# US 2025/0147518 A1 · Autonomous environmental perception, path planning and dynamic landing method and system of unmanned aerial vehicle

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0147518 A1
- Document ID: US20250147518A1
- Status: Published application
- Kind: A1
- Application number: 19/017,245
- Filed: 2025-01-10
- Published: 2025-05-08
- Earliest claimed: 2024-08-22
- Assignees: China Jiliang University
- Applicants: China Jiliang University
- Inventors: Wensong JIANG; Zai LUO; Linzhen SHI; Jie ZHU; Li YANG; Minyue LI; Yaru LI; Dian BIAN; Yaxiong HE
- Web page: https://patentgrep.com/patent/US20250147518A1

## CPC

- G05D 1/644: Optimisation of travel parameters, e.g. of energy consumption, journey time or distance
- G05D 1/2465: {using a 3D model of the environment}
- G05D 1/622: Obstacle avoidance (predicting or avoiding probable or impending collision of road vehicles B60W30/08)
- G05D 1/654: Landing (docking at a base station G05D1/661)
- G05D 2101/15: using machine learning, e.g. neural networks
- G05D 2107/30: Off-road
- G05D 2109/20: Aircraft, e.g. drones
- Y02T 10/40: Engine management systems
- G05D: SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- Y02T: CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION

## Abstract

An autonomous environmental perception, path planning and dynamic landing method includes: obtaining three-dimensional environment information in real time; determining a global starting point and a global end point, and generating an initial path; optimizing the initial path based on a local path optimization algorithm to obtain a first optimized path; when a perception threshold of the current position of the unmanned aerial vehicle is greater than a preset threshold, optimizing the initial path based on a frontier-perceived path optimization method to obtain a second optimized path and a local end point; when the unmanned aerial vehicle advances to the local end point, switching to optimizing the initial path in real time based on the local path optimization algorithm; and when the unmanned aerial vehicle arrives at the global end point, carrying out dynamic landing based on a deep reinforcement learning algorithm.

## Claims

### Claim 1

An autonomous environmental perception, path planning and dynamic landing method of an unmanned aerial vehicle, comprising the following steps:
S1. obtaining depth information and two-dimensional plane information of an environment in real time by the unmanned aerial vehicle, and generating three-dimensional environmental information in space through a three-dimensional reconstruction algorithm based on the depth information and the two-dimensional plane information;
S2. determining a global starting point and a global end point, and generating an initial path according to the global starting point and the global end point;
S3. optimizing the initial path in real time based on a local path optimization algorithm to obtain a first optimized path, and judging in real time whether a perception threshold of the current position of the unmanned aerial vehicle is greater than a preset threshold; when the perception threshold is greater than the preset threshold, proceeding into S4; otherwise, continuing to perform S3, until the global end point is reached, and proceeding into S7;
wherein a calculation formula of the perception threshold is as follows:
τ
per
=
Γ
optimiz
⁢
e
⁢
d
Γ
i
⁢
n
⁢
i
⁢
t
;
wherein τper represents the perception threshold, Γinit represents the initial path within the field of view of the unmanned aerial vehicle, Γoptimized represents the first optimized path within the field of view of the unmanned aerial vehicle, and τ represents the preset threshold, when τper>τ, proceeding into S4;
S4. perceiving the frontier of the field of view of the unmanned aerial vehicle, storing the perceived frontier in the form of space points and recording as frontier points, and performing distance optimization on the frontier points to obtain a frontier space point set P;
S5. according to the frontier space point set P, optimizing the initial path in real time based on a frontier-perceived path optimization method to obtain a second optimized path and a local end point;
S6. switching to optimizing the initial path in real time based on the local path optimization algorithm when the unmanned aerial vehicle advances along the second optimized path to the local end point; and
S7. carrying out dynamic landing based on a deep reinforcement learning algorithm after the unmanned aerial vehicle arrives at the global end point.

### Claim 2 (depends on 1)

The autonomous environmental perception, path planning and dynamic landing method according to claim 1, wherein in S4, performing distance optimization on the frontier points to obtain a frontier space point set P comprises:
constructing a cube area with a side length of 2 m with the frontier point as a central point, dividing the cube area into 8 sub-cube areas with a side length of m, and taking vertices of the sub-cube areas as collision judgment points to judge whether the distance between each collision judgment point and a target obstacle is less than a preset collision distance; and if the distance between at least one collision judgment point and the target obstacle is less than the present collision distance, deleting the frontier point;
traversing all frontier points by repeating the above steps, and constructing the frontier space point set P according to the frontier points finally retained.

