR UNMANNED AERIAL VEHICLE
BASED ON MULTI-SIGNAL ACQUISITION AND ROUTE PLANNING MODEL
Abstract
Disclosed is an obstacle avoiding method and apparatus for an
unmanned aerial vehicle based on a multi-signal acquisition and
route planning model. The method comprises: conducting signal
acquisition processing on a first environmental area to obtain an
initial millimeter-wave radar signal, an initial laser radar
signal, an initial image signal and an initial ultrasonic signal;
generating an initial three-dimensional environmental model
according to a preset dynamic environment real-time modeling
method; acquiring a motion parameter and a body shape parameter of
the unmanned aerial vehicle and inputting the parameters into an
initial route planning model corresponding to the initial
three-dimensional environmental model based on a genetic algorithm
to process to obtain an output of the initial route planning model;
judging whether the output is capable of avoiding an obstacle; if
yes, generating an obstacle avoiding flight instruction to require
the unmanned aerial vehicle to fly through the first environmental
area.
Inventors: |
Tang; Yu; (Guangzhou City,
CN) ; Luo; Shaoming; (Guangzhou City, CN) ;
Guo; Qiwei; (Guangzhou City, CN) ; Zhuang;
Xincai; (Guangzhou City, CN) ; Li; Jiahao;
(Guangzhou City, CN) ; Yang; Jiepeng; (Guangzhou
City, CN) ; Fu; Yiqing; (Guangzhou City, CN) ;
Hou; Chaojun; (Guangzhou City, CN) ; Zhuang;
Jiajun; (Guangzhou City, CN) ; Miao; Aimin;
(Guangzhou City, CN) ; Chu; Xuan; (Guangzhou City,
CN) ; Huang; Huasheng; (Guangzhou City, CN) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Guangdong Polytechnic Normal University |
Guangzhou City |
|
CN |
|
|
Family ID: |
1000005882563 |
Appl. No.: |
17/468864 |
Filed: |
September 8, 2021 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G01S 13/862 20130101;
B64C 39/024 20130101; G01S 13/865 20130101; G05D 1/106 20190501;
G01S 13/867 20130101; B64C 2201/141 20130101; G01S 13/933
20200101 |
International
Class: |
G05D 1/10 20060101
G05D001/10; G01S 13/86 20060101 G01S013/86; G01S 13/933 20060101
G01S013/933; B64C 39/02 20060101 B64C039/02 |
Foreign Application Data
Date |
Code |
Application Number |
Nov 24, 2020 |
CN |
202011326513.3 |
Claims
1. An obstacle avoiding method for an unmanned aerial vehicle based
on a multi-signal acquisition and route planning model, comprising:
S1, conducting signal acquisition processing on a first
environmental area to obtain an initial millimeter-wave radar
signal, an initial laser radar signal, an initial image signal and
an initial ultrasonic signal by utilizing millimeter-wave radar,
laser radar, a binocular vision camera and an ultrasonic
transceiver pre-arranged on the unmanned aerial vehicle; S2,
generating an initial three-dimensional environmental model
according to the initial millimeter-wave radar signal, the initial
laser radar signal, the initial image signal and the initial
ultrasonic signal and according to a preset dynamic environment
real-time modeling method; S3, acquiring a motion parameter and a
body shape parameter of the unmanned aerial vehicle and inputting
the parameters into an initial route planning model corresponding
to the initial three-dimensional environmental model based on a
genetic algorithm to process so as to obtain an output of the
initial route planning model, wherein the motion parameter at least
comprises a position parameter of the unmanned aerial vehicle, the
output of the initial route planning model comprises incapability
of avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the initial route planning model is incapability
of avoiding the obstacle, the output of the initial route planning
model is further provided with an initial obstacle avoiding route;
S4, judging whether the output is capable of avoiding an obstacle
or not; and S5, if yes, generating an obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the first environmental area in an obstacle avoiding manner along
the initial obstacle avoiding route; the step S1 of conducting
signal acquisition processing on a first environmental area to
obtain an initial millimeter-wave radar signal, an initial laser
radar signal, an initial image signal and an initial ultrasonic
signal by utilizing millimeter-wave radar, laser radar, a binocular
vision camera and an ultrasonic transceiver pre-arranged on the
unmanned aerial vehicle comprises: S101, dividing the first
environmental area into a first sub area, a second sub area, a
third sub area and a fourth sub area, and respectively conducting
signal acquisition processing on the first sub area, the second sub
area, the third sub area and the fourth sub area by utilizing the
millimeter-wave radar, laser radar, the binocular vision camera and
the ultrasonic transceiver pre-arranged on the unmanned aerial
vehicle to obtain the initial millimeter-wave radar signal, the
initial laser radar signal, the initial image signal and the
initial ultrasonic signal, wherein the first sub area and the
fourth sub area are of axial symmetry about an axle wire of the
unmanned aerial vehicle, and the second sub area and the third sub
area are of axial symmetry about an axle wire of the unmanned
aerial vehicle; and a first distance between a horizontal position
of any one point in the first environmental area and a horizontal
position of a head of the unmanned aerial vehicle is smaller than a
second distance between a horizontal position of the point in the
first environmental area and a horizontal position of a tail of the
unmanned aerial vehicle; and after the step S4 of judging whether
the output is capable of avoiding an obstacle or not, the method
comprising: S41, if the output of the initial route planning model
is incapable of avoiding the obstacle, generating a hovering and
overturning instruction of the unmanned aerial vehicle to require
the unmanned aerial vehicle to hover and overturn at 180 degrees
under a condition that a direction of the head is kept unchanged;
S42, respectively conducting signal acquisition processing on the
fourth sub area, the third sub area, the second sub area and the
first sub area by utilizing the millimeter-wave radar, laser radar,
the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain a secondary
millimeter-wave radar signal, a secondary laser radar signal, a
secondary image signal and a secondary ultrasonic signal; S43,
generating a secondary three-dimensional environmental model
according to the secondary millimeter-wave radar signal, the
secondary laser radar signal, the secondary image signal and the
secondary ultrasonic signal and according to the preset dynamic
environment real-time modeling method; S44, inputting the motion
parameter and the body shape parameter of the unmanned aerial
vehicle into a secondary route planning model corresponding to the
secondary three-dimensional environmental model based on a genetic
algorithm to process so as to obtain an output of the secondary
route planning model, wherein the motion parameter at least
comprises a position parameter of the unmanned aerial vehicle, the
output of the secondary route planning model comprises incapability
of avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the secondary route planning model is
incapability of avoiding the obstacle, the output of the secondary
route planning model is further provided with a secondary obstacle
avoiding route; S45, judging whether the output of the initial
route planning model is capable of avoiding an obstacle or not;
S46, if yes, generating a secondary obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the first environmental area in an obstacle avoiding manner along
the secondary obstacle avoiding route.
2. The obstacle avoiding method for an unmanned aerial vehicle
based on a multi-signal acquisition and route planning model
according to claim 1, wherein the step S2 of generating an initial
three-dimensional environmental model according to the initial
millimeter-wave radar signal, the initial laser radar signal, the
initial image signal and the initial ultrasonic signal and
according to a preset dynamic environment real-time modeling method
comprises: S201, conducting dynamic environment real-time modeling
processing by adopting a preset first spatial modeling tool
according to the initial millimeter-wave radar signal and the
initial image signal so as to obtain a first spatial model; S202,
meanwhile, conducting dynamic environment real-time modeling
processing by adopting a preset second spatial modeling tool
according to the initial millimeter-wave radar signal and the
initial ultrasonic signal so as to obtain a second spatial model;
S203, combining the first spatial model with the second spatial
model in parallel as an initial three-dimensional environment
model; the output of the initial route planning model comprises an
output corresponding to the first spatial model and an output
corresponding to the second spatial model; the output of the first
spatial model comprises incapability of avoiding the obstacle or
capability of avoiding the obstacle, and when the output of the
first spatial model is capability of avoiding the obstacle, the
output of the first spatial model is further provided with a first
initial obstacle avoiding route; the output of the second spatial
model comprises incapability of avoiding the obstacle or capability
of avoiding the obstacle, and when the output of the second spatial
model is capability of avoiding the obstacle, the output of the
second spatial model is further provided with a second initial
obstacle avoiding route; the step S4 of judging whether the output
of the initial route planning model is capable of avoiding the
obstacle or not comprises: S401, judging whether both the output
corresponding to the first spatial model and the output
corresponding to the second spatial model are capable of avoiding
the obstacle or not; S402, if both the output corresponding to the
first spatial model and the output corresponding to the second
spatial model are capable of avoiding the obstacle, calculating a
similarity value between the first initial obstacle avoiding route
and the second initial obstacle avoiding route according to a
preset route similarity calculating method; S403, judging whether
the similarity value between the first initial obstacle avoiding
route and the second initial obstacle avoiding route is greater
than a preset similarity threshold value or not; and S404, if the
similarity value between the first initial obstacle avoiding route
and the second initial obstacle avoiding route is greater than the
preset similarity threshold value, judging that the output of the
initial route planning model is capable of avoiding the
obstacle.
