Method for optimizing decision-making regulation and control, method for controlling traveling of vehicle, and related apparatus

Patent metadata

Field	Value
Publication number	US-12168455-B2
Application number	US-202218048679-A
Country	US
Kind code	B2
Filing date	Oct 21, 2022
Priority date	Apr 23, 2020
Publication date	Dec 17, 2024
Grant date	Dec 17, 2024

Abstract

In the method for optimizing decision-making regulation and control, a first traveling sequence is obtained, where the first traveling sequence includes a first trajectory sequence of the vehicle in information about a first environment and first target driving behavior output by a behavior decision-making layer of a decision-making and control system based on the information about the first environment. A second traveling sequence is obtained, where the second traveling sequence includes a second trajectory sequence output by a motion planning layer of the decision-making and control system based on preset second target driving behavior and the second target driving behavior. The behavior decision-making layer is optimized based on a difference between the first traveling sequence and a preset traveling sequence, and the motion planning layer is optimized based on a difference between the second traveling sequence and the preset traveling sequence.

First claim

What is claimed is:

1. A method for optimizing decision-making regulation and control, comprising:
obtaining a first trajectory sequence that comprises trajectory information of a vehicle in a first environment;
obtaining first target driving behavior information output by a behavior decision-making layer of a decision-making and control system based on information about the first environment;
combining the first trajectory sequence and the first target driving behavior information to obtain a first traveling sequence;
obtaining a second trajectory sequence output by a motion planning layer of the decision-making and control system based on preset second target driving behavior information;
combining the second trajectory sequence and the second target driving behavior information to obtain a second traveling sequence;
optimizing the behavior decision-making layer based on a difference between the first traveling sequence and a preset target teaching traveling sequence, wherein the target teaching traveling sequence comprises a teaching trajectory sequence and teaching driving behavior information; and
optimizing the motion planning layer based on a difference between the second traveling sequence and the target teaching traveling sequence.

2. The method according to claim 1 , wherein the optimizing the behavior decision-making layer comprises:
obtaining a first output obtained when the first traveling sequence is input into a determining model that is used to determine whether the input traveling sequence is the teaching traveling sequence, and optimizing the behavior decision-making layer based on the first output;
wherein the optimizing the motion planning layer comprises:
obtaining a second output obtained when the second traveling sequence is input into the determining model, and optimizing the motion planning layer based on the second output; and
wherein the method further comprises:
obtaining a third output obtained when the target teaching traveling sequence is input into the determining model, and
optimizing the determining model based on the first output, the second output, and the third output.

3. The method according to claim 2 , wherein the optimizing the motion planning layer based on the second output comprises:
optimizing the motion planning layer based on the second output by using a policy optimization algorithm.

4. The method according to claim 2 , wherein the optimizing the behavior decision-making layer based on the first output comprises:
obtaining a gradient of a first function, and
optimizing the behavior decision-making layer based on the gradient of the first function using a gradient descent algorithm, wherein an independent variable of the first function comprises the first output.

5. The method according to claim 4 , wherein the first function is defined as follows:

1
N

⁢

∑

j
=
1

N

[

1
-
ω

T
j

⁢

∑

t
=
1

T
j

log
⁢

D
ψ

(

s
t
c

,

a
t
c

,

c
j
c

)

-

λ
E

T
j

⁢

∑

t
=
1

T
j

c
j
e

⁢
log
⁢

pc
a

(

c
t
c

❘

s
t
e

,

a

t
-
1

e

)

-

λ
G

T

Assignees

• Huawei Tech Co Ltd

Inventors

• Wang Bin
• Zhuang Yuzheng
• Liu Wulong

Classifications

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title

• B60W2050/0006

  Digital architecture hierarchy · CPC title

• B60W50/0098

  Details of control systems ensuring comfort, safety or stability not otherwise provided for · CPC title

• B60W2050/0083

  Setting, resetting, calibration · CPC title

• B60W60/001

  Planning or execution of driving tasks · CPC title