### Claim 3 (depends on 2)

The autonomous environmental perception, path planning and dynamic landing method according to claim 2, wherein in S5, according to the frontier space point set P, optimizing the initial path in real time based on a frontier-perceived path optimization method to obtain a second optimized path and a local end point comprises:
S51. determining the local end point;
according to the view distance of the unmanned aerial vehicle, selecting the point on the initial path that is farthest from the current position of the unmanned aerial vehicle as the initial local end point; and if the initial local end point is within the obstacle range, selecting the second local end point by doubling the view distance of the unmanned aerial vehicle based on the initial local end point until the selected local end point is outside the obstacle range;
S52. determining decision indicators, and calculating a decision function according to the decision indicators; wherein the decision indicators comprise: adjacent point constraint, distance constraint, and direction constraint;
the expressions of the adjacent point constraint are as follows:
d
A
(
p
i
)
=

p
c
-
p
i

;
f
A
(
i
)
=
{
d
A
⁢
(
p
i
)
-
d
min
)
/
(
d
max
-
d
min
)
d
min
<
d
⁡
(
i
)
<
d
max
1
others
;
wherein dA represents the distance between adjacent points, fA represents a cost function of the adjacent points, pi represents the i-th frontier point in the frontier space point set P, pc represents the current position of the unmanned aerial vehicle, ∥·∥ represents the Euclidean distance operation, dmin represents the minimum value of the selected point range within the field of view of the unmanned aerial vehicle, and dmax represents the maximum value of the selected point range within the field of view of the unmanned aerial vehicle;
the expressions of the distance constraint are as follows:
d
L
(
p
i
)
=

p
g
-
p
i

;
f
L
(
i
)
=
d
L
(
p
i
)
/
(
d
L
⁢
max
-
d
L
⁢
min
)
;
wherein dL represents the distance between the current position pc of the unmanned aerial vehicle and the local end point pg, fL represents a distance cost function, pg represents the local end point, dLmin represents the minimum value of the distance from all frontier points in the frontier space point set P to the local end point pg, and dLmax represents the maximum value of the distance from all frontier points in the frontier space point set P to the local end point pg;
the expressions of the direction constraint are as follows:
d
D
(
p
i
)
=
pi
-
pc

p
i
-
p
c

·
v
c

v
c

+
1
;
f
D
(
i
)
=
(
d
D
(
p
i
)
-
d
D
⁢
min
)
/
(
d
D
⁢
max
-
d
D
⁢
min
)
;
wherein dD represents a specific value of the deviation from a current speed direction, fD represents a direction cost function, vc represents a current speed of the unmanned aerial vehicle, dDmin represents the minimum value of the deviation from the current speed direction, and dDmax represents the maximum value of the deviation from the current speed direction;
the calculation formula of the decision function is as follows:
p
b
=
arg
⁢
min
p
i
∈
P
⁢
λ
A
⁢
f
A
(
p
i
)
+
λ
L
⁢
f
L
(
p
i
)
+
λ
D
⁢
f
D
(
p
i
)
;
wherein λA, λL and λD represent the weight of the adjacent point constraint, the weight of the distance constraint, and the weight of the direction constraint respectively, and pb represents an optimal replanning point;
S53. generating the second optimized path according to the current position pc of the unmanned aerial vehicle and the optimal replanning point pb.

### Claim 4 (depends on 3)

The autonomous environmental perception, path planning and dynamic landing method according to claim 3, wherein in S6, switching to optimizing the initial path in real time based on the local path optimization algorithm when the unmanned aerial vehicle advances along the second optimized path to the local end point comprises:
S61. when the unmanned aerial vehicle advances along the second optimized path, each time a preset time threshold or a preset distance threshold is passed, reperforming S5 and updating the optimal replanning point pb and the second optimized path;
S62. recording a path length Lthresh that the unmanned aerial vehicle flies along the second optimized path from the current position, and when the path length Lthresh is greater than a preset path length, reselecting the local end point pg, reperforming S5 according to the new local end point, and updating the optimal replanning point pb and the second optimized path;
wherein the local end point pg is reselected, specifically as follows:
a new local end point is selected by doubling the view distance of the unmanned aerial vehicle based on the current local end point;
S63. switching to optimizing the initial path in real time based on the local path optimization algorithm when the unmanned aerial vehicle arrives at the local end point pg.