3. The obstacle avoiding method for an unmanned aerial vehicle
based on a multi-signal acquisition and route planning model
according to claim 1, before the step S3 of acquiring a motion
parameter and a body shape parameter of the unmanned aerial vehicle
and inputting the parameters into an initial route planning model
corresponding to the initial three-dimensional environmental model
based on a genetic algorithm to process so as to obtain an output
of the initial route planning model, wherein the motion parameter
at least comprises a position parameter of the unmanned aerial
vehicle, the output of the initial route planning model comprises
incapability of avoiding an obstacle or capability of avoiding the
obstacle, and when the output of the initial route planning model
is incapability of avoiding the obstacle, the output of the initial
route planning model is further provided with an initial obstacle
avoiding route, the method further comprising: S21, retrieving an
appointed standard environment model the most similar to the
initial three-dimensional environment model from a preset
environment model base, wherein the environment model base
pre-stores a plurality of standard environment models obtained by
modeling a plurality of air obstacle environments; S22, acquiring
an appointed route planning model corresponding to the appointed
standard environment model according to a mapping relationship
between the preset standard environment model and the route
planning model, wherein the different standard environment models
correspond to different route planning models based on the genetic
algorithm; and S23, marking the appointed route planning model as
the initial route planning model and generating a route planning
instruction, wherein the route planning instruction is used for
indicating the motion parameter and the body shape parameter of the
unmanned aerial vehicle and inputting the parameters into the
initial route planning model corresponding to the initial
three-dimensional environmental model based on the genetic
algorithm to process.
4. The obstacle avoiding method for an unmanned aerial vehicle
based on a multi-signal acquisition and route planning model
according to claim 1, wherein after the step S4 of judging whether
the output is capable of avoiding an obstacle or not, the method
comprises: S411, if the output of the initial route planning model
is incapable of avoiding the obstacle, controlling the unmanned
aerial vehicle to retreat at a preset first horizontal length under
a premise of keeping a current height; S412, conducting signal
acquisition processing on the second environmental area to obtain a
third millimeter-wave radar signal, a third laser radar signal, a
third image signal and a third ultrasonic signal by utilizing
millimeter-wave radar, laser radar, a binocular vision camera and
an ultrasonic transceiver pre-arranged on the unmanned aerial
vehicle, wherein the second environmental area is an area obtained
as the first environment area translates horizontally at the first
horizontal length along a retreating direction of the unmanned
aerial vehicle; S413, generating a third three-dimensional
environmental model according to the third millimeter-wave radar
signal, the third laser radar signal, the third image signal and
the third ultrasonic signal and according to the preset dynamic
environment real-time modeling method; S414, inputting the motion
parameter and the body shape parameter of the unmanned aerial
vehicle into a third route planning model corresponding to the
third three-dimensional environmental model based on a genetic
algorithm to process so as to obtain an output of the third route
planning model, wherein the output of the third route planning
model comprises incapability of avoiding an obstacle or capability
of avoiding the obstacle, and when the output of the third route
planning model is incapability of avoiding the obstacle, the output
of the third route planning model is further provided with a third
obstacle avoiding route; S415, judging whether the output of the
third route planning model is capable of avoiding an obstacle or
not; and S416, if yes, generating a third obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the second environmental area in an obstacle avoiding manner along
the third obstacle avoiding route.
5. An obstacle avoiding apparatus for an unmanned aerial vehicle
based on a multi-signal acquisition and route planning model,
comprising: an information acquisition unit for conducting signal
acquisition processing on a first environmental area to obtain an
initial millimeter-wave radar signal, an initial laser radar
signal, an initial image signal and an initial ultrasonic signal by
utilizing millimeter-wave radar, laser radar, a binocular vision
camera and an ultrasonic transceiver pre-arranged on the unmanned
aerial vehicle; an initial three-dimensional environmental model
generation unit for generating an initial three-dimensional
environmental model according to the initial millimeter-wave radar
signal, the initial laser radar signal, the initial image signal
and the initial ultrasonic signal and according to a preset dynamic
environment real-time modeling method; an initial route planning
model processing unit for acquiring a motion parameter and a body
shape parameter of the unmanned aerial vehicle and inputting the
parameters into an initial route planning model corresponding to
the initial three-dimensional environmental model based on a
genetic algorithm to process so as to obtain an output of the
initial route planning model, wherein the motion parameter at least
comprises a position parameter of the unmanned aerial vehicle, the
output of the initial route planning model comprises incapability
of avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the initial route planning model is incapability
of avoiding the obstacle, the output of the initial route planning
model is further provided with an initial obstacle avoiding route;
an obstacle avoiding judging unit for judging whether the output of
the initial route planning model is capable of avoiding an obstacle
or not; and an obstacle avoiding flight instruction generation unit
for generating an obstacle avoiding flight instruction to require
the unmanned aerial vehicle to fly through the first environmental
area in an obstacle avoiding manner along the initial obstacle
avoiding route if the output of the initial route planning model is
capable of avoiding the obstacle; conducting signal acquisition
processing on a first environmental area to obtain an initial
millimeter-wave radar signal, an initial laser radar signal, an
initial image signal and an initial ultrasonic signal by utilizing
millimeter-wave radar, laser radar, a binocular vision camera and
an ultrasonic transceiver pre-arranged on the unmanned aerial
vehicle comprises: dividing the first environmental area into a
first sub area, a second sub area, a third sub area and a fourth
sub area, and respectively conducting signal acquisition processing
on the first sub area, the second sub area, the third sub area and
the fourth sub area by utilizing the millimeter-wave radar, laser
radar, the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain the initial
millimeter-wave radar signal, the initial laser radar signal, the
initial image signal and the initial ultrasonic signal, wherein the
first sub area and the fourth sub area are of axial symmetry about
an axle wire of the unmanned aerial vehicle, and the second sub
area and the third sub area are of axial symmetry about an axle
wire of the unmanned aerial vehicle; and a first distance between a
horizontal position of any one point in the first environmental
area and a horizontal position of a head of the unmanned aerial
vehicle is smaller than a second distance between a horizontal
position of the point in the first environmental area and a
horizontal position of a tail of the unmanned aerial vehicle; and
after judging whether the output is capable of avoiding an obstacle
or not, the method comprising: if the output of the initial route
planning model is incapable of avoiding the obstacle, generating a
hovering and overturning instruction of the unmanned aerial vehicle
to require the unmanned aerial vehicle to hover and overturn at 180
degrees under a condition that a direction of the head is kept
unchanged; respectively conducting signal acquisition processing on
the fourth sub area, the third sub area, the second sub area and
the first sub area by utilizing the millimeter-wave radar, laser
radar, the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain a secondary
millimeter-wave radar signal, a secondary laser radar signal, a
secondary image signal and a secondary ultrasonic signal;
generating a secondary three-dimensional environmental model
according to the secondary millimeter-wave radar signal, the
secondary laser radar signal, the secondary image signal and the
secondary ultrasonic signal and according to the preset dynamic
environment real-time modeling method; inputting the motion
parameter and the body shape parameter of the unmanned aerial
vehicle into a secondary route planning model corresponding to the
secondary three-dimensional environmental model based on a genetic
algorithm to process so as to obtain an output of the secondary
route planning model, wherein the motion parameter at least
comprises a position parameter of the unmanned aerial vehicle, the
output of the secondary route planning model comprises incapability
of avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the secondary route planning model is
incapability of avoiding the obstacle, the output of the secondary
route planning model is further provided with a secondary obstacle
avoiding route; judging whether the output of the initial route
planning model is capable of avoiding an obstacle or not; and if
yes, generating a secondary obstacle avoiding flight instruction to
require the unmanned aerial vehicle to fly through the first
environmental area in an obstacle avoiding manner along the
secondary obstacle avoiding route.
6-7. (canceled)
Description
TECHNICAL FIELD
[0001] The application relates to the field of computers, in
particular to an obstacle avoiding method and apparatus for an
unmanned aerial vehicle based on a multi-signal acquisition and
route planning model, a computer device and a storage medium.
BACKGROUND
[0002] Featuring in lightness and flexibility, the unmanned aerial
vehicles have been applied to various industries. In the flying
process of the unmanned aerial vehicle, encountering an obstacle,
it is needed to avoid the obstacle. An existing unmanned aerial
vehicle obstacle avoiding scheme is to determine the position of
the obstacle by means of a single signal acquisition means, and
then the unmanned aerial vehicle flies in an obstacle avoiding
manner. Thus, a conventional unmanned aerial vehicle obstacle
avoiding scheme is narrow in application range. In a complex
environment, for example, in a complex agricultural environment (if
exist), it is hard to recognize the obstacle accurately and
efficiently and fly in an obstacle avoiding manner accurately, such
that the unmanned aerial vehicle has a greater potential safety
hazard.
SUMMARY
[0003] The application provides an obstacle avoiding method for an
unmanned aerial vehicle based on a multi-signal acquisition and
route planning model, including the following steps:
[0004] S1, conducting signal acquisition processing on a first
environmental area to obtain an initial millimeter-wave radar
signal, an initial laser radar signal, an initial image signal and
an initial ultrasonic signal by utilizing millimeter-wave radar,
laser radar, a binocular vision camera and an ultrasonic
transceiver pre-arranged on the unmanned aerial vehicle;
[0005] S2, generating an initial three-dimensional environmental
model according to the initial millimeter-wave radar signal, the
initial laser radar signal, the initial image signal and the
initial ultrasonic signal and according to a preset dynamic
environment real-time modeling method;
[0006] S3, acquiring a motion parameter and a body shape parameter
of the unmanned aerial vehicle and inputting the parameters into an
initial route planning model corresponding to the initial
three-dimensional environmental model based on a genetic algorithm
to process so as to obtain an output of the initial route planning
model, wherein the motion parameter at least comprises a position
parameter of the unmanned aerial vehicle, the output of the initial
route planning model comprises incapability of avoiding an obstacle
or capability of avoiding the obstacle, and when the output of the
initial route planning model is incapability of avoiding the
obstacle, the output of the initial route planning model is further
provided with an initial obstacle avoiding route;
[0007] S4, judging whether the output is capable of avoiding an
obstacle or not;
[0008] S5, if yes, generating an obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the first environmental area in an obstacle avoiding manner along
the initial obstacle avoiding route.
[0009] Further, the step S1 of conducting signal acquisition
processing on a first environmental area to obtain an initial
millimeter-wave radar signal, an initial laser radar signal, an
initial image signal and an initial ultrasonic signal by utilizing
millimeter-wave radar, laser radar, a binocular vision camera and
an ultrasonic transceiver pre-arranged on the unmanned aerial
vehicle includes:
[0010] S101, dividing the first environmental area into a first sub
area, a second sub area, a third sub area and a fourth sub area,
and respectively conducting signal acquisition processing on the
first sub area, the second sub area, the third sub area and the
fourth sub area by utilizing the millimeter-wave radar, laser
radar, the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain the initial
millimeter-wave radar signal, the initial laser radar signal, the
initial image signal and the initial ultrasonic signal, wherein the
first sub area and the fourth sub area are of axial symmetry about
an axle wire of the unmanned aerial vehicle, and the second sub
area and the third sub area are of axial symmetry about an axle
wire of the unmanned aerial vehicle; and a first distance between a
horizontal position of any one point in the first environmental
area and a horizontal position of a head of the unmanned aerial
vehicle is smaller than a second distance between a horizontal
position of the point in the first environmental area and a
horizontal position of a tail of the unmanned aerial vehicle;
[0011] after the step S4 of judging whether the output is capable
of avoiding an obstacle or not, the method comprising:
[0012] S41, if the output of the initial route planning model is
incapable of avoiding the obstacle, generating a hovering and
overturning instruction of the unmanned aerial vehicle to require
the unmanned aerial vehicle to hover and overturn at 180 degrees
under a condition that a direction of the head is kept
unchanged;
[0013] S42, respectively conducting signal acquisition processing
on the fourth sub area, the third sub area, the second sub area and
the first sub area by utilizing the millimeter-wave radar, laser
radar, the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain a secondary
millimeter-wave radar signal, a secondary laser radar signal, a
secondary image signal and a secondary ultrasonic signal;
[0014] S43, generating a secondary three-dimensional environmental
model according to the secondary millimeter-wave radar signal, the
secondary laser radar signal, the secondary image signal and the
secondary ultrasonic signal and according to the preset dynamic
environment real-time modeling method;
[0015] S44, inputting the motion parameter and the body shape
parameter of the unmanned aerial vehicle into a secondary route
planning model corresponding to the secondary three-dimensional
environmental model based on a genetic algorithm to process so as
to obtain an output of the secondary route planning model, wherein
the motion parameter at least comprises a position parameter of the
unmanned aerial vehicle, the output of the secondary route planning
model comprises incapability of avoiding an obstacle or capability
of avoiding the obstacle, and when the output of the secondary
route planning model is incapability of avoiding the obstacle, the
output of the secondary route planning model is further provided
with a secondary obstacle avoiding route;
[0016] S45, judging whether the output is capable of avoiding an
obstacle or not;
[0017] S46, if yes, generating a secondary obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the first environmental area in an obstacle avoiding manner along
the secondary obstacle avoiding route.
[0018] Further, the step S2 of generating an initial
three-dimensional environmental model according to the initial
millimeter-wave radar signal, the initial laser radar signal, the
initial image signal and the initial ultrasonic signal and
according to a preset dynamic environment real-time modeling method
includes:
[0019] S201, conducting dynamic environment real-time modeling
processing by adopting a preset first spatial modeling tool
according to the initial millimeter-wave radar signal and the
initial image signal so as to obtain a first spatial model;
[0020] S202, meanwhile, conducting dynamic environment real-time
modeling processing by adopting a preset second spatial modeling
tool according to the initial millimeter-wave radar signal and the
initial ultrasonic signal so as to obtain a second spatial
model;
[0021] S203, combining the first spatial model with the second
spatial model in parallel as an initial three-dimensional
environment model;
[0022] the output of the initial route planning model comprises an
output corresponding to the first spatial model and an output
corresponding to the second spatial model; the output of the first
spatial model comprises incapability of avoiding the obstacle or
capability of avoiding the obstacle, and when the output of the
first spatial model is capability of avoiding the obstacle, the
output of the first spatial model is further provided with a first
initial obstacle avoiding route; the output of the second spatial
model comprises incapability of avoiding the obstacle or capability
of avoiding the obstacle, and when the output of the second spatial
model is capability of avoiding the obstacle, the output of the
second spatial model is further provided with a second initial
obstacle avoiding route; the step S4 of judging whether the output
of the initial route planning model is capable of avoiding the
obstacle or not comprises:
[0023] S401, judging whether both the output corresponding to the
first spatial model and the output corresponding to the second
spatial model are capable of avoiding the obstacle or not;
[0024] S402, if both the output corresponding to the first spatial
model and the output corresponding to the second spatial model are
capable of avoiding the obstacle, calculating a similarity value
between the first initial obstacle avoiding route and the second
initial obstacle avoiding route according to a preset route
similarity calculating method;
[0025] S403, judging whether the similarity value between the first
initial obstacle avoiding route and the second initial obstacle
avoiding route is greater than a preset similarity threshold value
or not; and
[0026] S404, if the similarity value between the first initial
obstacle avoiding route and the second initial obstacle avoiding
route is greater than the preset similarity threshold value,
judging that the output of the initial route planning model is
capable of avoiding the obstacle.