### Claim 5 (depends on 4)

The autonomous environmental perception, path planning and dynamic landing method according to claim 4, wherein optimizing the initial path in real time based on the local path optimization algorithm to obtain the first optimized path, or optimizing the initial path in real time based on the frontier-perceived path optimization method to obtain the second optimized path further comprises:
inserting a middle point according to the current position of the unmanned aerial vehicle and the local end point, and generating the first optimized path or the second optimized path;
the calculation formula of the middle point is as follows:
p
m
=
p
c
+
p
g
2
+
v
c
3
2
·
a
max
·

v
c
·
v
c
×
(
p
g
-
p
c
)

v
c
⁢
||
·
||
p
g
-
p
c


(
v
c
×
(
p
g
-
p
c
)

v
c

·
||
p
g
-
p
c

)
3
;
wherein pm represents the middle point, and amax represents the maximum angular velocity of the unmanned aerial vehicle.

### Claim 6 (depends on 1)

The autonomous environmental perception, path planning and dynamic landing method according to claim 1, wherein in S7, carrying out dynamic landing based on a deep reinforcement learning algorithm after the unmanned aerial vehicle arrives at the global end point comprises:
S71. constructing a landing control agent based on a deep deterministic policy gradient algorithm;
S72. obtaining observation values of the unmanned aerial vehicle and the landing platform;
S73. defining a reward function; wherein the reward function comprises a horizontal direction reward function and a vertical direction reward function;
S74. judging whether the unmanned aerial vehicle meets landing conditions according to the observation values of the unmanned aerial vehicle and the landing platform and the reward function;
S75. controlling the unmanned aerial vehicle by a landing control agent according to the observation values of the unmanned aerial vehicle and the landing platform, the reward function, and whether the unmanned aerial vehicle meets the landing conditions until the unmanned aerial vehicle completes the dynamic landing.

### Claim 7 (depends on 1)

An autonomous environment perception, path planning and dynamic landing system of an unmanned aerial vehicle, configured to perform the autonomous environment perception, path planning and dynamic landing method according to claim 1, comprising the following modules:
a data acquisition module, configured to acquire depth information and two-dimensional plane information of an environment in real time by the unmanned aerial vehicle, and generate three-dimensional environmental information in space through a three-dimensional reconstruction algorithm based on the depth information and the two-dimensional plane information;
an initial path generation module, connected to the data acquisition module and configured to determine a global starting point and a global end point, and generate an initial path according to the global starting point and the global end point;
an optimized path generation module, connected with the data acquisition module and the initial path generation module, and configured to optimize the initial path in real time based on the local path optimization algorithm to obtain the first optimized path, and meanwhile, judge in real time whether the perception threshold of the current position of the unmanned aerial vehicle is greater than a preset threshold; wherein a calculation formula of the perception threshold is as follows:
τ
per
=
Γ
optimiz
⁢
e
⁢
d
Γ
i
⁢
n
⁢
i
⁢
t
;
wherein τper represents the perception threshold, Γinit represents the initial path within the field of view of the unmanned aerial vehicle, Γoptimized represents the first optimized path within the field of view of the unmanned aerial vehicle, and τ represents the preset threshold; if the perception threshold is greater than the preset threshold, perceive the frontier of the field of view of the unmanned aerial vehicle, store the perceived frontier in the form of space points and record as frontier points, and perform distance optimization on the frontier points to obtain a frontier space point set P; according to the frontier space point set P, optimize the initial path in real time based on a frontier-perceived path optimization method to obtain a second optimized path and a local end point; judge the distance between the current position of the unmanned aerial vehicle and the local end point in real time when the unmanned aerial vehicle advances along the second optimized path, and switch to optimizing the initial path in real time based on the local path optimization algorithm when the distance is less than the preset distance; otherwise, continue to optimize the initial path in real time based on the local path optimization algorithm until the global end point is reached; and
a dynamic landing module, connected to the optimized path generation module and configured to carry out dynamic landing based on a deep reinforcement learning algorithm after the unmanned aerial vehicle arrives at the global end point.