[0027] Further, before the step of acquiring a motion parameter and
a body shape parameter of the unmanned aerial vehicle and inputting
the parameters into an initial route planning model corresponding
to the initial three-dimensional environmental model based on a
genetic algorithm to process so as to obtain an output of the
initial route planning model, wherein the motion parameter at least
includes a position parameter of the unmanned aerial vehicle, the
output of the initial route planning model comprises incapability
of avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the initial route planning model is incapability
of avoiding the obstacle, the output of the initial route planning
model is further provided with an initial obstacle avoiding route,
the method includes:
[0028] S21, retrieving an appointed standard environment model the
most similar to the initial three-dimensional environment model
from a preset environment model base, wherein the environment model
base pre-stores a plurality of standard environment models obtained
by modeling a plurality of air obstacle environments;
[0029] S22, acquiring an appointed route planning model
corresponding to the appointed standard environment model according
to a mapping relationship between the preset standard environment
model and the route planning model, wherein the different standard
environment models correspond to different route planning models
based on the genetic algorithm;
[0030] S23, marking the appointed route planning model as the
initial route planning model and generating a route planning
instruction, wherein the route planning instruction is used for
indicating the motion parameter and the body shape parameter of the
unmanned aerial vehicle and inputting the parameters into the
initial route planning model corresponding to the initial
three-dimensional environmental model based on the genetic
algorithm to process.
[0031] Further, after the step S4 of judging whether the output is
capable of avoiding an obstacle or not, the method including:
[0032] S411, if the output of the initial route planning model is
incapable of avoiding the obstacle, controlling the unmanned aerial
vehicle to retreat at a preset first horizontal length under a
premise of keeping a current height;
[0033] S412, conducting signal acquisition processing on the second
environmental area to obtain a third millimeter-wave radar signal,
a third laser radar signal, a third image signal and a third
ultrasonic signal by utilizing millimeter-wave radar, laser radar,
a binocular vision camera and an ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle, wherein the second
environmental area is an area obtained as the first environment
area translates horizontally at the first horizontal length along a
retreating direction of the unmanned aerial vehicle;
[0034] S413, generating a third three-dimensional environmental
model according to the third millimeter-wave radar signal, the
third laser radar signal, the third image signal and the third
ultrasonic signal and according to the preset dynamic environment
real-time modeling method;
[0035] S414, inputting the motion parameter and the body shape
parameter of the unmanned aerial vehicle into a third route
planning model corresponding to the third three-dimensional
environmental model based on a genetic algorithm to process so as
to obtain an output of the third route planning model, wherein the
output of the third route planning model comprises incapability of
avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the third route planning model is incapability
of avoiding the obstacle, the output of the third route planning
model is further provided with a third obstacle avoiding route;
[0036] S415, judging whether the output of the third route planning
model is capable of avoiding an obstacle or not;
[0037] S416, if yes, generating a third obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the second environmental area in an obstacle avoiding manner along
the third obstacle avoiding route.
[0038] The application provides an obstacle avoiding apparatus for
an unmanned aerial vehicle based on a multi-signal acquisition and
route planning model, including:
[0039] an information acquisition unit for conducting signal
acquisition processing on a first environmental area to obtain an
initial millimeter-wave radar signal, an initial laser radar
signal, an initial image signal and an initial ultrasonic signal by
utilizing millimeter-wave radar, laser radar, a binocular vision
camera and an ultrasonic transceiver pre-arranged on the unmanned
aerial vehicle;
[0040] an initial three-dimensional environmental model generation
unit for generating an initial three-dimensional environmental
model according to the initial millimeter-wave radar signal, the
initial laser radar signal, the initial image signal and the
initial ultrasonic signal and according to a preset dynamic
environment real-time modeling method;
[0041] an initial route planning model processing unit for
acquiring a motion parameter and a body shape parameter of the
unmanned aerial vehicle and inputting the parameters into an
initial route planning model corresponding to the initial
three-dimensional environmental model based on a genetic algorithm
to process so as to obtain an output of the initial route planning
model, wherein the motion parameter at least comprises a position
parameter of the unmanned aerial vehicle, the output of the initial
route planning model comprises incapability of avoiding an obstacle
or capability of avoiding the obstacle, and when the output of the
initial route planning model is incapability of avoiding the
obstacle, the output of the initial route planning model is further
provided with an initial obstacle avoiding route;
[0042] an obstacle avoiding judging unit for judging whether the
output of the initial route planning model is capable of avoiding
an obstacle or not; and
[0043] an obstacle avoiding flight instruction generation unit for
generating an obstacle avoiding flight instruction to require the
unmanned aerial vehicle to fly through the first environmental area
in an obstacle avoiding manner along the initial obstacle avoiding
route if the output of the initial route planning model is capable
of avoiding the obstacle.
[0044] The application provides a computer device, including a
memory and a processor, the memory storing a computer program,
wherein the processor realizes the steps of any one method when
executing the computer program.
[0045] The application provides a computer readable storage medium,
having computer readable instructions stored therein, wherein the
steps of any one method is realized when the instructions are
executed by the processor.
[0046] According to the obstacle avoiding method and apparatus for
an unmanned aerial vehicle based on the multi-signal acquisition
and route planning model, the computer device and the storage
medium, the application provides an obstacle recognizing and
avoiding method by combining a dynamic environment real-time
modeling method and a genetic algorithm, which can be adapted to
any feasible environment, in particular to a complex agricultural
environment (the unmanned aerial vehicle needs to operate at a low
altitude to come across obstacles which are not encountered by the
common unmanned aerial vehicle, for example twigs), and the
unmanned aerial vehicle is particularly integrated with
millimeter-wave radar, laser radar, the binocular vision camera and
the ultrasonic transceiver for signal acquisition, such that an
integrated obstacle fast diagnosis model is established by making
full use of characteristics of high millimeter-wave radar moving
object capturing ability, high laser radar fine object sensitivity,
a wide binocular vision near distance view, good ultrasonic near
distance directivity and the like. A method for recognizing fixed
fine obstacles and protruding obstacles based on an actual scene in
a moving state of the unmanned aerial vehicle, an obstacle quick
detection device integrating millimeter-wave radar, laser radar,
binocular vision and ultrasonic waves is developed, and quick
diagnosing and obstacle avoiding problems of the fine objects and
protrusions are solved. The obstacle avoiding route planning is
conducted in real time by combining the moving characteristics of
the unmanned aerial vehicle, such that multidirectional effective
obstacle avoidance of the unmanned aerial vehicle within a 0-12 m/s
speed range is achieved.
BRIEF DESCRIPTION OF THE DRAWINGS
[0047] FIG. 1 is a flow diagram of an obstacle avoiding method for
an unmanned aerial vehicle based on a multi-signal acquisition and
route planning model of an embodiment of the application;
[0048] FIG. 2 is a structural schematic block diagram of the
computer device of an embodiment of the application.
[0049] Further description will be further made on implementation,
functional characteristics and advantages of the object of the
application with reference to drawings in combination of
embodiments.
DETAILED DESCRIPTION
[0050] In order to make purposes, technical schemes and advantages
of the disclosure clearer, the disclosure is further described in
detail below in combination with drawings and embodiments. It
should be understood that the specific examples described herein
are merely used for explaining the disclosure, instead of limiting
the disclosure.
[0051] Referring to FIG. 1, the embodiment of the application
provides an obstacle avoiding method for an unmanned aerial vehicle
based on a multi-signal acquisition and route planning model,
including the following steps:
[0052] S1, signal acquisition processing is conducted on a first
environmental area to obtain an initial millimeter-wave radar
signal, an initial laser radar signal, an initial image signal and
an initial ultrasonic signal by utilizing millimeter-wave radar,
laser radar, a binocular vision camera and an ultrasonic
transceiver pre-arranged on the unmanned aerial vehicle;
[0053] S2, an initial three-dimensional environmental model is
generated according to the initial millimeter-wave radar signal,
the initial laser radar signal, the initial image signal and the
initial ultrasonic signal and according to a preset dynamic
environment real-time modeling method;
[0054] S3, a motion parameter and a body shape parameter of the
unmanned aerial vehicle are acquired and the parameters are input
into an initial route planning model corresponding to the initial
three-dimensional environmental model based on a genetic algorithm
to process so as to obtain an output of the initial route planning
model, wherein the motion parameter at least includes a position
parameter of the unmanned aerial vehicle, the output of the initial
route planning model comprises incapability of avoiding an obstacle
or capability of avoiding the obstacle, and when the output of the
initial route planning model is incapability of avoiding the
obstacle, the output of the initial route planning model is further
provided with an initial obstacle avoiding route;
[0055] S4, whether the output is capable of avoiding an obstacle or
not is judged;
[0056] S5, if yes, an obstacle avoiding flight instruction is
generated to require the unmanned aerial vehicle to fly through the
first environmental area in an obstacle avoiding manner along the
initial obstacle avoiding route;
[0057] As described by the steps S1-S2, signal acquisition
processing is conducted on the first environmental area to obtain
the initial millimeter-wave radar signal, the initial laser radar
signal, the initial image signal and the initial ultrasonic signal
by utilizing millimeter-wave radar, laser radar, a binocular vision
camera and an ultrasonic transceiver pre-arranged on the unmanned
aerial vehicle; and an initial three-dimensional environmental
model is generated according to the initial millimeter-wave radar
signal, the initial laser radar signal, the initial image signal
and the initial ultrasonic signal and according to the preset
dynamic environment real-time modeling method. According to the
application, the millimeter-wave radar, the laser radar, the
binocular vision camera and the ultrasonic transceiver are
particularly selected to conduct signal acquisition processing
simultaneously, such that it is probable to acquire full and
accurate obstacle information in a complex air environment (for
example in an agricultural environment) within a short time by
means of characteristics of high millimeter-wave radar moving
object capturing ability, high laser radar fine object sensitivity,
a wide binocular vision near distance view, good ultrasonic near
distance directivity and the like. Further, the reason that the
millimeter-wave radar, the laser radar, the binocular vision camera
and the ultrasonic transceiver are particularly selected to conduct
signal acquisition processing simultaneously is not only that, and
is different as the specific execution process is different.
Therefore, detailed description will be made subsequently in
combination of specific steps. The dynamic environment real-time
modeling method may adopt any feasible method, for example, an
existing three-dimensional spatial modeling tool is adopted to
conduct modeling processing. As sufficient signal data in the
environment has been acquired, an initial three-dimensional
environment model with an obstacle (or the signal data shows that
there is no obstacle) can be simulated. However, the application is
characterized in real time modeling, which is due to signal
acquisition processing by four different signal collectors, such
that the environment signal is full enough. Thus, signal void can
be avoided, and therefore, it is probable to model in real
time.
[0058] Further, the step S1 of conducting signal acquisition
processing on a first environmental area to obtain an initial
millimeter-wave radar signal, an initial laser radar signal, an
initial image signal and an initial ultrasonic signal by utilizing
millimeter-wave radar, laser radar, a binocular vision camera and
an ultrasonic transceiver pre-arranged on the unmanned aerial
vehicle includes:
[0059] S101, dividing the first environmental area into a first sub
area, a second sub area, a third sub area and a fourth sub area,
and respectively conducting signal acquisition processing on the
first sub area, the second sub area, the third sub area and the
fourth sub area by utilizing the millimeter-wave radar, laser
radar, the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain the initial
millimeter-wave radar signal, the initial laser radar signal, the
initial image signal and the initial ultrasonic signal, wherein the
first sub area and the fourth sub area are of axial symmetry about
an axle wire of the unmanned aerial vehicle, and the second sub
area and the third sub area are of axial symmetry about an axle
wire of the unmanned aerial vehicle; and a first distance between a
horizontal position of any one point in the first environmental
area and a horizontal position of a head of the unmanned aerial
vehicle is smaller than a second distance between a horizontal
position of the point in the first environmental area and a
horizontal position of a tail of the unmanned aerial vehicle;
[0060] after the step S4 of judging whether the output is capable
of avoiding an obstacle or not, the method comprising:
[0061] S41, if the output of the initial route planning model is
incapable of avoiding the obstacle, generating a hovering and
overturning instruction of the unmanned aerial vehicle to require
the unmanned aerial vehicle to hover and overturn at 180 degrees
under a condition that a direction of the head is kept
unchanged;
[0062] S42, respectively conducting signal acquisition processing
on the fourth sub area, the third sub area, the second sub area and
the first sub area by utilizing the millimeter-wave radar, laser
radar, the binocular vision camera and the ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle to obtain a secondary
millimeter-wave radar signal, a secondary laser radar signal, a
secondary image signal and a secondary ultrasonic signal;
[0063] S43, generating a secondary three-dimensional environmental
model according to the secondary millimeter-wave radar signal, the
secondary laser radar signal, the secondary image signal and the
secondary ultrasonic signal and according to the preset dynamic
environment real-time modeling method;
[0064] S44, inputting the motion parameter and the body shape
parameter of the unmanned aerial vehicle into a secondary route
planning model corresponding to the secondary three-dimensional
environmental model based on a genetic algorithm to process so as
to obtain an output of the secondary route planning model, wherein
the motion parameter at least comprises a position parameter of the
unmanned aerial vehicle, the output of the secondary route planning
model comprises incapability of avoiding an obstacle or capability
of avoiding the obstacle, and when the output of the secondary
route planning model is incapability of avoiding the obstacle, the
output of the secondary route planning model is further provided
with a secondary obstacle avoiding route;
[0065] S45, judging whether the output is capable of avoiding an
obstacle or not;
[0066] S46, if yes, generating a secondary obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the first environmental area in an obstacle avoiding manner along
the secondary obstacle avoiding route.
[0067] Therefore, signals are utilized fully, such that the
obstacle avoiding flight of the application is more accurate and
efficient. As above-mentioned hereinbefore, the four signal
acquisition modes of the application are different in specific
execution processes, and reasons and advantages thereof are
different. Here, signal acquisition is conducted by dividing sub
areas. Specifically, if signal acquisition processing in all areas
of the first environment area is conducted respectively by adopting
the millimeter-wave radar, the laser radar, the binocular vision
camera and the ultrasonic transceiver. It is no doubt that the most
sufficient signal data can be acquired, and it takes a relatively
long time. The unmanned aerial vehicle of the application is in a
moving state, such that the shorter signal acquisition time is more
favorable. Hereby, the first environment area is divided into a
first sub area, a second sub area, a third sub area and a fourth
sub area and signal acquisition processing is conducted
respectively on the first sub area, the second sub area, the third
sub area and the fourth sub area by utilizing the millimeter-wave
radar, the laser radar, the binocular vision camera and the
ultrasonic transceiver pre-arranged on the unmanned aerial vehicle
to obtain ways of an initial millimeter-wave radar signal, an
initial laser radar signal, an initial image signal and an initial
ultrasonic signal, such that each signal collector only needs to
acquire signals in part of areas. Meanwhile, as the four signal
collectors are different in advantage, it is only needed to find a
smooth air road actually for obstacle avoidance of the unmanned
aerial vehicle, which is not high in requirement on the intact
spatial model. Therefore, by adopting the way of the application,
it is probable to find a proper obstacle avoiding route in a
certain sub area to avoid the obstacle accurately and it is shorter
in total time consumption. Further, the application adopting the
above-mentioned arrangement further has the advantage that under a
condition that the output of the initial route planning model is
incapable of avoiding the obstacle, it is only needed to hover the
unmanned aerial vehicle and overturn the unmanned aerial vehicle at
180 degrees by keeping the direction of the head unchanged, such
that the four different signal collectors can conduct signal
acquisition in the other sub different area to conduct modeling and
obstacle avoiding judgment again. At the time, the total time
needed to acquire the signals can be also saved by half.
[0068] Further, the step S2 of generating an initial
three-dimensional environmental model according to the initial
millimeter-wave radar signal, the initial laser radar signal, the
initial image signal and the initial ultrasonic signal and
according to a preset dynamic environment real-time modeling method
includes:
[0069] S201, conducting dynamic environment real-time modeling
processing by adopting a preset first spatial modeling tool
according to the initial millimeter-wave radar signal and the
initial image signal so as to obtain a first spatial model;
[0070] S202, meanwhile, conducting dynamic environment real-time
modeling processing by adopting a preset second spatial modeling
tool according to the initial millimeter-wave radar signal and the
initial ultrasonic signal so as to obtain a second spatial
model;
[0071] S203, combining the first spatial model with the second
spatial model in parallel as an initial three-dimensional
environment model;
[0072] the output of the initial route planning model comprises an
output corresponding to the first spatial model and an output
corresponding to the second spatial model; the output of the first
spatial model comprises incapability of avoiding the obstacle or
capability of avoiding the obstacle, and when the output of the
first spatial model is capability of avoiding the obstacle, the
output of the first spatial model is further provided with a first
initial obstacle avoiding route; the output of the second spatial
model comprises incapability of avoiding the obstacle or capability
of avoiding the obstacle, and when the output of the second spatial
model is capability of avoiding the obstacle, the output of the
second spatial model is further provided with a second initial
obstacle avoiding route; the step S4 of judging whether the output
of the initial route planning model is capable of avoiding the
obstacle or not comprises:
[0073] S401, judging whether both the output corresponding to the
first spatial model and the output corresponding to the second
spatial model are capable of avoiding the obstacle or not;
[0074] S402, if both the output corresponding to the first spatial
model and the output corresponding to the second spatial model are
capable of avoiding the obstacle, calculating a similarity value
between the first initial obstacle avoiding route and the second
initial obstacle avoiding route according to a preset route
similarity calculating method;
[0075] S403, judging whether the similarity value between the first
initial obstacle avoiding route and the second initial obstacle
avoiding route is greater than a preset similarity threshold value
or not; and
[0076] S404, if the similarity value between the first initial
obstacle avoiding route and the second initial obstacle avoiding
route is greater than the preset similarity threshold value,
judging that the output of the initial route planning model is
capable of avoiding the obstacle.
[0077] Therefore, it is more accurate to judge obstacle avoidance.
As above-mentioned hereinbefore, the four signal acquisition modes
of the application are different in specific execution processes,
and reasons and advantages thereof are different. Here, the initial
millimeter-wave radar signal and the initial image signal are
grouped to model, and then the initial laser radar signal and the
initial ultrasonic signal are grouped to model. By fully
considering the initial millimeter-wave radar signal, the initial
laser radar signal, the initial image signal and the initial
ultrasonic signal to establish a single spatial model, as it is
needed to consider the difference in the four signals, it takes a
long time to establish the spatial model, which is disadvantageous
to the integral obstacle avoiding scheme. Therefore, as the initial
millimeter-wave radar signal and the initial image signal are
grouped to model, and then the initial laser radar signal and the
initial ultrasonic signal are grouped to model, it is only consider
a difference between the two signals. Meanwhile, the four signals
will have positive influence to final obstacle avoiding judgment,
and therefore, information loss is not caused. Further, here, it
has a further important characteristic that in particular, initial
millimeter-wave radar signal and the initial image signal are
grouped to model, and then the initial laser radar signal and the
initial ultrasonic signal are grouped to model. It is because of
higher similarity between the model established by grouping the
initial millimeter-wave radar signal and the initial image signal
and a true space in an efficient modeling process within a short
time and higher similarity between the model established by
grouping the initial laser radar signal and the initial ultrasonic
signal and the true space, which may be associated with
characteristics of high millimeter-wave radar moving object
capturing ability, high laser radar fine object sensitivity, a wide
binocular vision near distance view, good ultrasonic near distance
directivity and the like. A more specific principle is still under
analysis, which is yet a finding of the application. The similarity
value between the first initial obstacle avoiding route and the
second initial obstacle avoiding route can be calculated by any
feasible method according to the preset route similarity
calculation method, for example, a length difference value and an
angle difference value are calculated respectively as being divided
into limited line segments and then the similarity value of a whole
curve is calculated comprehensively, which is actually similarity
calculation between two curved, and is not described repeatedly
herein.
[0078] As described in the step S3, a motion parameter and a body
shape parameter of the unmanned aerial vehicle are acquired and the
parameters are input into an initial route planning model
corresponding to the initial three-dimensional environmental model
based on a genetic algorithm to process so as to obtain an output
of the initial route planning model, wherein the motion parameter
at least includes a position parameter of the unmanned aerial
vehicle, the output of the initial route planning model comprises
incapability of avoiding an obstacle or capability of avoiding the
obstacle, and when the output of the initial route planning model
is incapability of avoiding the obstacle, the output of the initial
route planning model is further provided with an initial obstacle
avoiding route. The motion parameter, for example is position,
speed, acceleration and the like of the unmanned aerial vehicle,
and the body shape parameter, for example, a total length of a
body, a width and a height of the body and the like. It is because
of a relatively small obstacle avoiding space in the obstacle
avoiding process, and therefore, it is needed to consider the
motion parameter and the body shape parameter comprehensively to
determine whether the unmanned aerial vehicle can pass through or
not. The application adopts the initial route planning mode
corresponding to the initial three-dimensional environmental model
based on the genetic algorithm to judge whether the unmanned aerial
vehicle can avoid the obstacle or not and calculate the obstacle
avoiding route by utilizing a machine learning model. The genetic
algorithm is a method of searching for the optimum solution by
simulating a natural evolutionary process, which has a lot of
failure examples in an initial training stage. New training every
time may inherit experience of the previous failure sample till the
feasible obstacle avoiding flight route can be obtained
finally.
[0079] Further, before the step of acquiring a motion parameter and
a body shape parameter of the unmanned aerial vehicle and inputting
the parameters into an initial route planning model corresponding
to the initial three-dimensional environmental model based on a
genetic algorithm to process so as to obtain an output of the
initial route planning model, wherein the motion parameter at least
includes a position parameter of the unmanned aerial vehicle, the
output of the initial route planning model comprises incapability
of avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the initial route planning model is incapability
of avoiding the obstacle, the output of the initial route planning
model is further provided with an initial obstacle avoiding route,
the method includes:
[0080] S21, retrieving an appointed standard environment model the
most similar to the initial three-dimensional environment model
from a preset environment model base, wherein the environment model
base pre-stores a plurality of standard environment models obtained
by modeling a plurality of air obstacle environments;
[0081] S22, acquiring an appointed route planning model
corresponding to the appointed standard environment model according
to a mapping relationship between the preset standard environment
model and the route planning model, wherein the different standard
environment models correspond to different route planning models
based on the genetic algorithm;
[0082] S23, marking the appointed route planning model as the
initial route planning model and generating a route planning
instruction, wherein the route planning instruction is used for
indicating the motion parameter and the body shape parameter of the
unmanned aerial vehicle and inputting the parameters into the
initial route planning model corresponding to the initial
three-dimensional environmental model based on the genetic
algorithm to process.
[0083] Therefore, the initial route planning model is obtained. It
should be noted that it is hard to apply the genetic algorithm in a
common real-time obstacle avoiding scheme of the unmanned aerial
vehicle because the genetic algorithm can be regarded as a trial
and error algorithm which inherits experience of failed simulated
flight previously till the feasible obstacle avoiding flight route
can be obtained finally. As far as common real time obstacle
avoidance of the unmanned aerial vehicle is concerned, as the
spatial environment is a novel environment and it is needed to
obtain the obstacle avoiding route within a short time in a flight
state, it is contradicted with the mode based on a genetic
algorithm which takes a long time. An environment model base is
established in advance. The environment model base pre-stores a
plurality of standard environment models obtained by modeling a
plurality of air obstacle environments, such that the spatial
environment of the unmanned aerial vehicle is a known standard
environment (the quantity of models of the standard environment
modes is enough and the specificity is high enough, and in
particular, common environments of the unmanned aerial vehicle of
the application are collected in a targeted manner to model).
Training on the machine learning model based on genetic algorithm
is conducted on these standard environment models respectively,
such that corresponding route planning models are obtained
respectively after training of enough rounds is conducted, and
meanwhile, a mapping relationship between the standard environment
model and the route planning model can be established. An appointed
route planning model corresponding to the appointed standard
environment model can be obtained according to the mapping
relationship between the standard environment model and the route
planning model, and then the appointed route planning model is
marked as the initial route planning model.
[0084] As described in the steps S4-S5, whether the output of the
initial route planning model is capable of avoiding the obstacle or
not is judged; if yes, an obstacle avoiding flight instruction is
generated to require the unmanned aerial vehicle to fly through the
first environmental area in an obstacle avoiding manner along the
initial obstacle avoiding route. S5, as the output of the initial
route planning model is capable of avoiding the obstacle and the
output of the initial route planning model is provided with the
initial obstacle avoiding route, the unmanned aerial vehicle only
needs to fly in an obstacle avoiding manner along the initial
obstacle avoiding path to fly through the first environmental area
to complete the obstacle avoiding task.
[0085] Further, after the step S4 of judging whether the output is
capable of avoiding an obstacle or not, the method including:
[0086] S411, if the output of the initial route planning model is
incapable of avoiding the obstacle, controlling the unmanned aerial
vehicle to retreat at a preset first horizontal length under a
premise of keeping a current height;
[0087] S412, conducting signal acquisition processing on the second
environmental area to obtain a third millimeter-wave radar signal,
a third laser radar signal, a third image signal and a third
ultrasonic signal by utilizing millimeter-wave radar, laser radar,
a binocular vision camera and an ultrasonic transceiver
pre-arranged on the unmanned aerial vehicle, wherein the second
environmental area is an area obtained as the first environment
area translates horizontally at the first horizontal length along a
retreating direction of the unmanned aerial vehicle;
[0088] S413, generating a third three-dimensional environmental
model according to the third millimeter-wave radar signal, the
third laser radar signal, the third image signal and the third
ultrasonic signal and according to the preset dynamic environment
real-time modeling method;
[0089] S414, inputting the motion parameter and the body shape
parameter of the unmanned aerial vehicle into a third route
planning model corresponding to the third three-dimensional
environmental model based on a genetic algorithm to process so as
to obtain an output of the third route planning model, wherein the
output of the third route planning model comprises incapability of
avoiding an obstacle or capability of avoiding the obstacle, and
when the output of the third route planning model is incapability
of avoiding the obstacle, the output of the third route planning
model is further provided with a third obstacle avoiding route;
[0090] S415, judging whether the output of the third route planning
model is capable of avoiding an obstacle or not;
[0091] S416, if yes, generating a third obstacle avoiding flight
instruction to require the unmanned aerial vehicle to fly through
the second environmental area in an obstacle avoiding manner along
the third obstacle avoiding route.
[0092] Therefore, the unmanned aerial vehicle flies in an obstacle
avoiding manner again. When the unmanned aerial vehicle encounters
the obstacle, for example, a high density bush, the bush blocks all
routes in front, such that the output of the initial route planning
model is incapable of avoiding the obstacle. At the time, the
unmanned aerial vehicle avoids the obstacle again in a special
manner, i.e., the unmanned aerial vehicle is controlled to retreat
at a preset first horizontal length under the premise of keeping a
current height. On this basis, signal acquisition,
three-dimensional environment modeling, obstacle avoiding judgment
and obstacle avoiding flight are conducted again. As the unmanned
aerial vehicle retreats at the preset first horizontal length, the
obstacle avoiding environment of the unmanned aerial vehicle is
improved, and as the unmanned aerial vehicle retreats at the preset
first horizontal length, the safety of the unmanned aerial vehicle
can be guaranteed. Therefore, the unmanned aerial vehicle avoids
the obstacle by adopting a policy of retreating in order to
advance.
[0093] According to the obstacle avoiding method for an unmanned
aerial vehicle based on the multi-signal acquisition and route
planning model, the application provides an obstacle recognizing
and avoiding method by combining a dynamic environment real-time
modeling method and a genetic algorithm, which can be adapted to
any feasible environment, in particular to a complex agricultural
environment (the unmanned aerial vehicle needs to operate at a low
altitude to come across obstacles which are not encountered by the
common unmanned aerial vehicle, for example twigs), and the
unmanned aerial vehicle is particularly integrated with
millimeter-wave radar, laser radar, the binocular vision camera and
the ultrasonic transceiver for signal acquisition, such that an
integrated obstacle fast diagnosis model is established by making
full use of characteristics of high millimeter-wave radar moving
object capturing ability, high laser radar fine object sensitivity,
a wide binocular vision near distance view, good ultrasonic near
distance directivity and the like. A method for recognizing fixed
fine obstacles and protruding obstacles based on an actual scene in
a moving state of the unmanned aerial vehicle, an obstacle quick
detection device integrating millimeter-wave radar, laser radar,
binocular vision and ultrasonic waves is developed, and quick
diagnosing and obstacle avoiding problems of the fine objects and
protrusions are solved. The obstacle avoiding route planning is
conducted in real time by combining the moving characteristics of
the unmanned aerial vehicle, such that multidirectional effective
obstacle avoidance of the unmanned aerial vehicle within a 0-12 m/s
speed range is achieved.
[0094] Referring to FIG. 2, the embodiment of the application
provides an obstacle avoiding apparatus for an unmanned aerial
vehicle based on a multi-signal acquisition and route planning
model, including the following steps:
[0095] an information acquisition unit for conducting signal
acquisition processing on a first environmental area to obtain an
initial millimeter-wave radar signal, an initial laser radar
signal, an initial image signal and an initial ultrasonic signal by
utilizing millimeter-wave radar, laser radar, a binocular vision
camera and an ultrasonic transceiver pre-arranged on the unmanned
aerial vehicle;
[0096] an initial three-dimensional environmental model generation
unit for generating an initial three-dimensional environmental
model according to the initial millimeter-wave radar signal, the
initial laser radar signal, the initial image signal and the
initial ultrasonic signal and according to a preset dynamic
environment real-time modeling method;
[0097] an initial route planning model processing unit for
acquiring a motion parameter and a body shape parameter of the
unmanned aerial vehicle and inputting the parameters into an
initial route planning model corresponding to the initial
three-dimensional environmental model based on a genetic algorithm
to process so as to obtain an output of the initial route planning
model, wherein the motion parameter at least comprises a position
parameter of the unmanned aerial vehicle, the output of the initial
route planning model comprises incapability of avoiding an obstacle
or capability of avoiding the obstacle, and when the output of the
initial route planning model is incapability of avoiding the
obstacle, the output of the initial route planning model is further
provided with an initial obstacle avoiding route;
[0098] an obstacle avoiding judging unit for judging whether the
output of the initial route planning model is capable of avoiding
an obstacle or not; and
[0099] an obstacle avoiding flight instruction generation unit for
generating an obstacle avoiding flight instruction to require the
unmanned aerial vehicle to fly through the first environmental area
in an obstacle avoiding manner along the initial obstacle avoiding
route if the output of the initial route planning model is capable
of avoiding the obstacle.
[0100] Operations for execution by the units correspond to the
steps of the obstacle avoiding method for an unmanned aerial
vehicle based on a multi-signal acquisition and route planning
model of the previous embodiment one by one, which is not described
repeatedly herein.
[0101] According to the obstacle avoiding apparatus for an unmanned
aerial vehicle based on the multi-signal acquisition and route
planning model, the application provides an obstacle recognizing
and avoiding method by combining a dynamic environment real-time
modeling method and a genetic algorithm, which can be adapted to
any feasible environment, in particular to a complex agricultural
environment (the unmanned aerial vehicle needs to operate at a low
altitude to come across obstacles which are not encountered by the
common unmanned aerial vehicle, for example twigs), and the
unmanned aerial vehicle is particularly integrated with
millimeter-wave radar, laser radar, the binocular vision camera and
the ultrasonic transceiver for signal acquisition, such that an
integrated obstacle fast diagnosis model is established by making
full use of characteristics of high millimeter-wave radar moving
object capturing ability, high laser radar fine object sensitivity,
a wide binocular vision near distance view, good ultrasonic near
distance directivity and the like. A method for recognizing fixed
fine obstacles and protruding obstacles based on an actual scene in
a moving state of the unmanned aerial vehicle, an obstacle quick
detection device integrating millimeter-wave radar, laser radar,
binocular vision and ultrasonic waves is developed, and quick
diagnosing and obstacle avoiding problems of the fine objects and
protrusions are solved. The obstacle avoiding route planning is
conducted in real time by combining the moving characteristics of
the unmanned aerial vehicle, such that multidirectional effective
obstacle avoidance of the unmanned aerial vehicle within a 0-12 m/s
speed range is achieved.
[0102] Referring to FIG. 2, the embodiment of the present invention
further provides a computer device, wherein the computer device can
be a server, the internal structure of which may be shown in a
figure. The computer device includes a processor, a memory, a
network interface and a database connected via a system bus. The
processor designed by the computer is used for providing
calculation and control abilities. The memory of the computer
device includes a nonvolatile storage medium and an internal
memory. The nonvolatile storage medium stores an operating system,
a computer program and a database. The internal memory provides an
environment for operation of the operating system and the computer
program in the nonvolatile storage medium. The database of the
computer device is used for storing data used by the obstacle
avoiding method for an unmanned aerial vehicle based on the
multi-signal acquisition and route planning model. The network
interface of the computer device is used for connected
communication with an external terminal via a network. The computer
program is executed by the processor to implement the obstacle
avoiding method for an unmanned aerial vehicle based on the
multi-signal acquisition and route planning model.
[0103] The processor executes the obstacle avoiding method for an
unmanned aerial vehicle based on a multi-signal acquisition and
route planning model, wherein the steps included in the step
correspond to the steps of the obstacle avoiding method for an
unmanned aerial vehicle based on a multi-signal acquisition and
route planning model executing the previous embodiment one by one,
which is not described repeatedly herein.
[0104] Those skilled in the art can understand that the structure
illustrated in the figure is merely a block diagram of a partial
structure related to the scheme of the application and does not
constitute limitation to the computer device in the scheme of the
application applied thereto.
[0105] According to the computer device, the application provides
an obstacle recognizing and avoiding method by combining a dynamic
environment real-time modeling method and a genetic algorithm,
which can be adapted to any feasible environment, in particular to
a complex agricultural environment (the unmanned aerial vehicle
needs to operate at a low altitude to come across obstacles which
are not encountered by the common unmanned aerial vehicle, for
example twigs), and the unmanned aerial vehicle is particularly
integrated with millimeter-wave radar, laser radar, the binocular
vision camera and the ultrasonic transceiver for signal
acquisition, such that an integrated obstacle fast diagnosis model
is established by making full use of characteristics of high
millimeter-wave radar moving object capturing ability, high laser
radar fine object sensitivity, a wide binocular vision near
distance view, good ultrasonic near distance directivity and the
like. A method for recognizing fixed fine obstacles and protruding
obstacles based on an actual scene in a moving state of the
unmanned aerial vehicle, an obstacle quick detection device
integrating millimeter-wave radar, laser radar, binocular vision
and ultrasonic waves is developed, and quick diagnosing and
obstacle avoiding problems of the fine objects and protrusions are
solved. The obstacle avoiding route planning is conducted in real
time by combining the moving characteristics of the unmanned aerial
vehicle, such that multidirectional effective obstacle avoidance of
the unmanned aerial vehicle within a 0-12 m/s speed range is
achieved.
[0106] An embodiment of the application further provides a computer
readable storage medium, having the computer program stored
thereon, the computer program is executed by the processor to
realize the obstacle avoiding method for an unmanned aerial vehicle
based on a multi-signal acquisition and route planning model,
wherein the steps included in the step correspond to the steps of
the obstacle avoiding method for an unmanned aerial vehicle based
on a multi-signal acquisition and route planning model executing
the previous embodiment one by one, which is not described
repeatedly herein.
[0107] According to the computer readable storage medium, the
application provides an obstacle recognizing and avoiding method by
combining a dynamic environment real-time modeling method and a
genetic algorithm, which can be adapted to any feasible
environment, in particular to a complex agricultural environment
(the unmanned aerial vehicle needs to operate at a low altitude to
come across obstacles which are not encountered by the common
unmanned aerial vehicle, for example twigs), and the unmanned
aerial vehicle is particularly integrated with millimeter-wave
radar, laser radar, the binocular vision camera and the ultrasonic
transceiver for signal acquisition, such that an integrated
obstacle fast diagnosis model is established by making full use of
characteristics of high millimeter-wave radar moving object
capturing ability, high laser radar fine object sensitivity, a wide
binocular vision near distance view, good ultrasonic near distance
directivity and the like. A method for recognizing fixed fine
obstacles and protruding obstacles based on an actual scene in a
moving state of the unmanned aerial vehicle, an obstacle quick
detection device integrating millimeter-wave radar, laser radar,
binocular vision and ultrasonic waves is developed, and quick
diagnosing and obstacle avoiding problems of the fine objects and
protrusions are solved. The obstacle avoiding route planning is
conducted in real time by combining the moving characteristics of
the unmanned aerial vehicle, such that multidirectional effective
obstacle avoidance of the unmanned aerial vehicle within a 0-12 m/s
speed range is achieved.
[0108] Those skilled in the art can understand that implementation
of all or part of flows in the method of the embodiment is
completed by means of hardware related to the computer program or
instruction. The computer program can be stored in a nonvolatile
computer readable storage medium. When the computer program is
executed, it can include the flows of the embodiments of the
methods. Any citation of the memory, storage, database or other
media provided by the application and used in the embodiments can
include a nonvolatile and/or volatile memory. The nonvolatile
memory can include read-only memory (ROM), programmable ROM (PROM),
electrical programmable ROM (EPROM), an erasable programmable ROM
(EEPROM) or a flash memory. The volatile memory can include a
random access memory (Ram) or an external high speed cache memory.
As description rather than limitation, the RAM can be obtained in
various forms, for example, a static RAM (SRAM), a dynamic RAM
(DRAM), a synchronous DRAM (SDRAM), a double data rate SDRM
(SSRSDRAM), an enhanced SDRAM (ESDRAM), a synchronization link
(Synchlink) DRAM (SLDRAM), a memory bus (Rambus), a direct RAM
(RDRAM), a direct memory bus dynamic RAM (DRDRAM), a memory bus
dynamic RAM (RDRAM) and the like.
[0109] It should be noted that the terms "comprises", "include" or
any other variants herein are intended to cover nonexcludable
inclusion, such that the process, apparatus, article or method
including a series of elements only includes the elements, but also
includes other elements which are not limited clearly, or further
includes all inherent elements of the process, apparatus, article
or method. Under a circumstance of no more limitations, for the
elements defined by the term "include one", a condition that there
are additional same elements in the process, apparatus, article or
method including the elements is not excluded.
[0110] The above is merely preferred embodiments of the application
and does not hence limit the patent range of the application.
Equivalent structure or equivalent flow conversion made by means of
the contents of the description and drawings of the application are
applied to other related technical fields directly or indirectly,
which is, in a similar way, comprised in the protection scope of
the patent of the application.
* * * * *